Propensity Score Trimming Research Articles

High-dimensional Iterative Causal Forest (hdiCF) for Subgroup Identification Using Health Care Claims Data.

We recently developed a machine-learning subgrouping algorithm, iterative causal forest (iCF), to identify subgroups with heterogeneous treatment effects (HTEs) using predefined covariates. However, such predefined covariates may miss or poorly define important features leading to inaccurate subgrouping. To address such limitations, we developed a new semi-automatic subgrouping algorithm, hdiCF, which adapts methodology from high-dimensional propensity score for feature recognition in claims data. The hdiCF algorithm has 3 steps: 1) high-dimensional feature identification by International Classification of Diseases, Current Procedural Terminology, and Anatomical Therapeutic Chemical codes (in/outpatient diagnoses, procedures, prescriptions) and creation of ordinal variables by frequency of occurrence; 2) propensity score trimming and high-dimensional feature preparation; 3) iCF implementation to identify subgroups. We applied hdiCF in a 20% random sample of fee-for-service Medicare beneficiaries who initiated sodium-glucose cotransporter-2 inhibitors (SGLT2i) or glucagon-like peptide-1 receptor agonists to identify subgroups with HTEs for incidence of hospitalized heart failure. HdiCF findings were consistent with studies suggesting SGLT2i to be more beneficial for patients with pre-existing heart failure or chronic kidney disease. HdiCF is not dependent on prior hypotheses about HTEs and identifies subgroups with markers for potential HTEs in real-world evidence studies where active-comparator, new-user study designs limit the potential for unmeasured confounding.

Glucagon-like Peptide 1 Receptor Agonists and Asthma Exacerbations: Which Patients Benefit Most?

Rationale: Although recent evidence suggested that glucagon-like peptide 1 receptor agonists (GLP1RAs) might reduce the risk of asthma exacerbations, it remains unclear which subpopulations might derive the most benefit from GLP1RA treatment. Objectives: To identify characteristics of patients with asthma that predict who might benefit the most from GLP1RA treatment using real-world data. Methods: We implemented an active-comparator, new-user design analysis using commercially ensured patients 18-65 years of age from MarketScan data for 2007-2019 and identified two cohorts: GLP1RAs versus thiazolidinediones and GLP1RAs versus sulfonylureas. The outcome was acute exacerbation of asthma (hospital admission or emergency department visit for asthma) within 180 days after initiation. We applied iterative causal forest, a novel causal machine learning subgrouping algorithm, to assess heterogeneous treatment effects. In identified subgroups, we predicted propensity score, conducted propensity score trimming, and then estimated adjusted risk differences for the effect of GLP1RAs relative to comparators on asthma exacerbation using inverse probability treatment weighting in the propensity score-trimmed subpopulation. Results: Among 10,989 patients initiating GLP1RAs or thiazolidinediones and 17,088 patients initiating GLP1RAs versus sulfonylurea, GLP1RA initiators had fewer exacerbations, with adjusted risk differences of -0.5% (95% confidence interval [CI], -1.1% to 0.1%) and -1.6% (95% CI, -2.2% to -1.1%), respectively. In the GLP1RA versus sulfonylurea cohort, in which we observed a beneficial effect, our iterative causal forest analysis identified five subgroups with different treatment effects, defined by the number of emergency department visits, the number of prescriptions for short-acting β2-agonists, the number of prescriptions for inhaled steroids and long-acting β-agonists (either combination therapy or concurrent use), and age ≥ 50 years. Among these, patients with two or more emergency department visits during the 12-month baseline period had the largest absolute exacerbation risk reduction, with a decrease of 2.8% for GLP1RAs (95% CI, -4.8% to -0.9%). Conclusions: GLP1RAs demonstrated a beneficial effect on reducing asthma exacerbation relative to sulfonylureas. Patients with asthma with two or more emergency department visits (a proxy for disease severity) benefit most from GLP1RAs. Emergency department visit frequency, the number of maintenance and reliever inhalers, and age might help individualize prediction of the short-term benefit of GLP1RAs on asthma exacerbation.

Propensity Scores in Health Disparities Research: The Example of Cognitive Aging and the Hispanic Paradox.

Individuals of Mexican ancestry in the United States experience substantial socioeconomic disadvantages compared with non-Hispanic white individuals; however, some studies show these groups have similar dementia risk. Evaluating whether migration selection factors (e.g., education) associated with risk of Alzheimer disease and related dementia (ADRD) explain this paradoxical finding presents statistical challenges. Intercorrelation of risk factors, common with social determinants, could make certain covariate patterns very likely or unlikely to occur for particular groups, which complicates their comparison. Propensity score (PS) methods could be leveraged here to diagnose nonoverlap and help balance exposure groups. We compare conventional and PS-based methods to examine differences in cognitive trajectories between foreign-born Mexican American, US-born Mexican American, and US-born non-Hispanic white individuals in the Health and Retirement Study (1994-2018). We examined cognition using a global measure. We estimated trajectories of cognitive decline from linear mixed models adjusted for migration selection factors also associated with ADRD risk conventionally or with inverse probability weighting. We also employed PS trimming and match weighting. In the full sample, where PS overlap was poor, unadjusted analyses showed both Mexican ancestry groups had worse baseline cognitive scores but similar or slower rates of decline compared with non-Hispanic white adults; adjusted findings were similar, regardless of method. Focusing analyses on populations where PS overlap was improved (PS trimming and match weighting) did not alter conclusions. Attempting to equalize groups on migration selection and ADRD risk factors did not explain paradoxical findings for Mexican ancestry groups in our study.

Post-Authorization Safety Studies of Acute Liver Injury and Severe Complications of Urinary Tract Infection in Patients with Type 2 Diabetes Exposed to Dapagliflozin in a Real-World Setting.

At the time of dapagliflozin's approval in Europe (2012) to treat patients with type 2 diabetes mellitus, concerns regarding acute liver injury and severe complications of urinary tract infection (sUTI) led to two post-authorization safety (PAS) studies of these outcomes to monitor the safety of dapagliflozin in real-world use. To investigate the incidence of hospitalization for acute liver injury (hALI) or sUTI (pyelonephritis or urosepsis) among patients initiating dapagliflozin compared with other glucose-lowering drugs (GLDs). These two noninterventional cohort studies identified initiators of dapagliflozin and comparator GLDs in November 2012-February 2019 using data from three longitudinal, population-based data sources: Clinical Practice Research Datalink (UK), the HealthCore Integrated Research Database (USA), and the Medicare database (USA). Outcomes (hALI and sUTI) were identified with electronic algorithms. Incidence rates were estimated by exposure group. Incidence rate ratios (IRRs) were calculated comparing dapagliflozin to comparator GLDs, using propensity score trimming and stratification to address confounding. The sUTI analyses were conducted separately by sex. In all data sources, hALI and sUTI incidence rates were generally lower in dapagliflozin initiators than comparator GLD initiators. The adjusted IRR (95% confidence interval) pooled across data sources for hALI was 0.85 (0.59-1.24) and for sUTI was 0.76 (0.60-0.96) in females and 0.74 (0.56-1.00) in males. Findings from sensitivity analyses were largely consistent with the primary analyses. These real-world studies do not suggest increased risks of hALI or sUTI, and they suggest a potential decreased risk of sUTI with dapagliflozin exposure compared with other GLDs.

Post-Authorization Safety Study of Hospitalization for Acute Kidney Injury in Patients with Type 2 Diabetes Exposed to Dapagliflozin in a Real-World Setting.

Dapagliflozin is a sodium-glucose cotransporter 2 inhibitor approved to treat type2 diabetes mellitus (T2DM), among other conditions. When dapagliflozin was approved in Europe for treating T2DM (2012), potential safety concerns regarding its effect on kidney function resulted in this post-authorization safety study to assess hospitalization for acute kidney injury (hAKI) among dapagliflozin initiators in a real-world setting. The aim of this study was to evaluate the incidence of hAKI in adults with T2DM initiating dapagliflozin compared with other glucose-lowering drugs (GLDs). This noninterventional cohort study identified new users of dapagliflozin and comparator GLDs from November 2012 to February 2019 from three longitudinal, population-based data sources: Clinical Practice Research Datalink (CPRD; United Kingdom), the HealthCore Integrated Research Database (HIRD; United States [US]), and Medicare (US). Electronic algorithms identified occurrences of hAKI, from which a sample underwent validation. Incidence rates for hAKI were calculated, and incidence rate ratios (IRRs) compared hAKI in dapagliflozin with comparator GLDs. Propensity score trimming and stratification were conducted for confounding adjustment. In all data sources, dapagliflozin initiators had a lower hAKI incidence rate than comparator GLD initiators (adjusted IRRs: CPRD, 0.44 [95% confidence interval (CI), 0.22-0.86]; HIRD, 0.76 [95%CI, 0.62-0.93]; Medicare, 0.69 [95% CI, 0.59-0.79]). The adjusted IRR pooled across the data sources was 0.70 (95% CI, 0.62-0.78). Results from sensitivity and stratified analyses were consistent with the primary analysis. This study, with > 34,000 person-years of real-world dapagliflozin exposure, suggests a decreased risk of hAKI in patients with T2DM exposed to dapagliflozin, aligning with results from dapagliflozin clinical trials. European Union Post-Authorisation Studies Register, EUPAS11684; ClinicalTrials.gov, NCT02695082.

14-OR: Risk of Acute Kidney Injury in Real-World Use of Dapagliflozin

A noninterventional postauthorization safety study in adult patients with type 2 diabetes compared the incidence of hospitalized diagnoses of acute kidney injury (hAKI) in new users of dapagliflozin and new users of comparator glucose-lowering drugs (GLDs), matched by index year, age, sex, and region. Comparators included GLDs other than SGLT2 inhibitors or monotherapy of insulin, metformin, or sulfonylureas. Data from US Medicare (2014-2017), US HealthCore Integrated Research Database (HIRD) (2014-2019), and UK Clinical Practice Research Datalink (CPRD) (2012-2018) were included. Incidence rates of hAKI were calculated by exposure group and compared with adjusted incidence rate ratios (aIRRs) with covariate adjustment by propensity score trimming and stratification. A pooled aIRR was estimated by the Mantel-Haenszel method. The number of included person-years of dapagliflozin and comparator exposure, respectively, was 12,389 and 30,737 in CPRD, 12,575 and 111,919 in HIRD, and 9,208 and 141,313 in Medicare. The mean age (years) was 57 in CPRD, 52 in HIRD, and 71 in Medicare. Approximately 41% in CPRD, 47% in HIRD, and 50% in Medicare were women. All aIRRs and upper 95% confidence limits (CI) were below the null (figure); the pooled aIRR was 0.70 (95% CI, 0.62-0.78). This real-world study shows a decreased risk of hAKI associated with dapagliflozin compared with other GLDs and aligns with results from dapagliflozin clinical trials. Disclosure C. Johannes: Other Relationship; Self; AstraZeneca. H. Chen: None. A. Gilsenan: Other Relationship; Self; AstraZeneca. J. Layton: Other Relationship; Self; AstraZeneca. D. C. Beachler: Employee; Self; Anthem Inc. (HealthCore). R. M. Ziemiecki: Research Support; Self; Astellas Pharma Inc., AstraZeneca, Boehringer Ingelheim International GmbH, Janssen Pharmaceuticals, Inc., Pfizer Inc. L. Li: Other Relationship; Self; Pfizer Inc. H. E. Danysh: Other Relationship; Self; AstraZeneca. J. Dinh: None. P. R. Hunt: Employee; Self; AstraZeneca, Stock/Shareholder; Self; AstraZeneca. C. Karlsson: Employee; Self; AstraZeneca, Stock/Shareholder; Self; AstraZeneca.

805-P: Risk of Severe Complications of Urinary Tract Infection in Real-World Use of Dapagliflozin

A noninterventional postauthorization safety study in patients with type 2 diabetes compared the sex-specific incidence of severe complications of urinary tract infections (sUTI) (pyelonephritis or urosepsis) in new users of dapagliflozin and new users of other comparator glucose-lowering drugs (GLDs), matched by index year, age, and region. Comparators included GLDs other than SGLT2 inhibitors, or monotherapy of insulin, metformin, or sulfonylureas. Data from US Medicare (2014-2017), US HealthCore Integrated Research Database (HIRD) (2014-2019), and UK Clinical Practice Research Datalink (CPRD) (2012-2018) were analyzed. Incidence rates of sUTI were compared by exposure group with adjusted incidence rate ratios (aIRRs) with covariate adjustment by propensity score trimming and stratification. Pooled aIRRs were estimated by the Mantel-Haenszel method. The total number of person-years of dapagliflozin and comparator exposure, respectively, was 17,265 and 161,176 for females and 22,594 and 194,388 for males. Mean age (years) in CPRD, HIRD, and Medicare was 58, 52, and 72 in females and 59, 52, and 71 in males, respectively. All aIRR estimates were below the null but imprecise (figure). The pooled aIRR was 0.76 (95% CI, 0.60-0.96) in females and 0.74 (95% CI, 0.56-1.00) in males. This real-world study did not find an increased risk of sUTI in females or males with dapagliflozin compared with other GLDs. Disclosure C. Johannes: Other Relationship; Self; AstraZeneca. H. Chen: None. A. Gilsenan: Other Relationship; Self; AstraZeneca. J. Layton: Other Relationship; Self; AstraZeneca. D. C. Beachler: Employee; Self; Anthem Inc. (HealthCore). R. M. Ziemiecki: Research Support; Self; Astellas Pharma Inc., AstraZeneca, Boehringer Ingelheim International GmbH, Janssen Pharmaceuticals, Inc., Pfizer Inc. L. Li: Other Relationship; Self; Pfizer Inc. H. E. Danysh: Other Relationship; Self; AstraZeneca. J. Dinh: None. P. R. Hunt: Employee; Self; AstraZeneca, Stock/Shareholder; Self; AstraZeneca. C. Karlsson: Employee; Self; AstraZeneca, Stock/Shareholder; Self; AstraZeneca.

Association between bariatric surgery and risks of cancer among Chinese patients with type 2 diabetes mellitus: A retrospective cohort study.

To examine risks of cancers, obesity-related cancers (eg, cancers in digestive organs, breast, ovary, kidney, thyroid, and myeloma), cancer-related mortality, and all-cause mortality in patients with type 2 diabetes mellitus (T2DM) and obesity who underwent bariatric surgery. A retrospective cohort of 1944 T2DM patients with obesity (345 bariatric surgery patients and 1599 matched controls) who were free of cancer from 2006 to 2017 was assembled. One-to-five propensity score matching followed by propensity score trimming was used to balance baseline covariates. During a mean follow-up period of 37 months, there are risks that in 3.2%, 1.4%, 0.9%, and 3.2% of bariatric patients cancer, obesity-related cancer, cancer-related mortality, and all-cause mortality, respectively, would occur. Surgical patients were found to have reduced incidence rates (IRs) of obesity-related cancer (0.531/100 person-years, 95% confidence interval [CI]: 0.172-1.238/100 person-years) and cancer of breast and genital organs (0.394/100 person-years, 95% CI: 0.048-1.424/100 person-years) than matched control patients whose IRs for obesity-related cancer and cancer of breast and genital organs were 0.627/100 person-years (95% CI: 0.426-0.889/100 person-years) and 0.521/100 person-years (95% CI: 0.277-0.891/100 person-years), respectively. Patients in the surgical group had a significant reduction in risk of all-cause mortality (hazard ratio [HR]=0.508, P=.041). Effects of bariatric surgery on any cancers (HR=1.254, P=.510), obesity-related cancers (HR=0.843, P=.724), and cancer mortality (HR=1.304, P=.694) were not significant. Bariatric surgery was not associated with risks of overall cancer, obesity-related cancer, and cancer mortality among T2DM patients with obesity at 3 years.

Propensity score trimming mitigates bias due to covariate measurement error in inverse probability of treatment weighted analyses: A plasmode simulation.

Inverse probability of treatment weighting (IPTW) may be biased by influential observations, which can occur from misclassification of strong exposure predictors. We evaluated bias and precision of IPTW estimators in the presence of a misclassified confounder and assessed the effect of propensity score (PS) trimming. We generated 1000 plasmode cohorts of size N=10 000, sampled with replacement from 6063 NHANES respondents (1999-2014) age 40 to 79 with labs and no statin use. We simulated statin exposure as a function of demographics and CVD risk factors; and outcomes as a function of 10-year CVD risk score and statin exposure (rate ratio [RR]=0.5). For 5% of the people in selected populations (eg, all patients, exposed, those with outcomes), we randomly misclassified a confounder that strongly predicted exposure. We fit PS models and estimated RRs using IPTW and 1:1 PS matching, with and without asymmetric trimming. IPTW bias was substantial when misclassification was differential by outcome (RR range: 0.38-0.63) and otherwise minimal (RR range: 0.51-0.53). However, trimming reduced bias for IPTW, nearly eliminating it at 5% trimming (RR range: 0.49-0.52). In one scenario, when the confounder was misclassified for 5% of those with outcomes (0.3% of cohort), untrimmed IPTW was more biased and less precise (RR=0.37 [SE(logRR)=0.21]) than matching (RR=0.50 [SE(logRR)=0.13]). After 1% trimming, IPTW estimates were unbiased and more precise (RR=0.49 [SE(logRR)=0.12]) than matching (RR=0.51 [SE(logRR)=0.14]). Differential misclassification of a strong predictor of exposure resulted in biased and imprecise IPTW estimates. Asymmetric trimming reduced bias, with more precise estimates than matching.

Association of Gabapentinoids With the Risk of Opioid-Related Adverse Events in Surgical Patients in the United States

The use of gabapentinoids in multimodal postoperative analgesia is increasing; however, when coadministered with opioids, these drugs may potentiate central nervous system and respiratory depression. To evaluate the association between perioperative coadministration of gabapentinoids and opioids with inpatient opioid-related adverse events in surgical patients. This cohort study used propensity score trimming, stratification, and weighting of adults admitted for a major surgery between October 2007 and December 2017 who were treated with opioids on the day of surgery and included in the Premier Research database. Data analysis was conducted from February to April 2020. Gabapentinoids (gabapentin or pregabalin) coadministered with opioids starting the day of surgery vs opioid therapy without gabapentinoids. Primary outcome was opioid overdose. Secondary outcomes included respiratory complications, unspecified adverse effects of opioid use, and a composite of these 3 outcomes. Patients were followed up for as long as 30 days from the day of surgery until deviation from the initial treatment regimen or discharge. Gabapentinoids with opioids were administered to 892 484 of 5 547 667 eligible admissions (16.1%; mean [SD] age, 63.5 [11.8] years; 353 315 [39.6%] men). Among the 4 655 183 patients who received opioids only, the mean (SD) age was 63.7 (14.7) years, and 1 913 284 (41.1%) were men. Overall, 441 overdose events were identified, with absolute risks of 1.4 per 10 000 patients with gabapentinoid exposure and 0.7 per 10 000 patients receiving opioids only. Following propensity score trimming, the cohort included 737 383 patients exposed to gabapentinoids and 3 002 480 patients receiving opioids only. The primary analysis yielded the adjusted hazard ratio of 1.95 (95% CI, 1.49-2.55), and the number needed to treat for an additional overdose to occur was 16 914 patients (95% CI, 11 556-31 537 patients). Adjusted hazard ratios for secondary outcomes were 1.68 (95% CI, 1.59-1.78) for respiratory complications, 1.77 (95% CI, 1.61-1.93) for unspecified adverse effects of opioids, and 1.70 (95% CI, 1.62-1.79) for the composite outcome. The results were consistent across sensitivity analyses and subgroups identified by key clinical factors. In this real-world cohort study of patients who underwent major surgery, concomitant use of gabapentinoids with opioids was associated with increased risk of opioid overdose and other opioid-related adverse events; however, the absolute risk of adverse events was low.

Comparative safety and effectiveness of alendronate versus raloxifene in women with osteoporosis

Alendronate and raloxifene are among the most popular anti-osteoporosis medications. However, there is a lack of head-to-head comparative effectiveness studies comparing the two treatments. We conducted a retrospective large-scale multicenter study encompassing over 300 million patients across nine databases encoded in the Observational Medical Outcomes Partnership (OMOP) Common Data Model (CDM). The primary outcome was the incidence of osteoporotic hip fracture, while secondary outcomes were vertebral fracture, atypical femoral fracture (AFF), osteonecrosis of the jaw (ONJ), and esophageal cancer. We used propensity score trimming and stratification based on an expansive propensity score model with all pre-treatment patient characteritistcs. We accounted for unmeasured confounding using negative control outcomes to estimate and adjust for residual systematic bias in each data source. We identified 283,586 alendronate patients and 40,463 raloxifene patients. There were 7.48 hip fracture, 8.18 vertebral fracture, 1.14 AFF, 0.21 esophageal cancer and 0.09 ONJ events per 1,000 person-years in the alendronate cohort and 6.62, 7.36, 0.69, 0.22 and 0.06 events per 1,000 person-years, respectively, in the raloxifene cohort. Alendronate and raloxifene have a similar hip fracture risk (hazard ratio [HR] 1.03, 95% confidence interval [CI] 0.94–1.13), but alendronate users are more likely to have vertebral fractures (HR 1.07, 95% CI 1.01–1.14). Alendronate has higher risk for AFF (HR 1.51, 95% CI 1.23–1.84) but similar risk for esophageal cancer (HR 0.95, 95% CI 0.53–1.70), and ONJ (HR 1.62, 95% CI 0.78–3.34). We demonstrated substantial control of measured confounding by propensity score adjustment, and minimal residual systematic bias through negative control experiments, lending credibility to our effect estimates. Raloxifene is as effective as alendronate and may remain an option in the prevention of osteoporotic fracture.

P4783Treatment persistence of patients with atrial fibrillation on VKA or NOAC: Data from GLORIA-AF Phase III 1-year interim analysis

Abstract Background Oral anticoagulation (OAC) persistence is important for optimizing stroke prevention in patients with atrial fibrillation (AF); non-vitamin K oral anticoagulants (NOACs) generally show better persistence than vitamin K antagonists (VKAs), while the impact of dosing regimens remains unclear. We compared treatment persistence of NOACs and VKAs, and of NOAC dosing regimens in the prospective GLORIA-AF registry program. Methods Patients newly diagnosed with AF were enrolled in Phase III of GLORIA-AF (2014–2016) from 4 geographical regions (North America [NA], Europe, Asia and Latin America). Treatment persistence after 1 year for i) NOAC (dabigatran, rivaroxaban, apixaban, edoxaban) vs VKA, and ii) twice daily (bid, dabigatran, apixaban) vs once daily (od; rivaroxaban, edoxaban) NOAC treatment was analysed using multivariable Cox analysis; propensity score trimming was used to reduce bias due to unmeasured confounders. Missing data was handled by multiple imputation. Results Overall, 21,592 patients were enrolled (4970 [23%] patients on VKAs, 12,797 [59%] on NOACs, 2391 [11%] on antiplatelets, and 1426 [6.6%] received no therapy; 8 [0.04%] received other treatment). After trimming, 11,935 and 4484 patients treated with NOACs and VKAs, respectively, were compared. NOACs had better treatment persistence than VKAs (discontinuation hazard ratio [HR]=0.75, 95% confidence interval [CI] 0.69–0.81). Other relevant associations were decreased OAC persistence for symptomatic AF, NA and Asia regions (Table). There was no difference in treatment persistence for patients on a bid (N=7842) vs od (N=4098) NOAC (discontinuation HR=0.94, 95% CI 0.86–1.02). Conclusion In this 1-year interim analysis of GLORIA-AF Phase III, treatment persistence was improved with NOACs vs VKAs, whereas for NOACs, dosing regimen (bid vs od) had no impact on treatment persistence. Acknowledgement/Funding The GLORIA-AF Registry program was funded by Boehringer-Ingelheim

Multinomial Extension of Propensity Score Trimming Methods: A Simulation Study.

Crump et al. (Biometrika. 2009;96(1):187-199), Stürmer et al. (Am J Epidemiol. 2010;172(7):843-854), and Walker et al. (Comp Eff Res. 2013;2013(3):11-20) proposed propensity score (PS) trimming methods as a means to improve efficiency (Crump) or reduce confounding (Stürmer and Walker). We generalized the trimming definitions by considering multinomial PSs, one for each treatment, and proved that these proposed definitions reduce to the original binary definitions when we have only 2 treatment groups. We then examined the performance of the proposed multinomial trimming methods in the setting of 3 treatment groups, in which subjects with extreme PSs more likely had unmeasured confounders. Inverse probability of treatment weights, matching weights, and overlap weights were used to control for measured confounders. All 3 methods reduced bias regardless of the weighting methods in most scenarios. Multinomial Stürmer and Walker trimming were more successful in bias reduction when the 3 treatment groups had very different sizes (10:10:80). Variance reduction, seen in all methods with inverse probability of treatment weights but not with matching weights or overlap weights, was more successful with multinomial Crump and Stürmer trimming. In conclusion, our proposed definitions of multinomial PS trimming methods were beneficial within our simulation settings that focused on the influence of unmeasured confounders.

Evidence of potential bias in a comparison of β blockers and calcium channel blockers in patients with chronic obstructive pulmonary disease and acute coronary syndrome: results of a multinational study

ObjectivesA number of observational studies have reported that, in patients with chronic obstructive pulmonary disease (COPD), β blockers (BBs) decrease risk of mortality and COPD exacerbations. To address important methodological...

Are Partial Workplace Smoking Bans as Effective as Complete Smoking Bans? A National Population-Based Study of Smoke-Free Policy Among Japanese Employees.

Although complete workplace smoking bans are generally recommended rather than partial bans, the latter are widespread in many countries, especially Japan. Our objective was to compare complete workplace smoking bans and partial bans for associations with employee smoking and secondhand smoke (SHS)-related discomfort/ill-health. We also evaluated complete bans versus no ban and partial bans versus no ban. Eleven thousand ninety eligible employees (weighted number: 34 353 241) aged 20-64 years in 2011 (response rate: 62.5%) were analyzed using a nationally-representative, population-based cross-sectional study. Adjusted prevalence ratios for self-reported current smoking and SHS-related discomfort/ill-health according to workplace smoke-free policies were calculated, using conventional regression and propensity score (PS) weighting (targeting population of average treatment effect among both treated [TET] and untreated [TEU]). Both conventional regressions and PS weighting analyses showed complete bans were significantly associated with lower prevalence of current smoking and perceived SHS-related discomfort/ill-health among nonsmokers than partial or no ban. In contrast, partial bans were not significantly associated with either outcome compared with no ban. Using several PS trimming levels, we found interesting differences between TET and TEU in a comparison between partial and no ban: that is, significant associations in TET estimations, but none in TEU estimations. Although complete smoking bans were associated with lower levels of employee smoking and SHS-related discomfort/ill-health compared with no smoking ban, partial bans were not. Findings from PS weighting of TEU suggest that partial workplace bans may not be any more effective for Japanese employees than no ban. Therefore, complete bans may be strongly recommended for future implementation, but careful interpretation of the data is necessary because of the cross-sectional study design.

Risk of Venous Thromboembolism in Patients with Rheumatoid Arthritis: Initiating Disease-Modifying Antirheumatic Drugs

ObjectivesRecent research suggests that rheumatoid arthritis increases the risk of venous thromboembolism. This study compared the risk of venous thromboembolism in patients with newly diagnosed rheumatoid arthritis initiating a biologic disease-modifying antirheumatic drug (DMARD) with those initiating methotrexate or a nonbiologic DMARD. MethodsWe conducted a population-based cohort study using US insurance claims data (2001-2012). Three mutually exclusive, hierarchical DMARD groups were used: (1) a biologic DMARD with and without nonbiologic DMARDs; (2) methotrexate without a biologic DMARD; or (3) nonbiologic DMARDs without a biologic DMARD or methotrexate. We calculated the incidence rates of venous thromboembolism. Cox proportional hazard models stratified by propensity score (PS) deciles after asymmetric PS trimming were used for 3 pairwise comparisons, controlling for potential confounders at baseline. ResultsWe identified 29,481 patients with rheumatoid arthritis with 39,647 treatment episodes. From the pairwise comparison after asymmetric PS trimming, the incidence rate of hospitalization for venous thromboembolism per 1000 person-years was 5.5 in biologic DMARD initiators versus 4.4 in nonbiologic DMARD initiators and 4.8 in biologic DMARD initiators versus 3.5 in methotrexate initiators. The PS decile-stratified hazard ratio of venous thromboembolism associated with biologic DMARDs was 1.83 (95% confidence interval [CI], 0.91-3.66) versus nonbiologic DMARDs and 1.39 (95% CI, 0.73-2.63) versus methotrexate. The hazard ratio of venous thromboembolism in biologic DMARD initiators was the highest in the first 180 days versus nonbiologic DMARD initiators (2.48; 95% CI, 1.14-5.39) or methotrexate initiators (1.80; 95% CI, 0.90-3.62). ConclusionsThe absolute risk for venous thromboembolism was low in patients with newly diagnosed rheumatoid arthritis. Initiation of a biologic DMARD seems to be associated with an increased short-term risk of hospitalization for venous thromboembolism compared with initiation of a nonbiologic DMARD or methotrexate.

Effect of adjuvant chemotherapy on survival of patients with stage III colon cancer diagnosed after age 75 years.

Few patients 75 years of age and older participate in clinical trials, thus whether adjuvant chemotherapy for stage III colon cancer (CC) benefits this group is unknown. A total of 5,489 patients ≥ 75 years of age with resected stage III CC, diagnosed between 2004 and 2007, were selected from four data sets containing demographic, stage, treatment, and survival information. These data sets included SEER-Medicare, a linkage between the New York State Cancer Registry (NYSCR) and its Medicare programs, and prospective cohort studies Cancer Care Outcomes Research and Surveillance Consortium (CanCORS) and the National Comprehensive Cancer Network. Data sets were analyzed in parallel using covariate adjusted and propensity score (PS) matched proportional hazards models to evaluate the effect of treatment on survival. PS trimming was used to mitigate the effects of selection bias. Use of adjuvant therapy declined with age and comorbidity. Chemotherapy receipt was associated with a survival benefit of comparable magnitude to clinical trials results (SEER-Medicare PS-matched mortality, hazard ratio [HR], 0.60; 95% CI, 0.53 to 0.68). The incremental benefit of oxaliplatin over non-oxaliplatin-containing regimens was also of similar magnitude to clinical trial results (SEER-Medicare, HR, 0.84; 95% CI, 0.69 to 1.04; NYSCR-Medicare, HR, 0.82, 95% CI, 0.51 to 1.33) in two of three examined data sources. However, statistical significance was inconsistent. The beneficial effect of chemotherapy and oxaliplatin did not seem solely attributable to confounding. The noninvestigational experience suggests patients with stage III CC ≥ 75 years of age may anticipate a survival benefit from adjuvant chemotherapy. Oxaliplatin offers no more than a small incremental benefit. Use of adjuvant chemotherapy after the age of 75 years merits consideration in discussions that weigh individual risks and preferences.

Treatment Effects in the Presence of Unmeasured Confounding: Dealing With Observations in the Tails of the Propensity Score Distribution--A Simulation Study

Frailty, a poorly measured confounder in older patients, can promote treatment in some situations and discourage it in others. This can create unmeasured confounding and lead to nonuniform treatment effects over the propensity score (PS). The authors compared bias and mean squared error for various PS implementations under PS trimming, thereby excluding persons treated contrary to prediction. Cohort studies were simulated with a binary treatment T as a function of 8 covariates X. Two of the covariates were assumed to be unmeasured strong risk factors for the outcome and present in persons treated contrary to prediction. The outcome Y was simulated as a Poisson function of T and all X's. In analyses based on measured covariates only, the range of PS's was trimmed asymmetrically according to the percentile of PS in treated patients at the lower end and in untreated patients at the upper end. PS trimming reduced bias due to unmeasured confounders and mean squared error in most scenarios assessed. Treatment effect estimates based on PS range restrictions do not correspond to a causal parameter but may be less biased by such unmeasured confounding. Increasing validity based on PS trimming may be a unique advantage of PS's over conventional outcome models.