NURSING RESEARCH,

STEP BY STEP By Bernadette Capili, PhD, NP-C, and Joyce K. Anastasi, PhD, DrNP, FAAN
A series coordinated by the Heilbrunn Family Center for Research Nursing at Rockefeller University

Cohort Studies
A summary of study types, their strengths and limitations, and
results calculation and reporting.
Downloaded from http://journals.lww.com/ajnonline by BhDMf5ePHKav1zEoum1tQfN4a+kJLhEZgbsIHo4XMi0hCywCX1AWnYQp/IlQrHD3i3D0OdRyi7TvSFl4Cf3VC1y0abggQZXdtwnfKZBYtws= on 01/07/2022

Editor’s note: This is the seventh article in a series on clinical research by nurses. The series is designed to
give nurses the knowledge and skills they need to participate in research, step by step. Each column will
present the concepts that underpin evidence-based practice—from research design to data interpreta-
tion. The articles will be accompanied by a podcast offering more insight and context from the authors. To
see all the articles in the series, go to http://links.lww.com/AJN/A204.

C
ontinuing our discussion of observational
study designs, this column focuses on cohort
studies. The word “cohort” derives from
the ancient Roman military term for a group of
several hundred soldiers who march together to
achieve a tactical purpose.1 The epidemiology com-
munity began using the word during the 1930s to
mean a designated group of people with a com-
mon characteristic or characteristics (such as
smokers, ICU nurses, people exposed to lead in
drinking water) who are followed or traced over
a period of time.1 This group is followed longitu-
dinally, with periodic measurements to determine
the incidence of specific health outcomes or events.
Since cohort studies are observational, study par-
ticipants are monitored but study interventions
are not provided. This article describes prospec-
tive (following a group from the present into the
future) and retrospective (studying a group from
the past through to the present) cohort designs,
examines their strengths and weaknesses, and dis-
cusses methods for reporting the study results.
USES OF THE COHORT DESIGN
The cohort study design is excellent for understand-
ing an outcome or the natural history of a disease
or condition in an identified study population. Since
participants do not have the outcome or disease at
the start of the study, this study design can be used
to assess the relationship over time between expo-
sure and outcome.
A vital feature of a cohort study is selecting two
groups of study participants based on mutual char-
acteristics such as geographic location, birth year, or
occupation. Cohorts are also selected based on
exposure and nonexposure status. In such cases,
both groups should be similar except for exposure
status at study entry. For example, an investigator
could recruit people living with HIV (PLWH) who
smoke and who don’t smoke (have never smoked)
from the same community and follow them for five
years to determine the relationship between smok-
ing status and incidence of heart disease and stroke
in this population. Alternatively, the smokers could
be categorized by pack-years (less than five pack-
year smokers and more than five pack-year smok-
ers, for example) to determine whether heart dis-
ease and stroke are associated with the amount and
duration of smoking.
Prospective cohort studies are also referred to
as longitudinal studies. These studies are used to
answer a specific question or questions in a selected
area. Investigators recruit a sample of participants
and follow them over time, from the present to the
future. At predetermined time points, characteristics
are measured (via interviews, questionnaires, biolog-
ical assays, or physiological measures, for example)
to understand the relationship between the cohort
and the study outcome (see Figure 1).
During the recruitment phase, investigators need
to identify and exclude potential participants who
plan to move and thus may be difficult to reach
during the study’s follow-up period. The eligibility
criteria should reflect this consideration. Investiga-
tors should collect contact information from
enrolled participants: telephone number, e-mail
address, mailing address, and contact information
from at least two friends or family members in case
participants move or die during follow-up. Addi-
tionally, the study protocol should indicate periodic
contact with the participants, such as telephone
calls to provide assessment results, a study newslet-
ter, or study incentives (gift cards) to keep the par-
ticipants engaged.

ajn@wolterskluwer.com AJN ▼ December 2021 ▼ Vol. 121, No. 12 45

 NURSING RESEARCH,
STEP BY STEP
Using the HIV study example above, participants
are recruited from local New York City HIV pri-
mary care clinics and will be evaluated annually for
10 years to determine heart disease and stroke inci-
dence. PLWH are eligible to join if they smoke ciga-
rettes and have well-controlled HIV (undetectable
viral load). At study entry, individual smoking
exposure (pack-years) is determined, medical his-
tory is taken, and cardiovascular health is evalu-
ated. Participants identified at baseline to have a
history of heart disease or stroke are excluded. Par-
ticipants are categorized into two groups based on
smoking exposure: less than five pack-years or
more than five pack-years. The independent vari-
ables (outcome predictors: pack-years, blood pres-
sure, weight, waist circumference, lipid levels) and
the dependent variables (outcomes measured: inci-
dence of heart disease and stroke) are assessed
annually. The longitudinal design allows investiga-
tors to compare changes over time and determine if
the level of smoking exposure (pack-years) and
other independent variables are associated with the
outcome (incidence of heart disease and stroke).
Strengths and weaknesses. A primary strength
of the prospective cohort design is that it allows
investigators to determine incidence—the number
of new cases of an outcome occurring over time—in
this case, the incidence of new-onset heart disease
and stroke. In addition, measuring the independent
(predictor) variables before the onset of the out-
come strengthens the investigators’ ability to assess
the sequence of events and infer the causal basis of
an association between the predictor variables and
Figure 1. Prospective and Retrospective Cohort Designs
PROSPECTIVE
Start 2021
Outcome/Disease
Exposed
No Outcome/No Disease
Defined
Sample
Outcome/Disease
Unexposed
No Outcome/No Disease
RETROSPECTIVE
Start 2021
46 AJN ▼ December 2021 ▼ Vol. 121, No. 12 ajnonline.com
the outcome.2 A limitation of this design is that it
requires a large sample size when the incidence of
the outcome is smaller than the prevalence of the
exposure. Additionally, conducting the study may
be costly in terms of participant recruitment; the
number of staff needed; and the collection, storage,
and analysis of the outcome measurements. More-
over, some conditions, despite being relatively com-
mon (such as breast cancer or chronic obstructive
pulmonary disease), can occur at low rates in any
given evaluation period and not provide meaningful
results. In that case, participants need to be followed
for a longer duration, even though it increases the
cost as well as the possibility of participants with-
drawing from the study or being lost to follow-up.
Retrospective cohort studies are also called his-
torical cohort studies. The term “historical” is fit-
ting since data analysis occurs in the present but
participants’ baseline measurements and follow-ups
happened in the past. This type of study is feasible
if an investigator has access to a data set that fits
the research question. The data set must also have
adequate measurements of the predictor variables.
Generally, the data for a retrospective cohort
study were originally collected for other purposes—
for electronic health records (EHRs), for example,
or an administrative database such as Medicare.2
This study design’s primary goal is to examine
events or outcomes by reviewing past data in rela-
tion to predictor variables. Institutional review
board approval is required even though actual
patient interactions do not occur. For example, to
ascertain the incidence of heart disease and stroke
among PLWH who smoke, the EHRs of 500 HIV
patients from a local HIV primary clinic are exam-
ined over 10 years, from 2010 to 2020. In this
example, HIV patients are categorized by their
smoking exposure status: less than five pack-years
or more than five pack-years. The outcome of inter-
End 2031
est is the incidence of heart disease or stroke.
Strengths and weaknesses. The ability to analyze
an outcome based on data from already collected
measurements and participant follow-ups is one
strength of the retrospective cohort design. This
type of study is also inexpensive to conduct. One
limitation is that, because the data were collected
for other purposes, the data set may be incomplete
or inaccurate or include measurements that do not
fit the research question.2 In other words, the inves-
tigators do not have control over the data collection
methods and procedures.
METHODS FOR REPORTING RESULTS
End 2011 During a cohort study’s scheduled evaluation
periods, investigators determine the incidence of
Table 1. Calculating the Results of a Cohort Study
Disease No Disease
(Heart Disease/ (No Heart Disease/ Total Person-Time
Outcome Stroke) Stroke) Total (Years)
Exposed 125 375 500 (125 × 5) + (375 × 10) = 4,375
(Smoker)        a b (a + b) (a × 5a) + (b × 10b)

Unexposed 25 475 500 (25 × 5) + (475 × 10) = 4,875

(Nonsmoker)        c d (c + d) (c × 5a) + (d × 10b)

Total 150 850 1,000 9,250

(a + c) (b + d) (a + b + c + d) [(a × 5a) + (b × 10b)] +
[(c × 5a) + (d × 10b)]
a = exposed participant and acquires the outcome of interest
b = exposed participant and does not acquire the outcome of interest
c = unexposed participant and acquires the outcome of interest
d = unexposed participant and does not acquire the outcome of interest

Risk (cumulative incidence) of PLWH diagnosed with heart disease/stroke: (a + c)/(a + b + c + d) = 150/1,000 = 0.15 × 100 = 15%
Risk Ratio among PLWH who smoke for heart disease/stroke: [a/(a + b)] / [c/(c + d)] = (125/500) / (25/500) = 0.25/0.05 = 5
Rate (incidence rate) of heart disease/stroke among PLWH over 10 years: a + c/ [(a × 5a) + (b × 10b)] + [(c × 5a) + (d × 10b)] =
150/9,250 = 0.016 cases/person-year
Rate Ratio (IRR): a/[(a × 5a) + (b × 10b)] / c/[(c × 5a) + (d × 10b)] = 0.029/0.005 = 5.8

Interpretation of Risk Ratio or Rate Ratio

Risk Ratio or Rate Ratio = 1 Exposure is not preventive or harmful
Risk Ratio or Rate Ratio > 1 Exposure is harmful
Risk Ratio or Rate Ratio < 1 Exposure is protective

IRR = incidence rate ratio; PLWH = people living with HIV.

a
Participants diagnosed with heart disease/stroke at the end of 5 years of follow-up.
b
Participants with no heart disease/stroke at the end of the 10-year study duration.

the outcome of interest by counting the number Number of participants who

of participants who develop the outcome, such as
those PLWH in our example who develop heart
disease or stroke. Incidence is measured using
risks and rates.3 Both can provide additional infor-
mation about the exposure of interest (smoking,
nonsmoking) by calculating the risk ratio and rate
ratio.
Risk and risk ratio. Risk is also known as cumu-
lative incidence, and is defined as the number of par-
ticipants who develop the outcome of interest divided
by the total population of participants at risk. For
instance, investigators conduct a study to evaluate
the association between smoking and heart disease
and stroke among PLWH. They follow 1,000 PLWH
for 10 years, 500 of whom are smokers, 500 non-
smokers. Participants are evaluated annually. A total
of 125 heart disease and/or stroke cases were diag-
nosed in the smoking group while only 25 were diag-
nosed in the nonsmoking group (see Table 1). All
cases of heart disease/stroke were diagnosed at the
fifth year of follow-up.
Risk = develop the outcome
Total number of participants at risk
A total of 150 cases of heart disease and stroke
were identified in the cohort sample of 1,000 par-
ticipants. Based on these calculations, there was a
15% risk of developing heart disease or stroke
among the study participants. Additional analyses
used risk ratio to compare the risk between exposed
(smoker) and unexposed (nonsmoker) participants,
providing further information about the data. Risk
ratio illustrates the relative increase or decrease in
incidence of the outcome between the exposed and
unexposed groups.3
Risk Ratio = Riskexposed/Riskunexposed
Using the formula shown in Table 1, the risk
ratio was 5. The results demonstrate that PLWH
who were smokers (exposed) were five times more
likely to be diagnosed with heart disease or stroke

ajn@wolterskluwer.com AJN ▼ December 2021 ▼ Vol. 121, No. 12 47

 NURSING RESEARCH,
STEP BY STEP
Figure 2. Interpretation of Risk Ratio or Rate Ratio
Risk Ratio
or Rate
Ratio
Risk: less-exposed group Risk: more-exposed group
Less than 1 Greater than 1
1
than PLWH who were nonsmokers. To further
understand the meaning of the risk ratio results, if
the result had been equal to 1, then the exposure
(smoking) would not have affected the outcome. In
other words, the risk would have been the same for
the exposed and unexposed groups. If the risk ratio
had been less than 1, the exposure (smoking) would
have been protective for heart disease and stroke
(see Figure 2).
Rate and rate ratio. Rate is also known as inci-
dence rate and is defined as the number of partici-
pants who develop the outcome of interest (heart
disease and stroke) divided by the person-time
(days, months, years) at risk during follow-up.
­Person-time is the sum of each participant’s total
time free from the outcome of interest (no heart
­disease or stroke).
Rate = Number of new cases
Total person-time at risk
This measure provides the accumulated events
(cases of heart disease and stroke) and the speed at
which new health outcomes transpire in a study
cohort. Rate ratio is also used to compare and
understand the rate of speed (increase or decrease)
of a health outcome between the exposed and unex-
posed groups.
Rate Ratio (Incidence Rate Ratio [IRR]) =
IRRexposed/IRRunexposed
The calculated rate shown in Table 1 is 0.016,
indicating that 0.016 cases of heart disease and
stroke per person-year occurred in the sample.
The rate ratio was 5.8, indicating that heart dis-
ease and stroke rates were 5.8 times greater in
the exposed group than in the unexposed group.
As with risk ratio, if the result had been equal to
1, then the smoking exposure would not have
affected the outcome. If it had been less than 1,
then the smoking exposure would have been pro-
48 AJN ▼ December 2021 ▼ Vol. 121, No. 12 ajnonline.com
tective for heart disease and stroke. The further
the rate ratio is from 1 (null association, the
exposure is not preventive or harmful), the more
impact the exposure has on the study cohort.
REPORTING RECOMMENDATIONS
As with cross-sectional studies, which were discussed
in the last column, cohort studies also use the guide-
line Strengthening the Reporting of Observational
Studies in Epidemiology (STROBE)4 to explain
how a study is conducted and how the results are
obtained. This guideline provides specific recom-
mendations for cohort studies in its 22-item check-
list to guide investigators. The checklist provides
criteria for understanding research, including study
planning, conduct, findings, and conclusions. Addi-
tionally, the checklist contains information on how
a study might be replicated, how it can be used to
make clinical decisions, and what constitutes suffi-
cient information to be included in a systematic
review. See www.equator-network.org/reporting-
guidelines/strobe.
CONCLUSION
The cohort study design is appropriate to use to
determine the incidence of a health outcome or
event. It is especially helpful in understanding the
natural history of a disease or condition in an
identified study population. Additionally, the
cohort study allows an investigator to examine
the timing between an exposure and an outcome
or outcomes.
The next article in this series will discuss the
case–control study design. ▼
Bernadette Capili is director of the Heilbrunn Family Center
for Research Nursing at Rockefeller University, New York City.
Joyce K. Anastasi is the Independence Foundation Professor
of Nursing at the New York University Rory Meyers College
of Nursing, New York City. This manuscript was supported
in part by grant No. UL1TR001866 from the National Insti-
tutes of Health’s National Center for Advancing Translational
Sciences Clinical and Translational Science Awards Program.
Contact author: Bernadette Capili, bcapili@rockefeller.edu. The
authors have disclosed no potential conflicts of interest, finan-
cial or otherwise. A podcast with the authors is available at
www.ajnonline.com.
REFERENCES
1. Hood MN. A review of cohort study design for cardiovascu-
lar nursing research. J Cardiovasc Nurs 2009;24(6):E1-E9.
2. Hulley SB, et al. Designing clinical research. 4th ed.
Philadelphia: Wolters Kluwer/Lippincott Williams and
Wilkins; 2013.
3. Alexander LK, et al. Risk and rate in cohort studies. ERIC
Notebook 2015;2(7). https://sph.unc.edu/epid/eric.
4. von Elm E, et al. The strengthening the reporting of obser-
vational studies in epidemiology (STROBE) statement:
guidelines for reporting observational studies. Int J Surg
2014;12(12):1495-9.