Clinical pharmacists in German hospitals importantly contribute to optimise patients’ medication providing routine proactive medication management services. While in IW-2017 in every fourth patient day a pharmacist intervened, in IW-2019 this proportion increased to more than every third patient day. The noticeable increase in ME identified by CP before reaching the patient shows that PI become more proactive instead of reactive and patients are protected from ME early in the medication use process. This may be a positive effect of the increasing proportion of CPS performed on a daily basis. This deeper involvement in direct patient care might also reflect the observed changes towards more complex interventions in IW-2019, as expressed by the increase or reasons for PI regarding drug choice and dose adjustment as compared to the decrease reasons, that can easily be reduced or even avoided using electronic support (e.g. transcription errors, incomplete prescriptions, interactions).
As expected, the intervention rates are higher with rising professional experience and the absolute number of documented PI. However, we observed a small negative effect of performed medication analyses on the intervention rates in both IW. This effect might be explained on the one hand by the increasing use of computerised physician order entry (CPOE) with clinical decision support systems (CDSS) that are known to reduce prescribing errors [15, 16] and on the other hand by learning effects as a result of regular interdisciplinary exchange and training and education activities by clinical pharmacists. Taken together, we are confident that the intervention rates are a robust measure, considering the wide range of patient days and PI per participant. The high numbers of patient days reported by some pharmacists can be explained by medication validation within a completely digital closed loop medication management.
There are many studies about the impact of CPS, and especially for PI to reduce DRPs, ME or adverse drug events (ADE) [2, 17,18,19,20]. However, comparability is limited between these studies due to different definitions and methods and the lack of a standard denominator. Preventable ADE have been published to range between 4 to 26.5 per 1000 patient days in different settings [4, 21, 22]. With regard to PI, pharmacist validation of medication prescriptions in a French university teaching hospital resulted in a PI rate of 812 per 62,341 medication orders (1.3 per 100 medication orders) , while a mean rate of 4.5 PI per day and ward was determined in the setting of closed-loop medication management with pharmacist validation in a German university hospital .
As compared to the MEDAP study, where 62 pharmacists in the United States (in- and outpatient care) submitted 924 datasets on ME and related PI during a 2-weeks period , considerably more datasets have been provided in our study by similar numbers of pharmacists during a one-week period (2282 PI by 62 pharmacists and 2587 by 52 pharmacists, respectively). Regarding the classification of ME according to NCC-MERP, our results are in good accordance to those of the MEDAP study, with the vast majority of ME not resulting in patient harm (more than 95%). In another study, PI for community hospital inpatients were recorded and classified during a 14-day period from pharmacists of 15 organisations in the United Kingdom. In this study, mainly in rehabilitation patients, a PI was done in one of three charts for one or more medications with a total of 2758 PI in 4077 medication charts (63 PI per 100 medication charts), with two thirds being classified as prescribing errors. This rate markedly exceeds our rates of 27.5 (IW-2017) and 38.5 PI/100 patient days (IW-2019).
The acceptance rate in IW-2019 (88.4%) is comparable to other published data [14, 19, 22, 24,25,26]. Our results are especially in good accordance with the latest French national data of 34 522 PIs registered by 201 pharmacists working in 59 hospitals over a 30-months period, where an acceptance rate of 86% was reported with a significant association to the level of pharmacist integration in the ward . The lower rate in IW-2017 (77.9%) might be explainable by the relatively high proportion of PI with unknown acceptance (14.9%). This high rate might result from the much lower frequency of CPS (about 40% less than daily in IW-2017 as compared to about 13% in IW-2019), which makes it more difficult to completely track the acceptance of the PI. This is especially true in those cases, when the PI is not immediately accepted, for example if the junior physician on the ward needs to discuss the PI with a senior physician, patients have only short hospital stays and/or no electronic patient record is available.
The high numbers of both participating pharmacists and documented PI are clearly strengths of our study. Thus, we are confident, that our data represent a cross section of routine CPS in Germany. The fact that the majority of participants work in university hospitals reflects the general situation in German hospitals regarding the staffing levels of clinical pharmacists in different types of hospitals, with a median of 2.7 full-time equivalents per hospital for university hospitals and 1.0 for maximum care and general hospitals, respectively . Furthermore, the willingness to participate in scientific studies might be slightly higher for employees from university hospitals. Another advantage of our study is that we collected data longitudinally using the same methodology, to detect developments in the extent and characteristics of CPS. To the best of our knowledge, no comparable studies have been conducted before.
Possible limitations of our study are the heterogeneity of working environment of the participants concerning the frequency and manner of CPS, paper-based versus electronic prescribing (including CPOE-CDSS), although the PROTECTED-UK study did not show a correlation between the presence of electronic prescribing in critical care and intervention rates  and/or traditional supply of whole drug packages with distribution by the nursing staff versus patient-individual logistic provided in the pharmacy. On the other hand, using this cross-sectional approach, we gain insight into the daily routine of CPS in German hospitals. The relatively short time periods of five working days for each period as the basis for our study, might not fully represent everyday work, however we aimed to obtain a data collection of PI as complete as possible for a given time period. To have as many participants as possible and to obtain comparable data to similar studies from other countries we decided to perform our data collections on a five working day basis. As the participation in the DokuPIK IW was voluntary, it might mean that data might have been provided form more engaged pharmacists and therefore the results might not be fully generalizable to ‘pharmacy in Germany’ broadly. Nevertheless, the comparison between the two IW is supposed to provide reasonable reliable results.
In Germany the development of quality and/or key performance indicators for CPS in hospitals as part of the current guideline on inpatient care by hospital pharmacists  is about to be completed, based on the publications from other countries [29,30,31]. Therefore, the data from our study serve as valuable basis for quality indicators regarding the intervention rates of clinical pharmacists in Germany. In this context, ADKA DokuPIK as a structured and validated documentation tool, is intended to further serve as a national documentation platform for quality indicators and is going be developed further to fulfil these requirements.