Abstract
Background
Payment for performance (P4P), or financial rewards to healthcare workers contingent on their achievement of pre-defined performance targets, is gaining popularity as a mechanism to improve health service delivery. The underlying assumption is that health workers will respond to incentives by increasing effort, as the offer of additional funds increases their motivation to perform. There has been substantial research into the effects of P4P schemes on service delivery outcomes. However, there has been less research on how these schemes affect health worker behaviour which then impacts on the behaviour and care seeking of patients. In this presentation, we describe the steps taken to build an ABM for a P4P scheme in Tanzania, and highlight the contributions that ABM can make to understanding the behaviour and response of providers and patients to P4P.
Methods
Firstly, we defined the agent classes to be included in the ABM: patients and providers. Managers are modelled simply as system influences. We considered a single behaviour for patients: delivery at a facility or not; and a series of behaviours for providers: charging user fees; prescribing drugs; being kind. These behaviours were selected as they were significantly improved by P4P within an impact evaluation study. We describe the steps taken to build an ABM to examine the effect of P4P on these behaviours.
Results
We found three visualisations helpful in the pursuit of our aim. The first conceptual map connected variables that were found to be significantly affected by P4P and related to our outcomes of interest through statistical analysis of data from an impact evaluation of P4P. To identify other attributes that are relevant to decision making, we reviewed the empirical literature and ran regressions using baseline data from the impact evaluation to determine which patient/provider characteristics were associated with the behaviours of interest. These associations were then incorporated into the first conceptual map. The second conceptual map captured a broader set of connections than the data suggested, to allow for other intuitive links to be explored in the modelling, which is important when considering emergent behaviours and the generalisability of the model to other contexts. Thirdly, we drew decision trees to illustrate the different choices individual patients and providers have when making behavioural decisions affecting the rate of institutional deliveries. The ABM allows for heterogeneity in decision-making in that patients and providers are modelled to have specific characteristics and behaviors based on the data, whilst also allowing for the fact that not all patients will behave the same in response to the same provider in the same health facility.
Conclusions
We have described the steps used to build an ABM to explore how and why some facilities succeeded to increase the rate of institutional deliveries through P4P while others did not, including seeking lessons learned to highlight any early warning indicators of expected success or unintended negative effects. Our framework has generalisable methodological steps for others seeking to use ABM to better understand how P4P affects the behaviour of providers and patients.
Payment for performance (P4P), or financial rewards to healthcare workers contingent on their achievement of pre-defined performance targets, is gaining popularity as a mechanism to improve health service delivery. The underlying assumption is that health workers will respond to incentives by increasing effort, as the offer of additional funds increases their motivation to perform. There has been substantial research into the effects of P4P schemes on service delivery outcomes. However, there has been less research on how these schemes affect health worker behaviour which then impacts on the behaviour and care seeking of patients. In this presentation, we describe the steps taken to build an ABM for a P4P scheme in Tanzania, and highlight the contributions that ABM can make to understanding the behaviour and response of providers and patients to P4P.
Methods
Firstly, we defined the agent classes to be included in the ABM: patients and providers. Managers are modelled simply as system influences. We considered a single behaviour for patients: delivery at a facility or not; and a series of behaviours for providers: charging user fees; prescribing drugs; being kind. These behaviours were selected as they were significantly improved by P4P within an impact evaluation study. We describe the steps taken to build an ABM to examine the effect of P4P on these behaviours.
Results
We found three visualisations helpful in the pursuit of our aim. The first conceptual map connected variables that were found to be significantly affected by P4P and related to our outcomes of interest through statistical analysis of data from an impact evaluation of P4P. To identify other attributes that are relevant to decision making, we reviewed the empirical literature and ran regressions using baseline data from the impact evaluation to determine which patient/provider characteristics were associated with the behaviours of interest. These associations were then incorporated into the first conceptual map. The second conceptual map captured a broader set of connections than the data suggested, to allow for other intuitive links to be explored in the modelling, which is important when considering emergent behaviours and the generalisability of the model to other contexts. Thirdly, we drew decision trees to illustrate the different choices individual patients and providers have when making behavioural decisions affecting the rate of institutional deliveries. The ABM allows for heterogeneity in decision-making in that patients and providers are modelled to have specific characteristics and behaviors based on the data, whilst also allowing for the fact that not all patients will behave the same in response to the same provider in the same health facility.
Conclusions
We have described the steps used to build an ABM to explore how and why some facilities succeeded to increase the rate of institutional deliveries through P4P while others did not, including seeking lessons learned to highlight any early warning indicators of expected success or unintended negative effects. Our framework has generalisable methodological steps for others seeking to use ABM to better understand how P4P affects the behaviour of providers and patients.
| Original language | English |
|---|---|
| Publication status | Published - 14 Jul 2021 |
| Event | World Congress on Health Economics 2021 - Online Duration: 12 Jul 2021 → 15 Jul 2021 https://healtheconomics.confex.com/healtheconomics/2021/meetingapp.cgi/Home/0 |
Conference
| Conference | World Congress on Health Economics 2021 |
|---|---|
| Period | 12/07/21 → 15/07/21 |
| Internet address |