Figure 2: Diagrammatic representation of study design.
First, symptom history was captured using an automated questionnaire (devised by author [removed for blind peer review] in English, piloted in ENT clinics in Tanzania and England, and then translated into Khmer). An initial question asked participants to report ear discharge (otorrhoea), ear pain (otalgia) and/or hearing loss (we also captured symptoms of dizziness and tinnitus, but did not use these in subsequent analysis). Forced responses and skip logic (figure 3) were used to further define each symptom, including laterality, chronology, and (for otorrhoea) consistency. The algorithm also collected date on demographical information, past medical and surgical history and drug and allergy history. For this study the questionnaire was hosted on the Cupris platform (Cupris Ltd, London), and data were captured by one of three ENT specialist nurses who were native Khmer language speakers and had worked at CSC hospital for 1-6 years.