Permanent user bias in case-Crossover studies in pharmacoepidemiology

J. Hallas, S. V. Wang, J. J. Gagne, S. Schneeweiss, Anton Pottegård

Research output: Contribution to journalConference abstract in journalResearchpeer-review


Background: In pharmacoepidemiology, the case- crossover design is based on cases that have contrasting drug exposure at the time of an event and at a reference time in the past. If the drug in question should be taken permanently, only certain exposure patterns will occur. These patients cannot be unexposed at the event time and exposed at the reference time, while the opposite pattern can occur if the drug was initiated recently. The resulting odds ratio (OR) would thus be biased upward. As many drugs have a subpopulation of permanent users, this bias might pervade many case-crossover analyses of drug effects. Objectives: The aims of this study were to demonstrate this "permanent user bias" and to evaluate whether it can be remedied by including a control group (case-time-control design). Methods: Using nationwide Danish data resources, we conducted case-crossover and case-time-control analyses for combinations of three exposures that are often intended to be used permanently (statins, insulin, and thyroxin) and three outcomes (retinal detachment, wrist fracture, and ischemic stroke), where the true causal relations were expected to be null. Controls were matched on age, gender, and index date, and exposure was ascertained at 2-month intervals over the preceding 12 months. Results: For retinal detachment, the case-crossover OR was 1.60 (95% confidence interval (CI): 1.42-1.80) for statins, 1.40 (CI: 1.02-1.92) for thyroxin, and 1.53 (CI: 1.04-2.24) for insulin. Estimates for the control population were nearly identical, leading to near-null case-time-control estimates for the three drug classes. For the wrist fracture and stroke outcomes, case-time-control ORs were consistently above unity (1.09, 1.51, and 1.15 for wrist fracture, and 2.27, 1.87, and 1.67 for stroke), suggesting significant residual bias. Conclusions: In case-crossover studies of drugs, permanent users confer a moderate bias upward, which is partly remedied by using a control group. Additional research is needed to identify the optimal strategy for selecting this control group.
Original languageEnglish
Article number2
JournalPharmacoepidemiology and Drug Safety
Issue numberS1
Pages (from-to)1-2
Publication statusPublished - 2015
Event31st International Conference on Pharmacoepidemiology and Therapeutic Risk Management - Boston, United States
Duration: 22. Aug 201526. Aug 2015


Conference31st International Conference on Pharmacoepidemiology and Therapeutic Risk Management
CountryUnited States

Cite this