Jeroen van de Pol

143 5 Preferences of patients regarding community pharmacy services: A discrete choice experiment Study sample Participants were recruited from the AMP pharmacy patient panel. AMP Research & Consultation in healthcare (AMP Onderzoek & Advies in de zorg) [21] is a Dutch organization focusing predominantly on mystery shopping and provision of patient information leaflets in community pharmacy practice. The AMP patient panel holds about 40,000 individuals of whom 47.5% are male and 78% uses medication for chronic diseases. The panel was formed in 2015 and is comprised of individuals that visit a community pharmacy on a regular basis and consented to be contacted for research purposes. Individuals from the patient panel were invited once by e-mail by AMP to participate in this research. They were provided with a link that led to the online questionnaire including the CBC. No reminders were sent. Sample size General guidelines regarding CBC state that inclusion of 200 participants per subgroup is usually sufficient [3,18]. Other research utilizing a CBC for preferences of the general public or patients with regard to community pharmacy practice yielded response ranging from 194 to 9,202 individuals [14,20,22-24]. Our goal was to be able to distinguish 5 different groups, so a total of at least 1,000 individuals was deemed necessary. Data management Only the fully completed questionnaires were included for further analysis. Participants that did not complete the questionnaire fully were compared to participants who did, based on age and gender. Participants were excluded from analysis if thereweredoubtson the reliabilityof their answers. Potential unreliability was based on the amount of time spent on the questionnaire and the consistency of answers (constantly choosing the left- or right option). Participants finishing the questionnaire under 3 minutes were excluded, as this was not deemed possible. Participants were also excluded when they finished the questionnaire between 3 and 5 minutes and showed poor consistency in answers provided within the choice task (Root Likelihood < 0.6). Root likelihood is the mean of estimated probabilities associated with the different alternatives actually being chosen by participants (fit between utility estimates and choices made by participants). Lastly, participants were also excluded when choosing constantly the left- or right option in the choice task and provided poor consistency in answers provided (Root Likelihood < 0.6).