The quantity, reliability, transparency, reporting, and interpretation of pharmacovigilance signal detection studies in pregnancy: a meta-epidemiological study

To systematically review the characteristics of the available pharmacovigilance signal detection studies in pregnancy, and comprehensively assess the reliability, transparency, reporting, and interpretation of these studies. We searched five databases from inception to February 2024 to identify the available pharmacovigilance signal detection studies in pregnancy. We extracted three aspects of information (basic information, data processing modes, signal detection analyses) to assess the reliability, transparency, and reporting of each study. Moreover, we adopted the criteria of Mouffak et al.'s study to assess the misinterpretation of signal detection results in these studies. A total of 33 pharmacovigilance signal detection studies in pregnancy were identified. Among them, there were great methodological heterogeneities in the data processing modes (restriction to the population, comparator, standardization of drug names and adverse event names, the assigned roles of drugs, counting unit, etc.) and signal detection analyses (signal detection method, sensitivity analysis, subgroup analysis, adjustment for confounding factors, etc.). Moreover, 13 (39%) studies had at least one type of inappropriate interpretation and/or extrapolation of signal detection results. Our results reveals that the quantity of pharmacovigilance signal detection studies in pregnancy is relatively limited. Furthermore, the reliability, transparency, reporting, and interpretation of the existing studies are less optimistic. The main issues existing in the available pharmacovigilance signal detection studies in pregnancy consist of two aspects: (1) great methodological heterogeneities exist in the data processing modes and signal detection analyses among different studies and (2) inappropriate interpretation and extrapolation of signal detection results are frequent.