from EVIDENCE to POLICY THEWa BANK Learning what works for better programs and policies February 2015 RWANDA: Will More People be Tested for HIV if Clinics are Paid Extra? The world has made great strides toward ending AIDS. Yet The World Bank is working with the international the deadly disease remains a critical development challenge community and governments to improve health care, for poor countries. Sub-Saharan Africa, which has only and part of this is halting the spread of HIV and iden- 12 percent of the global population, is home to about 68 tifying infected people in order to make treatment pos- percent of all people living with HIV Improving rates of sible and help reduce transmission. Succeeding requires HIV testing in order understanding what programs are working, and how to identify and coun- these programs can be adjusted for specific goals. In sel infected people is Rwanda, impact evaluation researchers looked at the necessary for halting results of a national pay-for-performance scheme that 1 transmission of the vi- included extra payments for clinics that tested individu- rus and ensuring that als and couples for HIV infection. The evaluation found people who are in- that the payments increased the likelihood that people fected can get treated. who were part of a couple would get tested, showing The challenge is how that pay-for-performance could be a route for improv- to improve rates of testing, especially among couples where ing testing (and thus making available information on one partner is infected and either doesn't know or hasn't how to prevent HIV transmission) among those who told the partner. Increasingly, pay-for-performance is being face risk of infection from their partner. The results considered as an option for improving health care for preg- are particularly important for Sub-Saharan Africa, nant women and children. Development experts and poli- where according to 2009 World Health Organiza- cymakers are interested in whether bonus payments can tion data, nearly 80 percent of HIV-infected adults work in other areas of health care, such as improving the are unaware of their HV status, and more than 90 rate of HIV testing and treatment, especially in couples. percent dondt know if their partners are infected. The Government of Rwanda has made rebuilding institu- pregnant women in for prenatal care and vaccinating chal- tions destroyed in the 1994 civil war and genocide a prior- dren. The pay-or-performance program was expanded , ity. One area of focus has been the health system, especially tionwide in 2006, and at the governmenes request an im- treatment for women and children. In 2001, the govern- pact evaluation was built in to measure the impact on key ment started piloting programs that gave health facilities health indicators (Evidence to Policy March 2013, "Can bonus payments for meeting certain goals, such as getting Bonus Payments Improve the Quality of HealtH). The pay-for-performance program included bonus fected people in a relationship either didn't know they payments linked to testing, treatment and prevention were infected or were not always informing their part- of mother-child transmission. An estimated 3 percent ner, increasing the risk of transmission. In order to test of Rwandan adults were infected with HIV/AIDS, and whether bonus payments worked specifically for improv- the government especially wanted to increase testing to ing the rate of HIV testing, an impact evaluation focused help reduce transmission and offer treatment to those on HIV testing was included in the broader pay-for-per- who needed it. One particular concern was that HIV in- formance scheme. Evaluation The impact evaluation design took advantage of phased- only facilities offering HIM/AIDS services were included in implementation of the pay-for-performance program in the evaluation, with 10 in the treatment group and 14 at the district level. Districts in which non-governmen- in the comparison group. Health clinics in the treatment tal organizations were already running their own bonus groups started qualifying for the HIM-testing bonus pay- payment programs for health facilities were excluded. ments in January 2007 and this continued for 12 to 15 Using 2002 census data, the remaining districts were months, with payments made quarterly. The comparison grouped based on similarities for population density, group received payments for about 18 months. Payments rainfall, and economic activity. to the comparison group were calculated as an average of Health facilities in each group were randomly as- the payments made to the treatment group. signed either to the treatment group or to the com- A baseline survey of facilities was conducted in the parison group, though researchers were forced to make second half of 2006, and the follow-up survey took changes in the original assignment because of a govern- place between April 2008 and July 2008. A separate ment-led redrawing of administrative districts. Still, the household level survey sampled 1,000 households with treatment and control groups were comparable at base- an HIM-positive member, and 600 randomly sampled line, and researchers used a method called difference-in- neighboring households in the area served by the health difference to compare changes between the treatment facilities in the evaluation. HIM-positive individuals and control groups. were identified through the health facility or a group Facilities in the treatment group qualified for bo- for people with HIM/AIDS. They had to give informed nus payments based on how many people got tested consent before households were interviewed. for HIV Facilities in the comparison (control) group To measure the impact on testing of individuals, received the same average amount of money without people who were HIM positive were excluded from the any conditions. Giving the comparison group the same analysis, leaving 438 individuals in the treatment group average amount of money without linking it to perfor- and 445 in the comparison. Everyone was 15 years old mance results made it possible to isolate the impact of or older. In order to measure the increase in testing linking payments to performance. This made it possible among couples, the researchers focused on the sample to test whether any change in HIV testing was due to of adults identified as HIM negative (these were people the bonus payment structure. who were not identified as HIM positive in the previous In total, nine districts were in the treatment group and sample), who reported having sex in the previous 12 seven in the comparison group. Within those districts, months and who lived with their sexual partners. This policy note summarizes the results of the World Bank Group's Policy Research Working Paper 6364, "Using provider performance incentives to increase HIV testing and counseling services in Rwanda," by Damien de Walclue, Paul J. Gertler, Sergio Bautista-Arredlondlo, Ada Kwan, Christel Vermeersch, Jean de Dieu Bizimana, Agn2s Binagwaho, and Jeanine Condo. February 2013. Available at http://go.worldbank.org/MDE I NURDQO. The stidy was published in the Journal of Health Economics, 40 (201)a1-9, availble through www.elsevie0com/locate/econbase. People were more likely to get tested for HIV were part of a broader package of bonus payments linked if they lived in the areas served by health to HIV testing, counseling and treatment. facilities that qualified for bonus payments linked to testing. Health clinics were free to use the bonus payments as they wanted and most of the The likelihood that someone had been tested for HIV money went to improving pay for employees. significantly rose by 6.1 percentage points (or a 10.6 per- cent increase over baseline) if they went to a facility that Among the clinics that qualified for bonus payments, could receive bonus payments, compared with someone the money they received was equal to 14 percent of seen at a facility that received extra money regardless of how many people were tested. Before the program started, the overall testing rate was 53.9 percent. However, the increase was driven by more testing among individuals who lived with their sexual part- ners. The testing rate among couples before the pro- gram was launched was 69.4 percent. More than a year later, HIV testing rose by 10.2 percentage points com- pared with people seen at facilities that didn't qualify for bonus payments. In fact, there was little effect on the rate of HIV testing for people who weren't living with their sexual partner. The impact was the strongest among people whose partner had HI V/AIDS, their overall spending in 2007. On average, about 60 to 80 percent of the bonuses went to increased pay In couples where one partner was living with HIV/ for clinic workers. AIDS, testing of the other partner increased by 14.7 percentage points. Given the importance of knowing What the evaluation showed is that just giving someone's HIV status in order to get them into AIDS clinics more money may not be as effective treatment or to take precautions to reduce transmission as linking extra money to specific health among partners, these are important gains. activities, like HIV testing. Among all couples, the likelihood that both partners reported that they had been tested-whether previously By creating a comparison group of health clinics that or during this period-rose by 8.6 percentage points, an received similar payments without having to meet any increase of 12.7 percent. performance conditions, the researchers were able to really look at the link between pay-for-performance The payments were structured to promote and improving the rate of HIV testing. They found testing of couples. that clinics did a better job getting couples tested when their payments were linked to the number of people Clinics received US$4.59 for each couple tested, but tested, rather than when they received an additional only US$0.92 for testing an individual. These payments quarterly payment not linked to performance. This evaluation, believed to be the first to examine the impact of pay-for-performance on improving In sub-Saharan Africa.... HIV testing and related services, showed that giving * The number of children newly infected with HIV dropped by providers more money for testing couples worked. 24 percent between 2009 and 2011; The findings contribute to growing evidence that * HIV testing among women is on the rise; paying health facilities for performance is a workable, effective method for improving health system perfor- reAndvan it 6 o pr e n mance. The results also contribute to the relatively small literature on the effects of paying medical care Soune sAD einlFatSetfrsb-SaharanAAfrica 2012 providers for performance in developing countries. The Strategic Impact Evaluation Fund, part of the World Bank Group, supports and disseminates research evaluating the impact of development proects to help alleviate poverty. The goal is to collect and build empirical evidence that can help governments and development organizations design and implement the most appropriate and effective policies for better educational, health and job opportunities for people in developing countries. For more information about who we are and what we do, go to: http://www.worldbank.org/sief. The Evidence to Policy note series is produced by SIEF with generous support from the British government's Department for International Development. THE WORLD BANK THE WORLD BANK, STRATEGIC IMPACT EVALUATION FUND 1818 H STREET, NW WASHINGTON, DC 20433 Produced by the Strategic Impact Evaluation Fund Series Editor: Aliza Marcus; Writer: Roger Fillion