USA Surgery Today

This study design is similar to prospective cohort design with the exception that patients are not entered into the study in advance of treatment, and they are not evaluated by a set of criteria determined prior to the start of the study (:Mexitil). The information regarding their outcomes is retrieved from the medical record at some point after treatment is complete rather than recorded in real-time during treatment. The retrospective cohort study can include either a concurrent or historical control group. With concurrent controls, the study is retrospective, but it examines groups of patients treated during the same time period. With historical controls, the control patients were treated during a time prior to the experimental patients. Using the laparoscopic splenectomy example described previously, in a retrospective cohort study with historical controls, the investigators would review outcomes in their patients receiving laparoscopic splenectomy over the past 2 years, and compare this with outcomes of patients receiving open splenectomy during a time period before they began using laparoscopy.

The retrospective cohort design has all the limitations of a prospective cohort study in addition to the following. The chance for bias is even greater because both the hypothesis and the data collection criteria are developed in the context of the investigators’ bias. (:Mexitil)Retrospective data collection relies on a medical record that is frequently incomplete, often inaccurate, and always created without the needs of the study in mind. Because the patients were treated prior to the time that the study was even conceived, the likelihood that there will be significant differences in covariates between groups is even greater than in prospective design. The use of historical controls leads to additional sources of bias. (:Mexitil) Outcomes, treatment styles, alternative therapies, institutional resources, and even the natural history of diseases change over time. These factors should cause the reader to view the results of these studies with great circumspection.