Results of the thematic analyses are presented separately for GPs and paediatricians.
Participants
A total of 40 interviews (17 paediatricians; 10 from high-prescription and 7 from low-prescription districts, 23 GPs; 10 from high-prescription and 13 from low-prescription districts) were conducted. Participants had between 1 and 35 years of experience in their current positions (mean 13.4 years, SD 9.9 years). Interviews lasted an average of 44.8min (SD 7.1min, range 30–61min).
Paediatricians
TDF domains on region levels mentioned as influencing paediatricians’ prescribing behaviour (Fig.1) included context and resources (86 mentions), social influences (56 mentions), knowledge (36 mentions), skills (22 mentions), social/professional role (15 mentions), beliefs about consequences (15 mentions), beliefs about capabilities (9 mentions), goals (9 mentions), behavioural regulation (6 mentions), optimism (3 mentions) and emotions (2 mentions).
TDF domains mentioned as barriers (red) or resources (blue) by paediatricians
Context and resources
Regional context and resources can affect prescribing behaviour through multiple, direct and indirect pathways, according to the participating paediatricians. The distinction between contextual (i.e., factors specific to the region) and composition effects (i.e., factors resulting from the composition of the population within a region; [33])is particularly relevant.
Paediatricians mainly mentioned contextual factors, e.g., air pollution as a risk factor:
This area here is a former working-class area, air quality is poor, and this means we have more respiratory illnesses which are the most frequent reasons for antimicrobial prescriptions.
(A, paediatrician, urban area, high prescription rate)
Similar direct contextual effects are evident in the density of paediatricians:
…This means service provision for children in an emergency is limited, and they are rather seen by GPs. And the GPs are fantastic, […], but they don’t have our special training and might be a bit more anxious if they see a child with a high fever….
(B, paediatrician, rural area, high prescription rate)
This low density then results in overload of the paediatricians, which in turn can increase antimicrobial prescriptions:
I mean on a Monday in February I have seen about 200 children, or thereabouts. And then I can’t start discussing for ages, this just doesn’t work.
(C, paediatrician, rural area, high prescription rates)
Suboptimal transition from in-patient to out-patient care were also seen to increase antimicrobial prescriptions in districts with higher prescription rates:
…in the hospitals, they prescribe broad-spectrum antibiotics. And I have to say, after we have sat down together a year ago and have talked about outpatient antibiotic therapies, we had agreed on not prescribing some particular antibiotics. And now I see that these exact antibiotics are still being used in the hospital.
(D, paediatrician, rural area, high prescription rates)
Contextual effects however also can constitute resources for lower prescription rates, for example in high-quality laboratories and quick turnaround times:
This means we can get samples to them three or four times a day and are not dependent on pickups once a day like in the practices out there. This really is a resource I think.
(E, paediatrician, rural area, low prescription rates)
Social influences
Social influences have been mentioned frequently, both as social influences through patients and through other health care providers. In particular where patient characteristics are being discussed, such influences could also be classified as compositional context resources (see above). However, as most of the quotes illustrate, these compositional factors also contain social influences.
Social influences as factors affecting high prescription rates are mainly located on patient level, illustrated in the following quote referring to patients with Middle-Eastern migration history:
This is a totally different culture, also affecting ideas about illnesses. Their ideas are totally different, and antibiotics are seen as miracle drugs – they are over the moon if they can get an antibiotic.
(F, paediatrician, urban area, high prescription rates)
However, the demand by patients is also being attributed to context effects such as dominating agricultural influences:
I think that there are lots of expectations for antibiotics by patients. For example, I do have a mother who generally insists on getting an antibiotic for her child, and I wouldn’t prescribe it. And I tell you how she says it: ‘I also give this to my pigs, so it can’t be bad for my kids’. So I think that antibiotic practices in the farms around here, I think that this means they (antibiotics) are applied liberally and happily, and the parents have experience and want them for their kids as well.
(C, paediatrician, rural area, high prescription rates)
At the same time, social influences are seen as malleable influences, in particular in combination with skills and knowledge which can then contribute to improvements in prescription practice:
It has become much better, yes. They (patients) now understand it, they have gotten used to it. And now we have, when the doctor says, you don’t need an antimicrobial, then more than half of them don’t go and see another doctor immediately and say ‘I need an antibiotic’.
(G, paediatrician, urban area, low prescription rates)
Knowledge
Knowledge included both information on current recommendations for antimicrobial prescribing, information on local resistance prevalence, information on local and personal prescription rates, and training content relevant to prescribing antimicrobials.
Participants from low-prescription districts mentioned knowledge on current recommendations as a resource and linked this knowledge to lower prescription rates within their districts:
We feel quite well informed. And everyone builds on that through individual research, further training and talking to colleagues. And I think, else we wouldn’t see these numbers.
(H, paediatrician, rural area, low prescription rates)
In contrast, paediatricians from high-prescription areas mentioned increased effort in obtaining relevant information:
[…] There is no information in the district, you always have to look after this yourself.
(I, paediatrician, rural area, high prescription rates).
In districts that had employed a paediatrician-initiated education programme (AnTiB; [34]), this programme was mentioned as an explicit resource:
We used to have this little informal guideline here in (city), which is also lying around in out-of-hours paediatric services and which every paediatrician here is likely to have in their practice. It is very useful and if you are doing emergency shifts, you pull it out of the drawer, look at the dosage and then prescribe.
(J, paediatrician, urban area, low prescription rate).
In contrast, the lack of specific knowledge in paediatric emergency services is seen as a barrier to effective prescribing:
We live in one of the areas with the most children in Germany, and, you can’t make this stuff up, we don’t have a paediatric out-of-hours service. This means out-of-hours is staffed by colleagues, e.g., urologists who have no clue, who start googling first – and then quickly prescribe an antibiotic.
(B, paediatrician, rural area, high prescription rate)
Skills
Skills as mentioned by the paediatricians include both discipline-specific and generic skills such as language skills or interpersonal skills.
Lack of specific treatment skills are mentioned as barriers to lower prescription rates by paediatricians in high-prescribing districts:
Perhaps the experience that as a urologist, you might not have that much experience with these really high fever temperatures in toddlers under two years.
(K, paediatrician, rural area, high prescription rate).
Similarly, a lack of language skills both on the side of the prescribers and patients is being seen as a barrier, both to non-prescribing and to instructing parents to monitor their children’s health:
… there is such a large language barrier which prevents you from explaining what the parents have to look out for, what are the signs of deteriorations, when do they need to come back, well, that this is a problem overall.
(L, paediatrician, urban area, high prescription rate)
Social and professional role
Social and professional role are mainly seen as a resource for low prescription rates. The main effects are seen to be indirect, via social norms and better professional networks. In some areas, this professional role is a relevant part of paediatricians’ identity which is used to be a role model to other paediatricians.
I think there are these lighthouse or role model practices here, the bigger ones. And they do this on purpose, to set standards and blaze a trail, and the younger colleagues or others then orient themselves on them.
(M, paediatrician, urban area, low prescription rates)
In addition, the social influence through networks is being seen as strengthened through social and professional roles and identity:
So we do have quite a number of colleagues who are really well connected. They always participate in our quality groups, participate very reliably, and have good contact amongst themselves.
(M, paediatrician, urban area, low prescription rates)
Beliefs about consequences
Beliefs about consequences tend to be related to contextual and environmental resources or barriers as well as regional outcomes. A particularly strong motive seems to be using antibiotics to prevent potential risks.
Paediatricians from districts with high prescription rates discuss avoiding consequences in particular with regards to patient overload:
My personal record in winter was 209 children a day. […] I have briefly checked them and then prescribed an antibiotic, because even if most of it is viral, you have children with whooping cough and I tend to be generous, because the hospitals are full of pneumonia.
(F, paediatrician, urban area, high prescription rates)
Paediatricians from districts with low prescription rates on the other hand discuss low beliefs about negative consequences such as patients changing doctors due to low competition pressure:
So we don’t really have a competitive mindset here, because changes from one paediatrician to the other are really, really rare.
(M, paediatrician, urban area, low prescription rates)
Interestingly, beliefs about consequences in terms of developing resistant microbes differ between paediatricians from low- and high-prescribing districts. Whereas those from high-prescribing districts argue that the responsibility for resistances is mainly located in the agricultural sector:
I think that resistant microbes develop if the farms in the area use lots of antibiotics […] So the kids who have MRSA here, they are all from farms. So they didn’t get MRSA because we gave them antibiotics but because the farms at home use lots of antibiotics.
(C, paediatrician, rural area, high prescription rates),
Those from low-prescription districts tend to attribute resistance development to health care professional behaviour:
The less antibiotics one prescribes, and if this happens everywhere, then we can expect, that the development of resistances will be less bad than elsewhere.
(H, paediatrician, rural area, low prescription rates)
Beliefs about competences
Beliefs about competences mainly revolved around perceptions of competence to influence local resistance developments and largely mirror those exemplified in the beliefs about consequences section.
Goals
Both paediatricians from low- and high-prescribing districts explicitly mentioned goals to prescribe less antimicrobials, and mention that these goals are also shared by colleagues in the respective districts. Differences exist in the context within goals are mentioned – paediatricians from low-prescription districts mention the goal of lower prescriptions as part of a combinations of goals (e.g., optimal therapy or limiting resistance development), paediatricians from high-prescription districts concentrate on potentially more relevant goals than lower prescription rates:
…I think I can speak for most of my colleagues here, one tries to prescribe as little as possible. But if they really all read the reports, do they change their prescription behaviour, I doubt that. There are quite some other problems here that need solving as well.
(F, paediatrician, urban area, high prescription rates)
Behavioural regulation
Behavioural regulation had only six mentions, but these were mainly together with contextual factors in districts with high prescription prevalence to highlight that contextual factors can pose barriers which also affect the low likelihood to change through impeding behavioural regulation:
And I think that these are basically deeply rooted, historic, ritualized prescription patterns, which then manifest regionally such that it is really difficult to change this.
(D, paediatrician, rural area, high prescription rates)
General Practitioners (GPs)
TDF domains on district level that affected GP prescribing behaviour (Fig.2) included context and resources (159 mentions), social influence (60 mentions), knowledge (41 mentions), beliefs about consequences (29 mentions), social/professional role (16 mentions), skills (16 mentions), goals (6 mentions), and behavioural regulation (4 mentions).
TDF domains mentioned as barriers (red) or resources (blue) by GPs
Context and resources
Similar to the paediatric participants, GPs reported on a range of regional contextual factors that influenced prescribing behaviour. These can also be differentiated along contextual and compositional factors [33].
A combination of contextual (main industry in the region) and compositional (migrant workers in the main industry) is a good example for these influences:
With the (migrant) workers in the meat industry, we do have a lot of people who might have potentially problems in dental hygiene, infections due to cuts for example. This happens a lot, and then increases the prescription of (antimicrobials).
(N, GP, rural area, high prescription rates).
GPs also report on regional differences in the influence of pharmaceutical representatives in their practices. For example, a GP from a low-prescription rural district mentioned that their local quality circles “will not invite pharmaceutical representatives if possible”.
Social influence
Social influences differ between districts, according to GP participants, and similar to paediatricians, these influences come through colleagues and patients.
One example for a local social influence could be long established GPs who influence local quality circles:
…as a young and newly arrived doctor, I quit going to the quality circles because the old guard was so present and influenced communication, work and thinking about practices. However, we do have now a new generation of GPs and things change.
(O, GP, rural area, low prescription rates)
Patient-level influences are also perceived to differ between districts, with some of the differences in expectations to be prescribed antibiotics being attributed to cultural factors:
There is a group of patients who are really eager to get antibiotics and who are incredibly demanding. Germans from the former Soviet Republics, and we do have many of them in this district. For them, it (not being prescribed antibiotics) is not a real therapy, even if it is viral….
(P, GP, rural area, high prescription rates)
Similar to cultural factors, the age distribution in a district is perceived to affect prescription, with more older adults in a district being associated with higher antibiotic demand.
Similar to this influence on higher prescriptions, specific regional social influences are also perceived as being influential for low prescription rates:
I mean, (city) is a very special city. It’s an administrative centre, a big university city, so I think there are a lot of people with a relatively high educational attainment, relatively little industry and I guess it’s also related to the fact that people have a bit of a different attitude..
(Q, GP, urban area, low prescription rates)
Knowledge
Similar to the results in paediatricians, knowledge about current recommendations, information on local resistance, and training content relevant to prescribing antimicrobials were seen as relevant resources. One particular additional factor was that in one of the participating districts, the local university was seen as influential for particularly rational prescribing behaviour:
I think that this is due to the fact that here in (city) there are many doctors who have studied in (city). And I remember from my studies that antibiotic prescriptions were an important topic, and that in microbiology et cetera we were always being reminded that one does not just prescribe antibiotics but needs to justify this really well.
(R, GP, rural area, low prescription rates)
At the same time, similar to the paediatricians, a lack of knowledge in out-of-hours services is seen as a relevant factor for high prescription rates:
But there are many colleagues working in the out-of-hours primary care and doing GP tasks who have for example an anesthesia background, or something else from the hospital, they don’t know it any better.
(N, GP, rural area, high prescription rates)
These knowledge factors interact with resources and barriers on context level.
Beliefs about consequences
Similar to paediatricians, beliefs about consequences include beliefs about having to avoid liabilities, which are often mentioned in combination with structural and contextual factors:
And if something does go wrong, and that’s always a problem in outpatient settings, you are the one who screwed it up. That’s what all the colleagues are afraid of. So the fear of making a mistake and not prescribing the antibiotic is always bigger than the fear of damaging something with the antibiotic.
(S, GP, rural area, high prescription rates)
Losing patients to other practices in situations with strong competition was a strong belief about consequences in districts with high prescriptions:
You can say, No I am not going to prescribe this, but then you lose the patient, they are just going somewhere else.
(T, GP, urban area, high prescription rates).
At the same time, a lack of such perceived consequences has been perceived as a resource for lower prescriptions:
…at least we don’t have to bow to patient demands too much. It is very different here compared to (city) where I was before, in the inner city, where there was a lot of competition due to too many GPs. You are much more likely to give in to irrational demands then.
(U, GP, rural area, low prescription rates)
Social / professional role
Specific regional ideas on the professional roles are perceived to influence prescription behaviour, in particular in combination with specific aspects of rurality that could affect the composition of the local GP structure:
I just see what kind of colleagues – to say it cautiously – are coming to this region, who take over old practices or establish new ones. They are not necessarily the most committed doctors.
(V, GP, rural area, high prescription rates)
Skills
In districts with low prescription rates, skills were mainly being mentioned with regards to interpersonal skills regarding expectation management with patients, which were perceived to be higher in the respective districts:
…in fact, skills training in multiple areas. General communication skills, difficult patients, bad prognosis, diagnosis, or making the patients understand why a particular therapy is indicated – these are all key skills and have always been emphasized during our studies.
(R, GP, rural area, low prescription rates)
Goals
Similar to paediatricians, GPs from both low- and high-prescription districts mention the goal of low prescription rates, and assume that their colleagues in the district have similar goals. GPs in low-prescription districts mention this goal as part of multiple goals (ideal therapy, avoid resistance) in low-prescription districts, GPs in high-prescription districts mention this goal as having lower priority compared to competing demands.
Behavioural regulation
Similar to the paediatricians, a lack of behavioural regulation in combination with contextual measures such as relatively old GPs in the district was seen as a risk factor for higher prescriptions:
Prescription behaviour by older colleagues plays a role I think. You can see this when you look at the age structure of the GPs here. They tend to prescribe antibiotics quickly whenever there are respiratory infections.
(W, GP, rural area, high prescription rates)
