A retrospective study, sometimes called a historical cohort study, is a type of longitudinal study (research that follows the same people over time). Researchers look back through existing records to see who went on to develop an outcome of interest, such as a disease.
In a retrospective cohort study, the researcher identifies two groups from the historical record: one exposed to a certain factor, and one that was not exposed. The researcher then looks back in time. This shows how often an outcome, such as a disease, occurred in each group.
For example, a researcher might compare smokers with non-smokers. Looking back at medical records shows how lung cancer rates differ between the two groups.
This design is common in epidemiology, the study of how often diseases occur and what influences them. Researchers can use existing data to understand potential risk factors.
Key Takeaways
- Definition: A retrospective cohort study looks back at existing records to compare an exposed group with an unexposed group and see how an outcome, such as a disease, differed between them.
- Method: Researchers identify the two groups from historical data and track which factors differed between those who developed the outcome and those who did not.
- Speed and Cost: Because the data already exists, retrospective studies are quicker and cheaper than prospective studies, and are especially useful for studying rare diseases.
- Recall Bias: Retrospective studies rely on memory and old records, so they are more vulnerable to inaccurate or missing data than prospective studies, which measure events as they happen.
- Weaker Than Prospective: Retrospective designs are generally regarded as weaker evidence than a true prospective follow-up, because researchers cannot control what was originally recorded.
- vs Case-Control: Unlike case-control studies, which start with people who already have a disease, retrospective cohort studies start with a group before any members have developed it.
How a Retrospective Cohort Study Works
Researchers identify a cohort before its members develop a disease. They then mine existing data, such as medical records, for patterns and exposures to suspected risks.
In cohort studies, one group shares a common exposure factor. This exposed group is then compared with another group that was not exposed.
Consider exercise and diabetes as an example. Men over 60 who exercise daily could be compared with men over 60 who do not. Researchers then compare how many men in each group went on to develop diabetes. The comparison group shows what to expect without regular exercise.
Researchers collect data from existing records to study the relationship between exposure and outcome. They then calculate the relative risk, the outcome rate in the exposed group divided by the rate in the unexposed group. This shows how strongly the two groups differ in outcome.
Advantages of Retrospective Cohort Studies
Temporal Order and Relative Risk
Because the exposure was recorded before the outcome occurred, a retrospective cohort study can show that the exposure came first. A single snapshot of one moment cannot do this.
Researchers can then compare how often the outcome occurred in the exposed and unexposed groups and calculate the relative risk. This shows an association, not proof of cause. An unmeasured confounding variable may still explain the link.
Inexpensive and less time-consuming
Retrospective studies tend to be cheaper and quicker than prospective studies as the data already exists, and researchers do not need to recruit participants.
The time period has already elapsed. A prospective study must wait years, sometimes decades, for an outcome such as a disease to develop. A retrospective study can instead examine records, such as medical charts, that were created long ago and already cover that period.
Rare diseases are easier to study this way. A prospective study would need to recruit an extremely large cohort. Cohort designs suit rare exposures too, and existing records spare researchers from recruiting a new group to study them (Euser et al., 2009).
Limitations of Retrospective Cohort Studies
Bias and confounding variables
Most errors in retrospective studies come from confounding and bias. Both are more common than in prospective studies. Researchers should therefore choose a prospective design whenever one is possible.
A retrospective study cannot control what was measured. It also cannot control how consistently data were recorded at the time.
Like all observational research, it can reveal an association between exposure and outcome, but that association may be produced by a confounding variable the researcher never measured.
Recall bias
Participants might not remember whether they were exposed, when they were exposed, or other details that matter for the study.
A retrospective study asks people to reconstruct events from years earlier. Records and memories both degrade with time, and the loss can differ between people who did and did not develop the outcome.
A prospective study avoids this problem. It takes each measurement at the moment it happens, so researchers do not have to rely on anyone’s memory.
That is why recall bias is one of the central weaknesses of retrospective research, and why its findings are treated with more caution than prospective ones.
Missing data
Researchers using existing data rely on others to keep accurate records. Important information may never have been collected in the first place.
Retrospective studies cannot go back and collect it. The data were gathered for another purpose, such as patient care, before the study began.
This is one reason retrospective designs are generally regarded as weaker than a prospective follow-up. The researcher must work with whatever records happen to survive, rather than data collected specifically to answer the research question.
Retrospective vs Prospective Cohort Studies
Choosing between the two designs is usually a trade of temporal accuracy against speed (Euser et al., 2009). A prospective cohort study records each measurement as it happens but must wait for outcomes. A retrospective cohort study skips the wait but depends on the records that already exist.
| Retrospective cohort | Prospective cohort | |
|---|---|---|
| Data collection | Looks back at existing records, archives or recollections | Enrolls people before the outcome occurs and measures forward in real time |
| Main advantage | Efficiency: quicker and cheaper, because the time has already elapsed | Accuracy: each measurement is taken when it happens, so memory is not relied on |
| Main weakness | Incomplete records and recall distortion | Long wait, high cost and participants dropping out |
| Strength of evidence | Generally regarded as weaker | Among the strongest observational designs |
Examples of Retrospective Cohort Studies
Published examples of historical (retrospective) cohort studies include:
- Trafficking and severe mental illness: South London health records showed trafficked adults were more likely than matched non-trafficked adults to be compulsorily admitted (Oram et al., 2015).
- Vitamin D and dialysis: Among 51,037 hemodialysis patients, 2-year mortality was 13.8 per 100 person-years with injectable vitamin D and 28.6 without (Teng et al., 2005).
- Critical incidents: 3.5% of 110,310 procedures (Munting et al., 2015).
- Eosinophils and asthma: Patients with asthma and a high blood eosinophil count before admission were readmitted more often within a year, 6.6% versus 4.2% (Kerkhof et al., 2018).
- Trafficking and mental disorders: Among 120 women survivors, childhood sexual abuse and unmet needs after trafficking raised the odds of a mental disorder (Abas et al., 2013).
Frequently Asked Questions
1. What is the difference between case-control and retrospective cohort studies?
Case-control studies are usually, but not exclusively, retrospective. These studies are performed on individuals who already have a disease. Researchers then compare them with other individuals who share similar characteristics but do not have the disease.
A retrospective cohort study works the other way around. Researchers examine a group before any of its members develop the disease. They then track which factors differed between those who developed the condition and those who did not.
More simply, case-control studies start from the outcome and look back for earlier exposures. Cohort studies start from the exposure and follow people forward.
2. Is a retrospective study experimental?
No, retrospective cohort studies are observational. Researchers analyze a group of subjects. They do not manipulate variables or interfere with participants’ environment.
Researchers use existing data to investigate the target population, so no experimentation is necessary. Retrospective cohort studies examine the association between an exposure and an outcome, such as a disease. They cannot prove cause. An unmeasured confounding variable may explain the link.
Experimental studies are required to determine why a certain factor is associated with a particular outcome.
Retrospective cohort studies sit above cross-sectional studies and case-control studies in the hierarchy of evidence, the ranking of study designs by how securely they support causal claims. That is because they still show the exposure came first.
They sit below prospective cohort studies. They also sit below randomized experiments, which can manipulate and control variables directly.
It is random assignment, not the passage of time, that gives an experiment its stronger causal warrant (Concato et al., 2000). But no randomized controlled trial can assign a chronic illness or decades of exposure. For many questions, an experiment is simply not available.
Even so, across a broad sweep of clinical questions, well-designed observational studies did not systematically overestimate treatment effects compared with randomized trials on the same question (Benson & Hartz, 2000). The two designs suit different kinds of question.
References
Abas, M., Ostrovschi, N. V., Prince, M., Gorceag, V. I., Trigub, C., & Oram, S. (2013). Risk factors for mental disorders in women survivors of human trafficking: A historical cohort study. BMC Psychiatry, 13, Article 204. https://doi.org/10.1186/1471-244X-13-204
Benson, K., & Hartz, A. J. (2000). A comparison of observational studies and randomized, controlled trials. New England Journal of Medicine, 342(25), 1878–1886. https://doi.org/10.1056/NEJM200006223422506
Concato, J., Shah, N., & Horwitz, R. I. (2000). Randomized, controlled trials, observational studies, and the hierarchy of research designs. New England Journal of Medicine, 342(25), 1887–1892. https://doi.org/10.1056/NEJM200006223422507
Euser, A. M., Zoccali, C., Jager, K. J., & Dekker, F. W. (2009). Cohort studies: Prospective versus retrospective. Nephron Clinical Practice, 113(3), c214–c217. https://doi.org/10.1159/000235241
Hess, D. R. (2004). Retrospective studies and chart reviews. Respiratory Care, 49(10), 1171–1174. https://doi.org/10.4187/respcare.04491171
Kerkhof, M., Tran, T. N., van den Berge, M., Brusselle, G. G., Gopalan, G., Jones, R. C. M., Kocks, J. W. H., Menzies-Gow, A., Nuevo, J., Pavord, I. D., Rastogi, S., & Price, D. B. (2018). Association between blood eosinophil count and risk of readmission for patients with asthma: Historical cohort study. PLOS ONE, 13(7), Article e0201143. https://doi.org/10.1371/journal.pone.0201143
Munting, K. E., van Zaane, B., Schouten, A. N. J., van Wolfswinkel, L., & de Graaff, J. C. (2015). Reporting critical incidents in a tertiary hospital: A historical cohort study of 110,310 procedures. Canadian Journal of Anesthesia/Journal canadien d’anesthésie, 62(12), 1248–1258. https://doi.org/10.1007/s12630-015-0492-y
Oram, S., Khondoker, M., Abas, M., Broadbent, M., & Howard, L. M. (2015). Characteristics of trafficked adults and children with severe mental illness: A historical cohort study. The Lancet Psychiatry, 2(12), 1084–1091. https://doi.org/10.1016/S2215-0366(15)00290-4
Ranganathan, P., & Aggarwal, R. (2018). Study designs: Part 1 – An overview and classification. Perspectives in clinical research, 9(4), 184–186.
Song, J. W., & Chung, K. C. (2010). Observational studies: cohort and case-control studies. Plastic and reconstructive surgery, 126(6), 2234–2242. https://doi.org/10.1097/PRS.0b013e3181f44abc.
Teng, M., Wolf, M., Ofsthun, M. N., Lazarus, J. M., Hernán, M. A., Camargo, C. A., Jr., & Thadhani, R. (2005). Activated injectable vitamin D and hemodialysis survival: A historical cohort study. Journal of the American Society of Nephrology, 16(4), 1115–1125. https://doi.org/10.1681/ASN.2004070573
Further Information
- Cohort Effect
- Barrett, D., & Noble, H. (2019). What are cohort studies? Evidence-based nursing, 22(4), 95-96.
- Hess, D. R. (2004). Retrospective studies and chart reviews. Respiratory care, 49(10), 1171-1174.
- Euser, A. M., Zoccali, C., Jager, K. J., & Dekker, F. W. (2009). Cohort studies: prospective versus retrospective. Nephron Clinical Practice, 113(3), c214-c217.
