Analysis

Observations as well as interviews were transcribed verbatim. The primary transcription was made from the audio recordings, while the video recordings were used during transcription to clarify who was speaking. NVivo 9 software was used to store and structure the transcripts and code the data from both observations and interviews. Observations were analysed by means of directed content analysis (23). A detailed description of each observation was made, concentrating on the focal points mentioned in Supplementary Data File S1. The interview data were analysed using thematic content analysis (24). Two researchers, JvD and SB (an external and experienced coder), analysed all the transcripts independently and carried out the open coding of all quotes relevant to the aim of the study. Concepts were identified and grouped into subcategories. In the next step, the two researchers compared and discussed their codes until they reached consensus and subsequently categorized into the different subcategories. In case of disagreement, a third researcher (LvB) was asked for advice. In the final step, the researchers identified key theme categories into which the subcategories could be divided.