In our study, we evaluated a tailored multimodal intervention for GPs, developed with human-centered design methods.
The response rate to our questionnaire was 39%, which is comparable to other surveys in primary care [15, 32].
Overall, the intervention tools were well accepted and highly rated. Most respondents used more than one tool, with an average use of 3.1 of the six available intervention tools. A choice of different tools increases the likelihood of meeting different physician’s preferences.
The GP training on rational antibiotic use was the best-rated tool, with a “strong” or “very strong” influence on antibiotic prescribing behaviour by all training session attendees. This emphasizes the importance of training on the topic for GPs and is in line with our survey from the first project phase, where we explored the information sources of our target groups [28]. GP education has proven effective in the reduction of inappropriate antibiotic prescribing in primary care [27] and has been part of multifaceted interventions [33]. However, training on rational antibiotic use without pharmaceutical sponsoring are scarce in the outpatient sector in Germany. Our findings confirm that training on rational antibiotic use should firmly be integrated into GPs’ continuing medical education. To extend coverage, we developed online training (MOOC), which was used by a significant number of physicians after the intervention period.
Despite its good rating, GP training was not the most frequently used tool in our study. Approximately half of the respondents stated that they participated in one of the training sessions. Overall, over 300 physicians took part in our GP training. However, only those who later ordered additional material (i.e. tools) were registered as study participants. This is due to the certification procedure of the Medical Chamber for elective courses. Certified courses may not be part of a study.
On the other hand, the fact that nearly half of the respondents did not participate in the training shows that, in addition to the public announcement of the training, our other recruitment measures were also effective. Fewer personnel-consuming intervention tools could also be distributed without training as an anchor point.
At the time of the study, it was not possible to access prescriber feedback on antibiotic use from claims data in Germany. We therefore aimed to develop a simple application for GP self-monitoring. Compared to the other tools, the GP self-monitoring app was the least-used intervention tool. A frequently mentioned reason for that was a lack of technical prerequisites. Our budget only allowed for app development for one smartphone type; we chose the iOS®, but it is possible this was not best suited for the target group. Additionally, even if it was only one single tap (ARI without antibiotic prescription) or double tap (ARI with antibiotic prescription), it is an extra effort to actively use the application systemically, which might have overburdened already busy GPs. However, the small group who did use the app rated it as good; in cohorts with appropriate technical prerequisites and with higher digital affinity, the idea could work.
Most patients expect examination and explanation of their symptoms, rather than antibiotics [32, 34,35,36]. On the other hand, doctors often feel a certain AB prescription pressure exerted by the patient [19]. Fostering doctor-patient communication is an effective strategy to address this dilemma [37]. However, in Germany, a GP has an average of 8 min per patient [38] and, therefore, needs a time-effective tool. To address these needs, we developed information prescriptions, similar to ´viral prescription pads´ used in Northern America [39]. Internationally, such tools have already shown positive effects on the reduction of unnecessary antibiotics [22]. Furthermore, they correspond with the participatory approach in primary care, by increasing health competence of patients.
Printed information prescriptions were by far the most-used tool from our set. They were used twice as often as the digital version and more frequently in everyday work. Overall, half of those who used printed IPs also used digital versions, but only one person used digital IPs exclusively. Thus, in our responder cohort, digital IPs were a complement rather than a substitute for the printed versions. Printed IPs were short and ‘ready to use’, while the online platform provided a larger variety of content, with a broad choice of symptoms, handling instructions in several languages and the requirement of basic technical equipment (internet access), logging in, identification of relevant information for the patient, and printing. Hence, one could imagine that the printed versions were used preferentially for ‘standard’ patients and the digital content for patients that are more ‘complex’. However, what we can learn from this is that a considerable proportion of our cohort is not limited to digital or analogue procedures in their offices, but uses a combination of the two. Important for software developers is the finding that 31 of the 44 respondents who used the digital IP wanted it to be integrated with their medical software. Digital IPs were used more frequently by men than by women, which points towards different physician preferences and emphasizes the importance of diversity in intervention tools.
Overall, information prescriptions were used by 80% (n = 85) of respondents, of whom 40% (n = 43) used both types; this was possibly also due to different preferences of different patients.
The better-informed patients are, then the better they cooperate [40, 41]. Waiting room patient leaflet and posters were frequently used. By addressing patients before consultation, they served as preparation and, thereby, supported doctor-patient communication.
Apart from the GP self-monitoring app, all tools were well-accepted and respondents used them in a variety of combinations. Over half the respondents stated “participation in the study” itself had a “strong” or “very strong” influence on their individual prescribing behaviour. These findings suggest that the overall concept of a multimodal intervention (“tool box”), rather than a single tool, was convincing.