Abstract
This study examines whether hospitals without physician participation on their boards of directors deliver lower quality of care. Using data from California nonprofit hospitals from 2004 to 2008, the authors document that the absence of physicians on the board is associated with a decrease of 3 to 5 percentage points in 3 of 4 measures of care quality. This result was obtained using regression analysis, which controls for various hospital characteristics. The authors also identify factors that influence quality of care in hospitals. Specifically, hospital size, church affiliation, urban location, and system affiliation are positively associated with quality of care; proportion of Medicaid patient revenue and poverty level of the county in which the hospital is located are negatively associated with quality of care. These results highlight the importance of physician participation in hospital governance and indicate areas for hospitals and policy makers to focus on to enhance medical quality management.
Hospital boards usually have representation from a wide range of constituents, such as community leaders, business professionals, physicians, and government officials.1-3 Directors work together to fulfill various responsibilities, including fiduciary, monitoring, fund-raising, and communicating stakeholders’ needs to management.2,4 Because hospitals exist to provide health care for patients, overseeing quality of care is among the board’s most important responsibilities.5,6 The board can have a positive influence on quality of care by making medical quality management a top priority, facilitating the development of a supportive culture for quality improvement, and assisting the promotion of quality value and quality improvement techniques.6-8
Corporate governance literature suggests that board composition has important implications for organizational outcomes. For example, the presence of commercial bankers on the board improves a firm’s ability to access debt markets. 9 Also, firms that have directors with backgrounds in law and politics are more likely to succeed in influencing government decisions. 10 In this study, the authors argue that, because of physicians’ clinical expertise, the presence of physicians on the board is likely to increase the board’s emphasis on quality of care, enhance the effectiveness of oversight, and thus improve quality of care. Consequently, hospitals without physicians on the board are likely to have their quality of care compromised. The purpose of this study, therefore, is to examine whether the lack of physician participation on hospital boards makes a difference to quality of care.
This study is particularly timely as the federal government is imposing stringent regulatory requirements to strengthen board governance. Since 2008, the Internal Revenue Service has been requiring nonprofit organizations to disclose detailed information on board composition and compensation, indicating its recognition that the quality of board oversight determines whether organizational missions can be carried out effectively and efficiently.11,12 In an effort to enhance the quality of board monitoring of financial reporting, since 2003 the US Securities and Exchange Commission (SEC) has been requiring publicly traded companies to disclose whether they have at least one “audit committee financial expert” on their boards and to provide the reason if a company has no such expert, under the assumption that these professionals better understand internal control processes related to financial reporting.13,14 In the same spirit, because physicians thoroughly understand the process of providing health care, can we assume that hospitals without physician directors deliver lower quality of care as compared with hospitals that have physician directors?
Because of the scarcity of available data on quality of care and board composition, only a few empirical studies have investigated this question, and all of them used survey methods to obtain part or all of the data for statistical analysis. Weiner and colleagues obtained data from a 1989 national survey on hospital governance and a 1993 national survey on hospital quality improvement efforts.15,16 They find that physician participation on hospital boards increases the likelihood of adopting continuous quality improvement/total quality management (CQI/TQM) as well as the degree of clinical involvement in CQI/TQM. Jiang et al analyzed data from a 2006 survey by The Governance Institute and concluded that physician involvement in a board’s quality committee is associated with higher quality of care. 17 Moreover, Carman and colleagues conducted interviews with representatives of 10 hospitals and found that physician participation in quality improvement teams, which are independent of boards of directors, is positively associated with quality of care. 18 Weiner and colleagues used a 1997 national survey for hospital quality improvement practices and found similar results.19,20
The recent availability of hospital quality of care data from the Hospital Quality Alliance (HQA) provides an ideal opportunity to examine the impact of lack of physician participation on hospital boards on quality of care. The HQA is the largest and most comprehensive quality of care database available. 21 It is also the primary quality assessment program used by the Centers for Medicare and Medicaid Services and The Joint Commission, a nonprofit organization that accredits and certifies US health care organizations. 5 The authors merge this data set with California nonprofit hospitals’ disclosure reports, obtained from the Office of Statewide Health Planning and Development (OSHPD). California is the only state, to the authors’ knowledge, that makes data publicly available on individual board members’ occupations for all hospitals operating in the state. The present study, relying on these archival data sources, offers 2 advantages over previous survey-based studies. First, archival data mitigate nonresponse bias that could hamper the validity of survey-based studies. Second, the large numbers of variables available from the archival data make it possible to use regression analysis, which controls for various potential confounding factors and thus increases the interpretability and robustness of the result.
Methods
Quality of Care in Hospitals
To measure hospital quality of care, the authors use process of care data released by the HQA, available through Hospital Compare, a US Department of Health and Human Services database. Although hospitals are not mandated to report quality performance to the HQA, almost all of them do because the reported data are linked to Medicare and Medicaid financial incentives they may receive.5,22
The HQA database has 3 components: process of care quality, mortality, and patient satisfaction. This study focuses on process of care quality for 2 reasons. First, as opposed to mortality, it has greater variation among hospitals and is less likely to be influenced by factors outside the hospital’s control. 23 Second, it is a more objective measure as compared with patient satisfaction data.
Hospitals report process of care quality as a set of scores, with each score indicating the percentage of patients for which the hospital fulfilled a desirable clinical action; for example, to give patients who have pneumonia an oxygenation assessment. 22 The lowest possible score is 0%, reflecting a complete failure to follow the indicated action. The highest possible measure is 100%, reflecting full compliance. Eighteen such actions were in use consistently between 2004 and 2008, covering 4 categories: heart attack, heart failure, pneumonia, and surgery infection prevention. Details of the 18 actions are available in Supplemental Table A1 (available online at http://ajmq.sagepub.com/content/by/supplemental-data). Following previous literature, equal weight is given to individual scores within a category in order to obtain a quality measure for that category. 22 In other words, a hospital’s quality measure for a certain category is the average of the hospital’s scores for all actions related to that category. For example, a quality measure of 85% for heart attack means that the hospital carried out the indicated actions for 85% of patients who had heart attack.
Physician Participation on Hospital Boards
The data regarding physician participation on hospital boards are obtained from the OSHPD. The authors classify a board director as a physician if the director’s occupation is stated as physician, MD, or any other associated medical specialty (eg, internist, pediatrician, surgeon, radiologist, cardiologist, anesthesiologist). A hospital is considered as having no physician on its board if none of its board directors is listed as a physician, MD, or any other associated medical specialty. The authors merge data from OSHPD and data from HQA by using each hospital’s National Provider Identifier. The study sample covers a 5-year period between 2004 and 2008.
Statistical Analysis
The authors begin by examining the time trend of quality of care and physicians’ board participation. Next, the sample is split into 2 groups based on whether a hospital has a physician(s) on its board and compare the average quality of care between the 2 groups. However, the result from this step might not be conclusive. Because hospitals are not randomly assigned to the 2 groups, the between-group difference in quality, if any, can be attributed to factors other than physician participation in governance. For example, if most small hospitals do not have physicians on their boards and also deliver lower quality of care, then it is difficult to determine which factor (small size or lack of physician participation) leads to lower quality of care.
To overcome this difficulty, the authors use a regression model, Equation 1, that includes a set of control variables capturing various hospital characteristics that might potentially influence quality of care. By doing so, the authors can interpret the estimated coefficient for the key predictor variable (ie, the absence of physician directors) as the effect of the absence of physician directors on hospital quality of care when other characteristics stay unchanged. 24 The hospital characteristics controlled for include (1) board size as measured by the number of directors, (2) whether the hospital chief executive officer (CEO) is serving on the board, (3) hospital size as measured by the number of staffed beds, (4) number of residents, (5) church affiliation, (6) urban location, (7) system affiliation, (8) average length of stay (ALOS), (9) proportion of Medicare patient revenue among total revenue, (10) proportion of Medicaid patient revenue among total revenue, (11) case mix index (CMI), (12) profit ratio as measured by net income divided by total operating expenses, and (13) percentage of population below poverty line in the county where the hospital is located. Data about these control variables were obtained from the OSHPD and the US Census Bureau.
The authors also include indicator variables (ie, dummy variables) for each year to control for macroeconomic shocks that might influence all hospitals but vary from year to year. Because quality of care is measured as a continuous variable ranging between 0 and 1, the authors use the ordinary least squares (OLS) method to estimate the regression model.
Results
Physician Participation on Hospital Boards
As shown in Figure 1, between 2004 and 2008 the proportion of hospitals in the sample that have no physicians on their boards ranges from 6% to 8%. During this period, quality of care improved for all categories. The quality measure for surgery infection prevention experienced the most significant improvement, increasing from 58% in 2004 to 89% in 2008 (P < .01). The variation across categories also was reduced. In 2008, quality measures for all categories are within 7 percentage points of one another.
Time trend of quality of care and physician participation on boards.
In Figure 2, the authors compare the average quality of care between hospitals with physician directors and hospitals without physician directors. For all categories, hospitals without physician directors deliver lower quality of care. This between-group difference is statistically significant for the pneumonia and surgery infection prevention categories (P < .05 and P < .01, respectively). As mentioned earlier, this difference might be driven by variations in hospital characteristics. To address this concern, the OLS method is used, which controls for various hospital characteristics, to estimate Equation 1. Table 1 presents descriptive statistics by category for the variables used in the regression analysis. The number of observations (ie, N) varies by category because some hospitals do not have quality data available for all categories. The correlation coefficients among predictor variables are shown in Supplemental Table A2 (available online at http://ajmq.sagepub.com/content/by/supplemental-data).
Comparison of quality of care between 2 groups of hospitals.
Descriptive Statistics.
Abbreviations: ALOS, average length of stay; CEO, chief executive officer; CMI, case mix index.
W/O physician on board is defined as an indicator variable, which takes the value of one if the hospital has at least one physician on the board, and zero if otherwise.
# of physicians on board is the total number of physicians on the board.
Sample sizes vary across categories, because some hospitals do not have quality data available for all categories.
Table 2 documents the OLS regression result. Except for heart attack, the relationship between lack of a physician on the board and lower quality of care is statistically significant. The estimated effects for W/O physician on board are 4.3 percentage points lower in quality measure for heart failure, 3.1 percentage points lower in quality measure for pneumonia, and 5.6 percentage points lower in quality measure for surgery infection prevention. It is worth noting that the result is obtained after controlling for board size, CEO serving on the board, hospital size, teaching status, church affiliation, urban location, system affiliation, ALOS, proportions of Medicare and Medicaid patient revenue, CMI, profit ratio, and poverty level of the county in which the hospital is located. In other words, these control variables are assumed to remain constant when interpreting the estimated coefficient of W/O physician on board.
OLS Regression Estimation Result.
Quality of Careit = α a + β1W/O physician on boardit + β2Board sizeit + β3CEO as directorit + β4Hospital sizeit + β5Teachingit + β6Church affiliationit + β7Urban locationit + β8System affiliationit + β9ALOSit + β10Medicare proportionit + β11Medicaid proportionit + β12CMIit + β13Profit ratioit + β14Poverty levelit + ∑γtYeart + ε
Abbreviations: ALOS, average length of stay; CEO, chief executive officer; CMI, case mix index; OLS, ordinary least squares.
Estimation results for α and year dummies are not reported.
Standard errors, based on 2-tailed test, are in parentheses. *P < .05; **P < .01.
Sample sizes vary across categories because some hospitals do not have quality data available for all categories.
Other Factors
Table 2 also indicates estimated results for control variables. First, consistent with previous literature, hospital size, urban location, and system affiliation are associated with higher quality of care.25,26 This suggests that large hospitals, urban hospitals, and hospitals that are members of a health care system are likely to provide higher quality of care. Second, all else being equal, church-affiliated hospitals provide higher quality of care. Third, all else being equal, hospitals located in poorer counties and that have a higher proportion of Medicaid patients have lower quality of care. This finding echoes that of Asch et al that patients with higher annual household incomes experience higher quality of care. 27 Figure 3 summarizes these results.
Summary of estimation results.
Supplementary Analysis
The above analysis compares hospitals having no physician director with hospitals having a physician director. Next, the authors compare 3 groups of hospitals—those with no physician director, those with 1 director, and those with more than 1 director—by changing the key predictor variable in Equation 1. The authors replace W/O physician on board, an indicator variable (ie, dummy variable), with Extent of physician on board, a categorical variable that takes the value of 0, 1, and 2, if a hospital has none, 1, or more than 1 physician director, respectively. The estimated coefficient for Extent of physician on board captures the effect of the extent of physician participation in board governance on quality of care. The estimation result (abbreviated), as reported in Supplemental Table A3, Panel A (available online at http://ajmq.sagepub.com/content/by/supplemental-data), is consistent with the result reported in Table 2. Except for heart attack, in which no statistically significant result is found, greater physician participation is associated with higher quality of care. Specifically, 2 and more physicians on the board are better than only 1 physician on the board; 1 physician on the board is better than no physician on the board.
The authors also test whether the proportion of physicians on the board is associated with quality of care. The authors create a continuous variable, Proportion of physicians on board, measured as the number of physician directors divided by the board size, and use it as the predictor variable in the regression model. The estimation result is reported in Supplemental Table A3, Panel B (available online at http://ajmq.sagepub.com/content/by/supplemental-data). Only for heart attack does the proportion of physicians on the board have a statistically significant but small effect on quality of care. Specifically, a 10% increase in physician representation on the board (eg, physician representation on the board increases from 5% to 15%) leads to a 0.5% increase in quality measure for heart attack. This result, together with that reported in Supplemental Table A3, Panel A, suggests that the effect of physician participation on the board might be diminishing; that is, once there are a few physicians on the board, more physicians do not necessarily bring higher quality of care. One possible explanation is that physician directors, as a group, might be inclined to have similar opinions on how much emphasis the board should place on improving and overseeing quality of care.
Discussion
In today’s rapidly evolving health care environment, hospitals face increasing pressure to improve quality of care. 28 The board of directors is legally accountable for hospital quality of care and plays a critical role in overseeing and improving quality of care.5,29 This study investigates whether hospitals deliver lower quality of care if no physician serves on the board. Because of their expertise in medicine and their clinical training and experience, physicians thoroughly understand the importance of quality and the process of collecting and using quality data. Their involvement in governance, therefore, increases the board’s emphasis on care quality and its effectiveness in overseeing quality, and thus adds important value to the hospital.
After controlling for various factors that could potentially influence quality of care, statistical analysis shows that the lack of physician representation on hospital boards is associated with lower quality of care, evidenced by a 3 to 5 percentage point decrease in quality of care measures across 3 of 4 categories. In other words, physicians as directors add important value to hospital quality of care. In addition, the authors find that hospital size, church affiliation, urban location, and system affiliation have a positive effect on quality of care, while the proportion of Medicaid patient revenue and the poverty level of the community in which the hospital is located are associated with lower quality of care.
The statistical analysis of this study is based on data between 2004 and 2008. Since then the landscape of hospital governance has experienced changes; for example, more physicians participated on boards and more hospitals became affiliated with health care systems. These changes are not reflected in the study sample and thus might limit the relevance and external validity of this study. To address this concern, the authors examined recent surveys to understand the magnitude of these changes. According to the biennial surveys conducted by The Governance Institute, the average numbers of voting directors who are medical staff physicians were 2.2 in 2007, 2.1 in 2009, and 2.3 in 2011; the median numbers of voting directors who are medical staff physicians were 2 in 2007, 2 in 2009, and 1 in 2011; and the proportions of voting directors who are medical staff physicians are 16.5% in 2007, 16.0% in 2009, and 17.3% in 2011.30,31 The slight increase in mean accompanied by the slight decrease in median indicates that since 2007, some hospitals might have a greater number of medical staff physician directors, but the proportion of hospitals without medical staff physician directors might have experienced little change. It is worth noting that these statistics only refer to physicians who are members of the medical staff. The statistics pertaining to all physician directors, not available from the surveys, must be higher than the aforementioned medical-staff-physician statistics. However, the authors do not have reason to assume that the change in the all-physician statistics shows a different pattern than the change in the medical staff physician statistics. In addition, 28.5% of hospitals surveyed in 2009 reported that they do not have any physician board members who are active members of the medical staff but are not employed by the hospital, and 28.8% of hospitals surveyed in 2011 reported the same, showing little change as compared with the 2009 data.30,31 Admittedly, the surveys do not have aggregated information for physician board participation in general and collect data from subsamples of US hospitals. These results, nevertheless, suggest that physician board representation might not have changed significantly since 2007.
In recent years, many hospitals have been integrated into various forms of health care systems, a trend that has a profound effect on centralization of operation decisions and governance practices. According to the surveys conducted by the American Hospital Association (AHA), the proportions of US community hospitals that are affiliated with a health care system were 57.2% in 2008, 58.3% in 2009, 59.0% in 2010, and 60.5% in 2011, demonstrating a continuously increasing trend.32-35 The proportion of system-affiliated hospitals in the present study sample is 57% to 58% (as shown in Table 1), comparable to the figures reported by AHA surveys. Recognizing a health care system’s potential influence on board composition and quality management, the authors include system affiliation as a control variable in the statistical model. System affiliation takes the value of 1 if a hospital is in a system and 0 otherwise. The estimation result, obtained after controlling for any potential confounding effect of system affiliation, indicates a negative impact for lack of physician directors on quality of care (as shown in Table 2). In other words, holding system affiliation constant, the absence of physician directors has a negative effect on quality of care. The statistical analysis also suggests that system affiliation per se has a positive influence on care quality, as evidenced by 3 positive estimated coefficients for system affiliation (as shown in Table 2). Future studies on whether hospitals within the same system have similar board composition and whether they provide consistent quality of care are warranted to shed light on the extent to which system affiliation influences hospital quality management and governance.
This study has several additional limitations. First, because of the limited availability of data regarding the occupation of individual board directors, this study is confined to only one state, California. Second, the HQA program collects information only on major medical conditions, which might not accurately reflect overall hospital quality of care. Also, the quality of care data are self-reported by hospitals, and thus might suffer from data measurement errors and/or data manipulation. Finally, although various control variables have been included in the statistical model, the authors cannot be completely certain that no important factors have been omitted from the model.
This study has important implications for board selection strategies for the purpose of enhancing hospital quality of care. It also has the potential to inform policy with regard to medical quality management, supporting the suggestion by Jha and Epstein that policy makers who hope to improve care in US hospitals should target the boards of directors for intervention. 5 As the SEC encourages publicly traded companies to have at least one financial expert on their boards to ensure the quality of financial reporting, perhaps policy makers should suggest that hospitals have at least one physician serve on their boards to ensure quality of care.
Footnotes
Acknowledgements
The authors thank David Nash, MD, MBA (editor), and 2 anonymous reviewers for their valuable comments; Sylvia Hsu, PhD, CGA (York University), for her suggestions; John Herringer, RN, MS (The Joint Commission), for sharing his insight on hospital boards’ composition; and Sara Beazley, MA (American Hospital Association), and Kathryn Croom Peisert (The Governance Institute) for providing survey results.
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The authors received no financial support for the research, authorship, and/or publication of this article.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
