Association between tramadol use and seizures: A nationwide case–case‐time‐control study

Abstract

Purpose: Tramadol may lower the seizure threshold; however, there is no conclusive evidence to confirm this. This study aimed to determine whether the use of tramadol is associated with the occurrence of seizures. Methods:We conducted a case-case-time-control (CCTC) study by identifying patients who had received tramadol and seizure diagnosis in a nationwide healthcare database in South Korea between 2003 and 2015. Each case was matched for age and sex to one future case to adjust for time trends in exposure without selection bias from the use of an external control group. The use of tramadol was assessed during a risk period of 1-30 days, and two reference periods, 61-90 days and 91-120 days, preceding the first diagnosis of seizures. We calculated the adjusted odds ratio (aOR) by dividing the OR in cases (case-crossover) by the OR in future cases (control-crossover).We performed a dose-response analysis using the average daily dose. Results:We identified 2523 incident cases with matched future cases (mean age, 45.4 years; 50% men). The aOR for seizure with tramadol use was 0.94 (95% confidence interval [CI], 0.98-1.43) in the CCTC analysis, with a case-crossover OR of 1.19 (0.98-1.43) and control-crossover OR of 1.27 (1.03-1.56). The dose-response analysis showed a similar trend in the main analysis: a low-dose aOR of 0.80 (0.50-1.28) and a high-dose aOR of 0.92 (0.41-2.11). Conclusion:We could not identify a significant association between transient use of tramadol and incidence of seizures in clinical practice. Key Points• The effect of tramadol use on seizures remains unclear.• Reports have linked tramadol use to seizures with a potential mechanism, though the results between studies are inconsistent.• We conducted a nationwide case-based analysis that considered control-selection bias and minimized confounding by time-invariant factors.• We used the case-case-time-control design among case-crossover variants to adjust for bias from time trends of exposure in the case-crossover approach by using future-case control group.

Logistic regression analysis of the value of biomarkers, clinical symptoms, and imaging examinations in COVID-19 for SARS-CoV-2 nucleic acid detection

“…The case-case-time-control design is a case-only design. [11][12][13] By selecting future cases as controls, selection bias can be reduced, but there is also a certain time effect bias. Therefore, this method is mainly applicable to studies where the outcome variables will not affect the next outcome.…”