OBJECTIVES: Medico-Administrative DataBases (MADBs) are an important tool in pharmacoepidemiology, for the study of drug exposure and its effects in real world. Among the bias arising from the use of MADBs, the ENCePP describes the Immeasurable Time Periods (ITP) due to hospitalizations as a main source of exposure misclassification with major consequences on the evaluation of drugs. To date, despite the increasing number of studies carried out using MADBs, the risk of bias associated with ITP is only rarely assessed. The objective of this study was to carry out a systematic review of the literature available on ITP in MADBs.

METHODS: This systematic review was carried out on the MEDLINE and CENTRAL databases, and according to the recommendations of the Cochrane group. Moreover, all references cited in the articles were explored, and a manual search was carried out.

RESULTS: Twenty-nine publications that mentioned ITP in the MADBs were selected from the 41 identified references. ITP were poorly described in the guidelines and expert opinions (n = 5). Few systematic reviews on a pathology or methodological aspects of studies on secondary data identified very few studies considering ITP. Finally, the synthesis of the 15 studies on MADBs that mentioned ITP showed an evolutionary trend in accounting for ITP, with the development of time-dependent models. Only the recent study by Oh et al. showed, by means of an external validation, the differences in efficiency indicators (Odds Ratio) compared to the Gold Standard according to the methods tested.

CONCLUSIONS: This systematic review showed that the risks of bias linked to ITP are to date insufficiently controlled in pharmacoepidemiological studies on MADBs. Only one study included external validation to assess methods considering ITPs; other studies should confirm these results before a consensus on the choice of method(s) is established.