Table 1.

Research paradigm used. PHAR-QA, Quality Assurance in European Pharmacy Education and Training.

Step	Phase
1	A competence framework based on PHARMINE [6] and other published frameworks for practice in healthcare was ranked (4-point Likert scale) and refined by 3 rounds of a Delphi process [7], by a small expert panel consisting of the authors of this paper.
2	Following the 3rd Delphi round within the small expert panel above, the competences were ranked in two separate rounds by a large expert panel consisting of six groups, European academics, students and practicing pharmacists (community, hospital, industrial and pharmacists working in other professions), using the PHAR-QA SurveyMonkey® (SurveyMonkey Company, Palo Alto, CA, USA) questionnaire [8]. There were 68 competences proposed in the first round and 50 in the second, the difference being due primarily to the removal of the subject areas. Invitations were sent to the 43 countries of the European Higher Education Area that have university pharmacy departments (thus excluding countries, such as Luxembourg and the Vatican). Data were obtained from 38 countries (thus not including Armenia, Azerbaijan, Georgia, Moldova and Russia). In some figures, not all countries are represented, but data from all countries were included in the statistical analysis.
3	The first 6 questions were on the profile of the respondent (age, occupation, experience).
4	Respondents were then asked to rank clusters of questions on competences numbered 7–17 (numbering following on from the 6th question of the respondent profile). Questions in Clusters 7 through 10 were on personal competences and in Clusters 11–17 on patient care competences.
5	Respondents were asked to rank the proposals for competences on a 4-point Likert scale:
	(1) Not important = Can be ignored;
	(2) Quite important = Valuable, but not obligatory;
	(3) Very important = Obligatory, with exceptions depending on the field of pharmacy practice;
	(4) Essential = Obligatory.
	There was also a “cannot rank” possibility and the possibility of leaving an answer blank.
6	Ranking scores were calculated (frequency rank 3 + frequency rank 4) as the % of total frequency; this represents the percentage of respondents that considered a given competence as “obligatory”.
	The calculation of scores is based on that used by the MEDINE “Medical Education in Europe” study [9].
7	Leik ordinal consensus [10] was calculated as an indication of the dispersion of the data within a given group. Responses for consensus were arbitrarily classified as: <0.2 poor, 0.21–0.4 fair, 0.41–0.6 moderate, 0.61–0.8 substantial, >0.81 good, as in the MEDINE study [7].
8	For differences amongst groups and amongst competences, the statistical significance of differences was estimated from the chi-square test; a significance level of 5% was chosen. Correlation was estimated from the non-parametric Spearman’s “r” coefficient and graphically represented using parametric linear regression.
9	Respondents could also comment on their ranking. An attempt was made to analyse comments using the NVivo10® (QSR International Pty Ltd., Victoria, Australia) [11] and the Leximancer® (Leximancer Pty Ltd., Brisbane, Australia) [12] programs for the analysis of semi-quantitative data. In this study and the previous first round study, the word number of the comments was too small to draw significant conclusions.