GP practices were contacted 6 months after

interview to obtain MMR1 uptake data for participants’ children. Participants were classified to decisions groups as follows: ‘accepted MMR1 on time’ if child received MMR1 by the day he/she turned 14 calendar months old (UK immunisation schedule recommends MMR1 at 13 months [4]); ‘accepted MMR1 late’ if child received MMR1 after 14 calendar months old; ‘obtained singles’ if child received no MMR1 by time of data collection but GP confirmed singles had been given or the parent had intended to give singles; ‘accepted no MMR1 or singles’ if child received no MMR1 by time of data collection and the parent had intended to give neither MMR1 nor singles. Transcripts were analysed by a Microbiology inhibitor coder with background in psychology (KB) using a modified Grounded Theory approach [43], [44] and [45] using NVivo 8 (QSR International Crizotinib datasheet Inc.). Coding was completed before objective outcome data were obtained but the primary analyst was aware of each interviewee’s intended decision. Data were first broken into small sections of homogeneous content ranging in size from a few words to a paragraph, and grouped by that content into codes. Sections which covered the same content were grouped into the same code, and new codes were created as new content areas were found in the data.

Every section of data was grouped under at least one code, and sections with shared content but from different participants were grouped under the same code. The codes can be found in Supplementary Table 1. During the coding process, links between codes were identified and memoed, and through this process codes were linked together and synthesised into broad themes for reporting. Two measures were taken to counter analysis biases: eight transcripts distributed across the decision groups were analysed in duplicate by a second coder with background in medicine (SL) blinded to the first analyst’s codes and to the participant’s intended decision, and a further eight participants across the decision groups

provided a member check by reviewing the coding of their interviews. A qualitative approach to reliability was taken, whereby the two coders discussed their codes, identified discrepancies and reached consensus via discussion, tracing beyond the original subset where necessary to ensure any necessary amendments or additions were applied crotamiton to all relevant data in the full dataset. Twenty-four parents (all mothers) participated in interviews between June 2008 and March 2009. Their characteristics are shown in Table 1. Most participants were highly educated at-home mothers. Twelve participants were recruited through GP practices, 3 through mother-and-baby groups, 6 through online parenting forums and 3 through chain referral recruitment. Parents giving MMR1 on-time or late were mainly recruited through GPs or mother-and-baby groups, whilst parents giving singles or no MMR1 were mainly recruited through online forums and chain referral.