Highlights
-
•
Patients with STEMI at for-profit and public hospitals had greater socioeconomic needs than those at nonprofit hospitals.
-
•
For-profit hospitals had worse in-hospital mortality and 90-day readmission outcomes after STEMI than nonprofit hospitals.
-
•
Public hospitals also worse in-hospital mortality and readmission outcomes after STEMI than nonprofit hospitals.
ABSTRACT
Background
Hospital ownership type may influence acute cardiovascular (CV) disease disparities that persist across the U.S. We examined associations between hospital ownership type and in-hospital and readmission outcomes for ST-elevation myocardial infarction (STEMI) hospitalizations.
Methods
We performed a retrospective cohort study of hospitalizations for STEMI using the National Readmissions Database (2016-2022). Hospitals were categorized as nonprofit, for-profit, or public. Outcomes included in-hospital mortality and 90-day readmission for acute coronary syndrome, heart failure (HF), CV, and all causes. Associations were assessed using multivariable logistic and Cox proportional hazards regression, adjusting for patient, hospitalization, and hospital-level characteristics.
Results
Of 610,427 STEMI hospitalizations, 460,451 (75.4%) were at nonprofit, 88,965 (14.6%) at for-profit, and 61,011 (10.0%) at public hospitals. Compared with nonprofit hospitals, for-profit hospitals (adjusted odds ratio [OR] 1.09, 95% confidence interval [CI] 1.05-1.13) and public hospitals (aOR 1.17, 95% CI 1.12-1.22) were each associated with higher odds of in-hospital mortality. For-profit hospitals were associated with higher risk of 90-day readmission for acute coronary syndrome (adjusted hazards ratio [HR] 1.15, 95% CI 1.10–1.21), HF (aHR 1.08, 95% CI 1.03–1.13), CV (aHR 1.08, 95% CI 1.05–1.12), and all causes (aHR 1.13, 95% CI 1.10–1.16) relative to nonprofit hospitals. Public hospitals were associated with higher risk of 90-day readmission for HF (aHR 1.08, 95% CI 1.02–1.13) relative to nonprofit hospitals.
Conclusions
For-profit and public hospitals were associated with higher in-hospital mortality and 90-day readmission for various causes compared with nonprofit hospitals. These findings suggest that hospital-level factors may contribute to disparities in STEMI outcomes and warrant further investigation.
Graphical Abstract
With advances in reperfusion strategies and optimal medical therapy for ST-elevation myocardial infarction (STEMI), , STEMI mortality has declined over the last few decades. , Despite these advances, however, disparities in STEMI outcomes persist across the United States, and morbidity and mortality remain high among underserved populations. ,, These persistent gaps raise questions about how systems-level factors, including hospital characteristics, may influence STEMI outcomes. In particular, the growth of for-profit hospitals over the past couple of decades has reinvigorated attention to how hospital ownership type (for-profit, nonprofit, and public) may affect medical care and health outcomes. ,,
Prior research has demonstrated that hospital ownership type is associated with differences in staffing patterns and service offerings, often in ways linked to financial incentives. For example, for-profit hospitals have been shown to have lower investments in nursing services and lower nurse-to-patient staffing ratios compared with nonprofit hospitals, as well as be more likely to offer relatively profitable services such as cardiac surgery compared with nonprofit and public hospitals. , Though some prior studies have explored how hospital ownership type affects acute coronary syndrome (ACS) outcomes, these data have often predated the Affordable Care Act (ACA). The ACA introduced major changes to healthcare coverage that differentially affected revenue and operating margins for nonprofit, for-profit, and public hospitals, ,, with potential implications for care delivery and patient outcomes. One study of patients admitted for acute myocardial infarction (MI) in the mid-1990s found that patients at for-profit hospitals had higher 30-day and 1-year mortality than patients at nonprofit hospitals, whereas a study of patients with non-ST-segment elevation myocardial infarction in the early-2000s found that in-hospital mortality was similar between for-profit and nonprofit hospitals. Despite the major reforms and effects of the ACA, associations between hospital ownership type and ACS outcomes, particularly for STEMI, have not been well-characterized in the post-ACA era.
To address this gap, our study used a large, nationally representative database to identify hospitalizations for STEMI at nonprofit, for-profit, and public hospitals in the United States from 2016 to 2022, encompassing the post-ACA implementation period and early coronavirus disease 2019 (COVID-19) era. We described patient, hospitalization, and hospital-level characteristics by hospital ownership type, and we assessed associations between hospital ownership type and in-hospital mortality and readmission outcomes.
Methods
Data source
This study was a retrospective observational study using the National Readmissions Database (NRD). The NRD is a publicly available database developed for the Healthcare Cost and Utilization Project that captures nationally representative all-payer hospital admissions in the United States, constructed annually for each calendar year. Each record in the NRD represents an inpatient hospitalization, and records are assigned a unique identifier to track a patient’s hospitalizations across hospitals within a state and calendar year. As described by the NRD, the database does not include regional or state-level identifiers and is “designed to support national-level readmission analyses but not for use in regional, state-, or hospital-specific analyses.” Because the NRD is a publicly available and deidentified database, NYU Langone Health’s Institutional Review Board did not require human subject’s review.
Sample
We included index hospitalizations for adult patients (≥18 years old) who had a primary discharge diagnosis of STEMI from 2016 to 2022. This study period includes post-ACA implementation years and early COVID-19 years, and it corresponds with the period during which the NRD used International Classification of Diseases (ICD)-10 codes to classify diseases, thus minimizing potential misclassification from transitional coding practices. Table A.1 includes ICD-10 codes used to identify clinical characteristics. Charlson Comorbidity Index and Frailty scores were calculated as previously described. ,,,
Exposure
The exposure variable was hospital ownership type of the final discharge hospital, which included private nonprofit, private for-profit (investor-owned), and public (government-owned). The NRD obtained hospital ownership type data for each calendar year from the American Hospital Association Annual Survey of Hospitals. Updated annually, the American Hospital Association Annual Survey of Hospitals is a widely-used source of information about U.S. hospitals, including hospital ownership type, in the literature. ,
Outcomes
We compared index hospitalization and readmission outcomes among nonprofit, for-profit, and public hospitals, using nonprofit hospitals as the reference. The index hospitalization outcome was in-hospital mortality. Readmission outcomes included 90-day readmission for ACS, composite of STEMI, non-STEMI, and unstable angina), heart failure (HF), cardiovascular (CV) causes (composite of HF, acute MI, ischemic stroke, arrhythmias, arterial thromboembolism, and venous thromboembolism), and all causes, as well as death during any 90-day readmission. Patients who experienced index hospitalization in-hospital mortality were excluded from readmission analyses.
In supplemental analyses, we examined 30-day ACS, HF, CV, and all-cause readmission. Under the Hospital Readmission Reduction Program, hospitals paid by Medicare are financially penalized for 30-day unplanned readmissions after acute MI. It is possible that solely examining 30-day readmission could obscure potential differences in readmission across nonprofit, for-profit, and public hospitals, so we analyzed 90-day readmissions in our main analyses and 30-day readmissions in our supplemental analyses. In additional supplemental analyses, we examined cumulative number of ACS, HF, CV, and all-cause readmissions for the remaining calendar year after the index hospitalization.
Statistical analyses
Descriptive statistics were used to summarize index hospitalization characteristics and study outcomes by hospital ownership type. For groupwise comparisons, the Chi-square test was used for categorical variables, analysis of variance for normally distributed continuous variables, and Kruskal–Wallis for non-normally distributed continuous variables. Multivariable logistic regression was used to model in-hospital mortality during index hospitalization, with results presented as odds ratios (OR) and 95% confidence intervals (CI). Cox proportional hazards regression was used to model risk of readmission within 90 days after index hospitalization discharge and risk of death during any readmission within those 90 days, with results presented as hazard ratios (HR). Patients discharged after September were excluded from 90-day readmission analyses. Negative binomial regression was used to estimate the cumulative number of readmissions within the remaining calendar year, with results presented as incidence rate ratios. Additionally, the Mann–Kendall test was used to detect monotonic upward or downward trends in in-hospital mortality and 90-day readmissions among nonprofit, for-profit, and public hospitals across the study period.
In line with prior literature, all models were adjusted for characteristics at the patient, hospitalization, and hospital-level. , Patient characteristics included age, sex, insurance status, urban-rural status for patient’s zip code, median household income quartile for patient’s zip code, prior MI, prior coronary artery bypass grafting (CABG), HF, atrial fibrillation, anemia, thrombocytopenia, hypertension, Frailty score, and Charlson Comorbidity Index (CCI). Hospitalization characteristics included transfer status, percutaneous coronary intervention (PCI), PCI within 24 hours of admission, CABG, mechanical circulatory support (MCS) use, mechanical complication, cardiogenic shock, mechanical ventilation, palliative care, do-not-resuscitate status, and patient disposition. Hospital characteristics included hospital size, teaching status, hospital urban-rural status, and PCI-capable hospital status (defined as ≥1 PCI procedure recorded for the study year). Multicollinearity was evaluated using variance inflation factor scores. To prevent overfitting of models, variables with high degrees of multicollinearity (variance inflation factor >10) were excluded. Additionally, given the hierarchical structure of the NRD, all models were adjusted for clustering at the hospital-year level, as the NRD includes unique identifiers for hospitals within a calendar year. All analyses were two-tailed, with a P-value of <.05 considered significant. Analyses were conducted with SPSS version 29.0 (IBM) and Stata version 15 (STATA Corporation).
Two sensitivity analyses were performed. First, primary outcomes were investigated among index hospitalizations that included PCI to determine whether results persisted in this subgroup. Additionally, given the potential effect of the COVID-19 pandemic on STEMI care and outcomes in the study period, results for primary outcomes were stratified into pre-pandemic (2016-2019) and pandemic and post-pandemic (2020-2022) study years.
Results
Characteristics of STEMI hospitalizations
We included 610,427 index hospitalizations for STEMI from 2016-2022. Of these, 460,451 (75.4%) were at nonprofit hospitals, 88,965 (14.6%) were at for-profit hospitals, and 61,011 (10.0%) were at public hospitals. Patient, hospitalization, and hospital-level characteristics by hospital ownership type are shown in Table 1 .
Table 1
Patient, hospitalization, and hospital characteristics for STEMI hospitalizations by hospital ownership type.
|
All hospitalizations
N = 610,427 |
Nonprofit
n = 460,451 |
For-profit
n = 88,965 |
Public
n = 61,011 |
For-profit
vs nonprofit P -value |
Public
vs nonprofit P -value |
|
|---|---|---|---|---|---|---|
| Patient characteristics | ||||||
|
Age,
mean (SD) |
63.9 (13.0) | 64.0 (13.0) | 63.9 (13.0) | 63.4 (13.0) | .03 | <.001 |
|
Female sex,
n (%) |
185,967 (30.5) | 140,733 (30.6) | 27,257 (30.6) | 17,977 (29.5) | .66 | <.001 |
| Insurance, n (%) | <.001 | <.001 | ||||
| Medicare | 284,515 (46.6) | 214,851 (46.7) | 42,263 (47.5) | 27,401 (44.9) | ||
| Medicaid | 65,901 (10.8) | 50,143 (10.9) | 8,660 (9.7) | 65,901 (10.8) | ||
| Private insurance | 198,546 (32.5) | 153,612 (33.4) | 26,846 (30.2) | 18,088 (29.6) | ||
| Self-pay or no charge | 38,491 (6.3) | 25,394 (5.5) | 7,215 (8.1) | 5,882 (9.6) | ||
| Zip code in lowest quartile for median household income, n (%) | 161,332 (26.4) | 111,051 (24.1) | 29,203 (32.8) | 21,078 (34.5) | <.001 | <.001 |
| Zip code in non-metropolitan area, n (%) | 106,774 (17.5) | 79,063 (17.2) | 13,520 (15.2) | 14,191 (23.3) | <.001 | <.001 |
|
Comorbidities,
n (%) |
||||||
| Prior myocardial infarction | 72,801 (11.9) | 53,329 (11.6) | 12,534 (14.1) | 6,938 (11.4) | <.001 | .13 |
| Prior PCI | 74,247 (12.2) | 56,482 (12.3) | 10,753 (12.1) | 7,012 (11.5) | .13 | <.001 |
| Prior CABG | 24,260 (4.0) | 17,722 (3.8) | 4,327 (4.9) | 2,211 (3.6) | <.001 | .006 |
| Congestive heart failure | 232,401 (38.1) | 177,714 (38.6) | 32,438 (36.5) | 22,249 (36.5) | <.001 | <.001 |
| Atrial fibrillation | 90,934 (14.9) | 69,535 (15.1) | 12,801 (14.4) | 8,598 (14.1) | <.001 | <.001 |
| Anemia | 61,637 (10.1) | 45,685 (9.9) | 9,786 (11.0) | 6,166 (10.1) | <.001 | .15 |
| Thrombocytopenia | 24,009 (3.9) | 18,569 (4.0) | 3,328 (3.7) | 2,112 (3.5) | <.001 | <.001 |
| Hypertension | 447,256 (73.3) | 335,469 (72.9) | 67,369 (75.7) | 44,418 (72.8) | <.001 | .78 |
| Frailty score, mean (SD) | 1.0 (2.0) | 1.0 (2.0) | 0.9 (1.9) | 1.0 (2.1) | <.001 | .001 |
|
Charlson Comorbidity Index,
mean (SD) |
3.5 (2.2) | 3.5 (2.2) | 3.5 (2.2) | 3.4 (2.2) | .14 | <.001 |
| Hospitalization characteristics | ||||||
| PCI, n (%) | 501,327 (82.1) | 378,812 (82.3) | 72,552 (81.5) | 49,963 (81.9) | <.001 | .02 |
| PCI within 24 hours of admission, n (%) | 451,987 (74.0) | 341,309 (74.1) | 65,408 (73.5) | 45,270 (74.2) | <.001 | .69 |
| CABG, n (%) | 32,042 (5.2) | 23,692 (5.1) | 5,514 (6.2) | 2,836 (4.6) | <.001 | <.001 |
| Temporary MCS, n (%) | 21,741 (3.6) | 16,216 (3.5) | 3,064 (3.4) | 2,461 (4.0) | .25 | <.001 |
|
Type of MCS,
n (% of MCS) |
||||||
| IABP | 1,967 (9.0) | 1,385 (8.5) | 384 (12.5) | 198 (8.0) | <.001 | .41 |
| pLVAD | 18,202 (83.7) | 13,497 (83.2) | 2,646 (86.4) | 2,059 (83.7) | <.001 | .59 |
| ECMO | 3,296 (15.2) | 2,727 (16.8) | 143 (4.7) | 426 (17.3) | <.001 | .54 |
|
Mechanical complication,
n (%) |
2,971 (0.5) | 2,381 (0.5) | 294 (0.3) | 296 (0.5) | <.001 | .30 |
| Cardiogenic shock, n (%) | 81,942 (13.4) | 62,210 (13.5) | 11,734 (13.2) | 7,998 (13.1) | .01 | .01 |
| Mechanical ventilation, n (%) | 63,083 (10.3) | 46,384 (10.1) | 9,970 (11.2) | 6,729 (11.0) | <.001 | <.001 |
|
Palliative care,
n (%) |
25,142 (4.1) | 19,699 (4.3) | 3,022 (3.4) | 2,421 (4.0) | <.001 | <.001 |
| Do-not-resuscitate status, n (%) | 43,982 (7.2) | 33,265 (7.2) | 6,091 (6.8) | 4,626 (7.6) | <.001 | .001 |
| Interhospital transfer, n (%) | 15,706 (2.6) | 12,431 (2.7) | 1,869 (2.1) | 1,406 (2.3) | <.001 | <.001 |
|
Patient disposition,
n (%) |
<.001 | <.001 | ||||
| Rehab or other facility | 45,835 (7.5) | 34,193 (7.4) | 7,190 (8.1) | 4,452 (7.3) | ||
| Against medical advice | 6,205 (1.0) | 4,300 (0.9) | 1,304 (1.5) | 601 (1.0) | ||
| Other or unknown | 49,960 (8.2) | 7,405 (8.3) | 7,405 (8.3) | 5,489 (9.0) | ||
| Hospital characteristics | ||||||
|
Hospital size,
n (%) |
<.001 | <.001 | ||||
| Large | 357,024 (58.5) | 274,825 (59.7) | 34,168 (38.4) | 48,031 (78.7) | ||
| Medium | 173,688 (28.4) | 125,487 (27.2) | 38,923 (43.7) | 9,278 (15.2) | ||
| Small | 70,715 (13.1) | 60,139 (13.1) | 15,874 (17.8) | 3,702 (6.1) | ||
| Hospital location, n (%) | <.001 | <.001 | ||||
| Large metropolitan | 306,959 (50.3) | 234,257 (50.9) | 45,670 (51.3) | 27,032 (44.3) | ||
| Small metropolitan | 271,130 (44.4) | 203,287 (44.1) | 39,354 (44.2) | 28,489 (46.7) | ||
| Micropolitan | 30,197 (4.9) | 21,643 (4.7) | 3,820 (4.3) | 4,734 (7.8) | ||
| Rural | 2,141 (0.3) | 1,264 (0.3) | 121 (0.1) | 756 (1.2) | ||
| Hospital teaching status, n (%) | <.001 | <.001 | ||||
| Metropolitan teaching | 445,201 (72.9) | 347,311 (75.4) | 53,187 (59.8) | 44,703 (73.3) | ||
| Metropolitan non-teaching or non-metropolitan | 165,226 (27.1) |
53,140
(24.6) |
35,778
(40.2) |
16,308 (26.7) | ||
| PCI-capable hospital, n (%) | 606,101 (99.3) | 457,748 (99.4) | 88,277 (99.2) | 60,076 (98.5) | <.001 | <.001 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
