Department of Epidemiology, Biostatistics and Occupational Health, McGill University, Montreal, Quebec, Canada.
Centre for Clinical Epidemiology, Lady Davis Institute, Montreal, Quebec, Canada.
Pharmacoepidemiol Drug Saf. 2024 Oct;33(10):e70014. doi: 10.1002/pds.70014.
The optimal design for pharmacoepidemiologic drug-drug interactions (DDIs) studies is unclear. Using the association between concomitant use of sulfonylureas and warfarin and the risk of severe hypoglycemia as a case study, a DDI with little or no clinical impact, we tested whether the prevalent new-user design can be applied in the area.
Among all patients initiating sulfonylureas in the UK's Clinical Practice Research Datalink (1998-2020), we identified those adding-on warfarin while on a sulfonylurea. For each co-exposed patient, we defined a prescription-based exposure set including other sulfonylurea users not adding-on warfarin (comparators). Within each exposure set, we matched each co-exposed patient to five comparators on time-conditional propensity scores (TCPS) and followed them using an as-treated approach. Cox proportional hazards models estimated hazard ratios (HRs) and 95% confidence intervals (CIs) of severe hypoglycemia associated with concomitant use of sulfonylureas and warfarin compared to use of sulfonylureas alone. Sensitivity analyses addressed the impact of different potential sources of bias.
The study cohort included 17 890 patients co-exposed to sulfonylureas and warfarin and 88 749 matched comparators. After TCPS matching, patient characteristics were well-balanced between groups. Compared to use of sulfonylureas alone, concomitant use of sulfonylureas and warfarin was not associated with the risk of severe hypoglycemia (HR, 1.04; 95% CI, 0.92-1.17). Sensitivity analyses were consistent with the primary analysis (HRs ranging from 1.01 to 1.15, all not statistically significant).
Our study suggests that the prevalent new-user design could be used for the assessment of clinical effects of DDIs.
药物-药物相互作用（DDI）的最佳设计尚不清楚。本研究以磺酰脲类药物和华法林同时使用与严重低血糖风险之间的关联为例，该 DDI 具有很小或没有临床影响，旨在检验普遍的新用户设计是否适用于该领域。
在英国临床实践研究数据链（1998-2020 年）中所有开始使用磺酰脲类药物的患者中，我们确定了在使用磺酰脲类药物的同时加用华法林的患者。对于每个共同暴露的患者，我们定义了一个基于处方的暴露集，其中包括未加用华法林的其他磺酰脲类药物使用者（对照者）。在每个暴露集中，我们基于时间条件倾向评分（TCPS）将每个共同暴露的患者与五个对照者相匹配，并采用实际治疗方法对他们进行随访。Cox 比例风险模型估计了与单独使用磺酰脲类药物相比，同时使用磺酰脲类药物和华法林与严重低血糖相关的风险比（HR）和 95%置信区间（CI）。敏感性分析解决了不同潜在偏倚来源的影响。
该研究队列包括 17890 例同时暴露于磺酰脲类药物和华法林的患者和 88749 名匹配的对照者。经过 TCPS 匹配后，两组患者的特征得到了很好的平衡。与单独使用磺酰脲类药物相比，同时使用磺酰脲类药物和华法林与严重低血糖风险无关（HR，1.04；95%CI，0.92-1.17）。敏感性分析与主要分析一致（HR 范围为 1.01-1.15，均无统计学意义）。
我们的研究表明，普遍的新用户设计可用于评估 DDI 的临床效果。
