Factors affecting the
attrition of older people
in longitudinal (health
research) studies
Anthea Tinker and Gill Mein on behalf of the team:
Anthea Tinker, Gill Mein, Suneeta Bhamra,
Richard Ashcroft, Clive Seale (and the late Janet
Askham)
Presentation of the findings of the research to the
NatCen conference 23.3.09
The project
 Funded by Atlantic Philanthropies
 For 9 months from 1st March 2008
(extended until 30.12.08)
 Investigators from multidisciplinary
backgrounds (disciplines include social
policy, sociology, philosophy, nursing,
social sciences, biomedical ethics)
Rationale and research question
The research question was:
“What factors encourage older people to
remain as participants, or discourages
them from continuing to participate in
health related longitudinal or panel
studies?”
Aims
The main aim of the research was to
provide guidance to research teams
planning or carrying out surveys about
ways of increasing retention of older
participants and reducing drop out rates
Some early issues
 What is a longitudinal study?
 How far back should the study go? We took current




and completed studies within 20 years
What ages should be taken into account? We have
taken 55+
We have included only those that cover both men and
women
The use of proxies – some studies allow this and we
have included studies where someone answered the
questions on behalf of the respondent but not where
the respondent was replaced by someone else.
We included studies from the UK, Europe, North America
and Australasia
Methods
1
• Literature review
• Questionnaires to other researchers
2
• Secondary analysis of existing data
• Quantitative and qualitative
3
• Collection of new data
• Focus groups and telephone
interviews
1. Literature Review
 A review of existing qualitative and quantitative literature
on the retention of older people in longitudinal studies
that have met our criteria
 Factors where there is relatively clear evidence of the
association with attrition
 Factors where there is no clear evidence about the
relationship with attrition
 Reasons for drop-out in these studies
 How other longitudinal studies have tried to reduce dropout and retain older participants
Literature review - Findings
 We found a larger number of studies than
anticipated (45)
 Few studies mention attrition in publications.
 A meta analysis of attrition based on 12 studies of
people aged 65 and over showed attrition was
associated with being older, having poor functioning
and cognitive impairment, living alone and not being
married (Chatfield et al, 2005)
However this study did not examine all the variables
that we have and was limited to only 12 studies
Literature review - Findings
The following factors indicate that people are more likely to drop out of
longitudinal studies if:
 They are older
 They are cognitively impaired
 They are from a lower socio-economic status
 They have fewer years of education
 They do not have children
 They are not retired
 They are less socially active e.g. clubs
The relationship of the following with attrition is unclear:
 Gender
 Marital status
 Being in poor health
 Home ownership
Literature review - findings
Main reasons for drop-out:
 Too time consuming - contact is too frequent
 Questionnaires - difficult, intrusive, humiliating
 Medical Exams - tiring, dislike of blood samples and
cognitive tests
Drop-out can be reduced by:
 Personal gestures – b’day cards, letters from PI
 Feedback - on study progress and medical results
 Emphasising importance of study – media coverage
 Incentives? - small gifts, monetary
 Practical issues - home visits, transport to medicals
Questionnaires to researchers of
other longitudinal studies
 Questionnaires were sent to 38 studies which
included both men and women together,
identified from the literature review. 20 studies
responded.
 Researchers were asked about identifying
attrition, measuring attrition, and efforts to
reduce attrition.
Questionnaires to researchers on
studies of older people - findings
 Studies differed. Across country, locality based,
organisation based. All were aware of attrition.
 Very few had collected reasons for dropout. Few
had published details.
 Attrition was measured in a variety of ways
comparison with baseline, between waves,
annually.
 Participation was described differently in most
studies and terms used were: full, partially
active, partially withdrawn, withdrawn, active,
temporary refusal, and dead.
Methods used by other studies to
reduce attrition
 Information: newsletters, personal response to
queries, meetings, results feedback,
study website
 Incentives: fridge magnets, pens, money,
calendars
 Contact: newsletters, information meetings,
birthday cards, reminder telephone calls
 Efforts to reduce attrition:
tracing lost participants, completing questionnaires
online and a study website. Having local medical
examinations. Use close relatives/proxies
2. Secondary analysis of existing
data
 The Whitehall II study (the Stress and Health
Study) of civil servants recruited from 20
London based departments in 1985, when
participants were aged between 35-55 years
 The original group =10,308 people (a response
rate of 73%)
 Participants self complete questionnaires (every
two years) and medical examinations (every 5
years)
 There have been other sub studies e.g. an MRI
scan, vascular sub study etc.
Secondary analysis of existing
data – Whitehall II Study
The Whitehall II study was originally set up
to investigate the social gradient in health.
It has particularly looked at the relationship
between stress, health and occupational
grades
PI is Professor Sir Michael Marmot at
UCL. The Study uses newsletters to
maintain contact and participation
The Whitehall study: Ethical
issues
 All the active participants in the Whitehall II
study have given consent to being followed up.
The process for contacting them, interviews etc
have been approved by the UCL research
ethics committee
 The PI is a participant in the Whitehall II study,
she will not have access to the data of her
fellow participants except in an anonymised
manner (this has been declared to the relevant
ethics committees)
Secondary analysis of existing
quantitative data




Quantitative Analysis
To see if we can determine what drop-out is related to
and establish predictors of attrition (e.g. characteristics
of respondents)
The choice of variables was guided by the literature
review and the variables available in the Whitehall II
study data set
We examined: age, gender, marital status,
accommodation type, education level, employment
grade, and longstanding illness (at baseline)
A comparison of those who have stated they do
not wish to take any further part in the study (excluding
people who withdrew for other reasons e.g. death) with
those who have continued to fully participate
Summary of statistical findings
Using Logistic regression we found attrition in the Whitehall II study is
associated with:






Having a lower occupational grade
Being older
Being female
Being unmarried at baseline
Engaging in fewer social activities at phase 5
Renting house from council or privately - unfurnished
(rather than owning house or renting privately -furnished)
 Being less educated
Secondary analysis of
existing data
Qualitative Analysis
An analysis of the Whitehall II data base of
comments from the views of participants
who have dropped out of the study and
their reasons for doing so
Qualitative Analysis of existing
Whitehall II Data - results
For those who withdrew from the study and left
comments, the most frequent reasons for dropping
out were:
 Too busy/study takes up too much time
 Journey is too long/difficult
 Chronic ill health
 Being seen elsewhere e.g. GP
 No longer interested in study
 Retirement
 Relocation or travelling
3. The collection of new data
 Using the Whitehall II study participants, a
small qualitative study to gather data
about people’s experiences of, and views
about, participating in longitudinal
research.
 Data was collected through 3 focus groups
and telephone interviews.
Recruitment to focus groups
 100 invitations given to participants




attending the medical examination
35 responded
5 refused
30 agreed to take part in focus group or
telephone interviews
19 attended focus groups /or were
interviewed on the telephone
Focus groups - results
The transcriptions from the focus groups were
coded and themes were identified and were put
into categories of:
 Good and bad experiences of the study
(medical and questionnaire)
 Motivating factors for continuing in the study
 Suggestions to make future participation
easier
Focus group results – good things
 Liked having a medical examination
 Liked explanation of some of the tests
 Impressed with the offer of home visits
 Liked flexible appt system
 Enjoyed meeting ex-colleagues
 Appreciates use of headphones for people
who cannot hear
Focus group results - bad things





The memory questions in the medical
examination made participants feel “stupid”
“humiliated”
Lack of understanding of reasons for
doing measures
Really disliked the questionnaire
Found the questions repetitive, did not like
having to chose between boxes to answer
Lack of information about study
Focus group results –
Motivation for continuing with the study

The most important factor was having a
medical examination

Loved the prestige of being recognised as
part of the study, publicity, media

“Giving something back”

Enjoyed talking to friends/old colleagues
after medical examination – boosted
confidence as all the same
Suggestions for the future from focus
groups and telephone interviews (1).


More information about themselves, in relation to
previous phases, the average, the population, the
study in general
Information about the study. How the study benefits
others. Public presentation from PI

Increase the medical examination with popular tests
e.g. PSA, eye checks, hearing tests

Complete questionnaire online

Have an online resource to answer participants
queries
Suggestions for the future from focus groups
and telephone interviews (2)
 Would like to be sent a set of memory test
questions to prepare beforehand!
 Would like questions to reflect retirement
 Reduce the length of the questionnaire
 Addressing travel and access to medical
examinations
Recommendations from the
Whitehall study




Foster loyalty to the study
Invite people by departments
Introduce popular tests e.g. PSA, hearing, eyes
Give more information, individual, comparisons
to average, to population. Pitch information
between lay and scientific. Presentation by PI.
 Continue with, good refreshments, comfortable
examination surroundings and continue to offer
home visits
 Exit interview to record reasons for withdrawal
Recommendations contd.
 Continue to be flexible about support of
travel needs
 Examine ways of shortening the
questionnaires
 Ask participants what would help them
continue participating- more involvement
in questionnaires/medical tests
Conclusion
 Our major conclusion is that longitudinal studies
are facing serious problems of drop out and are
anxious to find ways of avoiding this
 Our contacts with other studies, including the
Whitehall II study indicate their willingness to
consider ways of retaining participants. We will
be producing a short summary to be widely
distributed of ways in which this might be
achieved
Further information
Article reporting the literature review Quality in
Ageing (Vol 9 issue 4 Dec 2008)
 anthea.tinker@kcl.ac.uk
 s.bhamra@natcen.ac.uk
 g.mein@sgul.kingston.ac.uk
 r.ashcroft@qmul.ac.uk
 c.seale@qmul.ac.uk