Data Analysis

Data analysis was done concurrently with the data collection using a Microsoft office word processor [25]. Interviews conducted on a day were transcribed from the audios and notes taken in the field before the next data collection. Information from the analysed data was used to enrich the subsequent interviews. Independent coding was done by two authors to keep validity and reliability of the theme development [26]. Consistency between the audio and the transcriptions was checked by two of the authors (author one and three), who speak the language used in the interview, independently. Furthermore, the audio records and transcripts were cross-checked by independent experts before the final analysis. Finally, all authors checked and agreed on the development of code categories and themes.Our study participants (CHWs) have embedded knowledge of the pastoralist communities and also understand the health system of the country, Their perspectives on community TB program performance are important to leverage in designing appropriate solutions for these unique marginalised communities. With this premise, we analysed the interview transcripts and audio records without requiring other data sources for triangulation.We followed an inductive content analysis approach [27]; that is identified codes, sorted into categories and developed themes which finally synthesized. We used a combination of ontological (capture participant realities) and epistemological (understand the phenomenon) approaches to explore community TB programs in the study setting. An interpretive description method was used to analyse the manifest content, Furthermore, a hermeneutic interpretation method was used to analyse latent contents [18, 19, 28]. The findings were reported using textual descriptions and quotes to illustrate ideas and illuminate experiences [29]. We also produced a concept map that shows the interrelation between themes emanating from the analysis.