Association of Medicare Enrollment With Increased Inpatient Coding Intensity

ABSTRACT

Objectives: Significant variation in coding intensity exists across patients and institutions, with important implications for reimbursement and risk-adjusted quality metrics. The degree to which coding intensity for hospitalized patients may be a function of primary payer is not well understood. We sought to measure differences in coding intensity between commercially insured and Medicare, Medicaid and Medicare, and self-pay and Medicare inpatient encounters for the same cohort of patients.

Study Design: Regression discontinuity, leveraging the fact that patients typically enroll in Medicare at age 65 years.

Methods: A multivariable linear regression was estimated to evaluate the relationship between the outcomes of interest and primary payer, controlling for age, age by payer interaction term, and inpatient visit count. Our analysis included Florida inpatients with at least 1 commercially insured, Medicaid, or self-pay inpatient hospitalization before age 65 years and at least 1 inpatient Medicare hospitalization at 65 years and older, with patients serving as their own controls. The outcome of interest was the number of hospital discharge diagnoses. Outcomes were measured separately for each group (commercial insurance to Medicare, Medicaid to Medicare, and self-pay to Medicare).

Results: Medicare inpatient encounters were associated with 0.8 (95% CI, 0.4-1.2), 1.0 (95% CI, 0.5-1.5), and 2.0 (95% CI, 1.2-2.8) more discharge diagnoses than commercially insured, Medicaid, and self-pay inpatient encounters, respectively.

Conclusions: Our findings suggest that Medicare inpatient encounters are associated with higher coding intensity than commercially insured, Medicaid, or self-pay inpatient encounters for those same individuals prior to age 65 years. This has important implications for the impact that insurance status may have on risk-adjusted quality measures.

Am J Manag Care. 2026;32(4):230-236. https://doi.org/10.37765/ajmc.2026.89919

Takeaway Points

Inpatient encounters for Medicare patients 65 years and older were coded more intensely than those for the same patients with commercially insured, Medicaid, or self-pay hospitalizations prior to age 65 years.

• Inpatient encounters with Medicare as the primary payer had higher levels of coding intensity as measured by number of discharge diagnosis codes, suggesting that insurance status may meaningfully impact coding intensity of inpatient hospitalizations.
• The finding of higher coding intensity for Medicare was also observed in the Medicare fee-for-service subanalysis but not in the Medicare Advantage subanalysis.

International Classification of Diseases diagnosis codes are used to summarize the acute and chronic medical issues addressed during a hospitalization. However, significant variation in coding intensity exists across patients and hospitals,^1-4 and this has important implications for reimbursement and risk adjustment.^4,5 Additionally, coding intensity variation may confound interpretation of disease trends, outcomes, and costs for hospitalized patients.^6,7 The extent to which coding intensity for hospitalizations relates to patients’ insurance status is poorly understood. We sought to quantify the change in coding intensity associated with Medicare enrollment by leveraging a regression discontinuity design and using patients with prior commercially insured, Medicaid, and self-pay hospitalizations to serve as their own controls.

METHODS

Data Source and Inclusion Criteria

We utilized Florida data from the Healthcare Cost and Utilization Project State Inpatient Database (SID)⁸ from 2016 through 2018. The SID contains all inpatient encounters for all payers. We chose Florida based on the cost of data acquisition and the presence of a revisit variable that allows linkage of patients across encounters. Our main analysis (model 1) included all patients with at least 2 inpatient encounters, one with commercial insurance, Medicaid, or self-pay as the primary payer (when aged 62-64 years) and another with Medicare as the primary payer (when aged 65-67 years). These 3 groups—commercial insurance to Medicare, Medicaid to Medicare, and self-pay to Medicare— were analyzed separately. In our 2 alternative models (models 2 and 3), we included all inpatient hospitalizations for patients aged 60 to 70 years with either Medicare, commercial insurance, Medicaid, or self-pay who had at least 2 inpatient hospitalizations. Data on hospital characteristics were obtained from the American Hospital Association Annual Survey.⁹

Outcome Measures

The number of discharge diagnoses for an inpatient encounter was the primary outcome (models 1-3). The secondary outcome (model 1) was the Elixhauser Comorbidity Index (ECI) score¹⁰ associated with an inpatient encounter. Additionally, we determined the 10 most frequent principal discharge diagnoses for patients aged 64 years and for those aged 65 years in each of the 3 analysis groups (commercial insurance to Medicare, Medicaid to Medicare, and self-pay to Medicare).

Statistical Design and Analysis for the Main Model (Model 1)

The main model (model 1) was a longitudinal cohort study using regression discontinuity at age 65 years, which is when most people become eligible for Medicare. The median (IQR) number of inpatient encounters per patient was measured. We characterized each of the 3 cohorts by mean (SD) number of diagnoses and mean (SD) ECI score stratified by age. We also described each cohort by sex and race.

For model 1, we estimated a multivariable linear regression with the number of discharge diagnoses as the outcome and primary payer (Medicare vs comparator insurance group) as the predictor of interest, adjusting for age (continuous variable centered at 65 years), an age by primary payer interaction term, and inpatient visit count (continuous variable based on the quarter of an inpatient hospitalization). We also estimated these regressions with ECI score as the outcome of interest.

Sensitivity and Subanalyses for the Main Model (Model 1)

For the primary outcome, we performed sensitivity analyses, varying bandwidth to include only ages 64 to 65 years and 63 to 66 years. In the former, we did not have a covariate for age because it was completely colinear with insurance status. We also performed a sensitivity analysis adjusting for hospital characteristics, including teaching status, hospital bed size, ownership, geography, and critical access status. We performed subanalyses evaluating Medicare fee-for-service (FFS) and Medicare Advantage (MA) encounters separately.

Alternative Models (Models 2 and 3)

We characterized the relationship between primary payer and coding intensity using 2 alternative models that employed a cross-sectional design. Unlike the main model (model 1), inpatient hospitalizations with all payers (Medicare, commercial, Medicaid, and self-pay) were included in the same analysis for models 2 and 3 (whereas model 1 analyzed 3 groups separately). These 2 models included all inpatient hospitalizations for patients aged 60 to 70 years who had at least 2 inpatient hospitalizations. The outcome for each model was the number of discharge diagnoses for an inpatient hospitalization. Each model adjusted for age (centered at 65 years as a continuous variable) and visit count (as a continuous variable). For model 2, the predictor of interest was primary payer as a categorical variable (commercial, Medicaid, and self-pay, with Medicare as the referent category).

For model 3, the predictor of interest was having Medicare at 65 years and older with a prior inpatient hospitalization. In this model, the primary independent variable was a binary variable, coded as 1 for any Medicare-covered inpatient hospitalization in patients 65 years and older occurring after a prior inpatient hospitalization (regardless of insurance for the first hospitalization). All other inpatient hospitalizations (including all first inpatient hospitalizations for a patient in the data set, all non-Medicare inpatient hospitalizations, and all inpatient hospitalizations occurring prior to age 65 years) were coded as 0. The goal of this coding structure was to compare the number of discharge diagnosis codes for nonfirst Medicare inpatient hospitalizations vs all other hospitalizations, using as much data as possible (with imperfect but a relatively high level of precision).

Although these 2 alternative models do not allow for causal inference as does model 1, they mitigate the problem of older Medicare beneficiaries naturally having more diagnoses as they experience more hospitalizations over time. This is because these 2 models include a broader range of hospitalizations including Medicare beneficiaries younger than 65 years who qualified early for Medicare enrollment and commercially insured patients older than 65 years (the age range for these 2 models is 60-70 years). Therefore, these models are more flexible and have a large sample size, allowing for more precise regression coefficient estimates with smaller CIs. The motivation behind these 2 models is similar, but model 2 is intended to measure coding differences among the major payers and model 3 is designed to quantify the coding intensity difference between a nonfirst Medicare hospitalization at 65 years and older vs all other hospitalizations.

RESULTS

The commercial insurance to Medicare, Medicaid to Medicare, and self-pay to Medicare cohorts included 22,351, 14,278, and 4931 encounters, respectively (Table 1). These groups contained 6014, 2680, and 1289 unique patients. The median (IQR) numbers of encounters per patient in the commercial insurance to Medicare, Medicaid to Medicare, and self-pay to Medicare groups were 4 (3-7), 7 (4-12), and 4 (3-8), respectively. Female patients comprised 49.0% of encounters and Black patients comprised 11.6% of encounters in the commercial to Medicare group (eAppendix Table 1 [eAppendix available at ajmc.com]). Female patients and Black patients made up 48.8% and 26.5% of the Medicaid to Medicare group, respectively. In the self-pay to Medicare group, female patients and Black patients made up 41.6% and 17.7% of the encounters. The mean (SD) numbers of discharge diagnosis codes per encounter were 14.3 (7.1), 15.6 (7.1), and 13.7 (6.7) in the commercial insurance to Medicare, Medicaid to Medicare, and self-pay to Medicare cohorts, respectively (Table 1 and Figure).

Commercial Insurance to Medicare Cohort

In the main analysis, Medicare encounters were associated with 0.8 (95% CI, 0.4-1.2) more discharge diagnoses than commercial encounters after adjusting for relevant covariates (Table 2 and eAppendix Table 2). The finding of higher coding intensity for Medicare compared with commercial encounters for the same patients was also observed in the sensitivity analyses and the Medicare FFS subanalysis but not in the MA subanalysis (Table 2 and eAppendix Tables 3-7). Insurance status was not associated with a significant difference in ECI score (Table 2 and eAppendix Table 8). In this cohort, commercially insured encounters for 64-year-olds and Medicare-insured encounters for 65-year-olds had the same top 10 principal discharge diagnoses (albeit in different orders) (eAppendix Table 9).

Medicaid to Medicare Cohort

Medicare encounters were associated with 1.0 (95% CI, 0.5-1.5) more discharge diagnoses than Medicaid encounters after adjustment for confounders in the main analysis (Table 3 and eAppendix Table 10). The findings in the sensitivity and subanalyses were similar, with results ranging from 0.7 (95% CI, 0.1-1.2) more discharge diagnoses in the MA subanalysis to 1.6 (95% CI, 1.3-1.9) in the sensitivity analysis including only 64- and 65-year-olds (Table 3 and eAppendix Tables 11-15). ECI scores were also significantly higher for Medicare encounters (Table 3 and eAppendix Table 16). Nine of the top 10 principal discharge diagnoses were the same for the 64-year-old Medicaid patients and 65-year-old Medicare patients in this cohort (eAppendix Table 17).

Self-Pay to Medicare Cohort

In the main analysis of the self-pay to Medicare cohort, Medicare encounters were associated with 2.0 (95% CI, 1.2-2.8) more discharge diagnoses (Table 4 and eAppendix Table 18). This was consistent across the sensitivity and subanalyses, with findings of a similar magnitude to the main analysis (Table 4 and eAppendix Tables 19-23). Medicare encounters were associated with an ECI score that was 0.5 (95% CI, 0.3-0.8) higher than self-pay encounters (Table 4 and eAppendix Table 24). In this cohort, 64-year-old self-pay patients and 65-year-old Medicare patients had 6 of their top 10 principal discharge diagnoses in common (eAppendix Table 25).

Alternative Models (Models 2 and 3)

In model 2, commercial insurance was associated with 2.57 (95% CI, 2.52-2.62) fewer discharge diagnosis codes than Medicare, Medicaid was associated with 1.07 (95% CI, 1.01-1.13) fewer discharge diagnosis codes than Medicare, and self-pay was associated with 3.8 (95% CI, 3.7-3.9) fewer discharge diagnosis codes than Medicare (eAppendix Table 26). In model 3, an inpatient hospitalization for a Medicare patient 65 years and older (that was not the first inpatient hospitalization for that patient in the data set) was associated with 1.0 (95% CI, 0.96-1.10) more discharge diagnoses than an inpatient hospitalization not in this group (eAppendix Table 27).

DISCUSSION

We utilized a regression discontinuity design to estimate the association of Medicare enrollment with coding intensity for a hospitalization, with patients serving as their own controls to further minimize confounding. Inpatient encounters with Medicare as the primary payer had higher levels of coding intensity, as measured by the number of discharge diagnosis codes, compared with encounters for commercially insured, Medicaid, and self-pay patients. These findings were robust to sensitivity analyses and subanalyses, with the except of the MA subanalysis in the commercial insurance to Medicare group. The magnitude of this disparity was highest in the self-pay to Medicare cohort and lowest in the commercial insurance to Medicare cohort. The ECI score findings mirrored the number of discharge diagnoses findings. Our findings in 2 models using different methodologies from the main model were similar with respect to coding intensity trends.

Quantifying coding intensity is challenging because there is no gold standard. Previous work has utilized variation in mean number of discharge diagnoses across hospitals^2,4 and regions¹¹ as a proxy. A prior analysis found that Medicaid patients discharged from a hospital serving the lowest quartile of private insurance payer mix and then later from a hospital serving the highest quartile had 1.4 (95% CI, 1.2-1.5) more diagnoses at the hospital with the larger commercial payer population.² Another study found that critical access hospitals with fewer financial incentives for high-intensity coding used 3.0 (95% CI, 2.7-3.2) fewer discharge diagnosis codes than non–critical access hospitals for patients with a number of common reasons for hospitalization and that this meaningfully impacted risk-adjusted mortality rates.⁴

Our analysis provides an empirical estimate of coding intensity increases associated with enrolling in Medicare, which often occurs at age 65 years. Most other insurance coverage shifts—like those related to changes in employment or marital status or a decision to enroll in insurance through the health insurance exchanges—are driven by patient decisions, so confounding by indication precludes meaningful comparisons across groups. For this reason, Medicare enrollment at age 65 years has been frequently used in regression discontinuity designs.^12-15

The relative contribution of reimbursement and risk-adjustment considerations in incentivizing higher coding intensity is unclear. We suspect that our findings reflect hospitals’ desire to optimize both reimbursement and risk adjustment. Reimbursement for Medicare FFS inpatient hospitalizations is based on the inpatient prospective payment system, which relies on diagnosis related groups (DRGs). Florida Medicaid inpatient reimbursement is also based on DRGs.¹⁶ The higher coding intensity for Medicare relative to Medicaid hospitalizations may be related to higher reimbursement rates for Medicare patients, making the marginal benefit of increased coding resources higher for Medicare patients. It may also reflect hospitals focusing coding resources on Medicare FFS patients rather than Medicaid patients to optimize performance on important risk-adjusted measures. For example, the CMS Hospital Value-Based Purchasing (VBP) Program¹⁷ risk-adjusted hospital mortality and readmission metrics, which include Medicare FFS but not Medicaid inpatient hospitalizations, are publicly reported and impact reimbursement.

For Medicare FFS patients (and Florida Medicaid patients), the DRG category is based on the principal discharge diagnosis, so utilizing many discharge diagnoses may not impact a patient’s DRG category. Our analysis showed significant overlap in principal discharge diagnoses between 64- and 65-year-olds for each cohort (eAppendix Tables 9, 17, and 25). However, within a DRG category (determined by the principal discharge diagnosis), secondary discharge diagnoses can impact the specific DRG (and thus reimbursement) through the presence of complications and comorbidities and major complications and comorbidities. In this way, listing more discharge diagnoses increases the expected reimbursement before the exact DRG is determined.

Reimbursement for MA and commercially insured inpatient hospitalizations is determined by contracts between insurance companies and hospitals, which can be DRG based or FFS based. Using more discharge diagnoses is less likely to impact reimbursement in an FFS arrangement. Reimbursement arrangements between hospitals and commercial payers are typically confidential, and to our knowledge, the relative prevalence of these 2 payment mechanisms is not known. However, prior work suggests that relative market power of hospitals and insurance companies in a geographic area determines which payment mechanism predominates.¹⁸ This difference between Medicare FFS and MA reimbursement may explain the higher-intensity coding in Medicare FFS compared with MA in the Medicare subanalyses in each cohort, although this comparison was indirect.

In addition to the different reimbursement mechanisms, hospitals’ attempts to optimize risk adjustment (which can impact reimbursement) might also explain the Medicare subanalysis findings. Medicare FFS but not MA hospitalizations are used to calculate CMS Hospital VBP Program metrics.¹⁷ Prior work suggests that including MA patients could have a significant impact on which hospitals fall into the top-performing quartile.¹⁹ Moreover, research actually suggests more intensive coding for MA than Medicare FFS patients in the ambulatory setting.^5,20 Although we did not directly compare Medicare FFS and MA patients, our findings suggest lower coding intensity for MA patients compared with Medicare FFS patients in the inpatient setting, which makes sense from both a reimbursement and a risk-adjustment perspective. This is because all Medicare FFS inpatient reimbursement and not all MA inpatient reimbursement uses the DRG system and because various CMS hospital-level quality measures such as risk-adjusted mortality and readmissions include Medicare FFS but not MA patients.

The magnitude of coding intensity increase was highest in the self-pay to Medicare group. Given that a significant number of self-pay patients are uninsured and unlikely to be able to pay their entire hospitalization bill, aggressive coding may not increase reimbursement. Additionally, these patients are less likely than insured patients to be included in risk-adjusted quality measures.

Finally, our analysis suggests that the impact of insurance status on coding intensity may confound the ECI score¹⁰ as a means of adjusting for medical complexity of hospitalized patients. This may be especially important when self-pay patients in the seventh decade of life are compared with similarly aged Medicare patients. However, the magnitude of this effect is small, and it is unclear whether it could meaningfully impact analyses using the ECI score to adjust for patient complexity.

Limitations

There are several noteworthy limitations. First, the methodology for our primary model required that each patient have at least 2 hospitalizations (one before age 65 years, prior to Medicare enrollment, and another at 65 years and older with Medicare). This model allowed for different slopes (modeling the relationship between age and number of discharge diagnoses) before and after aged 65 years and also adjusted for the inpatient visit count. However, there was a strong relationship between visit count and the outcome of interest, so residual confounding is still a concern. Models 2 and 3 were more flexible but also required at least 2 inpatient hospitalizations. Thus, our results may not be broadly generalizable if beneficiaries who have 2 inpatient hospitalizations within the study window are different from beneficiaries with only 1 inpatient hospitalization. Reassuringly, the significant overlap of principal discharge diagnoses for patients aged 64 and 65 years suggests that the most common reasons for hospitalization did not change appreciably over that year.

Second, our data include only Florida inpatient hospitalizations, so our results may not be generalizable to the extent that Florida is dissimilar to other states. State-specific health care market dynamics are likely to impact reimbursement and thus coding intensity. For example, managed care penetration for commercial insurance and Medicaid differs across states. Health care market competition—hospital or health system compared with insurer market power—is also likely to impact reimbursement and coding intensity, and this varies across states. Medicaid reimbursement rates also vary across states. A 2017 analysis of Florida found that Medicaid reimbursement was significantly above the national average.²¹ Additionally, coding intensity for self-pay patients may depend on the proportion of uninsured patients within a state, which likely depends on a number of factors that vary across states, such as whether they have adopted the Affordable Care Act’s Medicaid expansion, which Florida has not.²²

CONCLUSIONS

Inpatient encounters for Medicare patients 65 years and older were coded more intensely than commercially insured, Medicaid, or self-pay inpatient encounters for those same patients prior to age 65 years. This suggests that insurance status may meaningfully impact coding intensity of inpatient hospitalizations, which, in turn, may lead to reimbursement disparities and variable or biased risk adjustment across hospitals. Future studies should measure the extent to which this relationship between insurance status and coding intensity impacts risk-adjusted quality measures and also directly compare inpatient coding intensity between Medicare FFS and MA patients. Future studies could also further examine the issue of whether different codes are more common for different insurance types beyond principal discharge diagnoses, including looking at Z codes, which describe health factors such as personal and family history of diseases and generally do not influence billing or risk adjustment and are thus unlikely to account for the entire increase in coding intensity around age 65 years.

Author Affiliations: Department of Medicine, Johns Hopkins University School of Medicine (MIE, DJB), Baltimore, MD.

Source of Funding: Dr Ellenbogen was supported by the Agency for Healthcare Research and Quality (AHRQ) grant K08HS028673.

Author Disclosures: Dr Ellenbogen was a consultant for Garner Health at the time this research was done and subsequent to finishing the research has taken a full-time job at Garner Health and owns stock in the company; he also received the above-mentioned AHRQ K08 award. Dr Brotman reports no relationship or financial interest with any entity that would pose a conflict of interest with the subject matter of this article.

Authorship Information: Concept and design (MIE, DJB); acquisition of data (MIE, DJB); analysis and interpretation of data (MIE, DJB); drafting of the manuscript (MIE); critical revision of the manuscript for important intellectual content (MIE, DJB); statistical analysis (MIE); obtaining funding (MIE); and supervision (DJB).

Address Correspondence to: Michael I. Ellenbogen, MD, Johns Hopkins University School of Medicine, 600 N Wolfe St, Meyer 8-134P, Baltimore, MD 21287. Email: mellenb6@jhmi.edu.

REFERENCES

1. Ellenbogen MI, Weiner JP, Zhu Y, Swann J, Brotman DJ. Development of a hospital coding intensity measure based on sepsis diagnoses. J Hosp Med. 2024;19(6):505-507. doi:10.1002/jhm.13351

2. Dragan KL, Desai SM, Billings J, Glied SA. Association of insurance mix and diagnostic coding practices in New York state hospitals. JAMA Health Forum. 2022;3(9):e222919. doi:10.1001/jamahealthforum.2022.2919

3. Goldman LE, Chu PW, Osmond D, Bindman A. The accuracy of present-on-admission reporting in administrative data. Health Serv Res. 2011;46(6, pt 1):1946-1962. doi:10.1111/j.1475-6773.2011.01300.x

4. Kosar CM, Loomer L, Thomas KS, White EM, Panagiotou OA, Rahman M. Association of diagnosis coding with differences in risk-adjusted short-term mortality between critical access and non-critical access hospitals. JAMA. 2020;324(5):481-487. doi:10.1001/jama.2020.9935

5. Geruso M, Layton T. Upcoding: evidence from Medicare on squishy risk adjustment. J Polit Econ. 2020;128(3):984-1026. doi:10.1086/704756

6. Rubens M, Saxena A, Ramamoorthy V, et al. Increasing sepsis rates in the United States: results from National Inpatient Sample, 2005 to 2014. J Intensive Care Med. 2020;35(9):858-868. doi:10.1177/0885066618794136

7. Lindenauer PK, Lagu T, Shieh MS, Pekow PS, Rothberg MB. Association of diagnostic coding with trends in hospitalizations and mortality of patients with pneumonia, 2003-2009. JAMA. 2012;307(13):1405-1413. doi:10.1001/jama.2012.384

8. User support: SID overview. Agency for Healthcare Research and Quality Healthcare Cost and Utilization Project. Updated September 3, 2025. Accessed October 2, 2025. https://hcup-us.ahrq.gov/sidoverview.jsp

9. AHA Annual Survey Database. American Hospital Association Data and Insights. Accessed January 3, 2024. https://www.ahadata.com/aha-annual-survey-database

10. Moore BJ, White S, Washington R, Coenen N, Elixhauser A. Identifying increased risk of readmission and in-hospital mortality using hospital administrative data: the AHRQ Elixhauser Comorbidity Index. Med Care. 2017;55(7):698-705. doi:10.1097/MLR.0000000000000735

11. Song Y, Skinner J, Bynum J, Sutherland J, Wennberg JE, Fisher ES. Regional variations in diagnostic practices. N Engl J Med. 2010;363(1):45-53. doi:10.1056/NEJMsa0910881

12. Wallace J, Jiang K, Goldsmith-Pinkham P, Song Z. Changes in racial and ethnic disparities in access to care and health among US adults at age 65 years. JAMA Intern Med. 2021;181(9):1207-1215. doi:10.1001/jamainternmed.2021.3922

13. Bhaumik D, Ndumele CD, Scott JW, Wallace J. Association between Medicare eligibility at age 65 years and in-hospital treatment patterns and health outcomes for patients with trauma: regression discontinuity approach. BMJ. 2023;382:e074289. doi:10.1136/bmj-2022-074289

14. Park S, Wadhera RK, Jung J. Effects of Medicare eligibility and enrollment at age 65 years on the use of high-value and low-value care. Health Serv Res. 2023;58(1):174-185. doi:10.1111/1475-6773.14065

15. Wallace J, Song Z. Traditional Medicare versus private insurance: how spending, volume, and price change at age sixty-five. Health Aff (Millwood). 2016;35(5):864-872. doi:10.1377/hlthaff.2015.1195

16. States’ Medicaid fee-for-service inpatient hospital payment policies. Medicaid and CHIP Payment and Access Commission. Accessed May 10, 2024. https://www.macpac.gov/wp-content/uploads/2018/12/State-Medicaid-Fee-for-Service-Inpatient-Hospital-Payment-Policies.xlsx

17. Hospital Value-Based Purchasing Program. CMS. Accessed January 2, 2024. https://www.cms.gov/medicare/quality/initiatives/hospital-quality-initiative/hospital-value-based-purchasing

18. Cooper Z, Craig SV, Gaynor M, Van Reenen J. The price ain’t right? hospital prices and health spending on the privately insured. Q J Econ. 2019;134(1):51-107. doi:10.1093/qje/qjy020

19. Oseran AS, Wadhera RK, Orav EJ, Figueroa JF. Effect of Medicare Advantage on hospital readmission and mortality rankings. Ann Intern Med. 2023;176(4):480-488. doi:10.7326/M22-3165

20. Kronick R, Welch WP. Measuring coding intensity in the Medicare Advantage program. Medicare Medicaid Res Rev. 2014;4(2):mmrr2014.004.02.a06. doi:10.5600/mmrr.004.02.sa06

21. Medicaid hospital payment: a comparison across states and to Medicare. Medicaid and CHIP Payment and Access Commission. April 2017. Accessed May 2, 2024. https://www.macpac.gov/wp-content/uploads/2017/04/Medicaid-Hospital-Payment-A-Comparison-across-States-and-to-Medicare.pdf

22. Status of state Medicaid expansion decisions. KFF. May 8, 2024. Accessed June 7, 2024. https://www.kff.org/affordable-care-act/issue-brief/status-of-state-medicaid-expansion-decisions-interactive-map/