The Health Information Technology. HITECH Act Drove Large Gains In Hospital Electronic Health Record Adoption. Hospital EHRs

Transcription

1 doi: /hlthaff HEALTH AFFAIRS 36, NO. 8 (2017): Project HOPE The People-to-People Health Foundation, Inc. By Julia Adler-Milstein and Ashish K. Jha HITECH Act Drove Large Gains In Hospital Electronic Health Record Adoption Julia Adler-Milstein (juliaam@ umich.edu) is an associate professor in the School of Information and School of Public Health (health management and policy) at the University of Michigan, in Ann Arbor. Ashish K. Jha is the K. T. Li Professor of International Health at the Harvard T. H. Chan School of Public Health in Boston, and director of the Harvard Global Health InstituteinCambridge,bothin Massachusetts. ABSTRACT The extent to which recent large increases in hospitals adoption of electronic health record (EHR) systems can be attributed to the Health Information Technology for Economic and Clinical Health (HITECH) Act of 2009 is uncertain and debated. Because only short-term acute care hospitals were eligible for the act s meaningful-use incentive program, we used national hospital data to examine the differential effect of HITECH on EHR adoption among eligible and ineligible hospitals in the periods before ( ) and after ( ) implementation of the program. We found that annual increases in EHR adoption rates among eligible hospitals went from 3.2 percent in the pre period to 14.2 percent in the post period. Ineligible hospitals experienced much smaller annual increases of 0.1 percent in the pre period and 3.3 percent in the post period, a significant difference-in-differences of 7.9 percentage points. Our results support the argument that recent gains in EHR adoption can be attributed specifically to HITECH, which suggests that the act could serve as a model for ways to drive the adoption of other valuable technologies. The Health Information Technology for Economic and Clinical Health (HITECH) Act of 2009 was an ambitious policy effort to increase the adoption of electronic health records (EHRs). It was prompted by evidence that the use of EHRs can substantially improve the quality and efficiency of care delivered. 1 And while the act passed with strong bipartisan support, whether it has achieved its primary goal is debated. Proponents argue that before the act s passage, EHR adoption rates were low and increasing very slowly. Without a policy intervention, it might have taken many years, if not decades, for EHR adoption to reach nationwide scale. 2 Furthermore, in response to the recession, hospitals planned investments in EHRs could have been delayed if it had not been for the act. Critics disagree, arguing that EHR adoption rates were already increasing on their own and that the act s centerpiece financial incentives for providers demonstrating meaningful use of EHRs simply substituted public money for private funds. 3 That is, adoption rates might have been quite similar had the federal government not intervened, because hospitals and ambulatory practices would have adopted EHRs on their own in response to market pressures. The evidence to support either side of the debate is limited.while we know that EHR adoption rates have risen substantially since the enactment of HITECH, 4 it is unclear how hospital EHR adoption rates would have risen had HI- TECH incentives not been put into place. A study in the ambulatory care setting found that basic EHR adoption was not significantly altered by HITECH incentives, 5 but whether that was also the case for hospitals is unknown. It is critical to understand the extent to which HITECH directly increased EHR adoption for two reasons. First, 1416 Health Affairs August :8

2 the federal government has spent more than $30 billion of taxpayers money implementing HITECH provisions, 6 and it is important to assess whether the public has received a key component of what it was promised: incentive-driven adoption of EHRs. Second, HITECH is an unusual type of incentive program in that it pays hospitals and physicians not for outcomes or engagement in care processes, but for infrastructure (having and using an EHR). Understanding the impact of HITECH is therefore a rare opportunity to learn whether incentives can drive adoption of technology infrastructure, and it should inform policy makers in other countries planning similar programs. 7 Because it was not feasible to use a randomized trial to test the impact of HITECH, we used an alternative approach. The financial incentives for meaningful use of EHRs were made available to a certain group of hospitals (short-term acute care facilities) and not to others (long-term acute care, psychiatric, or rehabilitation facilities), which created a natural division to analyze. 8 These two groups of hospitals should be comparably motivated to pursue EHR adoption, as the clinical care benefits have been shown to be similar for both groups Therefore, using ineligible hospitals as our control group, we sought to assess whether HITECH incentives increased the trajectory of EHR adoption among eligible hospitals to a greater degree than was observed among ineligible hospitals. We also examined whether certain types of eligible hospitals, such as smaller or for-profit hospitals, were more likely than other types to respond to the incentives. Study Data And Methods Data Sources And Sample We used data for the period from the Annual Health Information Technology (IT) Supplemental Survey of the American Hospital Association. This period included three years before and five years after the initiation of the meaningful-use program. The AHA IT survey is sent to the CEO of every hospital in the United States. These executives are asked to complete the survey or delegate its completion to the most knowledgeable person in the organization. All nonrespondents receive multiple mailings and follow-up phone calls to achieve a high response rate. Hospitals are given the option of completing the survey online or by mail. Survey questions capture the extent of adoption of individual computerized clinical functions and when linked to the AHA s full Annual Survey additional hospital demographic characteristics. We included data for all nonfederal US hospitals from all of the years in which they responded to the AHA IT supplement survey. After we merged eight years of data, our sample consisted of 25,210 total observations and 5,119 unique hospitals that responded to one or more years of the AHA IT supplement survey between 2008 and 2015 (for more details about the sample, see online Appendix Exhibit A1). 12 Measures We considered general and medical-surgical acute care hospitals as eligible for meaningful-use incentives, and long-term acute care, psychiatric, and rehabilitation hospitals as ineligible.we used the AHA service code to identify the appropriate category for each hospital, which resulted in 4,268 eligible hospitals and 851 ineligible hospitals. Our primary measure was whether or not each hospital had adopted at least a basic EHR in each year. Using a definition from a previous study, 4 we considered a hospital to have at least a basic EHR if it reported full implementation of the following ten computerized functions in at least one clinical unit of the hospital: recording patient demographic information, physician notes, nursing assessments, patient problem lists, patient medication lists, and discharge summaries; viewing laboratory reports, radiologic reports, and diagnostic test results; and ordering medications. We captured hospital characteristics, including system affiliation, ownership (private forprofit, private not-for-profit, or public), size (based on the number of beds), region, teaching status (nonteaching or minor teaching, since no ineligible hospitals were major teaching hospitals that is, members of the Council of Teaching Hospitals), rural or urban location, and proportion of Medicaid admissions. We hypothesized that system-affiliated, private forprofit, small, rural, and nonteaching hospitals would respond more strongly to meaningful-use incentives. Analytic Approach We conducted a difference-in-differences of slope analysis in which we compared the difference between slopes of EHR adoption rates before ( ) and after ( ) implementation of meaningful-use incentives for eligible versus ineligible hospitals. As a robustness test, we limited the ineligible hospital comparison group to only long-term acute care hospitals, since compared to psychiatric and rehabilitation hospitals they are arguably more similar to general acute-care hospitals. As a second robustness test, we compared adoption rates among the subset of eligible and ineligible hospitals that did not have a basic EHR in the period Since four of the ten basic EHR functions were widely adopted in 2008 (recording patient demographic information and patient medication lists, and viewing laboratory August :8 Health Affairs 1417

3 reports and radiologic reports), we considered hospitals not to have a basic EHR if they lacked one or more of the remaining six basic EHR functions: recording physician notes, nursing assessments, discharge summaries, and patient problem lists; viewing diagnostic test results; and ordering medications. As a third robustness test, we included hospital margins (as reported on Medicare cost reports) to adjust for differences between eligible and ineligible hospitals in terms of resources that might be available to purchase an EHR. Our final robustness tests sought to address the potential for overestimating the impact of HITECH by failing to adjust for hospitals that were working toward EHR adoption before implementation of meaningful-use incentives and completed adoption after the incentives were available. Hospital EHR adoption typically occurs incrementally over multiple years, 13 and our main analytic approach gave HITECH credit for all hospitals that first reached the basic EHR threshold in the period after implementation of the incentives even if reaching that threshold was the continuation of plans made before implementation. We therefore created a new measure that captured both basic EHR adoption and basic EHR implementation if a hospital either reported full implementation of the ten computerized functions in at least one clinical unit or reported that it was beginning to implement one or more functions in at least one unit or have resources to implement in the next year. We then reran our primary model, substituting this variable in the years before the meaningful-use incentive program ( ) so that HITECH did not receive credit for hospitals that completed implementation of an EHR after the incentive program began but had already planned to implement the EHR before the incentives became available. We used a second approach to address this concern by running our primary model again but shifting the pre-post date forward one year (that is, we compared to ). This approach requires the strong assumption that all new basic EHR adoption in the one year following meaningful use incentives availability would have occurred on its own. To determine the differential impact of HITECH on EHR adoption based on hospital type, we conducted stratified analyses by system affiliation, ownership, size, teaching status, and urban versus rural location. We then compared the differences between slopes of EHR adoption rates before and after implementation of meaningful-use incentives for eligible and ineligible hospitals within each category. All models used ordinary least squares regressions, with hospital-level clustering to account for repeated observations. We chose ordinary least squares regression over logistic regression to enable us to directly interpret results in terms of percentages of EHR adoption. Models were adjusted for nonresponse bias by using weights derived from a regression model that predicted the likelihood of responding to the AHA IT survey based on hospital characteristics.we also ran an unweighted set of results. Limitations Our study had several important limitations. First, ineligible hospitals are different from eligible hospitals in terms of their patient populations and the care they deliver. Therefore, ineligible hospitals are not the perfect control group. However, because meaningful use is a national program, we lacked alternative comparison groups. Furthermore, given that all hospitals need to perform basic clinical tasks such as medication ordering and problem tracking, we expected that the two groups of hospitals would be similarly motivated to pursue EHR adoption. Second, although the overall response rates for the AHA IT supplement are high for a national institutional survey, nonresponders may have differed from responders. We used the statistical technique described above to account for this potential bias. However, these adjustments were not perfect. Finally, we were not able to disentangle the effect of the various individual components of HITECH on EHR adoption. While the meaningful-use incentive program was the centerpiece of HITECH, complementary programs such as the Regional Extension Center program and the EHR certification program alone or in combination with the meaningful-use program could have driven an increase in EHR adoption. Study Results Compared to eligible hospitals, ineligible hospitals were more likely to be small (72 percent versus 50 percent), located in the South (50 percent versus 38 percent), nonteaching (86 percent versus 75 percent), for profit (63 percent versus 17 percent), urban (84 percent versus 45 percent), and affiliated with a health care system (71 percent versus 59 percent) (Exhibit 1). Ineligible hospitals also had a slightly lower percentage of admissions from Medicaid, compared to eligible hospitals. Adoption Among Eligible Versus Ineligible Hospitals Among eligible hospitals, EHR adoption rates increased by an average of 3.2 percent annually in the period before implementation of the meaningful-use incentives (Exhibits 2 and 3). In the period after implementation, the 1418 Health Affairs August :8

4 average annual increase was 14.2 percent, reflecting a change of 11.1 percentage points. In comparison, among ineligible hospitals, the increase before implementation was 0.1 percent, and the increase after implementation was 3.3 percent. The difference-in-differences of the slope was 7.9 percentage points. When we compared eligible hospitals only to long-term acute care hospitals, our results were similar: The change from pre- to post-implementation among eligible hospitals was 11.1 percentage points (from 3.1 percent to 14.3 percent; p<0:001), while the change among longterm acute care hospitals was 2.9 percentage points (from 1.9 percent to 4.8 percent; p ¼ 0:156) (Appendix Exhibit A2). 12 The difference-in-differences of the slope was 8.3 percentage points (p <0:001). When we compared only eligible and ineligible hospitals that did not have an EHR before implementation of the meaningful-use incentives, we found that the EHR adoption rate for eligible hospitals increased by 16.5 percent per year, on average, compared to 5.5 percent for ineligible hospitals (Appendix Exhibit A3). 12 This reflects a difference-in-differences of 10.9 percentage points (p <0:001). Both our model that included hospital margins as a control and our unweighted model produced results that were almost identical to our primary results, with differences-in-differences of 7.97 percentage points and 7.60 percentage points, respectively (p <0:001 for both) (Appendix Exhibits A4 and A5). 12 Finally, in our model that substituted basic EHR adoption for either basic EHR adoption or planned implementation in the period before meaningful-use incentives, the difference-indifferences decreased to 3.86 percentage points (p <0:05) (Appendix Exhibit A6). 12 In our model that shifted the pre-post date forward by one year, the difference-in-differences decreased to 4.9 percentage points (p <0:001) (Appendix Exhibit A7). 12 Adoption Among Eligible Versus Ineligible Hospitals, By Hospital Type When we examined whether the difference-in-differences of slope varied based on system affiliation, ownership type, size, urban or rural location, and teaching status, the only significant difference that we observed was for ownership type (Exhibit 4). Among private for-profit hospitals, the difference-in-differences of the slope was 13.3 percentage points, significantly greater than the 1.8-percentage-point difference-in-differences among private not-for-profit hospitals. Exhibit 1 Hospitals eligible and ineligible for meaningful-use incentives to adopt at least a basic electronic health record, by selected characteristics Characteristic Eligible hospitals Ineligible hospitals Size Small 50.2% 72.4% Medium Large Region Northeast Midwest South West Teaching status Major Minor Nonteaching Ownership For profit Not for profit Public Location Urban Rural System affiliated No Yes Medicaid admissions as percent of total admissions 16.9% 10.9% SOURCE Authors analysis of data for 2013 from the American Hospital Association s Annual Health Information Technology Supplemental Survey. NOTES There were 4,268 hospitals in our data eligible to participate in the program of meaningful-use incentives, and 851 hospitals ineligible to participate. All comparisons were significant (p < 0:001) using chi-square tests for differences across categories. Exhibit 2 Percentages of hospitals that adopted at least a basic electronic health record system in , by eligibility for meaningful-use incentives SOURCE Authors analysis of data for from the American Hospital Association s Annual Health Information Technology Supplemental Survey. August :8 Health Affairs 1419

5 Exhibit 3 Average annual increase in the percentages of hospitals adopting at least a basic electronic health record system, by eligibility status, before and after implementation of meaningful-use incentives Before meaningful use ( ) After meaningful use ( ) Difference (percentage points) Eligible hospitals 3.2% 14.2% 11.1 Ineligible hospitals 0.1% 3.3% 3.2 Difference-in-differences 7.87 SOURCE Authors analysis of data for from the American Hospital Association s Annual Health Information Technology Supplemental Survey. NOTE All differences and differencesin-differences were significant (p < 0:001) from ordinary least squares regression. Discussion HITECH is a unique policy intervention that offered hospitals financial incentives in the form of bonus payments to speed the rate of EHR adoption. Even though the large observed increase in EHR adoption among eligible hospitals after meaningful-use incentives were introduced is compelling evidence of HITECH s effectiveness, it is possible that many hospitals would have adopted EHRs without the policy intervention. By using ineligible hospitals as a control group, we found that HITECH can be credited with increasing the rate of EHR adoption by 8 percentage points per year. A conservative estimate that did not give HITECH credit for any adoption that was planned before implementation of the incentive program reduced the effect size to 4 percentage points per year. For policy makers, our results are good news and suggest that HITECH could serve as a model for driving the adoption and use of other valuable technologies. However, our results do not imply that HITECH has been an unqualified success. Real challenges remain, particularly in the domains of interoperability and usability. Nonetheless, speeding the national timeline of hospitals EHR adoption was a key first step toward digitizing the health care system and facilitating the transition to paying for value. Our results raise the question of whether the annual 8-percentage-point increase attributable to HITECH is substantial and reflects good value for the $20.9 billion that was paid to hospitals through 2015 (with additional funding paid to eligible professionals) 6 as a result of their meeting meaningful-use criteria. Given that the level of EHR adoption among eligible hospitals in 2010 was 15 percent, an increase of 8 percentage points per year suggests that in five years the incentives moved US hospitals past the halfway mark. There are likely very few other policies that have driven such substantial change in such a short period. It is also worth noting that the incentive amount covers only a small fraction of the total cost for hospitals to adopt an EHR (10 percent, according to the American Hospital Association), 14 which suggests that the incentive was efficient. However, there were also incentives available for eligible professionals (who are largely ambulatory providers) that could have increased hospitals responsiveness by cross-subsidizing hospital costs and thereby could have contributed to the results that we observed. Ultimately, a comprehensive assessment of the value of HITECH requires quantifying the gains in terms of quality and efficiency of Exhibit 4 Differences in changes over time by eligibility for meaningful-use incentives between percentages of hospitals that adopted at least a basic electronic health record system, by hospital characteristics SOURCE Authors analysis of data for from the American Hospital Association s Annual Health Information Technology Supplemental Survey. NOTES The exhibit shows the results of a difference-in-differences analysis that compared adoption of electronic health record (EHR) systems by hospitals eligible to participate in the meaningful use of EHRs incentive program (implemented in 2011 under the Health Information Technology for Economic and Clinical Health [HITECH] Act of 2009) and those ineligible to participate, before and after the program s implementation. For example, among system-affiliated eligible hospitals, compared to system-affiliated ineligible hospitals, eligible hospitals had 7.9 percent point greater adoption in , compared to in Small is fewer than 100 beds. Medium is beds. Large is 500 or more beds. Minor teaching status is hospitals approved to participate in residency or internship training by the Accreditation Council for Graduate Medical Education. The error bars show 95% confidence intervals. The only category in which the difference-in-differences were different from each other was ownership, with for-profit private significantly greater than not-for-profit private (p < 0:001) Health Affairs August :8

6 Speeding the national timeline of hospitals EHR adoption was a key first step toward digitizing the health care system. care, where the evidence to date has been mixed Another important consideration in the value assessment is that HITECH did not target incentives only to hospitals that otherwise would not have adopted an EHR. 18 Thus, at least some of the hospitals that received incentives already had an EHR or would have adopted one on their own during the HITECH period even if the HITECH incentive program had not existed, which was a suboptimal use of public dollars. The results of our robustness tests that did not give HITECH credit for basic EHR adoption among hospitals that reported plans to adopt before 2011 suggest a 50 percent reduction in effect size (from 8 percentage points to 4 percentage points). However, our finding that, compared to private not-for-profit hospitals, private for-profit institutions responded much more strongly to the incentives bolsters the argument that the incentives were successful at moving nonadopters. We would expect private for-profit hospitals to behave in ways that help the institutions achieve their financial goals. Thus, their low EHR adoption rate in the period before the meaningful-use incentives and the large increase in that rate after the incentives became available suggest that the program was largely effective in targeting hospitals that would not have otherwise adopted an EHR. While a more efficient use of public dollars would have created an incentive program that targeted only nonadopters, such an approach would have equity implications: It would unfairly penalize hospitals that took a risk in investing in a technology with compelling patient benefits, particularly in the safety domain, but with an uncertain return on investment. Furthermore, it could have dampened hospitals willingness to invest in technology in the future because of concern about missing out on potential government incentives. Policy Implications Our findings have implications for future policy efforts that seek to spur technology adoption. They suggest that financial incentives tied to technology adoption are likely to substantially speed uptake across a range of hospital types. However, given that previous analyses of ambulatory care departments failed to find an effect of HITECH, 5 it is not clear which mechanisms are at play. It may be that hospitals have organizational capabilities to understand and react to complex incentive programs such as meaningful-use incentives that are lacking in ambulatory practices. It is also possible that the different results are explained by the increasing maturity of the EHR market. In 2009, when HITECH was passed, the ambulatory EHR market was less mature than the hospital EHR market; there were several thousand certified ambulatory EHR products from which to choose which could have made it more difficult for ambulatory care providers to respond. More broadly, our results raise the question of why HITECH was successful in driving change among hospitals when many other policy efforts that seek to change hospitals and ambulatory providers behavior have not succeeded. In particular, there is mixed evidence on the effectiveness of pay-for-performance programs in improving quality of care. 19 We suspect that a primary reason is the sense of inevitability: EHR adoption was likely in many hospitals long-term plans, and the availability of incentives may have simply moved it up on the priority list. In contrast, many hospitals might not have been aware of the magnitude of issues such as hospital-acquired infections or avoidable readmissions until they became the focus of national policies. It is also possible that hospitals felt more confident that they could be successful in responding to the meaningful-use incentive program, compared to in reducing hospitalacquired infections or avoidable readmissions, because implementing an EHR system was perceived as easier than meaningfully improving patient outcomes. While adopting an EHR is a complex undertaking, the process s individual steps (for example, selecting a vendor and planning for implementation) may seem more straightforward than those related to improving clinical outcomes. Conclusion We sought to determine the extent to which HITECH should be credited with the large increase in EHR adoption among hospitals that occurred following its passage. We found that HITECH drove annual gains in EHR adoption August :8 Health Affairs 1421

7 of 8 percentage points in the five years after implementation of the act s meaningful-use incentive program (with 4 percentage points as a conservative lower bound). Given the complexity of large-scale change in hospitals, this result is dramatic and suggests that HITECH can serve as a model for other countries seeking to increase EHR adoption among hospitals and for other policy efforts seeking to promote technology adoption more generally. This work was presented at the AcademyHealth Annual Research Meeting, Minneapolis, MN, June 16, The corresponding author had full access to the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. NOTES 1 Blumenthal D. Stimulating the adoption of health information technology. N Engl J Med. 2009; 360(15): Blumenthal D. Health information technology: what is the federal government s role? [Internet]. New York (NY): Commonwealth Fund; 2006 Mar [cited 2017 Mar 13]. (Publication No. 907). Available from: fund.org/~/media/files/ publications/fund-report/2006/ mar/health-informationtechnology what-is-the-federalgovernments-role/blumenthal_ hit_907-pdf.pdf 3 Woskie LR, Adler-Milstein J. Chapter 7: coordinator s corner. In: Robert Wood Johnson Foundation. Health information technology in the United States, 2015: transition to a post-hitech world [Internet]. Princeton (NJ): RWJF; [cited 2017 Mar 13]. p Available from: farm/reports/reports/2015/ rwjf Adler-Milstein J, DesRoches CM, Kralovec P, Foster G, Worzala C, Charles D, et al. Electronic health record adoption in US hospitals: progress continues, but challenges persist. Health Aff (Millwood). 2015;34(12): Mennemeyer ST, Menachemi N, Rahurkar S, Ford EW. Impact of the HITECH Act on physicians adoption of electronic health records. J Am Med Inform Assoc. 2016;23(2): CMS.gov. EHR incentive programs: data and program reports [Internet]. Baltimore (MD): Centers for Medicare and Medicaid Services; [last modified 2017 Mar 1; cited 2017 Mar 13]. Available from: programs/dataandreports.html 7 NHS England. Harnessing the information revolution: Wachter Review [Internet]. London: NHS England; 2016 [cited 2017 Jun 26]. Available from: info-revolution/wachter-review/ 8 Walker D, Mora A, Demosthenidy MM, Menachemi N, Diana ML. Meaningful use of EHRs among hospitals ineligible for incentives lags behind that of other hospitals, Health Aff (Millwood). 2016;35(3): Field TS, Rochon P, Lee M, Gavendo L, Baril JL, Gurwitz JH. Computerized clinical decision support during medication ordering for long-term care residents with renal insufficiency. J Am Med Inform Assoc. 2009;16(4): Judge J, Field TS, DeFlorio M, Laprino J, Auger J, Rochon P, et al. Prescribers responses to alerts during medication ordering in the long term care setting. J Am Med Inform Assoc. 2006;13(4): Gurwitz JH, Field TS, Rochon P, Judge J, Harrold LR, Bell CM, et al. Effect of computerized provider order entry with clinical decision support on adverse drug events in the long-term care setting. J Am Geriatr Soc. 2008;56(12): To access the Appendix, click on the Appendix link in the box to the right of the article online. 13 Adler-Milstein J, Everson J, Lee SY. Sequencing of EHR adoption among US hospitals and the impact of meaningful use. J Am Med Inform Assoc. 2014;21(6): American Hospital Association. Statement of the American Hospital Association to the U.S. Senate Committee on Health, Education, Labor and Pensions: America s health IT transformation: translating the promise of electronic health records into better care. [Internet]. Washington (DC): AHA; 2015 Mar 17 [cited 2017 Mar 14]. Available from: issues/testimony/2015/ statement-helpit.pdf 15 Adler-Milstein J, Everson J, Lee SY. EHR adoption and hospital performance: time related effects. Health Serv Res. 2015;50(6): Furukawa MF, Eldridge N, Wang Y, Metersky M. Electronic health record adoption and rates of in-hospital adverse events. J Patient Saf Feb 6. [Epub ahead of print]. 17 Jones SS, Rudin RS, Perry T, Shekelle PG. Health information technology: an updated systematic review with a focus on meaningful use. Ann Intern Med. 2014;160(1): Dranove D, Garthwaite C, Ody C, Li B. Investment subsidies and the adoption of electronic medical records in hospitals [Internet]. Cambridge (MA): National Bureau of Economic Research; 2014 Oct [cited 2017 Jul 5]. (NBER Working Paper No ). Available for download (fee required) from: 19 Van Herck P, De Smedt D, Annemans L, Remmen R, Rosenthal MB, Sermeus W. Systematic review: effects, design choices, and context of pay-for-performance in health care. BMC Health Serv Res. 2010; 10(1): Health Affairs August :8