Cohort Study and Quasi Experiment Types and Problems

Cohort As time Dimensional Study Design

    A cohort design is a time-dimensional design to examine sequences,
patterns of change or growth, or trends over time. A cohort is a group with
common characteristics or experiences during a given time period. 

    Cohorts
generally refer to age groups or to groups of respondents who follow each other
through formal institutions such as universities or hospitals or informal
institutions such as a family. 

    Populations also can be classified according to
other time dimensions, such as time of diagnosis, time since exposure to a
treatment, or time since initiating a behavior. A cohort might be graduates of
nurse practitioner programs in the years 1995, 2000, 2005 or siblings in
blended families. 

    Cohort designs were originally used by epidemiologists and
demographers but are increasingly used in studies conducted by nurses and other
researchers in the behavioral and health sciences.

Cohort Design Refers to a Quasi Experimental Design

    In the most restrictive sense, a cohort design refers to a
quasi-experimental design in which some cohorts are exposed to a treatment or event
and others are not. 

    The purpose of a cohort design is to determine whether two
or more groups differ on a specific outcome measure. 

    Cohort designs are useful
for drawing causal inferences in quasi-experimental studies because cohort
groups are expected to differ only minimally on background characteristics. 

    Recall that a quasi-experimental design lacks random assignment of subjects to
groups. 

    Although the groups in a cohort design may not be as comparable as
randomly assigned groups, archival records or data on relevant variables can be
used to compare cohorts that received a treatment with those that did not. 

    Because simple comparisons between cohorts may suffer from a number of design
problems, such as biased sample selection, intervening historical events that
may influence the outcome variable, maturation of subjects, and testing
effects, a strong cohort design can account for many of these threats to the
internal validity of a study.

 Types of Cohort Study Design 

    There are two major types of cohort design: cohort design with treatment
partitioning and the institutional cycles design. 

    In a cohort design with
treatment partitioning, respondents are partitioned by the extent of treatment
(amount or length) received. In the institutional cycles design, one or more
earlier cohorts are compared with the experimental cohort on the variable(s) of
interest. 

    The institutional cycles cohort design is strengthened if a
nonequivalent nontreatment group is measured at the same time as the
experimental group. 

    A well-planned cohort design can control for the effects of
age or experience. when these might confound results in a pre-test-post test
design or when no pretest measures of experimental subjects are available.
Cohort designs might utilize a combination of cross-sectional and longitudinal
data.

    The term cohort studies broadly refers to studies of one or more
cohort groups to examine the temporal sequencing of events over time. Cohort
studies may eventually lead to hypotheses about causality between variables and
to experimental designs. 

    Most cohort designs are prospective (eg, the Nurses’
Health Study, in which 100,000 nurses were enrolled in 1976 and have been
followed since) al- though some are retrospective.

    There are a number of types of cohort studies. The panel design, in
which one or more cohorts are followed over time, is especially useful for
describing phenomena. Trend studies are prospective designs used to examine
trends over time. 

    In trend studies, different subsamples are drawn from a
larger cohort at specified time points to look at patterns, rates, or trends
over time (Polit & Hungler, 1995). 

    Panel designs with multiple cohorts are
used to study change in the variable(s) of interest over time, to examine
differences between cohort groups in variables, and to identify different
patterns between groups. 

    In a panel study with multiple chorts , the groups can
enter the study at different points in time, and the effects of aging can be
differentiated from the effect of being a member of a particular cohort group
(Woods & Catanzaro, 1988). 

    A prospective study is a variation of a panel
design in which a cohort free of an outcome but with one or more risk factors
is followed longitudinally to determine who develops the health outcome. 

    The
prospective design is used to test hypotheses about risk factors for disease or
other health outcomes. Some authors limit the term cohort study to designs in
which exposed and non exposed subjects are studied prospectively or
retrospectively from a specific point.

Problems of Cohort Study 

    A major problem with prospective studies of all types is subject
attrition from death, refusal, or other forms of loss. The loss of subjects in
a prospective study may lead to biased estimates about the phenomena of
interests.