Concentration of Potentially Preventable Spending Among High-Cost Medicare Subpopulations: An Observational Study: Annals of Internal Medicine: Vol 167, No 10

Background: Little is known about whether potentially preventable spending is concentrated among a subset of high-cost Medicare beneficiaries. Objective: To determine the proportion of total spending that is potentially preventable across distinct subpopulations of high-cost Medicare beneficiaries. Design: Beneficiaries in the highest 10% of total standardized individual spending were defined as “high-cost” patients, using a 20% sample of Medicare fee-for-service claims from 2012. The following 6 subpopulations were defined using a claims-based algorithm: nonelderly disabled, frail elderly, major complex chronic, minor complex chronic, simple chronic, and relatively healthy. Potentially preventable spending was calculated by summing costs for avoidable emergency department visits using the Billings algorithm plus inpatient and associated 30-day postacute costs for ambulatory care–sensitive conditions (ACSCs). The amount and proportion of potentially preventable spending were then compared across the high-cost subpopulations and by individual ACSCs. Setting: Medicare. Participants: 6 112 450 Medicare beneficiaries. Measurements: Proportion of spending deemed potentially preventable. Results: In 2012, 4.8% of Medicare spending was potentially preventable, of which 73.8% was incurred by high-cost patients. Despite making up only 4% of the Medicare population, high-cost frail elderly persons accounted for 43.9% of total potentially preventable spending ($6593 per person). High-cost nonelderly disabled persons accounted for 14.8% of potentially preventable spending ($3421 per person) and the major complex chronic group for 11.2% ($3327 per person). Frail elderly persons accounted for most spending related to admissions for urinary tract infections, dehydration, heart failure, and bacterial pneumonia. Limitation: Potential misclassification in the identification of preventable spending and lack of detailed clinical data in administrative claims. Conclusion: Potentially preventable spending varied across Medicare subpopulations, with the majority concentrated among frail elderly persons. Primary Funding Source: The Commonwealth Fund.

Colonoscopy and Colorectal Cancer Mortality in the Veterans Affairs Health Care System: A Case–Control Study: Annals of Internal Medicine: Vol 168, No 7

Background: Colonoscopy is widely used in the Veterans Affairs (VA) health care system for colorectal cancer (CRC) prevention, but its effect on CRC mortality is unknown. Objective: To determine whether colonoscopy is associated with decreased CRC mortality in veterans and whether its effect differs by anatomical location of CRC. Design: Case–control study. Setting: VA–Medicare administrative data. Participants: Case patients were veterans aged 52 years or older who were diagnosed with CRC between 2002 and 2008 and died of the disease by the end of 2010. Case patients were matched to 4 control patients without prior CRC on the basis of age, sex, and facility. Conditional logistic regression was performed to calculate odds ratios (ORs) for exposure to colonoscopy, with adjustment for race, Charlson Comorbidity Index score, selected chronic conditions, nonsteroidal anti-inflammatory drug use, and family history of CRC. Measurements: Exposure to colonoscopy was determined from 1997 to 6 months before CRC diagnosis in case patients and to a corresponding date in control patients. Subgroup analysis was performed for patients who had undergone screening colonoscopy. Results: A total of 4964 case patients and 19 856 control patients were identified. Case patients were significantly less likely to have undergone any colonoscopy (OR, 0.39 [95% CI, 0.35 to 0.43]). Colonoscopy was associated with reduced mortality for left-sided cancer (OR, 0.28 [CI, 0.24 to 0.32]) and right-sided cancer (OR, 0.54 [CI, 0.47 to 0.63]). The results were similar for patients who had undergone screening colonoscopy (overall OR, 0.30 [CI, 0.24 to 0.38]). Sensitivity analyses that varied the interval between CRC diagnosis and colonoscopy exposure did not affect the primary findings. Limitation: Unmeasured confounding. Conclusion: In this study using national VA–Medicare data, colonoscopy was associated with significant reductions in CRC mortality among veterans and was associated with greater benefit for left-sided cancer than right-sided cancer. Primary Funding Source: U.S. Department of Veterans Affairs.

Should This Patient Have Weight Loss Surgery?: Grand Rounds Discussion From Beth Israel Deaconess Medical Center: Annals of Internal Medicine: Vol 166, No 11

Obesity is an important public health priority in the United States. One third of U.S. adults are obese and therefore can expect higher rates of diabetes mellitus, other obesity-related comorbidities, and mortality. In 2013, the American Association of Clinical Endocrinologists, the Obesity Society, and the American Society for Metabolic and Bariatric Surgery issued a guideline that recommended weight loss (bariatric) surgery for all patients with a body mass index (BMI) of 40 kg/m2 or higher and for those with a BMI of 35 kg/m2 or greater in the presence of at least 1 obesity-related comorbidity. Among the 3 most commonly performed surgeries, the amount of excess weight reduction ranges from 49% for laparoscopic adjustable gastric banding to 76% for Roux-en-Y gastric bypass. In accredited centers, perioperative mortality averages 0.3%. In this Beyond the Guidelines, 2 experts in obesity management, a bariatric surgeon and a general internist, discuss the role of weight loss surgery versus dietary and lifestyle modification, both in general and for a specific patient who is eligible for surgery. Ethnic and age-related variability in the effects of obesity on mortality, as well as potential long-term benefits and risks of weight loss surgery for patient subgroups, are discussed.

Holding Providers Accountable for Health Care Outcomes

Federal public reporting and payment programs have increasingly emphasized the measurement of outcomes (such as readmission, health care–associated infections, and mortality). Yet, the criteria for assessing whether outcome measures are accurate and valid enough to use for public reporting, payment, and accreditation are not well-defined. An outcome measure should be used only if the outcome can be influenced substantially by providers (that is, a strong process–outcome link exists) and statistical adjustment can be made for differences in patient populations across providers so that differences in outcomes are truly attributable to differences in the care provided. Validly distinguishing differences in quality of care across providers requires precision in both the design of the outcome measure and the actual recording of all the measure's elements. Four criteria are proposed to assess outcome measures. First, strong evidence should exist that good medical care leads to improvement in the outcome within the time period for the measure. Second, the outcome should be measurable with a high degree of precision. Third, the risk-adjustment methodology should include and accurately measure the risk factors most strongly associated with the outcome. Fourth, implementation of the outcome measure must have little chance of inducing unintended adverse consequences. These criteria were applied to 10 outcome measures currently used or proposed for accountability programs. Three measures met all 4 criteria; 5, including all 4 claims-based 30-day mortality measures, failed to meet 1 or more criteria. Patient-reported outcome measures are problematic, because low response rates may cause bias. These findings raise concerns and suggest the need for a national dialogue about how to judge outcome measures currently in use or proposed for the future.

Comparison of Five Major Guidelines for Statin Use in Primary Prevention in a Contemporary General Population

Background: Five major organizations recently published guidelines for using statins to prevent atherosclerotic cardiovascular disease (ASCVD): in 2013, the American College of Cardiology/American Heart Association (ACC/AHA); in 2014, the United Kingdom's National Institute for Health and Care Excellence (NICE); and in 2016, the Canadian Cardiovascular Society (CCS), the U.S. Preventive Services Task Force (USPSTF), and the European Society of Cardiology/European Atherosclerosis Society (ESC/EAS). Objective: To compare the utility of these guidelines for primary prevention of ASCVD. Design: Observational study of actual ASCVD events during 10 years, followed by a modeling study to estimate the effectiveness of different guidelines. Setting: The Copenhagen General Population Study. Participants: 45 750 Danish persons aged 40 to 75 years who did not use statins and did not have ASCVD at baseline. Measurements: The number of participants eligible to use statins according to each guideline and the estimated number of ASCVD events that statins could have prevented. Results: The percentage of participants eligible for statins was 44% by the CCS guideline, 42% by ACC/AHA, 40% by NICE, 31% by USPSTF, and 15% by ESC/EAS. The estimated percentage of ASCVD events that could have been prevented by using statins for 10 years was 34% for CCS, 34% for ACC/AHA, 32% for NICE, 27% for USPSTF, and 13% for ESC/EAS. Limitation: This study was limited to primary prevention in white Europeans. Conclusion: Guidelines recommending that more persons use statins for primary prevention of ASCVD should prevent more events than guidelines recommending use by fewer persons. Primary Funding Source: Copenhagen University Hospital.

Raising the Bar in Attribution

Cost-Effectiveness of Publicly Funded Treatment of Opioid Use Disorder in California

Background: Only 1 in 5 of the nearly 2.4 million Americans with an opioid use disorder received treatment in 2015. Fewer than half of Californians who received treatment in 2014 received opioid agonist treatment (OAT), and regulations for admission to OAT in California are more stringent than federal regulations. Objective: To determine the cost-effectiveness of OAT for all treatment recipients compared with the observed standard of care for patients presenting with opioid use disorder to California's publicly funded treatment facilities. Design: Model-based cost-effectiveness analysis. Data Sources: Linked population-level administrative databases capturing treatment and criminal justice records for California (2006 to 2010); published literature. Target Population: Persons initially presenting for publicly funded treatment of opioid use disorder. Time Horizon: Lifetime. Perspective: Societal. Intervention: Immediate access to OAT with methadone for all treatment recipients compared with the observed standard of care (54.3% initiate opioid use disorder treatment with medically managed withdrawal). Outcome Measures: Discounted quality-adjusted life-years (QALYs) and discounted costs. Results of Base-Case Analysis: Immediate access to OAT for all treatment recipients costs less (by $78 257), with patients accumulating more QALYs (by 0.42) than with the observed standard of care. In a hypothetical scenario where all Californians starting treatment of opioid use disorder in 2014 had immediate access to OAT, total lifetime savings for this cohort could be as high as $3.8 billion. Results of Sensitivity Analysis: 99.6% of the 2000 simulations resulted in lower costs and more QALYs. Limitation: Nonrandomized delivery of OAT or medically managed withdrawal. Conclusion: The value of publicly funded treatment of opioid use disorder in California is maximized when OAT is delivered to all patients presenting for treatment, providing greater health benefits and cost savings than the observed standard of care. Primary Funding Source: National Institute on Drug Abuse.

Technology and Medicine: Reimagining Provider Visits as the New Tertiary Care

Virtual Glucose Management in the Hospital Setting

Percutaneous Ablation Versus Partial and Radical Nephrectomy for T1a Renal Cancer: A Population-Based Analysis: Annals of Internal Medicine: Vol 169, No 2

Background: Stage T1a renal cell carcinoma (RCC) (tumors <4 cm) is usually curable. Nephron-sparing partial nephrectomy (PN) has replaced radical nephrectomy (RN) as the standard of care for these tumors. Radical nephrectomy remains the first alternative treatment option, whereas percutaneous ablation (PA), a newer, nonsurgical treatment, is recommended less strongly because of the relative paucity of comparative PA data. Objective: To compare PA, PN, and RN outcomes. Design: Observational cohort analysis using inverse probability of treatment–weighted propensity scores. Setting: Population-based SEER (Surveillance, Epidemiology, and End Results) cancer registry data linked to Medicare claims. Patients: Persons aged 66 years or older who received treatment for T1a RCC between 2006 and 2011. Interventions: PA versus PN and RN. Measurements: RCC-specific and overall survival, 30- and 365-day postintervention complications. Results: 4310 patients were followed for a median of 52 months for overall survival and 42 months for RCC-specific survival. After PA versus PN, the 5-year RCC-specific survival rate was 95% (95% CI, 93% to 98%) versus 98% (CI, 96% to 99%); after PA versus RN, 96% (CI, 94% to 98%) versus 95% (CI, 93% to 96%). After PA versus PN, the 5-year overall survival rate was 77% (CI, 74% to 81%) versus 86% (CI, 84% to 88%); after PA versus RN, 74% (CI, 71% to 78%) versus 75% (CI, 73% to 77%). Cumulative rates of renal insufficiency 31 to 365 days after PA, PN, and RN were 11% (CI, 8% to 14%), 9% (CI, 8% to 10%), and 18% (CI, 17% to 20%), respectively. Rates of nonurologic complications within 30 days after PA, PN, and RN were 6% (CI, 4% to 9%), 29% (CI, 27% to 30%), and 30% (CI, 28% to 32%), respectively. Ten percent of patients in the PN group had intraoperative conversion to RN. Seven percent of patients in the PA group received additional PA within 1 year of treatment. Limitations: Analysis of observational data may have been affected by residual confounding by provider or from selection bias toward younger, healthier patients in the PN group. Findings from this older study population are probably less applicable to younger patients. Use of SEER–Medicare linked files prevented analysis of patients who received treatment after 2011, possibly reducing generalizability to the newest PA, PN, and RN techniques. Conclusion: For well-selected older adults with T1a RCC, PA may result in oncologic outcomes similar to those of RN, but with less long-term renal insufficiency and markedly fewer periprocedural complications. Compared with PN, PA may be associated with slightly shorter RCC-specific survival but fewer periprocedural complications. Primary Funding Source: Association of University Radiologists GE Radiology Research Academic Fellowship and Society of Interventional Radiology Foundation.