Immortal time bias is a systematic error that occurs in epidemiologic research when an inappropriate analytical approach is used that gives the treatment group an unfair advantage relative to the comparison group. In the process, this systematic error (or bias) makes treatments look very protective and exaggerates treatment benefits. This bias is a well-established problem in the medical literature in general, and recent studies have examined its presence in pregnancy research. However, it typically occurs when comparing patients who receive a medication to those who do not. Little is known about its occurrence in studies comparing the safety of two medications in pregnancy. This methodological study will describe the occurrence of immortal time bias in such studies and illustrate the impact of its presence on estimated treatment effects using two case studies. The first case study will compare the rates of fetal death (spontaneous abortion or stillbirth) with the drug ondansetron (a drug used to decrease nausea) versus that of other anti-nausea drugs. The second case study will compare the rates of fetal death with fluconazole (a drug used to treat fungal infections) versus that of other antifungal drugs.
Immortal time bias is a systematic error that occurs in epidemiologic research due to the misclassification or exclusion of unexposed event-free person-time due to the use of time-fixed rather than time-dependent analyses. This bias typically occurs when comparing exposed to unexposed patients, and it biases estimates downward, often resulting in spuriously protective associations. Most recently, the presence of immortal time bias has been examined in perinatal and obstetrical epidemiology, an area where the use of time-dependent methods is particularly important given the inherent timing of various pregnancy outcomes. However, this previous research has focused on comparisons of exposed to unexposed women, and the impact of immortal time bias in drug safety studies involving an active comparator in pregnancy remains unknown. To address this methodological issue, we will compare the results of time-fixed and time-dependent analyses for two case studies, both of which will examine the outcome of fetal death, defined as a composite endpoint of spontaneous abortion or stillbirth. The first case study will compare the risk of fetal death with ondansetron to that of other antiemetics, and the second case study will compare the risk of fetal death with fluconazole to that of other antifungal medications. To minimize potential confounding, all analyses will adjust for high-dimensional propensity scores.
Fetal death, defined as a composite of spontaneous abortion or stillbirth.
Samy Suissa - Chief Investigator - Sir Mortimer B Davis Jewish General Hospital
Kristian Filion - Corresponding Applicant - McGill University
Colin Dormuth - Collaborator - McGill University
Josselin Cabaussel - Collaborator - Sir Mortimer B Davis Jewish General Hospital
Michael Paterson - Collaborator - Institute for Clinical Evaluative Sciences ( ICES )
pauline reynier - Collaborator - Sir Mortimer B Davis Jewish General Hospital
Robert Platt - Collaborator - McGill University
HES Admitted Patient Care;ONS Death Registration Data;Practice Level Index of Multiple Deprivation;Pregnancy Register