Characteristics of Top-Performing Hospitals Caring for High-Need Medicare Beneficiaries

Abstract

A small proportion of high-need (HN) Medicare beneficiaries account for a large share of medical expenditures in the United States. Identifying hospitals with the best outcomes for HN patients is central to identifying and spreading evidence-based practices to improve care for this population. The objective of this study was to identify and characterize top-performing hospitals for HN patients. Administrative claims data from 2013–2014 were used to identify HN beneficiaries and their treating hospital; hospitals were ranked based on their HN beneficiaries' outcomes in 2015. Hospitalization, mortality, and days spent in community were assessed, and all outcomes were risk standardized for age, sex, dual eligibility, and hospital referral region. American Hospital Association and aggregated inpatient claims data characterized hospitals. Logistic regression models estimated the odds of ranking in the top 20% on all outcomes. Of 2253 hospitals with at least 500 HN patients in the United States, 92 (4.1%) ranked in the top 20% across all outcomes. No hospital characteristics were associated with being top performing across all outcomes, but urban hospitals were significantly less likely to perform well on hospitalization and private, for-profit hospitals performed better on mortality. Small hospitals, Accountable Care Organization providers, and those providing palliative care services were more likely to rank highly on days spent in the community. Top-performing hospitals served fewer minority, dual eligible, and HN patients, suggesting that case mix may explain some of the differences in performance, and that additional work is needed to examine programs and practices at outstanding hospitals.

Introduction

A small percentage of patients account for a significant percentage of health care spending. A recent study estimates that 10% of patients account for more than 65% of spending and as few as 1% of patients account for more than 20% of spending.¹ Focusing on “high-utilizers” is emerging as a core strategy to reduce cost in the health care system, especially as it relates to better performance on population health and other value-based payment reforms such as accountable care organizations.^2,3 Yet, the question of what programs and hospital characteristics are related to the provision of the best quality of care for complex, high-need patients remains a debated issue.⁴ It is unclear how hospital performance compares between the overall population and subpopulations of patients who have complex chronic conditions, multimorbidity, functional impairment, or indicators of frailty. The importance of this question is magnified by value-based payment reforms and incentives that are being proposed and which may shape provider accountability for seriously ill patients in the future.

This work examines the characteristics of hospitals that are performing best for the high-need Medicare beneficiaries under their care, defined as having complex or multimorbid chronic conditions and health care utilization in a given year, or functional impairment in activities of daily living as assessed during post-acute care. This study examines select outcomes experienced by high-need beneficiaries to determine which hospitals perform best for this population, and to determine what, if any, hospital characteristics are associated with high performance. The ultimate goal of this work is to inform ongoing quality improvement interventions and reforms for this population of high-utilizers.

Background

The National Academy of Medicine (NAM) recently published a study entitled, Effective Care for High-Need Patients: Opportunities for Improving Outcomes, Value and Health.⁵ The report reiterates the finding that a small proportion of beneficiaries account for a large part of health spending.¹ Although some of these patients are nearing the final 6–12 months of life or are high cost because of a single episode, such as acute myocardial infarction or hip fracture (ie, “transient high-cost”), a recent study estimates that 28% of high-cost patients are “persistently” high cost,⁶ and maintain their high-cost status from year to year. Patients who are persistently high need often suffer from fragmented and lower quality care. Many assert that health care spending cannot be reduced without addressing the complex needs of this population.² However, the NAM report suggests a taxonomy of high-need patients consisting of 6 categories, each requiring a “customized strategy based on care requirements,”^5,6 making it difficult for hospital leaders to identify the most effective and highest impact solutions.

It is generally unknown which hospital characteristics are clearly associated with strong performance. In a systematic review, Brand at al concluded that the use of computerized physician order entry systems was the hospital characteristic most strongly related to performance, with other studies demonstrating a positive effect of hospital volume, nursing workforce design, and leadership on performance.⁴ More recently, Al-Amin et al examined the characteristics associated with a sustained high performance on Hospital Consumer Assessment of Healthcare Providers and Systems measures.⁷ They concluded that teaching hospitals, hospitals operating in markets with strong county-level competition, hospitals with better nursing staffing ratios, and hospitals with a lower ratio of Medicare beneficiaries were more likely to maintain high patient ratings over 5 years.⁷

Despite our understanding of some of the factors associated with overall hospital performance, there is a particular knowledge gap regarding the experiences of high-need beneficiaries, and the specific hospital characteristics and programs that predict top hospital performance for this patient population. Anderson et al focused on attributes that appeared to set apart programs that successfully improved at least 1 outcome among spending, satisfaction, and clinical outcomes for patients with disability or chronic conditions.⁸ They found that appropriate identification of high-risk patients and leadership were important attributes, as were program size, adaptability to local conditions, directly promoting interaction between patients and families, and focusing on care transitions.⁸ In the primary care setting, studies have shown that sustained participation in pay-for-value programs is associated with better outcomes among patients with ≥2 chronic conditions, but it is unclear whether these findings would extend to sicker patients receiving care at hospitals.⁹ With such a wide spectrum of high-need patient types, it is unclear if there are specific hospital characteristics associated with superior performance related to the care of such patients.

The objectives of this study were to identify top-performing hospitals caring for high-need Medicare beneficiaries, as well as to determine the characteristics associated with the likelihood of being a top-performing hospital for this subpopulation.

Methods

This study was considered exempt from approval by the Institutional Review Board at Brown University on the basis that it relies on analysis of secondary administrative health care data that are not collected for research purposes.

Study design

The research team conducted a retrospective analysis of national administrative data between 2013 and 2015 at both patient and hospital provider levels. High-need beneficiaries were defined as those having complex chronic conditions or multimorbidity and having used acute or post-acute health services in 2014.¹⁰ Any patients with ≥2 diagnoses indicative of frailty or with complete dependency in mobility or activities of daily living as indicated in post-acute care assessments also were considered high need. High-need beneficiaries who survived until the end of 2014 were then attributed to a hospital based on their last hospitalization in Medicare Provider Analysis and Review (MedPAR) inpatient claims in 2013–2014. On the basis of this attribution, the team finally aggregated high-need beneficiaries' 2015 outcomes to the hospital level, regardless of the actual hospital to which they were subsequently admitted. Only general acute care hospitals in the contiguous United States, caring for ≥500 high-need beneficiaries in 2015 were retained in the analysis in order to obtain stable estimates and given a focus on potentially actionable policies for large segments of the patient population. Approximately 20% of US hospitals did not meet this eligibility criterion about volume; therefore, smaller hospitals were generally excluded from the study. The team also excluded surgical hospitals where more than 75% of admissions were for surgical or orthopedic procedures because these hospitals serve different patient populations and are not directly comparable.

Data sources and variables

The following patient-level outcomes were assessed for high-need beneficiaries: hospitalization, mortality, and percentage of days alive spent in the community. For each day alive, beneficiaries were considered in the community if they were not in a hospital or nursing home. Outcome variables were obtained using the 2015 Medicare beneficiary summary file, as well as tracking inpatient days and nursing home days from MedPAR inpatient files and mandatory nursing home Minimum Data Set assessments throughout the year. These findings also were compared with the likelihood of hospitals obtaining a high star rating (4 or 5) on the Hospital Star Ratings during the same year, for their entire Medicare population.

A total of 6,465,948 high-need Medicare beneficiaries were identified in 2014 following the research team's previous work,¹⁰ and 78.66% of them were attributed to a hospital based on their last 2013–2014 hospitalization. Among unattributed beneficiaries, 90.45% had received care in a nursing home or from a home health agency in 2014, and 42.44% had a diagnosis of Alzheimer's disease and related dementias. These results confirm a different profile for unattributed high-need beneficiaries given that they receive care primarily in nursing homes or through home health agencies. In terms of the stability of hospital attribution over time, among the 20% high-need beneficiaries who were hospitalized in 2015, 65% were admitted to the same hospital in which they were hospitalized last during the previous 2 years.

Hospital characteristics were obtained from the 2013 American Hospital Association (AHA) survey, the 2014 Provider of Service Files for general acute care hospitals, and aggregated 2015 inpatient claims. Characteristics of interest targeted specific organizational domains, including hospital size, staffing, hospital type, medical school affiliation, Accountable Care Organization (ACO) membership, provision of palliative care services, and participation (any/number) in Model 2 Bundled Payments for Care Improvement (BPCI). Bed count was coded as a categorical variable according to its distribution (0–199, 200–399, ≥400). Also included were population density to compare rural and urban hospitals, a ratio of full-time registered nurse per bed, and type of hospital (ie, private for-profit, private non-for-profit, governmental, other). Centralization also was examined using a 4-response categorical variable, developed by Bazzoli et al, ranging from highly centralized to decentralized.¹¹ Finally, participation in bundled payments was obtained from the Model 2 participation list available on the Centers for Medicare & Medicaid Services (CMS) BPCI website (https://innovation.cms.gov/initiatives/BPCI-Model-2) and cross-checked with the CMS Center for Medicare and Medicaid Innovation demonstration participant list (as of May 2017) provided on the data.cms.gov website.

AHA survey data suffer from significant missing data patterns that were addressed before estimating the regression models. A total of 13 hospitals were excluded from analyses because they were missing all AHA survey values. In addition, 603 (26.61%) hospitals did not have a value for the cluster variable,¹¹ and therefore they were grouped with the 30 that were flagged as having insufficient data to determine this health systems characteristic. Moreover, 293 hospitals did not have information about whether or not palliative care services were provided at the health system level, and these were coded separately as unknown.

Statistical analyses

The research team adjusted for case mix in order to avoid penalizing hospitals with larger volumes of medically and socially vulnerable patients, which is a recurrent concern in the appraisal of value-based payment reforms.¹² Case-mix adjustments included percentage of Medicare days, percentage of dual eligible visits, percentage of high-need patients, percentage of Medicare Advantage visits, as well as the percentage of surgical and elective procedures, all aggregated at the hospital level for 2015. The 3 study outcomes also were risk standardized for age, sex, dual eligibility, and hospital referral region. Effectively, this means that the team estimated the likelihood of being high performing within a given hospital's own referral region, all previously enumerated case mix and population variables held equal. The 3 study outcomes were normally distributed and therefore ranking on these outcomes captures substantial variation in patient outcomes between hospitals. The team then estimated logistic regression models to test for an association between the characteristics of hospitals and whether the hospital ranked in the top 20% on all selected outcomes (henceforth referred to as the composite measure of top-performing hospitals).

Results

When examining hospital performance in a reduced sample of hospitals in the contiguous United States caring for a larger volume of high-need beneficiaries (n = 2253), 92 hospitals (4.1%) ranked in the top 20% across all 3 study outcomes. As shown in Table 1, the case mix of high-performing hospitals was significantly different: the mean percentage of high-need patients was lower among top-performing hospitals than other hospitals (mean 27.3% vs. 33.5%). At high-performing hospitals, a smaller percentage of admissions were for patients enrolled in Medicare Advantage, and greater numbers of elective procedures were carried out. Palliative care was offered in a higher proportion of high-performing hospitals (Table 1). As far as comparison with overall hospital performance is concerned, 88% of the 92 top-performing hospitals had ≥3 stars on the 5-star rating.

Table 1.

Descriptive Characteristics of Hospitals with ≥500 High-Need Patients, by Performance (n = 2253)

	Top performing n = 92	Other hospitals n = 2161	F/Chi²	P
Case Mix Adjustments - Mean (SD)
% Medicare Days	53.06 (10.78)	51.44 (12.18)	1.56	0.212
% Dual Eligible Visits	18.68 (8.63)	29.07 (13.12)	56.68	0.000
% High-Need Patients	27.33 (4.15)	33.47 (6.49)	80.71	0.000
% Medicare Advantage Visits	24.41 (13.22)	27.96 (15.40)	4.74	0.030
% Surgical Procedures	62.66 (16.03)	56.33 (11.60)	25.38	0.001
% Elective Procedures	34.10 (19.61)	18.38 (10.96)	166.86	0.000
Bed Count - %			6.45	0.040
1–199	57.61	44.42
200–399	29.35	36.05
≥400	13.04	19.53
Population Density - %			0.77	0.381
Rural	14.13	17.68
Urban	85.87	82.32
Staffing Ratio - Mean (SD)			2.26	0.13
Full-time RN per bed	1.41 (0.83)	1.30 (0.67)
Type of Hospital - %			5.71	0.126
Private Not-for-Profit	60.87	55.58
Private For-Profit	20.65	18.60
Governmental	4.35	12.73
Other	14.13	13.10
Centralization - %			3.35	0.341
Highly Centralized	20.65	14.95
Moderately Centralized	21.74	19.57
Decentralized	35.87	37.71
Unknown	21.74	27.76
Medical School - %			0.06	0.804
Affiliated	41.09	58.91
Nonaffiliated	42.39	57.61
ACO Membership - %			3.52	0.061
Member	21.74	14.62
Nonmember	78.26	85.38
Palliative Care (System)- %			8.03	0.018
Provided	23.91	16.52
Not Provided	57.61	71.31
Unknown	18.48	12.17
Type-2 Bundles
Any - %	16.30	13.51	0.10	0.747
Number of Bundles- Mean (SD)	3.07 (2.12)	5.09 (5.62)	1.94	0.165

ACO, Accountable Care Organization; RN, registered nurse; SD, standard deviation.

Table 2 displays the odds ratios of being a top-performing hospital for each outcome, and for being top performing on the composite top performance measure. Notably, none of the independent variables of interest were significantly associated with the odds of being a top-performing hospital across all outcomes. When examining the organizational characteristics related to each discrete outcome, no consistent pattern of relationships was found. Urban hospitals were less likely than rural hospitals to be a top-performing hospital for hospitalization. Top-performing hospitals for mortality were more likely to be private, for-profit and were less likely to be government-owned compared to private, not-for-profit facilities (Table 2). Small hospitals, those with ACO membership, and those that offered palliative care services as part of their health system had higher odds of performing well as far as the days in the community metric was concerned (Table 2). Centralization was not significantly associated with any of the other outcomes, nor were bundle participation, medical school affiliation, or staffing ratios. Comparison of these outcomes with the likelihood of obtaining a high star rating for the same year for the entire Medicare population shows that large and private-for-profit hospitals were significantly less likely to have a good star rating than smaller, not-for-profit private hospitals. Finally, moderately centralized hospitals were more likely to be high performing on the overall hospital star rating than decentralized hospitals (Table 2).

Table 2.

Adjusted Odds Ratios of Being a High-Performing Hospital on Risk Standardized Outcomes for High-Need Beneficiaries, n = 2253 (95% CI)

	Hospitalization	Mortality	Days in community	Overall top performing	Star rating 4 or 5
Bed Count
1–199	1.02 (.77–1.35)	1.26 (.95–1.66)	1.33 (1.01–1.75)	1.60 (.89–2.87)	1.08 (.85–1.38)
200–399	(ref)	(ref)	(ref)	(ref)	(ref)
≥400	.75 (.53–1.08)	1.16 (.83–1.62)	.89 (.64–1.25)	.73 (.33–1.62)	.69 (.50–.95)
Population Density
Rural	(ref)	(ref)	(ref)	(ref)	(ref)
Urban	.65 (.46–.92)	1.41 (.99–2.00)	.74 (.53–1.03)	.96 (.45–2.03)	1.26 (.92–1.73)
Staffing Ratio
Full-time RN per bed	.95 (.80–1.14)	1.07 (.90–1.26)	1.16 (.98–1.37)	1.06 (.77–1.46)	1.05 (.90–1.23)
Type of Hospital
Private Not-for-Profit	(ref)	(ref)	(ref)	(ref)	(ref)
Private For-Profit	.81 (.56–1.19)	1.63 (1.17–2.26)	1.16 (.82–1.62)	1.23 (.61–2.48)	.65 (.46–.90)
Governmental	1.26 (.86–1.83)	.50 (.33–.78)	.87 (.59–1.27)	.38 (.12–1.23)	.92 (.65–1.29)
Other	.94 (.66–1.32)	1.10 (.79–1.56)	1.21 (.87–1.69)	.92 (.45–1.89)	1.01 (.75–1.36)
Centralization
Highly Centralized	1.34 (.92–1.95)	1.13 (.78–1.63)	.90 (.63–1.31)	1.53 (.73–3.20)	1.30 (.93–1.80)
Moderately Centralized	1.18 (.84–1.65)	1.02 (.74–1.41)	.97 (.70–1.33)	1.26 (.64–2.48)	1.51 (1.13–2.02)
Decentralized	(ref)	(ref)	(ref)	(ref)	(ref)
Unknown	1.14 (.82–1.58)	.75 (.54–1.04)	.82 (.60–1.13)	.88 (.42–1.78)	1.17 (.88–1.57)
Medical School
Affiliated	1.00 (.76–1.30)	1.21 (.93–1.57)	1.22 (.94–1.58)	1.03 (.59–1.78)	.85 (.67–1.08)
Nonaffiliated	(ref)	(ref)	(ref)	(ref)	(ref)
ACO Membership
Member	1.20 (.87–1.65)	1.20 (.88–1.63)	1.86 (1.39–2.49)	1.67 (.92–3.03)	1.23 (.93–1.62)
Nonmember	(ref)	(ref)	(ref)	(ref)	(ref)
Palliative Care
Provided	.89 (.64–1.24)	1.12 (.82–1.56)	1.43 (1.04–1.95)	1.11 (.59–2.09)	1.30 (.99–1.73)
Not Provided	(ref)	(ref)	(ref)	(ref)	(ref)
Unknown	1.14 (.75–1.66)	1.36 (.96–1.94)	1.23 (.86–1.76)	1.91 (.92–3.96)	.72 (.50–1.05)
Type-2 Bundles
Any	1.06 (.68–1.64)	1.24 (.82–1.88)	1.13 (.74–1.73)	1.90 (.79–4.54)	1.14 (.78–1.66)
Number of Bundles	.95 (.89–1.03)	.96 (.90–1.02)	.95 (.89–1.02)	.88 (.72–1.07)	.98 (.93–1.03)

All models are adjusted for case mix and racial/ethnic composition. Values in bold are significant at P ≤ 0.05.

ACO, Accountable Care Organization; CI, confidence interval; RN, registered nurse.

Discussion

This work aimed to identify the organizational characteristics of top-performing hospitals caring for high-need Medicare beneficiaries. This study found that only a very small proportion of US hospitals (4.1%) caring for ≥500 high-need patients had much better rates of hospitalization, mortality, and percent days in the community, and ranked in the top 20% across these 3 risk-standardized outcomes for high-need Medicare beneficiaries under their care. High-performing hospitals care for a lower percentage of high-need patients, had higher percentages of overall surgical and elective procedures, and served a lower percentage of dual eligible and Medicare Advantage patients. These large differences in case mix suggest that there are ongoing selection issues at play, despite having risk-standardized outcomes and controlling for case-mix variables. This study also found that structural characteristics were not significantly associated with a composite measure of high performance for high-need beneficiaries. Successful hospitals appear to be idiosyncratic in the characteristics that differentiate them from other facilities. It appears that structural characteristics do not determine which hospitals have the leadership and proactive programs that make it possible for them to better meet the needs of their complex patients, and these remain difficult to measure.

Study findings also reflect the difficulty in using composite measures to identify top performance.¹³ The results resonate with previous work; for example Lehrman et al found that hospital quality of care was multidimensional and that smaller rural hospitals excelled at different metrics than large ones, such as patient experience.¹⁴ In addition to the large differences in case mix, one reason why not much difference in standard organizational characteristics is observed could be that this study lacks more intangible factors that could set these hospitals apart, such as leadership, organizational culture, or being a trendsetter.⁸ As it relates to high-need patients, it would be valuable to have access to a more thorough inventory of existing programs tailored to their health care needs, as well as additional research about how these programs are implemented, and in what health system context they appear to be most successful. Based on this literature and theory, one would hypothesize that hospitals in the top 20% may operate in smaller markets and thus establish networks to meet the local health care needs of their high-need population, both factors that, to date, are difficult to tease out in quantitative studies using publicly available hospital data. Therefore, additional research could examine the interaction between certain hospital characteristics and the markets within which they operate; in previous work such interaction was found to modify the impact of for-profit status as a factor of penetration of managed care within given markets.¹⁵

In particular, qualitative studies that capture less tangible organizational features are likely necessary to understand the different features of top-performing facilities for high-need beneficiaries. For instance, a systematic review of 19 qualitative studies examining the factors associated with high-performing hospitals identified 7 themes that are challenging to capture in quantitative research, including a positive organizational culture that could lead to uptake of innovation, as well as senior management support.¹⁶ There also is increasing consensus about the contextual factors that contribute to the success of quality improvement interventions, such as the composition and capabilities of the teams implementing them, which unfortunately are not often reported in a standardized manner across the literature.¹⁷

This study has several limitations. The research team attributed accountability for patient care to the last hospital facility in which the beneficiary was hospitalized in 2013–2014, but arguably, a different time frame could be used, and enrollment in a Medicare Advantage plan or ACO also could be considered as evidence of a different level of health system accountability. Although not all beneficiaries may be admitted to the same hospitals in 2015, as the health care system moves more toward accountable care, it is useful to understand how previous hospitals may take responsibility for future care. The team also chose to focus on hospitals because they suspect that a higher degree of heterogeneity exists among other organizational structures, such as ACOs or Medicare Advantage plans, particularly regarding the existence of potential programs and practices that can affect the care of high-need beneficiaries. Moreover, selection remains an issue that makes it difficult to draw strong comparisons between high- and low-performing hospitals. It seems that hospitals are increasingly serving niche markets and engaging in specialized procedures.¹⁸ The rapid evolution of hospital markets and the push for outpatient procedures may be changing the landscape of care provided to seriously ill patients and concurrently affecting this current appraisal of hospital performance. Standardizing the case mix across very different facilities remains a serious challenge and is a recurring concern throughout research about the effect of hospital pay-for-performance programs.^19

–22

Conclusion

Although a small percentage of US hospitals appear to excel at managing high-need Medicare beneficiaries across a range of important outcomes, there are few organizational characteristics that account for their overall superior performance. These results highlight the challenges in identifying strategies to improve care for this particular group of patients. These results also have implications for defining the accountability of health systems toward patients who are the highest health care utilizers and who have complex care needs.²³ The fact that the facilities in the top 20% care for seemingly healthier patients and that risk standardization did not eliminate these differences also raises pressing questions regarding the relationship between value-based payments and variation in health outcomes.

Footnotes

Author Disclosure Statement

Dr Mor is Chair of the Scientific Advisory Board and consultant at NaviHealth, Inc., as well as former director of PointRight, Inc., where he holds less than 1% equity. All other authors delare that they have no conflicts of interest.

Funding Information

This work was supported by a research grant from the Peterson Center on Healthcare (Grant #17021).

References

Dzau

, McClellan

, McGinnis

, et al. Vital directions for health and health care: priorities from a National Academy of Medicine initiative. JAMA, 2017; 317:1461–1470.

Blumenthal

, Chernof

, Fulmer

, Lumpkin

, Selberg

. Caring for high-need, high-cost patients—an urgent priority. N Eng J Med, 2016; 375:909–911.

Blumenthal

, Abrams

. Tailoring complex care management for high-need, high-cost patients. JAMA, 2016; 316:1657–1658.

Brand

, Barker

, Morello

, et al. A review of hospital characteristics associated with improved performance. Int J Qual Health Care, 2012; 24:483–494.

Long

, Abrams

, Milstein

, et al. Effective care for high-need patients. 2017. https://nam.edu/wp-content/uploads/2017/06/Effective-Care-for-High-Need-Patients.pdf Accessed April 8, 2019 .

Figueroa

, Zhou

, Jha

. Characteristics and spending patterns of persistently high-cost Medicare patients. Health Aff, 2019; 38:107–114.

Al-Amin

, Schiaffino

, Park

, Harman

. Sustained hospital performance on hospital consumer assessment of healthcare providers and systems survey measures: what are the determinants?. J Healthc Manag, 2018; 63:15–28.

Anderson

, Ballreich

, Bleich

, et al. Attributes common to programs that successfully treat high-need, high-cost individuals. Am J Manag Care, 2015; 21:e597–e600.

Cross

, Cohen

, Harris Lemak

, Adler-Milstein

. Sustained participation in a pay-for-value program: impact on high-need patients. Am J Manag Care, 2017; 23:e33–e40.

10.

Bélanger

, Silver

, Meyers

, et al. A retrospective study of administrative data to identify high-need Medicare beneficiaries at risk of dying and being hospitalized. J Gen Intern Med, 2019; 34:405–411.

11.

Bazzoli

, Shortell

, Dubbs

, Chan

, Kralovec

. A taxonomy of health networks and systems: bringing order out of chaos. Health Serv Res, 1999; 33:1683–1717.

12.

Bynum

, Lewis

. Value-based payments and inaccurate risk adjustment—who is harmed?. JAMA Intern Med, 2018; 178:1507–1508.

13.

Shwartz

, Restuccia

, Rosen

. Composite measures of health care provider performance: a description of approaches. Milbank Q, 2015; 93:788–825.

14.

Lehrman

, Elliott

, Goldstein

, Beckett

, Klein

, Giordano

. Characteristics of hospitals demonstrating superior performance in patient experience and clinical process measures of care. Med Care Res Rev, 2010; 67:38–55.

15.

Jiang

, Friedman

, Begun

. Sustaining and improving hospital performance: the effects of organizational and market factors. Health Care Manage Rev, 2006; 31:188–196.

16.

Taylor

, Clay-Williams

, Hogden

, Braithwaite

, Groene

. High performing hospitals: a qualitative systematic review of associated factors and practical strategies for improvement. BMC Health Serv Res, 2015; 15:244.

17.

Kringos

, Sunol

, Wagner

, et al. The influence of context on the effectiveness of hospital quality improvement strategies: a review of systematic reviews. BMC Health Serv Res, 2015; 15:277.

18.

Eastaugh

. Hospital specialization: benefits-focused product line planning. www.healthfinancejournal.com/index.php/johcf/article/view/18 Accessed October 23, 2019 .

19.

Hoyer

, Padula

, Brotman

, et al. Patterns of hospital performance on the hospital-wide 30-day readmission metric: is the playing field level?. J Gen Intern Med, 2018; 33:57–64.

20.

Doyle J Joseph

, Graves

, Gruber

. Evaluating measures of hospital quality. 2017. https://www.nber.org/papers/w23166. Accessed March 19, 2019 .

21.

Kahn CN

III

, Ault

, Potetz

, Walke

, Chambers

, Burch

. Assessing Medicare's hospital pay-for-performance programs and whether they are achieving their goals. Health Aff (Millwood), 2015; 34:1281–1288.

22.

Bhalla

. Population factors are driving hospital quality measures: what should be done?. Popul Health Manag, 2019; 22:187–190.

23.

Mor

, Rahman

, McHugh

. Accountability of hospitals for Medicare beneficiaries' postacute care discharge disposition. JAMA Intern Med, 2016; 176:119–121.