Determinants of Health Care Decisions: Children's Health in Mali
Lead Research Organisation:
Innovations for Poverty Action
Department Name: Principal Investigator
Abstract
The delivery and funding of primary healthcare in developing countries, particularly for children, is the subject of sustained debate -- not least in light of child morbidity and mortality rates that continue to fall short of the Millennium Development Goals. Following evidence of increased utilization and improved health outcomes under free care, many African countries have recently fully or partly abolished user fees. Arguably, however, free care can lead to overuse and moral hazard, for example by reducing prevention efforts. There is a tension between subsidizing care to reduce barriers to access arising from market failures, and maintaining the role of prices in efficiently directing resources to those who value them most.
The goal of this project is to study the effect of providing free healthcare and free health worker visits on three important decisions parents make for the health of their children: the time they wait before seeking care in an acute spell of illness, use of preventive care, and the use of formal vs. informal sector care. The combination of health worker visits with free care has the potential to increase access to care for poor families while reducing the inefficiencies associated with abolishing user fees - such as the overuse of primary care and the underuse of preventative care: health workers may provide better information to parents about the need for care, and also 'nudge' them towards better healthcare practices.
We study the interaction of these policies with three key characteristics of households which may lead parents to underuse care for their children if they face the market price: financial constraints and an inability to borrow or save; insufficient information about health and healthcare; and the possibility that parents' spending priorities differ from those of a policymaker. The channels through which policies affect decisions are important in understanding the efficiency gains and losses from abolishing user fees and the potential of health workers to counteract some of these losses.
The data for this project comes from a randomized controlled trial (RCT) conducted in a peri-urban area of Bamako, Mali. The RCT consists of two interventions; free primary care for children and free health worker visits. For each child we collect uniquely detailed data on socio-economic background as well as a complete health history over 10 weeks - detailing on a daily basis all symptoms that a child exhibits, and all healthcare sought by the mother. This data allows us to study in detail how free healthcare and health worker visits affect the healthcare behaviors of parents, and how the effects depend on the household's financial resources, its knowledge and beliefs about health issues, and its preference. The health worker intervention is in this context particularly interesting because it may on the one hand be able to address inefficiencies more directly and therefore be a viable alternative to free care, and on the other it may provide a means to counteract some of the inefficiencies of free care.
Our study focuses on the effect of policy on the healthcare choices that families make, as well as on healthcare outcomes. Because of this, it will help us to understand the constraints that prevent different types of families from obtaining high quality care. This in turn will allow us to develop a model which we can used to estimate the impact of policies other than those we test directly - for example a partial abolition of fees, or means tested subsidies. In this way, our approach will make a significant contribution to the evidence-based design of healthcare policy
The goal of this project is to study the effect of providing free healthcare and free health worker visits on three important decisions parents make for the health of their children: the time they wait before seeking care in an acute spell of illness, use of preventive care, and the use of formal vs. informal sector care. The combination of health worker visits with free care has the potential to increase access to care for poor families while reducing the inefficiencies associated with abolishing user fees - such as the overuse of primary care and the underuse of preventative care: health workers may provide better information to parents about the need for care, and also 'nudge' them towards better healthcare practices.
We study the interaction of these policies with three key characteristics of households which may lead parents to underuse care for their children if they face the market price: financial constraints and an inability to borrow or save; insufficient information about health and healthcare; and the possibility that parents' spending priorities differ from those of a policymaker. The channels through which policies affect decisions are important in understanding the efficiency gains and losses from abolishing user fees and the potential of health workers to counteract some of these losses.
The data for this project comes from a randomized controlled trial (RCT) conducted in a peri-urban area of Bamako, Mali. The RCT consists of two interventions; free primary care for children and free health worker visits. For each child we collect uniquely detailed data on socio-economic background as well as a complete health history over 10 weeks - detailing on a daily basis all symptoms that a child exhibits, and all healthcare sought by the mother. This data allows us to study in detail how free healthcare and health worker visits affect the healthcare behaviors of parents, and how the effects depend on the household's financial resources, its knowledge and beliefs about health issues, and its preference. The health worker intervention is in this context particularly interesting because it may on the one hand be able to address inefficiencies more directly and therefore be a viable alternative to free care, and on the other it may provide a means to counteract some of the inefficiencies of free care.
Our study focuses on the effect of policy on the healthcare choices that families make, as well as on healthcare outcomes. Because of this, it will help us to understand the constraints that prevent different types of families from obtaining high quality care. This in turn will allow us to develop a model which we can used to estimate the impact of policies other than those we test directly - for example a partial abolition of fees, or means tested subsidies. In this way, our approach will make a significant contribution to the evidence-based design of healthcare policy
Planned Impact
Our proposed research will have an impact in four distinct groups of beneficiaries. First, as discussed in the 'Academic Beneficiaries' section, we expect our findings, data and methodology to make significant contributions in the fields of health, development and behavioral economics, as well as public health and demography.
Second, we anticipate that our research will benefit policy making bodies at a local, national and international level. Improving healthcare access for mothers and children are two of the most important aims of global development policy, and are included in the Millennium Development Goals. Child health is a particularly urgent concern in Mali, where the under-five mortality rate is the second-highest in the world. Our project is motivated by the active policy debate about how to fund healthcare in developing countries. The consensus in the 1980's, endorsed in the 1987 Bamako initiative, was that user fees were the best way to deliver high-quality service and prevent waste and over-utilization. Free care could encourage overuse and moral hazard (behaviors that inefficiently increase the cost of care, such as foregoing prevention). Opponents of user fees point out that they reduce health care access for the poor and negatively affect health and economic welfare. Furthermore, providing healthcare for free may improve efficiency if it helps offset other barriers to access, such as credit constraints. As a consequence, aid organizations now frequently advocate free care, and many African countries have returned to (partially) free healthcare.
The debate about user fees centers on their impact on behavior, and the interaction between fees and household constraints. Our study will help to determine the effects of free medical care on important healthcare behaviors, such as when to seek care during a spell of illness or whether to use preventive care, and study the extent to which free care can offset barriers to access. Our findings will inform the debate by providing information on the efficiency gains and losses associated with free care. We can determine household circumstances in which the efficiency losses are small (or gains are large).
The policy intervention that we study is innovative in its combination of free care with healthcare worker visits. The latter have the potential to offset some of the efficiency losses associated with free care, by providing households with information about the health of the child (and so the need for treatment), and 'nudging' families towards better healthcare behaviors (such as the use of preventive care). Our study will help determine whether healthcare workers can successfully perform this role.
Finally, our project will culminate in the development of a dynamic structural model that can be used by policymakers to estimate the effects of related interventions on different populations prior to running further randomized controlled trials.
Third, our research will benefit third sector organizations involved in the delivery of healthcare in developing countries. Most immediately, Mali Health will use the results of our study to optimize their programs and adjust the elements of Action for Health for currently enrolled families as well as for future expansions. It is also the stated aim of our partner organization to disseminate best practice to other providers within West Africa and beyond.
Finally, the study will be of benefit to the general public by leading to more efficient healthcare delivery. Mali Health plans to offer a version of the Action for Health program to 2400 children and their families by the end of 2014, for an estimated number of 13300 direct beneficiaries. In the long term they plan to expand into other Bamako neighborhoods. Indirectly, if the policy lessons from our study are adopted by governmental and non-governmental providers, the number of beneficiaries can be far larger.
Second, we anticipate that our research will benefit policy making bodies at a local, national and international level. Improving healthcare access for mothers and children are two of the most important aims of global development policy, and are included in the Millennium Development Goals. Child health is a particularly urgent concern in Mali, where the under-five mortality rate is the second-highest in the world. Our project is motivated by the active policy debate about how to fund healthcare in developing countries. The consensus in the 1980's, endorsed in the 1987 Bamako initiative, was that user fees were the best way to deliver high-quality service and prevent waste and over-utilization. Free care could encourage overuse and moral hazard (behaviors that inefficiently increase the cost of care, such as foregoing prevention). Opponents of user fees point out that they reduce health care access for the poor and negatively affect health and economic welfare. Furthermore, providing healthcare for free may improve efficiency if it helps offset other barriers to access, such as credit constraints. As a consequence, aid organizations now frequently advocate free care, and many African countries have returned to (partially) free healthcare.
The debate about user fees centers on their impact on behavior, and the interaction between fees and household constraints. Our study will help to determine the effects of free medical care on important healthcare behaviors, such as when to seek care during a spell of illness or whether to use preventive care, and study the extent to which free care can offset barriers to access. Our findings will inform the debate by providing information on the efficiency gains and losses associated with free care. We can determine household circumstances in which the efficiency losses are small (or gains are large).
The policy intervention that we study is innovative in its combination of free care with healthcare worker visits. The latter have the potential to offset some of the efficiency losses associated with free care, by providing households with information about the health of the child (and so the need for treatment), and 'nudging' families towards better healthcare behaviors (such as the use of preventive care). Our study will help determine whether healthcare workers can successfully perform this role.
Finally, our project will culminate in the development of a dynamic structural model that can be used by policymakers to estimate the effects of related interventions on different populations prior to running further randomized controlled trials.
Third, our research will benefit third sector organizations involved in the delivery of healthcare in developing countries. Most immediately, Mali Health will use the results of our study to optimize their programs and adjust the elements of Action for Health for currently enrolled families as well as for future expansions. It is also the stated aim of our partner organization to disseminate best practice to other providers within West Africa and beyond.
Finally, the study will be of benefit to the general public by leading to more efficient healthcare delivery. Mali Health plans to offer a version of the Action for Health program to 2400 children and their families by the end of 2014, for an estimated number of 13300 direct beneficiaries. In the long term they plan to expand into other Bamako neighborhoods. Indirectly, if the policy lessons from our study are adopted by governmental and non-governmental providers, the number of beneficiaries can be far larger.
Publications
Dean M
(2014)
Credit Constraints and the Measurement of Time Preferences
in SSRN Electronic Journal
Dean M
(2021)
Credit Constraints and the Measurement of Time Preferences
in The Review of Economics and Statistics
Dean M
(2023)
The Effects of Community Health Worker Visits and Primary Care Subsidies on Health Behavior and Health Outcomes for Children in Urban Mali
in The World Bank Economic Review
| Description | The primary development resulting from the grant is the collection of a detailed data set recording the symptoms exhibited and healthcare received by approximately 1700 children in Bamako Mali. Daily information was collected on 13 symptoms related to diarrhea, respiratory disease, malaria, vaccine-preventable disease, and malnutrition, chosen based on the IMCI guidelines for community health workers. This is complemented by objective health data measured in weekly visits. In addition, we collect detailed information on any health consultations and medication purchases and treatments. These data were collected weekly over a period of 10 weeks in autumn 2013, and offer a unique opportunity to study not only what healthcare was received by these children, but also when it was received, and what types of symptoms led parents to seek care. The timing of healthcare seeking is a key element both in determining healthcare outcomes, and the efficient use of healthcare resources. Two features of our data set make it particularly well suited to studying the timing of healthcare seeking: the weekly collection of health data using health diaries recorded by the mother overcomes the well-known problems associated with recall of previous illness episodes, and the recording of data at the household, rather than at the healthcare provider means that we can overcome selection effects which might otherwise bias the estimates of policy impacts on when care is sought. The second result of our study are estimates of the impact of healthcare policies on the health seeking behavior of poor families - specifically the provision of free care at a local clinic and regular health worker visits. The study families were randomly assigned into four groups, receiving one, both or neither of these two interventions, allowing us to test the hypotheses that (a) free care encourages families to seek timely care (i.e. prevent underuse of healthcare by families of children who require care) and (b) health workers can help to mitigate inefficiencies by encouraging families to seek care when it is appropriate (i.e. prevent overuse of healthcare by families who are not sick). We define 'overuse' and 'underuse' relative to WHO guidelines for the circumstances under which care should be sought. Our results find significant support for hypothesis (a). Without free care, we find evidence for widespread underuse in our study population: many children who should be receiving care from a doctor do not receive such care. Free care increases the probability of seeking care on any given day by about 250%. Moreover, the majority of such visits occur on days on which care is required, meaning that free care leads to a large (but far from complete) reduction in underuse with little increase in overuse. In contrast, we find little effect of healthworkers on care seeking behavior. In part, this is because we find little evidence for overuse in the first place: even in the free care treatment there are very few cases in which treatment is sought when the WHO guidelines suggest it should not be. Therefore, while we do find some evidence that health workers reduce doctor visits, they tend to do so more in cases in which the child should be taken to the doctor - in other words health workers act to increase underuse rather than reduce overuse. A third outcome of our study is estimates of the impact of free care and health worker visits on health outcomes, health knowledge preventative behaviors. These estimates come from the 2013 survey and a more limited follow up survey run in 2014. Broadly speaking, we find that healthcare worker visits do have a significant impact on the use of some preventive measures (mosquito net usage, knowledge of oral rehydration recipes) but not on others (measured water chlorination, vaccination rates). We do find some long term impact of the program on health outcomes, particularly the reported amount of diarrhea. These results have formed the basis of policy recommendations to our partner organization, and will be disseminated in the policy document in spring 2016. A final, unexpected outcome of the research from the grant is a paper discussing a novel experimental technique for using intertemporal choice to measure credit constraints and the type of financial shocks hitting a household, as discussed in "Research Tools and Methods". |
| Exploitation Route | From an academic perspective we believe that our research opens up the question of when people seek healthcare, and how policies and household characteristics affect this choice as an important issue within healthcare economics. While our results measure the impact of relatively simple policies - free care and health worker visits - there are many more nuanced options that should be studied - such as health insurance, means testing, co-pays and targeted health education. The methodology we have developed in this grant could easily be adapted to test such policies, with the ultimate goal of developing policies that maximize access to healthcare while minimizing inefficiency. A related issue that is apparent in our data is the over-prescription of drugs - particularly antimalarials and antibiotics. The research team intends to explore the causes of over-prescription in future work. We believe that the experimental techniques we have developed for using intertemporal choice to measure credit constraints and the type of financial shocks hitting a household will be widely used by the large community within development economics which studies these issues. From a policy perspective, our partner organization Mali Health has already started to adapt its policies based on our findings. More broadly we expect to disseminate the lessons from the research to governmental and non-profit health providers through a number of channels detailed in our pathways to impact statement. These include the effectiveness of the program in increasing access and targeting care, as well as the specific areas in which the program did and did not lead to a measurable improvement. |
| Sectors | Communities and Social Services/Policy,Healthcare |
| Description | The results of our research have been used by Mali Health (our partner NGO) to improve the way in which they administer their healthcare program - specifically improving their health worker training and monitoring, and the focus on preventive care, specifically vaccinations. Mali Health and Innovations for Poverty Action are also in the process of disseminating this findings to a wider NGO audience. |
| First Year Of Impact | 2015 |
| Sector | Healthcare |
| Impact Types | Policy & public services |
| Description | Changes to the policies of Mali Health |
| Geographic Reach | Local/Municipal/Regional |
| Policy Influence Type | Influenced training of practitioners or researchers |
| Description | Development Frontiers Stage 2 |
| Amount | £200,000 (GBP) |
| Funding ID | ES/N00583X/1 |
| Organisation | Economic and Social Research Council |
| Sector | Public |
| Country | United Kingdom |
| Start | 08/2015 |
| End | 02/2017 |
| Title | Intertemporal Choice as a Measure of Credit Constraints |
| Description | Experiments which measure the rate at which a person will trade money today for money in the future have often been used as a measure of time preferences. We show that this is not in fact the case, as such measures respond to financial circumstances outside the experiment. Instead, such experiments provide a novel way of measuring a household's marginal rate of intertemporal substitution, which can in turn used to assess the credit constraints the household faces, the financial shocks that they are prone to, and to test models of intertemporal choice and consumption smoothing. |
| Type Of Material | Improvements to research infrastructure |
| Year Produced | 2014 |
| Provided To Others? | Yes |
| Impact | The method has only just been circulated, so none as yet, but we expect it to be used in future research. We have been approached by two groups who wish to use it to test the impact of interventions designed to help households mitigate the effect of financial shocks, such as lack of rainfall. |
| URL | http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2423951 |
| Title | Mali Health Choice Database |
| Description | As per our grant application we have collected weekly data over the course of 10 weeks on a sample of approximately 1700 children from 1100 households. These households were randomly assigned to one of four treatment groups - receiving both free care and health worker visits, free care only, health worker visits or neither. The data includes daily reports from the mother on the health of the child, as well as any treatment sought. It also includes demographic and financial data on the household, and measures of health knowledge and preventive behavior. |
| Type Of Material | Database/Collection of data |
| Provided To Others? | No |
| Impact | From a preliminary analysis of our data, our partner organization Mali Health has made several changes to the way in which they deliver their service, including changes of focus (for example towards ensuring children are vaccinated) and increased monitoring of their health workers. |
| Title | Mali Health Follow Up Survey |
| Description | In 2014 we ran a smaller follow up survey of the 1700 study children to assess the longer term impacts of free care and healthworker visits on preventative care, health outcomes and health seeking behavior |
| Type Of Material | Database/Collection of data |
| Provided To Others? | No |
| Impact | Currently preparing policy documents for wide dissemination on the basis of the data. |
| Description | Presentation (Berkeley) |
| Form Of Engagement Activity | A talk or presentation |
| Part Of Official Scheme? | No |
| Geographic Reach | Regional |
| Primary Audience | Other academic audiences (collaborators, peers etc.) |
| Results and Impact | talk sparked questions and discussion afterwards None noted |
| Year(s) Of Engagement Activity | 2014 |
| Description | Presentation (Columbia University) |
| Form Of Engagement Activity | Participation in an activity, workshop or similar |
| Part Of Official Scheme? | No |
| Geographic Reach | Regional |
| Primary Audience | Other academic audiences (collaborators, peers etc.) |
| Results and Impact | talk sparked questions and discussion afterwards none noted |
| Year(s) Of Engagement Activity | 2014 |
| Description | Presentation (Cristmas) |
| Form Of Engagement Activity | A talk or presentation |
| Part Of Official Scheme? | No |
| Geographic Reach | International |
| Primary Audience | Other academic audiences (collaborators, peers etc.) |
| Results and Impact | Talk sparked questions and discussion None noted |
| Year(s) Of Engagement Activity | 2015 |
| Description | Presentation (Kiel) |
| Form Of Engagement Activity | A talk or presentation |
| Part Of Official Scheme? | No |
| Geographic Reach | Regional |
| Primary Audience | Other academic audiences (collaborators, peers etc.) |
| Results and Impact | Talk sparked questions and discussion afterwards None noted |
| Year(s) Of Engagement Activity | 2015 |
| Description | Presentation (Stanford Institute for Theoretical Economics |
| Form Of Engagement Activity | A talk or presentation |
| Part Of Official Scheme? | No |
| Geographic Reach | International |
| Primary Audience | Other academic audiences (collaborators, peers etc.) |
| Results and Impact | Talk sparked questions and discussion None noted |
| Year(s) Of Engagement Activity | 2014 |
| Description | Presentation (Yale) |
| Form Of Engagement Activity | Participation in an activity, workshop or similar |
| Part Of Official Scheme? | No |
| Geographic Reach | Regional |
| Primary Audience | Other academic audiences (collaborators, peers etc.) |
| Results and Impact | talk sparked questions and discussion afterwards none noted |
| Year(s) Of Engagement Activity | 2014 |