Prospective survey-based study on the categorization quality of hospital pharmacists' interventions using DokuPIK.

Background There is a growing need to categorize pharmacists' interventions (PIs) in Germany to document their impact on solving or avoiding drug-related problems. Objective To validate the categorization of drug-related problems-one aspect of the categorical internet database DokuPIK, designed for recording routinely PIs. To identify case-specific predictive values. Setting German hospitals. Methods Within a prospective, nationwide survey-based study, 37 of 498 registered database users volunteered to evaluate 24 standardized case reports independently. Case evaluation was restricted to classify problems, based on 26 given categories with no limit on the number of item choices. Ratings were conducted electronically and anonymously. A gold standard of one or more problems per case was developed by majority consensus of five senior clinical pharmacists. Agreement of raters' case classification with the gold standard was assessed by calculating sensitivity, specificity and positive and negative predictive value and was reported as median and range. Main outcome Level of agreement. Results Independent assessment yielded a median agreement of 90% [79-94%]. Sensitivity and specificity were 37% [21-57%] and 99% [97-100%], respectively. Median positive and negative predicted value were both 90% [60-100%] and 90% [78-95%]. Mean case-specific agreement was robust (≥ 79%) with respect to a majority and maximum consensus (three and five out of five raters). Conclusion DokuPIK seems to have a high level of agreement and a good specificity according to the majority of clinical pharmacists in a panel of assessors. Despite the allowance of multiple choices, predictive values were high and indicated a well-constructed classification by pharmacists.