Additional file 3 of Minimizing population health loss due to scarcity in OR capacity: validation of quality of life input

Additional file 3: Fig. S1. The structure of the previously developed cohort state-transition model. Preop: preoperative state; Postop: postoperative state (6). Fig. S2. The model estimates for urgency based on the original quality of life estimates (upper panel) and the updated scores from both the original and the validation study (bottom panel). Fig. S3. The random effects of procedure on the standard deviation of the QoL estimates. These estimates are the random intercept values for procedure in a model with as independent variable the standard deviations of surgical procedures, also including hospital and pre- or postoperative as fixed effects (supplementary table 2). A random intercept above 0 indicates a higher than expected standard deviation, which we interpret as lower consensus between experts. A random intercept below 0 indicates a lower than expected standard deviation, which we interpret as higher consensus between experts. The overall standard deviation of the random effect was 0.005. Table S1. The estimates from the first mixed effects linear regression model. The dependent variable is the utility scores scored by the expert panel. Table S2. The estimates from the second mixed effects linear regression model. The dependent variable is the standard deviation of the utility scores per study center, pre- and postoperative state, and procedure. Table S3. The quality of life estimates and 95% CI derived from the original study and the validation study, stratified for preoperative and postoperative state, corresponding to figure 1 in the manuscript. Table S4. The difference in urgency of surgical procedures between the original and the updated quality of life estimates. Only the diseases which now include the new scores from the validation study are shown. This table corresponds to figure 4 in the manuscript.

补充材料3：图S1. 此前开发的队列状态转换模型（cohort state-transition model）的结构。Preop：术前状态（preoperative）；Postop：术后状态（postoperative）(6)。图S2. 模型紧急度估计结果：上半部分面板基于原始生活质量估计值生成，下半部分面板基于原始研究与验证研究的更新评分生成。图S3. 手术操作对生活质量（quality of life，QoL）估计值标准差的随机效应。该估计值为以手术操作标准差为自变量，同时纳入医院、术前/术后状态作为固定效应的混合效应模型中，手术操作对应的随机截距项（详见补充表2）。随机截距大于0代表标准差高于预期，我们将其解读为专家间共识度较低；随机截距小于0代表标准差低于预期，对应专家间共识度较高。该随机效应的总标准差为0.005。补充表S1. 首个混合效应线性回归模型（mixed effects linear regression model）的估计结果。因变量为专家小组评分的效用分值。补充表S2. 第二个混合效应线性回归模型的估计结果。因变量为各研究中心、术前与术后状态下以及不同手术操作对应的效用评分标准差。补充表S3. 来自原始研究与验证研究的生活质量（quality of life，QoL）估计值及95%置信区间（confidence interval，CI），按术前与术后状态分层，对应本文中的图1。补充表S4. 原始与更新后生活质量评分对应的手术操作紧急度差异。仅展示纳入了验证研究新评分的疾病。该表格对应本文中的图4。