*Geek Box: Nested Case-Control & Case-Cohort Studies
Prospective cohort studies take a cohort of people from the population, and follow them prospectively, i.e., over time. The advantage to this design is that exposures – like diet, smoking, etc. – can be assessed before a disease develops in the participants, and this reduces the potential for recall bias. Another advantage of prospective cohort studies is the potential sample size, which can range into the thousands to hundreds of thousands of participants.
However, this comes with a potential disadvantage in terms of gathering more granular detail regarding exposures and participant characteristics, which would be prohibitively expensive to do with a cohort of, for example, >100,000 people. To navigate this potential hurdle, some additional study designs have been used, in particular nested case-control and case-cohort designs.
In a nested case-control design, a number of healthy participants are selected as controls for each case of an outcome, e.g., each case of coronary heart disease that is diagnosed. The design is termed ‘nested’, because it is within a larger defined cohort. In this design, the researchers identify cases of their outcome of interest – we’ll use CHD for example – that have already occurred, or as they occur if the parent study is a prospective cohort, and then select up to 4 to 5 healthy controls to match with a case of CHD. The nested case-control study would then compare, for example, blood cholesterol levels and blood pressure in the healthy controls vs. the CHD cases.
While nested case-control studies have some advantages, they can face some logistical hurdles, in particular if more than one disease outcome is of interest – which is common in prospective cohort studies – then it can be highly inefficient, as each case of each different disease would require its own set of healthy controls to be matched with. As a result, the case-cohort design was proposed as an alternative to the nested case-control. In a case-cohort design, rather than select 4-5 healthy controls for each case, a random sample [known as a ‘subcohort’] is selected along with all identified cases of the disease outcome of interest. Thus, the case-cohort consists of the cases and the subcohort. The main advantage of the case-cohort design is that, by the selection of a random sample, multiple disease outcomes can be examined from the same subcohort. This is because the subcohort is not matched to specific cases, as occurs in a nested case-control design.
Both study designs share the advantage of lower cost and more efficiency in execution, and where the outcome of interest is a single disease endpoint, the difference between the two designs is small. However, where multiple disease endpoints are of interest, the case-cohort design is more efficient due to the random sampling of the subcohort.