Propensity Score Subclassification Research Articles

Contemporary Prestroke Dual Antiplatelet Use and Symptomatic Intracerebral Hemorrhage Risk After Thrombolysis

Intravenous alteplase (IV-tPA) can be administered to patients with acute ischemic stroke but is associated with symptomatic intracerebral hemorrhage (sICH). It is unclear if patients taking prestroke dual antiplatelet therapy (DAPT) are at higher risk of sICH. To determine the associated risk of sICH in patients taking prestroke dual antiplatelet therapy receiving alteplase for acute ischemic stroke using propensity score matching analysis. This cohort study used data from the American Heart Association and American Stroke Association Get With The Guidelines-Stroke (GWTG-Stroke) registry between 2013 and 2021. Data were obtained from hospitals in the GWTG-Stroke registry. This study included patients hospitalized with acute ischemic stroke and treated with IV-tPA. Data were analyzed from January 2013 to December 2021. Prestroke DAPT before treatment with IV-tPA for acute ischemic stroke. sICH, In-hospital death, discharge modified Rankin scale score, and other life-threatening systemic hemorrhages. Of 409 673 participants, 321 819 patients (mean [SD] age, 68.6 [15.1] years; 164 587 female [51.1%]) who were hospitalized with acute ischemic stroke and treated with IV-tPA were included in the analysis. The rate of sICH was 2.9% (5200 of 182 344), 3.8% (4457 of 117 670), and 4.1% (893 of 21 805) among patients treated with no antiplatelet therapy, single antiplatelet therapy (SAPT), and DAPT, respectively (P < .001). In adjusted analyses after propensity score subclassification, both SAPT (odds ratio [OR], 1.13; 95% CI, 1.07-1.19) and DAPT (OR, 1.28; 95% CI, 1.14-1.42) were associated with increased risks of sICH. Prestroke antiplatelet medications were associated with lower odds of discharge mRS score of 2 or less compared with no medication (SAPT OR, 0.92; 95% CI, 0.90-0.95; DAPT OR, 0.94; 95% CI, 0.88-0.98). Results of a subgroup analysis of patients taking DAPT exposed to aspirin-clopidogrel vs aspirin-ticagrelor combination therapy were not significant (OR, 1.35; 95% CI, 0.84-1.86). Prestroke DAPT was associated with a significantly elevated risk of sICH among patients with ischemic stroke who were treated with thrombolysis; however, the absolute increase in risk was small. Patients exposed to antiplatelet medications did not have excess sICH compared with landmark trials, which demonstrated overall clinical benefit of thrombolysis therapy for acute ischemic stroke.

Venous Thromboembolic Risk of Estradiol Valerate-Dienogest Compared with Ethinyl Estradiol-Levonorgestrel Combined Oral Contraceptives.

This pooled analysis compared the risk of venous thromboembolism (VTE) associated with combined oral contraceptives (COCs) containing estradiol (E2) valerate-dienogest with those containing ethinyl E2-levonorgestrel. Data were retrieved from two large, prospective, observational cohort studies. Propensity score subclassification was applied to balance baseline parameters between the COC user cohorts. Crude and adjusted hazard ratios (HRs) were calculated based on the extended Cox model. The pooled data set included 11,616 E2 valerate-dienogest users and 18,681 ethinyl E2-levonorgestrel users, contributing 17,932 and 29,140 women-years of observation, respectively. A significantly decreased VTE risk in E2 valerate-dienogest COCs compared with ethinyl E2-levonorgestrel COCs was observed (propensity score-stratified HR 0.46, 95% CI, 0.22-0.98). This pooled analysis expands data from a previous postauthorization safety study and provides valuable real-world safety information on the relative safety of current COCs.

IMPACT OF RAMIPRIL/AMLODIPINE DOSE COMBINATIONS OF SINGLE-PILL VERSUS FREE EQUIVALENT COMBINATIONS ON PERSISTENCE AND MAJOR ADVERSE CARDIOVASCULAR EVENTS: A REAL-WORLD ANALYSIS

Objective: Guidelines recommend use of single-pill combination (SPC) for management of hypertensive patients. Very few studies compared SPCs and free equivalent combinations (FEC) focusing on short-term persistence and major adverse cardiovascular events (MACE) outcomes. This study assessed the impact of different dose combinations of ramipril/amlodipine (R/A) SPC versus FEC on treatment persistence and MACE in patients with hypertension over a 5-year period. Design and method: This real-world, retrospective, observational cohort study was performed using National Health Insurance Fund of Hungary database. The study included hypertensive patients aged &gt; = 18 years who initiated on R/A SPC or FEC, of different dose combinations: R/A 5/5 mg, 5/10 mg, 10/5 mg and 10/10 mg, between 01 April 2012 and 31 December 2018. Primary endpoint was the long-term persistence rate, applying 90-days grace period, between R/A SPC and FEC. Secondary endpoint was the rate of MACE (composite of non-fatal myocardial infarction, stroke, hospitalization for heart failure and all-cause death). Analyses were performed using Cox-proportional hazard model and propensity score sub-classification method to adjust the imbalances in baseline characteristics. Results: Overall, 173,206 patients with hypertension were included (SPC group, N = 104,882 [61%], and FEC group, N = 68,324 [39%]); R/A 5/5 mg representing the largest proportion (N = 107,992 [62%]). In SPC and FEC groups, mean (SD) age was 61.31 (13.75) and 66.08 (13.10) years and proportion of male was 45.4% and 43.1%, respectively. As shown in Table 1, the rate of non-persistence was significantly lower with all R/A SPCs compared to FECs from Month 1 to Month 24 (P &lt; 0.001). Thereafter, the rate of non-persistence remained significantly lower in the SPC groups except in the R/A 5/5 mg dose. MACE was significantly reduced with all R/A SPCs compared to FECs (HR 0.68–0.73; P &lt; = 0.002). There was no impact of age and gender on the association between the treatment and persistence rate or MACE outcomes. Conclusions: Our results confirm the beneficial effects of SPCs compared to FECs in achieving long-term improvement on persistence rate and MACE outcomes in patients with hypertension and support current guidelines recommending a widespread use of SPCs.

Propensity score-based analysis of stereotactic body radiotherapy, lobectomy and sublobar resection for stage I non-small cell lung cancer.

We applied two propensity score-based analyses to simultaneously compare three treatment modalities—stereotactic body radiotherapy (SBRT), lobectomy, or sublobar resection (SLR)—for stage I non-small cell lung cancer (NSCLC), with the aim of clarifying the average treatment effect (ATE) and formulating a risk-adapted approach to treatment selection. A retrospective review of 823 patients aged ≥65 years who underwent SBRT, lobectomy, or SLR for stage I NSCLC was conducted. The following two analyses using machine learning-based propensity scores were performed: (i) propensity score weighting (PSW) to assess the ATE in the entire cohort, and (ii) propensity score subclassification (PSS) to evaluate treatment effects of subgroups. PSW showed no significant difference in the 5-year overall survival (OS) between SBRT and SLR (60.0% vs 61.2%; P = 0.70) and significant difference between SBRT and lobectomy (60.0% vs 77.6%; P = 0.026). Local (LR) and distant recurrence (DR) rates were significantly lower in lobectomy than in SBRT, whereas there was no significant difference between SBRT and SLR. PSS identified four subgroups with different patient characteristics: lobectomy-oriented (5-year cumulative incidences of non-lung cancer death, 7.5%), SLR-oriented (14.2%), SBRT-oriented (23.8%) and treatment-neutral subgroups (16.1%). Each subgroup showed different survival trends regarding the three treatments. The ATE of SBRT was not significantly different from that of SLR, but it was inferior to lobectomy. Four subgroups with different risks of non-lung cancer death and different survival trends for each treatment were identified. These would help decision-making for patients with stage I NSCLC.

Thromboembolic safety of norethisterone vs levonorgestrel in combined oral contraceptive users: a pooled analysis of 4 large prospective cohort studies

BACKGROUNDNorethisterone (acetate) and levonorgestrel are marketed globally as components of combined oral contraceptives. Although guidelines recommend both as first-line combined oral contraceptives, no direct, comparative safety studies are available.OBJECTIVEWe directly compared the thromboembolic event risk associated with the use of norethisterone acetate-containing and levonorgestrel-containing combined oral contraceptives.STUDY DESIGNData regarding the cohorts of interest, norethisterone/norethisterone acetate (ethinylestradiol ≤30 µg) and levonorgestrel (ethinylestradiol ≤30 µg), were retrieved from a pooled dataset comprising 4 prospective, noninterventional, active-surveillance cohort studies in 14 European countries, the United States, and Canada, with similar study design but differing medication cohorts. Baseline characteristics and parameters of reproductive, contraceptive, and medical history were summarized using descriptive statistics. Propensity score subclassification was applied to balance baseline parameters between cohorts. Time-to-event analysis of venous thromboembolic events was performed on the basis of the extended Cox model to calculate crude and adjusted hazard ratios, including 95% confidence intervals. The time of venous thromboembolic events was censored at the end of the observation period for women who did not have an event. Women who dropped out or were lost to follow-up without reported venous thromboembolic events were censored at the time they last confirmed that they did not have an event.RESULTSThe pooled dataset included 235,437 combined oral contraceptive users who were followed up for a total of 571,163 women years. Among these, 40,142 women were users of norethisterone/norethisterone acetate (ethinylestradiol ≤30 µg), and 39,098 women were users of levonorgestrel (ethinylestradiol ≤30 µg), contributing 61,976 and 84,816 women years of observation, respectively. The observed prevalence of prognostic factors at baseline showed typical features of US and European combined oral contraceptive users. Both cohorts showed a similar, low rate of thromboembolic events, and we could exclude a 1.5-fold increased venous thromboembolism risk for norethisterone/norethisterone acetate relative to levonorgestrel (adjusted hazard ratio, 0.73; 95% confidence interval, 0.48–1.11).CONCLUSIONThese data confirm the similar risk profiles of norethisterone/norethisterone acetate and levonorgestrel regarding thromboembolic events in routine combined oral contraceptive use of around 80,000 women from Europe and the United States/Canada. The analysis provides reassurance for both combined oral contraceptive users and clinicians regarding the safety of oral contraceptives and potentially opens discussion on norethisterone acetate as a potential gold standard therapy in clinical and postmarket research alongside levonorgestrel-combined oral contraceptives.

The Sensitivity of Small Area Estimates under Propensity Score Subclassification for Generalization

Over the past decade, statisticians have developed methods to improve generalizations from nonrandom samples using propensity score methods. While these methods contribute to generalization research, their effectiveness is limited by small sample sizes. Small area estimation is a class of model-based methods that address the imprecision due to small samples and prior research has shown its potential to improve precision in generalization studies with small samples. However, the validity of both small area methods and existing design-based methods depend on the assumptions of propensity score models, leaving open the question of whether a specific method is preferred when core assumptions are violated. This study compares the performance of design-based and model-based estimators of population parameters in generalization studies when the assumptions for propensity score models are violated. Using a simulation study and an empirical example, we assess the sensitivity of the estimators under various degrees of violations in assumptions, discuss the advantages and tradeoffs of each approach, and highlight the implications for generalization research.

Thromboembolic safety profile of low-dose estradiol (valerate) in combined hormonal preparations: Implications for the development of new hormonal endometriosis and uterine fibroid therapies

Background: Several promising new medications containing low-dose estradiol (E2) in combination with a progestin and an additional component, such as gonadotropin-releasing hormone antagonists, are currently being developed for use in pre-menopausal women. Objective: This pooled analysis was designed to estimate the thromboembolic safety profile of E2 and its ester, estradiol valerate (E2Val), when used in combined hormonal treatments in a pre-menopausal population. Methods: Data regarding users of combined oral contraceptives (COCs) and combined menopausal hormonal therapy (MHT) containing either E2/E2Val or ethinylestradiol (EE) ⩽ 30 µg were retrieved from five large prospective, non-interventional, cohort studies in Europe, US, and Canada with similar study design but differing medication cohorts. Propensity score sub-classification was applied to balance baseline parameters between cohorts and time-to-event analysis of venous thromboembolic events (VTE) was carried out based on the extended Cox model to calculate crude and adjusted hazard ratios (HR). Results: (1) Crude incidence rates of VTE were higher in MHT users compared to pre-menopausal COC users, (2) the VTE risk in menopausal users of E2/E2Val-norethindrone acetate was not higher than that in menopausal users of E2/E2Val-progestin (adjusted HR 0.71; 95% confidence interval, 0.41-1.26) and (3) the VTE risk in pre-menopausal users of E2/E2Val-progestin was similar, or lower, than pre-menopausal users of EE⩽30µg-progestin (adjusted HR 0.49; 95% confidence interval, 0.28–0.84). Conclusion: Our data presents a solid safety assessment of combined hormonal preparations containing E2/E2Val. We conclude that the risk of E2/E2Val-norethindrone acetate in pre-menopausal users is unlikely to be higher than the known risk of COCs containing EE ⩽ 30 µg-progestin.

Intra-Study Matching Considerations When Using Mixed Methods-Based Research Approaches: A Critical Dialectical Pluralistic Approach

The step of obtaining a sample(s) (i.e., sampling) in mixed methods-based research studies likely represents the least developed step in the research process, with only 21 Scopus-indexed works published on the topic to date. Consequently, the time is rife for mixed methods-based researchers to develop sampling designs that are more TREEful—that is, transparent, rigorous, equitable, and ethical—especially when sampling among/between phases/components. Because, more than the other 13 mixed methods-based research philosophies, critical dialectical pluralism especially is concerned with the welfare of research participants, and because the sampling step is subject to misuse and abuse of participants, the use of a critical dialectical pluralist lens to ensure that mixed methods-based sampling designs are as TREEful as possible has logical appeal. Therefore, in this editorial, we have provided a meta-framework,1 via a critical dialectical pluralism lens, for selecting samples for each of the following four types of relationships among/between phases/components identified by Onwuegbuzie and Collins (2007), namely, identical samples, parallel samples, nested samples, and multilevel samples. This lens has led to the identification of several options for minimizing, or at least reducing, what we refer to as identical sampling bias, parallel sampling bias, nested sampling bias, and multilevel sampling bias such that samples are optimally matched within a single mixed methods-based research study. In the context of mixed methods-based research, matching refers to the process of forming groups to make them as similar as possible with respect to extraneous or confounding factors (e.g., demographic variables [e.g., gender, age]; personality variables [e.g., resilience]; affective variables [e.g., motivation]). In particular, we outline the use of several matching techniques—specifically, exact matching, greedy matching, optimal matching, propensity score matching, subclassification, and magnitude coding—for addressing these different forms of bias. We encourage mixed methods-based researchers to explore using one or more of these matching techniques, whenever appropriate, regardless of their philosophical stance, in order to avoid researcher participants from being misrepresented.

Prospective, multicenter clinical trial comparing M6-C compressible six degrees of freedom cervical disc with anterior cervical discectomy and fusion for the treatment of single-level degenerative cervical radiculopathy: 2-year results of an FDA investigational device exemption study

BACKGROUND CONTEXTVarious designs of total disc replacement (TDR) devices have been compared to anterior cervical discectomy and fusion (ACDF) with favorable outcomes in FDA-approved investigational device exemption trials. The design of M6-C with a compressible viscoelastic nuclear core and an annular structure is substantially different than prior designs and has previously demonstrated favorable kinematics and clinical outcomes in small case series. PURPOSETo evaluate the safety and effectiveness of the novel M6-C compressible artificial cervical disc compared with ACDF for subjects with single-level degenerative cervical radiculopathy. STUDY DESIGN/SETTINGProspective, multicenter, concurrently and historically controlled, FDA-approved investigational device exemption clinical trial. PATIENT SAMPLESubjects with one-level symptomatic degenerative cervical radiculopathy were enrolled and assigned to receive M6-C or ACDF. OUTCOME MEASURESPain and function (Neck Disability Index, VAS), quality of life (SF-36), safety, neurologic, and radiographic assessments of motion (both flexion extension and lateral bending) were performed. The primary clinical endpoint was composite clinical success (CCS) at 24 months. METHODSUsing propensity score subclassification to control for selection bias, 160 M6-C subjects were compared to a matched subset of 189 ACDF controls (46 concurrent and 143 historical controls). RESULTSBoth ACDF and M6-C subjects reported significant improvements in patient-reported outcomes at all time points over baseline. Overall SF-36 Physical Component Score and neck and arm pain scores were significantly improved for M6-C as compared to ACDF treatment. CCS and mean Neck Disability Index improvements were similar between M6-C and ACDF. Correspondingly, there were significantly fewer subjects that utilized pain medication or opioids following M6-C treatment at 24 months relative to baseline. Range of motion was maintained in subjects treated with M6-C. Subsequent surgical interventions, dysphagia rates, and serious adverse events were comparable between groups. CONCLUSIONSM6-C treatment demonstrated both safety and effectiveness for the treatment of degenerative cervical radiculopathy. Treatment with M6-C demonstrated noninferiority for the primary endpoint, indicating a similar ability to achieve CCS at 24 months. However, for the secondary endpoints, M6-C subjects demonstrated significantly improved pain and function compared to ACDF subjects, while maintaining range of motion, improving quality of life, and decreasing analgesic and opioid usage at 2 years postoperatively relative to baseline.

Determination of the optimal number of strata for propensity score subclassification

In observational studies, propensity score methods are useful for estimating causal effects. We propose a new method for selecting the optimal number of strata for subclassification on the propensity score. The proposed method achieves “optimality” in terms of the MSE.

Fluid Management in Patients with Acute Respiratory Distress Syndrome and Diabetes Mellitus: A propensity score matched analysis of the fluid and catheter treatment trial.

Diabetes mellitus results in an attenuated inflammatory response, reduces pulmonary microvascular permeability, and may decrease the risk of developing acute respiratory distress syndrome (ARDS). Studies have shown that patients with ARDS are better managed by a conservative as compared to liberal fluid management strategy. However, it is not known if the same fluid management principles hold true for patients with comorbid diabetes mellitus and ARDS.As diabetes mellitus results in reduced pulmonary microvascular permeability and an attenuated inflammatory response, we hypothesize that in the setting of ARDS, diabetic patients will be able to tolerate a positive fluid balance better than patients without diabetes.The Fluid and Catheter Treatment Trial (FACTT) randomized patients with ARDS to conservative versus liberal fluid management strategies. In a secondary analysis of this trial, we calculated the interaction of diabetic status and differing fluid strategies on outcomes. Propensity score subclassification matching was used to control for the differing baseline characteristics between patients with and without diabetes.Nine hundred fifty-six patients were analyzed. In a propensity score matched analysis, the difference in the effect of a conservative as compared to liberal fluid management strategy on ventilator free days was 2.23 days (95% CI: -0.97 to 5.43 days) in diabetic patients, and 2.37 days (95% CI: -0.21 to 4.95 days) in non-diabetic patients. The difference in the effect of a conservative as compared to liberal fluid management on 60 day mortality was 2% (95% CI: -11.8% to 15.8%) in diabetic patients, and -7.9% (95% CI: -21.7% to 5.9%) in non-diabetic patients.When comparing a conservative fluid management strategy to a liberal fluid management strategy, diabetic patients with ARDS did not have a statistically significant difference in outcomes than non-diabetic patients.

ESTRADIOL IS ASSOCIATED WITH DECREASED CARDIOVASCULAR RISK COMPARED TO ETHINYLESTRADIOL WHEN CONTAINED IN COMBINED HORMONAL PREPARATIONS: IMPLICATIONS FOR THE DEVELOPMENT OF NEW ENDOMETRIOSIS THERAPIES

Combined estrogen-progestin preparations have a long history in women’s health and are currently used for multiple indications, including the treatment of endometriosis-related symptoms. Empirically, estradiol (E2) and its ester, estradiol valerate (E2Val) carry a reduced cardiovascular risk due to their reduced impact on the hepatic system compared to the most-widely used estrogen ethinylestradiol (EE). However, robust comparative studies in pre-menopausal women are limited. A new hormonal product containing E2 in combination with the progestin norethindrone acetate (NETA) and the gonadotropin-releasing hormone receptor antagonist, relugolix, has been developed for the treatment of uterine fibroids and endometriosis. This analysis tries to quantify the cardiovascular risk of E2/E2Val-NETA in pre-menopausal women. A pooled analysis of four large cohort studies assessing the risk of cardiovascular events in users of combined oral contraceptives (COCs). Data regarding users of COCs containing either E2/E2Val or EE≤30μg were retrieved from four large prospective, controlled, observational, cohort studies in 14 European countries, US and Canada. Baseline characteristics, including reproductive, contraceptive and medical history, were summarized using descriptive statistics. Propensity score sub-classification was applied to balance baseline parameters between cohorts and time-to-event analysis of venous and arterial thromboembolic events was carried out based on the extended Cox model to calculate crude and adjusted hazard ratios (HR) including 95%-confidence intervals (CI). The analysis was based on 210,000 women, who contributed 470,000 women years of exposure. The observed prevalence of prognostic factors at baseline showed typical features of US and European COC users. However, the mean age in the E2/E2Val cohort was higher compared to the EE≤30μg cohort (31.7 years ±9.76 and 26.1 years ±7.89, respectively). A time-to-event analysis of the VTE data was carried out based on the extended Cox model. Validity of the model was demonstrated by the standardized differences summarized over strata as weighted average yielded upon PS sub-classification, which were consistently <0.25 and mostly <0.1. A comparison between cohorts showed a decreased risk of venous thromboembolism in users of COCs containing E2/E2Val vs users of COCs containing EE≤30μg; adjusted hazard ratio: 0.49 (95% CI, 0.28-0.84). A risk reduction in the E2/E2Val cohort was also observed with regards to arterial thromboembolic events; adjusted hazard ratio: 0.26 (95% CI, 0.08-0.83). Data presents a solid safety assessment of combined hormonal preparations in over 200,000 pre-menopausal women and shows a decreased cardiovascular risk of E2/E2Val when compared to low-dose EE. This provides reassurance for current users of E2/E2Val-containing COCs and opens new perspectives for the development of safe and effective endometriosis treatments, such as the recently developed product containing E2/NETA, in combination with relugolix.

0613 The Effects of Diabetes on Obstructive Sleep Apnea Severity and Progression: A Propensity Score Analysis

Abstract Introduction Obstructive sleep apnea (OSA) is associated with increased glucose intolerance and insulin resistance, and commonly coexists with type 2 diabetes mellitus (T2DM). Research suggests an independent association between these two conditions. However, more research into the role of T2DM in the development of OSA, or vice versa, is needed. Methods Leveraging data from 139 participants with T2DM from the Sleep AHEAD cohort and 5,085 participants without T2DM from the Sleep Heart Health Study (SHHS), we conducted two complementary propensity score (PS) subclassification analyses to estimate the effect of T2DM on the oxygen desaturation index (ODI), both at baseline and over 4-5 years of follow-up. PS models included age, sex, race, body mass index, neck circumference, waist circumference, total cholesterol, HDL cholesterol, triglycerides, Epworth Sleepiness Scale, and comorbid hypertension or cardiovascular disease. Models evaluating ODI progression also included baseline ODI. Results The PS subclassification analysis identified 109 participants with T2DM and 480 without T2DM, balanced with respect to baseline covariates, for evaluation of the effect of T2DM on baseline ODI. On average, those with T2DM had a 9 events/hour greater ODI compared to those without (24.3 [20.8, 27.9] vs. 15.3 [13.6, 16.9] events/hour; p&amp;lt;0.0001). Among patients with baseline ODI≥5, a second PS subclassification identified 99 with T2DM and 227 without for evaluating the effect of T2DM on ODI progression. No difference in change in ODI was observed between those with and without T2DM (mean [95% CI] difference -0.25 [-10.7, 10.2] events/hour; p=0.963). Conclusion Using two robust PS subclassification designs to minimize selection bias, we evaluated the effect of T2DM on baseline ODI and ODI progression in adults with OSA after 4-5 years. Those with T2DM had more severe baseline ODI, but there were no meaningful differences in ODI progression. Results further our understanding of the association between these coexisting conditions. Support NIH grant HL070301

Comparing the Cardiovascular Risk of Norethindrone Acetate &amp; Levonorgestrel Pills: Is There a Difference? [10I

INTRODUCTION: Combined oral contraceptives containing norethindrone acetate (NETA) are widely used in the US and globally, but how they compare to levonorgestrel-containing pills is unknown. This pooled analysis compares the safety of NETA and levonorgestrel (LNG) in pills with ethinyl estradiol (EE) ≤30 mcg. METHODS: We pooled cohorts of NETA (EE ≤30 mcg) and LNG (EE ≤30 mcg) users from four large prospective, controlled, non-interventional, cohort studies in 14 European countries, US and Canada. Baseline characteristics, including reproductive, contraceptive and medical history, were summarized using descriptive statistics. Propensity score sub-classification was applied to balance baseline characteristics between cohorts. Time-to-event analysis of venous and arterial thromboembolic events was carried out based on the extended Cox model to calculate crude and adjusted hazard ratios with 95% confidence intervals. RESULTS: Approximately 80,000 women contributed 145,000 women years of exposure. The prevalence of prognostic factors at baseline showed typical features of US and European oral contraceptive users. A comparison between NETA vs LNG showed no significant difference in the risk of venous thromboembolism: adjusted hazard ratio 0.73 (95% CI 0.48-1.11). A similar result was observed considering the risk of arterial thromboembolism (adjusted hazard ratio 0.89 [95% CI 0.42-1.89]). CONCLUSION: Low-dose oral contraceptives containing NETA and LNG are associated with similar low risks of venous and arterial disease. This analysis provides reassurance for both women and clinicians regarding the safety of contemporary oral contraceptives.

Treatment Effect Decomposition and Bootstrap Hypothesis Testing in Observational Studies

Causal inference with observational data has drawn attention across various fields. These observational studies typically use matching methods which find matched pairs with similar covariate values. However, matching methods may not directly achieve covariate balance, a measure of matching effectiveness. As an alternative, the Balance Optimization Subset Selection (BOSS) framework, which seeks optimal covariate balance directly, has been proposed. This paper extends BOSS by estimating and decomposing a treatment effect as a combination of heterogeneous treatment effects from a partitioned set. Our method differs from the traditional propensity score subclassification method in that we find a subset in each subclass using BOSS instead of using the stratum determined by the propensity score. Then, by conducting a bootstrap hypothesis test on each component, we check the statistical significance of these treatment effects. These methods are applied to a dataset from the National Supported Work Demonstration (NSW) program which was conducted in the 1970s. By examining the statistical significance, we show that the program was not significantly effective to a specific subgroup composed of those who were already employed. This differs from the combined estimate—the NSW program was effective when considering all the individuals. Lastly, we provide results that are obtained when these steps are repeated with sub-samples.

Simultaneous record linkage and causal inference with propensity score subclassification.

We develop methodology for causal inference in observational studies when using propensity score subclassification on data constructed with probabilistic record linkage techniques. We focus on scenarios where covariates and binary treatment assignments are in one file and outcomes are in another file, and the goal is to estimate an additive treatment effect by merging the files. We assume that the files can be linked using variables common to both files, eg, names or birth dates, but that links are subject to errors, eg, due to reporting errors in the linking variables. We develop methodology for cases where such reporting errors are independent of the other variables on the files. We describe conceptually how linkage errors can affect causal estimates in subclassification contexts. We also present and evaluate several algorithms for deciding which record pairs to use in estimation of causal effects. Using simulation studies, we demonstrate that case selection procedures can result in improved accuracy in estimates of treatment effects from linked data compared to using only cases known to be true links.

A Comparison of Bias Reduction Methods: Clustering versus Propensity Score Subclassification and Weighting

ABSTRACTPropensity score (PS) adjustments have become popular methods used to improve estimates of treatment effects in quasi-experiments. Although researchers continue to develop PS methods, other procedures can also be effective in reducing selection bias. One of these uses clustering to create balanced groups. However, the success of this new method depends on its efficacy compared to that of the existing methods. Therefore, this comparative study used experimental and nonexperimental data to examine bias reduction, case retention, and covariate balance in the clustering method, PS subclassification, and PS weighting. In general, results suggest that the cluster-based methods reduced at least as much bias as the PS methods. Under certain conditions, the PS methods reduced more bias than the cluster-based method, and under other conditions the cluster-based methods were more advantageous. Although all methods were equally effective in retaining cases and balancing covariates, other data-specific conditions may likely favor the use of a cluster-based approach.

Optimally combining propensity score subclasses.

Propensity score methods, such as subclassification, are a common approach to control for confounding when estimating causal effects in non-randomized studies. Propensity score subclassification groups individuals into subclasses based on their propensity score values. Effect estimates are obtained within each subclass and then combined by weighting by the proportion of observations in each subclass. Combining subclass-specific estimates by weighting by the inverse variance is a promising alternative approach; a similar strategy is used in meta-analysis for its efficiency. We use simulation to compare performance of each of the two methods while varying (i) the number of subclasses, (ii) extent of propensity score overlap between the treatment and control groups (i.e., positivity), (iii) incorporation of survey weighting, and (iv) presence of heterogeneous treatment effects across subclasses. Both methods perform well in the absence of positivity violations and with a constant treatment effect with weighting by the inverse variance performing slightly better. Weighting by the proportion in subclass performs better in the presence of heterogeneous treatment effects across subclasses. We apply these methods to an illustrative example estimating the effect of living in a disadvantaged neighborhood on risk of past-year anxiety and depressive disorders among U.S. urban adolescents. This example entails practical positivity violations but no evidence of treatment effect heterogeneity. In this case, weighting by the inverse variance when combining across propensity score subclasses results in more efficient estimates that ultimately change inference. Copyright © 2016 John Wiley & Sons, Ltd.

Propensity score matching and subclassification in observational studies with multi-level treatments.

In this article, we develop new methods for estimating average treatment effects in observational studies, in settings with more than two treatment levels, assuming unconfoundedness given pretreatment variables. We emphasize propensity score subclassification and matching methods which have been among the most popular methods in the binary treatment literature. Whereas the literature has suggested that these particular propensity-based methods do not naturally extend to the multi-level treatment case, we show, using the concept of weak unconfoundedness and the notion of the generalized propensity score, that adjusting for a scalar function of the pretreatment variables removes all biases associated with observed pretreatment variables. We apply the proposed methods to an analysis of the effect of treatments for fibromyalgia. We also carry out a simulation study to assess the finite sample performance of the methods relative to previously proposed methods.

Social determinants of disparities in weight among US children and adolescents

PurposeTo explore whether contextual variables attenuate disparities in weight among 18,639 US children and adolescents aged 2 to 18 years participating in the National Health and Nutrition Examination Survey, 2001 to 2010. MethodsDisparities were assessed using the Symmetrized Rényi Index, a new measure that summarizes disparities in the severity of a disease, as well as the prevalence, across multiple population groups. Propensity score subclassification was used to ensure covariate balance between racial and ethnic subgroups and account for individual-level and contextual covariates. ResultsBefore propensity score subclassification, significant disparities were evident in the prevalence of overweight and/or obesity and the degree of excess weight among overweight/obese children and adolescents. After propensity score subclassification, racial/ethnic disparities in the prevalence and severity of excess weight were completely attenuated within matched groups, indicating that racial and ethnic differences were explained by social determinants such as neighborhood socioeconomic and demographic factors. ConclusionsThe limited overlap in covariate distributions between various racial/ethnic subgroups warrants further attention in disparities research. The attenuation of disparities within matched groups suggests that social determinants such as neighborhood socioeconomic factors may engender disparities in weight among US children and adolescents.