Overview of Study Design
University of Nairobi
Winter 2013
Ann Vander Stoep, PhD
Psychiatric Epidemiologist
What is research design
• A research design is the arrangement of
conditions for collection analysis of data in
a manner that aims to combine relevance
to the research purpose with economy in
procedure
• It is a strategy specifying which approach
will be used for gathering and analyzing
the data
Types of Studies
Descriptive vs. Analytic
Descriptive Studies: Descriptive studies
present general characteristics of the
distribution of a disease (person/place/time)
•
Exploratory: No specific hypotheses
Analytic Studies: Analytic studies test a specific
hypothesis to determine whether a factor influences
the risk of a disease
•
Usually compare two groups (diseased vs. nondiseased or exposed vs. non-exposed)
Approach to research
Quantitative vs. Qualitative
• Qualitative is based on measurement of
quantity or amount
• Qualitative research is concerned with
quality or kind e.g why do the people think
and do certain things(concerned with
assessment of attitudes, opinion and
behavior
Research approach
Prospective: follow-up of subjects to
determine outcome status occurs after
study begins
Retrospective: follow-up time between
exposure and outcome has already
occurred when study begins
Characteristics of Studies
Unit of study
• Individuals
• Groups (Ecological Study)
– Don’t know whether diseased individuals actually had the
exposure or not
Characteristics of Studies
Time
•
Single point in time (Cross-sectional)
•
Multiple points in time (Cohort, Case-Control)
Descriptive Studies
Purpose “Provide information on patterns of disease
occurrence within populations according to
characteristics such as age, gender, race,…”
Used when:
•
little is known about causes of a disease
•
objective is to generate hypotheses about etiology
•
objective is to follow disease trends over time
•
early stages of disease investigation
Types of Descriptive Studies
1. Case report
No denominator
2. Case series
3. Population-based studies
• Prevalence
• Incidence
Denominator
• Mortality
Incidence
• Incidence and prevalence are both
measures of the extent of disease in a
population.
• Incidence tells us about a change in
status from non-disease to disease, thus
being limited to new cases.
Prevalence
• Prevalence includes both new cases and
those who contracted the disease in the past
and are still surviving.
The Numerator
• Both prevalence and incidence consist of
a numerator and a denominator.
• The numerator of an incidence measure is
a count of the newly developed cases that
arise in a population over a specified time
interval such as one year.
Numerator
• The numerator of a prevalence measure
consists of a count of the total number of
individuals in a population suffering from
the disease at a point in time.
The Denominator
• The denominator of both measures
consists of the number of persons in the
population from which cases of disease
arise, for example, the population of a city,
county,state or country.
Descriptive Studies
Descriptive studies provide the data and
information that become the foundation for the
development of hypotheses….
…leading to…
…Analytic studies
Analytic Studies
Purpose
To investigate the relationship between an
exposure and a disease outcome
Types of Analytic Studies
Observational Studies
•
•
•
Researcher systematically “observes” the association
between “exposure” and “outcome”
Cross-sectional studies, Cohort studies, Case-control
studies
Document naturally occurring associations
Experimental Studies
•
•
•
Investigator assigns “exposure” of interest
Clinical trials, Community trials
Randomized, non-randomized (quasi-experimental)
Observational Analytic Studies
Cross-Sectional Studies
Purpose: To assess relationships between
“exposure” and “outcome” (risk factor and
disease) at one point in time.
Observational Analytic Studies
Cross-Sectional Studies
Procedure:
1.
Measure “disease” (outcome), exposure, and
other factors simultaneously in defined population
2.
Identify individuals with/without condition and
assess presence/absence of “exposures”
3.
Compare frequency of exposure among
“diseased” and “non-diseased”
Observational Analytic Studies
Cross-Sectional Studies
In cross-sectional studies,
it is not possible to determine the temporal
sequence (which came first).
Did the exposure precede the disease???
Or did the disease precede the exposure???
Observational Analytic Studies
Cohort Studies
Purpose: “A cohort study involves comparing
disease incidence over time between groups
that differ on their exposure to a factor of
interest.”
(Koepsell and Weiss, 2003)
Observational Analytic Studies
Cohort Studies
Procedure:
1.
2.
3.
4.
Subjects without disease are classified on the basis of
exposure to potential risk factor:
“Exposed” / “Unexposed”
Observed over time to determine who develops
disease
Disease incidence compared in Exposed and
Unexposed
Yields “Relative Risk”
Observational Analytic Studies
Cohort Studies
Used when:
• Exposure is uncommon
• Outcome is not too uncommon
• Exposure may be difficult to recall accurately if assessed
long afterwards
• Objective is to investigate many outcomes of one
exposure
Observational Analytic Studies
Case-Control Studies
Purpose: To investigate hypotheses about disease
etiology (causes, risk factors, etc.)
“At first, a case-control study may appear to be
attacking the problem backwards, by proceeding
from effect to cause.”
(Koepsell and Weiss, 2003)
Observational Analytic Studies
Case-Control Studies
Procedure:
1. Subjects selected based on disease status:
Cases (with disease); Controls (without disease)
2. Exposure history ascertained
3. Frequency of exposure compared in Cases and
Controls
4.
Yields “Odds Ratio” (an estimate of the Relative Risk)
Observational Analytic Studies
Case-Control Studies
Used when:
• Disease is uncommon in general population
• Long latency period between exposure and disease
• Want to investigate many different exposures for one
outcome
• Accurate recall is possible or records of exposure are
available
Experimental Analytic Studies
Purpose:
• Test hypothesis regarding etiology of disease
• Evaluate method of intervention or prevention
• Eliminate bias by individual subject characteristics
Experimental Analytic Studies
Procedure:
1.
Form study groups from same underlying
population
2.
Researcher controls conditions under which
subjects are exposed
3. Researcher imposes the exposure (e.g.
intervention or level of risk factor)
Experimental Analytic Studies
Used when:
• much is known about epidemiology of disease
• want to evaluate clinical treatment options
• want to evaluate prevention approaches
Experimental Analytic Studies
Randomized Trials
Randomized [clinical or controlled] trials:
• Subjects randomly assigned to Experimental or
Control groups
• “Gold standard” of epidemiological studies
• Eliminate selection bias
Experimental Analytic Studies
“Quasi-Experimental” Studies
Quasi-experimental studies:
• Subjects (or subject groups) not randomized
• Researcher imposes (or does not impose)
exposure (intervention) on groups
Experimental Analytic Studies
Community Trials
Community trials:
• Entire population of community does or does not
receive intervention
• Communities may be randomized or nonrandomized to intervention or control condition
• Same approach might be applied to classrooms,
schools, neighborhoods, regions (groups, not
individuals)
Ann takes over the exercise
Toxic Shock Syndrome (TSS)
Clinical description: “Typically begins
suddenly with high fever, vomiting, and
profuse watery diarrhea, sometimes
accompanied by sore throat, headache,
and myalgias. The disease progresses to
hypotensive shock within 48 hours, and
the patient develops a diffuse, macular,
erythematous rash with non-purulent
conjunctivitis. Urine output is often
decreased, and patients may be
disoriented or combative…”
MMR; May, 1980
Toxic Shock Syndrome (TSS)
Laboratory:
“elevated blood urea nitrogen…white blood cell
counts…; in 33 of 45 (73%) patients cultured,
Staphylococcus aureus was isolated from the
throat, cervix, vagina, or rectum”
TSS: Descriptive Epidemiology
1. 1978 - Todd et al. described toxic shock
syndrome (TSS) in 7 children
2. May 1980 – “Since October 1, 1979, 55 cases
have been reported to CDC. Fifty-two of these
(95%) have been in women. Of 40 patients in
whom a menstrual history was obtained, 95%
had onset of illness within the 5-day period
following onset of menses…. Seven deaths have
occurred, for a case fatality ratio of 13%.”
TSS: Case-Control Study
1980 – “Because of the striking association with menses,
…the CDC administered a telephone questionnaire to 52
women who had illness meeting a clinical case definition
of TSS and 52 age- and sex-matched controls. …No
significant differences were found between the patients
and controls in marital status, contraceptive methods,
brand of tampon or sanitary napkin used,
absorbency…However, 50 of 50 cases with onset during
menstruation (100%) used tampons as compared to 43
of 50 controls (86%) (p=.02). Among case-control pairs
who used tampons, more cases used tampons at all
times (day and night) while menstruating (p < .05).”
TSS: Case-Control Study
“Eighty women who had onset of TSS between October 1,
1979, and September, 1980, and 160 age- and sexmatched neighborhood controls participated in this study
to evaluate risk factors associated with TSS in women.
Of the 80 women, 76 had onset of illness during their
menstrual periods. The odds ratio for developing
menses-associated TSS with any use of tampons
compared with no use of Tampons was 18.01 (p <
0.001). When exclusive use of a particular tampon
brand was compared with exclusive use of all other
brands, Rely was the only brand with a significantly
increased odds ratio (2.49; p=0.005)”
TSS: Cohort Study
A group of women (n = 50) who had used Rely
Tampons during the previous year were
compared with another group of women (n=50)
who had used other brands of tampons during
the previous year. Of the Rely Tampon users, 5
had developed TSS; of the other tampon users,
only 2 developed TSS. Women who use Rely
tampons are at increased risk of TSS compared
to women who use other types of tampons
(Relative Risk = 2.5)
TSS: Randomized trial
100 women, aged 20-40, who were not pregnant
and had regular menses, were recruited for this
experimental study. After being informed of the
potential risks, they consented to participate.
The women were randomly divided into two
groups. One group was provided Rely tampons
and told to use them during the next year; the
other group was provided with another brand of
tampons. At the end of one year, 10 of the
women in the Experimental group had dropped
out of the study. Of those remaining, 4 had
developed TSS. Of the women in the Control
group, 2 had developed TSS.
STUDY DESIGNS
DESCRIPTIVE
ANALYTIC
* Case Report
OBSERVATIONAL
* Case Series
* Cross-sectional
* Cohort
* Survey
EXPERIMENTAL
*Comparative trial
* RTC
* Case-control
Ecological or Individual unit of study