Statistics

Study designs: Part 3 ‑ Analytical observational studies

Priya Ranganathan, Rakesh Aggarwal1
Department of Anaesthesiology, Tata Memorial Centre, Mumbai, Maharashtra, 1Director, Jawaharlal Institute of Postgraduate Medical
Education and Research, Puducherry, India

Abstract In analytical observational studies, researchers try to establish an association between exposure(s) and
outcome(s). Depending on the direction of enquiry, these studies can be directed forwards (cohort studies) or
backwards (case–control studies). In this article, we examine the key features of these two types of studies.

Keywords: Case–control study, cohort study, epidemiologic methods

Address for correspondence: Dr. Priya Ranganathan, Department of Anaesthesiology, Tata Memorial Centre, Ernest Borges Road, Parel, Mumbai ‑ 400 012,
Maharashtra, India.
E‑mail: [email protected]

INTRODUCTION the outcome. Thus, cohort studies are forward‑direction

studies (moving from exposure to outcome) and are
In a previous article[1] in this series, we looked at descriptive typically prospective studies (the outcome has not occurred
observational studies, namely case reports, case series, at the start of the study).
cross‑sectional studies, and ecological studies. As compared
to descriptive studies which merely describe one or more An example of cohort study design is a study by Viljakainen
variables in a sample (or occasionally population), analytical et al., which investigated the relation between maternal
studies attempt to quantify a relationship or association vitamin D levels during pregnancy and the bone health in
between two variables – an exposure and an outcome. As their newborns.[2] Maternal blood vitamin D levels were
discussed previously, in observational analytical studies, estimated during pregnancy. Children born to these mothers
the exposure is naturally determined as opposed to were then followed up until 14 months of age, and bone
experimental studies where an investigator assigns each parameters were evaluated. Based on the maternal serum
subject to receive or not receive a particular exposure. 25‑hydroxy vitamin D levels during pregnancy, children were
divided into two groups – those born to mothers with normal
COHORT STUDIES blood vitamin D and those born to mothers with low blood
vitamin D. The authors found that children born to mothers
A cohort is defined as a “group of people with a shared
with low vitamin D levels had persistent bone abnormalities.
characteristic.” In cohort studies, different groups of
people with varying levels of exposure are followed over Advantages of cohort studies
time to evaluate the occurrence of an outcome. These • For an exposure to be causative, it must precede the
participants have to be free of the outcome at baseline. outcome. In a cohort study, one starts with subjects
The presence or absence of the risk factor (exposure) in who are known to have or not have the exposure
each subject is recorded. The subjects are then followed up and are free of the outcome at the start of the
over time (longitudinally) to determine the occurrence of
Access this article online This is an open access journal, and articles are distributed under the terms of the Creative Commons
Attribution‑NonCommercial‑ShareAlike 4.0 License, which allows others to remix, tweak, and
Quick Response Code: build upon the work non‑commercially, as long as appropriate credit is given and the new creations
Website:
are licensed under the identical terms.
www.picronline.org
For reprints contact: [email protected]

DOI:
10.4103/picr.PICR_35_19 How to cite this article: Ranganathan P, Aggarwal R. Study designs: Part
3 - Analytical observational studies. Perspect Clin Res 2019;10:91-4.

© 2019 Perspectives in Clinical Research | Published by Wolters Kluwer - Medknow 91

 Ranganathan and Aggarwal: Case–control and cohort studies

study, and the outcome develops later. Hence, one is Uses of cohort studies
certain that the exposure preceded the outcome, and • Since cohort study design closely resembles the
temporality (and therefore probable causality) can be experimental design with the only difference being
established. In the above example, one can be certain lack of random assignment to exposure, it is
that the maternal vitamin D deficiency preceded the considered as having a greater validity compared to
bone abnormalities. the other observational study designs.
• For a given exposure, more than one outcome can be • Since one starts with subjects known to have or not
studied. In the above example, the authors compared have exposure, one can determine the risk of outcome
not only bone growth but also the age at which the among exposed persons and unexposed persons, as
babies born to low and high vitamin D mothers started also the relative risk.
walking independently. • In situations where experimental studies are not
• In cohort studies, often several exposures can be feasible (e.g., when it is either unethical to randomize
studied simultaneously. For this, the investigators participants to a potentially harmful intervention,
begin by assessing several 'exposures', for example, such as smoking, or impractical to create an exposure,
age, sex, smoking status, diabetes, and obesity/ such as diabetes or hypertension), cohort studies are a
overweight status in every member of a population. reasonable and arguably the best alternative.
The entire population is then followed for the outcome
of interest, for example, coronary artery disease. Variations of cohort studies
At the end of the follow‑up, the data can then be Sometimes, a researcher may look back at data which have
analyzed for several contrasting cohorts defined by already been collected. For example, let us think of a hospital
levels of each “exposure” – old/young, male/female, that records every patient's smoking status at the time of
smoker/nonsmoker, diabetic/nondiabetic, and the first visit. A  researcher may use these records from
underweight/ideal body weight/overweight/obese, 10 years ago, and then contact the persons today to check
etc. if any of them have already been diagnosed or currently
have features of lung cancer. This is still a forward‑direction
Limitations of cohort studies study (exposure traced forward among exposed and
• Cohort studies often require a long duration of follow‑up unexposed to outcome) but is retrospective (since the
to determine whether outcome will occur or not. This outcome may have already occurred). Such studies are
duration depends on the exposure‑outcome pair. In the known as 'retrospective cohort studies'.
above example, a follow‑up of at least 14 months was
used. An even longer follow‑up over several years or Large cohort studies, such as the Framingham Heart Study
decades may be necessary – for instance, in the above or the Nurses’ Health Study, have yielded extremely useful
example, if the investigators wanted to study whether information about risk factors for several chronic diseases.
maternal vitamin D levels influence the final height of a
person, they would have needed to follow the babies till CASE-CONTROL STUDIES
adolescence. During such follow‑up, losses to follow‑up,
and logistic and cost issues pose major challenges. In case-control studies, the researcher first enrolls cases
• It is not uncommon for one or more unknown (participants with the outcome) and controls (participants
confounding factors to affect the occurrence of without the outcome) and then tries to elicit a history of
outcome. For example, in a cohort study looking at exposure in each group. Thus, these are backward‑direction
coffee drinking as a risk factor for pancreatic cancer, studies (looking from outcome to exposure) and are always
people who drink a large amount of coffee may also be retrospective (the outcome must have occurred when the study
consuming alcohol. In such cases, the finding that coffee starts). Typically, cases are identified from hospital records,
drinkers have an increased occurrence of pancreatic death certificates or disease registries. This is followed by the
cancer may lead the investigator to incorrectly conclude identification and enrolment of controls.
that drinking coffee increases the risk of pancreatic
cancer, whereas it is the consumption of alcohol which Identification of appropriate controls is a key element of the
is the true risk factor. Similarly, in the above study, the case-control study design and can influence the estimate of
mothers with low and high vitamin D levels could have association between exposure and outcome (selection bias).
been different in another factor, e.g. overall nutrition or The controls should resemble cases in all respects, except for
socioeconomic status, and that could be the real reason the absence of disease. Thus, they should be representative
for the differences in the babies’ bone health. of the population from which the cases were drawn. For
92 Perspectives in Clinical Research | Volume 10 | Issue 2 | April-June 2019
 Ranganathan and Aggarwal: Case–control and cohort studies

instance, if cases are drawn from a community clinic, an overall food intake, milk intake, and outdoor play time.
outpatient clinic or an inpatient setting, the controls should These factors could influence both the likelihood of
also ideally be from the same setting. prior use of vitamin D supplements (exposure) and the
risk of fracture (outcome), affecting the measurement
Sometimes, controls are individually matched with cases for of their association.
factors (except for the one which is the exposure of interest) • The determination of exposure relies on existing
which are considered important to the development of the records or history taking. Either can be problematic.
outcome. For example, in a study on relation of smoking The records may not contain information on exposure
with lung cancer, for each case of lung cancer enrolled, or contain erroneous data (e.g., those collected
one control with similar age and sex is enrolled. This would perfunctorily). This is particularly challenging if the
reduce the risk of confounding by age and sex – the factors missing or unreliable data are more likely to be present
used for matching. Sometimes, the number of controls per in one of the two groups being compared – cases
case may be larger (e.g. two, three, or more). or controls (misinformation bias). During history
taking, cases may be more likely to recall exposure
Furthermore, to minimize assessment bias, it is important than controls (recall bias), for example, the mother
that the person assessing the history of exposure (e.g., of a child with a congenital anomaly is more likely to
smoking in this case) is unaware of  (blinded to) whether
recall drugs ingested during pregnancy than a mother
the participant being interviewed is a case or a control.
with a normal child. In the study by Anderson et al,[3]
For example, Anderson et al. conducted a case–control the mothers of children with fractures could have
study to look at risk factors for childhood fractures.[3] underestimated the amount of vitamin D their children
They recruited cases from a hospital fracture clinic and have received, believing that this was the reason for
individually matched controls (children without fractures) the occurrence of fracture.
from a primary care research network. The cases and • Finally, since case–control studies are backward‑directed,
controls were matched on age, sex, height, and season. there is no “at risk” group at the start of the study;
They found that the history of previous use of vitamin D therefore, the determination of “risk” (and relative risk
supplements was significantly higher in the children without or risk ratio) is not possible, and one can only estimate
fractures, suggesting an inverse association between “odds” (and odds ratio). For a detailed discussion on
vitamin D supplementation and incidence of fractures. this, please refer to a previous article.[4]

Advantages of case–control studies Uses of case–control studies

• Case-control studies are often cheap, and less • Case-control studies are ideal for rare diseases, where
time‑consuming than cohort studies. identifying cases is easier than following up large
• Once cases and controls are identified and enrolled, numbers of exposed persons to determine outcome.
it is often easy to study the relationship of outcome • Case-control studies, because of their simplicity and
with not one but several exposures. need for fewer resources, are often the initial study
design used to assess the relationship of a particular
Limitations of case–control studies exposure and an outcome. If this study is positive,
• In case-control studies, temporality (whether the outcome then a study with more complex and robust study
or exposure occurred first) is often difficult to establish. design (cohort or interventional) can be undertaken.
• There may be a bias in selecting cases or controls. For
instance, if the cases studied differ from the entire pool A special variation of case–control study design
of cases of a disease in an important characteristic, then Nested case-control design is a special type of case-control
the results of the study may apply only to the selected study design which is built into a cohort study. From the
type of cases and not to the entire population of cases. In main cohorts, participants who develop the outcome
the above example,[3] the cases and controls were derived (irrespective of whether exposed or unexposed) are chosen
from different sources, and it is possible that the children as cases. From among the remaining study participants
that attended the hospital fracture clinic had different who have not developed the outcome, a subset of
socioeconomic backgrounds to those attending the matched controls are selected. The cases and controls are
primary care facility from where controls were enrolled. then compared with respect to exposure. This is still a
• Confounding factors, as discussed in cohort studies, also backward‑direction (since the enquiry begins with outcome
apply to case-control studies. For instance, the children and then proceeds toward exposure) and retrospective
with fractures and controls could have had different study (since outcomes have already occurred when the
Perspectives in Clinical Research | Volume 10 | Issue 2 | April-June 2019 93
 Ranganathan and Aggarwal: Case–control and cohort studies

study starts). The main advantage is that since one knows REFERENCES
that the outcome had not occurred when the cohorts were
1. Ranganathan P, Aggarwal R. Study designs: Part 1 – An overview and
established, temporal relation of exposure and outcome classification. Perspect Clin Res 2018;9:184‑6.
is ensured. 2. Viljakainen HT, Korhonen T, Hytinantti T, Laitinen EK, Andersson S,
Mäkitie O, et al. Maternal vitamin D status affects bone growth in early
Financial support and sponsorship childhood – A prospective cohort study. Osteoporos Int 2011;22:883‑91.
Nil. 3. Anderson LN, Heong SW, Chen Y, Thorpe KE, Adeli K, Howard A,
et al. Vitamin D and fracture risk in early childhood: A case‑control
study. Am J Epidemiol 2017;185:1255‑62.
Conflicts of interest 4. Ranganathan P, Aggarwal R, Pramesh CS. Common pitfalls in statistical
There are no conflicts of interest. analysis: Odds versus risk. Perspect Clin Res 2015;6:222‑4.

94 Perspectives in Clinical Research | Volume 10 | Issue 2 | April-June 2019