There are various definitions for P4P for organizations. For example, "the use of payment methods and other incentives to encourage quality improvement and patient-focused high value care" [25]; "incentives (generally financial) to reward attainment of positive health results" [26]; and "transfer of money or material goods conditional on taking a measurable action or achieving a pre-determined performance target" [27]. In order to perform systematic review, we needed to identify our own definition of P4P. In this article, P4P was operationally defined by any kind of financial incentives or rewards to healthcare providers aiming to improve quality of care. That is, we decided to pay attention to the fact that "pay" refers to financial benefits from purchasers and "performance" means only quality performance (or outcome).

A flow chart of the brief search strategy is shown in Figure 1. One librarian searched for published articles in two electronic databases, PubMed and Scopus, using selected key words. The electronic search was performed from 22 to 23, December, 2011 using the following keywords: pay for performance; incentive or reimbursement; performance or quality; payment or purchasing; health personnel, healthcare provider, physician, nurses, or hospital; attitude, behavior, position, or response; survey or questionnaires. Publication time and type were not limited. Two reviewers (Jo MW and Lee JY) screened target articles for analysis from results of electronic searches using titles and abstract review. Table 1 presents the inclusion and exclusion criteria used in the target articles chosen by the reviewers. Then, two reviewers read the full version of the screened articles and chose the final articles. In addition, references and forward citations of screened articles were also searched without any time limitations. One unpublished report was also added by expert opinion. Disagreements on the final choice of articles were resolved by discussions on the operational definition of P4P among the authors. Because the objective of this study was not to draw a single conclusion, the quality and heterogeneity of studies were not used for exclusion criteria and just considered in discussions. Among 1835 articles from PubMed, Scopus, and expert opinion, on the basis of their title and abstract, 58 articles were selected for full-text review. This review process provided 14 articles eligible for our analysis: 13 from the reviewed articles and 1 mentioned in the reference lists of several reviewed articles.

General information from the studies was extracted from the 14 articles that were finally selected: citation, country, program, target and eligible population of study, study participants, and survey method (Table 2). Healthcare providers' attitudes on P4P were summarized in two ways (Tables 3 and 4). First, we gathered their general attitudes and opinions regarding the effects of P4P. General attitudes included three items: level of awareness, agreement or disagreement, and the reasons why they support or oppose P4P. Providers' opinions on the effects of P4P were reviewed in terms of their behavioral changes, effects on quality of care, and financial impact. Second, we rearranged their opinions regarding desirable P4P design and implementation methods, as well as their concerns. Because there were differences in the context including target population, design, or goal of P4P and the healthcare system, the original descriptions in the selected articles were used in this paper if possible.

Table 2 shows the study descriptions including the first author and publication year, countries, the specific P4P program, target population, eligible bias, target population, and survey methods. Among the 14 studies, 11 studies were conducted in US and least 3 studies analyzed P4P in the US and UK, Korea, and Canada. The targets of the surveys were individual physicians in 6 studies, hospitals or physician organizations in 5 studies, and directors or physician executives in 2 studies. The total number of respondents in each study ranged from 16 to 1243. In a survey method, a mailed survey including postal service (6) was most common followed by interview (4), web-based survey including online questionnaire (3), fax (2), and poll (1).

We summarized providers' general attitudes toward P4P and their view on the effects resulting from P4P (Table 3). Table 3 showed the possibility that some healthcare providers still have a low level of awareness about P4P [28-31]. For example, McDonald and Roland [28] indicated that many physicians were not aware of the target contents or had poor understanding of the relation between performance and incentives payments received. Also, Lee et al. [29] mentioned that the majority of healthcare providers did not know or understand what a P4P program was. Lastly, Young et al. [30] reported that physicians were fairly negative about their understanding of the details of P4P programs and the amount of incentive money being offered to them.

We investigated why healthcare providers support or oppose P4P (Table 3). Their opinions varied according to their P4P settings. Proponents stated that 1) if the measures are accurate, physicians should be given financial incentives for quality [29,30,32,33]; 2) financial incentives are an effective way to improve the quality of healthcare [29,30,32]; 3) the benefits outweigh adverse consequences [34]; and 4) financial rewards are more effective as an incentive compared to non-financial rewards such as peer recognition [32]. However, opponents were concerned that 1) P4P will not lead to improved quality of care [31,35]; 2) they were skeptical that P4P would appropriately capture the quality of care [29,36]; 3) the incentive program was perceived as something externally imposed and managed, which made physicians feel that their autonomy was being challenged or that they were not trusted to perform in the absence of incentive payments [28]; 4) P4P could become a new method of government control over healthcare organization [29], and 5) P4P would result in unintended consequences [28,29,31-33,35-37].

We have summarized the effects resulting from P4P program in terms of behavioral change, effect on quality of care, and financial impact in Table 3. Regarding behavioral change, some healthcare providers reported that they have changed their behaviors to meet the quality goals. These behavioral changes included the structural modification and process alteration at the physician or organization level. At the organizational level, for example, Pines et al. [35] reported a number of operational changes that are being implemented to improve time to antibiotics for pneumonia. Also, Damberg et al. [34] reported that P4P has directly affected organizational behaviors by increasing accountability for quality, influencing the speed of IT adoption for quality management, and creating greater organizational focus and support for quality programs and goals. In addition, other studies indicated that structure and process change to reach the quality standard occurred at the organizational level [29,38]. At the physician level, even though three studies reported that physicians changed their behaviors to obtain financial incentives [28,33,39], other studies denied the fact that P4P induced behavioral changes at the physician level [30,32,36].

As for quality, their opinions on the effect of P4P were still controversial (Table 3). Some studies reported that healthcare providers believe P4P will increase the quality of cares [29,30,32,37,40] while others believe that the effect on quality will be a lack of or limited impact [29,33-35,41]. Lastly, regarding financial impact, only one study reported healthcare providers believe there are significant financial impacts [39]. Instead, other healthcare providers thought that financial gains would be limited or minimal [36,41].

Table 4 showed healthcare providers' attitudes on desirable P4P design and concerns. In order to summarize providers' opinions, we categorized their points of view into eight general issues such as participation method, evaluation unit and reward recipient, quality indicators, funding, quality of data, additional costs, unintended consequences, and government or health plan's support.

In relation to participation method, only one study directly mentioned that healthcare providers prefer voluntary participation in P4P program, not mandatory participation [29]. Regarding evaluation unit and reward recipients, some providers preferred individual- (or physician-) based performance evaluation, and therefore financial incentives should be given to physicians who achieved the goal of quality [36]. However, other providers believe financial incentives should provide medical group or healthcare organization based on the result of organizational performance evaluation [29,35].

Also, many healthcare providers had serious concerns about current quality indicators [29,31-33,36,37]. They believed that current quality indicators did not appropriately reflect on their clinical situations. For example, they believe quality measures cannot be adequately adjusted for patients' medical conditions or socioeconomic status [32]. Therefore they wanted purchasers to work hard to make quality measures accurate and reflect clinical significance [37].

How to raise funds for financial incentives is another important issue. One article concerned that incentives are not new money so there are always winners and losers [37]. In addition, another study reported that additional funding for P4P should be prepared for financial rewards [29]. That is, healthcare providers prefer to receive additional financial incentives without penalties.

In order to evaluate quality, healthcare providers should gather correct and accurate quality data and should hand in the data to purchasers. However, they felt that this process is very stressful and they could pay additional costs [29,31,33,37]. Also, for accurate quality evaluation, they believe more data needed, in addition to claims data [29,31,33,37,40]. Therefore, they desired that government or health plans should support the gathering of quality data and should pay for the additional costs or reduce the costs of data collection [29,40].

Unintended consequences were one of the most vital issues in designing and implementing P4P. Even though one article reported that unintended consequences would not be so serious [30], most healthcare providers had serious concerns that P4P would induce unintended consequences [28,29,31-33,35-37].

Lastly, the providers felt that they need some support from the government or health plan to make the P4P program more successful. For instance, some articles mentioned that healthcare providers need technical and educational support such as IT infrastructure and electronic medica record [28,30,36,37,40,41].

The purpose of this study was to investigate healthcare providers' attitude toward P4P. To do so, ultimately purchasers can reflect their opinions on designing and implementing P4P programs. Therefore, the basic hypothesis was that they could correctly understand the concept, evaluation method, and the relationship between performance and rewards. However, our results indicated that some healthcare providers still have a low level of awareness about P4P [28-31]. If the low level of awareness or understanding about P4P among healthcare providers is a general phenomenon, that would be a serious barrier to implementing a P4P program. As noted, purchasers should note that the provider's support is essential for the success of P4P [31,32]. Therefore, before implementing P4P programs, purchasers should grasp whether providers are correctly aware of the P4P. If not, they should make an effort to increase the level of awareness such as understanding of the quality indicators and the criteria and methods for distributing financial incentives. This could be a meaningful starting point to reconsider current P4P program. If they have incorrect knowledge about P4P, and if this results in distorted attitudes toward the P4P program, then each P4P program could not get a sufficient degree of support from providers. In order to increase the level of awareness of providers' knowledge and attitudes about P4P, adequate educational support should be provided with them by the health plan or professional societies. Actually, one study reported that providers were more likely to support P4P programs when they had received information about these programs from their professional societies [24].

Purchasers should pay attention to the reason they oppose P4P. The reasons of opposing P4P may be quite different according to their P4P settings. However, our results showed very interesting phenomenon. That is, two opinions are separated based on different points of attitudes on the same issue. For example, proponents support P4P because it can lead to improve the quality of care and benefits outweigh adverse consequences. On the other hand, opponents disagree on P4P because it cannot lead to improvement in the quality of care and it would result in several unintended consequences. In summary, purchasers should listen to the voices of opponents. If their reasons of objection are correctable or based on misunderstandings, we can reflect their opinions or try to fix them.

To achieve the final goal of P4P, purchasers should make cooperative relationship and communicate with providers and professional societies. Like Healy and Braithwaite [44] mentioned that "command and control" would be replaced with "the new regulatory state" seeking flexible, participatory, and devolved forms of regulation. From this point of view, purchasers should have patience to persuade providers and facilitate their supportive participation. In particular, in the designing stage of P4P program, purchasers should actively communicate with providers and make an effort to reflect their opinions regarding its primary target, evaluation unit and reward recipients, participation methods, carrots or sticks, and funding methods. Of course, government or health plan cannot and should not accept all opinions from providers. However, P4P is also not a zero-sum game. They may often conflict with each other but can cooperate to make better P4P program.

Purchasers should work hard to develop accurate quality measures and reflect clinical significance [37]. According to our study, many healthcare providers had serious concerns about quality measures or indicators. However, there is no perfect evaluation method. Therefore, purchasers should make a continuous effort to monitor and revise current quality indicators regularly in order to reflect providers' professional opinions.

Purchasers should take actions to minimize the unintended consequences. From the articles reviewed in this study, we could summarize their concerns about unintended consequences [28,29,31-33,35-37]: avoiding high-risk patients; ignoring quality of care; neglecting compulsory services; inappropriate behavioral changes; and threatening professional autonomy. Maybe, each P4P program can confront different kinds of unintended consequences and its solutions would be also quite different. Although these undesirable consequences cannot be eliminated in the real world, some of them can be fixed or minimized by purchasers.

Purchasers should pay attention to whether their providers get ready to launch P4P. Also, purchasers should make a plan to decrease the providers' burdens and supportive environments. According to our results, some providers believe they were not ready to or appropriately prepared to initiate P4P [36] and had concerns about infrastructure for P4P introduction such as IT and the quality of data et al [28,30,31,36,37,40,41]. Technical or financial supports such as providing grants for establishing IT infrastructure, providing education sessions, and sharing costs for additional data collection can be useful strategies for soft-landing of P4P program.

There could be several limitations in this study. One is incompleteness of literature search. Although we used wide range of search terms and references of searched articles, some studies related with this topic might be missed. In addition we did not use comprehensive source of references such as expert's opinion, qualitative studies, secondary data analysis. Second, there is a possibility that providers might intentionally express negative responses to some questions in their survey in order to get a better strategic position in the future P4P. The third limitation is about the heterogeneity of the articles reviewed in the perspectives of healthcare settings, respondents, and questionnaires. As stated, this heterogeneity could be related with healthcare providers' attitudes in each article. Therefore, we intended to describe and summarize the authors' opinions on items by studies rather than provided a single result.