Association of Dual Medicare and Medicaid Eligibility With Outcomes and Spending for Cancer Surgery in High-Quality Hospitals

Key Points

Question  Does treatment at high-quality hospitals mitigate dual eligibility–associated disparities in outcomes and spending for cancer surgery?

Findings  In this cohort study of 119 757 Medicare beneficiaries, patients dually eligible for Medicare and Medicaid were significantly more likely to be discharged to a facility, and differences in postacute care persisted even after accounting for postoperative complications, contributing to variation in spending. Compared with the lowest-quality hospitals, dually eligible patients had improved rates of discharge to a facility and spending, but rates remained increased compared with Medicare beneficiaries even at the highest-quality hospitals, with both findings statistically significant.

Meaning  The findings of this study indicate that, even among the highest-quality hospitals, dually eligible patients may have poorer outcomes and higher spending.

Importance  Although dual eligibility (DE) status for Medicare and Medicaid has been used for social risk stratification in value-based payment programs, little is known about the interplay between hospital quality and disparities in outcomes and spending by social risk.

Objective  To assess whether treatment at high-quality hospitals mitigates DE-associated disparities in outcomes and spending for cancer surgery.

Design, Setting, and Participants  Retrospective cohort study from January 1, 2014, to December 31, 2018, evaluating inpatient surgery at acute care hospitals. A total of 119 757 Medicare beneficiaries aged 65 years or older who underwent colectomy, rectal resection, lung resection, or pancreatectomy were evaluated. Data were analyzed between November 1, 2020, and April 30, 2021.

Exposures  Medicare and Medicaid DE status and hospital quality.

Main Outcomes and Measures  Postoperative complications, readmission, and mortality by DE status and hospital quality.

Results  Overall, 119 757 Medicare beneficiaries underwent colectomy, rectal resection, lung resection, or pancreatectomy. The mean (SD) age was 75.3 (6.7) years, 61 617 (51.5%) were women, 7677 (6.4%) were Black, 106 099 (88.6%) were White, and 5981 (5.0%) identified as another race or ethnicity; 11.3% had DE status. Dually eligible patients were more likely to be discharged to a facility (colectomy, 15.0% [95% CI, 14.7%-15.3%] vs 23.9% [95% CI, 22.9%-24.9%]; proctectomy, 18.7% [95% CI, 18.0%-19.3%] vs 26.9% [95% CI, 24.9%-28.9%]; lung resection, 11.0% [95% CI, 10.7%-11.3%] vs 17.9% [95% CI, 16.8%-18.9%]; pancreatectomy, 23.5% [95% CI, 22.5%-24.4%] vs 30.0% [95% CI, 26.5%-33.5%]). Differences in postacute care use persisted even after accounting for postoperative complications and contributed to variation in spending. Compared with the lowest-quality hospitals, DE patients had improved rates of discharge to a facility (22.7% vs 19.3%) and spending ($22 577 vs $20 100) but rates remained increased compared with Medicare patients even at the highest-quality hospitals.

Conclusions and Relevance  The findings of this study indicate that, even among the highest-quality hospitals, DE patients had poorer outcomes and higher spending. Dually eligible patients were more likely to be discharged to a facility and therefore incurred higher postacute care costs. Although treatment at high-quality hospitals is associated with reduced differences in outcomes, DE patients remain at high risk for adverse postoperative outcomes and increased readmissions and postacute care use.

More than 12 million individuals in the US are dually eligible (DE) for and enrolled in both Medicare and Medicaid, representing almost 20% of Medicare beneficiaries.¹ These patients qualify for Medicare owing to age (65 years or older) or disability and for Medicaid owing to low income.¹ Dually eligible patients tend to have higher social support needs compared with beneficiaries enrolled in Medicare alone.^2,3 Furthermore, DE patients have been shown to have higher service use and higher spending.^4-6 Dually eligible patients are also known to have worse adherence to cancer-specific process measures across different specialties,^7-10 including lower receipt of recommended radiotherapy or chemotherapy in breast, endometrial, and colon cancer and lower likelihood of adequate lymph node harvest in gastric cancer.¹¹ Given these factors, DE patients represent a highly marginalized population, with social determinants of health contributing to adverse outcomes and increased use of health care services.¹² Although DE status has been used for social risk stratification in value-based payment programs, little is known about the interplay between hospital quality and delivery of equitable care.

While DE patients are known to have poorer outcomes, it is unclear how these outcomes are influenced by hospital quality. Treatment at poorer-quality hospitals may exacerbate disparities in outcomes associated with social determinants of health. Conversely, high-quality hospitals may be able to mitigate some of the increased risk and health care use associated with DE status. Previous studies have shown that quality improvement initiatives reduce racial and sex disparities among Medicare patients undergoing hemodialysis and racial and ethnic disparities in breastfeeding.^13,14 Among surgical patients, selective referral of medically high-risk patients to high-quality hospitals has been shown to improve disparities in outcomes.^15-18 In addition, improved surgical quality has been shown to lower costs.^19-23 The interplay between DE status and social risk has not been adequately studied. A better understanding of the effect of hospital quality on disparities in outcomes is critical to designing strategies to overcome social risk for adverse outcomes and increased use.

This study aimed to compare outcomes and spending at the highest- and lowest-quality hospitals to assess whether treatment at high-quality hospitals would attenuate disparities. We included high-risk cancer procedures, which are typically elective procedures amenable to selective referral strategies.

All claims came from the Medicare Provider Analysis and Review (MedPAR) file from 2014 to 2018 at nonfederal acute care hospitals. We identified patients undergoing elective surgery for colon, rectal, lung, and pancreatic cancer from the MedPAR file procedure codes using the International Classification of Diseases, Ninth Revision and International Statistical Classification of Diseases and Related Health Problems, Tenth Revision codes (eAppendix 1 in the Supplement). Our cohort included fee-for-service Medicare beneficiaries aged 65 to 99 years who had continuous Part A and B coverage 3 months before and 6 months after the surgical procedures specified. Participant race and ethnicity were included in the Medicare claims database used for this study, and therefore, those identifiers were used to identify participant race and ethnicity. We chose to include these data because bariatric surgery outcomes have been demonstrated to differ by race and ethnicity. Race and ethnicity comprised the following patient categories: Asian, Black, Hispanic, North American Native, White, other, and unknown. Patients were then identified as Medicare alone or DE using data from the Master Beneficiary Summary File. We excluded patients in Medicare Advantage, those who had a nonelective admission, and those with preoperative length of stay (LOS) longer than 1 day. We defined the highest-quality hospitals as those in the quintile with the lowest risk-adjusted and reliability-adjusted serious complication rates. As a secondary analysis of deidentified administrative claims data, this study was determined to be exempt from regulation and the need for patient consent by the University of Michigan institutional review board. We followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline for cohort studies, and a STROBE checklist is included as eAppendix 2 in the Supplement.

Our primary outcomes were complications, serious complications, failure to rescue, LOS, 30-day mortality, readmissions, postacute discharge disposition (home, home health, or facility), and 30-day episode payments (Medicare spending).

Any complication included deep venous thrombosis, gastrointestinal bleeding, myocardial infarction, pneumonia, postoperative hemorrhage, pulmonary failure, kidney failure, and surgical site infection (eAppendix 1 in the Supplement). Serious complications were defined as any complication with an LOS greater than the 75% percentile for each procedure.^22,24 Failure to rescue was defined as an in-hospital mortality following a complication.^25-27 Discharge to a postacute care facility included skilled nursing, inpatient rehabilitation, intermediate care, or long-term care.

Spending was derived from MedPAR, carrier, outpatient, and home health agency files. Total episode 30-day spending included the index admission, physician reimbursement, readmissions and emergency department visits, and postacute care spending. We used price standardization to adjust for differences in adjustments to Medicare payments that are unrelated to health care use.^28-30 All payments were adjusted to 2016 dollars. Episode payments were winsorized at the 1st and 99th percentiles to limit the effect of extreme outliers.

From November 1, 2020, to April 30, 2021, we evaluated patient demographic and hospital characteristic differences using unpaired, 2-tailed t tests and χ² tests. For each procedure, we used multivariable logistic regression to assess differences in risk of any complication, serious complications, failure to rescue, 30-day mortality, readmissions, and postacute discharge disposition between Medicare and DE patients. Multivariable linear regression was used to evaluate LOS and spending. Models were adjusted for age, sex, race and ethnicity, Elixhauser comorbidities, hospital volume, hospital bed size, teaching status, urban vs rural location, and year. In addition, we examined adjusted probabilities of postacute care use for patients without complications.

Risk- and reliability-adjusted estimates of serious complications rates were then used to divide hospitals into quintiles of quality for each procedure.³¹ Risk adjustment of hospital quality included patient age, sex, and Elixhauser comorbidities, as well as year. We then used multivariable regression to compare outcomes and spending at the highest- and lowest-quality hospitals, specifically focusing on the association of hospital quality with the differences in outcomes and spending between Medicare and DE patients.

Social vulnerability was measured by the Social Vulnerability Index (SVI). The SVI is a composite metric freely available at the county level from the US Centers for Disease Control and Prevention. The SVI is a composite scale derived from census tract–level information that ranges from 1 to 100. Indices are made publicly available at the census track and county level by the US Centers for Disease Control and Prevention. A score of 50 is the mean; a lower score indicated lower vulnerability and a higher score denoted higher vulnerability. The composite score is composed of 4 themes, each of which is also given a composite score. Themes include socioeconomic status, household composition and disability, minority status and language, and housing and transportation. Census tract–level SVI data were merged with the analytic cohort at the patient county level. All analyses were performed using Stata, version 16 (StataCorp LLC), and statistical tests were determined to be significant at P = .05 (2-tailed).

The study cohort included 119 757 Medicare beneficiaries who underwent 4 elective high-risk cancer procedures at the highest-quality hospitals: 53 751 colectomies, 13 705 rectal resections, 44 756 lung resections, and 7545 pancreatectomies (eAppendix 3 in the Supplement). Of these patients, the mean (SD) age was 75.3 (6.7) years, 61 617 (51.5%) were women, 7677 (6.4%) were Black, 106 099 (88.6%) were White, and 5981 (5.0%) identified as another race or ethnicity. In all, 11.3% had DE status, and the proportion of DE patients ranged from 8% (n = 582) for pancreatectomy to 13% (n = 1819) for rectal resection. Dually eligible patients were more likely to be younger, female, and Black, with more comorbidities and higher social vulnerability. Dually eligible patients were less likely to undergo colectomy (31.0% vs 27.1%), rectal resection (36.0% vs 31.8%), and lung resection (41.9% vs 35.7%) at the highest-quality hospitals.

Dually eligible patients had higher rates of any complication, serious complications, 30-day mortality, readmissions, discharge to a facility, and longer LOS compared with Medicare patients across all procedures in unadjusted analyses among all hospitals (eAppendix 4 in the Supplement). After multivariable adjustment, DE patients still had significantly higher estimated probabilities of serious complications for colectomy, rectal resection, and lung resection (Table 1). Across all 4 procedures, DE patients were more likely to have a longer LOS (colectomy, 6.8% [95% CI, 6.8%-6.9%] vs 6.4% [95% CI, 6.4%-6.5%]; proctectomy, 8.2% [95% CI, 8.1%-8.3%] vs 7.7% [95% CI, 7.6%-7.7%]; lung resection, 6.6% [95% CI, 6.5%-6.7%] vs 6.3% [95% CI, 6.2%-6.3%]; and pancreatectomy, 12.2% [95% CI, 11.9%-12.5%] vs 11.6% [95% CI, 11.5%-11.7%]) and to be discharged to a postacute care facility (colectomy, 23.9% [95% CI, 22.9-24.9] vs 15.0% [95% CI, 14.7-15.3]; proctectomy, 26.9% [95% CI, 24.9-28.9] vs 18.7% [95% CI, 18.0-19.3]; lung resection, 17.9% [95% CI, 16.8-18.9] vs 11.0% [95% CI, 10.7-11.3]; and pancreatectomy, 30.0% [95% CI, 26.5-33.5] vs 23.5% [IQR, 22.5-24.4]). In an effort to understand whether discharge disposition reflected perioperative complications or lack of social support, we examined rates of discharge to a facility among patients without complications. Even among patients without complications, DE patients were less likely to be discharged home.

Dually eligible patients had higher rates of any and serious complications for colectomy, rectal resection, and lung resection, as well as longer LOS and discharge to a facility (for all patients and for patients without complications) for all 4 procedures at both the lowest- and highest-quality hospitals (Table 2). All patients had better outcomes at the highest-quality hospitals. The disparity in outcomes between DE and Medicare patients was also significantly smaller at the highest-quality vs lowest-quality hospitals for serious complications and LOS for colectomy, rectal resection, and lung resection (Table 2; Figure 1A). For example, the absolute risk difference of any serious complication after colectomy was 4.9% at the lowest-quality hospitals, compared with 0.9% at the highest-quality hospitals. Although not all differences in DE outcomes were statistically significant, there was a similar direction of effect, with highest-quality hospitals having smaller but persistent disparities.

Among all hospitals, postacute care spending was significantly higher for DE patients for all 4 procedures (Table 3). Even among patients without complications, postacute care spending was significantly increased for DE patients across all 4 procedures.

Comparing the lowest- and highest-quality hospitals, spending was highest for DE patients at the lowest-quality hospitals across all procedures (eAppendix 5 in the Supplement). Postacute care spending was significantly higher for DE patients compared with patients with Medicare alone even at the highest-quality hospitals while undergoing colectomy, lung resection, and pancreatectomy. The difference in spending between DE and Medicare alone was similar at the lowest- and highest-quality hospitals (Figure 1B). For example, there was a difference of $2311 in total spending for colectomy at both the lowest- and highest-quality hospitals.

Among all hospitals, postacute care contributed to most variation in spending for colectomy, rectal resection, and lung resection, ranging from 35% in lung resection to 40% in colectomy (Figure 2). Index admission spending contributed the most to variation in spending for pancreatectomy (42%). The lowest-quality hospitals had similar trends in variation. However, among the highest-quality hospitals, postacute care contributed the most to variation, and index admission contributed the least across all 4 procedures.

In this cohort study of elective high-risk cancer procedures, we highlight disparities in outcomes and spending for DE vs Medicare patients. Both groups of patients generally experience better outcomes and lower spending at the highest-quality hospitals, and the disparities in outcomes and spending were mitigated to a limited degree by treatment at the highest-quality hospitals. However, disparities in outcomes and spending persisted even at the highest-quality hospitals. We found that DE patients were more likely to have serious complications, longer LOS, and to be discharged to a facility, even among those without complications. This remained true even at the highest-quality hospitals. Spending was highest for DE patients at the lowest-quality hospitals, with most variation in spending attributable to postacute care for all procedures except for pancreatectomy. In comparison, among the highest-quality hospitals with improved outcomes, postacute care contributed the most to variation across all procedures. Taken together, these findings suggest that while outcomes and spending are improved for DE patients at high-quality hospitals, disparities in outcomes for DE vs Medicare patients will not be rectified by treatment at high-quality hospitals alone.

Dual eligibility status has previously been shown to be a more powerful marker of social risk and poor outcomes than other factors, such as race and ethnicity and neighborhood factors.¹² Our results are consistent with previous research that has shown poorer outcomes in DE patients.^7-11 Higher rates of complications and nonhome discharge have important implications for DE patients undergoing cancer treatment because previous research has shown these are risk factors for delayed chemotherapy and worse survival.^32-34 However, our analysis builds on these findings by examining the role of hospital quality in mitigating these disparities. Previous research has shown that medically high-risk patients may benefit the most from treatment at high-quality hospitals.^15-18 Our findings expand this work by showing that patients with high social risk (using DE status as a surrogate marker) would also benefit from referral to high-quality hospitals. More important, our results suggest that clinical quality alone will mitigate only a portion of the disparities in outcomes and spending. Interventions specifically targeted at addressing specific elements of social risk (eg, food insecurity, poor housing conditions, and poor access to transportation) will likely be necessary to further improve the outcomes of DE patients relative to their Medicare counterparts.

Our study also highlights the intersection of the association of patient risk and hospital quality with spending. Previous research has shown that lower-quality hospitals with higher rates of complications have higher spending, and shifting medically high-risk patients to higher-quality hospitals has been shown to reduce spending.^{17,19,21,22,35} We demonstrated that shifting socially high-risk patients would also reduce spending, which is especially important among this population with high expenditures. Although DE patients make up 20% of Medicare beneficiaries, they account for more than 30% of spending and are the highest-cost population served by public health programs, at more than $300 billion in 2018.^36,37 Improving quality and reducing spending could have significant repercussions for annual Medicare spending. With most variation in spending attributable to postacute care, having a better understanding of whether more DE patients could be supported at home and ensuring discharge to high-quality facilities when needed will be an important research and policy priority.

This study must be interpreted in light of its limitations. First, MedPAR is based on administrative claims data, which is dependent on accuracy of coding. In addition, administrative data are limited in that they may not capture granular surgical complexity. Second, our study was limited to fee-for-service Medicare beneficiaries undergoing 4 elective cancer surgeries, which may not be generalizable to populations with different insurance coverage or undergoing other procedures. However, these procedures were chosen to represent a population that is amenable to selective referral owing to the elective nature of their procedures. Third, because we analyzed Medicare expenditures, health care use that is not directly reimbursed was not captured in our data. Fourth, there may be residual confounding that is not captured by administrative data. Fifth, we are unable to capture disparities in use of health care services, which may lead to limited screening and referrals.

In summary, this analysis of a large cohort of Medicare beneficiaries undergoing 4 elective cancer procedures demonstrated that DE patients have worse outcomes and higher spending, even at high-quality hospitals. Our results suggest that improvement in hospital quality may reduce but not eliminate disparities. To address disparities by social risk, these results call for a multifaceted approach, including quality improvement and investment in greater social support.

Accepted for Publication: December 7, 2021.

Published Online: February 23, 2022. doi:10.1001/jamasurg.2021.7586

Corresponding Author: Kathryn Taylor, MD, Department of Surgery, University of Michigan, 2800 Plymouth Rd, NCRC, Bldg 14, Room G100, Ann Arbor, MI 48109 (tkathryn@med.umich.edu).

Author Contributions: Dr Nathan had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Taylor, Diaz, Ibrahim, Nathan.

Acquisition, analysis, or interpretation of data: Taylor, Diaz, Nuliyalu.

Drafting of the manuscript: Taylor, Diaz.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Taylor, Diaz, Nuliyalu.

Obtained funding: Nathan.

Administrative, technical, or material support: Diaz, Nathan.

Supervision: Diaz, Nathan.

Conflict of Interest Disclosures: None reported.

Funding/Support: This study was supported in part by funding from the National Clinician Scholars Program at the University of Michigan (Drs Taylor and Diaz), grant K08HS024763 from the Agency for Healthcare Research and Quality (Dr Nathan), and grant R01AG039434 from the National Institutes of Health (Dr Nathan).

Role of the Funder/Sponsor: The funding sources had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Meeting Presentation: Some results from this study were presented at the Society of Surgical Oncology; March 18, 2021; virtual.