Terminology in (pharmaco)epidemiology is not always stringent. It is likely a consequence of the constantly evolving methodology and the cross disciplinary nature of epidemiology and, to some extent, something we must accept. Nevertheless, it is our opinion that we, as a research community, have a responsibility to aim for unambiguity in the reporting of our research. We need to ensure a satisfactory level of informativeness when it comes to deep-rooted terminology that influences the immediate understanding of our research and the judgment of its quality. The current use of the terms retrospective and prospective opposes this aim; the two terms are highly value-laden and ingrained in pharmacoepidemiology despite repeated discouragements of their use.

In our research group, we are currently performing a literature review where we collect pharmacoepidemiologic studies investigating drug effectiveness.1 During the screening process, it is necessary to understand the exact study design used in each study. We were surprised by the number of studies where the terms retrospective and prospective were used, how inconsistent the terms were used, and how much confusion they introduced. The review is limited to studies published in 2019, so the included studies represent contemporary pharmacoepidemiology.

First, a short recap of the terminology. When a study is described as either a retrospective or a prospective study, often different, and sometimes conflicting, meanings can be the intention. According to Modern Epidemiology by Rothman, the following two definitions are at play: (1) The terms are used to describe whether exposure is recorded before (prospective) or after (retrospective) the outcome event has happened.2 The latter may introduce recall bias. (2) The terms are also used to describe whether a study is designed before or after accumulation of person-time (regardless of how data are collected).2 In this understanding, a retrospective study is designed after the accumulation of person-time, and a prospective study is designed before the accumulation of person-time.

Thus, a study can be both retrospective and prospective, if these terms describe different aspects of the study. However, there are ambiguities even within each definition. For the first definition, it is unclear how to label the study design if some data is recorded before and some after the events happen. Also, the meaning of “recorded” is unclear—does the recording have to be by the researchers, or can it be a study-independent routine collection? For the second definition, a study can be designed before person-time accumulates, but it may have been designed for a different purpose—nonetheless, this study design is typically labeled as prospective. Furthermore, there may be other definitions, which could add to the confusion.

One may argue that the ambiguity with regards to the meaning of prospective and retrospective is then a matter of settling on one of the definitions above and resolving the remaining ambiguities within each definition. However, decades of epidemiological research and methodological development have passed without achieving this clarity. Furthermore, the strong and unfounded connotations of the two words, that is, a prospective study being inherently better than a retrospective study, are very difficult to overcome. As such, we do not think that the terminology of prospective and retrospective can be salvaged.

We are not the first to advise against these terms. Already in 1999, in a BMJ paper with the title “Prospective or retrospective: What's in a name? The shorthand classification of a study may create confusion”3 Vandenbroucke stated: “The words prospective and retrospective have lost all meaning. From spotting them in the abstract the reader gains no insight into the type of research that was performed.” Vandenbroucke concluded that use of the two terms should be abandoned and replaced by specific descriptions. The STROBE guidelines4 refer to the BMJ paper and also advice against simple use of the terms to describe a study: “In STROBE, we do not use the words prospective and retrospective, nor alternatives such as concurrent and historical. We recommend that, whenever authors use these words, they define what they mean. Most importantly, we recommend that authors describe exactly how and when data collection took place.”

Concerns over the use of prospective and retrospective have also been expressed in pharmacoepidemiology context. The use of prospective and retrospective was discouraged in Pharmacoepidemiology and Drug safety in the ISPE-endorsed guidance on self-controlled study designs due to the common misinterpretation of retrospective studies as inferior to prospective studies.5 Nonetheless, several papers describing their design as prospective or retrospective have since been published in Pharmacoepidemiology and Drug safety.

It is unclear to us why the use of prospective and retrospective is still so common. Perhaps, it is simply a matter of habit and convention. Another possibility is that some researchers consider the terminology useful. However, we have not seen any counter-arguments for why prospective and retrospective remain useful descriptors of an epidemiological study design.

Therefore, dear fellow pharmacoepidemiologists, our advice is to refrain from using the terms retrospective and prospective. Instead, we would like to encourage a more precise description of the study design and data collection.