Data analysis

The transcripts were deidentified using pseudonyms and removing participants’ department and location in WHO. Data analysis began simultaneously with data collection. Through this cycle, we explored ideas and issues that emerged through the analysis of early transcripts in subsequent interviews [11]. The cycle was done in the following steps:

Transcripts were analysed and coded using line-by-line coding in the form of open coding, describing and labelling the ideas and issues that arise in the transcripts. The initial codes and themes were used to focus subsequent interviews.

As more data were collected, axial coding was used as the next stage of analysis. Axial coding is the second pass of the data, and involves organising codes from different transcripts and reassembling them [15].

Memos were made throughout the analysis to encourage reflexivity and record the ideas that arise during the analysis process.

Data collection continued until the developed categories were saturated, that is, when new data no longer provided further insight into a particular category. Memos were a key tool in the coding process. They were written to make comparisons and to allow the researchers to piece together the process of the WHO officers’ work.

Throughout the data collection and analysis process the research group conducted fortnightly workshops to discuss interview techniques, transcripts and coding. Team members discussed salient themes that arose. This enhanced the trustworthiness of the data and encouraged the primary researcher’s (ZW) reflexivity towards the data. Feedback from these workshops challenged how the data was grouped and analysed. In turn, it resulted in the refinement of the data presentation.