Changes in Hospital Adverse Events and Patient Outcomes Associated With Private Equity Acquisition.

Importance: The effects of private equity acquisitions of US hospitals on the clinical quality of inpatient care and patient outcomes remain largely unknown. Objective: To examine changes in hospital-acquired adverse events and hospitalization outcomes associated with private equity acquisitions of US hospitals. Design, Setting, and Participants: Data from 100% Medicare Part A claims for 662 095 hospitalizations at 51 private equity-acquired hospitals were compared with data for 4 160 720 hospitalizations at 259 matched control hospitals (not acquired by private equity) for hospital stays between 2009 and 2019. An event study, difference-in-differences design was used to assess hospitalizations from 3 years before to 3 years after private equity acquisition using a linear model that was adjusted for patient and hospital attributes. Main Outcomes and Measures: Hospital-acquired adverse events (synonymous with hospital-acquired conditions; the individual conditions were defined by the US Centers for Medicare & Medicaid Services as falls, infections, and other adverse events), patient mix, and hospitalization outcomes (including mortality, discharge disposition, length of stay, and readmissions). Results: Hospital-acquired adverse events (or conditions) were observed within 10 091 hospitalizations. After private equity acquisition, Medicare beneficiaries admitted to private equity hospitals experienced a 25.4% increase in hospital-acquired conditions compared with those treated at control hospitals (4.6 [95% CI, 2.0-7.2] additional hospital-acquired conditions per 10 000 hospitalizations, P = .004). This increase in hospital-acquired conditions was driven by a 27.3% increase in falls (P = .02) and a 37.7% increase in central line-associated bloodstream infections (P = .04) at private equity hospitals, despite placing 16.2% fewer central lines. Surgical site infections doubled from 10.8 to 21.6 per 10 000 hospitalizations at private equity hospitals despite an 8.1% reduction in surgical volume; meanwhile, such infections decreased at control hospitals, though statistical precision of the between-group comparison was limited by the smaller sample size of surgical hospitalizations. Compared with Medicare beneficiaries treated at control hospitals, those treated at private equity hospitals were modestly younger, less likely to be dually eligible for Medicare and Medicaid, and more often transferred to other acute care hospitals after shorter lengths of stay. In-hospital mortality (n = 162 652 in the population or 3.4% on average) decreased slightly at private equity hospitals compared with the control hospitals; there was no differential change in mortality by 30 days after hospital discharge. Conclusions and Relevance: Private equity acquisition was associated with increased hospital-acquired adverse events, including falls and central line-associated bloodstream infections, along with a larger but less statistically precise increase in surgical site infections. Shifts in patient mix toward younger and fewer dually eligible beneficiaries admitted and increased transfers to other hospitals may explain the small decrease in in-hospital mortality at private equity hospitals relative to the control hospitals, which was no longer evident 30 days after discharge. These findings heighten concerns about the implications of private equity on health care delivery.

Prescription Drug Spending in Fee-for-Service Medicare, 2008-2019.

Importance: Prescription drug spending is a topic of increased interest to the public and policymakers. However, prior assessments have been limited by focusing on retail spending (Part D-covered drugs), omitting clinician-administered (Part B-covered) drug spending, or focusing on all fee-for-service Medicare beneficiaries, regardless of their enrollment into prescription drug coverage. Objective: To estimate the proportion of health care spending contributed by prescription drugs and to assess spending for retail and clinician-administered prescriptions. Design, Setting, and Participants: Descriptive, serial, cross-sectional analysis of a 20% random sample of fee-for-service Medicare beneficiaries in the United States from 2008 to 2019 who were continuously enrolled in Parts A (hospital), B (medical), and D (prescription drug) benefits, and not in Medicare Advantage. Exposure: Calendar year. Main Outcomes and Measures: Net spending on retail (Part D-covered) and clinician-administered (Part B-covered) prescription drugs; prescription drug spending (spending on Part B-covered and Part D-covered drugs) as a percentage of total per-capita health care spending. Measures were adjusted for inflation and for postsale rebates (for Part D-covered drugs). Results: There were 3 201 284 beneficiaries enrolled in Parts A, B, and D in 2008 and 4 502 718 in 2019. In 2019, beneficiaries had a mean (SD) age of 71.7 (12.0) years, documented sex was female for 57.7%, and 69.5% had no low-income subsidies. Total per-capita spending was $16 345 in 2008 and $20 117 in 2019. Comparing 2008 with 2019, per-capita Part A spending was $7106 (95% CI, $7084-$7128) vs $7120 (95% CI, $7098-$7141), Part B drug spending was $720 (95% CI, $713-$728) vs $1641 (95% CI, $1629-$1653), Part B nondrug spending was $5113 (95% CI, $5105-$5122) vs $6702 (95% CI, $6692-$6712), and Part D net spending was $3122 (95% CI, $3117-$3127) vs $3477 (95% CI, $3466-$3489). The proportion of total annual spending attributed to prescription drugs increased from 24.0% in 2008 to 27.2% in 2019, net of estimated rebates and discounts. Conclusions and Relevance: In 2019, spending on prescription drugs represented approximately 27% of total spending among fee-for-service Medicare beneficiaries enrolled in Part D, even after accounting for postsale rebates.

Admission diagnoses among patients with heart failure: Variation by ACO performance on a measure of risk-standardized acute admission rates.

BACKGROUND: A key quality metric for Accountable Care Organizations (ACOs) is the rate of hospitalization among patients with heart failure (HF). Among this patient population, non-HF-related hospitalizations account for a substantial proportion of admissions. Understanding the types of admissions and the distribution of admission types across ACOs of varying performance may provide important insights for lowering admission rates. METHODS: We examined admission diagnoses among 220 Medicare Shared Savings Program ACOs in 2013. ACOs were stratified into quartiles by their performance on a measure of unplanned risk-standardized acute admission rates (RSAARs) among patients with HF. Using a previously validated algorithm, we categorized admissions by principal discharge diagnosis into: HF, cardiovascular/non-HF, and noncardiovascular. We compared the mean admission rates by admission type as well as the proportion of admission types across RSAAR quartiles (Q1-Q4). RESULTS: Among 220 ACOs caring for 227,356 patients with HF, the median (IQR) RSAARs per 100 person-years ranged from 64.5 (61.7-67.7) in Q1 (best performers) to 94.0 (90.1-99.9) in Q4 (worst performers). The mean admission rates by admission types for ACOs in Q1 compared with Q4 were as follows: HF admissions: 9.8 (2.2) vs 14.6 (2.8) per 100 person years (P < .0001); cardiovascular/non-HF admissions: 11.1 (1.6) vs 15.9 (2.6) per 100 person-years (P < .0001); and noncardiovascular admissions: 42.7 (5.4) vs 69.6 (11.3) per 100 person-years (P < .0001). The proportion of admission due to HF, cardiovascular/non-HF, and noncardiovascular conditions was 15.4%, 17.5%, and 67.1% in Q1 compared with 14.6%, 15.9%, and 69.4% in Q4 (P < .007). CONCLUSIONS: Although ACOs with the best performance on a measure of all-cause admission rates among people with HF tended to have fewer admissions for HF, cardiovascular/non-HF, and noncardiovascular conditions compared with ACOs with the worst performance (highest admission rates), the largest difference in admission rates were for noncardiovascular admission types. Across all ACOs, two-thirds of admissions of patients with HF were for noncardiovascular causes. These findings suggest that comprehensive approaches are needed to reduce the diverse admission types for which HF patients are at risk.

Propensity scores with misclassified treatment assignment: a likelihood-based adjustment.

Propensity score methods are widely used in comparative effectiveness research using claims data. In this context, the inaccuracy of procedural or billing codes in claims data frequently misclassifies patients into treatment groups, that is, the treatment assignment ($T$) is often measured with error. In the context of a validation data where treatment assignment is accurate, we show that misclassification of treatment assignment can impact three distinct stages of a propensity score analysis: (i) propensity score estimation; (ii) propensity score implementation; and (iii) outcome analysis conducted conditional on the estimated propensity score and its implementation. We examine how the error in $T$ impacts each stage in the context of three common propensity score implementations: subclassification, matching, and inverse probability of treatment weighting (IPTW). Using validation data, we propose a two-step likelihood-based approach which fully adjusts for treatment misclassification bias under subclassification. This approach relies on two common measurement error-assumptions; non-differential measurement error and transportability of the measurement error model. We use simulation studies to assess the performance of the adjustment under subclassification, and also investigate the method's performance under matching or IPTW. We apply the methods to Medicare Part A hospital claims data to estimate the effect of resection versus biopsy on 1-year mortality among $10\,284$ Medicare beneficiaries diagnosed with brain tumors. The ICD9 billing codes from Medicare Part A inaccurately reflect surgical treatment, but SEER-Medicare validation data are available with more accurate information.

Octogenarians' post-acute care use after cardiac valve surgery and recovery: clinical implications.

Octogenarians receiving cardiac valve surgery is increasing and recovery is challenging. Post-acute care (PAC) services assist with recovery, yet services provided in facilities do not provide adequate cardiac-focused care or long-term self-management support. The purpose of the paper was to report post-acute care discharge rates in octogenarians and propose clinical implications to improve PAC services. Using a 2003 Medicare Part A database, we studied post-acute care service use in octogenarians after cardiac valve surgery. We propose expansion of the Geriatric Cardiac Care model to include broader clinical therapy dynamics. The sample (n = 10,062) included patients over 80 years discharged from acute care following valve surgery. Post-acute care services were used by 68% of octagarians following cardiac valve surgery (1% intermediate rehabilitation, 35% skilled nursing facility, 32% home health). The large percentage of octagarians using PAC point to the importance of integrating geriatric cardiac care into post-acute services to optimize recovery outcomes.

Colorectal Cancer Identification Methods Among Kansas Medicare Beneficiaries, 2008-2010.

INTRODUCTION: Population-based data are limited on how often colorectal cancer (CRC) is identified through screening or surveillance in asymptomatic patients versus diagnostic workup for symptoms. We developed a process for assessing CRC identification methods among Medicare-linked CRC cases from a population-based cancer registry to assess identification methods (screening/surveillance or diagnostic) among Kansas Medicare beneficiaries. METHODS: New CRC cases diagnosed from 2008 through 2010 were identified from the Kansas Cancer Registry and matched to Medicare enrollment and claims files. CRC cases were classified as diagnostic-identified versus screening/surveillance-identified using a claims-based algorithm for determining CRC test indication. Factors associated with screening/surveillance-identified CRC were analyzed using logistic regression. RESULTS: Nineteen percent of CRC cases among Kansas Medicare beneficiaries were screening/surveillance-identified while 81% were diagnostic-identified. Younger age at diagnosis (65 to 74 years) was the only factor associated with having screening/surveillance-identified CRC in multivariable analysis. No association between rural/urban residence and identification method was noted. CONCLUSION: Combining administrative claims data with population-based registry records can offer novel insights into patterns of CRC test use and identification methods among people diagnosed with CRC. These techniques could also be extended to other screen-detectable cancers.

Comparative Risk of Anaphylactic Reactions Associated With Intravenous Iron Products.

IMPORTANCE: All intravenous (IV) iron products are associated with anaphylaxis, but the comparative safety of each product has not been well established. OBJECTIVE: To compare the risk of anaphylaxis among marketed IV iron products. DESIGN, SETTING, AND PARTICIPANTS: Retrospective new user cohort study of IV iron recipients (n = 688,183) enrolled in the US fee-for-service Medicare program from January 2003 to December 2013. Analyses involving ferumoxytol were limited to the period January 2010 to December 2013. EXPOSURES: Administrations of IV iron dextran, gluconate, sucrose, or ferumoxytol as reported in outpatient Medicare claims data. MAIN OUTCOMES AND MEASURES: Anaphylaxis was identified using a prespecified and validated algorithm defined with standard diagnosis and procedure codes and applied to both inpatient and outpatient Medicare claims. The absolute and relative risks of anaphylaxis were estimated, adjusting for imbalances among treatment groups. RESULTS: A total of 274 anaphylaxis cases were identified at first exposure, with an additional 170 incident anaphylaxis cases identified during subsequent IV iron administrations. The risk for anaphylaxis at first exposure was 68 per 100,000 persons for iron dextran (95% CI, 57.8-78.7 per 100,000) and 24 per 100,000 persons for all nondextran IV iron products combined (iron sucrose, gluconate, and ferumoxytol) (95% CI, 20.0-29.5 per 100,000) , with an adjusted odds ratio (OR) of 2.6 (95% CI, 2.0-3.3; P < .001). At first exposure, when compared with iron sucrose, the adjusted OR of anaphylaxis for iron dextran was 3.6 (95% CI, 2.4-5.4); for iron gluconate, 2.0 (95% CI 1.2, 3.5); and for ferumoxytol, 2.2 (95% CI, 1.1-4.3). The estimated cumulative anaphylaxis risk following total iron repletion of 1000 mg administered within a 12-week period was highest with iron dextran (82 per 100,000 persons, 95% CI, 70.5- 93.1) and lowest with iron sucrose (21 per 100,000 persons, 95% CI, 15.3- 26.4). CONCLUSIONS AND RELEVANCE: Among patients in the US Medicare nondialysis population with first exposure to IV iron, the risk of anaphylaxis was highest for iron dextran and lowest for iron sucrose.

Total joint arthroplasty: trends in medicare reimbursement and implant prices.

Total joint arthroplasty (TJA) continues to be a popular target of cost control efforts. In order to provide a unique overview of financial trends facing TJA, we analyzed Medicare databases including 100% of beneficiaries, as well as industry surveys of implant list prices. Although there was a substantial increase in TJA utilization over the period 2000-2011 (+26.9%), growth has been stagnant since 2005. New coding schemes have made complicated cases more lucrative for hospitals (+2.5% to 6.5% per year), while reimbursements for uncomplicated cases have fallen (-0.7% to -0.6%). Physician reimbursements have declined on all case types (-2.5% to -2.1% per year), while list prices of orthopedic implants have risen (+4.8% to 5.5%). These trends should be kept in mind while contemplating future changes to TJA payment.

Continuity of care and its effect on prescription drug use among Medicare beneficiaries with hypertension.

BACKGROUND: Continuity of care is one of the pillars of primary care; yet, it is unclear whether having a regular physician influences the use of prescription drugs among the elderly. OBJECTIVE: To assess the effect of continuity of care on prescription drug use for Medicare beneficiaries with hypertension. METHODS: Medicare beneficiaries age 67 years and older were categorized in 3 groups according to the level of continuity of care experienced during the year preceding an initial diagnosis of hypertension. The group with the lowest level (poor continuity) was selected as control, and matched on previous use of health services, sociodemographic and clinical characteristics, using propensity scores, to the intermediate and high continuity groups. Fixed effect models examined the association of continuity of care to number of drugs classes, cost of drugs, use of thiazide diuretics, and adherence to treatment. RESULTS: There was no clear evidence that Medicare beneficiaries with greater continuity of care had increased adherence to antihypertensive drugs for most variables studied. The only major difference was that total number of classes of drugs purchased was 10% higher for the intermediate group, and 15% higher for the high Continuity of Care index group compared with the control group (P<0.05). CONCLUSION: A prior history of continuity of care may lead to purchasing more drugs overall but may not increase adherence to hypertension treatment among the elderly.

Multiple measurement of serum lipids in the elderly.

BACKGROUND: Although there is considerable interest in underutilization of lipid testing, little is known about the prevalence and factors associated with overtesting of serum lipids. METHODS: We assessed the number of different days in which outpatient lipid testing was performed in a 5% national sample of patients with parts A and B Medicare in 2006. Covariates included patient characteristics (age, race, prior diagnosis of lipid disorder, and other indications for lipid testing), number of usual care physicians (UCP), type of UCP, total outpatient physician encounters, and health referral region (HRR) characteristics (average per-patient Medicare expenditures and percent of patients seeing multiple UCPs). RESULTS: Among the 1,151,891 patients, 11.9% underwent 3 or more outpatient measurements of serum lipids. In multivariable analyses, the total number of UCPs providing care for the patient was associated with multiple lipid testing, independent of patient characteristics, indications for lipid testing, and total outpatient encounters. There was a strong association among HRRs between the rate of multiple lipid testing and average Medicare expenditures (r = 0.56). This was reduced after including the percentage of patients with more than 2 medical subspecialist UCPs in the HRR in a partial correlation (r = 0.31). CONCLUSIONS: Multiple lipid testing is associated with the presence of multiple providers, independent of indications for testing, comorbidity, and total physician visits. Much of the association of multiple lipid testing with medical expenditures at the level of HRR appears to be explained by differences in exposure to multiple providers.

Clinical characteristics and outcomes of Medicare patients undergoing total hip arthroplasty, 1991-2008.

CONTEXT: Total hip arthroplasty is a common surgical procedure but little is known about longitudinal trends. OBJECTIVE: To examine demographics and outcomes of patients undergoing primary and revision total hip arthroplasty. DESIGN, SETTING, AND PARTICIPANTS: Observational cohort of 1,453,493 Medicare Part A beneficiaries who underwent primary total hip arthroplasty and 348,596 who underwent revision total hip arthroplasty. Participants were identified using International Classification of Diseases, Ninth Revision, Clinical Modification codes for primary and revision total hip arthroplasty between 1991 and 2008. MAIN OUTCOME MEASURES: Changes in patient demographics and comorbidity, hospital length of stay (LOS), mortality, discharge disposition, and all-cause readmission rates. RESULTS: Between 1991 and 2008, the mean age for patients undergoing primary total hip arthroplasty increased from 74.1 to 75.1 years and for revision total hip arthroplasty from 75.8 to 77.3 years (P < .001). The mean number of comorbid illnesses per patient increased from 1.0 to 2.0 for primary total hip arthroplasty and 1.1 to 2.3 for revision (P < .001). For primary total hip arthroplasty, mean hospital LOS decreased from 9.1 days in 1991-1992 to 3.7 days in 2007-2008 (P = .002); unadjusted in-hospital and 30-day mortality decreased from 0.5% to 0.2% and from 0.7% to 0.4%, respectively (P < .001). The proportion of primary total hip arthroplasty patients discharged home declined from 68.0% to 48.2%; the proportion discharged to skilled care increased from 17.8% to 34.3%; and 30-day all-cause readmission increased from 5.9% to 8.5% (P < .001). For revision total hip arthroplasty, similar trends were observed in hospital LOS, in-hospital mortality, discharge disposition, and hospital readmission rates. CONCLUSION: Among Medicare beneficiaries who underwent primary and revision hip arthroplasty between 1991 and 2008, there was a decrease in hospital LOS but an increase in the rates of discharge to postacute care and readmission.

Post-acute care services received by older adults following a cardiac event: a population-based analysis.

BACKGROUND: Post-acute care (PAC) is available for older adults who need additional services after hospitalization for acute cardiac events. With the aging population and an increase in the prevalence of cardiac disease, it is important to determine current PAC use for cardiac patients to assist health care workers to meet the needs of older cardiac patients. The purpose of this study was to determine the current PAC use and factors associated with PAC use for older adults following hospitalization for a cardiac event that includes coronary artery bypass graft and valve surgeries, myocardial infarction (MI), percutaneous coronary intervention (PCI), and heart failure (HF). METHODS AND RESULTS: A cross-sectional design and the 2003 Medicare part A database were used for this study. The sample (n = 1493521) consisted of patients 65 years and older discharged after their first cardiac event. Multinomial logistic regression was used to examine factors associated with PAC use. Overall, PAC use was 55% for cardiac valve surgery, 50% for MI, 45% for HF, 44% for coronary artery bypass graft, and 5% for PCI. Medical patients use more skilled nursing facility care, and surgical patients use more home health care. Only 0.1% to 3.4% of the cardiac patients use intermediate rehabilitation facilities. Compared with those who do not use PAC, those who use home health care and skilled nursing facility care are older and female, have a longer hospital length of stay, and have more comorbidity. Asians, Hispanics, and Native Americans were less likely to use PAC after hospitalization for an MI or HF. CONCLUSIONS: The current rate of PAC use indicates that almost half of nondisabled Medicare patients discharged from the hospital following a cardiac event use one of these services. Health care professionals can increase PAC use for Asians, Hispanics, and Native Americans by including culturally targeted communication. Optimizing recovery for cardiac patients who use PAC may require focused cardiac rehabilitation strategies.

Measuring coverage for seniors in Medicare Part A and estimating the cost of making it universal.

That Medicare is universal for seniors is widely accepted by leading analysts. But in the context of developing detailed policies that seek to cover as many people as possible, it is inaccurate to make Medicare eligibility sound so simple and inclusive. To estimate the number of seniors without full federal Medicare Part A coverage, we examined data for uninsured seniors, seniors with Medicaid and no Medicare coverage of any kind, seniors with Medicare Part B but without Part A, and seniors bought into Part A by their state Medicaid programs. We found that in 2005, 1.6 million seniors--or 5 percent of the elderly U.S. population--were without a full federal Part A premium subsidy. The share of seniors without this benefit was notably higher in the nation's two largest states--California (12 percent) and New York (8 percent). We estimate that reforming Medicare Part A to make the benefit truly universal and fully federal would cost the federal government $6 billion in new spending in federal fiscal year 2011, an increase in baseline federal Medicare expenditures of 1.1 percent.

An examination of Alzheimer's disease case definitions using Medicare claims and survey data.

BACKGROUND: The prevalence and expenditure estimates of Alzheimer's disease (AD) from studies using one data source to define cases vary widely. The objectives of this study were to assess agreement between AD case definitions classified with Medicare claims and survey data and to provide insight into causes of widely varied expenditure estimates. METHODS: Data were obtained from the 1999-2004 Medicare Current Beneficiary Survey linked with Medicare claims (n = 57,669). Individuals with AD were identified by survey, diagnosis, use of an AD prescription medicine, or some combination thereof. We also explored how much health care and drug expenditures vary by AD case definition. RESULTS: The prevalence of AD differed significantly by case definition. Using survey report alone yielded more cases (n = 1,994 or 3.46%) than diagnosis codes alone (n = 1,589 or 2.76%) or Alzheimer's medication use alone (n = 1,160 or 2.01%). Agreement between case definitions was low, with kappa coefficients ranging from 0.37 to 0.40. Per capita health expenditures ranged from $16,547 to $24,937, and drug expenditures ranged from $2,303 to $3,519, depending on how AD was defined. CONCLUSIONS: Different information sources yield widely varied prevalence and expenditure estimates. Although claims data provided a more objective means for identifying AD cases, survey report identified more cases, and pharmacy data also are an important source for case ascertainment. Using any single source will underestimate the prevalence and associated cost of AD. The wide range of AD cases identified by using different data sources demands caution interpreting cost-of-illness studies using single data sources.

The impact of physician-owned specialty orthopaedic hospitals on surgical volume and case complexity in competing hospitals.

Published studies of physician-owned specialty hospitals have typically examined the impact of these hospitals on disparities, quality, and utilization at a national level. Our objective was to examine the impact of newly opened physician-owned specialty orthopaedic hospitals on individual competing general hospitals. We used Medicare Part A administrative data to identify all physician-owned specialty orthopaedic hospitals performing total hip arthroplasty (THA) and total knee arthroplasty (TKA) between 1991 and 2005. We identified newly opened specialty hospitals in three representative markets (Durham, NC, Kansas City, and Oklahoma City) and assessed their impact on surgical volume and patient case complexity for the five competing general hospitals located closest to each specialty hospital. The average general hospital maintained THA and TKA volume following the opening of the specialty hospitals. The average general hospital also did not experience an increase in patient case complexity. Thus, based on these three markets, we found no clear evidence that entry of physician-owned specialty orthopaedic hospitals resulted in declines in THA or TKA volume or increases in patient case complexity for the average competing general hospital.

Higher Surgeon Procedure Volume Is Associated with Improved Hemodialysis Vascular Access Outcomes.

The objective of this study was to examine the association between surgeon characteristics, procedural volume, and short-term outcomes of hemodialysis vascular access. A retrospective cohort study was performed using Medicare Part A and B data from 2007 through 2014 merged with American Medical Association Physician Masterfile surgeon data. A total of 29,034 procedures met the inclusion criteria: 22,541 (78%) arteriovenous fistula (AVF) and 6,493 (22%) arteriovenous graft (AVG). Of these, 13,110 (45.2%) were performed by vascular surgeons, 9,398 (32.3%) by general surgeons, 2,313 (8%) by thoracic surgeons, 1,517 (5.2%) by other specialties, and 2,696 (9.3%) were unknown. Every 10-year increase in years in practice was associated with a 6.9 per cent decrease in the odds of creating AVF versus AVG (P = 0.02). Surgeon characteristics were not associated with the likelihood of vascular access failure. Every 10-procedure increase in cumulative procedure volume was associated with a 5 per cent decrease in the odds of vascular access failure (P = 0.007). There was no association of provider characteristics or procedure volume with survival free of repeat AVF/AVG or TC placement at 12 months. A significant portion of the variability in likelihood of creating AVF versus AVG is attributable to the provider-level variation. Increase in procedure volume is associated with decreased odds of vascular access failure.

Medication adherence as a measure of the quality of care provided by physicians.

OBJECTIVES: To assess the extent to which medication adherence in congestive heart failure (CHF) and diabetes may serve as a measure of physician-level quality. STUDY DESIGN: A retrospective analysis of Medicare data from 2007 to 2009, including parts A (inpatient), B (outpatient), and D (pharmacy). METHODS: For each disease, we assessed the correlation between medication adherence and health outcomes at the physician level. We controlled for selection bias by first regressing patient-level outcomes on a set of covariates including comorbid conditions, demographic attributes, and physician fixed effects. We then classified physicians into 3 levels of average patient medication adherence-low, medium, and high-and compared health outcomes across these groups. RESULTS: There is a clear relationship between average medication adherence and patient health outcomes as measured at the physician level. Within the diabetes sample, among physicians with high average adherence and controlling for patient characteristics, 26.3 per 1000 patients had uncontrolled diabetes compared with 45.9 per 1000 patients among physicians with low average adherence. Within the CHF sample, also controlling for patient characteristics, the average rate of CHF emergency care usage among patients seen by physicians with low average adherence was 16.3% compared with 13.5% for doctors with high average adherence. CONCLUSIONS: This study's results establish a physician-level correlation between improved medication adherence and improved health outcomes in the Medicare population. Our findings suggest that medication adherence could be a useful measure of physician quality, at least for chronic conditions for which prescription medications are an important component of treatment.

Empirical-Based Typology of Health Care Utilization by Medicare Eligible Veterans.

OBJECTIVE: Up to 70 percent of patients who receive care through Veterans Health Administration (VHA) facilities also receive care from non-VA providers. Using applied classification techniques, this study sought to improve understanding of how elderly VA patients use VA services and complementary use of non-VA care. METHODS: The study included 1,721,900 veterans age 65 and older who were enrolled in VA and Medicare during 2013 with at least one VA encounter during 2013. Outpatient and inpatient encounters and medications received in VA were classified, and mutually exclusive patient subsets distinguished by patterns of VA service use were derived empirically using latent class analysis (LCA). Patient characteristics and complementary use of non-VA care were compared by patient subset. RESULTS: Five patterns of VA service use were identified that were distinguished by quantity of VA medical and specialty services, medication complexity, and mental health services. Low VA Medical users tend to be healthier and rely on non-VA services, while High VA users have multiple high cost illnesses and concentrate their care in the VA. CONCLUSIONS: VA patients distinguished by patterns of VA service use differ in illness burden and the use of non-VA services. This information may be useful for framing efforts to optimize access to care and care coordination for elderly VA patients.

The Role of Medicare's Inpatient Cost-Sharing in Medicaid Entry.

OBJECTIVE: To isolate the effect of greater inpatient cost-sharing on Medicaid entry among Medicare beneficiaries. DATA SOURCES: Medicare administrative data (years 2007-2010) were linked to nursing home assessments and area-level socioeconomic indicators. STUDY DESIGN: Medicare beneficiaries who are readmitted to a hospital must pay an additional deductible ($1,100 in 2010) if their readmission occurs more than 59 days following discharge. In a regression discontinuity analysis, we take advantage of this Medicare benefit feature to test whether beneficiaries with greater cost-sharing have higher rates of Medicaid enrollment. DATA EXTRACTION METHODS: We identified 221,248 Medicare beneficiaries with an initial hospital stay and a readmission 53-59 days later (no deductible) or 60-66 days later (charged a deductible). PRINCIPAL FINDINGS: Among beneficiaries in low-socioeconomic areas with two hospitalizations, those readmitted 60-66 days after discharge were 21 percent more likely to join Medicaid compared with those readmitted 53-59 days following their initial hospitalization (absolute difference in adjusted risk of Medicaid entry: 3.7 percent vs. 3.1 percent, p = .01). CONCLUSIONS: Increasing Medicare cost-sharing requirements may promote Medicaid enrollment among low-income beneficiaries. Potential savings from an increased cost-sharing in the Medicare program may be offset by increased Medicaid participation.