Medlineに登録された傾向スコアを用いた論文数 t oun C Calendar Year 『Some journals and their reviewers seem to be ‘enamored’ with propensity scores and may request such analyses as a ‘better tool’ to control for confounding.』 2012 John Seegerを改編して引用
Exposure propensity score
Patients never Patients always treated with study drug treated with study drug s ct je ubs of % 0 0.5 1 = treated with study drug = treated with comparison drug 2012 John Seegerを改編して引用
PS distribution after matching
Patients never Patients always treated with study drug treated with study drug s ct bje su of % 0 0.5 1 = treated with study drug = treated with comparison drug 2012 John Seegerを改編して引用
1. Survival among high-risk patients after bariatric surgery. Maciejewski ML LE. JAMA. 2011; 305: 2419–26. doi:10.1001/jama.2011.817.
2. Resource Utilization and Costs of Schizophrenia Patients Treated with Olanzapine versus Quetiapine in a Medicaid Population. Yu AP, Atanasov P, Ben‐Hamadi R, Birnbaum H, Stensland MD, Philips G. Value in Health. 2009; 12: 708– 15.
Diagram of Surgical Options. Image credit: Walter Pories, M.D. FACS. NIDDK
■研究デザイン Retrospective cohort study in 退役軍人医療センター
Patients 850人 Nonsurgical controls 41244人
VA indicates Veterans Affairs; BMI, BMI indicates body mass index; FY, body mass index; and DCG, fiscal year; VA, Veterans Affairs; and diagnostic cost group (a measure DCG, diagnostic cost group (a measure of comorbidity burden). of comorbidity burden).
『Unadjusted differences between patients undergoing or not undergoing bariatric surgery were compared with χ2 tests for categorical variables and 2-tailed unpaired t tests for continuous variables in the unmatched cohorts, McNemar tests and paired t tests in the matched cohorts,
and standardized differences to enable comparison of covariate imbalance between the matched and unmatched cohorts.
The association between bariatric surgery and al -cause mortality was examined in the unmatched cohorts with crude mortality rate comparisons and unadjusted and multivariable Cox proportional hazards regression models.』
Multivariable Cox model
※併存症の評価方法 Diagnostic cost group scores
『The diagnostic cost group score aggregates inpatient and outpatient diagnoses in the year before baseline, with scores greater than 1.0 implying above-average expected expenditures and scores less than 1.0 implying below-average expected expenditures. Diagnostic cost group scores were as predictive of veterans‘ 1-year mortality as other comorbidity scores and were highly predictive of mortality, use, and expenditures in bariatric surgery.』
『In a third analysis, we accounted for the nonequivalence (eg, selection bias) of the nonsurgical control cohort via propensity score matching with logistic regression….
A propensity score represents the predicted probability that a given patient will undergo bariatric surgery, and patients who had bariatric surgery procedures were matched to controls with a greedy algorithm. The propensity score model included interaction of age, age squared, diagnostic cost group, BMI, BMI squared, BMI cubed, sex, race, marital status, and Veterans Integrated Service Networks, as well as numerous 2-way interactions, and had a concordance index of 0.85.』
『 Despite the large (n = 41 244) sample of controls, we conducted one-to-one matching to avoid the possible bias of many-to-one matching. Each surgical case patient was matched to a single nonsurgical control if their predicted propensity scores were identical to 8 digits. If such a match was not found, the case patient was matched to a control on the basis of a 7-, 6-, 5-, 4-, 3-, 2-, or 1-digit match. This process matched 847 surgical case patients (of 850 possible; 99.6%) to 847 nonsurgical controls, and covariate balance between matched surgical case patients and nonsurgical controls was assessed via McNemar tests, 2-tailed paired t tests, and standardized differences. We then conducted an unadjusted Cox regression stratified on the matched pairs to account for the lack of independence between cohorts induced by matching and a Cox regression adjusted for differences in the year indicating zero time. Alternative propensity score analyses from many-to-one matching generated similar results, so we present the 1:1 matching results here. 』
Resource Utilization and Costs of Schizophrenia Patients Treated with Olanzapine versus Quetiapine in a Medicaid Population. AP, Atanasov P, Ben‐Hamadi R, Birnbaum H, Stensland MD, Philips G. Value in Health. 2009; 12: 708–15.
メディケイド被保険者の統合失調症患者で、 オランザピン vs クエチアピンの 年間医療コストと医療資源消費を比較
『The following process was applied to implement the propensity score method with optimal matching. First, patients’ baseline characteristics were profiled during a 12-month preindex period. These characteristics were compared between the two study cohorts. Second, propensity scores were generated using logistic regressions and given patient baseline covariates, which were selected using backward selection. Third, optimal matching was applied to match olanzapine and quetiapine patients based on propensity score using a SAS …. Finally, baseline characteristics were compared for the full and the matched cohorts over the 12-month preindex period. Once the propensity score matching process was completed, outcomes were compared using paired t tests for continuous variables, and McNemar’s tests for categorical variables for matched pairs. Each cost component was studied separately. 』