Calculating physician supply using a service day method and the income percentiles method: a descriptive analysis

BACKGROUND

It is important to have an accurate count of physicians and a measurable understanding of their service provision for physician resource planning. Our objective was to compare 2 methods (income percentiles [IP] and service day activities [SVD]) for calculating the supply of full-time (FT) and part-time (PT) primary care physicians (PCPs) as measures of both physician supply counts and level of provider continuity.

METHODS

Using an observational study design, we compared 2 methods of calculating the supply of PT and FT PCPs for 2011-2015. For the IP approach, the Canadian Institute for Health Information's method was applied to Alberta Health billing data. The SVD method calculated annual service days for fee-for-service PCPs. A simple descriptive analysis was conducted of the supply of PT and FT PCPs.

RESULTS

The 2 methods agreed on the FT versus PT status of 85.2% of PCPs in 2015 but disagreed on the status of 490 PCPs. A total of 239 PCPs were classified as working FT by the IP method but PT by the SVD method. Two hundred and fifty-one PCPs were classified as working PT according by the IP method but FT by the SVD method. The former group of 239 PCPs worked fewer days per week (3.22 v. 4.1) and fewer weekend days per year (8.6 v. 24.1), billed more per year ($300 327 v. $201 834) and saw more patients per day (26.8 v. 17.8) with less continuity of care (38.0% v. 72.0%) than the latter group of 251 PCPs.

INTERPRETATION

The SVD method provides a valid alternative to calculating GP supply that distinguishes groups of physicians that the standard IP methodology does not. Those groups provide very different service; policy-makers may benefit from distinguishing them.

Putting Providers At-Risk through Capitation or Shared Savings: How Strong are Incentives for Upcoding and Treatment Changes?

BACKGROUND

Alternative payment models, including Accountable Care Organizations and fully capitated models, change incentives for treatment over fee-for-service models and are widely used in a variety of settings. The level of payment may affect the assignment to a payment category, but to date the upcoding literature has been motivated largely incorporating financial penalties for upcoding rather than by a theoretical model that incorporates the downstream effects of upcoding on service provision requirements.

AIMS OF THE STUDY

In this paper, we contribute to the literature on upcoding by developing a new theoretical model that is applicable to capitated, case-rate and shared savings payment systems. This model incorporates the downstream effects of upcoding on service provision requirements rather than just the avoidance of penalties. This difference is important especially for shared-savings models with quality benchmarks.

METHODS

We test implications of our theoretical model on changes in severity determination and service use associated with changes in case-rate payments in a publicly-funded mental health care system. We model provider-assigned severity categories as a function of risk-adjusted capitated payments using conditional logit regressions and counts of service days per month using negative binomial models.

RESULTS

We find that severity determination is only weakly associated with the payment rate, with relatively small upcoding effects, but that level of use shows a greater degree of association.

DISCUSSION

These results are consistent with our theoretical predictions where the marginal utility of savings or profit is small, as would be expected from public sector agencies. Upcoding did seem to occur, but at very small levels and may have been mitigated after the county and providers had some experience with the new system. The association between the payment levels and the number of service days in a month, however, was significant in the first period, and potentially at a clinically important level. Limitations include data from a single county/multiple provider system and potential unmeasured confounding during the post-implementation period.

IMPLICATIONS FOR HEALTH CARE PROVISION AND USE

Providers in our data were not at risk for inpatient services but decreases in use of outpatient services associated with rate decreases may lead to further increases in inpatient use and therefore expenditures over time.

IMPLICATIONS FOR HEALTH POLICIES

Health program directors and policy makers need to be acutely aware of the interplay between provider payments and patient care and eventual health and mental health outcomes.

IMPLICATIONS FOR FURTHER RESEARCH

Further research could examine the implications of the theoretical model of upcoding in other payment systems, estimate the power of the tiered-risk systems, and examine their influence on clinical outcomes.

Early impact of the implementation of Medicaid episode-based payment reforms in Arkansas

OBJECTIVE

To evaluate episode-based payments for upper respiratory tract infections (URI) and perinatal care in Arkansas's Medicaid population.

STUDY SETTING

Upper respiratory infection and perinatal episodes among Medicaid-covered individuals in Arkansas and comparison states from fiscal year (FY) 2011 to 2014.

STUDY DESIGN

Cross-sectional observational analysis using a difference-in-difference design to examine outcomes associated with URI and perinatal episodes of care (EOC) from 2011 to 2014. Key dependent variables include antibiotic use, emergency department visits, physician visits, hospitalizations, readmission, and preventive screenings.

DATA COLLECTION

Claims data from the Medicaid Analytic Extract for Arkansas, Mississippi, and Missouri from 2010 to 2014 with supplemental county-level data from the Area Health Resource File (AHRF).

PRINCIPAL FINDINGS

The URI EOC reduced the probability of antibiotic use (marginal effect [ME] = -1.8, 90% CI: -2.2, -1.4), physician visits (ME = 0.6, 90% CI: -0.8, -0.4), improved the probability of strep tests for children diagnosed with pharyngitis (ME = 9.4, 90% CI: 8.5, 10.3), but also increased the probability of an emergency department (ED) visit (ME = 0.1, 90% CI: 0.1, 0.2), relative to the comparison group. For perinatal EOCs, we found a reduced probability of an ED visit during pregnancy (ME = 0.1, 90% CI: -0.2, -0.0), an increased probability of screening for HIV (ME = 6.2, 90% CI: 4.0, 8.5), chlamydia (ME = 9.5, 90% CI: 7.2, 11.8), and group B strep-test (ME = 2.6, 90% CI: 0.5, 4.6), relative to the comparison group. Predelivery and postpartum hospitalizations also increased (ME = 1.2, 90% CI: 0.4, 2.0; ME = 0.4, 90% CI: 0.0, 0.8, respectively), relative to the comparison group.

CONCLUSION

Upper respiratory infection and perinatal EOCs for Arkansas Medicaid beneficiaries produced mixed results. Aligning shared savings with quality metrics and cost-thresholds may help achieve quality targets and disincentivize over utilization within the EOC, but may also have unintended consequences.

Using machine learning to predict risk of incident opioid use disorder among fee-for-service Medicare beneficiaries: A prognostic study

OBJECTIVE

To develop and validate a machine-learning algorithm to improve prediction of incident OUD diagnosis among Medicare beneficiaries with ≥1 opioid prescriptions.

METHODS

This prognostic study included 361,527 fee-for-service Medicare beneficiaries, without cancer, filling ≥1 opioid prescriptions from 2011-2016. We randomly divided beneficiaries into training, testing, and validation samples. We measured 269 potential predictors including socio-demographics, health status, patterns of opioid use, and provider-level and regional-level factors in 3-month periods, starting from three months before initiating opioids until development of OUD, loss of follow-up or end of 2016. The primary outcome was a recorded OUD diagnosis or initiating methadone or buprenorphine for OUD as proxy of incident OUD. We applied elastic net, random forests, gradient boosting machine, and deep neural network to predict OUD in the subsequent three months. We assessed prediction performance using C-statistics and other metrics (e.g., number needed to evaluate to identify an individual with OUD [NNE]). Beneficiaries were stratified into subgroups by risk-score decile.

RESULTS

The training (n = 120,474), testing (n = 120,556), and validation (n = 120,497) samples had similar characteristics (age ≥65 years = 81.1%; female = 61.3%; white = 83.5%; with disability eligibility = 25.5%; 1.5% had incident OUD). In the validation sample, the four approaches had similar prediction performances (C-statistic ranged from 0.874 to 0.882); elastic net required the fewest predictors (n = 48). Using the elastic net algorithm, individuals in the top decile of risk (15.8% [n = 19,047] of validation cohort) had a positive predictive value of 0.96%, negative predictive value of 99.7%, and NNE of 104. Nearly 70% of individuals with incident OUD were in the top two deciles (n = 37,078), having highest incident OUD (36 to 301 per 10,000 beneficiaries). Individuals in the bottom eight deciles (n = 83,419) had minimal incident OUD (3 to 28 per 10,000).

CONCLUSIONS

Machine-learning algorithms improve risk prediction and risk stratification of incident OUD in Medicare beneficiaries.

Regional Variation in Rates of Total Knee Arthroplasty Among Medicare Beneficiaries

IMPORTANCE

Rates of total knee arthroplasty vary widely across the United States. Whether this variation is associated with differences in patient characteristics or physician practice is unknown.

OBJECTIVES

To determine regional variations in rates of total knee arthroplasty after accounting for the prevalence of knee arthritis and other potentially associated patient risk factors and to assess the correlation of these variations with measures of access to care and surgical indications.

DESIGN, SETTING, AND PARTICIPANTS

This retrospective national cohort study used Medicare data on more than 24 million deidentified beneficiaries annually from 2011 to 2015. Individuals included had fee-for-service coverage, were 65 to 89 years of age, and resided in 1 of 306 health referral regions. Data were analyzed from September 13, 2018, to August 15, 2019.

MAIN OUTCOMES AND MEASURES

Rate of primary total knee arthroplasty indexed to the national rate using observed to expected ratios. The expected numbers of arthroplasty procedures were derived from estimates based on beneficiaries' demographic and clinical characteristics. Observed to expected ratios were confounded by race/ethnicity; thus race/ethnicity-stratified analyses were conducted.

RESULTS

In 2011, there were 218 282 total knee arthroplasty procedures among 24 583 706 white Medicare beneficiaries (mean [SD] age 74.2 [6.9] years; 54.6% women). The rate of arthroplasty during the study period (5 years) was 9.3 per 1000 person-years. Adjustment for clinical characteristics reduced the spread in observed to expected ratios among regions by 29% compared with adjustment for age and sex alone. However, substantial variation remained, with observed to expected ratios that ranged from 0.61 in Newark, New Jersey, to 1.82 in Idaho Falls, Idaho. High ratios were primarily present in the upper Midwest, Great Plains, and Mountain West regions. Higher ratios were associated with regions where beneficiaries had fewer outpatient visits (Spearman correlation [r], -0.64; 95% CI, -0.70 to -0.56) and with regions having more surgeons per capita who performed knee arthroplasty (r = 0.27; 95% CI, 0.16-0.37). Higher ratios were associated with higher rates of arthroplasty procedures among beneficiaries with dementia (r = 0.36; 95% CI, 0.25-0.46), peripheral vascular disease (r = 0.52; 95% CI, 0.42-0.61), and skin ulcers (r = 0.43; 95% CI, 0.32-0.53), which are relative contraindications to arthroplasty.

CONCLUSIONS AND RELEVANCE

Substantial regional variation in rates of total knee arthroplasty remained after adjustment for patient characteristics. Coexistence of high observed to expected ratios and high rates among patients at greater surgical risk suggested overuse of knee arthroplasty in some regions.

Utilization of ACP CPT codes among high-need Medicare beneficiaries in 2017: A brief report

Importance

Medicare beneficiaries with high medical needs can benefit from Advance Care Planning (ACP). Medicare reimburses clinical providers for ACP discussions, but it is unknown whether high-need beneficiaries are receiving this service.

Objective

To compare rates of billed ACP discussions among a cohort of high-need Medicare beneficiaries with the non-high-needs Medicare population.

Design

Retrospective analysis of Medicare Fee-for-Service (FFS) claims in 2017 comparing high-need beneficiaries (seriously ill, frail, ESRD, and disabled) with non-high need beneficiaries.

Setting

Nationally representative FFS Medicare 20% sample

Participants

Medicare beneficiaries were assigned to one of the following classifications: seriously ill (65+), frail (65+), seriously ill & frail (65+); non-high need (65+); end stage renal disease (ESRD) or disabled (<65). All participants had data available for years 2016–2017.

Exposure

Receipt of a billed ACP discussion, CPT codes 99497 or 99498.

Main outcome and measure

Rates of billed ACP visits were compared between high-need patients and non-high-need patients. Rates were adjusted for the 65+ population for sex, age, race/ethnicity, Charlson comorbidity index, Medicare/Medicaid dual eligibility status, and Hospital Referral Region.

Results

Among the 65+ groups, those most likely to have a billed ACP discussion included seriously ill & frail (5.2%), seriously ill (4.2%), and frail (3.3%). Rates remained consistent after adjusting (4.5%, 4.0%, 3.1%, respectively). Each subgroup differed significantly (p < .05) from non-high need beneficiaries (2.3%) in both unadjusted and adjusted analyses. Among the <65 high need groups, the rates were 2.7% for ESRD and 1.3% for the disabled (the latter p < .05 compared with non-high needs).

Conclusions and relevance

While rates of billed ACP discussions varied among patient groups with high medical needs, overall they were relatively low, even among a cohort of patients for whom ACP may be especially relevant.

Potentially Preventable Intensive Care Unit Admissions in the United States, 2006-2015

Rationale: Increasing intensive care unit (ICU) beds and the critical care workforce are often advocated to address an aging and increasingly medically complex population. However, reducing potentially preventable ICU stays may be an alternative to ensure adequate capacity.Objectives: To determine the proportions of ICU admissions meeting two definitions of being potentially preventable using nationally representative U.S. claims databases.Methods: We analyzed claims from 2006 to 2015 from all Medicare Fee-for-Service (FFS) beneficiaries and from a large national payer offering a private insurance (PI) plan and a Medicare Advantage (MA) plan. Potentially preventable hospitalizations were identified using existing definitions for ambulatory care sensitive conditions (ACSCs) and life-limiting malignancies (LLMs).Results: We analyzed 420,369,434 person-years of insurance coverage, during which there were 99,793,416 acute inpatient hospitalizations, of which 16,646,977 (16.7%) were associated with an ICU admission. Of these, the proportions with an ACSC were 12.9%, 12.7%, and 15.8%, and with an LLM were 5.2%, 5.4%, and 6.4%, among those with PI, MA, and FFS, respectively. Over 10 years, the absolute percentages of ACSC-associated ICU stays declined (PI = -1.1%, MA -6.4%, FFS -6.4%; all P < 0.001 for all trends). Smaller changes were noted among LLM-associated ICU stays, declining in the MA cohort (-0.8%) and increasing in the FFS (+0.3%) and PI (+0.2%) populations (P < 0.001 for all trends).Conclusions: An appreciable proportion of U.S. ICU admissions may be preventable with community-based interventions. Investment in the outpatient infrastructure required to prevent these ICU admissions should be considered as a complementary, if not alternative, strategy to expanding ICU capacity to meet future demand.

Claims-based cardiovascular outcome identification for clinical research: Results from 7 large randomized cardiovascular clinical trials

BACKGROUND

Medicare insurance claims may provide an efficient means to ascertain follow-up of older participants in clinical research. We sought to determine the accuracy and completeness of claims- versus site-based follow-up with clinical event committee (+CEC) adjudication of cardiovascular outcomes.

METHODS

We performed a retrospective study using linked Medicare and Duke Database of Clinical Trials data. Medicare claims were linked to clinical data from 7 randomized cardiovascular clinical trials. Of 52,476 trial participants, linking resulted in 5,839 (of 10,497 linkage-eligible) Medicare-linked trial participants with fee-for-service A and B coverage. Death, myocardial infarction (MI), stroke, and revascularization incidences were compared using Medicare inpatient claims only, site-reported events (+CEC) only, or a combination of the 2. Randomized treatment effects were compared as a function of whether claims-based, site-based (+CEC), or a combined system was used for event detection.

RESULTS

Among the 5,839 study participants, the annual event rates were similar between claims- and site-based (+CEC) follow-up: death (overall rate 5.2% vs 5.2%; adjusted κ 0.99), MI (2.2% vs 2.3%; adjusted κ 0.96), stroke (0.7% vs 0.7%; adjusted κ 0.99), and any revascularization (7.4% vs 7.9%; adjusted κ 0.95). Of events detected by claims yet not reported by CEC, a minority were reported by sites but negatively adjudicated by CEC (39% of MIs and 18% of strokes). Differences in individual case concordance led to higher event rates when claims- and site-based (+CEC) systems were combined. Randomized treatment effects were similar among the 3 approaches for each outcome of interest.

CONCLUSIONS

Claims- versus site-based (+CEC) follow-up identified similar overall cardiovascular event rates despite meaningful differences in the events detected. Randomized treatment effects were similar using the 2 methods, suggesting claims data could be used to support clinical research leveraging routinely collected data. This approach may lead to more effective evidence generation, synthesis, and appraisal of medical products and inform the strategic approaches toward the National Evaluation System for Health Technology.

Trends in Retina Specialist Imaging Utilization from 2012 to 2016 in the United States Medicare Fee-for-Service Population

PURPOSE

To characterize temporal trends and regional variance in retinal imaging utilization in the United States Medicare fee-for-service population from 2012-2016.

DESIGN

Cross-sectional, retrospective database analysis.

METHODS

This study addresses office or operating-room based retinal imaging. Our study population included retina specialists, defined as ophthalmologists performing either intravitreal anti-vascular endothelial growth factor injections or posterior segment laser photocoagulation and no neodymium-doped yttrium aluminum garnet laser capsulotomy. We recorded fundus photography, optical coherence tomography (OCT), intravenous fluorescein angiography (IVFA), indocyanine-green angiography, and ophthalmic ultrasound (B-scan) billed in the Medicare fee-for-service population from 2012-2016. Imaging obtained on any platform or device was eligible for inclusion (eg, posterior pole imaging vs ultrawidefield imaging). The main outcome measure was the relative utilization of retinal imaging modalities.

RESULTS

National relative utilization of OCT increased from 61.5% in 2012 to 70.5% in 2016 (P < .001), while IVFA fell from 20.9% to 15.1% over the same interval (P < .001). Fundus photography decreased from 14.6% in 2012 to 11.7% in 2016 (P < .001). By 2016, the Midwest region had the highest relative utilization of OCT (75.2%) and lowest of IVFA (12.0%), while the West region had the lowest OCT (68.4%) and highest IVFA (17.0%).

CONCLUSIONS

Among retina specialists, OCT usage increased while the utilization of fundus photography and IVFA has declined. The Midwest region had the highest utilization of OCT and lowest of IVFA.

Physician practices in Accountable Care Organizations are more likely to collect and use physician performance information, yet base only a small proportion of compensation on performance data

IMPORTANCE

It is critical to develop a better understanding of the strategies provider organizations use to improve the performance of frontline clinicians and whether ACO participation is associated with differential adoption of these tools.

OBJECTIVES

Characterize the strategies that physician practices use to improve clinician performance and determine their association with ACOs and other payment reforms.

DATA SOURCES

The National Survey of Healthcare Organizations and the National Survey of ACOs fielded 2017-2018 (response rates = 47 percent and 48 percent).

STUDY DESIGN

Descriptive analysis for practices participating and not participating in ACOs among 2190 physician practice respondents. Linear regressions to examine characteristics associated with counts of performance domains for which a practice used data for feedback, quality improvement, or physician compensation as dependent variables. Logistic and fractional regression to examine characteristics associated with use of peer comparison and shares of primary care and specialist compensation accounted for by performance bonuses, respectively.

PRINCIPAL FINDINGS

ACO-affiliated practices feed back clinician-level information and use it for quality improvement and compensation on more performance domains than non-ACO-affiliated practices. Performance measures contribute little to physician compensation irrespective of ACO participation.

CONCLUSION

ACO-affiliated practices are using more performance improvement strategies than other practices, but base only a small fraction of compensation on quality or cost.

Production of physician services under fee-for-service and blended fee-for-service: Evidence from Ontario, Canada

We examine family physicians' responses to financial incentives for medical services in Ontario, Canada. We use administrative data covering 2003-2008, a period during which family physicians could choose between the traditional fee for service (FFS) and blended FFS known as the Family Health Group (FHG) model. Under FHG, FFS physicians are incentivized to provide comprehensive care and after-hours services. A two-stage estimation strategy teases out the impact of switching from FFS to FHG on service production. We account for the selection into FHG using a propensity score matching model, and then we use panel-data regression models to account for observed and unobserved heterogeneity. Our results reveal that switching from FFS to FHG increases comprehensive care, after-hours, and nonincentivized services by 3%, 15%, and 4% per annum. We also find that blended FFS physicians provide more services by working additional total days as well as the number of days during holidays and weekends. Our results are robust to a variety of specifications and alternative matching methods. We conclude that switching from FFS to blended FFS improves patients' access to after-hours care, but the incentive to nudge service production at the intensive margin is somewhat limited.

Trends in High- and Low-Value Cardiovascular Diagnostic Testing in Fee-for-Service Medicare, 2000-2016

IMPORTANCE

Owing to a rapid increase in rates of diagnostic cardiovascular testing in the 1990s and early 2000s, the Centers for Medicare & Medicaid Services implemented a series of payment changes intended to reduce overall spending on fee-for-service testing. Whether guideline-concordant testing has been subsequently affected is unknown to date.

OBJECTIVE

To determine whether changes in overall rates of use of diagnostic cardiovascular tests were associated with changes in high-value testing recommended by guidelines and low-value testing that is expected to provide minimal benefits.

DESIGN, SETTING, AND PARTICIPANTS

This retrospective cohort study assessed a national 5% random sample of Medicare fee-for-service beneficiaries aged 65 to 95 years from January 1, 1999, through December 31, 2016. Data were analyzed from February 15, 2018, through August 15, 2019.

EXPOSURES

Eligibility to receive high-value testing (assessment of left ventricular systolic function among patients hospitalized with acute myocardial infarction or heart failure) and low-value testing (stress testing before low-risk noncardiac surgery and routine stress testing within 2 years of coronary revascularization not associated with acute care visits).

MAIN OUTCOMES AND MEASURES

Age- and sex-adjusted annual rates of overall, high-value, and low-value diagnostic cardiovascular testing.

RESULTS

Mean (SD) age was similar over time (75.57 [7.32] years in 2000-2003; 74.82 [7.79] years in 2012-2016); the proportion of women slightly declined over time (63.23% in 2000 to 2003; 57.27% in 2012 to 2016). The rate of overall diagnostic cardiovascular testing per 1000 patient-years among the 5% sample of Medicare beneficiaries increased from 275 in 2000 to 359 in 2008 (P < .001) and then declined to 316 in 2016 (P < .001). High-value testing increased steadily over the entire study period for patients with acute myocardial infarction (85.7% to 89.5%; P < .001) and heart failure (72.6% to 80.1%; P < .001). Low-value testing among patients undergoing low-risk surgery increased from 2.4% in 2000 to 3.8% in 2008 (P < .001) but then declined to 2.5% in 2016 (P < .001). Low-value testing within 2 years of coronary revascularization slightly increased from 47.4% in 2000 to 49.2% in 2003 (P = .03) but then declined to 30.8% in 2014 (P < .001).

CONCLUSIONS AND RELEVANCE

Rates of overall and low-value diagnostic cardiovascular testing appear to have declined considerably and rates of high-value testing have increased slightly. Payment changes intended to reduce spending on overall testing may not have adversely affected testing recommended by guidelines.

Health System and Beneficiary Costs Associated With Intensive End-of-Life Medical Services

IMPORTANCE

Despite recommendations to reduce intensive medical treatment at the end of life, many patients with cancer continue to receive such services.

OBJECTIVE

To quantify expected beneficiary and health system costs incurred in association with receipt of intensive medical services in the last month of life.

DESIGN, SETTING, AND PARTICIPANTS

This retrospective cohort study used data collected nationally from Medicare and the Veterans Health Administration for care provided in fiscal years 2010 to 2014. Participants were 48 937 adults aged 66 years or older who died of solid tumor and were continuously enrolled in fee-for-service Medicare and the Veterans Health Administration in the 12 months prior to death. The data were analyzed from February to August 2019.

EXPOSURES

American Society of Clinical Oncology metrics regarding medically intensive services provided in the last month of life, including hospital stay, intensive care unit stay, chemotherapy, 2 or more emergency department visits, or hospice for 3 or fewer days.

MAIN OUTCOMES AND MEASURES

Costs in the last month of life associated with receipt of intensive medical services were evaluated for both beneficiaries and the health system. Costs were estimated from generalized linear models, adjusting for patient demographics and comorbidities and conditioning on geographic region.

RESULTS

Of 48 937 veterans who received care through the Veterans Health Administration and Medicare, most were white (90.8%) and male (98.9%). More than half (58.9%) received at least 1 medically intensive service in the last month of life. Patients who received no medically intensive service generated a mean (SD) health system cost of $7660 ($1793), whereas patients who received 1 or more medically intensive services generated a mean (SD) health system cost of $23 612 ($5528); thus, the additional financial consequence to the health care system for medically intensive services was $15 952 (95% CI, $15 676-$16 206; P < .001). The biggest contributor to these differences was $21 093 (95% CI, $20 364-$21 689) for intensive care unit stay, while the smallest contributor was $3460 (95% CI, $2927-$3880) for chemotherapy. Mean (SD) expected beneficiary costs for the last month of life were $133 ($50) for patients with no medically intensive service and $1257 ($408) for patients with at least 1 medically intensive service (P < .001).

CONCLUSIONS AND RELEVANCE

Given the low income of many elderly patients in the United States, the financial consequences of medically intensive services may be substantial. Costs of medically intensive services at the end of life, including patient financial consequences, should be considered by both physicians and families.

Managed care for long-stay nursing home residents: an evaluation of Institutional Special Needs Plans

OBJECTIVES

To evaluate the patterns of clinical service use for long-term nursing home residents enrolled in UnitedHealthcare's Medicare Advantage Institutional Special Needs Plans (I-SNPs), which provide on-site direct coordinated care for beneficiaries through the use of advanced practice clinicians.

STUDY DESIGN

Observational analysis of 8052 I-SNP members and 12,982 Medicare fee-for-service (FFS) long-term nursing home residents across 13 states.

METHODS

Multivariate analyses were performed to compare rates of emergency department (ED), inpatient, and skilled nursing facility (SNF) use between I-SNP members and Medicare FFS long-term nursing home residents.

RESULTS

In comparison with FFS institutionalized Medicare beneficiaries, I-SNP members had 51% lower ED use, 38% fewer hospitalizations, and 45% fewer readmissions, whereas their SNF use was 112% higher.

CONCLUSIONS

"At-risk" models, administered through specialized Medicare Advantage plans, that invest in clinical management in the nursing home setting have the potential to allow individuals to receive care on-site and avoid costly inpatient transfers.

Multi-Payer Advanced Primary Care Practice Demonstration on quality of care

OBJECTIVES

We evaluated whether primary care practices in the Medicare Multi-Payer Advanced Primary Care Practice (MAPCP) Demonstration improved the quality of care and patient outcomes for beneficiaries.

STUDY DESIGN

For our quantitative analyses, we employed a pre-post study design with a comparison group using enrollment data, Medicare fee-for-service claims data, and Medicaid managed care and fee-for-service claims data, covering the period 2 to 4 years before Medicare joined the state patient-centered medical home initiatives through December 2014. We used difference-in-differences (DID) regression analysis to compare quality and outcomes in the period before and after the demonstration began.

METHODS

We examined the extent to which MAPCP and comparison group beneficiaries received up to 11 process and preventive care measures, as well as 4 measures of potentially avoidable hospitalizations to assess patient outcomes.

RESULTS

Analyses of Medicare and Medicaid data did not consistently reflect the positive impacts intended by the demonstration. Our descriptive and DID analysis found an inconsistent pattern among the process-of-care results, and there were some significant unfavorable associations between participation in MAPCP and avoidable hospitalizations.

CONCLUSIONS

Our analyses showed few statistically significant, favorable impacts on quality metrics among Medicare or Medicaid beneficiaries receiving care from MAPCP practices.

Changes in ambulatory utilization after switching from Medicaid fee-for-service to managed care

OBJECTIVES

To observe any change in ambulatory care utilization after switching from Medicaid fee-for-service (FFS) to Medicaid managed care (MC).

STUDY DESIGN

We conducted a statewide longitudinal study of 21,048 adult Medicaid beneficiaries in New York State who switched from FFS to MC in 2011 or 2012, with 2 sets of controls (n = 21,048 with continuous FFS; n = 21,048 with continuous MC) who were matched on age, gender, dual-eligible status, and number of chronic conditions.

METHODS

We measured ambulatory care utilization in the 12 months before and 12 months after the switch date, using regression to adjust for case mix and account for matching.

RESULTS

Overall, switching from Medicaid FFS to Medicaid MC was associated with greater absolute decreases over time in ambulatory visits and providers compared with controls (-1.49 visits vs continuous FFS and -1.60 visits vs continuous MC; each P <.0001; -0.10 providers vs continuous FFS and -0.12 providers vs continuous MC; each P <.0001). The subset of switchers with 5 or more chronic conditions had the greatest absolute decreases in visits (-5.88 visits vs continuous FFS and -5.98 visits vs continuous MC; each P <.0001) and providers (-1.37 providers vs continuous FFS and -1.39 providers vs continuous MC; each P <.0001). Significant decreases in visits and providers were also observed for switchers with 3 to 4 chronic conditions but not for those with 0 to 2 chronic conditions.

CONCLUSIONS

Switching from Medicaid FFS to Medicaid MC was associated with a decrease in ambulatory utilization, especially for the sickest patients.

Medicare Shared Savings Program ACO network comprehensiveness and patient panel stability

OBJECTIVES

The current Medicare Shared Savings Program (MSSP) accountable care organization (ACO) attribution methodology creates unpredictability for ACOs that are developing and deploying strategic initiatives aimed at improving value. The goal of this study is to determine if ACO network comprehensiveness is associated with the stability of assigned Medicare beneficiaries from 2013 to 2014.

STUDY DESIGN

We utilized a beneficiary-level logistic regression model to determine association of network comprehensiveness with stable attribution to an MSSP ACO.

METHODS

Using 2013 and 2014 Medicare fee-for-service beneficiary and provider files, we developed a measure of network comprehensiveness based on 2013 provider contracts, determined beneficiary attribution, and generated market-level measures. Additional population and quality measures were obtained from the US Census and the ACO Public Use File.

RESULTS

Of the 1,317,858 observed beneficiaries, 84.38% were attributed to the same ACO in 2013 and 2014, and mean (SD) ACO network comprehensiveness was 0.30 (0.20). We found that a 0.10 increase in network comprehensiveness score significantly increased the odds of remaining attributed to the same ACO by 4.5% (P = .001). Patient panel stability was significantly associated with improved diabetes (P = .01) and hypertension (P = .02) control, timely access to care (P = .001), and delivery of health education (P = .03) over the 2-year period.

CONCLUSIONS

The comprehensiveness of an MSSP ACO's contracted provider network is associated with stable patient assignment year to year. Patient panel stability may aid in the longitudinal management of some conditions.

Variation in Practice Patterns and Reimbursements Between Female and Male Urologists for Medicare Beneficiaries

Importance

Previous assessments of practice patterns and reimbursements for female urologists relied on surveys or board certification logs. A current evaluation of the geographic distribution and practice patterns by female urologists would reveal contemporary patterns of access for Medicare beneficiaries.

Objective

To characterize the variation in practice patterns and reimbursements by urologist sex and the regional deficiencies in care provided by female urologists.

Design, Setting, and Participants

This population-based cohort study used the publicly available Centers for Medicare & Medicaid Services Provider Payment database to evaluate payments for US urologists. The cohort (n = 8665) included urologists who provided and were paid for 11 or more services to Medicare beneficiaries in 2016. Data collection and analysis were performed from October 3, 2018, through June 19, 2019.

Main Outcomes and Measures

Proportion of female-specific services, payments per beneficiary, and payments per work relative value unit (wRVU) by urologist sex were assessed. Density of female urologists across hospital markets was also identified.

Results

Among the 8665 urologists who received payments in 2016, 7944 (91.7%) were men and 721 (8.3%) were women. Female urologists, compared with male urologists, saw a lower proportion of patients with cancer (mean [SD], 16.3% [9.2%] vs 22.7% [8.8%]; P < .001) and a greater proportion of female Medicare beneficiaries (mean [SD], 52.8% [23.2%] vs 24.4% [10.3%]; P < .001). Female urologists generated a greater proportion of wRVU from urodynamics (median [IQR], 2.88% [1.26%-4.84%] vs 1.07% [0.31%-2.26%]; P < .001) and gynecological operations (median [IQR], 0.68% [0.45%-1.07%] vs 0.41% [0.20%-0.81%]; P < .001) than male urologists. In addition, female urologists, compared with their male counterparts, received lower median payments per beneficiary seen ($70.12 [interquartile range (IQR), $60.00-$84.81] vs $72.37 [IQR, $59.63-$89.29]; P = .03) and lower payments per wRVU ($58.25 [IQR, $48.39-65.26] vs $60.04 [IQR, $51.93-$67.88]; P < .001). One-third (103 [33.7%]) of 306 hospital referral regions had 0 female urologists, and 80 (26.1%) had only 1 female urologist.

Conclusions and Relevance

Female urologists were more likely to provide care for female Medicare beneficiaries, to receive lower payments per wRVU generated and beneficiaries seen, and to be difficult to access in certain geographic areas; these findings have policy-related implications and highlight the regional deficiencies in urological care and reimbursement discrepancies according to urologist sex.

Racial Disparities in Readmission Rates among Patients Discharged to Skilled Nursing Facilities

OBJECTIVES

Prior studies have reported mixed findings about the existence of racial disparities in readmission rates among Medicare Advantage beneficiaries, but these studies used data from one state, focused on black-white disparities, and did not focus on patients discharged to skilled nursing facilities (SNFs). The objective of the study was to characterize racial and ethnic disparities in rates of 30-day rehospitalization directly from SNFs among fee-for-service and Medicare Advantage patients.

DESIGN

A cross-sectional study of admissions to SNFs in 2015 was conducted.

SETTING

SNFs across the United States.

PARTICIPANTS

The sample included 1 500 334 white, 213 848 African American, and 99 781 Hispanic Medicare patients who were admitted to 13 375 SNFs.

MEASUREMENTS

The main outcome of interest was readmission, identified as patients sent back to any hospital directly from the SNF within 30 days of admission, as indicated on the Minimum Data Set discharge assessment.

RESULTS

Overall readmission rates for fee-for-service patients were 16.7% (95% confidence interval [CI] = 16.7%-16.8%) for whites, 18.8% (95% CI = 18.7%-19.0%) for African Americans, and 17.4% (95% CI = 17.1%-17.7%) for Hispanics. Readmission rates in Medicare Advantage were 14.7% (95% CI = 14.5%-14.8%) for whites, 16.8% (95% CI = 16.6%-17.1%) for African Americans, and 15.3% (95% CI = 14.9%-15.6%) for Hispanics. We also found that African Americans had about 1% higher readmission rates than whites, even when they received care within the same SNF. No statistically significant differences were found in the magnitude of within-SNF racial disparities in Medicare Advantage compared with Medicare fee-for-service.

CONCLUSION

We found racial disparities in readmission rates even within the same facility for both Medicare Advantage and fee-for-service beneficiaries. Intervention to reduce disparities in readmission rates, as well as more comprehensive quality measures that incorporate outcomes for Medicare Advantage enrollees, are needed. J Am Geriatr Soc 67:1672-1679, 2019.

Contemporary Trends in Clinical Outcomes among Dialysis Patients with Medicare Coverage

BACKGROUND

The dialysis patient population in the United States continues to grow. Trends in rates of death and hospitalization among dialysis patients have important consequences for outpatient dialysis capacity and Medicare spending.

OBJECTIVES

To estimate contemporary trends in rates of death and hospitalization among dialysis patients in the United States, overall and within subgroups.

METHODS

We used Medicare Limited Data Sets (100% sample) in 2014-2017 to estimate trends in rates of death and hospitalization among dialysis patients with Medicare Parts A and B enrollment. We used seasonal autoregressive integrated moving average models to identify secular trends in the incidence of outcomes.

RESULTS

There were 631,075 unique patients; 222,924 deaths; and 1,876,779 hospital admissions. Weekly risks of both death and hospitalization exhibited strong seasonality. However, overall weekly risks of death were 34.9, 35.4, 35.2, and 35.7 deaths per 10,000 patients in 2014-2017, respectively (p = 0.47, from a likelihood ratio test of secular trend). The overall weekly risk of hospitalization was 3.08, 3.05, 3.11, and 3.11% in 2014, 2015, 2016, and 2017, respectively (p = 0.30). There were significant secular trends in risk of death in subgroups defined by black race and residency in South Atlantic states (p < 0.05). There were also secular trends in risk of hospitalization in subgroups defined by age 20-44 years, concurrent enrollment in Medicaid, and residency in South Central states.

CONCLUSION

For the first time since the beginning of this century, rates of both death and hospitalization among dialysis patients with Medicare fee-for-service coverage have stagnated. The reasons for this change are unknown and require detailed assessment. Persistent lack of change in clinical outcomes may alter the future expectations about dialysis patient population growth.

Cancer care spending and use by site of provider-administered chemotherapy in Medicare

OBJECTIVES

To compare cancer care spending and utilization by site of provider-administered chemotherapy in Medicare.

STUDY DESIGN

A retrospective analysis using 2010-2013 Medicare claims.

METHODS

The study population was a random sample of Medicare fee-for-service beneficiaries with cancer who initiated provider-administered chemotherapy in a hospital outpatient department (HOPD) or physician office (PO). We assessed the following outcomes during the 6-month follow-up period: (1) spending on cancer-related outpatient services excluding chemotherapy, (2) spending on cancer-related inpatient services, (3) utilization of select cancer-related outpatient services (evaluation and management, commonly used expensive billing codes, and radiation therapy sessions), and (4) the number of cancer-related hospitalizations. We used regression analyses to adjust for patient health risk factors and market characteristics.

RESULTS

During the 6-month follow-up period, risk-adjusted spending on nonchemotherapy outpatient services was slightly lower among patients receiving chemotherapy in HOPDs than in POs ($12,183 [95% CI, $12,008-$12,358] vs $12,444 [95% CI, $12,313-$12,575]; P <.05). Risk-adjusted cancer-related inpatient spending was higher in the HOPD group than in the PO group ($3996 [95% CI, $3837-$4156] vs $3168 [95% CI, $3067-$3268]; P <.01). The HOPD group had fewer visits in all select outpatient services but had a higher number of hospitalizations than the PO group.

CONCLUSIONS

Differences in cancer care spending by site of chemotherapy (HOPDs vs POs) vary by service type. Those differences are partially driven by utilization differences. As the site of chemotherapy shifts from POs to HOPDs, spending and utilization patterns in both settings need to be monitored.

Association of decision support for hospital discharge disposition with outcomes

OBJECTIVES

To assess the association of a clinical decision support (CDS) algorithm for hospital discharge disposition with spending, readmissions, and postdischarge emergency department (ED) use.

STUDY DESIGN

A retrospective study in a cohort of fee-for-service Medicare patients 65 years or older linked to a database of patients receiving CDS.

METHODS

We evaluated (1) patients whose discharge disposition was concordant with the CDS recommendation versus those whose disposition was not and (2) patients receiving CDS for discharge disposition versus those not receiving CDS, regardless of concordance. Outcomes were spending over a 90-day episode, 90-day readmissions, and postdischarge ED utilization not associated with a readmission.

RESULTS

Analysis of concordant versus discordant cases showed decreased spending for concordant cases ($860 savings; 95% CI, $162-$1558; P = .016), a decrease in readmissions (adjusted odds ratio [OR], 0.920; 95% CI, 0.850-0.995; P = .038), and no change in rate of postdischarge ED use (adjusted OR, 0.990; 95% CI, 0.882-1.110; P = .858). Analysis of patients receiving CDS versus not receiving CDS showed no significant difference in spending ($221 savings; 95% CI, -$115 to $557; P = .198), ED use (adjusted OR, 0.959; 95% CI, 0.908-1.012; P = .128), or readmission rate (adjusted OR, 1.004; 95% CI, 0.966-1.043; P = .840).

CONCLUSIONS

Following the recommendation of a CDS algorithm for hospital discharge disposition was associated with lower spending, fewer readmissions, and no change in ED use over a 90-day episode of care.

Health Care Fragmentation in Medicaid Managed Care vs. Fee for Service

Managed care plans often attempt to control health care costs through strategies designed to decrease health care utilization. However, the extent to which the resulting patterns of utilization represent high-quality care (compared to fee-for-service products) remains controversial. The authors sought to compare patterns of ambulatory care (including how diffuse or fragmented the care patterns were) for Medicaid fee-for-service beneficiaries vs. Medicaid managed care beneficiaries. A serial cross-sectional study of adults (≥18 years old) was conducted using statewide Medicaid claims from New York State for calendar years 2010-2013. Beneficiaries were required to be continuously enrolled and have ≥4 ambulatory visits for each year they contributed data, yielding a sample of more than 1 million beneficiaries per year. Beneficiaries were characterized by age, sex, and case mix. For each year, ambulatory care patterns were compared across subgroups of beneficiaries using Poisson models (for numbers of visits and providers) and bounded Tobit models (for fragmentation scores). In 2010, among those who were not dual eligible, managed care beneficiaries had on average fewer visits (10.9 visits vs. 11.4 visits [P < 0.0001]) but more providers (3.8 providers vs. 3.3 providers [P < 0.0001]) and therefore more fragmentation (0.58 vs. 0.51 [P < 0.0001]) than fee-for-service beneficiaries, adjusting for age, sex, and case mix. These patterns persisted throughout the follow-up period and in sensitivity analyses. Less utilization is not necessarily more efficient care; a smaller number of visits spread across a larger number of providers creates more challenges for care coordination.

Testing the myth of fee-for-service and overprovision in health care

Paying on the basis of fee-for-service (FFS) is often associated with a risk of overprovision. Policymakers are therefore increasingly looking to other payment schemes to ensure a more efficient delivery of health care. This study tests whether context plays a role for overprovision under FFS. Using a laboratory experiment involving medical students, we test the extent of overprovision under FFS when the subjects face different fee sizes, patient types, and market conditions. We observe that decreasing the fee size has an effect on overprovision under both market conditions. We also observe that patients who are harmed by excess treatment are at little risk of overprovision. Finally, when subjects face resource constraints but still have an incentive to overprovide high-profit services, they hesitate to do so, implying that the presence of opportunity costs in terms of reduced benefits to other patients protects against overprovision. Thus, this study provides evidence that the risk of overprovision under FFS depends on fee sizes, patients' health profiles, and market conditions.