Additional file 1 of Going beyond the mean: economic benefits of myocardial infarction secondary prevention

Additional file 1: Supplementary figure legend 1. Indicators for trajectory model choice with respect to hypothesized number of distinct trajectory groups. The blue line indicates, for the validation data, the overlap between the training-based predictor and an independent k-means classification. Overall, classification overlap was in the order of 90% to 95% and peaked when four k-means groups were chosen. Supplementary figure legend 2. Cost decomposition by groups derived from the trajectory analysis, full sample. Supplementary figure legend 3. Cost decomposition by groups derived from the trajectory analysis, unexposed persons only. Supplementary Table 1. Comparison of clinical outcomes between compliers and non-compliers during observation period. Supplementary Table 2. Factors associated with compliance to 4-class secondary myocardial infarction prophylaxis (main analysis) or 3- or 4-class prophylaxis (sensitivity analysis). The results from the multivariable logistic regression model were used to calculate the inverse probability weights. Confidence intervals printed in bold face do not include 1, which indicates statistical significance at the 5% level. Supplementary Table 3. comparison of medication expenditures between compliers and non-compliers.

附加文件1：补充图注1。针对假设的不同轨迹组别数量，用于轨迹模型选择的评价指标。蓝线代表验证集数据中，基于训练集的预测器与独立k均值（k-means）分类结果的重合度。整体而言，分类重合度约为90%至95%，当选取4个k均值组别时重合度达到峰值。补充图注2：基于轨迹分析得到的组别成本分解（全样本）。补充图注3：基于轨迹分析得到的组别成本分解（仅未暴露人群）。补充表1：观察期内依从者与非依从者的临床结局对比。补充表2：与4类二级心肌梗死（myocardial infarction）预防方案依从性相关的影响因素（主分析），或3类、4类预防方案依从性相关影响因素（敏感性分析）。本研究采用多变量logistic回归模型的结果计算逆概率权重。加粗显示的置信区间不包含1，代表其在5%显著性水平下具有统计学意义。补充表3：依从者与非依从者的药物支出对比。