Do Commercial ACOs Save Money? Interpreting Diverse Evidence No abstract available

Five-year Impact of a Commercial Accountable Care Organization on Health Care Spending, Utilization, and Quality of Care Background: Accountable Care Organizations (ACOs) have proliferated after the passage of the Affordable Care Act in 2010. Few longitudinal ACO studies with continuous enrollees exist and most are short term. Objective: The objective of this study was to evaluate the long-term impact of a commercial ACO on health care spending, utilization, and quality outcomes among continuously enrolled members. Research Design: Retrospective cohort study design and propensity-weighted difference-in-differences approach were applied to examine performance changes in 2 ACO cohorts relative to 1 non-ACO cohort during the commercial ACO implementation in 2010–2014. Subjects: A total of 40,483 continuously enrolled members of a commercial health maintenance organization from 2008 to 2014. Measures: Cost, use, and quality metrics for various type of services in outpatient and inpatient settings. Results: The ACO cohorts had (1) increased inpatient and outpatient total spending in the first 2 years of ACO operation, but insignificant differential changes for the latter 3 years; (2) decreased outpatient spending in the latter 2 years through reduced primary care visits and lowered spending on specialists, testing, and imaging; (3) no differential changes in inpatient hospital spending, utilization, and quality measures for most of the 5 years; (4) favorable results for several quality measures in preventive and diabetes care domains in at least one of the 5 years. Conclusions: The commercial ACO improved outpatient process quality measures modestly and slowed outpatient spending growth by the fourth year of operation, but had a negligible impact on inpatient hospital cost, use, and quality measures.

Using Data From the Healthcare Cost and Utilization Project for State Health Policy Research Background: The Healthcare Cost and Utilization Project (HCUP), the nation’s most complete source of all-payer hospital care data, supports analyses at the national, regional, state and community levels. However, national HCUP data are often used in inappropriate ways in studies of state-specific issues. Objective: To describe the opportunities and challenges of using HCUP data to conduct state health policy research and to provide empirical examples of what can go wrong when using the national HCUP data inappropriately. Research design: Comparison of results from state-level analyses using national HCUP data and the state-specific HCUP data recommended by the Agency for Healthcare Research and Quality (AHRQ). Analyses included trends in state-specific rates of cesarean delivery and a difference-in-differences analysis of Connecticut’s Medicaid expansion. Subjects: Hospital discharges from the 2004 to 2011 HCUP Nationwide Inpatient Samples (NIS) and State Inpatient Databases (SID). Measures: Cesarean delivery rates, discharges per capita, and discharges by the payer. Results: State-level estimates derived from the NIS are volatile and often provide misleading policy conclusions relative to estimates from the SID. Conclusions: The NIS should not be used for state-level research. AHRQ provides resources to assist analysts with state-specific studies using SID files.

The Effects of the Affordable Care Act on Health Care Access and Utilization Among Asian American Subgroups Objectives: We examined changes in health care access and utilization associated with the Patient Protection and Affordable Care Act (ACA) for different Asian American subgroups relative to non-Latino whites (whites). Research Design: Using 2003–2017 California Health Interview Survey data, we examined changes in 4 health care access measures and 2 utilization measures among whites and 7 Asian American subgroups. We estimated the unadjusted and adjusted percentage point changes on the absolute scale from the pre-ACA to post-ACA periods. Adjusted estimates were obtained from multivariable logistic regression models that controlled for predisposing, enabling, and need factors. We also estimated the pre-ACA to post-ACA changes between whites and Asian American subgroups using a difference-in-difference approach. Results: After the ACA was implemented, uninsurance decreased among all Asian American subgroups, but improvements in disparities relative to whites in these measures were limited. In particular, Koreans had the largest absolute reduction in uninsurance (−16.8 percentage points) and were the only subgroup with a significant reduction in terms of disparities relative to whites (−10.1 percentage points). However, little or no improvement was observed in the other 3 access measures (having a usual source of care, delayed medical care in past year, or delayed prescription drug use in past year) and 2 utilization measures (having a physician visit or emergency department visit in past year). Conclusions: Despite coverage gains among Asian American subgroups, especially Koreans, disparities in access and utilization persisted across all Asian American subgroups.

Who Will be the Costliest Patients? Using Recent Claims to Predict Expensive Surgical Episodes Introduction: Surgery accounts for almost half of inpatient spending, much of which is concentrated in a subset of high-cost patients. To study the effects of surgeon and hospital characteristics on surgical expenditures, a way to adjust for patient characteristics is essential. Design: Using 100% Medicare claims data, we identified patients aged 66–99 undergoing elective inpatient surgery (coronary artery bypass grafting, colectomy, and total hip/knee replacement) in 2014. We calculated price-standardized Medicare payments for the surgical episode from admission through 30 days after discharge (episode payments). On the basis of predictor variables from 2013, that is, Elixhauser comorbidities, hierarchical condition categories, Medicare’s Chronic Conditions Warehouse (CCW), and total spending, we constructed models to predict the costs of surgical episodes in 2014. Results: All sources of comorbidity data performed well in predicting the costliest cases (Spearman correlation 0.86–0.98). Models on the basis of hierarchical condition categories had slightly superior performance. The costliest quintile of patients as predicted by the model captured 35%–45% of the patients in each procedure’s actual costliest quintile. For example, in hip replacement, 44% of the costliest quintile was predicted by the model’s costliest quintile. Conclusions: A significant proportion of surgical spending can be predicted using patient factors on the basis of readily available claims data. By adjusting for patient factors, this will facilitate future research on unwarranted variation in episode payments driven by surgeons, hospitals, or other market forces.

Using Diagnoses to Estimate Health Care Cost Risk in Canada Objective: Until recently, the options for summarizing Canadian patient complexity were limited to health risk predictive modeling tools developed outside of Canada. This study aims to validate a new model created by the Canadian Institute for Health Information (CIHI) for Canada’s health care environment. Research Design: This was a cohort study. Subjects: The rolling population eligible for coverage under Ontario’s Universal Provincial Health Insurance Program in the fiscal years (FYs) 2006/2007–2016/2017 (12–13 million annually) comprised the subjects. Measures: To evaluate model performance, we compared predicted cost risk at the individual level, on the basis of diagnosis history, with estimates of actual patient-level cost using “out-of-the-box” cost weights created by running the CIHI software “as is.” We next considered whether performance could be improved by recalibrating the model weights, censoring outliers, or adding prior cost. Results: We were able to closely match model performance reported by CIHI for their 2010–2012 development sample (concurrent R2=48.0%; prospective R2=8.9%) and show that performance improved over time (concurrent R2=51.9%; prospective R2=9.7% in 2014–2016). Recalibrating the model did not substantively affect prospective period performance, even with the addition of prior cost and censoring of cost outliers. However, censoring substantively improved concurrent period explanatory power (from R2=53.6% to 66.7%). Conclusions: We validated the CIHI model for 2 periods, FYs 2010/2011–2012/2013 and FYs 2014/2015—2016/2017. Out-of-the-box model performance for Ontario was as good as that reported by CIHI for the development sample based on 3-province data (British Columbia, Alberta, and Ontario). We found that performance was robust to variations in model specification, data sources, and time.

Modeling the Health and Budgetary Impacts of a Team-based Hypertension Care Intervention That Includes Pharmacists Objective: The objective of this study was to assess the potential health and budgetary impacts of implementing a pharmacist-involved team-based hypertension management model in the United States. Research Design: In 2017, we evaluated a pharmacist-involved team-based care intervention among 3 targeted groups using a microsimulation model designed to estimate cardiovascular event incidence and associated health care spending in a cross-section of individuals representative of the US population: implementing it among patients with: (1) newly diagnosed hypertension; (2) persistently (≥1 year) uncontrolled blood pressure (BP); or (3) treated, yet persistently uncontrolled BP—and report outcomes over 5 and 20 years. We describe the spending thresholds for each intervention strategy to achieve budget neutrality in 5 years from a payer’s perspective. Results: Offering this intervention could prevent 22.9–36.8 million person-years of uncontrolled BP and 77,200–230,900 heart attacks and strokes in 5 years (83.8–174.8 million and 393,200–922,900 in 20 years, respectively). Health and economic benefits strongly favored groups 2 and 3. Assuming an intervention cost of $525 per enrollee, the intervention generates 5-year budgetary cost-savings only for Medicare among groups 2 and 3. To achieve budget neutrality in 5 years across all groups, intervention costs per person need to be around $35 for Medicaid, $180 for private insurance, and $335 for Medicare enrollees. Conclusions: Adopting a pharmacist-involved team-based hypertension model could substantially improve BP control and cardiovascular outcomes in the United States. Net cost-savings among groups 2 and 3 make a compelling case for Medicare, but favorable economics may also be possible for private insurers, particularly if innovations could moderately lower the cost of delivering an effective intervention.

Comparative Responsiveness and Minimally Important Difference of Common Anxiety Measures Background: Anxiety is one of the most prevalent mental disorders and accounts for substantial disability as well as increased health care costs. This study examines the minimally important difference (MID) and responsiveness of 6 commonly used anxiety scales. Methods: The sample comprised 294 patients from 6 primary care clinics in a single VA medical center who were enrolled in a telecare trial for treatment of chronic musculoskeletal pain and comorbid depression and/or anxiety. The measures assessed were the Patient Reported Outcomes Measurement Information System (PROMIS) 4-item, 6-item, and 8-item anxiety scales; the Generalized Anxiety Disorder 7-item scale (GAD-7); the Symptom Checklist anxiety subscale (SCL); the Posttraumatic Stress Disorder Checklist (PCL); the Short Form (SF)-36 Mental Health subscale; and the SF-12 Mental Component Summary (MCS). Validity was assessed with correlations of these measures with one another and with measures of quality of life and disability. MID was estimated by triangulating several methods. Responsiveness was evaluated by comparing: (a) the standardized response means for patients who reported their mood as being better, the same, or worse at 3 months; (b) the area under the curve for patients who had improved (better) versus those who had not (same/worse). Results: Convergent and construct validity was supported by strong correlations of the anxiety measures with one another and moderate correlations with quality of life and disability measures, respectively. All measures differentiated patients who reported global improvement at 3 months from those who were unchanged, but were less able to distinguish worsening from no change. The area under the curves showed comparable responsiveness of the scales. The estimated MID was 4 for the PROMIS scales; 3 for the GAD-7; 6 for the PCL; 9 for the SF-36 mental health subscale; 5 for the MCS score, and 0.3 for the SCL anxiety scale. Conclusions: Six commonly used anxiety scales demonstrate similar responsiveness, and estimated MIDs can be used to gauge anxiety change in clinical research and practice.

Patient Possession of Excess Medication Supply in the VA: A Retrospective Database Study Background: Medication overlap leading to medication excess is a form of therapeutic duplication, itself a type of potentially inappropriate prescribing. Objective: To determine the prevalence of potential medication excess in the Veterans Health Administration (VHA) and identify associated medication-level, patient-level, and system-level factors. Research Design: A retrospective database study. Subjects: All veterans who received ≥1 prescription dispensed by a VHA pharmacy in fiscal year 2014. Measures: The primary outcome of “medication excess” was defined for each patient as the number of excess days’ worth of medications for all overlap episodes (concurrently dispensed medications with the same name for >10 d). Predictors included medication-level, patient-level, and system-level factors. Multivariable negative binomial regression analyses estimated the rate ratio of each predictor with medication excess. Results: Among 4,687,453 veterans, 64% had ≥1 medication overlap episodes. Patients were prescribed a median of 7 [interquartile range (IQR), 3–12] unique medications, had a median of 2 (IQR, 0–5) overlap episodes, and a median of 27 (IQR, 0–96) days of medication excess. In adjusted regression models, factors associated with greater risk of medication excess included having more comorbidities, multiple prescribers, a combination of filling locations (consolidated mail-order pharmacy vs. local pharmacy), and multiple prescription durations (≥90 d vs. less). Conclusions: Medication excess was high among VHA users, with nearly two-thirds of patients experiencing at least 1 duplicative medication. As systems such as mail-order pharmacies and 90-day supply are increasingly implemented to reduce costs and improve medication adherence, it is important to recognize the potential for systems-level inefficiencies and potentially inappropriate prescribing.

Use of Medicare Data to Identify Team-based Primary Care: Is it Possible? Background: It is unclear whether Medicare data can be used to identify type and degree of collaboration between primary care providers (PCPs) [medical doctors (MDs), nurse practitioners, and physician assistants] in a team care model. Methods: We surveyed 63 primary care practices in Texas and linked the survey results to 2015 100% Medicare data. We identified PCP dyads of 2 providers in Medicare data and compared the results to those from our survey. Sensitivity, specificity, and positive predictive value (PPV) of dyads in Medicare data at different threshold numbers of shared patients were reported. We also identified PCPs who work in the same practice by Social Network Analysis (SNA) of Medicare data and compared the results to the surveys. Results: With a cutoff of sharing at least 30 patients, the sensitivity of identifying dyads was 27.8%, specificity was 91.7%, and PPV 72.2%. The PPV was higher for MD-nurse practitioner/physician assistant pairs (84.4%) than for MD-MD pairs (61.5%). At the same cutoff, 90% of PCPs identified in a practice from the survey were also identified by SNA in the corresponding practice. In 5 of 8 surveyed practices with at least 3 PCPs, about ≤20% PCPs identified in the practices by SNA of Medicare data were not identified in the survey. Conclusions: Medicare data can be used to identify shared care with low sensitivity and high PPV. Community discovery from Medicare data provided good agreement in identifying members of practices. Adapting network analyses in different contexts needs more validation studies.