communitypharmacy.scot.nhs.uk/core_services/mas.html); frequency of use of the same pharmacy; most recent use of a pharmacy medicine;

most recent purchase of a pharmacy medicine and type of pharmacy normally used. Health status was measured using the question, Smoothened Agonist ‘in general would you say your health is’ with five response options (excellent, very good, good, fair, poor). The sample size was estimated to ensure inclusion of an adequate number of individuals who had purchased NPMs in the previous 2 weeks. A postal survey of 3000 individuals was estimated to generate 1350 usable forms (based upon a 50% response rate and a further 10% being returned by the post office unopened). It was estimated that 8% (n = 108)

of these respondents would have purchased a NPM[19] and 45% (n = 608) would have used a NPM in the previous two weeks.[20] Approximately 75% of consultations for NPMs are made as product requests, with the remaining 25% representing advice requests.[4, 7] It was estimated, therefore, that 25% (95% confidence Rucaparib interval (CI), 18.8 to 32.4) of consultations for NPMs would involve the provision of some (unprompted) information to MCAs. A minimum sample size of 104 + m (where m is the number of predictors) is suggested when testing individual predictors in regression models.[21] This reflects the standard approach to sample size calculations for TPB surveys. The predicted sample size of 1350 was therefore sufficient to examine the proposed predictors of self-reported behaviour, together with potential confounding factors such as consultation type and patient characteristics. Data were entered into SPSS (version 18) (PASW Statistics 18. SPSS Inc, Chicago, IL, USA). All TPB variables showed skewed distributions and thus medians (interquartile

ranges) are presented alongside from Cronbach’s alpha (Cα) to determine the internal reliability of the measures. Items with low internal consistency were removed. Univariate tests were used to investigate the relationship between demographic characteristics and ‘giving information’ (the behaviour) and BI (intention) as well as BI-WWHAM. The association between TPB variables and behaviour was explored first by Spearman rank correlations (rs) and then using logistic regression performed in three steps: step one explored the proximal predictors (i.e. those nearest to the behaviour: PBC and BI); step two added the distal predictors (i.e. those that operate via the proximal predictors: attitude and subjective norm) and step three added demographic and pharmacy behaviour variables that were related to behaviour. Linear regression was used to assess the relationship between TPB variables and BI to give information and BI-WWHAM.