17
Reqder
qpri ! 2Ol4
Programma donderdagLT april 2014
Care Onderzoekschooldag over kwal itatief onderzoek
Lokatie: De Moulin A, route 124
Ochtendprogrdmma:
9.30 - 10.45 uur
- Ll-.00 uur
11.00 - 12.45 uur
10.45
Analyse met behulp van materiaal diepte-interviews
Docent:
Dr. Gijs Hesselink/Dr. Marianne Dees
Tutor:
Drs. Jasper van Riet Paap
Koffie
Criteria voor beoordeling publicatie van
(kwa litatief) onderzoeksa rti ke!
Analyse opzet artikel met behulp van criteria
12.45
- 13.30 uur
Docent:
Dr. Marieke Groot
Tutor:
Drs. Jasper van Riet Paap
Lunch in restaurant de Haardt
Middagprogromma:
13.30
1.4.45
- 16.30 uur
- 15.00 uur
16.30 uur
!nterviewtraining
Docent:
Drs. Jeanette Heldens
Tutor:
Drs. Jasper van Riet Paap
Thee
Evaluatie en slot
Literqtuur
l7 qpri I v.rlt.
Lacey, Luff: Qualitative Data Analysis (2007l.
Wester: Analyse van kwalitatief onderzoeksmateriaal (2004)
RATS
guidelines: Clark JP. How to peer review a qualitative
manuscript. ln Peer Review in Health Sciences. Second edition.
Edited by Godlee F, Jefferson T. London: BMJ Books; 2003:2L9-
235.
Tong, Sainsbury and Craig: Consolidated criteria for reporting
qualitative research (CORfQ): a 32-item checklist for interviews
and focus groups (2007).
Wray, Lindsay, Crozier, Andrews and Leeson. Exploring
perceptions of psychological services in a children's hospice in the
United Kingdom. Pallitative and Supportive Care 20L3;1L:373-382.
IVaf,ftrnal ínsfffr"rf,e f,rrr
ËÍeatrtfr fr[esearrh
The NIHR ffiereanch Derign Serwice
fsr the East :Midlands
The NIHR Research Design Service
for Yorkshire & the Humber
Qualitative Data
Analysis
Authors
Anne Lacey
Donna Luff
This Resource Pack is one of a series produced by The NIHR RDS for the East
Midlands / The NIHR RDS for Yorkshire and the Humber. This series has been
funded by The NIHR RDS EM / YH.
This Resource Pack may be freely photocopied and distributed for the benefit of
researchers. However it is tne copyright of The NIHR RDS EM / YH and the
authors and as such, no part of the content may be altered without the prior
permission in writing, of the Copyright owner.
Reference as:
Lacey A. and Luff D. Qualitative Research Analysis. The NIHR RDS for the East
Midlands / Yorkshire & the Humber, 2007.
Anne Lacey
The NIHR RDS for the East Midlands /
Yorkshire & the Humber
lnformatics Collaboratory for the Social
Sciences (ICOSS)
The University of Sheffield
219 Portobello
Sheffield S1 4DP
Donna Luff
Harvard Paediatric Health Services
Research Program
Children's Hospital
Boston
USA
Last updated: 2009
The NIHR RDS for the East
Midlands
Division of Primary Care,
14th Floor, Tower building
University of Nottingham
www. rds+astm id la nds.
Leicester:
n ih
r.ac.
u
k
enquiries-LN R@rds+astmid lands.orq. u k
Nottingham: enquiries-N DL@rds-eastmid lands.ors. u k
University Park
Nottingham
NG7 2RD
Tel: 01 í 5 823 0500
The NIHR RDS for
Yorkshire & the Humber
ScHARR
The University of Sheffield
Regent Court
30 Regent Street
Sheffïeld
S1 4DA
Tel: 01 14 222 0828
@
www.rds-vh.nihr.ac.uk
Sheffield:
Leeds:
York:
rds-vh@sheffield.ac.uk
rds-vh@leeds.ac.uk
rds-vh@vork.ac.uk
Gopyright of The NIHR RDS Elf, , Ytl
(2ooe)
The N|HR RDS for the East Midlands / Yorkshire & the Humber 2009 QUALITATIVE DATA ANALYSIS
Table of Gontents
Page
............
1.
lntroduction
2.
What is Qualitative
3.
Theories and Methods in Qualitative Data Analysis
4.
Stages in Qualitative Data
5.
Ensuring
6.
Practicalities
7.
Software
4
Data?
5
.......
Analysis
Rigour
..........
3í
34
8. Summary
37
............
38
Glossary
Appendix
í-
20
26
Packages
References
I
39
Sample Transcript
Appendix 2 - Sample Transcript
(Derek)
41
(Sue)
45
The N|HR RDS for the East Midlands / Yorkshire & the Humber 2009 QUALITATIVE DATA ANALYSIS
í. lntroduction
This resource pack is designed for researchers working in health and social care
who have in mind, or have already embarked upon, a piece of qualitative
research. Qualitative methods, using narrative and observation rather than
numerical data, are increasingly being used in health care settings where they
are seen to 'reach the parts other methods cannot reach' (Pope and Mays 2006),
and they are now seen as part of the mainstream of methods in health services
research (Holloway 2005). However, as qualitative analysis is still new in some
fields of health research, it is relatively common for the qualitative researcher to
disappear under a sea of interview transcripts or field notes, with little in the way
of guidance as to how to proceed. Confusion and de-motivation is the usual
result!
lf you are not already familiar with the basics of qualitative research, we suggest
you first read the The NIHR RDS EM / YH Resource Pack in this series entitled
'lntroduction to Qualitative Research' by Beverley Hancock, Elizabeth Ockleford
& Kate Windridge (Updated 2007), or the relevant chapters of a text on research
methods such as Bowling e044.
We assume you already have some knowledge of qualitative data collection
methods such as participant observation, or in-depth interviewing. However
beyond that we make no assumptions. lf you have a large pile of tapes, field
notes, or transcripts sitting on your desk waiting to be analysed, this pack is
probably for you! Better still, if you are still at the stage of designing some
qualitative research, a knowledge of your proposed methods of analysis will
improve the design and save you a lot of trouble in the later stages.
After the introductory sections, you will find a brief theoretical review of the
different ways of analysing qualitative data, and you will be encouraged to decide
which level of analysis is the right one for you. Then we take you through the
various stages of analysis, using some sample interview transcripts to let you try
out the various processes for yourself. A discussion of processes that can help
ensure rigour follows. Finally we address some of the practicalities of qualitative
analysis, including use of computer software programmes.
:
.
To:.dÍedus§..§ómè of th,è'théoretiöal mödels'within whièh qualitative dataican
o
r
.
To understand the stages involved in qualitative data analysis, and gain some
experience in coding and developing categories.
ïö àssëss how rlgour can bè maiimÏsèd in quàlitàtive dàta analysis,
To appl, practical solutions to the process of qualitative data analysis.
be analysed, and select the most appropriate one for a particular piece of
researcÉ.
Thè NIHR RDS Íor the East Midlands / Yorkshire & the Humber 2009 QUALITATIVE DATA ANALYSIS
2. What is qualitative data?
you are probably familiar with the basic differences between qualitative and
quantitative research methods, and their different applications in dealing with
research questions posed in health care research. Qualitative research is
particularly good at answering the 'why', 'what' or'hoW questions, such as:
Why are some patients with diabetes reluctant to comply with dietary advice
and insulin regimes, despite their experience of diabetic complications?
What are the perceptions of carers living with people with learning disability,
as regards their own health needs?
How is the work of an Accident and Emergency Department affected by
frequent physical and verbal abuse towards staff ?
.
.
.
Each of these questions could be addressed using quantitative techniques such
as structured questionnaires, attitude scaling, measurement of standard
outcomes such as mortality, morbidity or staff absence rates. All of these can be
readily analysed statistically, and you will get some sort of answer to the
question. For instance, you may find that older people are more likely to report
compliance with dietary advice than younger ones, or that episodes of violence
and staff sickness rates are significantly correlated. But that would only answer
part of the question, or may lead you to make assumptions about causes and
effects that are invalid. The problem may be quite different from how we
conceptualise it. To find out more about the connection between age and
complianCe, or between sickness and abuse, or to really "get into the shoes" of a
carer, you will probably need to talk with people in some depth, or observe their
behaviour over a period of time.
Look at the last research question above, about abuse in an A&E department.
What kinds of qualitative data collection methods could be used to investigate this
topic? List four different sources of data that could be used by a researcher.
You have probably listed some of the following:
Transcipts of individual interuiews with members of staff in the department
Focus group transcripts
Field notes from observation of staff meetings
Copr'es of diary entries that staff members have been asked to complete each
day
Critical incident recordings from specific episodes of violence or abuse
Researcher memos and reflections
Video
@kingsomeSenSeofthesesortsofcollectionsofdata?
pages
or lots of megabytes
of typescript,
Qualitative data tends to take up many
on a disc! lt is usually in the form of words and narratives, but may include visual
images, videotape, or other media. Where do we go from here?
The N|HR RDS foÍ the East Midlands / Yorkshire & the Humber 2009 QUALITATIVE DATA ANALYSIS
2.í What do we mean
bY analYsis?
euantitative research techniques generate a mass of numbers that need to be
summarised, described and analysed. Characteristics of the data may be
described and explored by drawing graphs and charts, doing cross tabulalions
and calculating means and standard deviations. Further analysis would blild on
these initial finàings, seeking patterns and relationships in the data by performing
multiple regression, or an analysis of variance perhaps. Advanced modelling
techniques may eventually be used to build sophisticated explanations of how the
data addresses the original question. But many quantitative research projects
would never need to go that far; the question would be answered by simple
descriptive statistics.
So it is with Qua/fative data analysis. The mass of words generated by interviews
or observational data needs to be described and summarised. The question may
require the researchers to seek relationships between various themes that have
been identified, or to relate behaviour or ideas to biographical characteristics of
respondents such as age or gender. lmplications for policy or practice may be
derived from the data, or interpretation sought of puzzling findings from previous
studies. Ultimately theory could be developed and tested using advanced
analytical techniques.
There are no 'quick fix' techniques in qualitative analysis. Just as a software
package such as the Statistical Package for the Social Sciences (SPSS) won't tell
you wÉich of the myriad statistical tests available to use to analyse numerical
àata, so there are probably as many different ways of analysing qualitative data
as there are qualitative researchers doing it! Many would argue that this is the
way it should be - qualitative research is an interpretative and subjective
exércise, and the researcher is intimately involved in the process, not aloof from it
(Pope and Mays 2006).
However there are some theoretical approaches to choose from, which will be
explored in the following section. Furthermore, there are Some common
processes, no matter which approach you take.
Analysis of qualitative data usually goes through some or all of the following
stages (though the order maY vary):
o
.
.
o
Familiarisation with the data through review, reading, listening etc
Transcription of tape recorded material
Organisation and indexing of data for easy retrieval and identification
Anonymising of sensitive data
The NIHR RDS for the East Midlands / Yorkshire & the Humber 2009 QUALITATIVE DATA ANALYSIS
.
o
.
.
Coding (may be called indexing)
ldentification of themes
Re-coding
Development of provisional categories
Exploration of
.
o
re
lationsh ps between categories
i
ReÍinement of themes and categories
Development of theory and incorporation of pre-existing knowledge
Testing of theory against the data
Report writing, including excerpts from original data if appropriate (eg quotes
from interviews)
These stages will be explored further in Section 3, giving you an opportunity to
have a go at coding and developing themes from the sample transcripts you will
find in Appendix 1.
2.2What do you want to get out of your data?
It isn't always necessary to go through all the stages above, just as it isn't always
necessary to use multivariate modelling in statistics! Let's take the example of the
research question about the perceived health needs of carers.
r
What are the perceptions of carers living with people with learning disability,
as regards their own health needs?
You may simply be interested in finding out the community services that should
be provided to meet these perceived needs. You might want to know what sorts
of services are valued or requested by the majority of carers. Maybe several
respondents mention that they struggle with depression and loneliness.
There are three broad levels of analysis that could be pursued here:
o
o
One strategy would be to simply count the number of times a particular word
or concept occurs (eg loneliness) in a narrative. The qualitative data can then
be categorised quantitatively, and subjected to statistical analysis. Policy
decisions could be based on the result. This kind of analysis (sometimes
called content analysisl) is not truly qualitative, however, and will not be
discussed in any detail in this pack.
For a thematic analysis we would want to go deeper than this. All units of data
(eg sentences or paragraphs) referring to loneliness could be given
a
particular code, extracted and examined in more detail. Do participants talk of
being lonely even when others are present? Are there particular times of day
t However, please note that Beverley Hancock, in the introductory resource pack on qualitative
research, defines content analysis more widely. Her definition is closer to what we are calling
thematic analysis.
The NIHR RDS for the East Midlands / Yorkshire & the Humber 2009 OUALITATIVE DATA ANALYSIS
or week when they experience loneliness? ln what terms do they express
loneliness? Do men and women talk of loneliness in different ways? Are those
who speak of loneliness also those who experience depression? Themes
could eventually be developed such as 'lonely but never alone' or'these four
walls'.
.
For a theoretical analysis such as grounded theory (see Section 2) you would
want to go further still. Perhaps you have developed theories as you have
been anàlysing your data about depression being associated with perceived
loss of a 'normalr child/spouse. The disability may be attributed to an accident,
or to some failure of medical care, without which the person cared for would
still be 'normal'. You may be able to test this emerging theory against existing
theories of loss in the literature, or against further analysis of the data. You
may even search for 'deviant cases', that is data which seems to contradict
your theory, and seek to modify your theory to take account of this new
iinOing. This process is sometimes known as 'analytic induction', and is used
to build and test emerging theory.
So some decisions have to be made by the researcher as to the questions she or
he is asking of the data, and the depth of analysis that is required. lt may even
come down to the amount of time you have available, or your ease of access to
adequate resources. Certainly there is no need to do more analysis than your
question demands, but seemingly simple questions have a habit of becoming
more complex along the way! ln the next section we look at different theories and
methods used in qualitative data analysis.
The N|HR RDS for the East Midlands / Yorkshire & the Humber 2009 QUALITATIVE DATA ANALYSIS
3. Theories and methods in qualitative
data analysis
ln Section 1 we looked at some common features of qualitative analysis. ln this
section, we will focus on distinct approaches to undertaking qualitative analysis.
There is no one right way to analyse qualitative data, and there are several
approaches available. Much qualitative analysis falls under the general heading
oi;thematic analysis'. The common features of undertaking a thematic analysis of
qualitative data will be outlined in Section 3. However, there are particular
'schools of thought', or theoretical approaches to qualitative analysis, which it is
important to be familiar with, both for designing your own research and for
critically appraising qualitative research evidence. The particular approach you
take to any given study will depend on many factors, not least: the research
question, the iime you have available and funders' priorities. You will always need
to consider the overall aims of the analysis -what do you want the data to
contribute to, for example answering a specific policy-related question or
generating new conceptual or theoretical understandings in a particular area.
Some approaches to qualitative analysis are probably more familiar to you than
others. For example, many studies in the health field, in particular nursing, state
that they take a 'Grounded Theory' approach to analysis. The aim of this section
is to iniroduce you to two distinct approaches to qualitative analysis. Firstly
Grounded Theory, and secondly Framework Analysis, which is increasingly
commonly used in health-related research.
The key features of each approach will be outlined in turn. You will then be asked
to consider what approaches might be suitable for particular research projects
and to think about how your choice of analysis might affect how you go about
your research.
3.1 Grounded Theory
Grounded Theory evolved out of research by sociologists Glaser and Strauss
(1967). Glaser and Strauss were concerned to outline an inductive method of
be generated
qualitative research which would allow social theory
systematically from data. That is, theories would be 'grounded' in rigorous
empirical research, rather than produced in the abstract. What is a theory? ln this
pack we take this to mean a generalisable idea or concept, gleaned from data,
inat fretps to understand the social world in a new or more sophisticated way, and
that can be tested in different settings.
to
Grounded Theory is a methodology; in other words, it is a way of thinking about
and conceptualising data. lt is an approach to research as a whole and as such
can use a range of different methods. However, researchers frequently use the
analysis procedures outlined in grounded theory without taking on board the
whoÈ methodological approach to research design. Grounded Theory analysis is
inductive, in that the resulting theory 'emerges' from the data through a process
of rigorous and structured analysis.
The NIHR RDS for the East Midlands / Yorkshire & the Humber 2009 QUALITATIVE DATA ANALYSIS
It is important to emphasise that what distinguishes grounded theory from many
other approaches to qualitative analysis is this emphasis on theory as the final
output of research (Strauss and Corbin, 1998). Whereas other forms of qualitative
analysis may legitimately 'stop' at the levels of description or simple
interpretation, ttre aim of grounded theory is theoretical development. lts focus
then is clearly on what has also been called 'analytic induction'
What do grounded theorists mean by theory? A 'grounded theory' consists of
'plausible ielationships' (Strauss & Corbin, 1998) among sets of concepts, which
are directly developed from data analysis. Theory, in this sense, provides a set of
testable propositions that help us to understand our social world more clearly,
rather than an absolute 'truths'.
The appeal of grounded theory analysis is the structured and detailed procedures
for the generation of theory from data. Grounded Theory starts with a clear, but
often bóad, research question. This question identifies the general area to be
studied. The research then traditionally proceeds in stages, with the analysis
performed after one stage of fieldwork determining what or who will be studied
next and which methods will be used. As a result, qualitative and quantitative
methods can be used within the same study at different stages.
ln terms of analysing qualitative data generated, at the heart of grounded theory
is the idea of the constant comparative method. ln this method, concepts or
categories emerging from one stage of the data analysis are compared with
concepts emerging from the next. The researcher looks for relationships between
these concepts and categories, by constantly comparing them, to form the basis
of the emerging theory. The researcher continues with this process of constant
comparison untilthey reach what is called 'theoretical saturation', that is no new
significant categories or concepts are emerging.
ln terms of the process of doing grounded theory analysis, the researcher
typically goes through several procedures. These are not linear stages, rather the
process óf grounded theory is cumulative and can involve frequent revisiting of
data in the iight of the new analytical ideas that emerge as data collection and
resses:
open coding (initialfamiliarisation with the data)
.
.
o
.
.
o
.
o
delineation of emergent concepts
coflc€ptual coding (using emergent concepts)
refinement of conceptual coding schemes
clustering of concepts to form analytical categories
searching for core categories
core categories lead to identification of core theory
testing
of
emerging theory
by reference to other research and
to
social/cultural/economic factors that affect the area qt§lqqfl
Grounded Theory analysis requires'theoretical sensitivity'. This is described as
an ability 'to see the research situation and its associated data in new ways, and
to explore the data's potential for developing theory' (Strauss and Corbin,
The N|HR RDS for the East Midlands / Yorkshire & the Humber 2009 QUALITATIVE DATA ANALYSIS
1990:44). This is a creative process but must be grounded in a scientific
approach. ln Grounded Theory texts, there are also discussions of how to
distance oneself from assumptions before the fieldwork and from the emerging
analysis, in order to return repeatedly to the data. All theoretical developments
are to be seen as provisional until proven by the data and by validation from
others. There is a strong tradition in grounded theory that 'how to do it' can only
be leamt from experience. Mentoring and working in teams have been seen as
important in developing research skills and in ensuring rigour of analysis and
theory generation.
An example of how Grounded Theory analysis proceeds in a research project is
given below. The example comes from a published study (Kendall 1998) of
families of children with Attention Deficit Hyperactivity Disorder (ADHD). The full
reference for the article is given in the reference list at the end of this section. ln
their reporting of methodology, the autho(s) outline how they went about
conducting grounded theory analysis:
The NIHR RDS for the East Midlands / Yorkshire & the Humber 2009 QUALITATIVE DATA ANALYSTS
andtranScribed.Transcriptionsoftheinterviews
comparative method (Glaser, 1976, 1993;
constant
the
using
wàre ànarysed
Glaser anó Strauss,lS6Z;. Data analysis was iterative with data collection. Data
were analysed as they were collected through the process of coding. Through
open codihg, common themes of everyday life were identified and examined in
relation to tÉe context, meanings, and circumstances of living with ADHD or with
an ADHD family member. lnterviews were coded by conceptualising underlying
patterns in the data. lnitial data analysis guided further and more focused data
follection, leading to further conceptualisation of the data and refinement of the
coding schemes.-As part of the analysis, similarities and differences about the
compÍed codes were clustered together to create categories. Conceptual
saturation was reached when no new categories were generated.
Through the process of open coding and theoretical coding, the basic social
psychólogical problem (or core variable) and the basic social process emerged.
Theoreticàl memos were written throughout the coding process to track
conceptual decisions and ideas as they were occurring. Theoretical memos also
were coded using theoretical coding and served as the basis for writing the
grounded theory luring the final phase of the analysis' A computer software
package, The Ethnograph 95, was used to manage the data.
Credibility of the data was established using the techniques of persistent
observation (recurring observations of family members during and between
individual and family interviews), peer debriefing (presenting analyses and
conceptual abstractions of the data to other expert qualitative researchers to
explore inquirer biases and to clarify the meanings and the basis for
interpretations), and member checks (presenting the analysis oÍ the data to
informants for their confirmation or revision) (Lincoln & Guba, 1985). ln addition,
two health professionals who were considered 'experts' on ADHD, and who also
had children with ADHD, reviewed the theoretical codes and grounded theory and
verified that the findings reflected their experience.' (Kendall J, 1998).'
There have been many developments in Grounded Theory since the original text
in 1967. lf you are interested in pursuing Grounded Theory research, these can
be exploreà in tf,e suggested further reading. Strauss and Corbin (1990) in
particular have produced new approaches to, and details of, coding and guides
on how to do grounded theory analysis. Grounded Theory has also been hugely
influential in génerally changing the way that qualitative analysis is undertaken.
Most approaèhes to qualitaÍve analysis, for example, stress the importance of
interpretation being'grounded' in detailed data analysis.
The N|HR RDS for the East Midlands / Yorkshire & the Humber 2OOg QUALITATIVE DATA
ANALYSIS
12
Further Re
unded Theory ln Gerrish K and Lacey A (2006) Ihe
Research Process in Nursing Oxford: Blackwell.
Kendall, J (1gg8)'Ouflasting disruption: the process of reinvestment
ADHD children' Quatitative Health Research 9:2' 166-181'
Glaser,
B and Strauss, A (1967) The
Discovery
of
in families with
Grounded Theory' London:
Weidenfield & Nicolson.
Gtaser, B (1978) Theoreticat sensitivity. california: sociology Press.
Strauss,
A (1987)
Quatitative Anatysis
for Social ScrenÍlsfs. New York:
Cambridge
University Press.
Strauss,
A and Corbin, J (1990)
Basrbs
of Qualitative
Research: Grounded Theory
Procedures and Techniques. London: Sage'
3.2 Framework AnalYsis
A second, more recent, approach to qualitative analysis is gaining popularity in
health - related research, namely Framework Analysis (Ritchie and Spencer,
1gg4). ln contrast to grounded theory, Framework Analysis was explicitly
deveíoped in the contexi of applied policy research. Applied research aims to
meet speciflc information needs and provide outcomes or recommendations,
often within a short timescale. Framework Analysis shares many of the common
features of much qualitative analysis which were outlined in Section 1, and of
what is often called 'thematic analysis'. The benefit of Framework Analysis is that
it provides systematic and visible stages to the analysis process, so that funders
and others, can be clear about the stages by which the results have been
obtained from the data. Also, although the general approach in Framework
Analysis is inductive, this form of analysis allows for the inclusion of a priori as
well às emergent concepts, for example in coding. This can be important in many
applied studies, where there are specific issues that the funders or other
stakeholders want to be addressed.
Framework Analysis has 5 key stages. These can be undertaken in a linear
fashion and theréfore all data can be collected before analysis begins, although
framework analysis can equally be used when data collection and analysis occur
concurrently.
Key stages of Framework AnalYsis
o
o
.
o
o
Familiarisation
ldentifying a thematic framework
lndexing
Charting
Mapping and lnterpretation
The NIHR RDS for the East Midlands / Yorkshire & the Humber 2009 QUALITATIVE DATA ANALYSIS
Familiarisation: whole or partial transcription and reading of the data.
ldentifying a thematic framework: this is the initial coding framework which
is Oevétopéd both from a priori issues and from emerging issues from the
familiarisation stage. This thematic framework should be developed and
refined during subsequent stages.
lndexing: the process of applying the thematic framework to the data, using
numericàl or textual codes to identify specific pieces of data which correspond
to differing themes (this is more commonly called coding in other qualitative
analysis approaches -see Section 3).
Charting: using headings from the thematic framework to create charts of
your datà so that you can easily read across the whole dataset. Charts can be
éitner thematic foi each theme across all respondents (cases) or by case for
each respondent across all themes.
Examples:
Thematic Ghart
Case
1
Case 2
Case 3 etc
Theme 2
Theme 3
etc
Theme
Case Chart
Theme
1
Case
ln the chart boxes, you could put line and page references to relevant passages
in the transcripts. You might also want to include some text, eg key words or
shortened quotations as a reminder of what is being referred to. For example, the
case chart below is from a qualitative study of complementary therapy provision
via primary care, which used Framework Analysis (Luff and Thomas, 1999)' The
themes boxes contain both paraphrases of key issues as well as snippets of data,
to 'jog' the researcher's memory about the content of the themes. Alongside this
texi, íhere are page and line references to aid easy retrieve of the original data in
the transcripts:
The N|HR RDS for the East Midlands / Yorkshire & the Humber 2009 QUALITATIVE DATA ANALYSIS
Example Gase Chart
GP Number
1
Theme: Seryice
rationale
Theme:
Role of
complementary
theraoies
Theme:
Limits to
integration
Public wants them
(2:25)
For conditions
'orthodox medicine
'Conceptual
differences' (7:15)
fails'(5:12)
Providing choice
for patients (4:5)
Allowing patient to
take responsibility
(5:22)
Treating the whole
oerson (6:1)
.
Range of and
differences
between therapies
(7:31)
Money (7:17,8:8,
9:24\
Mapping and lnterpretation: this means searching for patterns,
associations, concepts, and explanations in your data, aided by visual
displays and plots. Ritchie and Spencer, (1994) suggest that at this stage, the
qualitative analyst might be aiming to define concepts, map the range and
nature of phenomena, create typologies, find associations within the data,
provide explanations or develop strategies. They emphasis that which of
these areas the analyst chooses to focus on will depend both on the themes
that have emerged from the data and the original research question. Ritchie
and Spencer acknowledge that 'this part of the analytical process is the most
difficult to describe'(í994:186). They offer several examples from their own
research, including diagrams they have used to visually present their ideas, to
try and demonstrate this process. Miles and Huberman (1994) also offer a
wide range of display ideas, which may be useful for exploring your data in
the context of Framework Analysis. The central aim of these techniques is to
enable you to visually display ideas from the data as an aid in developing and
testing interpretations.
For example, using the study of complementary therapy provision in primary care
see there were significant
mentioned above, we were interested
complementary practitioners in
and
GPs
among
differences/patterns in attitudes
professional group and
between
some areas (i.e. to look for associations
back up
attitudes). The need for, and type of, research required
possible
association
complementary therapy provision was one such area of
which emerged from analysis of the data collected. To aid us in exploring this
area, we vióually plotted our respondents 'positions', using their identification
number, on the current state of 'proof in two key areas of research (efficacy and
cost-effectiveness).
to
if
to
A shortened version of our working diagram is presented below:
The N|HR RDS for the East Midlands / Yorkshire & the Humber 2009 OUALITATIVE DATA ANALYSIS
Need more efficacy proof
l, 5,6,3,4
9,7
Proof
suÍficient
18, 2, 10, 8, 15, 1 I
GPs
8,2,
9
10
18, 15
7
, 12, l, 5, 6,3, 4
Need more cost-effectiveness
The diagram seemed to confirm our ideas, rruhich had developed from our charts,
memos and the raw data, that GPs were more likely to think that more research
was needed, particularly in the area of efficacy, than complementary
practitioners. lt also suggested this tendency, but within a more mixed picture, in
relation to cost-effectiveness research. This diagram was used by the
researchers as a starting point to explore other possible patterns or associations,
for example, were there further similarities among the clustered GPs or CPs like
age, Sex, type of general practice in which they worked, other attitudes etc.
Suggested further reading:
ichie, J and Spencer, L (1994), 'Qualitative data analysis for applied policy
', in Bryman and Burgess, eds., Analysing Qua/ifative Data, London:
Routledge, p173-194.
Miles, MB and Huberman, AM (1994) Qualitative Data Analysis; An expanded
sourcebook. London: Sage.
The NIHR RDS for the East Midlands / Yorkshire & the HumbeÍ 2009 QUALITATIVE OATA ANALYSIS
Consider the following research project. Which of the two approaches to analysis
might you take and whY?
1. The problem of non-attendance in an out-patient department
2. A study to explore the organisation and integration of an acupuncture service
within a piimary care team from the perspective of professional participants.
1. A grounded theory approach would be appropriate. The research proiect is
broad and exploratory in nature. There are few or no a priori issues or necessary
starÍing poinis suggósÍed (eg a specified sample) for the research beyond the
out-palient settin§.- Such a study could easity be concerned with generating
theory about non-attendance behaviour, which could then be ÍesÍed across a
rangó of settings concemed with DNA patterns. lt could provide hypothesis for
subsequent deductive approaches as wel/.
2. Framework anatysis seems most appropriate here. The question is specffic
and suggesfs a pre-de signed sample (professional participanÍs) as well as some
a priorí rssues (organisation and integration), which will need to be explicitly
addressed. The stidy may generate theories that could be tested elsewhere, but
the primary concern appears to be with description and interpretation of what is
happening in a specific setting.
The N|HR RDS for the East Midlands / Yorkshire & the Humber 2009 QUALITATIVE DATA ANALYSIS
17
you have chosen one of the analysis approaches for the research
projects
above. Now consider how you would go about each of the studies, given the
analysis approach you are using:
1. The problem of non-attendance in out-patients.
you could begin in several ways. You might for example want to talk first to
hospital doctors, nurses and receptionrbfs Ío get their perceptions of reasons
for'non-attendance. You might try and contact people who have not attended
directty
groups.
or instead ask the general pubtic for their views, for example in focus
Grounded theory analysis of your first stage of fieldwork data should then
determine where you decide to look next and the methods you employ.
As your research progresses, you will be constantly comparing the concepts.
and categoies emerging from each stage of the analysis until you feel
safrcfi'ed íhat you haie covered the question as fully as possrb/e, in other
words that you have reached theoretical saturation.
tn developing and testing your emerging theory about why people do not
attend sc;heduled appointments with hospital doctors, you will also want to
bring in existing research evidence in the area and consider the social
organisation of hospitals.
2. A study to explore the organisation and integration of an acupuncture service
within a primary care team from the perspective of professional participants.
.
You are tikety to have already identified the primary care team in question, but
if not you wóuld need to find a team which offers an acupuncture setvice who
are willing to particiPate.
.
you would then identify who were the relevant professional patticipants
(GP/s, acupuncturist/s, practice nurse/s etc) and think about how you might
access their percePtions.
o
lnterviews seem an obvious choice of method, but you might also consider
focus groups or observation instead of, or as well as, interviewing. You might
develóp a semi-structured interview schedule based on existing literature in
the area and the a priori organisation and integration issues.
.
You would then pilot the schedule with similar professiona/s if at allpossrb/e.
Thè NIHR RDS for the East Midlands / Yorkshire & the Humber 2009 QUALITATIVE DATA ANALYSIS
the constraints on your proiect (time etc) you would then aim to
Deperdlrg
", as the interviews progress or you may conduct all the data
do someànalysis
collection before undeftaking the analysis.
your initial familiarisation and coding sÍages woutd draw on the a prioriissues
as well as identifying emergent themes from the data. As you move through
Íhe sÍages of Frimiwork Alnatysis, you will devetop Íhese themes by charting
and màpping exercrses, as well as referring back to the original transcripts.
The output from your analysis might be a largety descriptive piece. o.n. the
sevice, but wilt piobably aim to offer interpretation of the key rcsge9 that have
emerqed so thai the reóearch can inform future developments of this kind.
4. Stages in Qualitative AnalYsis
Having looked at the general principles of qualitative analysis (Sectio.n 1) -and
then tilo specific approàcfres (Section 2), we now turn to the practicalities. What
practical piocesses are involved in actually carrying out qualitative data analysis?
How long does it all take?
The NIHR RDS for the East Midlands / Yorkshire & the Humber 2OO9 QUALITATTVE DATA
ANALYSIS
í9
The answer to the last question is probably 'how long have you got?' As
discussed in Section 2, a framework analysis approach is much more suited to a
limited time frame than a grounded theory approach. But there's no denying that
any qualitative analysis is a very time consuming and demanding process. ln this
seótion we will take you through some of the early stages in qualitative analysis,
following a general thematic analysis approach. We will also highlight points
where framework analysis and grounded theory would take a separate route.
4.1 Transcription
Almost all qualitative research studies involve some degree of transcription - the
data may be tape recorded interviews, focus groups, video recordings, or
handwritten field notes. lt is not appropriate, usually, to write up summary notes
unless the words are transcribed verbatim, the
from a tape recording
transcription by only including those sections that
the
researcher is likely to bias
seem relevant or interesting to them. Many researchers would also include some
non-verbal cues in the transcript - silence may communicate embarrassment or
emotional distress, or simply a pause for thought. Words such as 'well.... er. '. ... I
suppose...... are important elements of a conversation and should not be
ignored. Laughter or gestures may also give added meaning to the spoken word.
li someone eÉe is transcribing your material, it is important to tell them how much
of this non-verbal matter to include. lÍ you have never transcribed material, it is a
useful exercise to do a little yourself.
-
The N|HR RDS for the East Midlands / Yorkshire & the Humber 2OO9 QUALITATIVE DATA
ANALYSIS
20
perhaps a member of your family, or a friend or
colleague. As1 them to take part in an informal interview with you, lasting perhaps
ten minutes. You could choose to ask them about their definition of a healthy
person, then try and probe what it is for them that makes that person healthy.
Tape record the interview, then transcribe it into a word processing package in
your own time, including as much non-verbal material as you can.
Find a willing volunteer
.
-
How long did the transcription take you, compared with the original interview?
Unless you're a very skilled typist and have a particulaly clear tape, it is likely
that the transcribing process took at least four times as long as the interview. lt
may have taken evt:en longer. You wilt probably have found that your first attempt
at {ranscribing was amazingly time consuming. However you will be pretty familiar
with the data by the time you finished!
o
Highlight the non-verbal communication you were able to include. What does
it tell you, in addition to the words you have recorded?
Depending on the fluency or otheruise of your volunteer, you will probably have a
few 'er's ànd 'um's, but this adds to the realism and credibility of your data, as
well as giving clues as to how the respondent was feeling. Recording laughter,
movements ànd 'asides' a/so helps to make the data live when you re-read it
several days or weeks later.
.
Look at the questions you asked, and any comments you made. Have you led
the respondent in any way, or missed important clues that they gave you?
You may well find you have interrupted your respondent, or asked an
inappropiate question; transcription is a salutary experience in evaluating your
own interview tech niques.
.
Listen to the tape again, with the transcript in front of you. Have you changed
any of the words from the tape, perhaps just to make a sentence more
grammatically correct? Have you transcribed everything accurately?
Changing words or phrases to make them grammatically correct is not necessary,
and may inadverÍentty change fhe sense of what was said lf necessary,
cottoquiàlisms and odd usage of language can be explained in the accompanying
text when you come to write a rePort.
The NIHR RDS for the East Midlands / Yorkshire & the Humber 2009 QUALITATIVE DATA ANALYSIS
4.2 Organising your Data
After transcription, it is necessary to organise your data into easily retrievable
sections. You may wish to give each interview a number or code, or to break up
field notes into sections identified by date, or by context. lnterviewees will need to
be given pseudonyms or referred to by a code number. A secure file will be
neeóed that links pseudonyms and code numbers to the original informants, but
as with any reseaich this file is confidential and would usually be destroyed after
completion of the project. Similarly names and other identifiable material should
be removed from the transcripts. Narrative data needs to be numbered using line
or paragraph numbers, so that any unit of text you use can be traced back to its
original context. lf you are using a software package to do the analysis, this may
be done for you automatically by the computer, but you may need to decide upon
your unit of analysis - whether you wish each word, each line, each sentence or
each paragraph to be numbered. lf you are working with 'hard' copies (i.e. type or
hand writtén sheets) you would be well advised to make several copies of each at
this point, to avoid losing data when the analysis stages begin. Similarly word
processed files should be backed up and stored independently.
4.3 Familiarisation
The above procedures will have begun the process of familiarisation. By this we
mean the researcher listening to tapes and watching video material, reading and
re-reading the data, making memos and summaries before the formal analysis
begins. Ínis is an essential stage, and is particularly important if the main
researcher has not gathered allthe data themselves.
4.4 Coding
Before we start this section, it would be helpful if you read through the sample
interview transcripts at Appendix 1 and 2. They are excerpts from a series of
interviews carried out with people who had experienced a heart attack a year
previously, to explore perceptions of health and of their own recovery. lnterviews
took place in the respondents' home, and lasted about an hour'
After familiarisation with the material, we need to do some preliminary coding
(this would be called open coding in grounded theory). Lets assume we are
interested in ideas of how the respondents conceptualise the heart attack, its
cause and its impact on their lives. Certain ideas crop up in the transcript readily,
and we can give these a preliminary code. For example, in Section 1 of Derek's
interview, he talks about the 'shock' of having a heart attack.
The NIHR RDS for the East Midlands / Yorkshire & the Humber 2009 QUALITATIVE OATA ANALYSIS
H,ghlight those lines in Derek's interview which refer to the shock of having a
heart attack.
you should have found two instances where Derek used this word (lines 14 and
16), and two from the interuiewer (line 19). However there are other phrases that
Dérek uses fhaÍ coutd also be interpreted as refening to the shock of the
diagnosis. For example, you may want to highlight 'l were one of them people
thal never thought it could happen' (line 14) as also expressing s
you now have your first provisional code, called 'shock'. lt is likely to be modified
later, but servés to begin the process of categorising and analysing the data.
Codes can be identified by 'in vivo' terms that the respondents themselves use.
For example 'just plod on; (lines 28 and 159) typifies Derek's description of his
attempts to live with his disability. Or codes may be named by the researcher, to
include a variety of ways that respondents express an underlying concept. An
example of this from Derek's interview might be 'confusion', a code that would
include Derek's comment 'ljust can't work it out sometimes' (line 75).
lf you are using framework analysis, you may at this stage have a priori concepts
you wish to usè as codes. You may want to know, for instance, about how people
who have had a heart attack conceptualise the causes of the attack. From
existing literature, you may know that these can be divided into physical causes,
psycho-logical cauóes, ideas of luck, genetic inheritance and so on. You could
ihén search the data for material that could be coded under these headings.
.
Using highlighter pens of different colours, or codes in the margin, read
through the rest of Derek's transcript, developing codes from the data.
.
Do the same with Sue's interview, noting which ideas are similar to those in
Derek's interview, and which you have to develop new codes for.
you witt have found some codes that can be applied to ideas in both interviews.
'Fear' crops up in both, particutarty in the context of having another heart attack.
Both interviewees talk about not taking enough exercise, and having to cut down
on work activity. Both discuss /ifesfyle changes they have made since the heart
attack. But you will have found some ideas in each that are not mirrored in the
other. Derek tatks abouf chesÍ pain, and taking medication for it. Sue doesnï
mention pain at all, but does mention medication.
You may well have noticed a difference in the way the two respondents talk about
their médication Sue sees it as a way of actively controlling her disease and
promoting health, Derek sees his angina Spray as a last resort, to be used if he
can't maÀage any other way. ln looking at these differences, you are beginning to
refine youróodes - after using 'medication' as a preliminary code, we could now
begin io analyse this concepts further. Maybe we decide to place the code
-
The NIHR RDS for the East Midlands / Yorkshire & the Humber 2009 QUALITATIVE DATA
ANALYSIS
23
,medication' into a category called 'managing the symptoms" 'taking control of the
disease', or'coping with chest pain'.
However you may then notice that other data also falls into these new revised
categorieó. 'Managing the symptoms' could also includes Derek's comment'as
roon-as the painlomes on tstop....l don't mess about'(line 62) S9:9me
degree of re-coding now needs to be done, and many units of data will fall into
more than one categorY.
.
f-oof through the data again, with the idea of forming these broader
categories.
o
Write down, on a Separate piece of paper, four or five broader ideas and,
undemeath each, a list of codes that could contribute to each.
One such category could be'coping with returning to work'. Codes that contribute
to that may be:
o
o
.
o
Taking more time
'l'll do them tomorrow'
Struggling with PhYsicalwork
Cutting down on stres{heavY wPlk
4.5 Themes
You have now begun to identify themes or emergent concepts, and will engage in
re-coding to develop more well defined categories.
One of these categories could be 'physical causes of the heart attack'. Both
respondents discusó this at various points, but this category could also have been
identified a priorifrom the literature in framework analysis'
y"* list of codes, and identify those that would inform this category
,perceivàd pfrysical causes'. Then look back through the interviews and see if
of
theie are any other references that you have missed.
L""k th*.gh
Derek expresses bewilderment that he should have had a heart attack at all,
given hii physical fitness and activity (ine9 13-16)..H..e wonders whether a
óhange to'móre mechanisation at work may have precipitated that attack, as he
at
was lualking /ess. Sue expresses her betieí that she is now fitter that she was
any
taking
not
was
she
that
the time o{her heart attack (tines 82-83), and sÍaÍes
exercise in the years before she had her hearÍ attack. She a/so sfresses the large
in her ir
workload she was
ANALYSIS
The N|HR ROS for the East Midlands / Yorkshire & the Humber 2009 QUALITATIVE DATA
With a framework approach, it would be likely that some of the themes emerging
from the data would also be the identified issues with which you began your
research. Your data would conflrm their importance, in this case, and enable you
to explore them further.
lf, on the other hand, you are taking a grounded theory approach, you would try
to ensure that all the emergent themes were generated from the data itself,
although you might later incorporate other theoretical ideas in your analysis. A
ground]ed-theoryapproach would also be prepared to test ideas generated in
early data analysis in further data collection.
How Oo the two respondents conceptualise 'health'? What theories about lay
perceptions of health could you test further in later data collection?
lines 92-95 Sue expresses something about being healthy despite having a
diagnosis of heaft drsease. She a/so defines health as having a balance between
woík and rest (tines 102-103). Derek defines health in functional terms, being
able to climb hiils without symptoms (line 111). Maybe the differences are related
to gender are women more tikely to define health in complex, holistic terms,
whéreas men relate it more to physicat functioning? You could test out this idea in
future interviews with both
ln
-
Obviously we cannot go very far with the process of analysis with the sample
transcripis provided here. Normally you would have much more data to analyse,
and the processes described above would generate many themes
and
categories, which would need refining and developing. Using grounded theory,
early data would be subjected to preliminary analysis, then emerging theory
testód in subsequent data collection. Collection of data would continue, ideally,
until 'saturation' is reached - i.e. no new themes are emerging, and theoretical
ideas have been tested satisfactorily. Using framework analysis, indexing would
be followed by mapping and charting. ln practice, research studies are often
bound by constraints of time and resources, and analysis has to be brought to a
close wÉen specific questions have been answered. Framework analysis is an
approach moie suited to research asking specific questions and with limited
timescales than grounded theory.
The N|HR RDS for the East Midlands / Yorkshire & the Humber 2009 QUALITATIVE DATA ANALYSIS
5. Ensuring Rigour
Reliability and validity are important issues in all research including qualitative
research. Demonstràting that your qualitative data analysis is rigorous is
especially important given a common criticism (from those less favourable to
qualitativL research) ihat qualitative results are anecdotal. lncreasingly, journal
editors and funders are using 'checklists' of criteria or questions for assessing the
reliability and validity of qualitative research sent to them (see for example, the
British Medical Journal guidelines for qualitative research on
http://www.bmi.oro.uk4. On many levels then,
issues in your analYsis.
it is important to address
these
ln this section, we will discuss the issues of reliability and validity in the context of
qualitative data analysis.
lf you are familiar with quantitative research, you will
see that these terms are interpreted somewhat differently and have different
implications in qualitative research. We will then look at other ways of
demonstrating the robustness of qualitative analysis, namely triangulation and
respondent validation.
5.í Reliability
ln terms of assessing qualitative research the emphasis is on the reliability of the
methods employed, Vou need to demonstrate to the reader that the methods you
have used are reproducible and consistent. However, unlike in quantitative
research, external replication may not be the most appropriate measure. lnstead,
in demonstrating the reliability of your analysis you would need to consider the
following:
.
.
.
Describing the approach to and procedures for data analysis
Justifying why these are appropriate within the context of your study
Clearly documenting the process of generating themes, concepts or theories
from the data audit trail
.
Referring to external evidence, including previous qualitative and quantitative
studies, to test the conclusions from your analysis as appropriate
The N|HR RDS for the East Midlands / Yorkshire & the Humber 2009 OUALITATIVE DATA ANALYSIS
5.2 Validity
Here the emphasis is on the validity of the interpretation. The ability of the
findings to represent the 'truth' may not be appropriate if we accept the existence
and importance of multiple 'truths'. Rather, validity will be judged by the extent to
which an account seems to fairly and accurately represent the data collected. ln
terms of presenting the analysis then, reflection is required on:
o
The impact of your research design and approach to analysis on the results
you present.
.
The consistency of your findings, for example has analysis been undertaken
by more than one researcher (often referred to as inter-rater reliability).
o
The extent to which you represented all relevant views, for example checking
for'negative' or deviant cases to test your interpretations.
.
Adequate and systematic use of the original data (for example using
quotations, and not all from the same person!) in the presentation of your
analysis so that readers are convinced that your interpretations relate to the
data gathered.
5.3 Other ways of demonstrating reliability and validity
Triangulation
Evidence that the qualitative researcher has undertaken 'triangulation' is
frequently seen as demonstrating rigour. Triangulation means gathering and
analysing data from more than one source to gain a fuller perspective on the
situation you are investigating. This may be more or less important, or possible,
depending on your research question and setting. For example, consider the
hypothetical study of an acupuncture service in primary care, discussed in
Section 3. ln this study, triangulation might mean that you conduct observation of
the operation of the service and a review of service records in addition to semistructured interviews, aS a way of gaining different insights into the Same
situation. However, triangulated data should not be simply used to 'check' the
conclusions from one data source against another. Often the data from one
source will contradict or question the findings from another. This is not
necessarily a failure of the research in itself, as 'real' life research situations are
inevitably complex. lndeed a key strength of triangulation is the possibility of
uncovering this complexity and of finding different views. The contradictions and
differences within the data collected should spur the researcher on to further
analysis, and sometimes, to further investigation until some 'sense' can be made
of what is happening. Evidence that the analyst has used triangulation in this way
and has effectively drawn the analysis of different forms of data together
demonstrates rigour, rather than simply the use of different sources.
The NIHR RDS for the East Midlands / Yorkshire & the Humber 2009 QUALITATIVE DATA ANALYSIS
Researcher perspective
Feminist researchers such as Reinhaz (1992) have highlighted the necessity to
make clear the perspectives of the researcher themselves in demonstrating the
validity (they would call it credibility) of the research. Because qualitative analysis
is an interpietative process, the preconceptions, assumptions and 'worldview' of
the researcher are likely to influence the process and any emerging theory,
despite use of rigorous approaches such as we have described. This needs
acknowledging, by making the researcher more 'visible' in the analysis process.
Other writels óall this reflexivity', A reflexive account is an honest attempt by the
researcher to declare their conceptualjourney through the research, perhaps by
including sections of their own reflective diaries as they undertook the research.
Respondent Validation
Qualitative researchers frequently feed back the findings from their research to
their participants in some way. The range of feedback to respondents varies. ln
some cases transcripts or quotations may be sent back simply to check accuracy
or consent for use, in other cases respondents may be asked to comment on the
interpretation or drafts of the report. Feedback to respondents has been seen as
important in involving participants in the research process and, for some critical
social scientists, in addressing concerns about the researcher having sole power
of interpretation. Many funders and reviewers consider 'respondent validation' of
qualitative research to be a mark of quality, and evidence of respondent
validation' of findings is increasingly seen as a way of demonstrating rigour.
However, the decision to involve respondents in feedback or validation may
legitimately vary from study to study. lt might be more helpful in terms of
assessing rigour to concentrate on evidence that the researchers have:
.
.
.
Considered the issue of feedback to respondents
Provided reasons for their decision to provide feedback or not
Explained how they have gone about any feedback, the type of feedback
provided, and why
o
Explained how any feedback from respondents has been used in the analysis
and interpretation.
Respondent feedback generates important issues for the analyst to consider, not
least what to do about competing interpretations. lt is important to consider that,
for reasons of confidentiality, individual respondents may not have access to the
full range of views that the researcher has found; that respondents may have
competíng perspectives (for example, doctors and receptionists may view the
same situation differently); and that they may have particular personal,
professional or political reasons for disliking the researchers interpretations,
however legitimate these interpretations may be. lt is important to consider how
far it is the researchers' job to question 'taken for granted' assumptions or
particular views, even when this may be unpopular with Some respondents.
The N|HR RDS for the East Midlands / Yorkshire & the Humber 2009 QUALITATIVE DATA ANALYSIS
Write down brief notes in response to the following questions:
Think of a qualitative project you would like to undertake:
1. What would be the advantages of feedback to respondents in your proposed
research?
Ihese witlvary with the proiect, but common advantages include:
. lt can be used to check the accuracy of both the recording of data (by giving
respondents transcripts or quotations) and the analytical interpretations.
.
Some qualitative researchers are concemed about the power of the
researcher to interpret people's words or actions. Feedback to respondenfs is
seen as one way of redressing this power imbalance in the research process.
.
External reviewers increasingty see rï as one way of demonstrating the validity
of your research.
2. What issues would you need to consider when planning respondent feedback?
Again this list is not exhaustive, but includes common tssues'
Respon dent feedback witt generate new 'data'. lt is important to consider how,
and at what stage of the research, feedback will be incorporated into the
analysis or final report.
.
.
dent disinteresÍ -respo ndents may have differing /evels of interest in
and commitment to the research and you may end up only getting feedback
from those who have a pafticular perspective. /f is important Ío discuss
feedback with respondents during the initial sfages of the research, so that
they have a clear idea of what is expected of them, and to gain their consent
to iake part. /Í rs a/so worth considering the most appropriate and engaging
way of feeding back resu/Ís to respondents -would verbal or written formats or
seminars/conierences be most hetpful? A final question ,s how far
researchers shoutd go in 'chasing' those respondents who do not take part in
feedback. To what extent is it acceptable to pursue people for feedback that
Respon
they may not want to give?
.
Seerng or hearing their words or actions fed back to them may cause
respondents to reconsider what they have said or done and what they are
happy Ío see reported. They may also not agree with the way you have
interpreted what they or other people have said or done. Before beginning the
research it is important to think about the kind of feedback that you want from
The N|HR RDS for the East Midlands / Yorkshire & the Humber 2009 QUALITATIVE DATA ANALYSIS
respondents and how far you are wilting to change your analysis, beyond
facÍuat inaccuracies, in light of this feedback. Ihis a/so needs to be clearly
explained and agreed with respondents.
.
Finally, when using verbatim quotes, it is quite common for respondents to
feetthat they appear inarticulate! tt may be helpful to tell patÍicipants how you
wil be presenting their words in advance so as to pre-empt negotiations about
íng up'quotations.
The NIHR RDS for the East Midlands / Yorkshire & the Humber 2009 QUALITATIVE DATA ANALYSIS
6. Practicalities
You will by now be aware, if you weren't before you started this pack, that
qualitative analysis is a complex and time consuming business. lt is no easy
option for those who don't fancy getting to grips with statistics. Neither is it a
journalistic enterprise for seeking overall impressions from the data - the last
section on rigour demonstrated the importance of ensuring adherence to a
systematic and scientific method.
So how is such a process to be managed, without sinking beneath a pile of
transcripts and field notes?
o
Be organised. ldentify each set of data as you generate it. lnterview
transcripts, field notes, photographs, videos, documents, and any other items
of source material need to be given an identification code that makes them
readily retrievable. All data should include a date, some indication of context,
and an anonymised identifier that will enable the researcher to identify the
source. A complete list of data sources can then be compiled and used for
reference throughout the analysis.
.
The business of coding is managed in a variety of ways by different
researchers. lt is possible to buy computer software packages that will
manage this process electronically, and there will be a short discussion of
these in section 6. Assuming you are doing this process manually, or by using
an ordinary word processor, there are two main systems used by most
qualitative researchers:
transcripts (having previously made copies!) into bite-size pieces for
analysis. So you may have phrases, sentences or paragraphs as your
main unit of analysis, and these can be arranged physically in groups
or electronically in computer files according to initial coding. Some
people do this by pasting the text units onto index cards, which can
then be sorted and re-sorted easily. Others do it by pasting text units
onto large sheets of paper representing codes or larger categories and
themes. Obviously each text unit needs to be traceable back to its
context, hence the importance of organisation as described above.
Also, many items of data will carry two or more codes; in this case
several copies will need to be made so that the item can be contained
in several codes or categories. A cross-referencing system may also
need to be develoPed.
process, using separate colours for each code or category. Obviously
there is a finite number of colours available for such highlighting, and
this method would not be suitable for complex analyses where many
hundreds of codes were being used. Problems may also be
encountered where a piece of text needs to be coded under two or
more colours - confusion may result! However the advantage of this
method is that the text does not need to be cut up, and so text units
The NIHR RDS for the East Midlands / Yorkshire & the Humber 2009 QUALITATIVE DATA ANALYSIS
remain
in context. For relatively
straightforward
and
pragmatic
analyses, it may be the preferred method.
A combination of the above methods can also be used, perhaps cutting
and
pasting into initial categories using a word processing package, then printing out
each category and colour coding to look for over arching or recurring themes.
o
Keep a record of your thoughts and theories as you go along, often called a
journal or memo keeping. Define your inclusion criteria for particular codes,
record why you decided to re-code a particular set of data, write down
emerging theories and questions that can be tested later. This record willform
the basis of your narrative analysis for your final report, and will enable you to
follow your thought processes when you feel you have lost your way.
examples oÍ verbatim daÍa that support your argument. This will save
you searching for them later, and ensures you are grounding your
analysis in real evidence.
means to support their analysis. lf you sketch out any of these as you
go along, keep it in your journal, no matter how scrappy it is. lt may
well be useful in the eventual analysis. An example of a flow chart is
included at the end of this section (see Figure 1).
Finally, qualitative analysis is a lengthy process. You need to allow perhaps a
third of the total time taken for your project for the analysis. lt may take longer if
you are using grounded theory, where analysis and data collection take place
concurrently. lt is also a resource intensive task, especially if you are using two or
more researchers to do the analysis to improve rigour.
Reports of qualitative analysis tend to be wordy, and it is often difficult to pare the
report down to the required word limit for a journal article. Journals which accept
a lot of qualitative material have more generous word limits (usually up to 5000
words), but others (such as the British Medical Journal) may require you to
reduce your article to 2000 words. This may feel like treachery, forcing you to
omit much that gives richness and credibility to your analysis.
The NIHR RDS for the East Midlands / Yorkshire & the Humber 2009 QUALITATIVE DATA ANALYSIS
Figure í
Pat's Progress
Pre+xbting
stroke
(!{ever regained her
c- ,i,\
,L?-Z
o
àtéo
i*x
toàa%)
-<%
Admitted to
long-term
care
.."§i
The NIHR RDS Íor the East Midlands / Yorkshire & the Humber 2009 QUALITATIVE DATA ANALYSIS
7. Gomputer software Packages for
Qualitative Analysis
Since the mid 1990s, various packages have been developed to aid the process
of management of qualitative data during the analysis process. The earliest of
these, Ethnograpfr, was developed before the widespread adoption of the
Windows operating system, and was somewhat cumbersome to use because of
the need to prepare data specifically for entry into the software. However, like all
later packages, the system allowed the researcher to organise, code and search
data using computer technology that saved a great deal of the 'paper chase'
described in Section 5.
There are many popular packages now in use by health services researchers
such as AtlasTi, NUD*IST and NVivo, and these will be discussed briefly below.
However all computer software packages have basic similarities.
7.1 What will a computer software analysis package do
for me?
Data storage and management
Software packages will allow you to enter your raw data directly into the package,
and will then hold your documents securely, much as a word processing package
does. lt is possible in most of the packages to enter rich, formatted text directly;
others will require a standardised format such as plain text. The newest packages
will also handle visual material such as photographs, diagrams, video and links to
WebPages. Most packages will also let you annotate and edit the material once it
is entered, although the ease with which this can be done is variable. The
package will generally have some form of automatic indexing of material, and will
allow you to add your own identification information such as date and context.
You can store huge quantities of data on the packages, provided your computer
has sufficient capacity, but beware of being tempted to gather more data than you
can analyse!
Data searching and retrieval
All packages can search textual data for particular words or phrases. lf you wish
to count frequency of certain words for content analysis, this can be done easily.
Boolean operators such as AND, OR, NOT, NEAR can be used to refine
searches and test out emerging theory. The packages will retrieve data with
appropriate context - the word you are searching for needs to be seen in the
context of the sentence or paragraph from which it comes. Retrieval also includes
identification of the data - you will know which interview or field note the data
came from, and whereabouts in the text it was sited.
The NIHR RDS Íor the East Midlands / Yorkshire & the Humber 2009 QUALITATIVE OATA
ANALYSIS
34
Coding
The proóess of coding and re-coding is made a great deal simpler by using a
computer package. Small sections of data can be highlighted and assigned to a
pre-existing or new code in a matter of seconds. ltems that have been coded are
stored and can be searched in the same way as documents. Codes can be given
titles and descriptions chosen by the researcher. They can also be combined with
other codes, subdivided, or built into conceptual models to develop theory.
Developing and testing theory
Packages vary in the extent to which they allow theoretical modelling, but all will
enable relationships between coded data to be explored and displayed. One of
the most popular systems, NVivo, uses a hierarchical system which takes a 'top
down' approach, dividing and subdividing major concepts into their constituent
elements. This may be ideal for a policy orientated approach such as framework
analysis but may be less good for grounded theory. AtlasTi, for example, is less
prescriptive, and allows diagrammatic representation of relationships between
concepts. This fosters a more inductive approach, allowing theory to be built
'upwards'from the data itself.
Writing reports
The software packages will produce reports as requested by the researcher
printing out the entire dataset under one code, for example, or reproducing a
section of a document. The results of a search as described above can be
reported upon. This makes it quite easy to incorporate verbatim quotes or visual
material into an analytical account or article. lt is also possible to write a 'journal'
within the package, enabling the researcher to record memos and ideas
throughout the analysis process. This journal can similarly be printed out as a
textual document.
7.ZWhat will a software package NOT do for me?
ln common with most software, analysis packages are a tool that can aid the
researcher, but they cannot replace the human element! A package cannot 'do'
the analysis, because it lacks the capacity to think, reflect and analyse.
Computer-aided analysis can be deceptively easy - coding and searching, for
instance, is quick and satisfying, but it is then possible to keep the analysis at a
superficial level, without the deep engagement with the data that is a hallmark of
good qualitative research. At the end of the day, there are no short cuts to the
àemanding process of reading and re-reading the data, sorting, categorising and
analysing the data and building and testing theories.
ln making a decision whether or not to use a software package, it is worth asking
a few questions:
The NIHR RDS for the East Midlands / Yorkshire & the Humber 2009 QUALITATIVE DATA ANALYSIS
.
Do I have the resources to buy the package? (Most cost between 8200-€500,
but site licences may be available at a lower individual rate for universities
and research units)
o
.
Do I have the time and inclination to learn how to use the package?
Will I be able to use the software again? lf so, the investment of time and
money may be more cost effective
o
Do I have a lot of data to manage? (eg more than 6-10 hours of interview or
equivalent). lf not, it may be as easy to use manual methods
.
o
Do I like working 'on screen' or do I prefer paper-based methods?
Are others involved in the research able and willing to use the software
package?
It is a good idea to visit the websites of the many packages available, and
download the free demonstration versions available. This will give you an
opportunity to go through tutorials and get a feel for the package before you
decide whether to buy. Some web addresses are given below.
The Computer Assisted Qualitative Data Analysis networking project (CAODAS)
at University of Surrey provides an excellent website where you can download
very helpful information about the many packages now available. The project
also runs training sessions on the most popular packages, and guidance on
which package to choose.
htto://caqdas.soc. su rrev. ac. uk
The latest versions of software packages available in 2006, and websites, are
listed below
ATLAS.ti (Version 5)
http i/wtirrw. atlasti. de
:
HypeTRESEARCH
http ://unnrw. resea rchwa re. co m
MAXqda2
vtnivw.maxqda.de
NVivoT (supersedes NVivo and N6/NUD.IST)
http //www. qsrintern atio n a l. com
:
QDA Miner
http ://www. provalisresea rch. com
QUALRUS
The NIHR RDS for the East Midlands / Yorkshire & the Humber 2009 QUALITATIVE DATA ANALYSIS
htto ://rnnryw. o u al rus. co m
TRANSANA2 (specialises in audio and video data)
rrurvw.transana.oro
8. Summary
We hope that this pack has introduced you to the basics of qualitative data
analysis. Although demanding and time consuming, we hope you will discover,
through analysing your data, that qualitative research is a rewarding and
satisfying experience.
It remains true that it is a process best learned by doing, and, where possible, by
working alongside others more experienced in the art. lf you are truly trailblazing
and working alone with qualitative data, we suggest you forge some links with a
local university department where there are staff who are experienced in
qualitative analysis. Staff at your local RDSU should be able to point you in the
right direction.
This pack should have enabled you to:
.
.
.
o
o
.
Understand
the differences between various approaches to
qualitative
analysis, particularly grounded theory and framework analysis.
Decide on the level of analysis necessary for your project.
Work through the early stages of analysing qualitative data by transcribing,
coding and developing themes and categories.
Become aware of processes that can be used to improve and demonstrate
rigour in qualitative analysis.
Understand the practicalities and resource demands of qualitative data
analysis.
Become familiar with the functions of qualitative data analysis software, and
know where to access further information and demonstration copies of three
packages.
The NIHR RDS for the East Midlands / Yorkshire & the Humber 2009 QUALITATIVE DATA ANALYSIS
References
Bowling, A 2nd ed (2002) Research Methods in Health: lnvestigating health and
health services Open University Press, Buckingham.
Glaser, B and Strauss,
Weidenfleld & Nicolson.
Holloway
I
A (1967) The Discovery of Grounded Theory.
(2005) Qualitative methods
London:
in health care Maidenhead,
Open
University Press.
Lewins
A and Silver C
(2007) lJsing software for qualitative data analysis
London: Sage Publications.
Miles, MB and Huberman, AM (1994) Qualitative Data Analysis: An expanded
sourcebook London: Sage.
Murphy, E; Dingwall, R, Greatbatch, D; Parker; S, Watson, P (1998) Qualitative
research methods in health technology assessment: a review of the literature
Health Tech nology Assessme nt 2, 16.
Pope, C; Mays, N. eds. 3'd edition (2006) Quatitative Research in Health Care.
BMJ Publishing Group.
Pope, C; Ziebland, S; Mays, N (2000) Analysing qualitative data British Medical
Journal 320,114-116.
Richie, J and Spencer, L (1994), 'Qualitative data analysis for applied policy
research', in Bryman and BurgeSS, eds., Analysing Qualitative DaÍa, London:
Routledge, p173-194.
Rheinharz
S
(1992) Feminist methods
in
social research Oxford: Oxford
University Press.
Strauss,
A and Corbin, J
(1990) Basics of Qualitative Research: Grounded
Theory Procedures and Techniques. London: Sage.
Strauss, A and Corbin, J (1998) Grounded Theory Methodology: An overview in
Denzin and Lincoln, eds, SÍraÍegies of Qualitative lnquiry, London: Sage, p. 581
83.
Websites
http :/icaodas. soc. su rrev. ac.
u
k
http://onlineqda. hud. ac. uUlntroduction/index. php
The NIHR RDS for the East Midlands / Yorkshire & the Humber 2009 QUALITATIVE DATA
ANALYSIS
38
Glossary of Terms
Coding
This is the process of deciding how to conceptually divide
up raw qualitative data. Sections of text transcripts, for
example, may be marked by the researcher in various
ways (underlining in a coloured pen, given a numerical
reference, or bracketed with a textual code in the margin).
These sections contain data which the researcher is
interested in exploring and analysing further. ln the early
stages of analysis, most if not all sections of the text will be
marked and given different 'codes' depending on their
content. As the analysis progresses these codes will be
refined or combined to form themes or categories of
issues. (NB: ln Framework Analysis this process is referred
to as'indexing').
Familiarisation
The process of becoming increasingly familiar with the
data you have collected. This initially takes place in several
ways, for example through undertaking transcription,
reading the transcripts, field notes or observation notes
gathered, producing summaries
of
interviews,
or
re-
listening to tape recordings.
Framework
Analysis
An approach to qualitative analysis which was developed
in the context of applied policy research. lt provides
systematic analysis stages which are clearly defined and
easily accessible to others (eg funders). lt is particularly
well suited to qualitative research where there are pre-set
questions that need to be addressed (a prioriissues) and
where the timescale is short.
influential approach to qualitative research and
analysis which has been widely adapted and used in the
health and social sciences. ln a 'classic' grounded theory
study, data is simultaneously collected, coded and
conceptualised throughout the project. Grounded Theory
offers a rigorous approach to generating theory from
qualitative data. lt is particularly well suited to exploratory
studies where little is known and to research that is
explicitly concerned with theory generation.
Grounded Theory
A highly
lnductive
Research which is concerned with uncovering meanings,
explanations or hypotheses from within the data gathered,
rather than testing pre-existing hypotheses (deductive
research).
The NIHR RDS for the East Midlands / Yorkshire & the Humber 2009 QUALITATIVE DATA
ANALYSIS
39
lnter-rater
reliability
Used to refer to the process of checking the analysis and
interpretation of qualitative data by having more than one
researcher undertake some or all of the analysis stages.
The aim is to provide a check on the consistency and
transparency of the analysis.
Themes
A theme is generated when similar issues and ideas
expressed by participants within qualitative data are
brought together by the researcher into a single category or
cluster. This 'theme' may be labelled by a word or
expression taken directly from the data or by one created
by the researcher because it seems to best characterise
the essence of what is being said.
Appendix
1
- Sample Transcript (Derek)
The NIHR RDS for the East Midlands / Yorkshire & the Humber 2009 QUALITATIVE DATA
ANALYSIS
40
lnterview w1h Derek, aged 48, gardener, working for the local council. Wife (W) was present for
the latter part of the interview and contributed. Derek had a heart attack one year previously.
SECTION
í
A: I just want to get you talking about what the heart attacks meant to you and if it's changed your
life.-But just to get you started, tell me a bit about how it happened.
DEREK: I've no idea whatsoever.
A: You don't remember it?
DEREK: Yes, Oh, yes. As though it were yesterday, but, I mean I were one of those people who
nobody ever thought.....it was such a shock to everybody. I were one of them people who never
thought it could nàppen, I'm a six-footer, twelve and a half stone. I were an active...at that time I
were in an active job, all day every day. lt were a shock.
A: A shock? Why was it such a shock?
DEREK: Why, all my family, they've had it, it's like hereditary type thing and thafs always- been'
But with beiÀg on tfris iob-l aWàys thought, well, I'm alright, you know, being fit, never fat l've
always....l usàd to be iÀ a steel firm and all that...l was a bit heavier then....... all my family died
throrigh it. So when it did happen...well, this is it, but when you think about it, l'm the only one
what's had one but still here type thing.
A: Did it make you feel frightened, or wonied, or....?
DEREK: At first, but then, l'd say no I just, you know....plod on, you know, carry on. I just don't
know why it happened really. You know, I mean I'm walking ten mile a day, everything was
physical on our job.
A: Yes, so you're very fit....
DEREK: So, just no idea. Apart from....
A: Go on
DEREK: I said, last six months, I've been on mowing machines, cutting....l've been on one of them
for 15 years, then all of a sudden everything got automated like and we've gone on tractors, given
vans tó drive around in. Thats how it is for last six months, but I makes me wonder if that's.'...you
know lwasn't....
A: What, that happened six months before your heart attack, did it?
DEREK: Yes, yes, you know, I weren't walking around as much or, ......still going out every day
and having to dig, rake .........
SECTION 2
A: So a year later are there any changes in your lifestyle, or the way you think about your health?
DEREK: l'm a bit weary. Yes, no...l still get odd pains and it upsets me when I'm walking up hill, I
get a pain. I don't take one of these sprays I carry on thinking it'll go away but it doesn't. so l finish
Ip naving to stop, take it, stand there for a bit and I feel daft just stood, you know, while the pain
goes off. Then I'm all right, like.
A: What about at work? Do you do what you used to do?
DEREK: As much as, yes. But its just that I do it....as soon as the pain comes on I stop, you know,
or I say, right, that's enough.....At one time I would have carried on and wherever pain were you'd
carry on and hope... ....like, but, now I stop now and I don't mess about.
A: So you are back to quite a lot of physical......
The NIHR RDS Íor the East Midlands / Yorkshire & the Humber 2009 QUALITATIVE DATA
ANALYSIS
41
DEREK: Well it still hurts, you know. There's no way I can go like I used to...like I used to go
out...l,ll dig this hote, l'll cnof tnis tree down, l'll cut all this privet. You do it, but it takes a bit longer.
A: And you can go walking like you used to?
DEREK: No, walking doesn't worry me no. As I say, I have to take spray, pain'll come on, take the
spray. lt's just differént things you think to yourself, why is it come on.now, like, why didn't it come
on yàsterO'ay when I walkeó about five mile? You know, you think things like that, I do anyway' I
jusican't work it out why I'll walk to town, about two mile and no trouble, yet another day I'll get to
iop of this hill, must be about 100 yards and l'll have to stop, ljust can't work it out sometimes.
A: Have you cut down on things outside work?
DEREK: Some nights 1....1' mean l'm working all day, come home have a cup of tea, go and have a
shower, come back down , have my tea and I don't feel like moving, you know. Whereas at one
time I would have gone out in gardén, cut grass, now its....l have to think about it for a couple of
days. I think its just idleness, l'm not sure.
SECTION 3
A: Do you feel, kind of back to normal health wise, compared with a year ago?
DEREK: Not really, no. Do you mean before the heart attack?
A: Before the heart attack
DEREK: No.
A: ln what ways are you not back to normal?
DEREK: lt's only these pains that's put me off, you know. They seem to be more regular than what
they were before.
When I first come out of hospital, to be honest I didn't feel much, any different really, you know it
were in here that it could happen again, like,......but they found it, I'm lucky they're treating me like,
so I were alright weren't l?
Then l, they got a bit more regular these pains, cause I were going .... wi' me dad, I mean, (father
died of a nèàrt attack) they gave me two or three of these sprays and I never used them. I don't
think I used one for months. Then all of a sudden l'm using them three or four times a day. I don't
take it every time I get pain I take it when l'm at work usually, when I'm struggling. Then it gives me
a headache and....
A: Can you say what being healthy means, for you?
DEREK: Being able to walk up hills without getting breathless or pain.
W: He used to run uP hills before
DEREK: Yes. Now l'm having to say to her, slow down.
W: He didn't though because a few months before he had this heart attack he started slowing
down. I noticed a difference in him. He used to come up this hill, he'd be really, you know
everybody's hot when you're coming up, but he'd be really fighting for breath and he were like that
for quite à while before he had heart attack and then he put a fence up in the garden and that, he'd
done it without....in fact, I think it were that that brought it on. Brought it to a head.
A: Yes, so you were having some symptoms before?
DEREK: No.
W: No. When you were putting that fence up you were forever...this....
DEREK: lndigestion
The NIHR RDS for the East Midlands / Yorkshire & the Humber 2009 QUALITATIVE DATA
ANALYSIS
42
W: yes. He were putting it down to indigestion. You know, when you've got a feeling, because I
wanted to pay somebody to do it and - Oh, no I do it all day long at work...you know what I mean.
But I think this had been coming on for a while.
DEREK: Thats what l've just said, that it seems to, when l've been driving I've gone from walking
every day all day, to drivíng, sat down seven hours, weren't l. lt must be....after being, doing that
and ihen-a few weeks working....alright, you're still digging and that, but you're not actually -getting
breathless. As l've been walking, now l'm driving a tractor or driving a lorry, six months of doing
that and then walking home, it used to take it out of you. lts like with football isn't it, if you've not
played football for a few weeks you can't expect to go out and play football for an hour.
W: We was away a fortnight ago at Newquay. We just went for a week's break and we walked,
must have walked five miles one day on top of cliff and wind were howling and that takes your
breath and he was right as rain.
DEREK: Another day I was struggling on the flat wasn't l?
W: Yes, we'd only gone a few yards and we had to stop.
DEREK: lts that what annoys, it does get to me does that, I've just no idea why. But I suppose you
get used to it.
A: Do you see yourself getting better....?
I hope so. I don't know what you call better. I mean I don't suppose anybody gets....fully
:.-=*=^,
A: You expect to be able to do things...
DEREK: lf I can just plod on like I'm doing, you know, if I can...l'd be happy enough.
tf I could keep like this for twenty years l'll be really happy. As I say, its just odd bits, odd times you
get a bit down and depressed sometimes.
W: And soon get tired don't You?
DEREK: Well, sometimes you forget, I'm saying that I'm out of breath, but sometimes you're
walking as fast as what I used to be, then you think to yourself I suppose I ought to slow down.
W: He doesn't walk as fast as he did. Because I always used to have to tell him to slow down and
used to have to run sometimes to keep up with him and now he tells me to slow down. Don't you?
But...l mean we go at a fair pace, he's not dawdling, but I know I used to have to tell him to slow
down, but now its boot on other foot.
DEREK: Well I get a bit behind sometimes when I'm at work and I can't do something, you know,
or it really sbwà me down or if its come on really bad. lt frightens you a bit and you think, l'm not
going to do that, someone else'11......so it gets you a bit down then. But then l'll think.......well if I
were in an office, you know, just sat there writing or something like that. I don't think you'd even
notice aught would you?
A: So its because you're doing a physicaljob that you notice it?
DEREK: Well, thats what l've put it down to. You know, I think, if I were sat at a desk, if I were sat
here all night, or sat at a desk I don't think you'd feel anything would you because you're not really
exerting yóurself, so I come out of it then, thinking well, at least l'm getting a bit gf .exercise at
work. I fàep like I say, wondering whether to walk to work and walk back, but I don't know. I like
me bed.
The NIHR RDS for the East Midlands / Yorkshire & the Humber 2009 QUALITATIVE OATA
ANALYSIS
43
Appendix 2 - Sample Transcript (Sue)
lnterview with Sue, aged 52, secondary school teacher. Heart attack 14 months previously.
SECTION í
A: whilst you were in hospital, when you knew you've had a heart attack, how did you feel about it?
SUE: I was angry, I was angry that this could happen to me, somebody who'd not been ill. I'd
done as mucn àd I could have done to lead a healthy sort of lifestyle, so I was angry, I was
frightened, I was worried, I was frightened about another one occuning, I was worried about the
efiects, long-term effects on the family and on myself. There were those sort of things going
through my mind.
A:
But now, a year later, looking back, how has your life been affected by that dramatic thing?
SUE: I've cut down on my work load. I don't do half as much as I used to.
A: Are you doing the same job?
SUE: I'm doing the same job. I'm trying to isolate my home life from my work life. Before I used to
bring work home, now I go in early in the morning, I stay late at night. I say to myself, 'Well, I've
not iinished it, so what?' whereas at one time I would - l'd got a set of books to mark, say. They'd
got to be
marked
now I'll walk away, think, 'lt doesn't matter, I'll do them tomorrow.'
A: What else has changed?
The NIHR RDS for the East Midlands / Yorkshire & the Humber 2009 QUALITATIVE DATA ANALYSIS
SUE: I try to take more exercise, I don't worry about things as much.now as I used to, er, l'm still in
some *iys scared of it happening again. lí I get a bit a bit of a twinge I think, 'Oooh,' you know
what lmean?
I think l,m a much calmer person now. I don't think I get as up tight about things as I used to. Mind
you, bear in mind, just before this happened we'd had an OFSTED inspection, and I'd be.en for 2
years Oefore had bèen asked to take over this job l'm doing now as Head of Science to allow one
job
öf my colleagues to retire. And I was told it wasonly for a year. I then discovered l'd got the
I
prepare
inspection,
for
it,
and
to
to
work
doing
permanentlyl and the department needed a lot of
won't worry about that again.
There are more importànt things in life than getting a good report at the end of an inspection'
Those sort of things.
A:
But you have returned to the same job, you haven't changed, not reduced hours or anything?
SUE: No, no, no, I'm probably doing what the vast majority of people do now. I was, I'll admit
I
was a workaholic and a worrier, er, but l've got a department that is extremely supportive, and will
say to me, 'What time are you going home tonight? You're not taking any work with you, are you?'
anà ask me how I am and let me know they're concerned about me. They were absolutely
magnificent allthe time I was off. I had visits nearly every day.
tn tàA it was like open house, you know, people were coming at lunchtime, they were coming after
work and so on, because the school's only up the road. And er a lot of support from them, from my
immediate colleagues, and the pastoral team said, 'Well, we're not going to give you a form, we'll
attach you to a year,' so I haven't got form responsibilities'
A:
So you've been able to cut down on some responsibilities.
SUE: Yes. And I used to be union rep, l've given that up. I used to be on the PTA Committee,
l've given that up, er what else did I do? There were a few other things. ln fact, my work was
dividéd amongst 6 people while I was off. So I think that sort of told them just how much of a load
l'd been carrying.
A:
Um. Was it hard to go back then?
SUE: Yes. One of my colleagues said to me, 'You know, when you c;lme back to school, you
were scared.' So they picked up allthose things and acted on them. And my husband had been
very supportive, and my daughter very supportive as well. Although she was very very angry at
what had happened when she was in the middle of doing 'A' levels.
A:
Do you think it's changed your outlook about the future?
SUE: Yes, I try to think now, 'Well, do what we want to do. Don't put it off.' We've put off so
much, we've put off going on holiday, we've put off having what you might call life's luxuries that
we could have afforded, but well no we'd better save up for our old age. You think to yourself,
'There might not be an old age.' I've got a bit more, I won't say totally, a bit more of the sort of
approachio life - well life could end tomorrow so make the most of it now. And I'm trying to do that
- but breaking the habits of a lifetime is a bit difficult.
A: So your outlook is different?
SUE: Well, I feel I'm leading more of a normal sort of life now, not a life that's dictated by my work.
Um I did a lot of things that I would never have had time to do before, I felt much more rested. I'd
never ever had a complete rest. l'd never had a break, I mean the longest I'd ever been away from
work was er from the end of January to the middle of June after my daughter was born.
sEcTtoN 2
A:
Now, a year afterwards, do you feel your health is back to normal?
SUE: Yes, in fact I'm probably more healthy than I was, than I was over the 2 - 3 years leading
to it because I was I'd not had any exercise, l'd not been active'
A: When you say I feel healthier than I was, how do you define healthy, for you?
Thé NIHR RDS for thé East Midlands / Yorkshire & the Humber 2009 QUALITATIVE DATA ANALYSIS
up
SUE: Well, I'm having the exercise so that when I go up the hills I'm not as out of breath as I was.
I feel I've got more stamina. I suppose that's fitness really rather than health.
And the tàa tnat I'm taking medication, because I discovered my cholesterol level was high, so
thats down. I'm taking the aspirins so my clotting time is down and er l'm taking the beta blockers,
which I had done befóre, I'd taken beta blockers after my daughter was born because I started
having high blood pressure after my daughter was born, that didn't come down. And I went to see
the doctor.
I have to go 3, 4 times a year, and the last time I went she said to me, 'You know, I don't think of
you as someone who suffers from heart trouble. I said, 'Well, I don't either, but it's nice to hear you
iay that.' That was sort of another 'well, yes, I am a healthy individual'. So it's, I think it's
psychological as well as physical.
A:
Psychological health? You think that's better?
SUE: Yes. I went to see the occupational health nurse and had a couple of sessions with her and
I found what she had to say very beneficial as well. The thing that sticks out in my mind was'You
go to work, you do a good day's job, you give value for money, you come away and have enough
énergy and iime to lead a life.' So thats another, looking at it from another healthy point of view
that I'm now getting much more of a balance between work and rest and relaxation.
A: Do you worry about the future at all?
SUE: Only in as much as I would like somebody, and I daren't ask for it, I would like somebody to
have a look at all the vessels in my heart and say, 'Well, yes, they are all all right, this was just a
minor blip'. But I don't think they'll do that.
A: An angiogram?
SUE: Yes. No they didn't feel it was necessary and when I sort of, to a certain extent, trust the
medical profession because I have a friend who is very into cardiovascular stuff and he said, 'Have
you had an angiogram?' I said no and I didn't take it any further. Part of me wants to know and
ihe other doesn't although if there was anything l'd be quite prepared to take prophylactic
measures rather than waiting till something happens.
I feelthat apart from somebody looking at the blood vessels, l'm doing -- l'm not getting as much
exercise as I ought to because it has tailed off I must admit. I used to walk every single day, fast
Íor half an hour. Now I might manage it 3 times a week. But still l'm doing the aerobics.
A: So you're feeling fitter.
SUE: Yes. I mean I do, I do everything that I have done before. But I was told by the people at
the hospital that they would epect me to be back to normal, and l'm starting to learn what normal
is to me - tearing down a corridor, leaping about in the lab trying explain things, you know. I work
in a very physical way. lf I'm going to explain to the kids how big the surface area of the intestine
is, l'll leap down the lab, you know. I sort of leap down the lab and I think -'Am I all right doing this
sort of thing? Yes, yes you are because you do the aerobics.' You can do all that and keep up
with people a lot younger than myself.
A: So it's been quite a positive experience?
SUE: Yes, I think so, yes. I mean there is the negative part, in that I know l've got some damage.
I haven't enquired how much, because yes I can cope with the exercises so I must have a
reasonably efficient pumping system. There is just that thing that atways niggles at the back of
your mind now and again - will it happen again? And yet I know so many people who have had a
heart attack and 20, 30 years on they're still perfectly OK.
The NIHR RDS for the East Midlands / Yorkshire & the Humber 2009 QUALITATIVE DATA
ANALYSIS
46
!
Analyse van kwalitatief ond erzoel<smateriaal
F Wester
Dit artikel is het derde uit een serie van vier over de methoden
van kwalitatief onderzoek. H&W publiceert steeds meer kwalitatief onderzoek. Een heldere beschriiving van de methodologie kan
ook gewone lezers helpen om de waarde van dergeliik onderzoek
beter in te schatten.
ln september verscheen een korte inleiding van Phiiipsen en Vernooyr; Hak richtte zich in het oktobernummer op waarnemingsmethoden in kwalitatief onderzoek2 en in het decembernummer
gaan Van Zwieten en Willems in op de strategische betekenis van
het begrip oblectiviteit in het denken over kwalitatief onderzoek
De kern
> Net als bii ieder ander ondezoek is de analyse in kwalitatief
onderzoek gericht op de systematische beantwoording van een
goed geformuleerde vraagstelling.
> omdat in kwalítatief onderzoek vraagstellingen en begrippenkader meestal nader moeten worden uitgewerkt, bestaat het
onderzoek uit een aantal fasen waarin de analyse een steeds
ander karakter heeft.
Veel verslagen van kwalitatief onderzoek in tiidschriften beper'
ken zich tot een explorerende analyse waarvan de uitkomsten
I
lnleiding
niet systematisch zijn onderzocht.
> Het toelichten van ontwikkelde categorieën via cltaten uit het
ln dit artikel staat de kwalitatieve analyse centraal E'erst zal ik een
globale theoretische schets geven van het verloop van een kwalitatieve analyse. Daarna zal ik een aantal praktische handvatten
> Het gebruik van
aanreiken om het analyseproces in een verslag van een kwalitatief
onderzoek op waarde te kunnen schatten.
Vaak wordt analyseren als een min of meer op zichzelf staande
fase in het onderzoek beschouwd, het onderzoeksmateriaal is verzameld en wordt vervolgens geanalyseerd. Dit is - overÍgens niet
alleen voor kwalitatief onderzoek - een beperkte weergave van
wat analyseren inhouclt. Dit wordt duideliik als men stilstaat bii
de betekenis van het woord 'analyse', Volgens Yan Dale gaat het
daarbii om 'ontbinding', 'het uiteenleggen in bestanddelen'3 Zo
opgevat wordt tiidens het analyseren het onderzoeksmateriaal
uiteengelegd naar de belangrilkste bestanddelen Men kan hierbii
denken aan de verschillende onderwerpen en indelingen die met
de vraagstellingen samenhangen.
Dit wiist op de belangrijke rol die het analytisch kader (met zowel
theoretische als empirische termen) bii het analyseren heeft' Uit
het theoretisch kader kan ook worden afgeleid welke begrippen,
variabelen of classificaties met elkaar samenhangen en hoe die
samenhang uitvalt. Naast analyse in de zin van uiteenrafeling is dus
synthese in de zin van 'samenstelling van een overzichtsbeeld' of
patronen zoeken met behulp van een dergeliik kader mogeliik!
Welnu, in kwalitatief onderzoek is de analyse gericht op de uitwerking van het analytische kader van de onderzoeker. In die zin
wordt kwalitatief onderzoek ook wel als fornr ulerendonderzoek aangeduid: aan het eind van het onderzoek zal de onderzoeker het
ondezoeksprobleem in meer passende termen kunnen beschriiAuteursgegevens
Prol dr. E WesteÍ, Radboud Universiteit, sectie Communicatieweten'
schap, Postbus 9104,6500 HK Niimegen.
interviewmateriaal is geen onderbouwing van de relevantie van
zo'n indeling,
focusgroepen om snel veel re§pondenten te
ondervragen, is voor een systematische analyse dan ook veelal
ongunstig.
ven. Dit moet breed worden opgevat: het gaat zowel om het scher-
per formuleren van de onderzoeksvragen, het toespitsen van het
waarnemingsinstrument als het benoemen van ordeningscategorieën en patronen om gegevens te bewerken en te ordenen ProbJeemstelling, waarneming en analyse moeten in overeenstemming met elkaar worden uitgewerkt. De kwalitatieve analyse wordt
dan ook niet alleen gekenmerkt door een serie technieken of
hulpmiddeten om bepaalde gegevens te bewerken (analyse in
engere zin), het gaat ook om de reflectie op en de formulering van
onderzoeksvragen, waarnemingsprocedures en ordeningscategorieën (analyse in bredere zin) en dus om hulpmiddelen om dat
ref
Een schets van de kwalitatieve analyse
ln de voorafgaande bildragen is duideliik geworden dat kwalitatief
onderzoek wordt gekenmerkt door haar open karakterr'2 De
onderzoeker start met een deels open analytisch kader, dat in de
loop van het onderzoek verder moet worden uitgewerkt Het
onderzoek is een leerproces waarbij de onderzoeker ziin voorlopige ideeën op het onderzoeksveld moet afstemmen, waarnemingsprocedures moet uitproberen en vraagstellingen toespitsen'
Perioden van waarneming en analyse wisselen elkaar af, gestuurd
door voortdurende reflectie op de resultaten daarvan.a
Of het nu gaat om procedures als de gefundeerde theoriebenadering, een beschrijvende samenvatting van interviewmateriaal, een
etnografische analyse van observatiemateriaal of een narratieve
analyse van tv-drama, bii al deze werkwiizen komen een aantal
vergeli jkbare analysekenmerken en analysehandelingen terug die
Correspondentie: nwester@maw.ru.nl
Mogeliike belangenveïstrengeling;
n
lectieproces te sturen.
iets aangegeven.
met het open karakter van kwalitatief onderzoek samenhangen.a-8
Huisarrs & Werenschap
47(12) november 2004
s65
Gefaseerde werkwijze
Een van de belangrilkste kenmerken van de kwalitatieve analyse
is dat het hier gaat om een complex proces waarin achtereenvolgens een aantal tussenstappen worden gezet die deels op elkaar
voortbouwen. Daarna kan de eindanalyse plaatsvinden om de
definitieve probleemstelling te beantwoorden Het gaat hier dus
qet'aseerde werhwiizen, waarin deelanalyses plaatsvinden om tussendoelen te bereiken. De meeste stapsgewiize analyseprocedures kennen een eerste fase van exploratie en afstemming op het
onderzoeksveld. Daarnaast ziln er fasen waarin specifieke producten worden uitgewerkt (biivoorbeeld variabelen) ln de laatste
om
De rode draadl Het ondenoek van Freeman en §weèney
Als rode draad in de serie over kwalitatief onderzoek gebruiken we
een publicatie in de BMI.
Freeman en Sweeney deden een kwalitatief onderzoek om een
antwoord te kriigen op de vraag waarom huisartsen zich niet aan
evidence-based richtliinen houden. Ze hielden drie focusgroepen
van in totaal negentien huisartsen (dertien mannen, zes vrouwen)
in het zuidwesten van Engeland' De drie groepen bestonden uit
een mix van stads- en plattelandshuisartsen, afkomstig uit verschillende, geografisch van elkaar gescheiden gebieden Tiidens
u_.1 de huisartsen een
de groepsbiieenkomsten presenteerde
":n
hii de richtliin niet gevolgd had, hoewel hii
deze wel
fase vindt de eindanalyse vanuit een definitieve vraagstelling op
casus waarin
al het onderzoeksmateriaal plaats.
Voor de meest bekende werkwiize, de gefundeerde-theoriebenaderlng (kader\, die zo veel mogeliik van de empirisch verkregen
gegevens uitgaat hebben wii een stapsgewiize procedure uitgewerkt voor theorieontwikkeling in vier fasen,q'10 E'lke fase is een
besteed aan de arts-patiëntrelatie en de gevoelens die het consult opriep bii de huisarts. Alle groepsbiieenkomsten werden
deelonderzoekie waarin steeds nieuw materiaal wordt verzameld,
de analyse op bepaalde vragen wordt gericht en gereflecteerd
wordt wat de uitkomsten betekenen voor vraagstelling en begrippenkader. Zo kan de onderzoeker de opgedane inzichten steeds
weer toetsen aan nieuw materiaal en een theorie opbouwen die
dicht bij het onderzoeksmateriaal blijft,
ng
is stap voor stap
gefundeerde-theoriebenadering
bii
de
Het doel
in het
past
verschiinselen
de
op
die
een theorie te ontwikkelen
Oefu ndeerds-theoriebenaderi
veld. Vanuit een globaal idee wordt op basis van het systematisch
verzamelen en analyseten van waarnemingsgegevens dit idee uit-
kende. De groep discussieerde vervolgens over de redenen waarom de richtliin niet gevolgd was. Daarbii werd veel aandacht
opgenomen en voor de analyse uitgetypt. De auteurs deden drie
analyses gezamenliik, de rest individueel. Ze bespraken samen de
resultaten van de analyses om gemeenschappeliike thema's vast
te stellen.
Uit het ondezoek bteek dat huisartsen positief stonden tegenover evidence-based richtliinen en die ook vaak iÍ.nplementeerden. Barrières die implementatie verhinderden, waren onder
andere de persoonliike ervaringen van de huisarts. de artspatiëntrelatie, het verschil tussen eerste en tweede liin en logistieke problemen. Het implementeren van evidence is niet het
resultaat van een eenvoudig lineair proces, maar van egn geza'
menlilke beslissing van huisarts en patiënt. En daarbii is soms de
conclusie dat de regels liever niet toegepast moeten worden.
gewerkt tot antwoorden op beredeneerde onderzoeksvragen'
Freeman en Sweeney (zíe kader\ deden een onderzoek naar de
redenen en omstandigheden waarin huisartsen geen gebruik
maken van de evidence die zii wel kennen. Ook in het artikel van
Hak en het artikel van Van Zwieten en Willems in het volgende
nummer van H&W kwam en komt dit onderzoek aan de orde Een
gefaseerde opbouw van de analyse is in dit onderzoek niet te herkennen. Hoewel zii zich voor hun explorerende analyses hebben
laten leiden door procedures (zoals coderen) ontleend aan de
gefundeerde-theoriebenadering, stellen zii expliciet dat het niet
hun doel was om theorie te ontwikkelen, maar algemene thema's
te ontlenen aan de groepsgesprekken.lt Het is een typisch product van een explorerende analyse, herkenbaar maar met deels
vage omschriivingen van en overlap tussen de onderscheiden
thema's. ln volgende fasen zou men gericht vanuit deze thema's
de gesprekonderwerpen kunnen toespitsen en de analyse kunnen
richten op verschilpunten en achterliggende mechanismen. Dit
zou beter kunnen verklaren waarom huisartsen zo te werk gaan.
Tekst als materiaal
Freeman AC, Sweeneg K. Whg general practitianers do not implement eu.t'
dence: q ualitativ e studu. BMI 200 I : 3 23 : I I 00'2. (zte www'b mi.tom voor het
volledige artikell
ten, observatiemateriaal of interviews, de onderzoeker mist veelal een uitgewerkt kader om dit materiaal meteen te ordenen,
zoals dat in een vragenlijst met antwoordcategorieën gebeurt De
onderzoeker zal dus op een of andere manier de ruwe antwoorden
of observaties moeten noteren. Bovendien wil men in kwalitatief
onderzoek het perspectief van de onderzochten vastleggen. Dat
betekent dat men het belangriikvindt in het onderzoeksmateriaal
de handelingen, opvattingen en denkbeelden van de onderzoeksgroep zo veel mogeliik ín hun eigen woorden vast te leggen. Voor
interviews betekent dit meestal dat de vraaggesprekken op band
worden opgenomen,
Omdat de waarneming in met name de beginfase van een kwali-
tatief onderzoek onzeker is, past de onderzoeker
controleprocedures
toe repli catie:
(
herha
Ii n
g
va
daarnaast
n waa rnem ing: tria n-
gulatie: iets vaststellen met behulp van verschillende methodes Van Zwieten en Willems gaan in hun artikel in het volgende H&W-
Een tweede kenmerk van de genoemde procedures voor kwalitatieve analyse is dat de analyse wordt toegepast op, meestal
nummer hier nader op in) waardoor het materiaal nog eens
omvangrijker wordt. Maar ruw materiaal in de zin van audio- of
omvangrijk, tekstmateriaal. Of het nu gaat om (media)documen-
videobanden kan moeiliik worden geanalyseerd, er moet dus een
566
47(12) november 2004
Huisarts & WetenschaP
O
.,
andere vorm van trdnscriptieplaatsvinden om de analyse ervan
te vergemakkeliiken.
Dit betekent dat de kwalitatieve analyse plaatsvindt op de uitqeschre'
of docuven versies van observaties, gespreksf ragmenten, interviews
menten. Ook in het ondetzoek van Freeman en Sweeney werden de
groepsgesprekken op band opgenomen en daarna uitgeschreven'
waarbii de beide onderzoekers, die niet zelf de gesprekken hadden
gevoerd, ieder voor zich de transcripten analyseerden
ll
(zoals
Wat de onderzoeker wel of niet in het transcript opneemt
van
aarzelingen, stemhoogte, een veelbetekenende blik) hangt af
zijn aandachtspunten. Voor het meeste interviewonderzoek vol'
staat men met het weergeven van de woordeliike tekst' hier en
daar aangevuld met toelichtend commentaar' Een weergave van
een gesprek van arts en patiënt vraagt veelal om interactionele en
contextuele toelichting
bliikt
bii sommige gespreksfragmenten
Vaak
de noodzaak daarvan pas tiidens een eerste analyse van het
materiaal.
Zelfs een kleinschalig onderzoekie met vijftien vraaggesprekken
van een uur levert zo al een stapel papier op van honderden bladzijden. Het open karakter van het onderzoek maakt het bovendien
om soepel met de transcripties om te kunnen gaan'
belangriik
omdat altiid aanvullingen noodzakeliik kunnen bliiken Het
gebruik van computerprogramma's waarmee men eenvoudig met
het tekstmateriaal kan omgaan is voor de analyse dan ook steeds
belangriiker geworden.
De interpreterende analyse: open coderen
geproDe kern van de kwalitatieve analyse is het lezen van de zo
duceerde teksten in de drievoudige betekenis van waarnemen'
namelijk van tekens zoals woorden en zinnen, selecteren en interpreteren. De lezing is selectief, omdat de lezing gericht is op de
beantwoording van bepaalde vraagstellingen. In het onderzoek
van Freeman en Sweeney gaat het om de vraag waarom de huisartsen geen gebruik maken van de evidence die zij wel kennen'
Men zou het gespreksmateriaal ook kunnen analyseren vanuit
andere vraagstellingen (biivoorbeeld over de huisarts-patiëntrelatie), De lezing is interpreterend omdat de onderzoeker aan de
hand van de tekst antwoorden op die vragen formuleert in termen
van het analytisch kader (begrippen, interviewonderwerpen,
variabelen) dat in het onderzoek wordt gehanteerd'
Het zal duideliik ziin dat in het begin van het onderzoek' als vraagstelling en analytisch kader nog minder precies ziln uitgewerkt,
dit proces een ander karakter heeft dan in de latere fasen wanneer
vraagstelling en analytisch kader vaststaan. In navolging
van
Strauss kan men dan ook een onderscheid maken in opencoderenin
de verkennende fase, geichl coderen in de fase voor het uitwerken
van dimensies en variabelen (zie Cericht coderen en vergeliikende ana'
lyse) en selectief coderen in de eindfase lzie: ProÍíelkaarten) bii het
tentatief en herhalend, waarbii de onderzoeker zo veel mogeliik
relevante trefwoorden in de kantliin bii de tekstsegmenten
plaatst. Elk segment wordt gelezen vanuit elke afzonderliike
onderzoeksvraag. De trefwoorden in de kantliin geven aan dat er
een onderwerp voorkomt en wat daarover gezegd wordt Voor een
deel zi!n die trefwoorden gebaseerd op de theorie en de veldkennis die de onderzoeker vooraf ontleend heeft aan eigen ervaring
of de literatuur, maar het kan ook om nieuwe trefwoorden gaan'
Daarnaast leest de onderzoeker elk segment vanuit het perspectief van de respondent: wat brengt deze, naast de onderwerpen
van de onderzoeker, nog naar voren En zoals hierboven aangegeven, wordt dit proces meermalen herhaald, omdat latere segmen-
ten of die van andere respondenten iets duideliik kunnen maken
wat eerder nog niet was onderkend.
Dit proces van open coderen levert een groot aantal trefuoorden
op, zodat de onderzoeker behoefte kriigt aan hulpmiddelen om de
koppeling van trefwoorden en materiaalsegmenten soepel te
beheren, overzicht te houden over de gebruikte trefwoorden, segmenten met hetzelfde trefwoord met elkaar te kunnen vergeliiken
en de trefwoorden te ordenen en groeperen rond een centraal
onderwerp, Het gebruik van de computer als hulpmiddel is dan
ook al gauw noodzakeliik om het een en ander goed te beheren'
Het lezen van het onderzoeksmateriaal is aldus een creatief proces, dat gestuurd wordt door voortdurende reflectie op analysevragen, waarnemingsmateriaal (de vraaggesprekken) en ordeningscategorieën. Daarbij komt de onderzoeker tot vele ideeties,
inzichten en beslissingen, biivoorbeeld over de verandering van
trefwoorden, de herformulering van onderzoeksvragen, de aanpassing van de topiclijst of het doorvragen in vervolginterviews
bii bepaalde onderwerpen. Die ideeën en beslissingen, en de achtergronden daarvan, kunnen eenvoudig verloren gaan als ze niet
in memo's worden vastgelegd. Het gaat hierbii niet alleen om de
ondersteuning van het geheugen van de onderzoeker' Voor een
belangriik deel is het schrilven van memo's ook het expliciteren en
de ideeën, inzichten en beslissingen die tiidens het
formulerenvan
reflecteren ziin opgekomen en die vervolgens in de analyse moeten worden toegepast. Bovendien maakt de onderzoeker ook
gebruik van literatuur over het relevante onderzoeksveld, die hii
met het eigen werk moet verbinden. Het uitschriiven en vastleggen van deze inzichten in memo's is een belangriik hulpmiddel
om de analyse cumulatief te laten ziin en tot herÍormuleringen
van vraagstellingen en analysekader te komen.
Freeman en Sweeney hebben het proces van open coderen niet
expliciet beschreven.rrWel geven zii zicht op het reflectieproces:
zil hebben de groepsgesprekken onafhankeliik van elkaar gecodeerd en hebben de uitkomsten vergeleken om gemeenschappe-
beantwoorden van de vraagstelling.5 Bii alle drie de vormen van
coderen legt de onderzoeker een koppeling tussen segmenten uit
het materiaal en de vraagstellingen die centraal staan'
liike thema's te formuleren. Het liikt erop dat zii niet de systematiek hebben gehanteerd die hierboven wordt voorgesteld om aan
de hand van de codes stapsgewiis precieze analytische categorieën uit te werken. Zii ziin meer uitgegaan van hun professione-
Het proces van open coderen in de verkennende fase is vooral open,
huisartsen zo handelen. De zes thema's die zil onderscheiden, ziin
Huisarts & werenschaP
47(12) november 2004
le achtergrond om 'redenen'te kunnen onderscheiden waarom
567
redenen soms in de zin van achtergronden, motieven of condities
die een rol spelen.
Gericht coderen en vergelijkende analyse
Als de exploratiefase is afgerond en de onderzoeker zicht heeft op
de centrale onderwerpen die van belang zijn - bij Freeman en
Sweeney ziin dat de zes onderscheiden thema's - zal hil het analr
lytische kader rond deze onderwerpen verder kunnen uitwerken
Daartoe selecteert
hij nieuw materiaal, bilvoorbeeld viif inter-
views die met een aangepaste li!st van gesprekstopics gericht op
deze centrale onderwerpen ziin uitgevoerd' Dit nieuwe materiaal
wordt vanuit de geformuleerde vraagstellingen gelezen, waarbii
nu zo veel mogeliik gebruik wordt gemaakt van de trefwoorden die
hij al heeft geordend, Bii dit qericht coderen wordt de toepasbaarheid van trefwoorden getoetst, wat veelal tot aanscherping en
herformulering leidt. Alleen als bestaande trefwoorden niet passen, worden nieuwe trefwoorden geformuleerd Zo ontstaat een
relatief groot bestand van gecodeerd materiaal waarbii de centraIe onderwerpen van het onderzoek op verschillende manieren op
diverse plaatsen in het onderzoeksmateriaal voorkomen
Door een vergeliikende analgse van segmenten waarin een dergeliik
centraal onderwerp aan de orde komt, kunnen de trefwoorden
rond dat onderwerp worden geabstraheerd in overeenkomsten'
verschillen en variaties op achterliggende dimensies Met deze
dimensies kan al het materiaal vervolgens worden beschreven'
Freeman en Sweeney waren niet gericht op het analytisch uitwer'
ken van de door hen onderscheiden thema's, omdat zii (en de
tiidschriftredactie) de resultaten van hun exploratieve analyse al
interessant genoeg vonden. Verdere analyse had ongetwiife)d tot
scherpere onderscheidingen geleid in motieven, dan wel achtergronden of condities die een rol spelen
Profielkaarten, oveaichten en tabellen
Zodra de analyse eenmaal zicht heeft gegeven op belangriike abstracte ordeningscategorieën als dimensies, kriigt de onderzoeker
behoefte aan overzichten hoe die categorieën in nieuw materiaal
voorkomen en met name hoe zii aan de onderzoekseenheden
(gesprekken, respondenten, organisaties) ziin gerelateerd De
onderzoeker wil dus bestanden creëren waarin het materiaal per
eenheid is geordend naar de belangriikste aandachtspunten, om
zo een profíel te kunnen maken van alle eenheden met behulp van
de dimensies van elk onderwerp. Dergeliike profielkaarten vor-
men het basismateriaal voor de vergeliikende analyse op een
ander niveau, biivoorbeeld een vergeliiking van respondenten ln
het onderzoek van Freeman en Sweeney kan men denken aan een
beschriiving van wat elke huisarts heeft gezegd in termen van elk
van de genoemde redenen om geen gebruik te maken van de
beschikbare evidence. De overeenkomsten en verschillen van de
respondenten met betrekking tot de centrale onderwerpen kunnen in variabelen worden uitgedrukt; soms gebeurt dat in de vorm
van
ttlpen
die worden onderscheiden (biivoorbeeld de
orilinteerde versus
de receptgeoriënteerde huisarts) op grond van meer-
dere variabelen tegeli ik.
568
patirintge'
Wanneer voor alle centrale onderwerpen dergeliike variabelen ziin
uitgewerkt, kan al het onderzoeksmateriaal vanuit het nu vast-
staande analytisch kader worden gecodeerd (selectief coderen)'
Daarmee kunnen overzichten en tabellen worden gemaakt, waarin ook naar samenhang tussen variabelen kan worden gezocht'
Rapportage
Ten slotte zal de onderzoeker verslag doen van het onderzoek in
de vorm van een onderzoeksrapport of een artikel. Het schriiven
van een dergeliik onderzoeksverslag is op zich weer een vorm van
analyse, waarbii vraagstelling, methoden, resultaten en conclu-
sies in overeenstemming met elkaar en met de theoretische en
veldspecifieke literatuur moeten worden geformuleerd Bii de
beschrilving van de bevindingen en hun betekenis wordt veelai
gebruikgemaakt van citaten uit het onderzoeksmateriaal die deze
bevindingen kunnen illustreren. ln de rapportage wil de onderzoeker de beschikking hebben over voorbeeldseqmenten die relatief
zelfstandig te lezen ziin en goed aansluiten bii het ontwikkelde
analysekader. Dit soort centrale voorbeeldsegmenten ziin veelal
in reflectiememo's al eerder besproken en vaak al van extra trefwoorden voorzien, zodat de selectie achteraf eenvoudig kan
plaatsvinden.
De rol van de comPuter
In het voorafgaande mag al duidelilk geworden ziin dat de kwalitatieve analyse een groot aantal administratieve handelingen met
zich meebrengt.
dit soort werkzaamheden ziin specifieke computerprogramma's ontwikkeld (zoals Atlas-ti, Kwalitan, Nvivo, The Ethnograph, Winmax), die in verschillende mate mogeliikheden bieden
om dit soort activiteiten te ondersteunen.r0 De computer analy-
Voor al
seert niet, maar ondersteunt de analyserende onderzoeker, soms
met opties (woordenoverzicht, zoeken naar thema's, automatisch
coderen, systematisch trefwoorden wiizigen, woorden in context
weergeven) die het monnikenwerk sterk verlichten Bovendien
zorgt de computer voor systematiek in de analyse, biivoorbeeld
door alle segmenten met een specifiek trefwoord te selecteren in
plaats van enkele voorbeelden die de onderzoeker zich herinnert'
Aandachtspunten bij het lezen van een analyse
Niet elk kwalitatief onderzoek start met een globale vraagstelling
en/of een beperkt analytisch kader en is gericht op de ontwikkeling van gefundeerde theorie (meer hierover in het vierde artikel
in deze serie van Van Zwieten en Willems). Soms gaat het slechts
om de illustratie van verschillende perspectieven die worden
gehanteerd of de beschriiving van een procesverloop bii twee
casussen. Bovendien zal niet iedere onderzoeker de reconstructie
van het actorperspectief of de leefwereld van de onderzochten
even belangrijk vinden. De meest eenvoudige vorm van kwalita-
tieve analyse is een analyse van de antwoorden op een open
vraag in een vragenliist, waarbil men passende antwoordcategorieën wil ontwikkelen. ln ander onderzoek streeft men een samenvatting na van wat de respondenten denken over enkele centrale
47(12) november 2OO4
Huisarts
& WetenschaP
t
,n"rrlr'r, zoals in het geval van Freeman en sweeney In dit soort
kwalitatief onderzoek gaat het meer om empirische verkenning
Abstract
dan om uitwerking van een theorie zoals hierboven beschreven is'
om bii lezing van een artikel over kwalitatief onderzoek de analyse goed te kunnen volgen is duideliikheid over de doelstellin4 van het
onderzoek dan ook uiterst relevant.
2004i47 ll2\:565'70.
Een tweede aandachtsp
u n
t ls het globale
verloop van de analyse en voor'
F.
Analysis of qualitative research material. Hui§arts Wet
oualitative research is characterised by its anal$ical goals' the development of categoÍies, the elaboration of concepts oÍ the foÏmulation oÍ a
theory, Because of this analytical openness, the research design shows
successive phases, each with its own obiective and specific demands for
data collection. textual analysis and analytical reflection. Many reports
al de sgstematiek daawan. Een van de belangriikste aspecten hierbil
is de gefaseerde opzet van het onderzoek en vooral de afwisseling
on qualitative research fall to present a systematic analysis to suppoÍt
the relevance of developed categories. Three steps in qualitàtive analysis
van waarneming en analyse die daarmee samenhangt Het is in
kwalitatief onderzoek niet verstandig om eerst al het materiaal te
verzamelen en dan met de analyse te beginnen ln de explorerende fase van het onderzoek weet de onderzoeker nog niet precies
aÍe presented here: interpretative analysis using open coding, comparative analysis using axial coding and integrative analysis using profile
wat hii wil weten, en dus zal dit materiaal onvolledig ziin gelet op
het uiteindeliike analysekader. Dat betekent dus dat onvolledige
interviews moeten worden aangevuld, dan wel van de uiteindelii-
comparing and reÍlecting on text segments
ke analyse moeten worden buitengesloten.
Bovendien willen we de kwaliteit van de categorieën die we gefor-
muleerd hebben, toetsen aan ander materiaal dan het materiaal
waaraan zii ziin ontleend Dit zal meestal betekenen dat we nieuw
materiaal moeten verzamelen waarbii wii gerichter moeten observeren of interviewen om te ontdekken of die categorieën al of niet
aanwezig ziin.
Een tweede aspect van de systematiek van de analyse is de mate
waarin de analyse is gebaseerd op aLhet relevante materiaal' Bilvoor-
beeld moeten categorieën relevant ziin op het niveau van de
respondent of de onderzoekscase en niet alleen relevant zijn in
enkele fragmenten uit het materiaal van een respondent of onderzoekscase. Bovendien moet duideliik zijn of categorieën geba-
seerd zijn op overeenkomsten en verschillen tussen alle respondenten of onderzoeksgevallen, of dat de onderzoeker zich heeft
gebaseerd op enkele interessante gevallen. Het is luist op deze
punten dat het verslag van de analyse van Freeman en Sweeney
tekortschiet.
O
Wester
cards and selective coding. The computer can be of great help in dealing
with the administrative work related to the cyclical process of reading,
waarvan onduidelilk is of het nu
zijn (zie hiervoor bi! open codemotieven
redenen, condities of
liin en logistieke problemen),
ren
).
De empirische onderbouwing geschiedt meestal met illustraties
uit onderzoeksmateriaal, biivoorbeeld citaten uit de interviews'
Daarbii moet in het biizonder aandacht worden geschonken aan
de verschillende manieren waarop de categorie voorkomt Freeman en Sweeney hebben dit zeer uitgebreid gedaanr elk hoofd-
thema lichten zii met meerdere citaten toe We moeten hierbii
bedenken dat onderzoekers in artikelen niet altiid de ruimte hebben om aandacht te schenken aan de empirische illustratie van
alle ontwikkelde categorieën. Bovendien bestaat het gevaar dat
men de citaten uit het materiaal vanuit een ander gezichtspunt
leest dan de onderzoeker met ziin categorie bedoelt. De presentatie van citaten moet dan ook goed worden ingeleid en de inter-
pretatie ervan worden toegelicht, want bii lezen van teksten is
niets vanzelfsprekend, Dat de lezer ook nog iets anders uit het
citaat weet te halen, hoeft dus niet te betekenen dat de onderzoeker fout zit!
,"n
derde aandachtspunt is het onderscheid tussen de analqtísche
termen die de onderzoeker als uitgangspunt heeft genomen en
waarmee de probleemstelling en de onderzoeksopzet ziin uitgewerkt en de termen die als product van de analyse moeten worden gezien. De eerste termen worden vooral gebruikt als interpreta'
moet worden omschreven wat eronder moet worden verstaan. Maar van de tweede soort teÍmen willen we weten
hoe ze tot stand ziln gekomen. Hier willen we inzicht kriigen in de
analyse wat betreft de Íormulering van analytische categorieën
tíekader, waarvan
die aan de hand van het onderzoeksmateriaal ziin ontwikkeld. Die
formulering moet allereerst onderbouwd worden met empirisch
materiaal, daarnaast is van belang hoe de categorie zich verhoudt
tot de bestaande vetdspecifieke of theoretische onderzoeksliteratuur. Beide ziln van belang om te overwegen waarom de categorie
in deze bewoordingen is geformuleerd, lk wiis op de hoofdthema's
van Freeman en Sweeney (de persoonliike ervaringen van de huisarts, de arts-patiëntrelatie, het verschil tussen eerste en tweede
Huisarts & WerenschaP
Een vierde aandachtspunt is de onderbouwing van gevonden patronen
in het materiaal. Zo leidt kwalitatieÍ onderzoek nogal eens tot de for-
mulering van een typologie gebaseerd op de verschillen en overeenkomsten op meerdere dimensies. Men zou de indeling van
Freeman en Sweeney als een aanzet tot een dergeliike typologie
kunnen zien. Nou lukt het altiid wel om een indeling te maken,
maar de vraag is of de indeling relevant is. De relevantie van een
bepaalde indeling kan bliiken uit de samenhang met iets anders,
bijvoorbeeld verschillen tussen respondenten in voorkeuren of
gedragingen.
Dit kan blilken door de typologie en de gedragingen in een overzicht of tabel op elkaar te betrekken. In kwalitatief onderzoek kan
men een verdeling wel getalsmatig weergeven, maar een statistische onderbouwing is door kleine aantallen en/of niet-aselecte
47(12) november 2004
steekproeftrekking meestal niet mogelÍik. Wel kan men nagaan ol
alle eenheden volgens het patroon kunnen worden geordend, en
s69
t
er condities ziin waarom sommige eenheden afwijken van het
",
patroon. Ook kan men nagaan of een ordening volgens aan de
Iiteratuur ontleende alternatieve hypothesen al of niet past, dan
wel of het patroon aansluit bii de theorie of de veldspecifieke literatuur.
Conclusie
Via een cyclisch proces van lezing, vergeliiking en reflectie werkt
de onderzoeker tentatief geformuleerde onderzoekstermen uit tot
Literatuur
I Philipsen H, Vernooii'Dassen M. Kwalitatief onderzoek: nuttig'
onmisbaar en uitdagend. Huisart Wet 2004;47:454'7'
2 llak T Waarnemingsmethoden in kwalitatief onderzoek Huisarts Wet
2004:47,502'8
3 Geerts G, Heestermans H. Van Dale Groot Woordenboek der Nederlandse Taal. UtÍechti Van Dale Lexicografie, 1984:177
4 Wester F. Strategieën voor kwalitatief onderzoek, Muiderberg'
Coutinho, 1987.
5 Strauss À. Corbin I. Basics of Oualitative Research. t,ondon: Sage,
I
weinheim: DSV 1987.
7 Spradley lP Participant observation. New York: Holt, Rinehart E
Winston. 1980.
8 wester F, Verbrugge N. Op zoek naar boodschappen in sitcoms
begrippen of variabelen die goed aansluiten bij het onderzoeksmaterjaal waarin de perspectieven van de onderzochten naar
voren komen. Hoewel het hier gaat om een deels onvoorspelbaar
creatief proces, kunnen stapsgewilze procedures voor sturing zorgen. Daarmee is succes weliswaar niet verzekerd, maar de navolgbaarheid en de overdraagbaarheid van het onderzoeksverslag
kunnen zo wel worden bevorderd,
Kij kgaovetenschap
Empirische wetenschap kwantificeert en analyseert er lustig op
los. Daar kleeft een nadeel aan: afstandeliikheid. Anders gezegd'
er is sprake van een groeiende distantie tussen wetenschappeliike modellen en de poedelnaakte - vooral beleefde - werkeliik-
heid. Hoogteraar medische sociologie Gerhard Niihof zegt er dit
over in zijn afscheidscollege:r 'Nadruk op theorie, het gebruik van
afstandeliike methoden, afstandelijke veronderstellingen en de
overname van in de samenleving dominante categorieën, leiden
tot een gedistantieerde sociologie, bedacht en gepraktiseerd op
afstand van de alledaagse sociale werkelilkheid',
Niihof houdt zich al tientallen iaren bezig met ziekteverhalen. Al
die kennis kwam in een ander perspectief te staan vanaf het
moment dat hii darmkanker kreeg. Hii schreef over die ervaringen
het boek Ziekenwerk (zie ook het artikel van Vandamme en Oderwald in dit nummer op p.597). In het voorwoord noemt hij dat verslag een reisverslag, want 'ziekte betekent ook het betreden van
onbekend terrein'.2 Het woordie 'onbekend' is van belang in deze
zin. In ziin afscheidsrede komt hii er nameliik op terug, maar in
een andere hoedanigheid, nameliik als spiegelbeeld, het 'gewone', het 'alledaagse'. Heel ziin wetenschappeliike carrière ziet hij
als een speurtocht en een pleidooi voor een grotere aandacht
voor het alledaagse en het gewone in een mensenleven. Hii richtte zich daarbii op het onderzoeken van het 'ongewone'. Want het
leven van een chronisch zieke wordt als 'ongewoon' ervaren. Niet
zozeer in de beleving van de patiënt als wel in de ogen van anderen, buitenstaanders. Het ongewone kan daarbii talloze vormen
aannemen: in lichaamsuitingen (niet rechtop kunnen lopen), ver-
t
s7o
990.
ó Mayring P Oualitative Inhaltsanalyse. Crundlagen und lbchniken'
Sociologische Gids 20O0;43:243'67
.
9 Claser BC, Strauss AI-. The discovery of grounded theory: strategies
for qualitative research. Chicago: Àldine, 1967.
R Peters V Kwalitatieve analyse, uitgangspunten en procedures. Bussum: Coutinho, 2004.
Freeman AC, Sweeney K. Why general practitioners do not implement
evidence: qualitative study. BMI 2001;323:l 100-2.
l0 Wester
I
I
lies van gratie, traagheid en biivoorbeeld verlies van vertrouwen'
Het iukvan het gewone is zwaar. In onze cultuur is het gewone de
maat der dingen.
Via de methode van het levensverhaal probeerde hii te achterhalen wat chronisch zieken, en via hen ook mensen die hen bezig
zien, ongewoon vinden in hun gedrag. Levensverhalen noemt hii
'kijkgaten' op de samenleving en vooral op de manier waarop
zingeving toekennen aan zÍekte en gezond-ziin 'Door
mensen
gebruik te maken van een weinig gedistantieerde methode als die
van het levensverhaal en van een analysemethode die zo dicht
mogelilk bij de onderzoeksdata bliift, kan ik zichtbaar maken hoe
een chronische ziekte dit gewone, vanzelfsprekende en natuurliike sociale leven verstoort, zowel het publieke als het persoonliike.' De opbrengst? Ziin onderzoek laat zien hoe precair het gewone leven van gewone mensen is. 'Onderzoek van chronisch zieken
laat zo de fragiliteit zien van het gewone leven Het toont de
kwetsbaarheid van mensen die hun leven vanzelfsprekend als
gewoon en natuurliik leven.'
De afscheidsoratie van Niihof is een warm pleidooi voor de menselijke factor in wetenschappeliik onderzoek. Voor betrokkenheid
in plaats van afstandelijkheid. Voor zingeving door individuen in
plaats van abstracte ciifers uit populaties, Onverminderd is Niihof
een man die gelooft in ziln missie. Het wrange bii dit alles is dat
de Universiteit van Amsterdam bii het afscheid van Niihof de leer-
stoel ophief..
.
Frans Meulenberq
1
NUÍroÍ C. Ongewoon ziehenleven. Amsterdau: Het Spinhuis. 2004.
Een kleine sociologie van alledaags ziekenleG. Ziekenwerk
ven. Amsterdam: Aksant, 2001.
2 Niihof
47(12) november 2004
-
Huisarrs & wetenschaP
Qualitative research review guidelines
THIS SHOULD BE INCLUDED IN THE
MANUSCRIPT
ASK THIS OF THE MANUSCRIPT
R Relevance
* RATS
of study question
Research question explicitly stated
the research question interesting?
ls the research question relevant to clinical
practíce, public health, or PolicY?
ls
Research question justified and linked to the
existing knowledge base {empirical research,
theory, policy)
A Appropriateness of qualitatiue method
ls qualitàtlve methodology
the best approach
for the study aims? '
: " lnterviews: experience, perceptions,
.
r
't'
r
::
§tudy design descrihed and justified i.e., why
was a, particu'lar rnethod {e.g., interviews}
chosen?
behaviour, practice, Process
Focu} gr,aups: Er0uP dYnamics,
eonvenience, non-sen§itive topics
EtÍt(1.§graphy;culture, organizational
behaviour, interactian
documents, art,
, Textual analysís:
representatio ns, c0nversatíons
T Transparency
of procedures
Sampling
Are the participants selected the most
appropriate to provide acces§ to the type of
knowledge sought by the studY?
ls the sampling strategy appropriate?
Criteria for selecting the study sample
justified and explained
. theareticol: based on preconceived or
emergent theory
o purposive: diversity of opinion
c volunteer; feasibility, hard-to-reach
groups
Rgcruitment
,:
l
Was:recruitment conducted usin§ appropriate
Details of how recruitment was conducted
methods?
and by whom
t:
Ís
the sampling strategy appropriate?
Could there be selection
bias?
:'
Details of who chose not to participate and
why
Doto callection
Was collection of data systematic and
Method(s) outlined and examples given (e.9.,
comprehensive?
interview questions)
AsK THI§ OË ïHË MANUSCRIPT
THI§ SHOULD BE INCLUDED IN ïHE
MANU§CRIPT
Are characteristics of the study group and
setting clear?
Study group and setting clearly described
Why and when was data collection stopped,
and is this reasonable?
End of data collection justified and described
fis/* of rese archers
the researcher{s} appropriate? Flow mÍght
they, biàs (good and bad) the conduct ofthe
§o ïhe researchers occupy dual roles
study and results?
this discussed? Do the researcher{s) critically
exarnine their own influence on the
formulation of the research question, data
ls
{clinic,ian ànd,IÉs€Brcher)? Are the ethics
of
collection, and interpretatisn?
Ethics
Was informed consent sought and granted?
lnformed consent process explicitly and
clearly detailed
Were participants' anonYmitY and
Anonymity and confidentiality discussed
csnfidentiality ensured?
Was approval from an appropriate ethics
committee received?
§ Soundness
Ethics approvalcited
of interpretive approach
An*'lysis
ls the type of.analysis a,ppropriate fór the type
justified
""ïotn*nto,
. .
r
:Analytic approach described in depth and
**p torxory, descri ptio*,
,, hy§othesis generating
:, '
ïrarnewark: e.g., policy
.
constsntcomporison/graundedtheory:
theory generating, analYtica l
'tadícatars aÍ quality: Descripticn of how
themes were derived from the data
(.inductive or deductive)
Evídence of alternative explanations being
sought
Ànalysis and presentation of negative or
Are the interpl'etations clÉarly presented and
ad§q,uately suppCIrted by the evidence?
deviant cases
Are quotes used ahd àre these appropriate and
Description of the basis on which quotes
were chssen
effective?
§emi-quantification when appropriate
lllumination of context and/or meaning,
richly detailed
Was trustworthiness/reliability of the data a'nd
interpretations checked?
Method of reliability check described and
justified
e.g-, was an audit trail. triangulation, or
A§K ïHIS OF THE MANU§CRIPT
THIS SHOULD BE INCLU§ED IN THE
MANUSCRIPT
member checking emPloYed? Did an
independent analyst review data and contest
themës? How were disagreements resolved?
Discussion ond presentation
Are findings sufficiently grounded in a
theoretical or conceptual framework?
ls adequate account taken of previous
knowledge and how the findings add?
Are the limitations thoughtfully considered?
ls
the manuscript well written and accessible?
Findings presented with reference to existing
theoreticaland empirical literature, and how
they contribute
Strengths and limitations explicitly described
and discussed
Evidence of following guidelines (format,
word count)
Detail of methods or additional quotes
contained in appendix
Written for a health sciences audience
Are r,ed flags presènt? These are comínon
features of ill-conceived or poorly executed
qualitative'stud'tes, are a cau§e for concern, and
rllust be viewed critically. ïhey rnight be fatal
flaws; or they may result from lack of detail or
ilar.ity.
Graunded thËary: not a simple conte nt
analysis but a complex, sociological,,theory
generating approach
Jargan: descriptions that are trite, pat or
jargon filled should be viewed sceptically
Over interpretotion; interpretation mu§t be
grounded in "accountsl' and semi-quantified
if possible or appropriate
Seems,onecdotol, sef e vident: rnay be a
superficial analysis, not rooted in conceptual
fràmework or linked to'previous knowledge,
and lacking depth
Cansent proress thinly'discussed; may not
have met ethics requirements
Ooctor-researcher: consider the ethical
implications for patients and the bias in data
colleition
a
nd interpretation
guidelines modified for BioMed Central are copyright Jocalyn Clark. They can be found in
Clark JP: t-íaw ta peer review a qualitotive manuscript.ln Peer Review in Healttt Sciences' Second
ïhe
RATS
edition. [dited by Godlee
F, Jefferson T. London: BMJ Books; 2003:219-235
lntsrnationol JaLrnol for Quolity in Heo/th Corcr Volume 19, Number 6: pp 349 357
Advmce Access Publicationr I 4 September 2007
I
0. I 093
/intqhc/mzm042
Consolidated criteria for rePorting
qualitative research (COREQ): a 32'item
checklist for interviews and focus grouPs
ALLISON TONGI2, PETER SAINSBURYI3 AND JONATHAN CRAIGI'2
lschool
of public t{ealth, gniyersity of
Sl,clncl,,
sPopulation l{ealth,
NS1p'2145, Àustralia, ancl
2Centre
lbr l(ic{ney Research, The Chilclren's Hospital at \x/estmead,
Sout}r V/est Arca Health Sen'ice, NSW 2170,.r\ustralia
NSw 200ó, Àustralia,
Sy'dr"rey
Abstract
Background. (]unlitative research explores complex phenomena encountered by clinicians, health care ptoviders, policy,
makers an4 consumers. Although paitial checklists are available, no cr.»nsoliclated repotting framework exists for any type of
U
qualitative design.
À
o
À
Objective. To clevelop a checklist for
explicit ancl comprehensive teport.ing
oí
qualitative studies (indepth intetviev's and
o
{
=
§
o
focus groups).
Methods. \Xë performecl a comprehensive search in Cochrane and Campbell Protocols, Medline, CINAHL, slstematic teviews
clualitative suclies, auth,rr or reviewer guidelines of maior medical iournals and reference lists of relelant publications for
er,isting chccklists use4 t6 assess qualitative studies. Sevenry-si:r items frcm 22 checklists were compiled into a comprehensive
list. Ali items were gnruped into three clomains: (i) research team and reflexivi§', (ii) srudy design and (@ data anallsis and
repr:rting. Duplicate ii.*i ,"d those that were ambigprous, too broadly defined and impractical to assess were temoved.
ol
Results. Irems m6st frequentl.v included in the checklists related to sampling met}od, setting for data collectiorq method of data
collection, respondent valiclation of findings, method of recording dau, description oí the derivation of themes and inclusion of
suppr:rting quotations. §íe groupecl all items into three ck:mains:
(f
research team and teflexiviq', (Q study design and (iir) data
in COREQ, a
research team, study methods, contex(t
9
a
a'
Ë
3
qx
È
analysis and reporting.
Conclusions. The criteria inciudecl
a
32-item checklist, can he.lp researchers to report important aspects of the
of the studli findings, analinis and interpretations.
xp
o
o
o
Keywords: focus groups, interviews, qualitative tesearch, research design
Qualitative research explores complex phenomena eÍrcountered
by clinicians, health care providers, policy makers ancl consumers in health care. Poody designed studies and inadequate
reporting can lead to inappropriate application of qualitative
research in decision-making health care, health policy and
hature research.
Ii<rrmai reporting lpidelines have been developed fcrt ran-
randomized controlled ttials [5]. Sptematic reviews of qualitative research ahnost alwaln shcnv that key aspects of study
desi€F are not reported, and so there is a clear need for a
CONSOM-equivalent for qualitative research [6].
The Uniform Requirements for Manuscripts Submitted to
Biomeclical Journals published by the International C,r:mmittee
of Medical Journal Eclitors (ICMJE) do not provide reporting
guidelines for qualitative studies. Of all the mainstream biomeclical iounals Fig. 1), only the British lvÍedical Journal @MJ)
has criteria for reviewing qualitative research. However, the
guidelines for authors specifically record that the checklist is
not routinely used. In addition, the checklist is not compre-
better understaÍ1d the design, concluct, anallsis and findin5 of
published stuclies. Thjs process allows users of published
research to be more fuller informed when they critically
appraise studies relerant to each checklist and decide upon
applicabitity of research findings to their local settings. Empiric
studies have shown that the use of the CONSORT statement
is associate<l with improvements in the quality of repotts of
hensive ancl does not provide specific guidance to assess some
of the criteria. Although checklists for critical appraisal ate
available for qualiutive tesearch, there is no widely endorsed
domized controlled trials (CONSORD t1], diagnostic test
studies (STARD), meta-anal,vsis of RCTs (QUORONf) [2],
ol:servational studies (STROBE) [3] and meta-analvses of
observational studies (,\IOOSE) [4] These aim to improve
the qua.lity of reporting these study qpes ancl allorv readers to
repoting framework for any qpe of qualitative research [].
V/e have developed a formal teporting checklist for
in-depth interviews and focus €Jroups, the most comrnon
methods for data collection in qualitative health research.
Address reprint requests to: Allison Tong, Centre for Kidney Research, The Children's Hospital at Westmead, NSW 2145,
Ausrralia. Tel: *61 -2-9845-lr482; Fax: +61-2-9845- 149 l; E-mail: allisont@health,usyd.edu.au, allisont@chw.edu.au
lnternational Journal for Quality in Heakh Care vol. l9 no. 6
((:) The Author 2007. Published by OxÍord Univershy Press on behalf oí lnternational Society íor Quality in Health Care; all rights reserved
o
@
Eo
6
C
o
p
{p
o
N
A. Tong et ol
Excluded n = 7
Title and abstract review
Excluded (n = 445)
reviewer
4
appraisal
3
No author
guidelines for
qualitative studies
Duplicate
checklist
No appraisal checklist 218
127
Qualitative
58
Primary
31
Mixed
6
Comment or
5
Duplicate article
methods
fieldwork
methods
debate
or
checklist
U
o
4
Full text analysis
5
o
§
À
o
Excluded (n = 77\
No appraisal checklist 33
25
Comment or
11
Duplicate article
checklist
Qualitative
debate
or
methods
o-
o
J
5
8
:
G
b
3
Items identified from the 22 tools/checklists
n = 76 items
24
Research team and
25
Study
27
Data analysis and
design
reflexivity
reporting
eo
?
o
(I1
§
7(
N
llems excluded
n=46
l.
Reasons for exclusion
- Duplicate item or
overlapping definition
- Not specific to qualitative
studies
- lmpractical to assess
- Ambiguous, obscure
definition
Items not found in existing
checklists
n=2
Development
of the COREQ
guidelines provided by BMJ,
JÀMA,
Checklist. *References Í26, 271, tR.f.t
of lnternal Medicine, NEJM.
o
@
o
@
o
!
o
{p
N
r,...
*Autho, and reviewer
16, 2S-32),
Basic definitions
Qualitative studies use non-quantitative methods to contrib-
complete and transparent reporting among rese,uchers and
ute new knowledge ancl to provide new perspecuves in
indirectly improve the rigor, comprehensiveness and cteclibility of interview ancl focus-group studies.
health care. Although qualitative research encomPàsses a
broad range of study methods, most qualitative research
350
e
N
Lancet, Ànnals
These two methocls are particularly useftrl for eliciting
patient and consumer priorities and needs to imptove the
qualiq.' of health care [8]. The checklist aims to Promote
o
ó
COREQ 32-item checklist
Research team and reflexivity 8
15
Study design
I
Data analysis and reporting
I
@
d
Reasons for inclusion
- Suitable for assessment
- Clear deÍinitlon
- Key characteristic for qualitative
research
Figure
o
Consolidated criteria for rePorting qualitative research
publications in health care describe the use of inten'iervs and
focus goups [81.
lnterviews
ln-clepth and semi-structured interviews expiore the experiences of participants and the metnings drev attribute to
them. Researchers encourage participants to talk about issues
pertinent to the research question by asking open-ended
questions, usua.lly in oÍle-to-one interviews. The interviewer
might re-word, re-order or clari$' the questions to futther
im'estigate topics inur.rduced by the respondent. ln <lualitative
health research, in-depth interv.iews are often used to study
the experiences and mean.ings of disease, and to explore per-
sonal and sensitive themes. They can also help to identi§potentially modifiable factots for improving health cate [9].
validation of findings, method of recording data, description
of the derivation of themes and inclusion of supporting
quotations. We gtouped all items into three domains: (i)
research team and reflexiviry (ii) study design and (iii) data
analysis and reporting. (see Tables 2-4)
§íifiin each domain we simplified all relevant items by
removing duplicates and tlose that were ambi€+rous, too
broadly defined, not specific to qualitative reseaÍch, or
impractical to assess. ril'here necessary, the remaining items
\t'eÍe rephrased lor clariq'. Ilased upon coflsensus among the
authors, two new items that weÍe consideted relevant for
reporting qurrlitative research were added. The two new itcms
were identi$,'ing tl-re :ruthors who conducted the interview or
focus group and reporting the presence of non-participants
during the interview or focus gtoup. The COREQ checklist
fot
explicit and comprehensive reporting
studies consists of 32 criteria,
plement each item (fable 1).
r:f
qualitative
with a descriptoÍ to
suP-
Focus groups are semi-structured discussions with groups of
4-12 people that aim to explore a sPeci{ic set of issues [101.
Moderatr>rs often commence the focus gÍouP by askin€l
broad questions about the topic of interest, before asking the
focal questions. Although participants individually ansver the
t-acilitator's questions, they are cncouraged to talk and interact
with each other [1 11. This technique is bui]t on t]re notion
that the group interaction encourages respondents to explote
and clari$ individuai ancl shared perspectives [12]. Focus
groups àÍe used to exPlore views on health issues, ProgÍams,
interl'entions and research.
Methods
Development of a checklist
,fearch $rategy. §íe pertbrmed a comprehensive search íor
published checldists used to assess oÍ review <lualitative
studies, and lpidelines for reporting quditativc studies in:
Medline (1966*Week I Àpril 2006), CINAHI- 0982rVeek 3 April 2006), Cochrane and Campbell protocols,
systemàtic reviews of qualitative stuclies, author or reviewer
guidelines of major medical iournals and reference lists of
relevant publications. We identified the terms used to index
the relelant articles alreadv in our possession :rnd performed
a broad search using those search terms. The electronic
databases were searched using terms and text words for
research (standards), health services tesearch (standards) and
quaiitative studies (e valuation). Duplicate checklists and
detailed instructions for conducting and analvsing qualitative
studies were excluded.
Datut extraction.
{
o
p
À
o
r
Focus groups
exuacted
o
From each of the included publications, we
all criteria for
assessing
or Íeportir€l <lualitative
studies. Seventv-six items l«»m 22 checklists were compiled
into a comprchensive List. Wb recorded the ftequency oí each
item across all the publications. Items most frequently
included in the check.lists related to sampling method, setting
fot <latt collection, method oí data collection, respondent
o
COREQ: content and rationale
(see Tables I )
Domain l: research team and reflexivity
:
&
Í)
o
characteristics: Qualitative researchers closely
a
with the research process and participants and are
therefore unable to completely avoid personal bias. Instead
researchers should recognize and clarifr for readers their
o
(i) Personal
engage
identity, creclen[ials, occupation, gender, experience and uaining. Subsequendy this improves the credibiliry of the findinp
by giving readers the ability to assess how these factots
might have influenced the researchers' observations and
interpretations [13-15].
(ii) Relationship u'ith participants: The relationship and
extent of interaction bet'ween the researcher and their participants should be clesctibed as it can have an effect on the
participants' Íesponses and also on the tesearchers' understanding of the phenomena [16]. For examplc, a clinicianresearcher may have a deep understanding of patients' issues
but their involvement in patient c te m^y inhibit frank discussion with patient-participants when patients believe that
their responses will affect theit treatment. For transparenqi
the investigator should identi$, and state their assumptions
and personal interests in the tesearch topic.
Domain 2: study design
(i) Theoretical ftamework:
Researchers shoulcl clari§, the
theoretical framewotks underpinning their study so teadets
can understand hou, the researchers explored their research
questions and aims. Theotetical ftameworks in quaiitative
research include: gtounded theor1,, to build theories from the
data; ethnographr,; to understand the culture o[ gtoups with
shared characteristics; phenomenology, to describe the
meaning and significance of experiences; discourse analysis,
tr: analyse linguistic expression; and content analysis, to systematically otganize data into a structurecl format [10].
?
o
,A
§
7(
D
o
o
x
o
o
o
o
(r,
Eo
o
é
@
{
N
N
N
A. Tong et
Table
I
o,.
Cgnsoliclated criteria
fbr reporting qualitative studies (COREQ: 32-item checklist
No ltem
Guicle questions /description
Domain 1: Research team and teflexivity
Personal Characteristics
or focus grc.»up?
1.
2.
3.
4.
5.
Interviewerf facrlitator
Expericnce and training
Reiationship with pnrticipants
\fi4-rat experience
6.
7.
Relationship established
\X'as a
Participant knowledge of the
interviewer
lntervien'crcharacteristics
\!hat did the participants know about the re searcher?
\\hich authot/s
conductecl the interview
()ccupation
\X,hat wete the researcher'.s cteclentials? I:,.C. PhD Mt)
What was their occupation at tlre time of the study?
Gender
\)7as
Credentials
8.
the researchet male or female?
or training did the researcher have?
rel'ltionship established prior to study commencement?
e.g. pertonal goab, rea.ron.r
for
doing the
reseanh
\Yhat characterisdcs were reported about the interviewer/íacilitator? e.g. Biat, asttttttptions,
realom and interest.r in tbe re*arch topic
Domain 2: study design
15.
of non-participants
16. Description of sample
Presence
I)ata collection
17. lnterview guide
18. Repeat inten iews
19. Audkr/visual recording
20. lrield notes
21. l)utation
22. Datt saturation
23. Transcripts returned
o
a
s,
o
a
Ë
c
V'hete was the data collected? e.g. houe, clinic, workpku
§7as anyone else present besicles the patticipants and researchers?
\{'}rat :rre the important characteristics of the sample? e4. darugraphic data,
i
date
È
X
§
ril/ere questions, promPts, guides provided by the authoÍs?
Vas it pilot tested?
\('ere repeat inter'"'iews carried out? If .ves, how many?
Did the tesearch use audio or visual recording to collect the data?
Were field notes made during and/or aftet the interview or focus group?
\['hat vras the duration o[ the interviews or foctls group?
r'X/as
data s2rtumtion discussed?
Were transcripts retutned to participants for comment and/or correction?
l)ata analysis
V
o
o
6
o
o
o
3
U)
o
ró
d
of- data coders
25.
Description of the coding tree
26. Derivation of themes
27. Software
28. Participant
checking
o
N
How many data codets coded the data?
{
Did authors provide a desctiption of the coding tee?
W'ere themes identified in advance ot derived írom the data?
\X/hat softw'are, if applicable, was used to manage the data?
Did participants provide feedback on the frndings?
l.J
o
N
Reporting
29. Quotations
presented
30. f)au and {indings consistcnt
31. Clariry of major tl:remes
32. Cbnty of minor themes
findings? \rVas each
W'ere participant quotations presented to illustrate the themes
partidpant
namber
quotation identified? e1.
\X/as there consistency between the data presented and the Éndings?
/
\Y'ere mrrjor themes clearly ptesented in the findings?
Is there a description of diverse cases or discussion of
(ii) Participant selection: Researchers should report how
participants wete selectecl. tJsually purposive samPliog is
used which involves se.lecting participants r»'ho share particular characteristics ancl have the potential to provide dch, relevant and diverse data pettinent to the research question
352
{
À
tbeory,
Domain 3: analysis and findingsz
24. Numl:er
o
o
À
o
Theoretical framework
9. Methodological orientation and What methodological orientation was stated to underpin the study? eg. groantled
divorrse anallsit, ethnograplry, phenornmloglt, content analysis
Theory
Participant selectic»n
How were participants selected? ug. parposiuq conrcnience, consecutiue, Youball
10. Sampling
I-Iow were participànts apptoached? eg. face-to-iace, telepltone, mail, emuil
11. Method of apptoach
How many participants were in the study?
12. Sample size
How many people refused to PàrticiPxte or dropped out? Reasons?
13. Non-participation
Setting
14. Setting of d:rta collection
U
minor themes?
[13, 17]. Convenience sampling is less optimal because it
to caprure important perspectives from difficultto-teach people [16]. R-igorous rttempts to recruit participauts
and reasons for non-participation should be stated to Íeduce
the likelihood of making unsupported statemenLs P8].
may fail
Consolidated criteria íor rePorting qualitative research
à
É
§
aa
c.I
!
a
o
ii
cc
§,
€
c.)
èJ
F
d)
É
F:
U
.=
G-
o
,;
aaa
o
a
I
\,
e
À
o
.E
È
fiï
o
bo
.+
:6
§
o.
,9
§
€
aa
ía
a
:d
.E
:
G
I
o
:6
!sI
'n
à
U
o
ö
U
I
a
c,)
F:
N
\c
o
Í))
Eo
ë)
d
o
p
5
{N
p
Ë
o
q)
-3
P
e
-o
Eo'd
.,eii.í
ts E I
o
A
èo
«
x::
;'
Er
.È !o
a?o
jE * Ë
PË
s!"8
r
Ëp H-*
iÉze
E*r,
e.ïí.r të É F5 íE?
EiEËt:EHÈËË{ ËEfl
;EíïÈ:eàïsïfl=ËHi
o.
GI
N
(tr^O-.OuË"'É?a.:ocÉ,A0lP
(j
E
o
èl
5
o
É
öl
o
_g
.d
f-
o
o
:§
.?
(J
xE
a
o
:^
d
o
O
§
?
:in
E
o
b
aaa
: b.{'
-o
o
ts
aaa
:-
O
p
d
I
o
\
a
a
q
E
o
.o
à.
.E
5
5
0
&-.c
=U
d
o
>
o
-:
€i s ë
E *,A
È''E
; €
EËi Ë
E
E
g
o
ó
H
§,: Í E si: e EE Ui Crï É aË;; _+E b
§ !Ë E,+ië E r.U;ië iËfl §: ÉEëÉ H í§
eE,rfi À § § § É § ÀË óE,i§ ë,9 "q § H É E 3 § §
o
.D
3
o
A. Tong et ol
:h:
-à
aaa
::
.Fi
aaa
a
:Öl:
i:
.ó
'.é:
::
.ó
:.i:
aa
::
:6:
:6i
:*:
:Ë::
;-:
:r:
:*:
'\o
:.ö:
::
:*:
::i
:*;
:[.:
::
:-:
::
: i.'
:ó:
::
.tr:*:
:..:*
:-
aa
a
a
i
aaa
o
a
:
u
a
a
E
o
a
a
aa
a
a
a
a
a
aaa
aaaa
.j.
:
a
U
o
O
aa
{
=
§
a.
o
'.ir
+
o
aaa
ào
a
a
o.
|
:-j-:
a
a
aaaa
a
aaa
,!
a
ëo
|
:ro
a
a
:È
:-:.4
a
U
aa
b
a
q1
:
a
aaa
aa
:co
a
aaa
aa
i,.É
a
aaaaa
:61'
aaaaaaa
:6
D
I
-I
:§t
U
0
>-
o
: l'--
É
:c{
E
aaoaaaaa
:st
eo i.
oa
E
a
EE.
u
O
0
c.cEe3trà
!EE4a
a
aaa
o
=
h.eqc
:
u:
.:lc
'oË.
qc
ó=
o
PsE
E
o
;;
i
Éq
{
p
N
o
N
E
E
:^í
".E
F"SX.
ë-
u
Ëq9
€«§g
l.i
ri
,i6IA
o.
u
U
s
«
e
o
§
Ë
I
ó
g
E
l-
d
a
irE i'íaÍí s3Ëïai :*i,
EËEE
o
.t
-o
.d
6
öc éxó ,íF*p
;
r itLi i§
ËE
ï€q
irf
E
3T
Eàí §'
ËE-Eë
c-rE
!.98
Eq
Eq;
d*'Fui - =
-
U
o
o
a
o
E
ËË
'q
d
o
'É
c
.é
a'l
354
aaaaaaa
öt
o.
c
à
a
E
È ,.!; ;Ë 1
Í
* , $
È § :E Ee .s.:,
-o
X
§
o
o
xo
:
:t
a
Io
iÏË u §Ï ËËit *Ët ï s ÈË ï§ -{
íS˧ § gËEËEË ÈÉ gf
Ë
˧!Ë
§E EË É È
-o
À
I
c
Consolidated criteria for rePorting qualitative research
aaaaa
!
€
I
§L
o
aaaaaa
aa
aa
a
o
rr
F
a
E
o
-
a
aaaa
6
aa
U
o
o
.E
D
7
À
o
a.
o
€
aa
.s
!'i
'ë
o.
a
o
9
,!
aaaa
a
aaa
È
o
o
U
aaaaa
?
s
t
,E
p
a
3
N
X
§
o
o
aaaaaaaaa
F
o
60
cc
+.
aaaaaaaa
3
\c
d
(.)
rr
.aaaaaaaa
F
N
aaaaaaaa
CN
aa
aaaaaaaaa
I
€o
ca
o
o
à
o
!o
d
o
O
a
N
a
N
o
N
a
5'r
É
U
U
ËË
c + 6
o.
Ë
'oí.,
s§
-o
o.
öl
c!
'^
0
o
.t
(,
t-
o
o
a
i;
ó
.d
à
aa
E
o
fr_oEB8"El
EÉ à
§
u « dr-1
ÍIx
=Ë
í
'o
Ë
i
e"E h.E q Eu Ë
Ë Ë
.E
60
à" .3 Ëe íÈ
e
E
2
5
o
E
"?rH§§;«;; Ëíi x ïËe x sE : Ë
sË Eós€:&E
§9
ËË;ïíïEuui*.
*iË"t€È Ë i€":,§
BÏ ":E;.§
EÉ
i Ë E§;E ËiïiË EEitËËíË [If $Ë+ïËf àiï
Ë'ËrEÉg gË
gË
g
ËÈ § ËË Ë É i íË Ë"! ëj
ËE Ë§Ë Ë Ë
a
E
a
-ó
o.
o
o
355
A. Tong et
ol.
Researchers should report the sample size of their study to
enable readers to assess the diversity ofperspectives included'
(iii) Setting: Researchers should clescribe the context in
which the data were collected because it illuminates why participants responded in a particulat way. Iror instance, participants might l:e more reservecl and [ee1 cliserrpowered talking
participants to add ttansparency and uust§/orthiness to their
fin<iings and interpretations oí the clata [lfl. Readers should
be able to assess the consistency between the data PÍ€sented
ancl the study findings, including the both maior and minr:r
themes. Summaty Íindingn, interpretations ancl theories gcnerated should be cleaily presented in qualitative research
in a hospital setting. The Presence of non-participants during
intcrviews or focus gtoups should be reportecl às this can
also affect the opinions expressed bv participants' For
example, parent intervieu'ees might be reluctant to talk on
sensitive «rpics ií their children are pÍesent. Participant
characteristics, such as basic demogrrrphic data, should be
reported so readeÍs can consicler the relevance of the findings and interpretations to their crwn situation. This alscr
publications.
alk:ws readers to assess whether perspectives from different
gÍoups wefe exPlofed rrnd compared, such as patients and
health care proviclers [13, 191.
(iv) Data collecdon: The <luestions and prompts used in
data collection should be providecl to enhance the reaclers'
undetstandinl5 of the researcher's focus and to g.ive readers the
ability to assess whether participants wcre encouraged to
openly convey their viewpoints. Researchers should also repott
whether repeat interviews were conducted as this can influence
the Íapport de'r'eloped berween the tesearcher and participants
ancl affect the richness of data c.»btained. The method of
recording
the prtticipants' words should be
reported.
Generally, audio recordinp; and ttanscription more accurately
reflect the participants' views than contemporaneous
tesearcher notes, more so if participants checked thcir own
transcript for accuracy 119-211. Reasons íor not audio recording sl-rould be ptovidecl. ln addition, field notes maintain contextual detàils ancl non-verbal expressions for data anallnis and
interpretation Í19, 24. Dutation of the interview or fbcus
group should be reported as this affects the amount of data
obtained. Researchers should alsr: clari§' whether ParticiPants
were tecruited until no new relevant kncxvledge was being
obtained from new participants (data saturation) 123,24).
Domain 3: analysis and findings
(i) Data analysis: Specifying the use oí multiple coclers or
other methocls of reseatcher triangulation can indicate a
broader and more complex understanding of the phenomenon. The credibility of the findings can be assessed if the
process of coding (selecting significant sections from participant statements), and the derivation and identification of
themes are made expiicit. Descripdons of coding and
memoing demonstrate hour the researchers perceived, examined and developed their understanding of the clata [17, 191.
Researchers sometimes use software packages to assist with
stoÍage, searching and coding of qualitative data. In addition,
obtaining feedback from participants on the research findings
adds validity to the researcher's interpretations b,v ensuring
that the participants' own meanings and perspectives are
represented and not curtailed by the researchers'own agenda
ancl knowledge [231.
(ii) Reportjng If supporting quotations are provided,
researchers should inclucle quotadons from diÍferent
356
Discussion
The COREQ checklist was developed to pmmote explicit
and comprehensive reporting of quaiitative studies (interviews and fircus groups). The checklist consists of items
specific to reporting qualitative studies and precludes generic
criteria that are applicable to all rypes of research rePorts.
COREQ is a comprehensive checklist that covers necessary
components of srudy design, which should be reported. The
criteria included in the checklist can help researchers to
report important asPects of the research team, study
methods, context oí the studl', findings, anaiysis and
interpretations.
At present, we acknowleclge there is no empitic basis that
shows that the intoduction of COREQ vill improve the
qualitl,' of reporting of qualitative research. Hovrever this is
no difFerent than when CONSORÏ QUOROM and other
reporting checklists wete introduced. Subsequent tesearch
has shown that these checklsts have impnrvecl the qualiry" of
reporting of study types relevant to each checklist 15, 25),
and we believe that the effect of COREQ is likely to be
similar. Despite differences in the obiecdves and methocls of
quantitative ancl qualitative methods, the undedying aim of
transparency in research methods and, tt the least, the theoredcal possibility of the reader being able to duplicate the
study methods should be the aims of both methodological
approaches. There is a perception among reseaÍch funcling
agencies, clinicians and policy makers, that qualitative
research is 'second class' research. Initiatives like COREQ
are designed to encourage imprc.rvement in the qualit,v of
reporting of qualitative studies, which rvill indirectly leacl to
improved conduct, and greatet recognition o[ qualitative
research as inherendy equal scientific encleavor compated
with quantitative research that is usecl to assess the quality
and safetv of lrealth care. We invite readers to comment on
COREQ to improve the checklist.
References
1. Mohcr D, Schulz
K§
z\ltman D. The CONSORT statcment:
revised recommenclations for improving the qualitv of reports
of parallel-group rznc'lomized mals. J4t\L4 200 1 ;285 : 1 9 87 - 9 1.
S rl a/. Imptoving the tluality of
meta-analyses of randomised conttolled trials: the
QUOROM statement. Quality of Reporting of lvleta-analyses'
2. N{oher D, Cook DJ, I'iastwood
reports
I -ann
t
of
1.999 :354 :1
896- 900.
o
4
o
p
À
o.
+
o
g
€
a,
o
a
a
3
G
È
X
p
o
6
F
o
@
o
o
U'
Eo
o
d
@
{
N
N
N
Consolidated criteria íor rePorting qualitative research
3. STROBII Starcnrent: Streng'thening the reporting ot obserrztional stuclies in epidemiologir http://u'stt*strobe-statement'
23. PopavJ, Rogers A, \!'ilÍams G. Rationale and stanclarc{s fot the
systemàtic reviev' of qualitative literature in health services
research.
org/Checkliste.html
4. Stroup f)f, Berlin JÀ, NÍotton SC et al. N{eta-analysis of
obser-
24. Blumer
proposal tbr reporting'
vational studies
Meta-anall,sis Of Obsetvational Studies in lipidcmiology
O{OOSE) grory. IAA'IÀ 2r)00;283:2008 - 1'2.
in
5. N{oher
f),
Jones
ep.idemiolngy:
À,
I-epagc
Lr,,
Jaclad ÀR, Ross C
2005;58: 1081
-
Cit
5
Citiqrte.r oJ' Rescarcb,
I
in the Sodal 'lciencet'
Carc 2005:9:575-82.
26. Críttcal Skills Àppraisat Programue (CASP) 10 ()uestions to
help .vou make sense of qualitative research: lV{ilton Keynes
Prirnarl. Care Trusg 2002.
of qualitative
stuclies exploring parental beliets and attitucles toward chilclhoocl vaccination identiÍiecl common barriers to vaccination'
I Clh Epideniol
-
25. I)elane-v À, Bagshau, SN{, Ferland L et al. A systemàtic evaluation of the quà[ty of meta-anl,vses in the critical care literature'
l-. tJse of the CONSORT
a/ a/' Systematic review
H.
Rer 1 998;8:341
New Bruoswick, NJ: Transaction Books, 1979.
a
Statement ar"rd <luality of reports «rf randomized trials' A comparative bef<rre-and-al-ter eva.luation, J,'1M'7 201)7;285:1992- 5'
(r. i\,tills
Qual Íl ealth
1,, Ritchie J, kwis J et dl. StMliA in ptalitatitte F'ruhratirtn:
Wanavork Jàr Asstssing l\enarch Ët'idence. Lnndon: Cabinet
()f6ce. Government Chief Social Researcher's Of6ce, 2003.
27. Spencer
A
8.
7. KnaÍl KÀ, Lloward NJ' lnterpreting and teporting qualitative 2g. Campbell R" pound I] pope C rr al l:i'a,uating meta-crhnogaaphy:
resr:arch. Ilt.t Nrn Hulth 1984;7:7*74.
à synthesis of qualitative'research on lay exferience of cliabetes
anddiabetes care.Soc.lcilÍed2003;56:ó71-84'
8. Sofaer S. eualitatiye research methrxls. Int J Qrat Llealth Care
2tJ02;74:329-36'
29. Feder GS, Hutson N{, Rrmsay I r/ a/. \lbmen exposed to intiwhen they
9. Vright Àts, Flolcombc C, Salmon P Doctor's communication mate Parmer violence: exPectations ancl experiences
of qualitameta-analysis
a
professionals:
health
carc
encounteÍ
stuclyr
qual.itative
cancer:
in
brcast
respect
an6
care,
of irost,
tir.'e studies. .Arcb Intem Med 2006;166:22-37 '
Bi\ll 2O04;328:864*8.
10. Uamputtcrng p, Ez.zy Í). puatitatiue Rt.warch Mdltod.r. 2nd
Nlelbàurne, vict.ria: oxfor--d Universitl' press, 2005.
Groups. ;1
edn.
30. Pound
thesis
Ilocas
Grotp:
as
Prat:tical ()ttide
pualitatircRz.naal:.
California: Sage, 1988.
Britten N, i!{otgan I\'t rlal. Resisting medicincs: a s1'n<lualitative studies of medicirle taking Jar sci Med
2005Í1:t33--ss
.far
2000. 31. Smith I-K, Pope
Newburl, Park,
;iit,at#tÏ:ïer
11. Krueger RÀ, Case,v N,1.. Foatt
Apptíd Rueaoh. Thousancl Oaks CÀ: §age Publications,
12. N{organ DL.
P,
of
14. I\{dterud
K.
lines. I.trcet
Quatitative research:standards challenges
ritera-
guide-
2001;358:483-8.
17.
/
I;ant Pract
In,i TS, Frankel RM. E'aluating the quality of
1995i41:279-85.
L, Turlpon
18. Altheidc D, lohnsonJ. Criteda for assessing intetpretive validiq'
irr clualitative research. ln f)enz.in N, I-incoln Y (eds). I-Iandhook
oJ'
Qnalilatite Ilesearcb.
"Íhotstncl Oaks CA: Sage Publications,
1y)4.
20. Seale C, Silverman S. linsuring rigour in qualitative reseatch.
I::ar J Pablic Healtb 1991 ;7 :37 9 -84.
21. Scheff'f
Single case analysis in the health sciences. L)ur.l Ptblic
Healtb 7995;5:72-4.
R. Iri,valuating qualitetive research.
1997:5:232-5.
22. BlutÍ
3(r. Hoddinott R Pill R. À review of
research in general ptacticc. More methodological
tecently published qualitative
quest.ions
than answers? Fa» Pmct 1997;74:313-9.
37. Kuzel ÀJ, Iingct lD, Addison RR et al. Desirable feanrtes of
19. Fossey Ii,, Í{arvey C, i\'{cDermott F el r/. Understanding and evaluating qualitative reseatch. Àast N Z J Pgchialry 200236:777 -32'
Br .l
Mid»ifet1
À
+
o
5
g
e
9
o
3
clualitative research. l;an Pract Ra
J 1994;74'.369-18.
Pl. et al. Crirical apptaisal
qualitative research s:r:udies. lndian J Pulialr
38. Treloar C, Champness S, Sirnpson
checklist
for
2000.67:347
-5\.
39. Cesario S, 1\Íorin
K, Santa-Donato À. Iivaluating the level of
evidence in qnalitative research.
2001137:708-14.
J
Accepted for publication 7 July 2007
È
xÈ
@
f,
qualitative
35. Dixon-Woocls N.t, Shaw RL, Àgarwal S eí at. 11-te problem of
appraising qualiative research' Qtal .ïaJ' Heahb Carc
in
articles
research
qualitative
Àppraising
.1.
2004-73:223-5'
A4ed'Ibaclt2O05;27:71-5.
nredicineandmedicaleducation'
Ca1te
o
§
o"
o
m
research: a proposal pr<i-term. rl Gin lnten Med 19g1;6:485-6.
15. Llays N, Pope C. Qualitative tcsearch in health carc: asscssing 34. Boulton lr{, Fitzpatdck R, Swinburn C. (]ualitatir-e research in
<1ualit1, in qualitative reseatch. BMJ 2000;320:50-2.
health care: II A structurecl review ancl evaluation of stu4ies.
1(r. l:ilder NC, William 1.. Reoding and evaluating qualitative .l .EtalClinPraa1.996l'2:171-9.
tesearch studies.
{
?
t'
33.
o
a
C, Botha JL. Patients' hclp-seeking cxperiences
ptesentation: a <lualitative s'\'nthesïs' Idncet
xiffik';ï:ï*:Ï':ïïi[Ï'í:i1J'ï,ïiïï::'Ï:,is
or quaritative research' Ann bitn Med 2004;
'n"
"'"
ÏÏ,;Xï1Ï.Ï::Ïm,ï.,#r,firïïiir;:ï;.^'
llrlrïif""
13. Giacomini N{ri, cook D.}. users'gui<ies to the mecricar
U
Obne, C)ytecol Neanatal Nzrr
o
o
o
v,
o
.6
o
d
o
{
p
N
o
N
Palliatiue and Supportiue Carc (20L8)' 11, 373-382.
O Cambridge University Press, 2012 1478-9515/12 $20.00
doi: 10. 1017iS147895 1512000284
ORIGINALARTICLE
Exploring perceptions of psychological services in a
children's hospice in the United Kingdom
JO WRAY,I BRUCE LINDSAYz KENDA CROZIER,2 LAUREN ANDREWS,o
JANET LEESONs
AN»
lGreat Ormond Street Hospital for Children NHS Tlust, London, United Kingdom
2University of East Anglia, Norwich, United Kingdom
sEast Anglia's Children's Hospices, Cambridge, United Kingdom
(Rrcslvro Februaty 17,20121' Accopmo March 15, 2012)
ABSTR.ACT
Background.: The provision of emotional and psychological suppol! for all tamily members who
need-it is an essential element of holistic palliative care. Within East Anglia's Children's
Hospice, teams of professionally trained and experienced workers offer psychosocial support to
ail fàmily members at all times during the child's and family's journey. However, theeffectiveness and appropriateness ofcurrent psychosocial provision is unclear, as is the
requirement for any additional psychological services.
Óbjectiue: The purpose of this siudy wàs to elicit perceptions about current p_sych-ological
suppórtwithin tËe hóspice from a group ofstakeholders (parents, hospice staff, and external
professionals).
- Method: Forty-five parents participated in family focus groups, telephone interviews,_
individual interviews in theirhome, àr a web-based survey. Ninety-frve hospice staff(including
nurses, carers, play specialists, therapists, and family support practitioners) and 28 external
staff (including physiclans, nurses, and commissioning managers) were seen using a mixture of
focus group and individual meetings. Focus groups and meetings were held at the hospice
building oi at an external venue. Interviews were recorded and transcribed verbatim and
analyzed using thematic coding.
Ràsults: Tfdmain themes adàressing perceptions of current psychological provision emerged:
"understanding psychological support" and "unmet psychological need." Subthemes linked to
support includéd choice, staff rolèi and labels, communication, and flexibility, whereas the
thómes within unmet need had a stronger focus on people and problems.
Significance of results: Understanding different userperspectives is an important firrst step in
enhàncing current psychological provision; operationalizing the findings will be challenging.
KEYWORDS: Psychological support, Children, Perceptions, Hospice, Family support
Every life-limited child or young person in the UK,
regardless ofrace, religion, age or where they live
has access to sustainable, holistic, family-centred
INïB,ODUCTION
The vision of the Association for Children's Palliative
Care (ACT) and Children's Hospices UK (CHUK) is
and high quality palliative care (ACT/CHUK,
that:
2009)
within this vision is the provision of
psychological services as constituent elements of
each stage of the ACT pathway, supporting the inclusion of psychological services for young people
Encompassed
Address correspondence and reprint requests to: Jo Wray,
Centre for Nursing and Allied Health Rcsearch, Great Ormond
Street Hospital for Children NHS Tlust, Great Ormond Street,
London WC1N 3JH, Unitcd Kingdom. E-mail: jo.wray@btopen
world.com
373
WraY et al.
374
and their families living with life-limiting and lifethreatening conditions outlined in Aiming High for
Disabted Chitd,ren (DH, 2007) andBetter Care: Better
.Liues (DH,2008).
Psychology has made significant contributions to
the study of death, dying, palliative care, and bereavement, not only from an academic perspective
but also from assessment of, and interventions
with, patients receiving end-of-life care and their fa-
milies (Haley et al., 2003). As the emotional support
of such patients and families is the shared responsi-
bility of all involved professionals (Crawford,2004)
psychologists also have a role in facilitating the delivery ofeffective support by/through others.
Adult patliative care has been the focus of much of
the research into the psychological impact of life-lim-
iting/life-threatening conditions and their treatment, particularly for patients with cancer, with
considerably less attention being given to children,
and in particular children with complex needs arising from disability or illness, or a combination of
both, where families are living with a condition
over the longer term. Similarly, clinical psychological
services have greater prominence in adult palliative
care, as is evidenced by the British Psychological Society's (BPS) publication of a framework for psychologists working with adults receiving end-of-life care
(British Psychological Society, 2008). No equivalent
guidance exists for those working with children receiving end-of-life care. It is also evident that within
the adult population there is inequality of access to
palliative care services, with the OÍfice for National
Statistics reporting in 2005 that 957o ofpeople in palliative care units had a diagSosis ofcancer. This has
implications for the caseload of professionals working within those units. \Mithin pediatric palliative
care cancer has also been the area in which most
psychological research and clinical energies have
been focused, with a wealth of studies documenting
the psychological concomitants of this disease and
its treatment for children and their families, both in
the short and longer term. However, there is a broad
range and complexity of childhood pathologies lead-
ing to palliative care requirements. Furthermore,
as advances continue to be made in the medical and
nursing management of children and young people
with complex health needs, increasing numbers
will live longer and the numbers and complexity of
such patients accessing palliative care services will
continue to grow, with repercussions for a1l involved
professionals.
For the majority of children and young people
with
life-limiting/life-threatening conditions who also
have complex health needs, it is parents, and particularly mothers, who are the main carers. The amount
of support received from statutory and voluntary
agencies varies, but it is evident that the strain on
a family resulting from looking after such children
can be considerable (Brehaut et al., 2011), with elevated levels of stress, distress, and exhaustion reported (Yantzi et a1., 200?; Rodriguez & King, 2009).
However, it is also important to recognize that such
stress is often in the context of positively valuing
the child and the caring role (Stainton & Besser,
1998; Oulton & Heyman, 2009). Frequently families
become isolated from their friends and extended family and this lack of support can further increase
stress levels (Carnevale et al., 2006). Even when sup-
port is available, some parents will not
access
it
be-
cause of their concerns about the risk of emotional
or physical harm to their children from doing so (Oulton & Heyman, 2009). The costs of raising a child
with disability (Sloper & Beresford, 2006) and the
need for specialized equipment to care for such a
child can result in parents feeling like prisoners in
their own homes (Brinchmann, 1999) and the caregiving responsibility can extend to other family mem-
bers, including siblings (Roberts & Lawton, 2001;
Heaton et al., 2005). Furthermore, the children
themselves can become isolated as care becomes focused on their medical and physical needs, some-
times to the detriment of their emotional, social,
and spiritual needs.
With the introduction of palliative care services to
a child and family there is the opportunity to alleviate some of these problems. Elements of care may
be delivered in a range of settings (home, school/college, hospital, hospice) and professionals, including
psychologists and others providing psychosocial support, may work at any of these locations' Evaluation
of palliative care has focused primarily on assessing
aspects such as the impact of short break/respite
care on parental stress (Sherman, 1995), access to,
and satisfaction with, services (Maynard et al.,
2005; Eaton, 2008; Grinyer et a1., 2010), and parental
perceptions of bereavement services (Davies, 2005;
Wilkinson et al., 2007; Agnew et al., 2010)' ÏVhat
has not been well studied to date is the need for, or
impact of, specific psychological interventions with
children and/or families. Although the consensus is
that psychological support should be an integral
part ofpediatric palliative care, the lack ofrobust outcome measures and the widespread use of nonvalidated satisfaction measures as tools for evaluating the
effectiveness ofinterventions have hampered the development of an evidence base. Furthermore, family
and staÍÏperceptions ofpsychological need within the
hospice framework have not been well studied.
In 2010, the Department of Health made funding
available for service improvement initiatives in children's palliative care services in the United Kingdom.
One ofthe funded projects focused on the provision of
375
Perceptions of psychological seruices in a hospice
psychoiogical services in a children's hospice (East
Anglia's Children's Hospice [EACH]), with the aim
of reviewing the way supportive care is currently delivered, and for this to inform the development of a
model ofpsychological support that assesses need in
a consistent and systematic way using evidence-based
techniques and tools, identifying and targeting areas
ofassessed need, and providing appropriate levels of
intervention to meet that need GACH, 2010). The
findings concerning perceptions of existing psychological support and the need for psychological services are reported in this article.
METHOD
of the evaluators involved in this study were independent ofthe hospice and were unknown to any of
the participants prior to the start of the study. Al1
had a background in health and were experienced researchers. Three of the evaluators (JW, BL, and KC)
were appointed as individual project leaders for this
and two other Department of Health projects; JL was
appointed as the project manager for all three projects and LAwas the project assistant.
Participants who had used or provided services
were recruited: families, external staÍf, and internal
staff The hospice cared for 464 families in 2009 (including bereaved and prebereaved), with increasing
numbers of end-of-life referrals being received as a
consequence of the introduction of local neonatal
and oncology pathways. The hospice offers both inhouse and community services covering a wide geographical area across four rural counties. StaÍïtend
to be based in one of the three localities, with relatively few working across sites. Links with local services vary between localities, as does the provision of,
and access to, community based National Health Service (NHS) services.
Families were invited to participate via the organ-
All
izational newsletter, circulated to -350 families
across the region, and through fliers and posters in
the hospice buildings, and at hospice-run events.
A]1 families who accessed the services of EACH
were eligible for participation. Families were given
a choice about whether they wished to participate
in a focus group, individual interview, or online survey. Hospice staÍï were invited to participate via
email (sent to all hospice care staÍÏ) or by direct con-
tact with the project manager, and external staff
were contacted via the local palliative care stratery/steering groups.
All participants received information about the project and were asked to complete a consent form' AII
were made aware that participation was voluntary
that they could withdraw at any time, and that data
collected for the evaluation would be anonymized
and d.estroyed on completion of the project. At the
start of each meeting clarification was provided about
the project brief; the role and professional background
ofthe facilitator(s); and the plan for the transcription,
analysis, and dissemination of the findings. For the
three focus groups involving parents, lunch was provided and parents were offered the opportunity to
have their children (ineluding sibiings) looked after
by hospice staÍf.
The interviews were undertaken by either one or
two of the authors (JW, BL, and/or JL) and used a
guided conversation technique. The question schedules were developed by two of the authors (JWand
JL) and before use were sent to and agreed to by
the chair of the Hospice Family Forum (family schedule only). Questions were developed drawing on the
outcomes of previous satisfaction survey§ for families, staff, and external staff and expectations from
the project brief. Participant experiences and expectations of psychological services were explored and
participants were asked to identify any gaps in service provision. Focus gïoups were held either in a
quiet room at the hospice or at a venue independent
of the hospice; individual interviews were held in
the hospice, at the participant's home (for parents),
or at the participant's place of work (for external professionals). Focus groups lasted from 1 to 2 hours and
individual interviews from 45 to 90 minutes. Individual and telephone interviews were conducted at a time
convenient to participants, which sometimes involved
evenings. Focus groups with parents were held on
the weekend to increase the likelihood of participation, whereas meetings with hospice staff and external professionals were held during the working day.
Thirty-seven parents participated in one ofthree
family focus groups (n: L1'), telephone interviews
h: 11), or individual interviews (n : 15) in their
home. A further eight parents completed the online
survey. Ninety-flve hospice staff (including nurses,
carers, therapists, and family support practitioners)
and 28 external staff (including physicians, social
care staff, nurses, and commissioning managers)
were seen using a mixture of focus group and individual meetings.
Analysis and Reporting
All participants agreed to have their interviews recorded, with the exception of two meetings at which
detailed notes were taken. Interviews were transcribed and analyzed used basic thematic coding to index
and manage the data, and the main themes were further coded and clustered into subthemes. The analysis was undertaken by the first authoq who read and
reread the transcripts to become familiar with the
data. The first and last author agreed the thernes.
WraY et al.
376
As the evaluation was small and related to a com-
paratively limited group of staff and families, illustrative quotes from participants have only been
used if their source was able to remain anonymous
and confidentiality was not compromised. Quotes
have been identified by participant type and meeting
number. Specific participant demographic data were
not collected for this evaluation, other than broad
categorization of participant type.
over (thereby removing any sense of choice), such
as end-of-life moving into post-bereavement care.
This has been and continues to be a source ofdistress
for some, particularly in those situations in which individual staffhave had a longstanding relationship
with a family. In some cases the transfer of support
was perceived to be very abrupt and the need for a
more gradual transition was highlighted by both staff
and families.
RESULTS
Roles and Labels
In this section, quotations are juxtaposed with dis-
The EACH model of care is focused on a holistic
vision, and within that framework the need for
psychological care is currently addressed by avariety
of professional groups. However, there is a lack of
clarity about who does or should be undertaking
this and what constitutes "psychological support."
Similarly, families have varying expectations and experiences ofhow their psychological needs are - or
are not - addressed and by whom.
Within the current organizational structure of the
hospice, the specialist interventions for providing
psychological care are delivered by the family support teams (FST) but, as indicated by both parents
and staÍï, nursing staff from the care teams have a
crucial and valued role to play in supporting families.
There is, however, some confusion surrounding the
distinction between "support" and "therapy'" In
some situations, families clearly wanted "support"
in terms of some practical help and a "listening
ear," rather than the more therapeutic intervention
that they were offered:
cussion to illustrate the themes from the perspective
of the parents, hospice staÍf, and external professionals. TWo main themes addressing perceptions
of current psychological provision emerged: "understanding psychological support" and "unmet psychological need."
Understanding Psychological Support
Choíce
Choice is one of the tenets of the EACH model and includes giving families choice with regard to the type
of support they would like and where they would like
it. Families appreciated this choice when they were
able to exert it and also felt that having choice resulted in them not feeling pressured about which aspects of the service they used. However, when
choices could not be accommodated, disappointment
and feeling that their needs had not been met could
result:
and regs that we as parents don't really know
That is where for me personally Family Support
would have been really good [help with practica] issues at time of deathl. I didn't want therapy
[Parent, 20I.
lParent,
Sornetimes I think it is difficult to access things
and whether that's a communication thing or rules
'Whereas many parents developed a good rapport
with particular staff, some did expre§s diÍficulty engaging with specific individuals and wanted more
choice about whom they could see. The perception
of some staff was that the families did not have a
choice about whom they saw because ofthe rigidity
of some of the roles (discussed later):
Sometimes we need to respect what the family say
they want. We're letting them choose where they
receive care; we should let them choose who they
wish to share things with IEACH staff, 32].
Both families and staffexpressed dissatisfaction with
situations in which one professional, such as a nurse,
had to stop supporting a family and another staff
member, such as a family support practitioner, took
181.
The approach to clarifying the individual professional
specialities of the FST to families varies, with some
choosing to operate entirely under the generic term
of family supporb practitioner (FSP), whereas others
provide more information regarding their own
specialist training. This did cause some confusion
for families and in some cases the perception was
that it had been detrimental to the therapeutic re-
lationship, with families expressing uncertainty
about whether an individual FSP was "qualified" to
offer a particular therapy, as the following quote
from a parent shows:
Now
I don't know that they're trained counsellors,
I don't know that they're BACP accredited, I
don't know what their clinical backgrounds are.
lParent,
181.
Perceptions of psycholagical seruices
377
in a hospice
Nurses and care assistants also described some tensions around, "being allowed" to support families.
Although they may not provide psychosocial support
as a formal part of their job, they do offer this informally, and this is something that families appeared
to value and need, and that the staÍf themselves considered to be part oftheir caring ro1e. Their input was
focused around the fact that they were there when families needed them, offering spontaneous supportive
care in contrast to the more structured sessions provided by other professionals:
\You've given end-ofJife care
in the Hospice,
during the night shift Mum and Dad want to talk
to you, you can't very well tell them "you have to
wait" [EACH staff, L1].
Some staff mentioned feeling disempowered by the
rigidity of the existing model in defining roles:
We've got carers who've got a lot of skills in that
area [psychosocial support] and they feel pushed
aside, excluded from being able to do that, told
"that's not your role" IEACH staÍf, 32].
Similarly, other professional groups of therapists
who work at EACH, but are external to both the
care and FSP teams, see themselves as providing informal psychosocial support:
Everything you do you are providing psychosocial
support TEACH staff, 321.
These professional groups identified the importance
of being able to provide
"psychosocial support"
when and where it was needed, with flexibility in
terms of who provided that support, but at the
same time recognizing that where specialist interventions are indicated staff should be directing families to providers of those interventions. Such an
approach was also endorsed by families, some of
whom saw EACH professionals external to the care
or FSP teams as their primary source of support.
A further issue related to staÍï roles was the blurring ofprofessional boundaries as the range ofstaff
backgrounds diversified and inclividual staff groups
took on more work. This was mentioned in connection with staff practices:
It's very difficutt for children, isn't it' . .. If they\e
got an emotional need ... well "where do I take
this then, do I take it to that person? Do I take it
to that person?" If they are doing similar things
then I think that's confusing TEACH sLafï,2).
Ctarifying the roles of professionals from different
professional backgrounds and having increasecl insight into what individuals do within a specifrc professional group will become increasingly important
for service users and service providers as EACH extends and develops its services.
Effectiveness and Appropriateness of
Services
The provision of a diverse range of services is a
strength of the current EACH model, with both "support" and more specialized "therapy" being constituent elements of the EACH package of care' It was
clear that some families wanted support in the form
of practical help and having someone there:
- sit down; I'll make you a cup
to me that's psychological [Parent,43l.
Just someone saying
of tea
-
Others wanted more therapeutic input - "I guess I
could say [I wantl more of a therapy support" ltele-
phone surveyl. However, some families felt that
they were being given a service which they did not
want or did not think was appropriate for them at
that time, and one staff member commented "They
don't want a professional psychosocial chat" TEACH
staff, 101. Many families do not choose to access family support services from the hospice but instead
just use the respite and day care facilities, whereas
for others the focus of their hospice experience is
the family supporb element - "The thing that drew
me to [the hospice] was the Family Support" [Parent,
33]. Others saw the care provided by EACH as more of
a package,
and accessed both the respite/day care ser-
vices as well as the therapeutic elements, seeing
EACH services as "an absolute lifeline" [Parent,33J.
Flexibility and its Impact on Care
Psychological and supportive care was provided in a
variety of locations (hospice buildings, home, hospital, school) and to a range ofservice users (children,
siblings, parents, other extended family members,
I'm finding that the boundaries are actually getting quite close now, and that's something that
IVe got to be quite aware of [EACH staff, 2].
and sigaificant others). Flexibility, both in terms of
where and when the services were delivered and
the way in which care was provided, was a recurrent
theme throughout many sessions, with both negative
As well an being issue for staff, the merging of professional boundaries was also seen as having direct
implications for children:
and positive views being expressed. As the following
quote illustrates, staÍï could see that flexibility may
come with a price in terms of what the service can
WraY et al.
378
offer, with this staffmember expressing the difficulty
of offering a specific piece of therapeutic work with a
family when they are also required to undertake
other types of activities with other families that
might interfere with the therapeutic session:
do question this notion of how flexible the service
is, from a therapeutic point of view. . .. actually the
benefits therapeutically are being compromised by
the level of flexibility that we are meant to lwork
I
tol [EACH staÍf 1].
offer "psychosocial support," and are working at least
at level 1, with some staÍÏworking at level 2 or 3. However, specialist mental health interventions (level 4)
are not currently being provided. One of the objectives of EACH is to develop as a specialist provider
of children's palliative care services, and the provision of some psychological services at level 4 would
impor[ant contribution to achieving that aim. A
different issue, however, is whether it should be a
psychologist providing this service and whether it
be an
should be provided by EACH or referred to the statu-
A further issue with flexibility and delivery of services concerned the geographically harder-to-reach families, (some may be 90 minutes from the nearest
hospice location), where travel times for either the family or staJf are at the limits of what is practical' A
number of staff groups talked about the time to get
to some families and external professionals also commented that for some families the hospice building
was too far. For some families, distance precluded
them from attending some of the groups' Different
approaches to psychological support ofthese families
need to be explored.
Communication
Whereas some families felt well informed about what
was available and knew how to aceess support services, others were less clear about what different therapies could offer and how to get referred for specialist
interventions. A number of external professionals
also expressed a lack of knowledge about what EACH
offered with regard to specific therapies and services:
"I might not be clear what the psychosocial support
might be" [External professional, 161' Some also commented on a lack of awareness of the professional
training of the FSPs: "I don't know if they have the
knowledge, skills and expertise for the parents
they are struggling" [External professional, 521.
Clinical Excellence, z}O[),all staÍf working at EACH
if
Unmet Need
Both families and staff identified areas of unmet
need, both directly from their o'rvn perspective and
also indirectly (e.g., parents identifying the need for
psychologicai input for staÍï to help them support
their children more adequately). IJnmet need was described in terms of the people who might meet the
need and the areas in which additional input was required.
The Need for a Psychologist
In accordance with National Institute for Health and
Clinical Excellence (NICE) guidelines for supportive
and palliative care (National Institute for Health and
tory sector. How the service is structured and the
specific skills that are required are also areas that
need to be addressed.
Views differed as towhether there was a need for a
psychologist at all, whether such a person should be
employed by EACH or be externally based with contracted sessions, and the extent to which that person
should work with the children and families and/or
with the staff. Some external professionals felt that
EACH met the psychological needs of families very
well and that a psychologist would not be their priority
for service development. Others, however, saw that
there was a need for children and families to have better access to psychological support and that this was a
gap in the service, with some external professionals
clearly looking to EACH to provide psychology input'
What is needed is a trained psychologist - e.g' a
child might be cognitively normal but as they go
through life they have issues they need to deal
with. . ..not a lot of support for them. . .'it's hard to
find where they can get that help, support and understanding to work through the issues as a family
or as an individual [External professional, 52].
The gap in psychology services was recognized by
EACH staffand external professionals as being an issue in statutory services, not just EACH, with recog-
nition of the difficulties of accessing
psychology
support through Child and Adolescent Mental Health
Services and/or Children's Community Health Services. Some parents also commented that they had
asked for a referral to a psychologist but had not
seen anyone. Other external professionals identified
that some children had access to their own team's family support service and in some cases children, particularly ifthey were oncology patients, had external
psychology provision. Nevertheless, the value to external organizations of EACH having a psychologist
in their team was also identified:
EACH could have a fairly senior psychologist
who could work with our psychologist that would
be really good lExternal professional, 621.
If
379
Perceptions of prychological seruices in a hospice
EACH staff identified that as the expansions in
If we don't pull together there is a risk that chii-
symptom management, end-of-life care, and bereavemenl were happening, the psychosocial side should
their support addressed [EACH staff, 17].
dren and families may only have some aspect of
be expanding too:
They're full to the brim and can't offer as much as
think could be useful to some families
I
Managing Challenging Behavior
TEACH
A number of children whom EACH looks after have
Whereas it was evident that further therapeutic support would be valued, participants differed in what
input to manage behavioral issues was a recuïTent
theme throughout staff and parent sessions' Currently staÍï may refer to external agencies and will
also work with behavior teams at school or in the
community, but there was recognition of the potential
staff,4l.
Skills
they thought was needed. For example, one staff
member suggested, "It would be great if the Hospice
employed a part-time family therapist" TEACH
staff, 1l whereas another participant commented "I
question how we assess that actually a family therapist is needed or whether actually a family counsellor
is needed" IEACH staff, 11. It was also recognized
that the important factor was for support to be as holistic as possible.
challenging behavior and the need for some specialist
beneflts of having someone at EACH who could help
them recognize and work with complex behavior
issues.
In some situations with an older child with particu-
larly challenging behaviour, staff could feel anxious
about and untrained to deal with the behavior:
It makes you feel particularly on edge. . .'it's like
working blindfolded, you don't know what's going
to happen...It affects the way you care. A behaviour management programme for some of our children would make me feel secure and I would feel
The hospice offers holistic care and as part ofthat
holistic care the psychological aspect needsto be an
important aspect of that care' And in order for a
child to have the best care that they can possibly
have, the mental wellbeing not only of themselves
but of their family is going to have a huge impact
Similarly, parents also spoke of the need for some
TEACH staÍf,91.
help managing their children's behavior:
Other specific therapies mentioned by both staffand
families included cognitive behavior therapy and
psychotherapy, although it was acknowledged that
even if a therapist with one or more of those specific
qualifications was employed and an individualwould
benefit from one of those approaches, EACH might
not be the best or most appropriate agency to meet
the needs of everyone. In particular, it was felt that
some mental health needs might be better addressed
by other external agencies rather than the hospice.
I'm wondering about cognitive behaviour therapy,
maybe.
...whilst being holistic is a positive
to get sidetracked
you...need to be careful not
into stuffthat maybe another agency would be better meeting those needs. . ..IEACH staff, 11.
The complexity ofthe different systems in which children and families receive care and the relationships
between them were hightighted by staffand families.
Systemic working and helping staffto think systemically were identified as key components of a psychol-
ogy role. Good communication and collaborative
working with other staff were perceived by staÍï as
being vital:
that I would be giving a better level of
care
TEACH staÍï,391.
OÍïen his behaviour is awful. ' .and it's one of the
goes. . .Someone who could look
at. . .the complex health needs but also you've got
the behaviour side of it [Parent, 20].
first things that
Another parent talked about her child's erratic behavior when he became frustrated because he could
not communicate:
There's nowhere to take that. It can be very' very
stressful. . ..it would be useful to have a behavioural specialist on board lParent' 33]'
Worhing with Children and Young People
Although the emphasis on the value of a psychologist
was primarily identified in relation to parents and
staff, together with a liaison role with communitybased professionals, some staÍï did identify the need
for a psychologist to work directly with the sick child.
The comment was made that children rarely talk to
nursing staÍï about dying; howeveq this was recognized by staÍïas something that must be frightening
for them. Some anxieties about whether nursing stafl
should be talking to children about dying were also
WraY et al.
380
voiced. One member of sta{ï described being "put on
the spot" by a child:
. . .there was a little girl in for end-of-life who came
out of her room and wanted to know what would
happen to her. . ..I wasn't expecting this, she just
said "what will happen to me?". Firing all these
questions at me. That was tricky. That's out of my
comfort zone IEACH staff, 11,].
In the same meeting another staÍÏmember described
such situations as ". . .very challenging because that's
not our remit".
Working with Families
In addition to the specific area of behavior management, other areas were identified in which a psychol-
ogist would be a valuable resource for working
with families, including facilitation of adaptive coping with the concept of a hospice and the ramifications of that. As one parent said, "Hospices, you
think of them as a place to die and not very nice"
lParent,43l.
Comingto terms with the diagnosis of their child's
eondition and their prognosis, diagnosis ofpsychopathology in parents or other family members and the
implementation of specialist psychological interventions (Level 4), issues ofparenting, and psychological
interventions in the management of complex grief
were also identified as areas in which a psychologist
could work with families. As a parent said,
Sometimes it would be useful to see someone - to
know if you are handling something in the right
way. I have asked for a psychologieal referral but
there is no-one out there [Parent, telephone interview 31.
Another parent commented:
I think EACH should try to incorporate more sensitivity about living with the threat of death and how
it impacts on everyone, dealing with the physical
pain and difflcult emotions, dealing with behaviour problems, using well-grounded psychological
and developmentally appropriate techniques
[Parent, on-line surveyl.
The need for support for siblings was also identified:
do wonder about a primary mental health worker
role. . .the siblings often need support. . . younger
I
children have emotional needs - they might be
bed-wetting, doing all sorts and things [External
professional, L91.
Working with Staff
Support for staff, by providing supervision to individuals working with particularly difficult situations,
was also identified: 'Wouldn't it be nice to have a psychologist who could look at a case study with us"
TEACH staff,9l. Other stafftalked about awider supportive, teaching, and supervisory role:
The [psychologistl to be there for staff' . ,to start encouraging staffto think outside certain boxes, to facilitate group work with staff. . ' It could be peer
gï'oup superyision, it could be training,.. TEACH
staff,2l.
Another member of staff commented:
Awish list for me would be to have someone who is
very experienced psychotherapeutically as a resource to staff IEACH staff, 11.
Other professionals identified the need for a stronger
therapy structure within EACH and suggested that a
therapy lead was required, identified as:
. just bringing that perspective to the table. In doing so we'd be acknowl-
Someone who is strategic.
edging the importance and significance of
therapy TEACH staff,
321.
DISCUSSION
A number of qualitative approaches were used to obtain the views ofparents, hospice staff, and external
professionals about the provision of psychological
services. The findings about current and future provision, together with information about how need is
assessed and interventions are evaluated, formed
the basis of a series of recommendations for the development of a model of psychological support.
TWo main themes addressing perceptions of
psychological need were identified: understanding
psychological support and unmet need. The concept
of psychological support was understood in a number
of different ways, with staff and families having dif-
fering views as to what constituted psychological
support, who should and does provide it, and the
way in which it should be provided' As has been reported by others (Steele et al., 2005), the hospice
was seen as a lifeline for parents, providing them
with highly valued support and respite from caring'
However, several elements of the current model of
service delivery that were introduced to improve
the care offered to children and families conversely
elicited negative responses from families and staff.
For example, the flexibility in terms of how, where,
Perceptions of psychological seruices
in a hospice
381
and when support is provided can compromise the
benefits of care and therapy when staff are unable
to be as flexible as the service and families expect
them to be. Similarly, the notion of choice can cause
tensions about what families want and what the hospice is able to deliver with its available resources, and
this can extend to the provision ofpsychological support. Frictions were also evident in relation to differing perceptions and expectations of who was, and
should be, providing what kind of support, blurring
of role boundaries, and what type of service EACH
should be aiming for. Others have also identified difficulties regarding role definitions among members
of the multidisciplinary team and the potentially deleterious effect this can have on the effective and efficient Í'unctioning of the whole care team (Junger
et a1.,2A07; O'Connor & Fisher, 2011).
Similarly, a number of different unmet needs were
identified, in terms of specific areas in which psychological input would be beneficial, such as managing
challenging behavior, supporting staff to work with
families to ensure consistency of approach and personnel, and the broader need for psychological support for staff. What was evident throughout,
however, was that the holistic care of the child and family was regarded by families and staffalike as a key
component of hospice provision, as has been identified previously (Kirk & Pritchard, 2011). Although
there are gaps in current service provision there is,
importantly, a motivation and desire to address this.
The challenge now is to ensure that there is a unified
understanding ofwhat psychological support is and
how it should be delivered, so that expectations are
realistic and the needs of families and staff are met.
LIMITATIONS
Because of the short time frame set by the Depart-
ment of Health for completion of the project
(6
months) and the need to capture the views of a diverse
range ofparticipants, it was not possible to interview
participants more than once, or for participants to be
involved in reviewing the findings (although the final
report has subsequently been made available to all
participants). One aim of the project was to engage
with a diverse range of participants,
including
harder-to-reach families and professionals with limited time availability, so we adopted a mixed-methods
approach utilizing approaches that were individual or
Broup, face-to-face or telephone, and involved contact
at home, at work, or at an external venue, as well as
using an anonymous web-based survey, to facilitate
the inclusion of as many participants as possible
within the time frame of the project. The importance
of involving key stakeholders and obtaining multiple
perspectives when evaluating services has been re-
cognized (Hiatt et al., 2007), but a further limitation
is the absence of accounts from children and young
people who use the hospice, primarily because most
do not have verbal communication skills. Although
research has been conducted previously with children
with communication issues (Beresford et al., 2004;
Mitchell & Sloper, 2011), we had insufficient time to
engage, obtain consent from, and work with a gïoup
ofyoung people with learning disabilities in a manner
that would have ensured that their voice was properly
heard and documented. A number of siblings attended the family focus groups and their use of services
and what they liked/would like to change was explored (via an art medium) but this did not speciflcally address psychological need, and is not reported
here. The need for a gïeater emphasis to be placed
on obtaining the views and experiences of children
using hospices and their siblings has been identified
previously, and should be addressed as serwices are
developed andimplemented (Davies et al., 2005; Malcolm et al., 2008).
CONCLUSION
For psychological services to fulfll their potential in
the delivery of care to children with life-limiting conditions and their families, including end-of-life care,
there are a number ofchallenges that need to be addressed. These are related to articulating the role of
psychological support as part of a multidisciplinary
approach to care, training, and supervision needs of
individuals filling the role, to ensure appropriate service provision and to acknowledge competence, and to
obtain a realistic understanding by all stakeholders of
what is achievable and reasonable, especially across
an organization that covers such a wide geographical
region. It is evident that the role of psychologist would
be challenging, but as one participant commented:
"It's a complicated job. Doing support with staff and
parents. Maybe this is the Messiah, I don't know,"
ACIil\OWLEDGMET\ïTS
This work was Íunded by the Department of Health
through East Anglia's Children's Hospices (EACH). The
authors thank the families, external professionals, and
EACH staff who have given up their time to share their
stories and experiences with them, as well as thanking
the EACH stafffor allowing their practice to be put under
scrutiny.
REFERENCES
Association for Children's Palliative Care/Children's Hospices U.K. (2009). Right People, Right Place, Riglrt
ïlme. Bristol: ACT
Wra.Y
382
(2010)' Bereaveneeds assessment in specialist palliative care:
revierv of the literattre. Palliatiue Medícine, 24,
R., Tavlor,
Agnew,
"mentA., Manktelow
A
B',9t-{,
46-59.
Beresford, 8., Tbzer, R., Rabiee' P., et al.
QAlq' Developing
an approach to involving children with autistic spectrum àisorders in a social care research project' Btitish
J ournal o f Learnin g D i s abílit ie s, 32, 1"80 - 185'
British PsycLological Societv. (2008)' The Rop of Psvchology in End of Life Care. Leicester: British Psychological
Society.
Brehaut,i.C., Garner, R.8., Miller, A'R., Lach, et al' (2011)'
Changes over time in the health of caregivers of children
with Èealth problems: Growth-curve Íindings from a 10'
year Canadiàn population-based stucly. American Journal of Public. Health, 101,2308-2376.
Brinchmann, B.S. (1999). When the home becomes a
prison: Living with a severely disabled child' Nursing
Ethics,6,
L37
-143.
Carnevale,'F.A., Alexander, E., Davis, M., et al' (2006)'
Daily living with distress and enrichment: The moral
families with ventilator-assisted children
"*p"riutt."if
at home. Ped,iatrics,
117, e48-60.
Crawford, A.A. (2004)' Psychological models that help hospice workers perforrn mental status evaluations' The
American Journal of Hospice and Palliatiue Care,21,
26L-266.
Davies,8., Collins, J. B', Steele, R., et al. (2005). Children's
perspectives ofa pediatric hospice prograrn. Journal of
Palliatiue Care, 2 7, 252-261.
Davies, R. (2005). Mothers'stories of loss: Their need to be
u,ith their dying child and their child's body after death'
Journal of Chilcl Health Care,9,2BB-300.
DH. (2007). Aimíng High for Disabled Child,ren'
DH. (2008). Better Care: Better Liues.
East Anglia's Children's Hospice. (2010). Department of
Heattlh 130 Million Funding to Support New Child.ren's
ancl Young People's Palliatiue Care Projects in 20102011: Application by East Anglía's Child.ren's Hospices'
Eaton, N.-(2008). 'I don't know how we coped before': A
study ofrespite care for children in the home and hospíce. Journal of Clinical Nursing, 17, 3t96-3204'
Grinyer, A., Payne, S. & Barbarachild, Z. (2010)' Issues of
pàwer, contiol and choice in children's hospice respite
care sórvices: A qualitative study. International Journal
of' Palliatiue Nursíng, 16, 505-510.
Haley, W.E., Larson, D.G', Kasl-Godlel J., et al. (2003)'
Roies for psychologists in end-of-life care: emerging
models of practice'- P rofes sional P ry chology : Rp search
and Practice, 34, 626'633.
Heaton, J., Noyes, J., Sloper, P., et al. (2005). Families'experiences oi caring for technology;dgne1{e4 children:
À temporal perspóctive' Health & Social Care in the
Community, 13, 441- 45O.
Hiatt, K., Steile, C., Mulsow, M., et al. (2007). The importance'of perspective: evaluation of hospice care from
multiple itokóhold"t". The Am'erican Journal of Hos'
pice and Palliatiue Care, 24, 37 6-382.
et al-
S., Pestinger, M., Elsner, F., et al' (2007)' Criteria
Junger,
Tí. Éo.ó".tful iruitiprofessional cooperation in palliative care teams. Pal[iatiue Medicine,21,347 -354'
f§"t, S. C pdkhard, E. (2011). An exploration of parents'
and young people's perspectives of holn!ry support'
40'
C hilà : C are,- H eàlth an d D euelop ment, 3 8, 32 Malcolm, C., Forbat, L., Knighting, K., et al. (2008)' Exploring the experiences and perspectives offamilies using a
chïdren's'hospice and professionals providing hospice
care to identify Í'uture rlsearch priorities Íbr children's
hospice care. Palliatiue Med'icine, 22, 921-928'
Maynard, L., Rennie, T., Shirtliffe, J', et a,l. (2005)' Seeking
ànd using families' views to shape ch!l<lren's-hospice services. Iníernational Jow'nal of Palliatiue Nursing, 11,
624-630.
Mitchell, W. & Sloper,
P. (2011).
Making choices in mylife:
listening to th"iduus and experiences ofyoung people in
the UK*who communicate non-verbally' Child'ren and
Youth Seruices Reuiew, 33, 521-527 .
National Institute for Health and Clinical Excellence
(2004). Improuing supportiue and palliatiue care lbr
adults uith canceL The m.a,nual.
O'Connor, M. & Fisher, C' (2011). Exploring the dynamics
of interdisciplinary palliative care teams in providing
psychosociaÍ care:- "Everybody thilks lhat everybody
èan do it and th ey caÍt't" . Journal ofPalliatiue Meclicine,
14,191-196.
Oulton, K. & Heyman, B. (2009). Devoted protection:
how parents ofchildren with severe learning disabilities manage visk. Health, Rish and Society, 1"1,
303-319.
Roberts, K. & Lawton, D. (2001). Acknowledging the extra
care parents give their disabled children' Child: Care,
Health and Deuelopment, 27, 307-319.
Rodriguez, A. & KinÀ, N. (2009). The lived experience of
paienting a child *ith a lif"-li*iting condition: A focus
àn the mintal health realm. Palliatiue and' Supportiue
Care, 7, 7 -12.
Sherman, B.R. (1995). Impact of home-based respite care
on families of children with chronic iilnesses' ChiftJ
Health Care,24,33-45.
Sloper, T. & Beresford, B. (2006). Families Yith disabied
ihlldrett. Br it ish M e d.i cal J our nal, 3 3 3, 928 - 929'
Stainton, T. & Besser, H. (1998). The positive impact of children with an intellectual disabitity on the family' Journal of Intellectual and Developmental Dísability' 23,
57
-74.
Steele, R., Davies,8., Collins, J.B., et al' (2005)' End-of-life
care in a childrenls hospice progïam. Journal of Pallia'
tiue Care, 21, 5-71.
Wilkinson, S., Croy, P., King, M., et al. (2007)' Are we get-
ting it right? Pàrents' pèrceptions of hospice child be."u:rr"*ut1 support service§. Palliatiue Medicíne, 21,
401-407.
Yantzi, N.M., Rosenberg, M.W' & McKeever' P' (2007)' Getting out of the house: The challenge§mothers face when
theír children have long-term care needs' Health & So'
cial Care in the Com.ntuníty, 15,45-55'