Vertical Integration Into Skilled Nursing Facilities and Hospital Readmission Rates

The Long-Term Effect of Financial Penalties on 30-Day Hospital Readmission Rates

BACKGROUND

Although the immediate effect of financial penalties imposed by the Hospital Readmissions Reduction Program (HRRP) was a decrease in 30-day hospital readmission rates, the long-term effects are unclear. The authors studied 30-day readmissions before and immediately after HRRP penalties and during the most recent period before the COVID-19 pandemic and examined whether readmission trends differed between penalized and non-penalized hospitals.

METHODS

Centers for Medicare & Medicaid Services hospital archive data and US Census Bureau data were used to analyze hospital characteristics, including readmission penalty status, and hospital service area (HSA) demographic information, respectively. These two datasets were matched by HSA crosswalk files, available through the Dartmouth Atlas files. Using data from 2005-2008 as baseline, the authors examined hospital readmission trends before (2008-2011) and after penalties (during three periods: 2011-2014, 2014-2017, 2017-2019). Mixed linear models were used to examine readmission trends through periods, and differences by hospital penalty status without and with adjustment for hospital characteristics and HSA demographic information.

RESULTS

For all hospitals combined, rates for 2008-2011 vs. 2011-2014 were as follows: pneumonia, 18.6% vs. 17.0%; heart failure (HF), 24.8% vs. 22.0%; acute myocardial infarction (AMI), 19.7% vs. 17.0% (p < 0.001 for all three conditions). Rates for 2014-2017 vs. 2017-2019 were as follows: pneumonia, 16.8% vs. 16.8% (p = 0.87), HF, 21.7% vs. 21.9% (p < 0.001); AMI, 16.0% vs. 15.8% (p < 0.001). Compared to penalized hospitals, using difference-in-differences, non-penalized hospitals had a significantly greater increase for two conditions between the 2014-2017 and 2017-2019 periods: pneumonia 0.34%, p < 0.001; and HF 0.24%, p = 0.002.

CONCLUSION

Long-term readmission rates are lower than pre-HRRP rates, with recent trends decreasing further for AMI, stabilizing for pneumonia, and increasing for HF.

Association of Skilled Nursing Facility Ownership by Health Care Networks With Utilization and Spending

IMPORTANCE

Health care mergers and acquisitions have increased vertical integration of skilled nursing facilities (SNFs) in health care networks. While vertical integration may result in improved care coordination and quality, it may also lead to excess utilization, as SNFs are paid a per diem rate.

OBJECTIVE

To determine the association of vertical integration of SNFs within hospital networks with SNF utilization, readmissions, and spending for Medicare beneficiaries undergoing elective hip replacement.

DESIGN, SETTING, AND PARTICIPANTS

This cross-sectional study evaluated 100% Medicare administrative claims for nonfederal acute care hospitals performing at least 10 elective hip replacements during the study period. Fee-for-service Medicare beneficiaries aged 66 to 99 years who underwent elective hip replacement between January 1, 2016, and December 31, 2017, with continuous Medicare coverage for 3 months before and 6 months after surgery were included. Data were analyzed from February 2 to August 8, 2022.

EXPOSURES

Treatment at a hospital within a network that also owns at least 1 SNF based on the 2017 American Hospital Association survey.

MAIN OUTCOMES AND MEASURES

Rates of SNF utilization, 30-day readmissions, and price-standardized 30-day episode payments. Hierarchical multivariable logistic and linear regression clustered at hospitals was performed with adjusting for patient, hospital, and network characteristics.

RESULTS

A total of 150 788 patients (61.4% women; mean [SD] age, 74.3 [6.4] years) underwent hip replacement. After risk adjustment, vertical SNF integration was associated with a higher rate of SNF utilization (21.7% [95% CI, 20.4%-23.0%] vs 19.7% [95% CI, 18.7%-20.7%]; adjusted odds ratio [aOR], 1.15 [95% CI, 1.03-1.29]; P = .01) and lower 30-day readmission rate (5.6% [95% CI, 5.4%-5.8%] vs 5.9% [95% CI, 5.7%-6.1%]; aOR, 0.94 [95% CI, 0.89-0.99]; P = .03). Despite higher SNF utilization, the total adjusted 30-day episode payments were slightly lower ($20 230 [95% CI, $20 035-$20 425] vs $20 487 [95% CI, $20 314-$20 660]; difference, -$275 [95% CI, -$15 to -$498]; P = .04) driven by lower postacute payments and shorter SNF length of stays. Adjusted readmission rates were particularly low for patients not sent to an SNF (3.6% [95% CI, 3.4%-3.7%]; P < .001) but were significantly higher for patients with an SNF length of stay less than 5 days (41.3% [95% CI, 39.2%-43.3%]; P < .001).

CONCLUSIONS AND RELEVANCE

In this cross-sectional study of Medicare beneficiaries undergoing elective hip replacements, vertical integration of SNFs in a hospital network was associated with higher rates of SNF utilization and lower rates of readmissions without evidence of higher overall episode payments. These findings support the purported value of integrating SNFs into hospital networks but also suggest that there is room for improving the postoperative care of patients in SNFs early in their stay.

AI Models for Predicting Readmission of Pneumonia Patients within 30 Days after Discharge

A model with capability for precisely predicting readmission is a target being pursued worldwide. The objective of this study is to design predictive models using artificial intelligence methods and data retrieved from the National Health Insurance Research Database of Taiwan for identifying high-risk pneumonia patients with 30-day all-cause readmissions. An integrated genetic algorithm (GA) and support vector machine (SVM), namely IGS, were used to design predictive models optimized with three objective functions. In IGS, GA was used for selecting salient features and optimal SVM parameters, while SVM was used for constructing the models. For comparison, logistic regression (LR) and deep neural network (DNN) were also applied for model construction. The IGS model with AUC used as the objective function achieved an accuracy, sensitivity, specificity, and area under ROC curve (AUC) of 70.11%, 73.46%, 69.26%, and 0.7758, respectively, outperforming the models designed with LR (65.77%, 78.44%, 62.54%, and 0.7689, respectively) and DNN (61.50%, 79.34%, 56.95%, and 0.7547, respectively), as well as previously reported models constructed using thedata of electronic health records with an AUC of 0.71–0.74. It can be used for automatically detecting pneumonia patients with a risk of all-cause readmissions within 30 days after discharge so as to administer suitable interventions to reduce readmission and healthcare costs.

Evaluating Telehealth Adoption and Related Barriers Among Hospitals Located in Rural and Urban Areas

Purpose

To assess telehealth adoption among hospitals located in rural and urban areas, and identify barriers related to enhanced telehealth capabilities in the areas of patient engagement and health information exchange (HIE) capacity with external providers and community partners.

Methods

We used the 2018 American Hospital Association (AHA) Annual Survey and IT Supplement Survey. We applied state fixed effects multivariate analyses and Oaxaca decomposition to estimate the variation of outcomes of interest by hospital geographies.

Findings

Our research showed substantial differences in telehealth adoption among hospitals located in rural, micropolitan, and metropolitan areas, where adoption rates increase with urbanicity. Rural hospitals were least likely to have telehealth systems with patient engagement capabilities such as the ability to view their health information online and electronically transmit medical information to a third party. They were also the least likely to report that clinical information was available electronically from outside providers. Our model explained 65% of the rural/urban difference in telehealth adoption, 55% of the number of telehealth services adopted, and 43%‐49% of the rural/urban difference in telehealth barriers.

Conclusion

Findings demonstrated significant barriers to telehealth use among hospitals located in rural and urban areas. For rural hospitals, barriers include lack of HIE capacity among health care providers in the community, and lack of patient engagement capability.