Understanding the municipal-level design and adaptation of pay-for-performance schemes across two states of Brazil

Abstract

The design of complex health systems interventions, such as pay for performance (P4P), can be critical to determining such programmes’ success. In P4P programmes, the design of financial incentives is crucial in shaping how these programmes work. However, the design of such schemes is usually homogenous across providers within a given scheme. Consequently, there is a limited understanding of the strengths and weaknesses of P4P design elements from the implementers’ perspective. This study takes advantage of the unique context of Brazil, where municipalities adapted the federal incentive design, resulting in variations in incentive design across municipalities. The study aims to understand why municipalities in Brazil chose certain P4P design features, the associated challenges and the local adaptations made to address problems in scheme design. This study was a multiple case study design relying on qualitative data from 20 municipalities from two states in Northeastern Brazil. We conducted two key informant interviews with municipal-level stakeholders and focus group discussions with primary care providers. We also reviewed municipal Primary Care Access and Quality laws in each municipality. We found substantial variation in the design choices made by municipalities regarding ‘who was incentivized’, the ‘payment size’ and ‘frequency’. Design choices affected relationships within municipalities and within teams. Challenges were chiefly associated with fairness relating to ‘who received the incentive’, ‘what is incentivized’ and the ‘incentive size’. Adaptations were made to improve fairness, mostly in response to pressure from the healthcare workers. The significant variation in design choices across municipalities and providers’ response to them highlights the importance of considering local context in the design and implementation of P4P schemes and ensuring flexibility to accommodate local preferences and emerging needs. Attention is needed to ensure that the choice of ‘who is incentivized’ and the ‘size of incentives’ are inclusive and fair and the allocation and ‘use of funds’ are transparent.

Introduction

Pay for performance (P4P), or the provision of financial rewards based on the achievement of pre-specified performance indicators, is a widely used intervention to improve the availability, quality and utilization of health care. It has been applied in high-income settings such as the UK’s (UK) National Health Service (Campbell et al., 2009; Sutton et al., 2012) and many low- and middle-income countries (LMICs) (Singh et al., 2021; Diaconu et al., 2021; Junior et al., 2022). P4P is typically talked about as a single, uniform intervention. However, there is, in fact, substantial variation in incentive structures and scheme design across settings (Eijkenaar, 2013). There is consensus that these design details are crucial, particularly when determining the effect of P4P incentives on equity (Van Herck et al., 2010).

As P4P incentive design is usually homogenous across providers within a given scheme, there is still limited understanding of how design elements contribute to overall programme outcomes and their distribution. One study (Ryan et al., 2012) found that a design only incentivizing high performers resulted in a pro-rich concentration of payouts. In contrast, a design that incentivized high and moderate performance and improvements in achievement closed the equity gap in payouts. A recent systematic review of P4P schemes found that programme design appeared to affect programme effectiveness, with payment per output adjusted for service quality yielding the greatest positive effects on outcomes (Diaconu et al., 2021). Nonetheless, programme design is typically under-reported within research on P4P (Ogundeji et al., 2016; Kovacs et al., 2020b).

A further consideration is that the design of P4P schemes within a given setting often evolves over time in response to changing priorities or to address constraints or challenges encountered during programme implementation. However, there has been limited attention to documenting these design adaptations and the reasons for them, particularly in LMIC, despite their potential relevance for programme effectiveness. A recent study reported that significant adaptations were made to a P4P scheme during its implementation in response to limitations in the initial programme design and contextual pressures in Zimbabwe (Kadungure et al., 2021).

Brazil’s Improving Primary Care Access and Quality (PMAQ) is a P4P scheme introduced in 2011. The decentralization of Brazil’s health system, prevalent also in other policy areas such as education and environment, allows municipalities to participate in health policymaking, including P4P schemes. The federal government designed PMAQ, leaving some design features for the municipalities to adapt, such as whether and how to incentivize Family Health Teams (FHTs). As a result, PMAQ was unique relative to other P4P schemes elsewhere (typically uniform within a country) (Macinko et al., 2017; Junior et al., 2022). In this study, we take advantage of PMAQ’s local-level adaptability to examine P4P’s design variations within a single P4P scheme and explore how programme design affects implementation in different settings.

The study aimed to understand the design choices made by 20 municipalities across two states in Brazil. We explored why municipalities made certain design choices, the challenges that arose from these decisions and the local adaptations made to address problems in the scheme design.

Methods

Study design and setting

PMAQ was a federal programme making financial payments to municipalities based on the performance of primary healthcare providers. FHTs provide primary care for a catchment population of ∼3500 people per team. Each team comprises at least one physician, nurse, nurse assistant and full-time community health worker (Macinko and Harris, 2015). Municipal participation in PMAQ was voluntary. Participating municipalities could retain PMAQ funds at the municipal level, use the funds to improve facility infrastructure and/or pay incentives to health workers in FHT. PMAQ comprised three rounds of implementation, with Round 1 (November 2011 to March 2013), Round 2 (April 2013 to September 2015) and Round 3 (October 2015 to December 2019). PMAQ covered an increasing proportion of municipalities across rounds (71% in Round 1, 91% in Round 2 and 96% in Round 3) (Secretaria de Atenção Primária à Saúde, 2013, 2014; 2018).

At the beginning of each round, a performance assessment determined the monthly PMAQ payment to be made throughout the round. The assessment covered hundreds of indicators, including infrastructure, structural quality of care, such as drug and equipment availability, processes of care (e.g. the content of antenatal care), service utilization, health outcomes and management activities. The assessment consisted of three components: self-assessment, where teams reported their own performance, accounting for 10% of the total score; routine monitoring (20–30% of the overall score), which was collected from reports submitted by teams to the health information system; and external evaluation which was conducted by external evaluators through their visits to the facility (accounting for 60–70% of the total score) (Kovacs et al., 2020a).

At the municipal level in Brazil, two main bodies are responsible for health policymaking and administration. The first is the municipal health secretariat, which administers health policies and programmes, including PMAQ. The secretariat coordinates and cofinances health programs as well as health services delivery (Adriano Massuda et al., 2020). The second body is the Municipal Health Councils, which are representative bodies responsible for overseeing and monitoring health policies and services at the municipal level. These councils typically consist of representatives from various sectors of society, including government officials, healthcare professionals, community leaders and ordinary citizens. The councils play a crucial role in ensuring accountability, transparency and citizen participation in the planning and implementation of health policies and programs within their respective municipalities (Coelho et al., 2006).

This study was a multiple case study design relying on qualitative data (Paparini et al., 2020). Ten municipalities were selected at random from each of the two states in Northeastern Brazil: Pernambuco and Paraíba. Table 1 presents the characteristics of our sample municipalities, showing substantial variation in the population size, number of FHTs and socio-economic status.

Conceptual framework

We used two conceptual frameworks to guide our analysis. First, we used a P4P design typologies framework to describe and compare the design choices across municipalities (Kovacs et al., 2020b). Design features refer to the variables across which the PMAQ configuration changed among municipalities. We refer to the decisions made by municipalities regarding the allocation of PMAQ funds as their ‘design choices’ (Box 1). Some design features were determined at the federal level. These features include the safeguard-against-gaming, which was built into the external assessment; the FHT being the unit whose performance was assessed; and the assessment indicators (Junior et al., 2022). In this paper, while we report on the challenges at both municipal and federal levels, we focus on design modifications at the municipal level.

Second, to describe the design adaptations made, we used the framework for reporting adaptations and modifications to evidence-based interventions (Stirman et al., 2019). We focus on the adaptation process (what is modified and who was involved in the process) and the reasons for the adaptation [the reason and goal of the change, which could be related to tailoring the intervention to the local context (exogenous factors) or overcoming challenges inherent within the design (endogenous factors)].

Data collection

In each municipality, we conducted two key informant interviews with a member of the Municipal Health Secretariat, who was responsible for PMAQ management in the municipality (hereafter management representative or managers), a member of the Municipal Health Council, who was involved in deciding how to allocate PMAQ funds (20 interviews in total), and focus group discussions with two FHTs (40 focus groups in total). The council members in our sample comprised governmental officials (n = 4), healthcare workers (n = 15), service users (n = 1).

Additionally, we reviewed the local municipal laws which set out the incentive design in each municipality. The topic guide covered implementation challenges, modifications of PMAQ and decision-making processes. As this study aims to capture municipal adaptations, perspectives from federal actors were not included. The interviews and focus group discussions were conducted in person in Portuguese and facilitated by one to two researchers. Data were recorded during the interviews and transcribed. All participants consented, and data were anonymized to ensure confidentiality.

Data analysis

We applied thematic analysis to the qualitative data to determine the key challenges in relation to each design feature and identify adaptations made to address these challenges (Braun and Clarke, 2006). Deductive coding relied on the constructs in the conceptual frameworks, such as design features and modifications. We also inductively coded new themes emerging from the data.

All researchers familiarized themselves with the conceptual frameworks and agreed on the initial coding framework, which was piloted by two researchers on three interviews, leading to a revision to the framework. Subsequently, one researcher coded the remaining data and reviewed and discussed emerging themes with the research team through regular meetings. The analysis reached theoretical saturation in the municipalities sampled. Data were managed using NVivo 13.

Results

Design at the municipality level

Table 2 summarizes the initial P4P design features before adaptation, aggregated by the state. Almost all (n = 19) municipalities rewarded FHT workers with bonuses, except for one municipality in Pernambuco, M12, which did not give any bonuses to workers throughout the PMAQ implementation period. Three municipalities (M9, M3 and M13) started giving bonuses in the third round of PMAQ following tension between teams and municipal managers, which led to negotiations.

As for the ‘basis for payment’, all municipalities in our sample received PMAQ funds according to their PMAQ score. This is because this design feature was determined at the federal level. The majority of municipalities (n = 16) paid their teams based on the PMAQ score. However, two municipalities paid the same amount to all FHTs irrespective of the PMAQ score. One municipality, M8, did not report the ‘basis for payment’. Ten municipalities transferred more than 50% of the PMAQ funds to FHT, with the remainder being kept by municipalities. The municipal laws generally did not specify how municipal funds should be used, but interviewees indicated that funds retained by municipalities were typically used to support health facility infrastructure, FHT training and/or management activities. Although the distribution of design features was similar across the two states, the ‘frequency’ of PMAQ incentive payments to teams differed markedly, from a majority receiving payments monthly (n = 6) in the Pernambuco state to a majority receiving payments semi-annually (n = 6) in the Paraíba state.

Clinically trained professionals received bonuses in all municipalities that gave bonuses to staff (n = 19). Fifteen municipalities also paid bonuses to technicians, and eight paid incentives to administrative and support staff. Within teams, clinical staff typically received the highest amounts, while three municipalities divided funds equally between team members irrespective of cadre. Supplementary Appendix 1 details the design features of each municipality in the sample.

Design challenges and adaptations

While many PMAQ design features remained the same throughout the programme, others underwent notable adaptations. In Table 3, we show the challenges associated with different design features as identified by respondents, while design adaptations are shown in Table 4.

What is incentivized and basis for payment

Respondents perceived the external evaluation, which assessed performance in relation to pre-defined indicators, to be unfair and felt that there was a lack of transparency around how the PMAQ score was determined. The most frequently reported challenge was that the external evaluation score was primarily based on the availability of physical resources or staff in the FHT, which was outside the control of the team and reliant on municipal support.

We have terrible infrastructure. […] We work in a rented house, […] Many times, we get a bad grade because of the infrastructure, not because [of] the working process.’ Municipal health council representative (M4).

While many municipalities chose to allocate PMAQ funds to teams based on the performance category they fell into, some municipalities found that paying in this way disadvantaged some teams, especially those with poor infrastructure who received lower performance scores. To address this, four municipalities adapted this feature, distributing PMAQ funds equally among teams. Some managers explained that this was because they believed all teams were putting in equal effort.

As everyone was producing, doing the same, we paid everyone equally (Management representative, M18).

However, one manager disapproved of paying teams equally because it disassociated the payment from team performance, potentially undermining team motivation. One municipality shifted from initially paying equally to paying based on the PMAQ score, with the aim of explicitly linking payment with performance.

We never made classifications. We pay equally. I’ve always complained “why don’t we encourage teams to perform their duties better”, right? How about if we rank them by the production/work of the teams? The team that produces better receives more. The one that produces less receives less. Each team will strive to work harder. But here it has always been distributed equally. That’s why I suffer a lot to coordinate the teams (Management representative, M7).

It is worth noting that despite the challenges reported with linking the payment to performance scores, in most municipalities, the performance score remained the most common mechanism for allocating funding to teams.

Another challenge related to the lack of transparency of the performance evaluation process, which was compounded by the complexity of the formula underpinning the calculation of the performance score. This was a challenge with the federal programme design reported in many municipalities. Several teams reported dissatisfaction with the lack of feedback on the external evaluation assessment, which left teams uncertain about how to improve their performance in the following rounds. In two municipalities, respondents reported a lack of understanding of the self-reported assessment form. In one municipality, this costs several teams 10% of the total score. One municipality reported an adaptation, whereby the allocation of funding to teams was based on user satisfaction determined through a survey.

A lot of teams had a lower grade because they weren’t responding [to self-assessment]. These teams already lost 10%. […] there are professionals that are committed and they in fact return the self-assessment every 6 months, but there are some who want the money, but do not want the obligations (Management representative, M5).

In some municipalities, dissatisfaction with the external evaluation process at the team level resulted in tensions between teams and managers where teams erroneously believed that municipal managers were involved in the external evaluation. As one manager stated, ‘they [FHT staff] blame the municipality. They think the municipality evaluates’. Some managers reported they had tried to raise concerns about the evaluation with the external evaluators, but without any response. As one manager in M18 explained, they tried to get a grading of a health team changed as they felt it was unfair.

We tried to justify it, but they didn’t open anything for us or reply to it either.

Frequency and payment lag

PMAQ funds were normally transferred from the federal level to municipalities on a monthly basis, but municipalities could decide on the frequency with which to pass funds on to teams. Payment frequency was fairly responsive to staff perceptions and needs with numerous reported adaptations across municipalities. For example, annual payment compounded by lengthy payment delays caused tensions between teams and municipal management, which was resolved by increasing the payment frequency to monthly in M6. However, in other municipalities, there was a reluctance to pay the bonus on a monthly basis to avoid it being perceived as a salary supplement that was disassociated from performance. Indeed, it was stated by the National Department that the PMAQ payments were not intended to be seen as a salary supplement, but as a separate performance bonus. This principle underlined why some municipalities, such as M19, did not want to pay the PMAQ benefits on a monthly basis. As one manager stated:

There are cities paying it monthly, but the Law itself says the PMAQ bonus cannot be paid monthly to not confuse the bonus payment with the salary amount (Municipal Health council representative, M4).

On the other hand, sometimes changes in payment frequency, especially a shift to less frequent payments, caused frustration among some respondents. Some team members also reported payment delays, which was demotivating.

it is something that we used to receive monthly, and then it changed to quarterly, now we do not know, annual. It was a huge gap, more than one year went by without receiving it (Health worker, M5).

Health workers also complained about the time lapse between the external evaluation and when the payment was made, which could be up to a year or longer. This was further compounded by the lengthy intervals between the external evaluations themselves (three evaluations in a 10-year period), which created a disconnect between efforts made and eventual adjustments to bonuses received. Those delays were perceived to have undermined the principle of teams getting rewarded for their effort. This issue was even greater when there was staff turnover (discussed further later). In addition, the infrequent performance assessment gave rise to ‘gaming’ with some teams intensifying their performance around the time of the external evaluation. However, the external evaluation process and frequency were determined at the federal level and were not amenable to municipal-level adaptation.

Now I’m going to be very honest. I think [PMAQ] motivates on time. We see the most committed professionals when they know that the Ministry of Health could arrive at any time to make an evaluation. Once the evaluation happened, I think the staff breath [a sigh of relief] and say’ it already happened (Management representative, M16).

Use of funds

Almost all municipalities in our sample shared the PMAQ payments between team staff and the municipal government. Several concerns were reported by teams regarding the use of PMAQ funds by municipal managers primarily due to issues of transparency of fund spending and perceived unfairness. In M4, it was reported by team workers that it was unclear how PMAQ municipal funds were used by the municipal government. This was partly attributed to a lack of clarity in the municipal law regarding how the management funds could be spent.

the PMAQ Law is very open. It left a lot of loopholes. Some of the resources that were for infrastructure were not being used for it. And the same for the professionals. We used to receive the money in the wrong way because there were different interpretations of the current law (Municipal health council representative, M4).

Everyone knew we weren’t getting anything from PMAQ. It wasn’t coming to us [FHT staff]. So, the funding part is a very obscure thing, it is something that we cannot talk much about it, got it? (Health worker, M5)

Some municipalities that allocated more than half of PMAQ funds to the municipality reported that this created tensions with teams who felt they should receive a higher share of the funds. In M19, many teams refused to join the third round of PMAQ as they were not satisfied with the limited percentage of funds allocated to them. M17 and M3 municipalities increased the share of PMAQ funding allocated to teams following pressure from the teams. In some cases, this required changes to the municipal law to enable the allocation of funds to teams.

There was a Union negotiation, and in this negotiation, the professionals wanted a higher share of funds from PMAQ. The payment to the professionals already happened in the previous cycle, but the professionals wanted a larger amount and there was a very big standstill (Management representative, M19).

In the municipality that did not allocate any of the PMAQ funds to teams, the FHT challenged the management to pay them, and the new management team had promised to reward them. However, this did not ultimately materialize due to insufficient PMAQ funds following low PMAQ scores during the external evaluation, which was attributed to the lack of incentive teams had received from the management in previous rounds.

This current Mayor, he accepted the decision of the previous Mayor, who also did not pay. Previously the current mayor even told the professionals he would pay in this cycle now. But in this last cycle there was a… a great squeeze of resources. There was no way to pay (Management representative, M12).

Adaptations were sometimes driven by a desire for inclusiveness in terms of staff bonus eligibility. In one municipality, the management reduced its PMAQ share by 5% in order to pay bonuses to administrative staff as well as medical staff in teams.

It was 40% (for management), but then, with the joining of general services assistants [to the bonus recipients], the management took 5% off it [their bonus], to be able to reward this class of workers (Health worker, M20).

Some adaptations to the use of funds aimed at improving equity. In one municipality (M4), the management redirected part of the funds from good performing teams to pay teams who performed poorly. In another (M5), the municipal management used their municipal resources to pay poor performers or teams that had been disqualified from PMAQ, in order to improve equity and support teams who were persistently performing poorly due to a lack of resources.

If I have a team that got a low grade and it has a poorer structure and the other had a good grade, the management takes from the one that got a good grade and tries to invest in that worse one that could not improve by itself (Management representative, M4).

In some municipalities, the decision-making process around the allocation of PMAQ funds was influenced by the perceptions of PMAQ as a way of increasing staff pay, of rewarding staff for effort or performance or of improving quality of care.

There was one of the speakers who came to [M9], she said, look, this issue of rewarding the professionals is at the discretion of each city, because between working in a city so hot like this one in a room that has air conditioning, or getting a bonus, I prefer air conditioning to work in comfort, but here [in M10] we tried to, in addition to giving that comfort to the professionals, give the bonus too because this is also a way of encouraging the (healthcare) professionals (Management representative, M10).

Who was incentivized

As most municipalities allocated funding to teams as a whole, challenges in some municipalities arose where team members received PMAQ funds despite limited or no contribution (i.e. free riders). Some municipalities considered evaluating and paying bonuses based on individual rather than team-level performance to address this issue, although they did not implement this change, likely due to the absence of individual performance monitoring data.

We always have those in the teams who have not contributed to anything and yet are receiving. So, we were even discussing whether we would change it to a different format, an individual performance bonus since we heard PMAQ might end. […] Let each one receive according to their own production (Management representative, M5).

In the initial round of PMAQ, bonuses were restricted to clinically trained cadres where payment was passed on to staff, while support and administrative workers (e.g. janitorial and reception) and technical staff were not eligible for bonus payments. Due to the contribution of these support workers to the FHT performance, several municipalities subsequently amended the PMAQ law to enable them to receive bonus payments.

In 2017 we realised the need to also include the janitorial service, which is extremely important within a health unit. So today, in addition to this team that I mentioned earlier, we have janitorial services and receptionists also receiving funds from PMAQ (Management representative, M20).

Community health agents (CHAs) were initially ineligible to receive bonuses within the federal programme design. However, due to pressure from the CHA union, they were subsequently included. However, their inclusion was contested by municipal health managers because it was felt that they contributed less to the performance indicators included in the evaluation than other cadres.

High staff turnover throughout PMAQ sometimes led to the delinking of PMAQ payments from performance, as those who were evaluated were often different from those receiving bonuses. This was a particular issue for staff who were not on permanent contracts. To address this concern, the management in M2 only allocated PMAQ funds to staff who were present during the external evaluation and still employed by the municipality. However, the management representative in M8 commented that this adjustment would concentrate the PMAQ funds among too few staff members.

You evaluate for two years that are already in the past, people have changed, not everyone is permanent, there are new providers. So, in the first year, we made payments for the people who were there, who continued working, based on the previous evaluation. In the second year, the same thing. In the third year, no, we made payments to the ones who were working the year before, even though [the evaluation] referred to two years back(Management representative, M2).

As CHAs typically had permanent contracts, they were more likely to receive PMAQ funds, when these were allocated to staff present during the external evaluation, than those on fixed-term contracts, who may have left at the time of bonus payment. Although this was reported as a concern in some municipalities, no adjustments were made to address this issue.

Incentive size

The incentive size differed across professional cadres in many municipalities, causing concerns about fairness in many municipalities. In some cases, this led to certain cadres refusing to participate in PMAQ. To address this challenge, some municipalities opted to pay each staff member equally irrespective of their profession, recognizing that everyone is contributing in their own way to team performance. This approach was reported to contribute to team building. In other cases, support and administrative cadres did not receive equal amounts to clinically trained professionals.

…it’s not because you’re a doctor that you should earn more than the general services assistant. I mean, the doctor’s not going to do the job for the general services assistant. So, I believe nothing’s fairer than everyone getting an equal share (Health worker, M20).

If the team reached the X grade and the value is X, then all workers get the prize, it is not a salary top up, this award is given equally to all team members. That resulted in fewer conflicts (Management representative, M5).

The number of team members eligible for bonus payments also had bearing on the incentive size. Since the PMAQ bonus was based on the team, rather than individual, performance, most municipalities allocated a fixed amount to teams of the same performance grade. As a result, a large team would receive a smaller amount per member than a team with fewer members, even when both teams achieved the same performance grade. This led to fairness concerns among health managers in M5 and M20. To address this issue, M5 management adjusted the allocation of PMAQ funds for team size such that team members from teams with the same performance grade would receive equal amounts per capita regardless of team size.

Look, here in the city, […] the grades, practically all were the same, right? They had excellent grades. But there’s a team that we know with 15 professionals. There’s another team that has only 10. So, the value, whether liking it or not, is higher. It’s bigger for the one with the least professionals, you know (Management representative, M20).

Decision-making process regarding programme design and adaptation

Municipal managers and health council representatives in their interviews reported that a range of organizations were involved in the negotiations regarding scheme design and adaptation at the municipal level. Unions representing different cadres of health workers played an important role in both municipal- and national-level negotiations regarding the allocation of PMAQ funds. The union that represented CHAs was recognized as being especially powerful, ensuring that municipal laws allowed CHAs to be paid as part of the FHT. Nursing unions were also powerful, and nurses were seen as key to running the health teams.

The nurses and the community health agents are the ones questioning the most. The other professional ranks go with the flow. Social Workers too (Management representative, M4).

Most municipal managers and council representatives reported that negotiation around the design of PMAQ sometimes led to discontent and resulted in conflict within health teams and between teams and managers and at times health councils. Design features which were especially contentious included ‘what was incentivized’ (the reliance on the performance score largely determined by the external evaluation), the ‘use of funds’ and ‘who was incentivized’.

There was a Union negotiation, and in this negotiation, the professionals wanted a higher share of the PMAQ funds (Management representative, M19).

The extent of conflicts regarding incentive design varied. In one municipality, unions representing dentists and nurses, unhappy about how funds were managed, advocated that their members did not participate in the third round of PMAQ. Exerting pressure by threatening not to participate in future rounds of PMAQ was a common strategy adopted by teams, with nurses playing a key role.

The nurse and dentist did not accept to earn less, having the same grade, as the medical professional. And the doctors did not accept to lower the amount they received, the percentage. And then a… a mess (Management representative, M19).

In one municipality, teams exerted pressure on municipal managers to sign up to PMAQ. However, the management then did not support teams with PMAQ and offered no guidance to teams to prepare for the external evaluation. In other cases, pressure from teams and unions resulted in their direct involvement in creating the municipal law together with municipal managers.

The battle happened in the second cycle because the other government management had promised to pass on the funds [.] [but] they (team) didn’t receive anything, so they didn’t want to sign [to join PMAQ in the next round]. So, we started creating the municipal law so I participated in the process of making the law. We held meetings in 100% of the teams listening team by team for us to make up this law (Management representative, M4).

Introducing design changes required consensus across members of the municipal council, which included municipal health managers, staff members and civil society organizations. Where such consensus was lacking, it was difficult to bring about change. One manager complained that they were unable to make modifications that they felt would improve the programme partly because they were the minority at the health council and ‘the majority always wins’. However, introducing changes to some design features created confusion among health workers.

[the municipality] has autonomy to distribute [funds] as they think it should be. (…) they created the law, then they changed the law. That’s why we can’t really know the way the [payment] transfer is done. Sometimes they make annual transfers, sometimes semi-annual (Health worker, M5).

There was evidence of diffusion of information and experiences with different designs across municipalities. Several managers reported that staff and unions were aware, or had heard rumours, of what was happening in other municipalities, which sometimes influenced their decision-making.

The Unions brought a lot of examples from other cities including from the metropolitan region and the countryside that started to use the PMAQ bonus paid to professionals monthly, like a pay rise (Management representative, M19).

Upcoming elections, or changes in political leadership, were also identified as reasons for implementing or blocking certain adaptations.

Discussion

This study aimed to understand the design choices made by municipalities and subsequent challenges and design adaptations. We also explored the underlying decision-making process regarding incentive design. To our knowledge, this is the first study to comprehensively describe the design of PMAQ at the municipal level and to examine challenges associated with incentive design features and subsequent modifications together with drivers for these at the local level.

We found substantial variation in the design choices made by municipalities in terms of ‘who was incentivized’, the ‘size’ and ‘frequency of bonus payment’. While the team performance score determined both the level of funding received by municipalities and the allocation of funds across teams in most cases, some municipalities changed payment rules to improve equity and reduce perceived unfairness among primary care workers.

We identified two key design features that led to challenges and adaptations—‘who was incentivized’ and ‘incentive size’. When perceived as unfair, these design features resulted in tension within and between teams from different family health units. For example, if key players were left out of bonus payments, or if the size of the incentive was based on, e.g. cadre rather than the level of contribution to performance targets, tension led to poorer performance and staff or teams dropping out of PMAQ. Equally, when these features were considered inclusive and fair, this resulted in team cohesiveness and bonding. This finding is consistent with qualitative studies in Tanzania and Brazil that found perceived unfairness of P4P often demoralized team members, adversely impacting performance (Chimhutu et al., 2016; Silva et al., 2023). Furthermore, we found that adaptations were mainly motivated by a sense of universality or inclusiveness and equity, ensuring that no one was left out in the allocation of funds. However, despite the importance of these features, they often receive relatively little attention in the design phase of P4P programmes.

Other design features of note were ‘what is incentivized’ and ‘the basis for payment’. While most municipalities used the performance score as the basis for payment, there was a high level of dissatisfaction with the external evaluation process, which accounted for 50–60% of the overall performance score. In particular, teams were dissatisfied with the assessment’s focus on infrastructure, the lack of clarity and transparency over what would be assessed, and the communication of performance results. While the performance assessment process determining the level of funding allocated to municipalities was outside of the control of municipalities, the municipalities could choose to use different criteria to allocate funding to teams. However, in practice, only one municipality proposed alternative performance criteria by which to assess teams (based on patient satisfaction). Others opted to remove the performance element altogether and allocate equal levels of funding to teams, which had its own challenges. Fairness was again the motive behind adaptations to the ‘use of funds’, particularly when healthcare staff did not approve the division of funds between the municipal management and the teams.

Furthermore, two contextual issues emerged as being influential in terms of whether design features resulted in challenges and the effectiveness of design adaptations. First, where facilities did not have the available infrastructure, they performed poorly and the design adaptations generally failed to address this. The external evaluation was skewed towards structural quality, and facilities with limited facility infrastructure struggled to achieve a high-performance score even if they put in a lot of effort to improve service delivery. Several municipalities tried to address this by ensuring that poor-performing teams also received PMAQ funds, although these funds were usually paying health workers rather than investing in facilities. By default, the PMAQ scheme did not seek to improve facility readiness nor guarantee that a minimum amount of funds be used to support facility operations or infrastructure. Nonetheless, municipalities could decide to allocate funding towards this for selected teams. However, we were unable to determine to what extent municipalities using funding in this way from our data. Yet, the relevance of facility readiness in the context of PMAQ is aligned with a realist review which identified facility readiness (staffing levels and quality of facility infrastructure and equipment) as a key contextual factor underpinning programme effectiveness (Singh et al., 2021).

The second contextual factor is transparency in relation to governance at the municipal level. Where there was a lack of transparency about the use of PMAQ funds at the municipal level, this created distrust and tension among teams in relation to the programme design, an occurrence which was more likely when laws were ambiguously worded. However, this was generally not addressed within design adaptations as it was an issue outside the scope of PMAQ. The importance of transparency and the extent to which health system hierarchies promoted this were highlighted elsewhere as a key contextual factor relevant to P4P (Fillol et al., 2019). While P4P schemes can sometimes improve the governance function of health systems where incentives align with this goal (Mayumana et al., 2017), the PMAQ experience shows that this is not guaranteed.

We found that some design choices presented trade-offs of relevance for policymakers. First, in terms of payment frequency, monthly payments could have been perceived as a salary rise, potentially disassociating the bonus from performance. In contrast, annual payments were generally seen as not frequent enough, causing frustration. However, there was a general preference for monthly payments among staff, and many municipalities tried to accommodate this. Second, a balance needs to be struck between allocating funds to providers vs funds to support service delivery. Most municipalities chose to allocate funds to staff and facilities, with only one municipality opting not to pay workers and no municipalities only paying workers. Bonus payments to workers appear to be a key element driving performance in the Brazilian context, with other research showing that performance was significantly associated with the allocation of PMAQ funds to the staff (Fardousi et al., 2022).

The study has several limitations. First, it was based on interviews with respondents who interacted with PMAQ over multiple points of time, and therefore, some further adaptations may have occurred after data collection in some municipalities. Second, we may have encountered recall bias, where participants may have focused on the most recent challenges, overlooking earlier issues. The adaptations and challenges documented in this study do not encompass all potential adaptations to PMAQ in Brazil. Municipalities not included in our sample, particularly those with a population below 10 000, might have implemented different adaptations. PMAQ implementation underwent dynamic changes with staff and management turnover design changes, especially in areas where there was no law or ambiguous wording of provisions. Social desirability bias could potentially have diluted some of the findings. Lastly, our study reports the challenges associated with incentive design choices from a health system perspective. The health outcomes resulting from those choices are important priorities for future research (Fardousi et al., 2022).

Conclusion

This study provides insights into P4P design choices and their implications at the local level in Brazil, with lessons for such schemes elsewhere. The significant variation in design choices across municipalities and FHTs’ response to them highlights the importance of considering the local context when designing and implementing P4P programs, as well as ensuring that such schemes are sufficiently flexible to accommodate local preferences and emerging needs.

Particular attention should be paid to ‘who is incentivized’ and the ‘size’ of incentives within P4P design to ensure inclusivity and perceived fairness, maximizing programme success. Furthermore, our study highlights the critical role of transparency in the allocation and use of funds to build confidence in scheme integrity and fairness. We demonstrated the value of two frameworks for studying P4P design. We encourage further analysis and reporting of P4P design within future research and documentation of adaptations and the reasons for these in other settings.

Supplementary data

Supplementary data is available at Heapol Journal online.

Data availability

The data underlying this article cannot be shared publicly to ensure confidentiality of the participants. The data will be shared on reasonable request to the corresponding author.

Funding

Funding for this study was provided jointly by the Medical Research Council (MRC), Newton Fund and the Brazilian National Council for the States Funding Agencies (CONFAP) under the UK to Brazil Joint Health Systems Research Call (grant MR/R022828/1). The MRC grant was awarded to J.B. and T.P.-J. Funding from CONFAP came from Fundação de Amparo à Pesquisa do Distrito Federal, Fundação de Amparo à Ciência e Tecnologia do Estado de Pernambuco and Fundação de Apoio à Pesquisa do Estado da Paraíba. CONFAP funding was awarded to E.D.S. The funders had no role in study design, data collection and analysis, publication decisions or manuscript preparation.

Acknowledgements

We are grateful to Allan Nuno Alves de Sousa, Olivia Lucena, Davllyn Anjos, Ilano Barreto and Wellington Carvalho for their valuable comments and insights throughout the project. We would like to thank the study participants who shared their PMAQ experiences.

Author contributions

N.F, J.B., G.D.G.J. and H.S. contributed to the conceptualization of this study. Formal analysis and interpretation was conducted by N.F. with support from J.B., G.D.G.J. and H.S.; M.O.S.d.S. and G.D.G.J. collected the study data; N.F. and J.B. produced the first draft; critical revision was provided by H.S., K.S.d.B.S., M.O.S.d.S., A.F.B.B., J.S., G.D.G.J., E.d.S., L.G. and T.P.-J. All co-authors gave the final approval of the final version to be submitted.

Reflexivity statement

This study is part of a collaborative research project between several Brazilian institutes and the London School of Hygiene and Tropical Medicine. The research gave voice to a diverse range of actors involved in the design of PMAQ, including front line primary healthcare workers, municipality managers and council representatives. The co-authors of this paper represent diverse backgrounds in discipline, gender and seniority levels. Six co-authors are female (H.S., K.S.d.B.S., M.O.S.d.S., A.F.B.B., J.S. and J.B.). Five co-authors are male (N.F., G.D.G.J., E.d.S., L.G. and T.P.-J.). Nine co-authors are Brazilian and based in Brazil, with two co-authors based in the sampled states—Pernambuco (K.S.d.B.S.) and Paraíba (L.G.). While the first co-author is based at the London School of Hygiene and Tropical Medicine (LSHTM), he is an LMIC national (Syrian). T.P.J. and J.B. are based at LSHTM and have extensive experience in health systems and policy research in LMICs. N.F. and M.O.S.d.S. are early career researchers.

Ethical approval.

The study received ethics approval from the University of Brasilia (Brasília, Brazil; CAAE 30424620.4.0000.8093) and the London School of Hygiene & Tropical Medicine (London, UK; 15805).

Conflict of interest:

None declared.

References