Aaron L. Schwartz

Early Performance of Accountable Care Organizations in Medicare

BACKGROUND In the Medicare Shared Savings Program (MSSP), accountable care organizations (ACOs) have financial incentives to lower spending and improve quality. We used quasi-experimental methods to assess the early performance of MSSP ACOs. METHODS Using Medicare claims from 2009 through 2013 and a difference-in-differences design, we compared changes in spending and in performance on quality measures from before the start of ACO contracts to after the start of the contracts between beneficiaries served by the 220 ACOs entering the MSSP in mid-2012 (2012 ACO cohort) or January 2013 (2013 ACO cohort) and those served by non-ACO providers (control group), with adjustment for geographic area and beneficiary characteristics. We analyzed the 2012 and 2013 ACO cohorts separately because entry time could reflect the capacity of an ACO to achieve savings. We compared ACO savings according to organizational structure, baseline spending, and concurrent ACO contracting with commercial insurers. RESULTS Adjusted Medicare spending and spending trends were similar in the ACO cohorts and the control group during the precontract period. In 2013, the differential change (i.e., the between-group difference in the change from the precontract period) in total adjusted annual spending was -

Comparison of Low-Value Care in Medicaid vs Commercially Insured Populations.

144 per beneficiary in the 2012 ACO cohort as compared with the control group (P=0.02), consistent with a 1.4% savings, but only -

Choosing Wisely for Syncope: Low‐Value Carotid Ultrasound Use

3 per beneficiary in the 2013 ACO cohort as compared with the control group (P=0.96). Estimated savings were consistently greater in independent primary care groups than in hospital-integrated groups among 2012 and 2013 MSSP entrants (P=0.005 for interaction). MSSP contracts were associated with improved performance on some quality measures and unchanged performance on others. CONCLUSIONS The first full year of MSSP contracts was associated with early reductions in Medicare spending among 2012 entrants but not among 2013 entrants. Savings were greater in independent primary care groups than in hospital-integrated groups.

Focusing on High-Cost Patients — The Key to Addressing High Costs?

IMPORTANCE Reducing unnecessary tests and treatments is a potentially promising approach for improving the value of health care. However, relatively little is known about whether insurance type or local practice patterns are associated with delivery of low-value care. OBJECTIVES To compare low-value care in the Medicaid and commercially insured populations, test whether provision of low-value care is associated with insurance type, and assess whether local practice patterns are associated with the provision of low-value care. DESIGN, SETTING, AND PARTICIPANTS This cross-sectional study of claims data from the Oregon Division of Medical Assistance Programs and the Oregon All-Payer All-Claims database included Medicaid and commercially insured adults aged 18 to 64 years. The study period was January 1, 2013, through December 31, 2013. MAIN OUTCOMES AND MEASURES Low-value care was assessed using 16 claims-based measures. Logistic regression was used to test the association between Medicaid vs commercial insurance coverage and low-value care and the association between Medicaid and commercial low-value care rates within primary care service areas (PCSAs). RESULTS This study included 286 769 Medicaid and 1 376 308 commercial enrollees in 2013. Medicaid enrollees were younger (167 847 [58.5%] of Medicaid enrollees were aged 18-34 years vs 505 628 [36.7%] of those with commercial insurance) but generally had worse health status compared with those with commercial insurance. Medicaid enrollees were also more likely to be female (180 363 [62.9%] vs 702 165 [51.0%]) and live in a rural area (120 232 [41.9%] vs 389 964 [28.3%]). A total of 10 304 of 69 338 qualifying Medicaid patients (14.9%; 95% CI, 14.6%-15.1%) received at least 1 low-value service during 2013; the corresponding rate for commercially insured patients was 35 739 of 314 023 (11.4%; 95% CI, 11.3%-11.5%). No consistent association was found between insurance type and low-value care. Compared with commercial patients, Medicaid patients were more likely to receive low-value care for 10 measures and less likely to receive low-value care for 5 others. For 7 of 11 low-value care measures, Medicaid patients were significantly more likely to receive low-value care if they resided in a PCSA with a higher rate of low-value care for commercial patients. CONCLUSIONS AND RELEVANCE Oregon Medicaid and commercially insured patients received moderate amounts of low-value care in 2013. No consistent association was found between insurance type and low-value care. However, Medicaid and commercial rates of low-value care were associated with one another within PCSAs. Low-value care may be more closely related to local practice patterns than to reimbursement generosity or insurance benefit structures.

Local Access to Family Planning Services and Female High School Dropout Rates.

Background The United States spends more than

Tracking Labor Demand with Online Job Postings: The Case of Health IT Workers and the HITECH Act

750 billion annually on tests and procedures that do not benefit patients. Although there is no physiological indication for carotid ultrasound in “simple” syncope in the absence of focal neurological signs or symptoms suggestive of stroke, there is concern that this practice remains common for routine syncope workups. Methods and Results We used a 5% random‐sample Medicare claims database to evaluate large‐scale national trends in utilization of low‐value carotid ultrasound imaging for simple syncope. We found that 16.5% of all Medicare beneficiaries with simple syncope underwent carotid imaging and 6.5% of all carotid ultrasounds ordered in 2009 were for this low‐value indication. These findings were complemented by a manual chart review of 313 patients at a large academic medical center who underwent carotid ultrasound for simple syncope over a 5‐year period. For the 48 (15.4%) of 313 patients with stenosis ≥50%, carotid ultrasound did not yield a causal diagnosis. Only 2% of the 313 patients imaged experienced a change in medications after a positive study, and <1% of patients underwent a carotid revascularization procedure. Conclusions These data suggest that carotid ultrasound for patients with uncomplicated syncope are still commonly ordered and may be an easy target for institutions striving to curtail low‐value care.

Discretionary Interpretations of Accountable Care Organization Data—Reply

In reducing wasteful health care utilization, patient-focused strategies targeting high-cost patients may be less effective than systems-focused strategies intended to reduce low-value services for everyone. But current incentives favor a focus on high-cost patients.

Measuring Low-Value Care in Medicare

OBJECTIVE: To assess whether geographic access to family planning services is associated with a reduced female high school dropout rate. METHODS: We conducted a retrospective cross-sectional study. We merged the location of Planned Parenthood and Title X clinics with microdata from the 2012–2013 American Community Surveys. The association between female high school dropout rates and local clinic access was assessed using nearest-neighbor matching estimation. Models included various covariates to account for sociodemographic differences across communities and male high school dropout rates to account for unmeasured community characteristics affecting educational outcomes. RESULTS: Our sample included 284,910 16- to 22-year-old females. The presence of a Planned Parenthood clinic was associated with a decrease (4.08% compared with 4.83%; relative risk ratio 0.84, P<.001) in female high school dropout rates. This association was consistent across several model specifications. The presence of a Title X clinic was associated with a decrease (4.79% compared with 5.07%; relative risk ratio 0.94, P=.03) in female high school dropout rates, an association that did not remain significant across model specifications. CONCLUSION: Local access to Planned Parenthood is associated with lower high school dropout rates in young women.

Performance Differences in Year 1 of Pioneer Accountable Care Organizations

Growth in the health information technology (health IT) workforce will be necessary for the widespread adoption of electronic health records called for by the Health Information Technology for Economic and Clinical Health (HITECH) provisions of the American Recovery and Reinvestment Act. However, the health IT workforce is difficult to track using existing sources of data. We introduce a novel method for measuring labor demand in markets not defined by standard industrial or occupational codes. Drawing from 84 million online help wanted postings, we create a dataset of 434,000 health IT–related job listings from 2007 to 2011 whose descriptions contain key phrases such as “electronic health record” or “clinical informatics.” We find that health IT–related job postings have grown substantially over time, tripling as a share of healthcare job postings since 2007. Trend‐break and difference‐in‐difference analyses suggest that health IT–related job postings accelerated following HITECH. According to our preferred specification, the legislation was associated with an 86 percent increase in monthly postings, or 162,000 additional postings overall.

Changes in Low-Value Services in Year 1 of the Medicare Pioneer Accountable Care Organization Program

In Reply Dr Brem believes our findings—that computer-aided detection (CAD) does not improve diagnostic accuracy of mammography—are curious and “incongruent with the overwhelming majority of peer-reviewed, published studies on the topic.” Her perspective is puzzling and ignores the large metaanalysis of 10 CAD studies which found that CAD significantly increased recall rates (and thus false-positive examinations) with no significant improvement in cancer detection rates compared with readings without CAD.1 Dr Brem selects 4 studies to support her perspective.2-5 Three studies, 2 from Europe2,3 and 1 from the United States,4 compared double reading to single reading with CAD. None measured single reading without CAD, and thus none were designed to evaluate the specific contribution of CAD to mammography performance. The single-site report4 from the United States by a single author that Dr Brem selects found no significant differences in sensitivity of single reading with CAD by a specialized mammographer compared with double reading by a generalist and a specialized mammographer.4 Dr Brem’s fourth reference is a retrospective cohort study5 of mammograms billed with and without CAD in a population of elderly women (80% aged 70-89 years and none younger than 67). In addition, no interpretations of any of the mammograms were available (the study was of Medicare claims data), and the authors could not control for confounding variables such as prior mammography, breast density, hormonal therapy, or radiologist. Fenton et al6 conclude from their study that diagnostic mammography, ultrasonography, and analysis of biopsy specimens were all significantly increased in this elderly population when CAD was used, and diagnoses of ductal carcinoma in situ increased significantly, as did surgery and potentially radiation in this elderly patient population. The benefits of these outcomes in women aged 70 to 89 years are unclear. More importantly, Dr Brem fails to cite the results and conclusions published subsequently by Fenton et al in their study6: “In conclusion, we found that, among large numbers of diverse facilities and radiologists, the use of computer software designed to improve the interpretation of mammograms was associated with significantly higher false positive rates, recall rates, and biopsy rates and with significantly lower overall accuracy in screening mammography than was nonuse.” As a practicing breast imager, I share Dr Brem’s passion to provide optimal care to patients. This optimal care includes providing high-quality examinations that add value and avoiding examinations that do not. As a researcher, I am equally passionate about the potential we have to extract more medical value from image data. The field of deep learning applied to image analysis is in its infancy and with rigorous and careful research will no doubt lead to new discoveries that add real value to image interpretation and our patients. Will rapid computing power and processing allow new frontiers in CAD and diagnosis? I would argue the potential is limitless. Are currently available commercial CAD products providing value to our patients? Careful scientific study has not found this to be true.