| Summary: | Abstract Background Choosing who should be recruited as a community health worker (CHW) is an important task, for their future performance partly depends on their ability to learn the required knowledge and skills, and their personal attributes. Developing a fair and effective selection process for CHWs is a challenging task, and reports of attempts to do so are rare. This paper describes a five-stage process of development and initial testing of a CHW selection process in two CHW programmes, one in Malawi and one in Ghana, highlighting the lessons learned at each stage and offering recommendations to other CHW programme providers seeking to develop their own selection processes. Case presentation The five stages of selection process development were as follows: (1) review an existing selection process, (2) conduct a job analysis, (3) elicit stakeholder opinions, (4) co-design the selection process and (5) test the selection process. Good practice in selection process development from the human resource literature and the principles of co-design were considered throughout. Validity, reliability, fairness, acceptability and feasibility—the determinants of selection process utility—were considered as appropriate during stages 1 to 4 and used to guide the testing in stage 5. The selection methods used by each local team were a written test and a short interview. Conclusions Working with stakeholders, including CHWs, helped to ensure the acceptability of the selection processes developed. Expectations of intensiveness—in particular the number of interviewers—needed to be managed as resources for selection are limited, and CHWs reported that any form of interview may be stressful. Testing highlighted the importance of piloting with CHWs to ensure clarity of wording of questions, interviewer training to maximise inter-rater reliability and the provision of guidance to applicants in advance of any selection events. Trade-offs between the different components of selection process utility are also likely to be required. Further refinements and evaluation of predictive validity (i.e. a sixth stage of development) would be recommended before roll-out.
|
|---|
