Lessons from the Parachute Trials, Prospective vs Retrospective

By | April 26, 2011


Janus, the two headed god, traditionally guarded Roman homes because of his ability to look forward and backward at the same time.

While chronically the adventures of our poor plodding parachute research professionals is amusing, it is also informative about the nature of medical knowledge and research. To begin with, the idea, satirically put forth by the Lancet, of doing a randomized double blind study on the efficacy of parachutes highlights the diverse and multifaceted nature knowledge. It is not surprising therefore, that no single method of investigation is universally applicable. At the outset the distinction needs to be made between prospective studies and retrospective ones. They differ fundamentally in scope, expense, moral questions regarding the treatment of the participants and potential weaknesses in the data.

With respect to the first of these concerns retrospective studies are often more representative of the community as a whole. They can include large numbers of patients (see the concern over sample size in the first paragraph of the Blog on Day 1 of the Trials) and will reflect actual standards of practice in the community. On the other hand they are subject to biases and confounding factors if the study population and the control group differ somehow in an important, but unappreciated way.

Expense is another major advantage of retrospective studies (see the concerns about cost at the end of the first paragraph of the Blog on Day 4 of the Trials). Since we are studying things that have already happened, we have only to properly collect and analyze the data. This becomes more difficult and expensive as the size of the study and control populations increase, but small pilot studies involving small numbers of patients will often give clues as to the population sizes required to make a decision and the most cost effective way to assemble the data.

Moral questions are rarely a major issue (see the concern for the safety of the students in the second paragraph of the Blog on Day 4 of the Trials), since the events we study are in the past. Consent to participate is not usually necessary, since patients are presumably being given the standard diagnostic work up and treatment.

Finally, with regard to the potential weaknesses of the data, it is important to remember that weaknesses are present in both types of research, but they differ in type. Retrospective studies are very good at finding relatively large differences between two types of populations, but often cannot say exactly what caused these differences. Follow up studies can help narrow the list of potential causes, but may not be able to definitively assign risk. Prospective studies are better at assessing the relative contributions of several causal events, but results are reliable only for the populations actually studied. If the study populations differ in any important way from the population as a whole or even any important subset of that population, then extrapolations to the real world can be just as dangerous as with retrospective studies (see the objection of the editors in the second sentence of the last paragraph of the Blog on Day 4 of the Trials).

Leave Your Comment