Anticipation and response: pandemic influenza in Malawi, 2009

Sambala, Evanson Z.; Manderson, Lenore

After each interview, the first author listened and relistened to audio recordings to gain familiarity with the data and enable iteration, and entered notes into a data analysis logbook. Note taking was repeated after full transcription, during the coding of the data, and again at the time of writing. All transcribed interviews were exported to NVivo 8 to facilitate coding and thematic organisation. We used the six thematic areas of preparedness and responses set out in the WHO checklist [18] as a framework of analysis to identify respondents' views on the levels of preparedness, responses, strengths and weakness. These included planning and coordination, surveillance, communication, public health interventions, patient management and maintaining essential services. With these thematic areas in mind, coding was adopted, with the text examined closely, line by line, to identify both pre-defined and new themes. Generating codes involved three stages. The first was open coding, where interesting features of the data were identified, labelled and defined. Initial codes matching data extracts were collated by labelling and assigning a selection of unique identifiers of text within each data item. The second phase proceeded by reviewing and refining themes in which the connections between concepts (such as planning and public health infrastructure) were explored to help build categories and interrelationships. Here, we considered whether predefined themes and sub-themes formed a coherent pattern and if not, whether this was problematic. Themes that did not fit particular data extracts were redefined or discarded. The third phase involved searching selective coding, where predefined themes were further defined and refined. Throughout the process, codes identified in the data, predefined themes and categories were subjected to an iterative process, involving constant testing of data, confirming or negating the concepts, until the patterns in the data were clearly understood. This provided structure to the extracted data and interpretation in the final analysis.