Developing a selection tool for Community Health Workers in sub-Saharan Africa

Community Health Workers (CHWs) provide basic health care and advice to local communities. They often work in remote, rural areas where access to formal health care is poor. Historically, CHWs have provided support for single diseases, such as HIV/AIDS and there is good evidence that their work has a positive impact on the health of their clients. This evidence, together with the need to improve access to health care in sub-Saharan Africa has led to a rapid increase in the number of CHWs being recruited. The nature of the CHW role is also being expanded to include significantly more tasks, including screening for malnutrition and chronic disease, deworming and advice on contraception. Successful completion of this broad range of tasks requires CHWs to be knowledgeable and skillful in a variety of health areas and they also need to demonstrate qualities such as empathy, time management and diligence. Although all CHWs receive basic training, not all of these job requirements can be taught, so applicants should be able to demonstrate some of these before beginning training. Competition for CHW roles is often high, so it is essential to select those applicants most likely to be able CHWs, and least likely to resign, in order to maximise the health gains that they provide for their clients.

The involvement of the local community in selection is critical, but is unlikely to be sufficient by itself because community members may not be able to assess applicants on their literacy, basic health knowledge and the extent to which they can be trained. In addition, community involvement may not always be fair: one study found that just under half of CHWs were related to their village chief. Even though investments in selection for other jobs usually have a high positive payback, the resources available for CHW selection are limited. This means that an effective and affordable method of choosing between applicants for CHW roles is required. This project will use a selection tool already being used in Kenya and Uganda, and we will start by examining how well scores on this tool predict the later performance of CHWs. We will then develop the tool for use in other countries; for this project we will work in Neno in Malawi and Mpohor in Ghana. Here, we will begin by producing a 'job specification' for each area, listing all of the knowledge, skills and attributes needed to do the job well. This will involve analysing documents relating to the tasks completed by CHWs, observing CHWs at work and interviewing CHWs, community leaders and CHW programme providers. We will then host a series of workshops, to which CHW programme providers, CHWs and community leaders will be invited, during which we will co-design the selection tool. Between the workshops we will run small-scale tests to provide initial feedback to the development team and after the final workshop we will run a larger-scale test to determine how to proceed at the end of this project. If the tests show that the tool has promise, we will apply for further funding to run a full-scale pilot. Our ultimate aim is that the selection tool can be used across settings and countries with only minor modifications. Once developed and fully tested, the tool would be made freely available to any CHW programme providers who wanted to use it and we would publicise its availability using various CHW networks. We would provide written manuals and training for new users to help them use the tool effectively.

There is good evidence that Community Health Workers (CHWs) can provide effective disease-specific care in their local communities. Recent scale-up of the role, tasks and workload of CHWs has however increased the job requirements of CHWs in terms of the knowledge, skills and attributes required to perform the job well, and there are now reports of difficulties in retaining CHWs, in their ability to perform all the tasks required of them and maintaining their diligence. An improved selection process for CHWs would mitigate these problems by identifying CHWs most likely to perform well and be retained in post.

This project will use an existing selection tool (used by the charity Living Goods in Kenya and Uganda) and co-design appropriate versions of the tool for initial testing in two other local contexts, Mophor district in Ghana and Neno district in Malawi. We will start by establishing the predictive validity of the existing tool in use in Kenya and Uganda using previously-collected data on selection, training and job-performance. At the same time, we will review the CHW programmes in Ghana and Malawi to determine differences in required tool content (for example due to differences in the skills required to perform the tasks required in each programme). We will do this using task inventory analysis, observational job analysis of high-performing CHWs and job-element method-based semi-structured interviews with key stakeholders in each district. We will also establish 'design requirements' for tools such as format and maximum costs. We will use the results of these studies to co-design tools and undertake initial alpha and beta testing in Mpohor and Neno. Our final outcome will be a revised tool suitable for roll-out and summative evaluation in other areas.

The impacts expected from this early stage development project (18-24 months after commencement) are:
1. Increased awareness of the need to consider how Community Health Workers (CHWs) are selected;
2. Increased awareness of selection tool development methods; and
3. Increased awareness of methods of evaluating selection processes.

These impacts will accrue to CHW programme providers (including non-government organisations and local and national governments) and academics. They will be realised by a communications plan that involves all stakeholders and relevant CHW networks, as well as engagement with the academic community via peer-reviewed journal articles and a conference presentation.

Medium-term impacts, following tool refinement and further testing and evaluation (24-36 months after commencement of the follow-on work that would form the main study) are expected to include:
4. An assessment of whether the tool is fit for purpose across a variety of CHW programmes;
5. A framework for evaluation of selection tools that can be used by CHW programme providers (and others involved in the selection of healthcare professionals);
6. Free provision of the selection tool for local adaptation and use, with appropriate written manuals (co-designed by stakeholders) and training when requested; and
7. A process of monitoring use of the tool by others.

These impacts will accrue to CHW programme providers (including non-government organisations and local and national governments) and academics. They will be realised by development of a selection tool website which will be publicised through CHW networks as well as to other stakeholders, and by further academic dissemination.

Ultimate outcomes, following use of the tool in practice (>36 months after commencement of the main study), are expected to include:
8. Selection of CHWs who are likely to remain in post, have the knowledge, skills and attributes required to successfully complete their training and subsequently perform well on-the-job and who will do so diligently;
9. Enhanced health outcomes for the clients of CHWs; and
10. Academic evidence about the effectiveness of the tool in different contexts.

These impacts will accrue to CHW programme providers (including non-government organisations and local and national governments) but primarily to their clients. These impacts require that the selection tool is used appropriately by CHW programme providers; hence our medium-term intention to provide written manuals and user-training and to monitor use of the tool so we can be reassured that it continues to meet the standards expected of any selection method (e.g. validity, reliability and fairness).