Objective: To examine spillovers from a federal policy, managed care market, and community perspective. Data Sources/Study Setting: We studied spillovers from a federal policy and managed care market perspective using the Health Care Utilization Project’s (HCUP) State Inpatient Database (SID). American Hospital Association (AHA) data, Interstudy Commercial Managed Care, and Area Health Resource File (AHRF). Medicare Advantage county-level payment schedules originate from CMS. We examined community uninsurance spillovers using 2011-2015 Medical Expenditure Panel Survey (MEPS), the Area Health Resource File (AHRF), and the Small Area Health Insurance Estimator (SAHIE). Study Design: Ordinary Least Squares (OLS) and difference-in-difference regression analyses were used to examine a federal policy spillover on hospital readmissions. We used OLS and instrumental variable (IV) estimation to examined Medicare Advantage (MA) spillovers on Medicare fee-for-service (FFS) hospital readmissions. We used logistic regression to examine community uninsurance spillovers on the privately insured. Principal Findings: After the HRRP, Medicare FFS saw a decrease in 30-day preventable condition- and all-cause readmissions. Medicare Advantage saw a positive spillover after the HRRP. MA market penetration has no effect on Medicare FFS hospital readmissions. High community uninsurance rates are associated with less access to behavioral health related outpatient/office-based and prescription utilization. Conclusions: HRRP had a positive spillover on MA hospital all-cause readmissions. MA market penetration has no effect on Medicare FFS readmissions. High levels of community uninsurance are associated with poorer access to outpatient/office-based and prescription behavioral related services.

Academic Health Centers (AHCs) hold a unique place in today’s health care environment. They service their communities through a tripartite mission of education, research, and provision of complex care to disadvantaged populations. To achieve this mission, AHCs face challenges in funding and cost containment compared to non-AHCs. Additionally, the implementation of government programs like the Hospital Readmission Rate Program (HRRP) and Value Based Purchasing (VBP) have the potential to affect AHCs differently from non-AHCs. While AHC’s unique features are known and there has been research to date on HRRP and VBP, literature has yet to statistically explore the financial differences between AHCs and non-AHCs and how HRRP and VBP may have differentially affected AHCs compared to non-AHCs. The objectives of this study are to explore financial differences between AHCs and non-AHCs and the impact that HRRP and VBP has had on these two types of organizations through the use of a contingency theory framework. Contingency theory is an organizational theory that seeks to explain variations in organizational performance over time by studying internal and external environmental influences. Guided by Contingency Theory, the study used a non-randomized, quasi-experimental, retrospective study design to evaluate two hypotheses. The study sample consisted of a total of 10,157 (991 AHCs) US non-rural hospital years from 2011 through 2015. The study used operating margin and total margin as the key measures of hospital financial performance for the dependent variables. HRRP and VBP were combined into a single independent variable along with hospital type differentiating AHCs from non-AHCs. Covariates of Herfindahl-Hirschman Index, Medicaid expansion, health system affiliation, and ownership structure were used to control for other environmental influences. A repeated measure analysis of variance was employed to test the difference between the two hospital groups in isolation of HRRP, VBP, and covariates and a repeated measure analysis of variance with covariance was used to test the full model, which incorporated HRRP, VBP, and covariates. The results of the analysis support the significance of HRRP and VBP on hospital operating margin, but the results did not support a differential effect of these programs on AHCs as compared to non-AHCs. While the results did not support the two main hypotheses, it did provide valuable insight into the financial differences between AHCs and non-AHCs and the importance of VBP and HRRP on hospital financial performance. The results also provide important policy implications and thoughts on potential managerial actions given the HRRP and VBP programs.

Background: The 30-day readmission rate is considered a quality of care measure for providers and has become important because providers might face reduced reimbursement from any increase in unplanned readmissions Objective: The aim of the first chapter is to investigate the waiting-length of stay (WLOS) and post-length of stay (PLOS) on the 30-day readmission. In the second chapter, we examined the hospital procedural volume and hospital quality on the 30-day readmission. Our objective in the third chapter is to examine the zip code-level SES factors on the 30-day readmission rates. Participants: patients undergoing isolated coronary artery bypass grafting (CABG) in Virginia Methods: A retrospective study design has been conducted using a multi-level logistic model of increasing complexity for all three chapters. The sample used was from the Virginia Cardiac Surgery Quality Initiative (VCSQI) of the periods 2008-2014, the dataset included patient characteristics. Afterward, we merged the sample with both the Virginia Health Information (VHI) to obtain hospital characteristics (ownership, teaching status, and location), and Agency for Healthcare Research and Quality (AHRF) to obtain county-socio-economic status (SES) characteristics (education, employment, and median household income), the previous SES was used for chapter’s one and two. In chapter three, instead of AHRF, we merged the sample with the American Community Survey (ACS) to obtain zip code-SES characteristics (employment, median household income, education, median house price). The main outcome was the 30-day readmission rate. The analytical sample of chapter one n = 22,097, in chapter two the sample n = 25,531, while in chapter three the sample n= 25,829. We conducted a sensitivity analysis in all three chapters. In chapter one we analyzed the data at the patient level, in chapter two we analyzed the data at the hospital level, while in chapter three we conducted the analysis at the area zip code level. Results: In chapter one, we found that readmitted patients after a prolonged PLOS had increased odds of readmission, by 68.7%, compared to readmitted patients with a shorter PLOS in the fully adjusted model; while, WLOS was not significant at the P < 0.05. In chapter two, the fully adjusted model displayed significant results with a reduced odds in readmissions by 22.8% in the middle-volume hospitals compared to the low-volume hospitals, while the middle-quality hospitals had increased odds of readmission by 23.5% compared to the low-quality hospitals. In chapter three, statistically, we did not find that area zip code-SES had an effect on the 30-day readmission rate. While, geographically, we found that addresses of individuals were clustered in certain areas of Virginia. Conclusion: In chapter one, patients undergoing CABG and experience a prolonged PLOS of > 6 days are at risk to be readmitted within 30-days of the procedure. In chapter two, the higher volume hospitals (middle-volume) compared to low-volume hospitals showed a significant reduction in odds in the 30-day readmissions, especially after adjusting the model with hospital quality. In chapter three, even though, there was no association of area-SES with 30-day readmission, in the maps, we found a cluster of patient addresses in the southern parts of Virginia with an increased readmission, which is considered underprivileged area; and the fact might be due to the proximity of these areas to cardiovascular hospitals. Policy Implication: In chapter one, the study provided a model for clinicians to stratify patients at risk of readmission, especially patients with risks of staying longer in the hospital after CABG. In chapter two, policymakers and the CMS should find new ways to help hospitals with low-volumes to reduce their isolated-CABG readmission rates and be able to compete with high-volume hospitals. In chapter three, no significant correlation between area-SES and readmission for patients who underwent CABG was found; these backs prior notion that SES should not be adjusted for the reimbursement penalties of the Hospital Readmission Reductions Program (HRRP) on hospitals

Objectif : Estimer la contribution statistique des caractéristiques des hôpitaux et des caractéristiques liés aux patients sur la probabilité de réadmission des patients qui ont subi une PCNL, une procédure endoscopique à haut risque de morbidité, dans les hôpitaux aux États-Unis en 2014 et évaluer les prédicteurs des taux de réadmissions d’une PCNL. Méthode : Nous avons identifié tous les patients qui ont subi une PCNL dans les hôpitaux aux États-Unis en 2014 (janvier-novembre) en utilisant la banque de données nationale de réadmission (NRD). L’issue d’intérêt était une réadmission non planifiée 30 jours après une PCNL. À l’aide d’un modèle multi-niveaux à effets mixtes, nous avons estimé l’association statistique entre les caractéristiques hospitalières ainsi que les caractéristiques individuelles liés aux patients sur la probabilité de réadmission. Un effet aléatoire associée à l'hôpital a été utilisé pour estimer le taux de réadmission au niveau hospitalier. Un pseudo R-carré a été calculé pour évaluer la contribution de chaque catégorie de variables sur les taux de réadmission. Résultats : Notre échantillon pondérée était constitué de 6 974 personnes ayant subi une PCNL dans 485 hôpitaux aux États-Unis en 2014. Le taux de réadmission à 30 jours était de 8,5 % (IC à 95 % 7,4 – 9,7). Après ajustement, les caractéristiques hospitalières n’étaient pas associées à une probabilité accrue de réadmission. Le sexe féminin était associé à une diminution de la probabilité de réadmission (IC à 95% 0.54 – 0.93). Les hôpitaux individuelles n’ont contribué qu’à une infime partie à la probabilité d’être réadmis de leurs patients. Les caractéristiques liés aux patients expliquaient davantage la variabilité dans la probabilité de réadmission que les caractéristiques hospitalières (pseudo-R2 9.50% vs 0.03%). Conclusion : Le risque d’être réadmis après une PCNL varie énormément entre les hôpitaux. Une fraction minime de cette variabilité peut être expliqué par les caractéristiques hospitalières contrairement aux caractéristiques des patients. Ces résultats soulignent les limites potentielles de l’utilisation des réadmissions comme mesure de la qualité des soins. / Objective: Estimate the relative contribution of hospital and patient factors to readmission after a typical high-risk endoscopic procedure, percutaneous nephrolithotomy (PCNL). Methods: We utilized the Nationwide Readmission Database to identify the patients who underwent PCNL in the United States hospitals in 2014 (January-November). The main outcome was unplanned 30-day readmission following a PCNL. Using a multilevel mixed-effects model, we estimated the statistical association between patient and hospital characteristics and readmission. A hospital-level random effects term was added to estimate hospital-level readmission. To assess the relative contribution of each group of variables on readmission rates, a pseudo-R2 was calculated to assess the contribution of hospital effects to the model of readmission. Results: We identified a weighted sample of 6,974 individuals who underwent PCNL at 485 hospitals in the United States in 2014. The 30-day readmission rate was 8.5% (95% CI 7.4 – 9.7). In our adjusted model, hospital characteristics were not associated with increased likelihood of readmission. Female sex was the only characteristic associated with decreased likelihood of readmission (95% CI 0.54 – 0.93). Individual hospitals contributed marginally to their patients probability of readmission. Patient level characteristics explained far more of the variability in readmissions than hospital characteristics (pseudo-R2 9.50% vs 0.03%). Conclusion: The risk of readmission after a PCNL is highly variable in between hospitals. The statistical contribution of individual hospitals and hospital characteristics to the probability of readmission following a PCNL was minimal compare to patient characteristics. These findings underscore the potential limitations of using 30-day post-discharge readmissions as a hospital-level quality metric.