In his paper, ‘Incentivizing improvements in health care delivery’, Adam Oliver discusses recent efforts to manage the performance of health care workers in the United States and United Kingdom. He explains that various types of performance management have been deployed in response to quality and safety concerns and reviews the available evidence as to their levels of effectiveness and unintended consequences. Overall, the results of performance management programs seem to be mixed. Oliver ultimately recommends ‘public ranking with modest financial incentives’, based in part on the successes of the U.S. Hospital Quality Incentive Demonstration Project and the U.S. Veterans Health Administration (VHA).Footnote 1 Performance management, as Oliver makes clear, is the application of economic and psychological (behavioral economic) principles to the health care workplace. It is an attempt to shape the behavior of workers, including professional and non-professional staff, i.e., to induce them to do things they would not otherwise do.
It is perhaps beyond the scope of Oliver’s paper to evaluate performance management in general as a means of improving the quality of health care. His charge seems to be to review and assess its varieties. His discussion, however, hints at a more fundamental question about this approach, specifically: what are the limits of a focused instrumentality in a context as rich, fluid and collaborative as the delivery of health care? For example, Oliver refers to the opportunistic quality of performance measurement and the limits imposed by ‘target fatigue’. He raises the possibility of ‘gaming’ the system and even outright cheating, as well as the prospect that secular trends in health care account for some of performance management’s successes. In addition, what is to be made of ‘integration’, which is clearly important but poorly conceptualized? Oliver seems to suggest that these are minor challenges to a sound regime of performance management. They may, however, be indicative of more systematic limitations of this approach, which conceives of performance as behaviors and of workers as objects of management by others.
The task of performance management strategies is to construct the right incentives to change discrete behaviors in a positive direction without changing others in a negative one. For example, pay-for-performance schemes ‘have to be cleverly designed’ so as to achieve sustained improvement; workers must be incentivized in just the right way, in just the right amount and during just the right time period to effect changes on specific measures. ‘Performance’ is construed as selected health care behaviors and ‘management’ as the inducement of behavioral change among the ‘targets’ (Stone, Reference Stone2012) of a given method. In this sense, performance management views health care quality as incremental rather than global and health care workers as objects rather than subjects, participants or contributors. Whether or not performance management is effective in the mission it defines, it seems not to address the dense human context of health care delivery or even the other instrumentalities that may already be in place (Stone, Reference Stone2012).
As Oliver makes clear, performance management is predicated on the identification of workers’ economic or psychological self-interest and the ability of policy makers and others to enlist this self-interest in satisfying performance indicators of various kinds. Health care workers, including physicians, are the targets of this inducement, and any self-interested behaviors of gaming and cheating are not unexpected. Although performance management may ‘creat[e] workers … who identify with the goals of the firm’, this is no less instrumental for engineering intrinsic as opposed to extrinsic motivation. In both instances, workers are acted upon rather than engaged. Like inducements for ‘teaching to the test’ in education, performance management in health care responds to a perceived crisis that workers putatively do not recognize, do not know how to fix or do not care about fixing. This may describe some health care workers in some settings, but others have called for and instigated quality improvement practices. Oliver suggests the contribution that knowledgeable and intrinsically motivated personnel might make to improving the quality of care.
Might performance management schemes actually frustrate the efforts of conscientious health care workers? First, performance management methods center on performance indicators or quality standards that may or may not be met by health care workers. Indicators may be chosen based on criteria ranging from prevalence of disease to availability of guidelines to ease of monitoring. Many proponents prefer that they be ‘evidence based’, i.e., derived from randomized controlled trials (RCTs) or meta-analyses of these trials. Although RCTs have long been considered the ‘gold standard’ for determining effectiveness, they cannot predict effectiveness for every patient. The statistical methods employed analyze mean effectiveness across treatment and control groups, and even when an intervention is highly effective for the treatment group, it may include patients who did not improve and who may have gotten worse. Performance management, however, makes few allowances for the heterogeneity of treatment effects or, for that matter, for patient values or preferences. Although some performance indicators may represent universally beneficial practices, such as infection control, others are inappropriate to the circumstances of some patients. Whether performance indicators are process or outcome measures, they neglect important heterogeneities, and may force physicians and others to ignore the best interests of the patient or risk financial or reputational loss (Tanenbaum, Reference Tanenbaum2012). Pogach et al. (Reference Pogach, Tawari, Maney, Mangala, Miller and David2007), e.g., report that implementation of a common performance indicator, A1(c) rate of <7%, would actually decrease benefit or increase risk to one-third of diabetes patients in 144 VHA medical centers owing to physical or mental comorbidities. One physician has reported that in the ICU where he works, tight control of blood sugar is a quality metric that must be met in every case. Should one of his patient’s level rise, even for good reason, the physician faces a pointed lecture about the need to adhere to rules, as payers also use these metrics, and the potential loss of his faculty status owing to the poor quality of medical care he renders (Groopman and Hartzband, Reference Groopman and Hartzband2009).
Second, health care workers may face pressure to appear to satisfy performance indicators that are actually impossible to satisfy. In late May 2014, the acting Inspector General (IG) of the United States issued an interim report citing the manipulation of waiting lists at one and possibly more hospitals in the VHA system. A total of 1700 patients at the veterans medical center in Phoenix were omitted from the official waiting list and may never have received care. The IG suggests that data may have been falsified to create more favorable performance reviews, which in turn resulted in bonuses and salary increases. A total of 42 additional VHA facilities are under investigation (Oppel and Shear, Reference Oppel and Shear2014). Strategies for meeting wait-time performance standards appear to have varied among VHA medical centers, among them: the omission noted above; recording the date of a patient’s first appointment as the date of his first request for an appointment; and creating ‘ghost’ clinics to which patients are assigned for non-existent care (Oppel and Goodnough, Reference Oppel and Goodnough2014). The Secretary of the Department of Veterans Affairs, a member of the President’s Cabinet, has resigned in the wake of these revelations (Jaffe and O’Keefe, Reference Jaffe and O’Keefe2014).
A scandal is not a study, and the IG’s report is only preliminary and suggestive. Still, the problem at the VHA alerts us to the pitfalls of managing through self-interest. Perhaps, more importantly, it points out what can happen when performance management schemes are implemented without regard for real constraints on performance, i.e., when behavioral change is simply assumed to be within a worker’s control. In the case of the VHA (Oppel and Goodnough, Reference Oppel and Goodnough2014), an acute shortage of physicians, especially primary care physicians, seems to be responsible for long wait-times, but workers were pressured – by the system of rewards and by superiors affected by the system of rewards – to show that wait-times decreased. The VHA has experienced a 50% rise in primary care appointments over the last three years, but only a 9% increase in primary care physicians during the same time period. Medical center directors’ performance evaluations and consequent material rewards were nonetheless tied to a maximum 14-day wait for veterans seeking medical care. A shortage of primary care physicians is endemic to the US health care system, and the VHA may also suffer from lower physician salaries and personnel policies favoring mid-level administrators. None of these factors, however, are under the control of most workers whose performance is being managed and who may face the choice of capitulating, ‘whistle-blowing’ or exiting the health care system dedicated to serving veterans.
Oliver offers a careful review of performance management methods and their effectiveness measured in their own terms. These methods have also been politically effective, at least in the United States (Tanenbaum, Reference Tanenbaum2009). Despite its shortcomings, performance management enjoys a simple logic of economic or psychological inducement, which allows policy makers to be seen as doing something about the quality of health care. It appeals to the American preference for technical solutions, which can be viewed as ideologically neutral. It applies only to the current performers in a given setting, who may be less powerful than the interests shaping the health care system as a whole. Performance management, however, takes a narrow approach to improving health care quality, targeting selected, evidence-based behaviors, by self-interested workers, who are able to change what they do. One hopes performance management can improve hand-washing on the wards, but is it equipped to improve global quality or the practitioners who must achieve it?
