Data analysis
The data needed to determine the Activity Standard concerned the tasks required to deliver pharmaceutical care, the times these tasks should take and the frequency that they should be delivered. These data were analysed using descriptive statistics, using the mode value to identify consensus. The extent to which consensus was achieved was depicted using a ‘RAG’ rating (see Table 2) to give greater clarity on the strength of the consensus for each component, since there is no universal definition of consensus[21, 22]. This was particularly necessary where consensus was not reached for tasks to include in the CPWC, which then needed a finite value for the time taken to complete them and the frequency they should be undertaken for the algorithm in the CPWC to function. For times and frequencies associated with these tasks the typical binary approach of consensus/no-consensus was not practical. The RAG rating identified where agreement was widespread (green >70%) and where there were elements of greater variety of opinion (amber >50%). Where consensus (i.e. greater than 50% agreement on a specific figure) could not be achieved for the time a task took, the value for the algorithm was derived from the data provided. To be meaningful in calculating staffing levels nationally, these derived figures had to represent UK national practice and so the data was analysed by mean and median as well as mode to determine a ‘national best representative’ figure. Similarly, for the frequency of task completion where consensus was not achieved, especially for patient-dependent activities, responses from the exemplar patient questions in Round two were used to calculate an ‘average’ frequency for the purposes of algorithm development. To complete the WISN algorithm (Figure 1), the ‘unavailable’ staff time was calculated from a mean of reported data.