Modeling dose-response relationships of drugs is essential to understanding their safety effects on patients under realistic circumstances. While intention-to-treat analyses of clinical trials provide the effect of assignment to a particular drug and dose, they do not capture observed exposure after factoring in nonadherence and dropout. We develop a Bayesian method to flexibly model the dose-response relationships of binary outcomes with continuous treatment, permitting multiple evidence sources, treatment effect heterogeneity, and nonlinear dose-response curves. In an application, we examine the risk of excessive weight gain for patients with schizophrenia treated with the second-generation antipsychotics paliperidone, risperidone, or olanzapine in 14 clinical trials. We define exposure as total cumulative dose (daily dose × duration) and convert to units equivalent to 100 mg of olanzapine (OLZ doses). Averaging over the sample population of 5891 subjects, the median dose ranged from 0 (placebo randomized participants) to 6.4 OLZ doses (paliperidone randomized participants). We found paliperidone to be least likely to cause excessive weight gain across a range of doses. Compared with 0 OLZ doses, at 5.0 OLZ doses, olanzapine subjects had a 15.6% (95% credible interval: 6.7, 27.1) excess risk of weight gain; corresponding estimates for paliperidone and risperidone were 3.2% (1.5, 5.2) and 14.9% (0.0, 38.7), respectively. Moreover, compared with nonblack participants, black participants had a 6.8% (1.0, 12.4) greater risk of excessive weight gain at 10.0 OLZ doses of paliperidone. Nevertheless, our findings suggest that paliperidone is safer in terms of weight gain risk than risperidone or olanzapine for all participants at low to moderate cumulative OLZ doses.

BACKGROUND: Medicare insurance claims may provide an efficient means to ascertain follow-up of older participants in clinical research. We sought to determine the accuracy and completeness of claims- versus site-based follow-up with clinical event committee (+CEC) adjudication of cardiovascular outcomes. METHODS: We performed a retrospective study using linked Medicare and Duke Database of Clinical Trials data. Medicare claims were linked to clinical data from 7 randomized cardiovascular clinical trials. Of 52,476 trial participants, linking resulted in 5,839 (of 10,497 linkage-eligible) Medicare-linked trial participants with fee-for-service A and B coverage. Death, myocardial infarction (MI), stroke, and revascularization incidences were compared using Medicare inpatient claims only, site-reported events (+CEC) only, or a combination of the 2. Randomized treatment effects were compared as a function of whether claims-based, site-based (+CEC), or a combined system was used for event detection. RESULTS: Among the 5,839 study participants, the annual event rates were similar between claims- and site-based (+CEC) follow-up: death (overall rate 5.2% vs 5.2%; adjusted κ 0.99), MI (2.2% vs 2.3%; adjusted κ 0.96), stroke (0.7% vs 0.7%; adjusted κ 0.99), and any revascularization (7.4% vs 7.9%; adjusted κ 0.95). Of events detected by claims yet not reported by CEC, a minority were reported by sites but negatively adjudicated by CEC (39% of MIs and 18% of strokes). Differences in individual case concordance led to higher event rates when claims- and site-based (+CEC) systems were combined. Randomized treatment effects were similar among the 3 approaches for each outcome of interest. CONCLUSIONS: Claims- versus site-based (+CEC) follow-up identified similar overall cardiovascular event rates despite meaningful differences in the events detected. Randomized treatment effects were similar using the 2 methods, suggesting claims data could be used to support clinical research leveraging routinely collected data. This approach may lead to more effective evidence generation, synthesis, and appraisal of medical products and inform the strategic approaches toward the National Evaluation System for Health Technology.

Importance: Risk adjustment models using claims-based data are central in evaluating health care performance. Although US Centers for Medicare & Medicaid Services (CMS) models apply well-vetted statistical approaches, recent changes in the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) coding system and advances in computational capabilities may provide an opportunity for enhancement. Objective: To examine whether changes using already available data would enhance risk models and yield greater discrimination in hospital-level performance measures. Design, Setting, and Participants: This comparative effectiveness study used ICD-9-CM codes from all Medicare fee-for-service beneficiary claims for hospitalizations for acute myocardial infarction (AMI), heart failure (HF), or pneumonia among patients 65 years and older from July 1, 2013, through September 30, 2015. Changes to current CMS mortality risk models were applied incrementally to patient-level models, and the best model was tested on hospital performance measures to model 30-day mortality. Analyses were conducted from April 19, 2018, to September 19, 2018. Main Outcomes and Measures: The main outcome was all-cause death within 30 days of hospitalization for AMI, HF, or pneumonia, examined using 3 changes to current CMS mortality risk models: (1) incorporating present on admission coding to better exclude potential complications of care, (2) separating index admission diagnoses from those of the 12-month history, and (3) using ungrouped ICD-9-CM codes. Results: There were 361 175 hospital admissions (mean [SD] age, 78.6 [8.4] years; 189 225 [52.4%] men) for AMI, 716 790 hospital admissions (mean [SD] age, 81.1 [8.4] years; 326 825 [45.6%] men) for HF, and 988 225 hospital admissions (mean [SD] age, 80.7 [8.6] years; 460 761 [46.6%] men) for pneumonia during the study; mean 30-day mortality rates were 13.8% for AMI, 12.1% for HF, and 16.1% for pneumonia. Each change to the models was associated with incremental gains in C statistics. The best model, incorporating all changes, was associated with significantly improved patient-level C statistics, from 0.720 to 0.826 for AMI, 0.685 to 0.776 for HF, and 0.715 to 0.804 for pneumonia. Compared with current CMS models, the best model produced wider predicted probabilities with better calibration and Brier scores. Hospital risk-standardized mortality rates had wider distributions, with more hospitals identified as good or bad performance outliers. Conclusions and Relevance: Incorporating present on admission coding and using ungrouped index and historical ICD-9-CM codes were associated with improved patient-level and hospital-level risk models for mortality compared with the current CMS models for all 3 conditions.

Background Current strategies for ensuring the postmarket safety of medical devices are limited by small sample size and reliance on voluntary reporting of adverse events. Prospective, active surveillance of clinical registries may provide early warnings in the postmarket evaluation of medical device safety but has not been demonstrated in national clinical data registries. Methods and Results The CathPCI DELTA (Data Extraction and Longitudinal Trend Analysis) study was designed to assess the feasibility of prospective, active safety surveillance of medical devices within a national cardiovascular registry. We sought to assess the ability of our surveillance strategy to avoid false safety alerts by conducting an active safety surveillance study of aspiration thrombectomy catheters using data within the National Cardiovascular Data Registry CathPCI registry, where no difference in safety outcomes were anticipated for the primary in-hospital safety outcome of death and major adverse cardiovascular events (MACE). We performed a propensity-matched analysis of 5 aspiration thrombectomy catheter devices used during percutaneous coronary intervention among 95 925 patients presenting with ST-segment-elevation myocardial infarction between January 1, 2011 and September 30, 2013. After 33 months of surveillance, no safety alerts were triggered for the primary safety endpoints of death or MACE, with no between-catheter differences observed. The absolute risk of death during acute hospitalization ranged from 5.11% to 5.32% among the most commonly used aspiration thrombectomy catheter devices, with relative risks for death ranging from 0.96 to 1.03. The absolute risk of MACE ranged from 9.78% to 10.18%, with relative risks for MACE ranging from 0.99 to 1.02. There were no statistically significant differences in the rates of death or MACE between any of the aspiration thrombectomy catheter devices analyzed. Conclusions The CathPCI DELTA study demonstrates that prospective, active safety surveillance of national clinical registries is feasible to provide near-real-time safety assessments of new medical devices.

Importance: Predicting payments for particular conditions or populations is essential for research, benchmarking, public reporting, and calculations for population-based programs. Centers for Medicare & Medicaid Services (CMS) models often group codes into disease categories, but using single, rather than grouped, diagnostic codes and leveraging present on admission (POA) codes may enhance these models. Objective: To determine whether changes to the candidate variables in CMS models would improve risk models predicting patient total payment within 30 days of hospitalization for acute myocardial infarction (AMI), heart failure (HF), and pneumonia. Design, Setting, and Participants: This comparative effectiveness research study used data from Medicare fee-for-service hospitalizations for AMI, HF, and pneumonia at acute care hospitals from July 1, 2013, through September 30, 2015. Payments across multiple care settings, services, and supplies were included and adjusted for geographic and policy variations, corrected for inflation, and winsorized. The same data source was used but varied for the candidate variables and their selection, and the method used by CMS for public reporting that used grouped codes was compared with variations that used POA codes and single diagnostic codes. Combinations of use of POA codes, separation of index admission diagnoses from those in the previous 12 months, and use of individual International Classification of Diseases, Ninth Revision, Clinical Modification codes instead of grouped diagnostic categories were tested. Data analysis was performed from December 4, 2017, to June 10, 2019. Main Outcomes and Measures: The models' goodness of fit was compared using root mean square error (RMSE) and the McFadden pseudo R2. Results: Among the 1 943 049 total hospitalizations of the study participants, 343 116 admissions were for AMI (52.5% male; 37.4% aged ≤74 years), 677 044 for HF (45.5% male; 25.9% aged ≤74 years), and 922 889 for pneumonia (46.4% male; 28.2% aged ≤74 years). The mean (SD) 30-day payment was $23 103 ($18 221) for AMI, $16 365 ($12 527) for HF, and $17 097 ($12 087) for pneumonia. Each incremental model change improved the pseudo R2 and RMSE. Incorporating all 3 changes improved the pseudo R2 of the patient-level models from 0.077 to 0.129 for AMI, from 0.042 to 0.129 for HF, and from 0.114 to 0.237 for pneumonia. Parallel improvements in RMSE were found for all 3 conditions. Conclusions and Relevance: Leveraging POA codes, separating index from previous diagnoses, and using single diagnostic codes improved payment models. Better models can potentially improve research, benchmarking, public reporting, and calculations for population-based programs.

OBJECTIVES: Beginning in 2002, all 14 Massachusetts nonfederal cardiac surgery programs submitted Society of Thoracic Surgeons (STS) National Database data to the Massachusetts Data Analysis Center for mandatory state-based analysis and reporting, and to STS for nationally benchmarked analyses. We sought to determine whether longitudinal prevalences and trends in risk factors and observed and expected mortality differed between Massachusetts and the nation. METHODS: We analyzed 2003 to 2014 expected (STS predicted risk of operative [in-hospital + 30-day] mortality), observed, and risk-standardized isolated coronary artery bypass graft mortality using Massachusetts STS data (N = 39,400 cases) and national STS data (N = 1,815,234 cases). Analyses included percentage shares of total Massachusetts coronary artery bypass graft volume and expected mortality rates of 2 hospitals before and after outlier designation. RESULTS: Massachusetts patients had significantly higher odds of diabetes, peripheral vascular disease, low ejection fraction, and age ≥75 years relative to national data and lower odds of shock (odds ratio, 0.66; 99% confidence interval, 0.53-0.83), emergency (odds ratio, 0.57, 99% confidence interval, 0.52-0.61), reoperation, chronic lung disease, dialysis, obesity, and female sex. STS predicted risk of operative [in-hospital + 30-day] mortality for Massachusetts patients was higher than national rates during 2003 to 2007 (P < .001) and no different during 2008 to 2014 (P = .135). Adjusting for STS predicted risk of operative [in-hospital + 30-day] mortality, Massachusetts patients had significantly lower odds (odds ratio, 0.79; 99% confidence interval, 0.66-0.96) of 30-day mortality relative to national data. Outlier programs experienced inconsistent, transient influences on expected mortality and their percentage shares of Massachusetts coronary artery bypass graft cases. CONCLUSIONS: During 12 years of mandatory public reporting, Massachusetts risk-standardized coronary artery bypass graft mortality was consistently and significantly lower than national rates, expected rates were comparable or higher, and evidence for risk aversion was conflicting and inconclusive.

BACKGROUND: Telomeres cap and protect DNA but shorten with each somatic cell division. Aging and environmental and lifestyle factors contribute to the speed of telomere attrition. Current evidence suggests a link between relative telomere length (RTL) and depression but the directionality of the relationship remains unclear. We prospectively examined associations between RTL and subsequent depressive symptom trajectories. METHODS: Among 8,801 women of the Nurses' Health Study, depressive symptoms were measured every 4 years from 1992 to 2012; group-based trajectories of symptoms were identified using latent class growth-curve analysis. Multinomial logistic models were used to relate midlife RTLs to the probabilities of assignment to subsequent depressive symptom trajectory groups. RESULTS: We identified four depressive symptom trajectory groups: minimal depressive symptoms (62%), worsening depressive symptoms (14%), improving depressive symptoms (19%), and persistent-severe depressive symptoms (5%). Longer midlife RTLs were related to significantly lower odds of being in the worsening symptoms trajectory versus minimal trajectory but not to other trajectories. In comparison with being in the minimal symptoms group, the multivariable-adjusted odds ratio of being in the worsening depressive symptoms group was 0.78 (95% confidence interval, 0.62-0.97; p = 0.02), for every standard deviation increase in baseline RTL. CONCLUSIONS: In this large prospective study of generally healthy women, longer telomeres at midlife were associated with significantly lower risk of a subsequent trajectory of worsening mood symptoms over 20 years. The results raise the possibility of telomere shortening as a novel contributing factor to late-life depression.

BACKGROUND: Coercive measures is a topic that has long been discussed in the field of psychiatry. Despite global reports of reductions in the use of restraint episodes due to new regulations, it is still questionable if practices have really changed over time. For this study, we examined the rates of coercive measures in the inpatient population of psychiatric care providers across Finland to identify changing trends as well as variations in such trends by region. METHODS: In this nationwide registry analysis, we extracted patient data from the national database (The Finnish National Care Register for Health Care) over a 20-year period. We included adult patients admitted to psychiatric units (care providers) and focused on patients who had faced coercive measures (seclusion, limb restraints, forced injection and physical restraints) during their hospital stay. Multilevel logistical models (a polynomial model of quadratic form) were used to examine trends in prevalence of any coercive measures as well as the other four specified coercive measures over time, and to investigate variation in such trends among care providers and regions. RESULTS: Between 1995 and 2014, the dataset contained 226,948 inpatients who had been admitted during the 20-year time frame (505,169 treatment periods). The overall prevalence of coercive treatment on inpatients was 9.8%, with a small decrease during 2011-2014. The overall prevalence of seclusion, limb restraints, forced injection and physical restraints on inpatients was 6.9, 3.8, 2.6 and 0.8%, respectively. Only the use of limb restraints showed a downward trend over time. Geographic and care provider variations in specific coercive measures used were also observed. CONCLUSIONS: Despite the decreasing national level of coercive measures used in Finnish psychiatric hospitals, the overall reduction has been small during the last two decades. These results have implications on the future development of structured guidelines and interventions for preventing and more effectively managing challenging situations. Clinical guidelines and staff education related to the use of coercive measures should be critically assessed to ensure that the staff members working with vulnerable patient populations in psychiatric hospitals are ethically competent.

Importance: Medicare and other organizations have focused on improving quality of care for patients with acute myocardial infarction (AMI) over the last 2 decades. However, there is no comprehensive perspective on the evolution of outcomes for AMI during that period, and it is unknown whether temporal changes varied by patient subgroup, hospital, or county. Objective: To provide a comprehensive evaluation of national trends in inpatient outcomes and costs of AMI during this period. Design, Setting, and Participants: This cohort study included analysis of data from a sample of 4 367 485 Medicare fee-for-service beneficiaries aged 65 years or older from January 1, 1995, through December 31, 2014, across 5680 hospitals in the United States. Analyses were conducted from January 15 to June 5, 2018. Main Outcomes and Measures: Thirty-day all-cause mortality at the patient, hospital, and county levels. Additional outcomes included 30-day all-cause readmissions; 1-year recurrent AMI; in-hospital mortality; length of hospital stay; 2014 Consumer Price Index-adjusted median Medicare inpatient payment per AMI discharge; and rates of catheterization, percutaneous coronary intervention, and coronary artery bypass graft surgery. Results: The cohort included 4 367 485 Medicare fee-for-service patients aged 65 years or older hospitalized for AMI during the study period. Between 1995 and 2014, the mean (SD) age of patients increased from 76.9 (7.2) to 78.2 (8.7) years, the percentage of female patients declined from 49.5% to 46.1%, the percentage of white patients declined from 91.0% to 86.2%, and the percentage of black patients increased from 5.9% to 8.0%. There were declines in AMI hospitalizations (914 to 566 per 100 000 beneficiary-years); 30-day mortality (20.0% to 12.4%; difference, 7.6 percentage points; 95% CI, 7.3-7.8 percentage points); 30-day all-cause readmissions (21.0% to 15.3%; difference, 5.7 percentage points; 95% CI, 5.4-6.0 percentage points); and 1-year recurrent AMI (7.1% to 5.1%; difference, 2.0 percentage points; 95% CI, 1.8-2.2 percentage points). There were increases in the 2014 Consumer Price Index-adjusted median (interquartile range) Medicare inpatient payment per AMI discharge ($9282 [$6969-$12 173] to $11 031 [$8099-$16 861]); 30-day inpatient catheterization (44.2% to 59.9%; difference, 15.7 percentage points; 95% CI, 15.4-16.0 percentage points); and inpatient percutaneous coronary intervention (18.8% to 43.3%; difference, 24.5 percentage points; 95% CI, 24.2-24.7 percentage points). Coronary artery bypass graft surgery rates decreased from 14.4% to 10.2% (difference, 4.2 percentage points; 95% CI, 3.9-4.3 percentage points). There was heterogeneity by hospital and county in the mortality changes over time. Conclusions and Relevance: This study shows marked improvements in short-term mortality and readmissions, with an increase in in-hospital procedures and payments, for the increasingly smaller number of Medicare beneficiaries with AMI.

Importance: Although current federal quality improvement programs do not include cancer surgery, the Centers for Medicare & Medicaid Services and other payers are considering extending readmission reduction initiatives to include these and other common high-cost episodes. Objectives: To quantify between-hospital variation in quality-related outcomes and identify hospital characteristics associated with high and low performance. Design, Setting, and Participants: This retrospective cohort study obtained data through linkage of the California Cancer Registry to hospital discharge claims databases maintained by the California Office of Statewide Health Planning and Development. All 351 acute care hospitals in California at which 1 or more adults underwent curative intent surgery between January 1, 2007, and December 31, 2011, with analyses finalized July 15, 2018, were included. A total of 138 799 adults undergoing surgery for colorectal, breast, lung, prostate, bladder, thyroid, kidney, endometrial, pancreatic, liver, or esophageal cancer within 6 months of diagnosis, with an American Joint Committee on Cancer stage of I to III at diagnosis, were included. Main Outcomes and Measures: Measures included adjusted odds ratios and variance components from hierarchical mixed-effects logistic regression analyses of in-hospital mortality, 90-day readmission, and 90-day mortality, as well as hospital-specific risk-adjusted rates and risk-adjusted standardized rate ratios for hospitals with a mean annual surgical volume of 10 or more. Results: Across 138 799 patients at the 351 included hospitals, 8.9% were aged 18 to 44 years and 45.9% were aged 65 years or older, 57.4% were women, and 18.2% were nonwhite. Among these, 1240 patients (0.9%) died during the index admission. Among 137 559 patients discharged alive, 19 670 (14.3%) were readmitted and 1754 (1.3%) died within 90 days. After adjusting for patient case-mix differences, evidence of statistically significant variation in risk across hospitals was identified, as characterized by the variance of the random effects in the mixed model, for all 3 metrics (P < .001). In addition, substantial variation was observed in hospital performance profiles: across 260 hospitals with a mean annual surgical volume of 10 or more, 59 (22.7%) had lower-than-expected rates for all 3 metrics, 105 (40.4%) had higher-than-expected rates for 2 of the 3, and 19 (7.3%) had higher-than-expected rates for all 3 metrics. Conclusions and Relevance: Accounting for patient case-mix differences, there appears to be substantial between-hospital variation in in-hospital mortality, 90-day readmission, and 90-day mortality after cancer surgical procedures. Recognizing the multifaceted nature of hospital performance through consideration of mortality and readmission simultaneously may help to prioritize strategies for improving surgical outcomes.

Importance: Although studies have described differences in hospital outcomes by patient race and socioeconomic status, it is not clear whether such disparities are driven by hospitals themselves or by broader systemic effects. Objective: To determine patterns of racial and socioeconomic disparities in outcomes within and between hospitals for patients with acute myocardial infarction, heart failure, and pneumonia. Design, Setting, and Participants: Retrospective cohort study initiated before February 2013, with additional analyses conducted during the peer-review process. Hospitals in the United States treating at least 25 Medicare fee-for-service beneficiaries aged 65 years or older in each race (ie, black and white) and neighborhood income level (ie, higher income and lower income) for acute myocardial infarction, heart failure, and pneumonia between 2009 and 2011 were included. Main Outcomes and Measures: For within-hospital analyses, risk-standardized mortality rates and risk-standardized readmission rates for race and neighborhood income subgroups were calculated at each hospital. The corresponding ratios using intraclass correlation coefficients were then compared. For between-hospital analyses, risk-standardized rates were assessed according to hospitals' proportion of patients in each subgroup. These analyses were performed for each of the 12 analysis cohorts reflecting the unique combinations of outcomes (mortality and readmission), demographics (race and neighborhood income), and conditions (acute myocardial infarction, heart failure, and pneumonia). Results: Between 74% (3545 of 4810) and 91% (4136 of 4554) of US hospitals lacked sufficient racial and socioeconomic diversity to be included in this analysis, with the number of hospitals eligible for analysis varying among cohorts. The 12 analysis cohorts ranged in size from 418 to 1265 hospitals and from 144 417 to 703 324 patients. Within included hospitals, risk-standardized mortality rates tended to be lower among black patients (mean [SD] difference between risk-standardized mortality rates in black patients compared with white patients for acute myocardial infarction, -0.57 [1.1] [P = .47]; for heart failure, -4.7 [1.3] [P < .001]; and for pneumonia, -1.0 [2.0] [P = .05]). However, risk-standardized readmission rates among black patients were higher (mean [SD] difference between risk-standardized readmission rates in black patients compared with white patients for acute myocardial infarction, 4.3 [1.4] [P < .001]; for heart failure, 2.8 [1.8] [P < .001], and for pneumonia, 3.7 [1.3] [P < .001]). Intraclass correlation coefficients ranged from 0.68 to 0.79, indicating that hospitals generally delivered consistent quality to patients of differing races. While the coefficients in the neighborhood income analysis were slightly lower (0.46-0.60), indicating some heterogeneity in within-hospital performance, differences in mortality rates and readmission rates between the 2 neighborhood income groups were small. There were no strong, consistent associations between risk-standardized outcomes for white or higher-income neighborhood patients and hospitals' proportion of black or lower-income neighborhood patients. Conclusions and Relevance: Hospital performance according to race and socioeconomic status was generally consistent within and between hospitals, even as there were overall differences in outcomes by race and neighborhood income. This finding indicates that disparities are likely to be systemic, rather than localized to particular hospitals.

Importance: The US Hospital Readmissions Reduction Program (HRRP) was associated with reduced readmissions among Medicare beneficiaries hospitalized for acute myocardial infarction (AMI), heart failure (HF), and pneumonia. It is important to assess whether there has been a signal for concomitant harm with an increase in mortality. Objective: To evaluate whether the announcement or the implementation of HRRP was associated with an increase in either in-hospital or 30-day postdischarge mortality following hospitalization for AMI, HF, or pneumonia. Design, Setting, and Participants: In this cohort study, using Medicare data, all hospitalizations for AMI, HF, and pneumonia were identified among fee-for-service Medicare beneficiaries aged 65 years and older from January 1, 2006, to December 31, 2014. These were assessed for changes in trends for risk-adjusted rates of in-hospital and 30-day postdischarge mortality after announcement and implementation of the HRRP using an interrupted time series framework. Analyses were done in November 2017 and December 2017. Exposures: Announcement of the HRRP in March 2010, and implementation of its penalties in October 2012. Main Outcomes and Measures: Monthly risk-adjusted rates of in-hospital and 30-day postdischarge mortality. Results: The sample included 1.7 million AMI, 4 million HF, and 3.5 million pneumonia hospitalizations. Between 2006 and 2014, in-hospital mortality decreased for the 3 conditions (AMI, from 10.4% to 9.7%; HF, from 4.3% to 3.5%; pneumonia, from 5.3% to 4.0%) while 30-day postdischarge mortality decreased from 7.4% to 7.0% for AMI (P for trend < .001), but increased from 7.4% to 9.2% for HF (P for trend < .001) and from 7.6% to 8.6% for pneumonia (P for trend < .001). Before the HRRP announcement, monthly postdischarge mortality was stable for AMI (slope for monthly change, 0.002%; 95% CI, -0.001% to 0.006% per month), and increased by 0.004% (95% CI, 0.000% to 0.007%) per month for HF and by 0.005% (95% CI, 0.002% to 0.008%) per month for pneumonia. There were no inflections in slope around HRRP announcement or implementation (P > .05 for all). In contrast, there were significant negative deflections in slopes for readmission rates at HRRP announcement for all conditions. Conclusions and Relevance: Among Medicare beneficiaries, there was no evidence for an increase in in-hospital or postdischarge mortality associated with HRRP announcement or implementation-a period with substantial reductions in readmissions. The improvement in readmission was therefore not associated with any increase in in-hospital or 30-day postdischarge mortality.

High-dimensional data provide many potential confounders that may bolster the plausibility of the ignorability assumption in causal inference problems. Propensity score methods are powerful causal inference tools, which are popular in health care research and are particularly useful for high-dimensional data. Recent interest has surrounded a Bayesian treatment of propensity scores in order to flexibly model the treatment assignment mechanism and summarize posterior quantities while incorporating variance from the treatment model. We discuss methods for Bayesian propensity score analysis of binary treatments, focusing on modern methods for high-dimensional Bayesian regression and the propagation of uncertainty. We introduce a novel and simple estimator for the average treatment effect that capitalizes on conjugacy of the beta and binomial distributions. Through simulations, we show the utility of horseshoe priors and Bayesian additive regression trees paired with our new estimator, while demonstrating the importance of including variance from the treatment regression model. An application to cardiac stent data with almost 500 confounders and 9000 patients illustrates approaches and facilitates comparison with existing alternatives. As measured by a falsifiability endpoint, we improved confounder adjustment compared with past observational research of the same problem.

BACKGROUND: Technological advances can improve care and outcomes but are a primary driver of health care spending growth. Understanding diffusion and use of new oncology therapies is important, given substantial increases in prices and spending on such treatments. OBJECTIVES: Examine diffusion of bevacizumab, a novel (in 2004) and high-priced biologic cancer therapy, among US oncology practices during 2005-2012 and assess variation in use across practices. RESEARCH DESIGN: Population-based observational study. SETTING: A total of 2329 US practices providing cancer chemotherapy. PARTICIPANTS: Random 20% sample of 236,304 Medicare fee-for-service beneficiaries aged above 65 years in 2004-2012 undergoing infused chemotherapy for cancer. MEASURES: Diffusion of bevacizumab (cumulative time to first use and 10% use) in practices, variation in use across practices overall and by higher versus lower-value use. We used hierarchical models with practice random effects to estimate the between-practice variation in the probability of receiving bevacizumab and to identify factors associated with use. RESULTS: We observed relatively rapid diffusion of bevacizumab, particularly in independent practices and larger versus smaller practices. We observed substantial variation in use; the adjusted odds ratio (95% confidence interval) of bevacizumab use was 2.90 higher (2.73-3.08) for practices 1 SD above versus one standard deviation below the mean. Variation was less for higher-value [odds ratio=2.72 (2.56-2.89)] than lower-value uses [odds ratio=3.61 (3.21-4.06)]. CONCLUSIONS: Use of bevacizumab varied widely across oncology practices, particularly for lower-value indications. These findings suggest that interventions targeted to practices have potential for decreasing low-value use of high-cost cancer therapies.

Postmarket comparative effectiveness and safety analyses of therapeutic treatments typically involve large observational cohorts. We propose double robust machine learning estimation techniques for implantable medical device evaluations where there are more than two unordered treatments and patients are clustered in hospitals. This flexible approach also accommodates high-dimensional covariates drawn from clinical databases. The Massachusetts Data Analysis Center percutaneous coronary intervention cohort is used to assess the composite outcome of 10 drug-eluting stents among adults implanted with at least one drug-eluting stent in Massachusetts. We find remarkable discrimination between stents. A simulation study designed to mimic this coronary intervention cohort is also presented and produced similar results.