Cohort Studies: Prospective and Retrospective

Cohort studies are the strongest observational design for establishing temporal sequence — proving that exposure came before outcome. You classify participants by exposure status at baseline, then follow them over time to see who develops the outcome. The word 'cohort' comes from the Latin 'cohors,' meaning a group of Roman soldiers who march together — and this is exactly what the design does: it marches a group of people forward through time. A prospective cohort study recruits participants now and follows them into the future. The researcher defines the exposed and unexposed groups at the start, collects baseline measurements, and then monitors the cohort at regular intervals for the development of outcomes. This is considered the gold standard of observational research because: the researcher controls data collection quality, exposures are measured before outcomes occur (eliminating temporal ambiguity), and multiple outcomes can be studied simultaneously. The major disadvantages are cost (potentially millions of dollars), time (may require decades of follow-up for diseases like cancer), and attrition (participants drop out, move away, or die from other causes). A retrospective cohort study uses historical records: you identify a past cohort (for example, all employees who worked at a chemical plant between 2000 and 2010), classify them by exposure status using historical records, and trace their outcomes forward to the present using medical records, death certificates, or registry data. The key insight is that this is NOT a case-control study — you are still following the same cohort logic (exposure → outcome), just using data from the past. Both prospective and retrospective cohort studies can calculate incidence rates, relative risk (RR), absolute risk difference, and number needed to harm (NNH). The key practical differences between prospective and retrospective cohorts are: prospective collects data you design (higher quality, you choose what to measure), costs more, and takes longer. Retrospective uses existing records (faster and cheaper, but limited to whatever variables were recorded, with potential data quality issues like missing data or inconsistent coding). A common confusion is between a retrospective cohort study and a case-control study — they are fundamentally different designs. A retrospective cohort starts with exposure and follows forward (even if using past records). A case-control study starts with the outcome and looks backward. Major threats to cohort studies include attrition bias (if people who leave the study differ systematically from those who stay — for example, sicker patients may drop out more often), confounding by indication (in pharmacoepidemiology, sicker patients receive more aggressive treatment, so the treatment group appears to have worse outcomes), healthy worker effect (employed cohorts are healthier than the general population), and information bias. Strategies to handle confounders include multivariable regression, propensity score matching, and stratification. In later levels of this course, you will learn how to conduct each of these study designs step by step from zero — from writing the protocol to collecting data to analyzing results and writing the manuscript.