Structural, institutional and organizational factors associated with successful pay for performance programmes in improving quality of maternal and child health care in low and middle income countries: a systematic literature review

Background Pay for Performance (P4P) mechanisms to health facilities and providers have been implemented in several low- and middle-income countries (LMIC) to improve maternal and child health (MCH). These are tied to predetermined quality and quantity indicators. There is limited synthesized information on the structural, institutional and organizational factors that influence the success of P4P programmes with respect to quality of care. This review, which builds on a previously published review sets out to synthesize existing literature on the factors that influence the outcome of P4P programmes and quality of care. Methods A literature review was conducted of published studies documenting implementation of P4P interventions and quality of care in low and middle income countries. Records published between June 2014 and September 2017 were selected and combined with articles from January 1990 to June 2014 previously identified by colleagues. Results 13 studies were included in the final analysis. The majority of studies found a positive impact on quality of care scores and at least one study showed significant reductions in mortality outcomes in newborns. One study from Afghanistan did not show any positive effects. Structural factors associated with likely success of P4P programmes included: explicit acceptance and understanding by health workers; limiting the number of indicators measured with inputs from health workers. Organisational factors included sufficient incentive payments. Notably the main positive outcome identified was facility financial autonomy from additional payments. Verification by external assessors revealed no major manipulation to achieve payment trigger levels. The primary institutional factors identified that P4P programmes fared better when introduced alongside other health reforms and increased funding. Conclusions This review has found that P4P is not a uniform intervention, but rather a range of approaches with a substantial variation and complexity in how programmes incorporate quality of care considerations. P4P has shown to have an impact on the quality of a number of limited aspects of maternal and child health care. Further research is needed to understand whether additional aspects of the quality of MCH care could be positively influenced by P4P programmes and how health worker motivation and acceptance are linked to this.

VIEWPOINTS RESEARCH THEME 6: PERFORMANCE-BASED FINANCING

RESULTS
Searches from the databases and others resources identified 155 records. Screened records were 82 after removing duplicates and excluding records that did not mention P4P and quality. From 12 articles eligible for fulltext assessment, only 5 were included in the review. Details of the study selection are given in Figure 1. Table 1 outlines the characteristics of the studies included in this review, including those identified by Das [2]. 13 studies (including 8 from Das and colleagues review) and 7 Impact Evaluations of P4P programmes were identified that investigated the effect of P4P on quality of maternal and child care in low-and middle-income countries. These studies indicated that P4P did positively affect the quality of maternal and child care to varying levels. Many of the studies found positive effects. For example, in P4P districts in Afghanistan providers spent more time with patients; conducted a more complete history and examination and provided more counseling [13]. The Philippines demonstrated a 7%-9% improvement in General Self Reported Health and age adjusted wasting over time in the P4P group. Authors estimated the large impact of higher quality care with 294 cases of wasting averted and 229 more children reporting at least good health [1]. Talukder et al. found average quality of care scores to be higher in the intervention sites, and that the visits conducted by the quality assurance groups acted as refresher trainings for the providers [14]. In Burundi, both the average quality score and the number of women having institutional deliveries increased significantly in the P4P group [7]. In addition patients' chance of feeling cured was higher under P4P programme in Burundi [8]. Van de Poel also estimated that deliveries increased in a public facility by 7.5% [12]. Significant improvements in the P4P group were also seen for institutional deliveries and preventative care visits for child in Rwanda [6].

Study characteristics
The Plan Nacer programme in Argentina demonstrated a significant positive effect on increasing prenatal visits and provision of tetanus toxoid as well as a very significant reduction in neonatal mortality (74%) in the beneficiary group. Interesting, there was also a positive spillover effect with an overall 22% reduction in neonatal mortality (beneficiaries and non-beneficiaries) using the same clinics [11]. In the DRC, 5 out of 6 indicators related to patient perception of quality improved in the P4P sites, some significantly [9].
On the other hand, some studies found that P4P did not show any demonstrable effect on certain indicators. For example, in Afghanistan there was found to be no difference in improving skilled birth attendance or postnatal coverage between intervention and control districts [13]. In addition, the study in Burundi did not find any effect on the use of vaccinations or modern family planning. In Cambodia, P4P did not have a significant effect on antenatal care or vaccination [12].
Overall, the studies revealed the following key elements that contribute to the successful impact of P4P on quality of maternal and child-care.

Structural factors
Perception and acceptance of P4P by health workers 3 studies included in this review discussed the importance of health worker attitudes towards P4P with the literature indicating the need for consultation with, and buy in from the health workers in order for the programme to have an impact. In Egypt, Huntingdon et al. conducted interviews with physicians in the Primary Health Care Units and the district health care officers where the P4P scheme was implemented. The results revealed mixed feedback on the design and functioning of the incentive payment scheme. VIEWPOINTS RESEARCH THEME 6: PERFORMANCE-BASED FINANCING   Healthcare providers voiced concerns that national level decision makers without consulting local administration selected indicators and that too many indicators were used to calculate incentives. There were also problems with delays in receiving incentives that created an atmosphere of distrust and uncertainty [10]. In another study conducted in the DRC, health workers from the P4P group complained about the P4P system and the frustration they had from the inefficiency of their strong efforts to increase the demand -"If there is no patient, we can't do more than working 26 days" [5].
In Afghanistan, when health workers were surveyed only 37.9% in the P4P sites recognized that they had received any payment for P4P intervention even though 86.7% of the P4P health facilities reported that they had received performance payments [13].
Consultation with the health providers on the identification of suitable indicators, transparency on how incentives will be calculated and timely disbursement of payments would result in clearer understanding and ownership with the potential of improved quality of care outcomes.

Health worker motivation
It has been suggested that P4P would lead to improved quality of care by motivating health care providers. Of the 13 studies included in this review half considered aspects of health worker motivation and its impact on quality of care within a P4P programme. The evidence from these studies does not necessarily support the view that motivated health care workers will deliver better quality of care. Indeed the literature indicates there is a more complex relationship between incentives and motivation. Engineer and colleagues in Afghanistan suggest that the linkages between payment and motivation of workers to improve targeted services require more finely-tuned understanding of human motivation, as well as more sophisticated approaches to managing organizations and individuals beyond performance payments (eg, taking into account organizational culture, leadership, management and psychology, among other things) [13].
In the DRC, it was found that the introduction of financial incentives led to concrete changes in health workers behaviors. For example, health workers were found to be present at the health facility more often, they organised more preventive health sessions at the facility and conducted more community outreach to sensitize the population on the services offered [5]. The study in Rwanda found similar results, the incentive payment gave providers the motivation to translate their prenatal care knowledge into better practice [6]. Another study has also demonstrated the positive effect of measuring quality without incentives, whereby the act of measurement and feedback in itself led to improvement from awareness and consequent motivation to perform better [3,4].
Examples of motivational outcomes from 3 studies are summarized in Table 3.

Indicators and quality measures
It is vitally important to identify indicators and quality measures that are meaningful, measurable and based on best practice clinical protocols. The types and numbers of quality indicators measured varied in the studies identified and included quantitative and qualitative outcomes.

VIEWPOINTS RESEARCH THEME 6: PERFORMANCE-BASED
FINANCING Some examples of quantitative measures include: volume of services, child vaccination rates, contraceptive prevalence rates, institutional delivery rates, prevalence of low birthweight, neonatal mortality, wasting, use of priority services such as beginning antenatal care within first 20 weeks of pregnancy. Qualitative measures included patient satisfaction and health worker satisfaction and motivation.

Monitoring and verification
The majority of the studies in this review (9 out of 12) examined the need for monitoring and verification within P4P programmes. Measuring change in quality can be difficult, time consuming, costly and subjective. To overcome these challenges, Peabody and colleagues in the Philippines found Clinical Performance Vignettes to be valuable tools as they provide a detailed measure of the clinical encounters they capture [3,4].
Independent assessors are vital to ensuring verification of data. In Afghanistan the monitoring and verification systems used were quite comprehensive. The total amount of financial incentive paid was adjusted by a quality score based on a National Monitoring Checklist (NMC), which was assessed quarterly by an independent team of provincial officers and consisted of items related to equipment functionality, drug availability, quality of medical charts and number of households visited by Community Health Workers. Health facilities submitted monthly reports on the volume of services provided, which were verified quarterly by independent monitors, record-matching and random home visits of patients reported as service users. Systematic audits of 1100 Health Facility visits verified over 95% of the medical records used for payments, and random sampling of over 29 000 household visits based on medical records verified 89% of the reported services. The community and Health Management Information System verification analysis suggested that there was no major manipulation of the payment triggers by the health facilities, suggesting that the reporting of results for payments was likely to be largely genuine [13]. In Van de Poel' s study there also was no evidence of over reporting in response to financial incentives [12].
Other researchers also utilised independent and blinded assessors in their studies [4,9]. These assessors and interviewers as well as being independent required specific training to ensure adequate capacity to assess quality [4,7].

SuPerviSion of, feedbACk to And motivAtion of heAlth workerS Study
Approximately 50% of providers in the intervention districts reported the benefits of teamwork to ensure appropriate distribution of responsibilities as well as to improve quality of care compared to only 6% in the control districts. Health providers in the intervention districts were twice as likely to receive periodic supervisory visits.
Talukder et al, 2015 [14] No difference found in indices for motivation and job satisfaction in either the intervention or the control group. The level of performance of health workers was not communicated back to them in either group Engineer et al, 2016 [13] PBF schemes brought about a significant increase in job satisfaction and a decrease in attrition, but had no significant effect on motivation.
Shen et al, 2017 [15] PBF -performance-based financing VIEWPOINTS RESEARCH THEME 6: PERFORMANCE-BASED FINANCING In Argentina' s Plan Nacer payment of financial incentives were divided with 60% of the maximum payment disbursed monthly based on the number of verified registered beneficiaries; and up to 40% of the maximum transferred every 4 months after verification and certification that the province actually met the quality targets [11].
Results from Egypt are suggestive that care providers do respond to incentives but they must be carefully integrated into a well-known and established quality of the care monitoring system [10].

Financial incentive arrangements
The studies in the review revealed a number of financial incentive arrangements from bonuses directed towards individual doctors [3,4] to incentives paid directly to facilities [5] or provinces [11].
In Rwanda, Basinga [6] and colleagues demonstrated larger effects on services for which facilities receive larger financial incentives and those over which the provider has greater control (eg, prenatal care quality and tetanus vaccination during a prenatal care visit) and are less dependent on patients' health-seeking behaviour (eg, timely prenatal care visits). This finding was supported by Van de Poel [12] who hypothesied that this was likely to be due to the marginal cost of finding and convincing pregnant women to come for regular check-ups that is high compared with the small monetary incentive. On the other hand where the financial incentives was higher, for example for institutional delivery (particularly when implemented as a per case payment) and the health worker had to exert less effort (it is easier to encourage women who have already come into contact with the facility to give birth in it) the impact of the incentive was greater [7].
Huillery and Seban in the DRC also noted that the autonomy of payment allocation among facility staff in the P4P group led to a more egalitarian distribution of payments among workers [5]. P4P benefitted non-technical workers (pharmacists, managers, secretaries, receptionists and maintenance workers) who are not in the governmental payroll and therefore do not receive a share of the fixed payment but who can all contribute to the quality of child and maternal care. Health Facility managers in Afghanistan distributed the performance incentives in a range of ways, which included giving individual bonuses proportional to the health worker' s salary, giving them in equal amounts to all staff, or giving them based on their determination of an individual' s contribution [13].
As seen in the study from Afghanistan, other aspects that were considered in the provision of incentives were baseline conditions and expected improvements. The NGOs delivering the services negotiated with the MOPH to adjust their payment to account for the differences in insecurity and geographical inaccessibility that varied by facility [12].
The size of the incentives paid for services varied between studies, for example in Afghanistan the bonus amounts paid were initially about 6%-11% above the base salary, and increased to about 14%-28%, depending on the health worker' s cadre [13]. In the Philippines the bonuses were equal to about 5 percent of a physician' s salary [3,4].
Countries allocated different incentive amounts to various services. In Rwanda, the highest payment was for institutional deliveries (US$ 4.59), whereas the payment rate for an initial prenatal visit was only US$ 0 · 09 [8]. Argentina utilised a different approach by equally dividing the performance payment among ten indicators, with 4% assigned to each, totaling up to 40%. If the target is met, the province receives the full 40% percent of the capita for that indicator. If it does not meet the target, it receives nothing for that indicator [11].
Performance related payments were generally made every four months [11,13].
The methods by which the total incentivised payment amount was calculated varied in study sites. For example, in Burundi the total payment to a facility was calculated as a weighted sum of the number of provided services in the previous 3 months times their unit payment multiplied by the quantity bonus, which ranged between 1 and 1.25 depending on the score obtained from evaluation of facilities [7,8].
Peabody and colleagues perceived that quality effects seen with incentives provided to individuals may also be possible through indirect financial incentives that operate at the system level. These effects on quality affected performance earlier and to a greater degree than measurement and feedback of performance alone [3].
VIEWPOINTS RESEARCH THEME 6: PERFORMANCE-BASED

Country context
The extent to which the P4P scheme actually had on the improved quality of care has to be viewed within the economic, policy and overall context of the country. 4 studies in this review cited specific contextual issues. Basinga and colleagues in Rwanda note that the P4P scheme was implemented in the context of a larger health sector reform [6]. In the Philippines it is important to note at the time of the study, the increase in the prevalence of wasting was due to severe weather disturbances (hurricanes) in 2006 that affected food supply, shelter, and infrastructure and led to outbreaks of waterborne diseases [4]. In Cambodia and Burundi, the introduction of P4P schemes, as in most other contexts, was accompanied by an increase in budgets [7,12].

DISCUSSION
This review reports the synthesized findings from 13 studies and 7 Impact Evaluations on the structural, institutional and organizational factors associated with successful P4P programmes in improving quality of maternal and child health care in low-and middle income countries.
In general, the review suggests that P4P approaches to health delivery can be effective at improving both coverage and quality of targeted maternal and child health services. However, the improvements achieved are not uniform and can be seen in coverage of preventive services in some programmes and for some conditions but not others.
There has been concern that P4P programmes may negatively affect outcomes that are not incentivized.
Most of the studies in this review did not address this issue. However, a recent Impact Evaluation of P4P programme in Zimbabwe [16] found that none of the non-incentivised services investigated showed a decline in the number of cases treated, which would be the case if task shifting was occurring and affecting these services.
The perception and acceptance of P4P programmes by health workers needs careful consideration during planning and implementation. Early consultation with health workers regarding which indicators are to be measured and how the incentive will be calculated could prevent issues seen in Egypt where health workers expressed frustration at having these decisions made at the National level [10]. In addition, the overall number of indicators measured needs to be carefully considered and should cover all aspects of quality and not focus on structural quality as was found in the review conducted by Gergen [17]. Checklists seem to increase in length with time [18]; deliberate review of checklists is required to prevent them becoming too long and cumbersome. Lack of understanding can undermine the potential impact of P4P programme by limiting the behavioural response of health workers. In addition, clear communication about the structure of P4P programmes to health workers will likely improve the acceptance of them. In this regard, careful thought should be given to select indicators that will be acceptable to providers but can also maximize the efficiency of spending.
Many researchers have investigated health worker motivation and there is evidence showing that direct incentives such as P4P as well as organisational incentives such as supervision combined with institutional rewards or punishments do lead to improved quality of care [19]. Qualitative work conducted as a part of the Impact Evaluation of Afghanistan showed that the P4P programme was a good motivator even though salaries and incentives were not always received on time [16]. Studies in this review in DCR and Rwanda reported that incentives improved motivation resulting in higher health worker presence at facilities; more facility based preventative sessions and more community outreach [5,6]. However, of concern are findings from Zimbabwe' s Impact Evaluation where health workers despite being motivated by incentives expressed their dissatisfaction with the size of incentive relative to their tasks and overall higher workload. This may result in a decline in effect of the incentive as time progresses [16].
Financial arrangements for incentive payments were varied, both in size and recipient. One of the main positive outcomes identified was the autonomy provided to facilities by some programmes (Argentina, DRC, and Zimbabwe). These countries welcomed the ability to distribute the incentive payments in an egalitarian manner among facility staff as well as being able to better allocate scarce resources to best suit their needs [11,16]. The health facility is then able to utilise the incentives to address broader health systems challenges such as drug availability.

VIEWPOINTS RESEARCH THEME 6: PERFORMANCE-BASED FINANCING
Another obvious but important aspect is the need to have adequate levels of incentives or else there may be a limit to the possible gains that can be achieved through P4P programmes as health workers may not feel the added effort is worth the reward. In Misiones province of Argentina the strongest evidence for sustained impact from P4P was seen with a substantial 3-fold increase in incentives [16].
Demand-side incentives need to also be considered in a P4P programme as they can work alongside supply-side incentives. The increase in health seeking behaviour, allows more opportunity for health workers to provide quality care and ultimately impact maternal and child health outcomes. The Impact Evaluation in Afghanistan identified the lack of attention to demand-side considerations as one of the flaws of the RBF pilot implemented there [16].
As seen in the studies from Rwanda, Cambodia and Burundi, P4P programmes have often been introduced alongside other health reforms and increased funding. The results from the HRITF Impact Evaluations suggest that P4P programmes should indeed be part of broader health system reforms and complementary intervention. The programmes can be seen as entry points in tackling wider systems issues.
Monitoring and verification is essential to ensure quantity and quality objectives are being met. Feeding performance data back to providers facilitates performance improvement. The Impact Evaluations reviewed establishes the importance of continued innovations on ways to intelligently measure and incentivize quality measures of care in maternal and child health, which are more complex than coverage indicators. It is suggested that the 'easier' structural quality indicators are addressed first and then programmes can move onto introducing process measures of clinical care. This will allow health providers to address less complex quality of care issues first, develop better understanding of RBF and quality of care, and then shift gradually toward more demanding measures of care under the RBF programmes [16].

Limitations
The focus on examining the quality of maternal and child health care is relatively recent and hence there are only a limited number of published articles. This review examined primarily peer reviewed articles. A limitation of this review is not having access to unpublished findings.

CONCLUSIONS
This review has found that P4P is not a uniform intervention, but rather a range of approaches.
There is substantial variation and complexity in how programmes incorporate quality of care considerations. There are differences in how quality is included in the payment formula, how many and what indicators are utilised in checklists, and how they are measured.
P4P has shown to have an impact on the quality of a number of limited aspects of maternal and child health care and supports the findings of Das and colleagues [2]. In addition to previous findings of an increase in prenatal visits, provision of antenatal tetanus toxoid, institutional deliveries and preventative visits for children aged under 5, a significant reduction in neonatal mortality was found. Patient experience is not a common performance criteria measured though where is has been studied it has been reported to be positively impacted by P4P programmes.
Many of the P4P programmes have some documented or perceived positive spillover effects on individual provider activity and the health system as a whole. From the literature examined, improved generation and use of data are possibly the most important positive spillover effect of the P4P programmes.
Further research is needed to understand whether additional aspects of the quality of maternal and child health care could be positively influenced by P4P programmes and how health worker motivation and health worker acceptance are linked to this.