Assessment of the E-value in the presence of bias amplification: a simulation study.

BACKGROUND: The E-value, a measure that has received recent attention in the comparative effectiveness literature, reports the minimum strength of association between an unmeasured confounder and the treatment and outcome that would explain away the estimated treatment effect. This study contributes to the literature on the applications and interpretations of E-values by examining how the E-value is impacted by data with varying levels of association of unobserved covariates with the treatment and outcome measure when covariate adjustment is applied. We calculate the E-value after using regression and propensity score methods (PSMs) to adjust for differences in observed covariates. Propensity score methods are a common observational research method used to balance observed covariates between treatment groups. In practice, researchers may assume propensity score methods that balance treatment groups across observed characteristics will extend to balance of unobserved characteristics. However, that assumption is not testable and has been shown to not hold in realistic data settings. We assess the E-value when covariate adjustment affects the imbalance in unobserved covariates. METHODS: Our study uses Monte Carlo simulations to evaluate the impact of unobserved confounders on the treatment effect estimates and to evaluate the performance of the E-Value sensitivity test with the application of regression and propensity score methods under varying levels of unobserved confounding. Specifically, we compare observed and unobserved confounder balance, odds ratios of treatment vs. control, and E-Value sensitivity test statistics from generalized linear model (GLM) regression models, inverse-probability weighted models, and propensity score matching models, over correlations of increasing strength between observed and unobserved confounders. RESULTS: We confirm previous findings that propensity score methods - matching or weighting - may increase the imbalance in unobserved confounders. The magnitude of the effect depends on the strength of correlation between the confounder, treatment, and outcomes. We find that E-values calculated after applying propensity score methods tend to be larger when unobserved confounders result in more biased treatment effect estimates. CONCLUSIONS: The E-Value may misrepresent the size of the unobserved effect needed to change the magnitude of the association between treatment and outcome when propensity score methods are used. Thus, caution is warranted when interpreting the E-Value in the context of propensity score methods.

Use of Electronic Health Records to Characterize Patients with Uncontrolled Hypertension in Two Large Health System Networks.

Background: Improving hypertension control is a public health priority. However, consistent identification of uncontrolled hypertension using computable definitions in electronic health records (EHR) across health systems remains uncertain. Methods: In this retrospective cohort study, we applied two computable definitions to the EHR data to identify patients with controlled and uncontrolled hypertension and to evaluate differences in characteristics, treatment, and clinical outcomes between these patient populations. We included adult patients (≥ 18 years) with hypertension receiving ambulatory care within Yale-New Haven Health System (YNHHS; a large US health system) and OneFlorida Clinical Research Consortium (OneFlorida; a Clinical Research Network comprised of 16 health systems) between October 2015 and December 2018. We identified patients with controlled and uncontrolled hypertension based on either a single blood pressure (BP) measurement from a randomly selected visit or all BP measurements recorded between hypertension identification and the randomly selected visit). Results: Overall, 253,207 and 182,827 adults at YNHHS and OneFlorida were identified as having hypertension. Of these patients, 83.1% at YNHHS and 76.8% at OneFlorida were identified using ICD-10-CM codes, whereas 16.9% and 23.2%, respectively, were identified using elevated BP measurements (≥ 140/90 mmHg). Uncontrolled hypertension was observed among 32.5% and 43.7% of patients at YNHHS and OneFlorida, respectively. Uncontrolled hypertension was disproportionately higher among Black patients when compared with White patients (38.9% versus 31.5% in YNHHS; p < 0.001; 49.7% versus 41.2% in OneFlorida; p < 0.001). Medication prescription for hypertension management was more common in patients with uncontrolled hypertension when compared with those with controlled hypertension (overall treatment rate: 39.3% versus 37.3% in YNHHS; p = 0.04; 42.2% versus 34.8% in OneFlorida; p < 0.001). Patients with controlled and uncontrolled hypertension had similar rates of short-term (at 3 and 6 months) and long-term (at 12 and 24 months) clinical outcomes. The two computable definitions generated consistent results. Conclusions: Our findings illustrate the potential of leveraging EHR data, employing computable definitions, to conduct effective digital population surveillance in the realm of hypertension management.

Use of Electronic Health Records to Characterize Patients with Uncontrolled Hypertension in Two Large Health System Networks.

Background: Improving hypertension control is a public health priority. However, uncertainty remains regarding the optimal way to identify patients with uncontrolled hypertension using electronic health records (EHR) data. Methods: In this retrospective cohort study, we applied computable definitions to the EHR data to identify patients with controlled and uncontrolled hypertension and to evaluate differences in characteristics, treatment, and clinical outcomes between these patient populations. We included adult patients (≥18 years) with hypertension receiving ambulatory care within Yale-New Haven Health System (YNHHS; a large US health system) and OneFlorida Clinical Research Consortium (OneFlorida; a Clinical Research Network comprised of 16 health systems) between October 2015 and December 2018. We identified patients with controlled and uncontrolled hypertension based on either a single blood pressure (BP) measurement from a randomly selected visit or all BP measurements recorded between hypertension identification and the randomly selected visit). Results: Overall, 253,207 and 182,827 adults at YNHHS and OneFlorida were identified as having hypertension. Of these patients, 83.1% at YNHHS and 76.8% at OneFlorida were identified using ICD-10-CM codes, whereas 16.9% and 23.2%, respectively, were identified using elevated BP measurements (≥ 140/90 mmHg). Uncontrolled hypertension was observed among 32.5% and 43.7% of patients at YNHHS and OneFlorida, respectively. Uncontrolled hypertension was disproportionately higher among Black patients when compared with White patients (38.9% versus 31.5% in YNHHS; p<0.001; 49.7% versus 41.2% in OneFlorida; p<0.001). Medication prescription for hypertension management was more common in patients with uncontrolled hypertension when compared with those with controlled hypertension (overall treatment rate: 39.3% versus 37.3% in YNHHS; p=0.04; 42.2% versus 34.8% in OneFlorida; p<0.001). Patients with controlled and uncontrolled hypertension had similar rates of short-term (at 3 and 6 months) and long-term (at 12 and 24 months) clinical outcomes. The two computable definitions generated consistent results. Conclusions: Computable definitions can be successfully applied to health system EHR data to conduct population surveillance for hypertension and identify patients with uncontrolled hypertension who may benefit from additional treatment.

Assessment of the Social Security Administration Death Master File for Comparative Analysis Studies of Peripheral Vascular Devices.

OBJECTIVES: The objective of this study was to assess the reliability the Social Security Administration Death Master File (SSADMF) for evaluating mortality in comparative peripheral vascular device studies. METHODS: We leveraged 2 versions of an administrative claims data set that were identical except for the source of mortality data. The SSADMF was the primary source of mortality records in one version. The SSADMF was combined with mortality from Medicare beneficiary records in the other. Our study was set in the context of a comparative effectiveness analysis of recent Food and Drug Administration interest involving peripheral paclitaxel-coated devices. Mortality of patients with Medicare Advantage insurance coverage from 2015 to 2018 who underwent femoropopliteal artery revascularization with a drug-coated device (DCD) or non-DCD was assessed through 2019. Covariate differences between treatment groups were adjusted by inverse propensity treatment weighting. The hazard ratio of DCD to non-DCD mortality was estimated using Cox regression. RESULTS: The cumulative incidences of mortality differed substantially between versions of the data. Nevertheless, we could not reject the null hypothesis that the hazard ratios of the SSADMF (1.05; 95% confidence interval 0.95-1.17) and the Master Beneficiary Summary File/SSADMF (1.03; 95% confidence interval 0.96-1.11) were the same (P = .63). CONCLUSIONS: The SSADMF is a common source of mortality records in the United States that can be linked to real-world data sources but is known to underreport mortality rates. We find that the SSADMF provides a reliable source of all-cause mortality for a comparative study assessing the safety of peripheral vascular devices.

Association between hospital-insurer contract structure and hospital performance.

OBJECTIVES: To describe the association between the form of hospitals' contracts-either markup from a benchmark or a discount from a list price-and performance: price, charge, cost, and length of stay. STUDY DESIGN: Retrospective observational study using administrative claims data matched with hospital characteristics from the American Hospital Association Annual Survey and the Healthcare Cost Reporting Information System. Data include a balanced panel of 1889 general acute care hospitals for the years 2011 to 2015. METHODS: Inpatient hospital commercial claims data from the Health Care Cost Institute were used to classify claims by contract type based on claim-line billed and allowed charges. Hospital-level performance measures-prices, charges, costs, and length of stay-were analyzed using linear regression models to identify the association of each measure with contract types. All measures were risk adjusted to control for differences in hospital case mix, and the regression specifications controlled for numerous hospital characteristics. RESULTS: Our estimate of the distribution of contract types in the data is similar to estimates using other methods. We find that discounted charges contracts are associated with higher prices and higher costs but not higher charges. Fixed-rate contracts are associated with lower prices as well as lower costs. CONCLUSIONS: Limited research exists on the relationship between contract structure and hospital performance. Our results suggest that hospital performance is related to contract structure, possibly due to factors such as differences in bargaining strategies or ex post incentives.

Long-term safety of drug-coated devices for peripheral revascularisation.

BACKGROUND: Meta-analyses of randomised trials of paclitaxel-coated peripheral devices found an association with worse long-term survival. AIMS: We aimed to assess long-term mortality in patients treated with drug-coated versus non-drug-coated devices who are insured by Medicare Advantage (MA), an alternative to traditional Medicare that represents >30% of the Medicare eligible population. We analysed data from an MA administrative claims data source that includes both inpatient and outpatient femoropopliteal artery revascularisation procedures. METHODS: Patients treated with or without drug-coated devices for femoropopliteal artery revascularisation from 4/2015-12/2017 were studied using Optum's De-identified Clinformatics Datamart Database. Mortality was assessed up to December 2019 using Kaplan-Meier cumulative mortality curves and Cox proportional hazard models. Inverse probability of treatment weighting was used to adjust for differences between groups. RESULTS: Of 16,796 patients revascularised, 4,427 (26.4%) were treated with drug-coated devices: 3,600 (81.3%) balloons and 827 (18.7%) stents. The median follow-up was 2.66 years (IQR 2.02-3.52). Treatment with drug-coated devices was associated with similar long-term mortality to non-drug-coated devices (adjusted HR 1.03, 95% CI: 0.96-1.10; p=0.39). Results were comparable for patients treated with balloons alone (adjusted HR 1.00, 95% CI: 0.92-1.08; p=0.96) or stents (adjusted HR 1.02, 95% CI: 0.88-1.18; p=0.78). These findings did not differ based on treatment setting, disease severity, age, sex or comorbidity burden (interaction p>0.05 for all). CONCLUSIONS: In this large cohort, there was no evidence of increased long-term mortality following treatment with drug-coated devices.

Treatment Of Opioid Use Disorder Among Commercially Insured US Adults, 2008-17.

There is abundant literature on efforts to reduce opioid prescriptions and misuse, but comparatively little on the treatment provided to people with opioid use disorder (OUD). Using claims data representing 12-15 million nonelderly adults covered through commercial group insurance during the period 2008-17, we explored rates of OUD diagnoses, treatment patterns, and spending. We found three key patterns: The rate of diagnosed OUD nearly doubled during 2008-17, and the distribution has shifted toward older age groups; the likelihood that diagnosed patients will receive any treatment has declined, particularly among those ages forty-five and older, because of a reduction in the use of medication-assisted treatment (MAT) and despite clinical evidence demonstrating its efficacy; and treatment spending is lower for patients who choose MAT. These patterns suggest that policies supporting the use of MAT are critical to addressing the undertreatment of OUD among the commercially insured and that further research to assess the cost-effectiveness of treatment with versus without medication is needed.

Area-level deprivation and preterm birth: results from a national, commercially-insured population.

BACKGROUND: Area-level deprivation is associated with multiple adverse birth outcomes. Few studies have examined the mediating pathways through which area-level deprivation affects these outcomes. The objective of this study was to investigate the association between area-level deprivation and preterm birth, and examine the mediating effects of maternal medical, behavioural, and psychosocial factors. METHODS: We conducted a retrospective cohort study using national, commercial health insurance claims data from 2011, obtained from the Health Care Cost Institute. Area-level deprivation was derived from principal components methods using ZIP code-level data. Multilevel structural equation modeling was used to examine mediating effects. RESULTS: In total, 138,487 women with a live singleton birth residing in 14,577 ZIP codes throughout the United States were included. Overall, 5.7% of women had a preterm birth. In fully adjusted generalized estimation equation models, compared to women in the lowest quartile of area-level deprivation, odds of preterm birth increased by 9.6% among women in the second highest quartile (odds ratio (OR) 1.096; 95% confidence interval (CI) 1.021, 1.176), by 11.3% in the third highest quartile (OR 1.113; 95% CI 1.035, 1.195), and by 24.9% in the highest quartile (OR 1.249; 95% CI 1.165, 1.339). Hypertension and infection moderately mediated this association. CONCLUSIONS: Even among commercially-insured women, area-level deprivation was associated with increased risk of preterm birth. Similar to individual socioeconomic status, area-level deprivation does not have a threshold effect. Implementation of policies to reduce area-level deprivation, and the screening and treatment of maternal mediators may be associated with a lower risk of preterm birth.

Health Care Spending Under Employer-Sponsored Insurance: A 10-Year Retrospective.

Using a national sample of health care claims data from the Health Care Cost Institute, we found that total spending per capita (not including premiums) on health services for enrollees in employer-sponsored insurance plans increased by 44 percent from 2007 through 2016 (average annual growth of 4.1 percent). Spending increased across all major categories of health services, although the increases were not uniform across years or categories. Growth rates for total per capita spending generally slowed after 2009 but increased between 2014 and 2016. Spending on outpatient services grew more quickly (average annual growth of 5.7 percent) compared to spending on the other types of services. However, the overall distribution of spending across categories remained largely unchanged. In the context of the dramatic economic and policy events that have taken place since 2007-including the Great Recession, the Affordable Care Act, and numerous medical innovations-this assessment of ten-year spending trends provides insights into how the largest insured population in the US contributes to health care spending growth.

Medicare Competitive Bidding Program Realized Price Savings For Durable Medical Equipment Purchases.

From the inception of the Medicare program there have been questions regarding whether and how to pay for durable medical equipment, prosthetics, orthotics, and supplies. In 2011 the Centers for Medicare and Medicaid Services (CMS) implemented a competitive bidding program to reduce spending on durable medical equipment and similar items. Previously, CMS had used prices in an administrative fee schedule to reimburse for these items. We compared prices from Round 1 of the Medicare competitive bidding program, which were established for the periods 2011-13 and 2014-16, to prices paid by national commercial insurers for the same types of items in 2011-14. Our results suggest that the initial years of the program produced prices comparable to those obtained, on average, by large commercial insurers-sophisticated purchasers that presumably were able to negotiate prices with suppliers of durable medical equipment and similar items.

Association Between Maternal Comorbidities and Emergency Department Use Among a National Sample of Commercially Insured Pregnant Women.

OBJECTIVES: Evidence suggests that, despite routine engagement with the health system, pregnant women commonly seek emergency care. The objectives of this study were to examine the association between maternal comorbidities and emergency department (ED) use among a national sample of commercially insured pregnant women. METHODS: We conducted a retrospective cohort study using multipayer medical claims data maintained by the Health Care Cost Institute for women ages 18 to 44 years with a live singleton birth in 2011 (N = 157,786). The association between common maternal comorbidities (e.g., hypertension, gestational diabetes) and ED use during pregnancy was examined using multilevel models, while controlling for age, region, and residential zip code. RESULTS: Twenty percent (n = 31,413) of pregnant women had one or more ED visit (mean ± SD = 1.52 ± 1.15). Among those who used the ED, 29% had two or more visits, and 11% had three or more visits. Emergency care seekers were significantly more likely to have one or more comorbid condition compared to those with no emergency care: 30% versus 21%, respectively (p < 0.001). Pregnant women with asthma had 2.5 times the likelihood of having had any ED visit (adjusted odds ratio [AOR] = 2.46, 95% confidence interval [CI] = 2.32-2.62). There was a significant increase in the probability (approximately 50%) of ED use among pregnant women with diabetes (AOR = 1.47, 95% CI = 1.33-1.63) or hypertension (AOR = 1.49, 95% CI = 1.43-1.55) or who were obese (AOR = 1.55, 95% CI = 1.47-1.64). Increased odds associated with gestational diabetes were more modest, resulting in a 13% increased odds of using the ED (AOR = 1.13, 95% CI = 1.07-1.18). Less than 0.6% of pregnant women (n = 177) received emergency care that resulted in a hospital admission. The admission rate was 0.4% (189 admissions/47,608 ED visits). CONCLUSIONS: Among pregnant women, comorbidity burden was associated with more ED utilization. Efforts to reduce acute unscheduled care and improve care coordination during pregnancy should target interventions to patient comorbidity.

Maternal Medical Complexity: Impact on Prenatal Health Care Spending among Women at Low Risk for Cesarean Section.

BACKGROUND: Obstetric procedures are among the most expensive health care services, yet relatively little is known about health care spending among pregnant women, particularly the commercially-insured. OBJECTIVE: The objective of this study was to examine the association between maternal medical complexity, as a result of having one or more comorbid conditions, and health care spending during the prenatal period among a national sample of 95,663 commercially-insured women at low risk for cesarean delivery. METHODS: We conducted secondary analyses of 2010-2011 inpatient, outpatient, and professional claims for health care services from the Health Care Cost Institute. Allowed charges were summed for the prenatal and childbirth periods. Ordinary least squares regressions tested associations between maternal health conditions and health care expenditures during pregnancy. RESULTS: Thirty-four percent of pregnant women had one or more comorbidities; 8% had two or more. Pregnant women with one or more comorbidities had significantly higher allowed charges than those without comorbidities (p < .001). Spending during the prenatal period was nearly three times higher for women with preexisting diabetes compared with women with no comorbid conditions. Average levels of prenatal period spending associated with maternal comorbidities were similar for women who had vaginal and cesarean deliveries. Patient characteristics accounted for 30% of the variance in prenatal period expenditures. CONCLUSIONS: The impact of maternal comorbidities, and in particular preexisting diabetes, on prenatal care expenditures should be taken into account as provider payment reforms, such as pay-for performance incentives and bundled payments for episodes of care, extend to maternal and child health-related services.

Prices For Common Medical Services Vary Substantially Among The Commercially Insured.

Using a national multipayer commercial claims database containing allowed amounts, we examined variations in the prices for 242 common medical services in forty-one states and the District of Columbia. Ratios of average state prices to national prices ranged from a low of 0.79 in Florida to a high of 2.64 in Alaska. Two- to threefold variations in prices were identified within some states and Metropolitan Statistical Areas.