Using theory to explore facilitators and barriers to delayed prescribing in Australia: a qualitative study using the Theoretical Domains Framework and the Behaviour Change Wheel

Background Delayed antibiotic prescribing reduces antibiotic use for acute respiratory infections in trials in general practice, but the uptake in clinical practice is low. The aim of the study was to identify facilitators and barriers to general practitioners’ (GPs’) use of delayed prescribing and to gain pharmacists’ and the public’s views about delayed prescribing in Australia. Methods This study used the Theoretical Domains Framework and the Behaviour Change Wheel to explore facilitators and barriers to delayed prescribing in Australia. Forty-three semi-structured, face-to-face interviews with general practitioners, pharmacists and patients were conducted. Responses were coded into domains of the Theoretical Domains Framework, and specific criteria from the Behaviour Change Wheel were used to identify which domains were relevant to increasing the use of delayed prescribing by GPs. Results The interviews revealed nine key domains that influence GPs’ use of delayed prescribing: knowledge; cognitive and interpersonal skills; memory, attention and decision-making processes; optimism; beliefs about consequences; intentions; goals; emotion; and social influences: GPs knew about delayed prescribing; however, they did not use it consistently, preferring to bring patients back for review and only using it with patients in a highly selective way. Pharmacists would support GPs and the public in delayed prescribing but would fill the prescription if people insisted. The public said they would delay taking their antibiotics if asked by their GP and given the right information on managing symptoms and when to take antibiotics. Conclusions Using a theory-driven approach, we identified nine key domains that influence GPs’ willingness to provide a delayed prescription to patients with an acute respiratory infection presenting to general practice. These data can be used to develop a structured intervention to change this behaviour and thus reduce antibiotic use for acute respiratory infections in general practice. Electronic supplementary material The online version of this article (doi:10.1186/s12875-017-0589-1) contains supplementary material, which is available to authorized users.


Background
Antibiotic use contributes to the development of bacterial resistance, [1,2] which is a significant threat to public health [3]. Acute respiratory infections (ARIs) constitute more than half of all problems treated with antibiotics in general practice [4], but many will resolve without the need for antibiotics [5][6][7]. Antibiotic resistance appears to be reversible if the selection pressure from antibiotic use is removed [8].
"Delayed prescribing" (also known as "wait-and-see" or "back-up prescription" [9,10]) is one approach that reduces antibiotic use -93% of those given an immediate prescription will fill it compared to 32% of those given a delayed script [11]. For patients suffering from cough or common cold, there are no significant differences in duration and severity measures for pain, malaise, fever, cough and rhinorrhoea between patients given a delayed, immediate or no antibiotic prescription. For patients suffering acute otitis media and sore throat, immediate antibiotics do provide slightly quicker relief than those given a delayed prescription but there is an increase in adverse reactions (including nausea, vomiting and diarrhoea) [12,13]. Patients asked to delay antibiotics were only slightly less satisfied (87%) with care received than those given an immediate antibiotic prescription (92%) [11]. Providing a delayed prescription does not impact re-consultation rates [11] or increase emergency department presentations, [13] is associated with better outcomes [14] and is popular with patients [15].
Delayed prescribing can be used for acute otitis media, acute sore throat/pharyngitis/tonsillitis, the common cold, acute rhinosinusitis and acute cough/bronchitis. Delayed prescribing is the process of a general practitioner (GP) making available an antibiotic prescription during the consultation, but asking patients to delay its use, to see if symptoms will resolve first. Patient concerns and expectations should be addressed when GPs and patients agree to using delayed prescribing. Advice should be given about using the delayed antibiotic if symptoms do not stabilise in accordance with the expected course of the illness or symptoms worsen significantly. Prescriptions can be made available in different ways: patient recontacting the practice subsequently to request it by phone; patients collecting their pre-written prescriptions from the practice after the consultation if their condition worsens; post-dating the prescription; and giving prescriptions to patients while asking them to wait some days before having it dispensed. This last method seems most suitable for the Australian fee-forservice health system. Delayed prescribing has not been implemented in Australian general practice, and local facilitators and barriers to the uptake of this intervention have not been explored. It is the authors belief that delayed prescribing could be one successful strategy Australian GPs could utilise to reduce the overuse of antibiotics.
A growing body of research [17][18][19] supports the use of psychological theories to identify local facilitators and barriers to changing clinicians' behaviour [20,21]. The Behaviour Change Wheel (BCW) is an eight-step guide to designing interventions using a theoretical approach [22]. The first four steps help researchers understand the target behaviour: (1) define the problem in behavioural terms, being specific about the target population and the behaviour itself; (2) select the target behaviour from a list of potential competing behaviours; (3) specify the target behaviour in terms of who needs to do what, when, where, how often and with whom; (4) identify what needs to change in the person or environment in order to achieve the desired behaviour by analysing interview data using the Theoretical Domains Framework (TDF) [23]. The TDF comprises 14 theoretical 'domains' representing a range of possible theory-based facilitators and barriers to behaviour change [24].
The primary aim of this study was to use the BCW and the TDF to define delayed prescribing by GPs in behavioural terms, and to identify GPs' facilitators and barriers to changing this behaviour. The secondary aim was to explore pharmacists' and the public's views on delayed prescribing, using the TDF as a framework for analysis.

Methods
GPs, pharmacists and members of the public completed a single face-to-face, one-on-one, semi-structured interview (February 2014 to July 2015) with the principal researcher (LS, who had previous interviewing experience and no prior relationships with any participant).

Inclusion and exclusion criteria
Eligible GPs were practising and regularly seeing patients with ARIs. Eligible pharmacists were practising in community pharmacy, and eligible members of the public were those between 18 and 65 years whom accepted the invitation to participate in the research.

Recruitment strategy
We recruited participants from the Gold Coast and the Sunshine Coast, Queensland, and continued until data saturation was reached. LS transcribed the interviews. Recruitment ceased when all domains had been populated and no new themes within each domain emerged [25].
We recruited GPs practising on the Gold Coast by telephone using snowball sampling. One researcher (CDM) provided a list of 8 GPs to approach for interview. LS telephoned them directly (two agreed). After each interview was completed, the researcher requested a name of another GP to contact. This continued until 5 GPs had been interviewed and no new contacts occurred. GPs on the Sunshine Coast were recruited in person using convenience sampling. Medical practices were identified using an internet search. LS walked into medical practices and requested to see the practice manager. LS extended an invitation to participate which the practice manager distributed via email to all GPs in that practice. GPs contacted the researcher directly by telephone or email. Thirteen GPs agreed to be interviewed. Due to difficulties in recruiting GPs, GPs recruited later in the study (mainly the Sunshine Coast) were offered compensation for the time taken for interview (between $70 and $100 depending on length of interview; figures derived from 2015 Royal Australian College of General Practitioners Medicare Benefits Schedule fee summary) [26]. Only half of the GPs interviewed took the offer of compensation. The researcher arranged an appointment with GPs, who were aware that the consultation was for research, not medical purposes.
We recruited community pharmacists on the Sunshine Coast using convenience sampling and face-to-face discussion with pharmacy managers, who either agreed to be interviewed or allocated a member of staff for interview.
We recruited members of the public (in person) in shopping malls, schools (parents only), universities and coffee shops, and some sitting on park benches, using maximum variation sampling (across gender, age and educational status). LS kept a table to document age, gender and educational status of the public interviewed. Gaps in the table identified the ideal demographic of next participant.

Setting
Interviews took place in locations convenient for participants: (1) GPsat home, in general practices and coffee shops; (2) pharmacistsin their pharmacies; (3) the publicat home, outside school gates and in coffee shops. No others were present during interviews. Figure 1 shows the process of data collection and analysis.

Data collection and analysis
Step 1 (Define the problem in behaviour terms): We (LS and AMcC) used three questions to define the problem [22]: (1) What behaviour?; (2) Where does the behaviour occur; (3) Who is involved in the behaviour?
Step 2 (Select the target behaviour): We (LS and AMcC) generated a list of all potential GP behaviours relevant to the target behaviour, including influences out of scope (see Additional file 1). To identify the most important target behaviour, we used four criteria [22] and generated a rating system for each criteria: (1) How much of an impact will changing this target behaviour/issue have on outcome? (1 = highly unlikely, 2 = likely and 3 = very likely); (2) How likely it is that the behaviour can be changed? (1 = yes, 0 = no); (3) How likely it is that the behaviour (or group of behaviours) will have a positive or negative impact on other related behaviours? (1 = negative or no impact, 2 = potential impact, 3 = positive impact); (4) How easy will it be to measure the behaviour? (1 = difficult, 2 = possible, 3 = easy). A total score for each behaviour on the list was generated. LS and AMcC used this score along with discussion to reach consensus about which behaviour to target.
Step 3 (Specify the target behaviour): We defined the target GP behaviour from Step 2 using the following questions [22]: (1) Who needs to perform the behaviour? (2) What does the person need to do differently? (3) When will they do it? (4) Where will they do it? (5) How often will they do it? (6) With whom will they do it?
Step 4 (Identify what needs to change to achieve target behaviour): We collected two rounds of interview data for this step (Fig. 1). Round 1 interview questions explored broad factors influencing GPs' , pharmacists' and the public's beliefs on delayed prescribing (see Additional file 2). We did not design this topic guide using the TDF and had not defined the problem in behavioural terms when these interviews were conducted, but did complete steps 1-3 before analysis of Round 1 data took place, allowing us to identify domains not explored that may have been important. We piloted interview questions with one GP, 3 pharmacists and 3 members of the public. No pilot interviews were used in the final data set. No repeat interviews were required. Two GPs requested a copy of the transcripts before analysis. Transcribed interviews were sent via secure email, with a request to reply within 2 weeks if issues were identified; no replies were received. Field notes were made after each interview for use during analysis.
LS recorded and transcribed verbatim all Round 1 interviews. Two researchers (LS and AMcC) coded each transcript independently by assigning each relevant piece of data to one or more TDF domains using NVivo (10 QSR International). All coding was discussed and agreed by the two researchers, one of whom (AMcC) had previous experience using the TDF [27]. From this process, we identified that some domains had not been fully explored (skills, beliefs about capability, optimism, intentions, goals, reinforcement and emotion). We then modified the topic guide (interview questions) to target the domains that required further exploration (see Additional file 3). Two researchers (LS and AMcC) analysed the interviews as previously described for Round 1. In addition, within each category, the researchers (LS and AMcC) identified facilitators and barriers to the target behaviour. The process was iterative and, as new data were added to TDF domains, facilitators and barriers were modified, expanded or re-categorised.
We (all authors) agreed relevant domains (i.e. those directly influencing the target behaviour) by consensus using three questions: (1) What needs to happen for the target behaviour to occur? (2) Is there need for change?

Results
Eighteen GPs (8 females, mean age 43 years [range 29-61 years], mean 13 years in practice [range 1-33 years]) with 7 GPs using delayed prescribing, 2 regularly. Two GPs interviewed were married to each other and worked at the same practice but had opposing views on delayed prescribing. Nine pharmacists and three pharmacy assistants (11 female, mean age 36 years [range 24-56 years], mean 10 years in practice [range 1-34 years]) participated. Approximately 35 members of the public were approached for interviews. Fourteen agreed to be interviewed (9 female, mean age 40 years [range 20-61 years], educational status ranged from completion of secondary school to completion of tertiary degrees). One pharmacist withdrew due to time constraints; no GPs or members of the public withdrew. Mean interview time was 29 min (range 13-45 min). Figure 2 illustrates the key results from steps 1 to 4. We defined the problem as the need for GPs to change from prescribing immediate antibiotics to delayed prescription, and specified when and how this should happen. We identified nine relevant domains that influence this behaviour, briefly discussed below. Table 2 provides example quotations from participants that pertain to facilitators and barriers within these domains. The 5 domains identified as irrelevant were behavioural regulation, professional/social role and identity, beliefs about capabilities, reinforcement, and environmental context and resources (Table 1). GPs believed prescribing is part of their role, with delayed prescribing an extension of this role (professional/social role), and that they have the capability to do it (beliefs about capabilities). GPs did not believe systems aimed at monitoring (behavioural regulation), or punishments and rewards (reinforcement), would be appropriate. GPs believed resources (environmental context and resources) are important to help patients manage their symptoms and follow instructions for delayed prescribing; however, they also believed these resources are not relevant to their adopting delayed prescribing.

Knowledge
All GPs had heard of delayed prescribing but did not use it correctly or only used it selectively. GPs knew that delayed prescribing involves giving an antibiotic prescription to the patient with instructions on how long to wait and when to take the antibiotics if symptoms worsen.
"…if you are no better in a couple of days, fill the script. Um, so I am thinking how often I do it, um, I do do that occasionally, especially if it is near the weekend or something and they can't come back to see me." (GP16) All pharmacists were knowledgeable about management of ARI symptoms. The public reported needing to know when to start antibiotics (if given a delayed prescription) and how to treat their symptoms.
"So I would need to know when to take the antibiotic, so when is the cold getting worse, how long this takes, the expected length of time for the cough and how it would impact my breathing." (Pt15) All participants knew antibiotic use caused antibiotic resistance.     Delayed prescribing would provide an educational opportunity "…if we take the time to educate patients here, then they learn…they will get better without antibiotics." (GP19) "…like really, this is a great educational moment" (Ph12) "then maybe that message might get to people" (Pt07) Delayed prescribing would support professional satisfaction "…I just know I have done the right thing." (GP05) "…it [the reason to adopt delayed prescribing] is going to have to be a professional incentive, you will know you have stopped them getting an antibiotic when you thought it wasn't needed". (Ph10) n/a Barriers Patients misusing prescriptions "Some people are just going to go out and fill the prescription anyway." (GP19) "It would be interesting to see how many delay." (Ph12) "Oh 95% would walk in and get the antibiotic." (Pt06)

Missing a serious diagnosis
"…will I miss a pneumonia or a sepsis in the next few days?" (GP05) n/a "I would be worried of it [the illness] getting worse…" (Pt15) Losing business "…so losing patients" (GP05) "We might actually lose business because we wouldn't fill a prescription."  "Most people don't want medication these daysthey want the vitamins and the zinc! It' s such a better way to do 'sick' , well, I think anyway." (Ph12) "I am so vocal that I don't want antibiotics, so they don't feel the need to do that with me." (Pt03) Trust in patients "Then they [the patient] learn for the rest of their livesthey will get better without antibiotics but not just thatthey also learn that we have trusted them and it all worked out well." (GP19) -"They [GP] are kind of going well, I am trusting, that would mean they would be trusting me as a mum, to know when my child was getting worse or better. And that I can make that judgement call myself" (Pt04) Relationships with regular patients make it easier to refuse antibiotics "Most patients that are regular that we see all the time, they are more likely to know what we will do and again if they know you reasonably well, it is easier to get them not to take thingsso if they haven't got a relationship with you, you are more likely to prescribe." (GP04) "…mainly because we have regular customers who believe what I say. We do have that relationship. They know I have been right before and I am not making things up or trying to sell them stuff" (Ph07) "I wouldn't have a problem at all [if my GP asked me to wait], I have a reasonably good relationship with my GP, I don't always agree with what he says but… then if he asked me to wait, that would be fine with me." (Pt14)

Some patients just want reassurance
"'You are doing it all ' correctly' and then they [patients] feel reassured and maybe that has a bit of a placebo effect that they are speaking to the doctor and he has said everything is going to be OK" (GP05) -"just going to the doctor and him telling me I am going to be OK, if he said I didn't need antibiotics then I would be fine with this, really, I don't mind being told not to take tablets or to wait for that matter" (Pt13) GPs don't feel pressured by patients who demand antibiotics "Well, I just I guess I tend to be a bit stubborn at times. If they really get up my nose, then I explain as nicely as possible all the reasons that I ah, believe it' s viral. But I will dig my heels in if they really insist" (GP11) n/a "Another doctor, a younger doctorhe works at the hospital but he also has rooms at the surgery, he wouldn't give me an antibiotic. He just flatly refused." (Pt02) GPs are a trusted profession "A position of trust I think. Being a doctor. Patients listen to you." (GP15) "…because not only are you getting the information from the doctor who you trust…" (Ph08) "I do think GPs are very trusted people." (Pt09)

Barriers
Expectations "The common cold is very very common of course and people come in expecting to have antibiotics." (GP14) "Do you think it is because the patient goes to see the doctor expecting something?" (Ph09) "…in my generation we were brought up to believe that antibiotics fix everything." (Pt10)

Skills
GPs considered they did not need any additional physical skills to conduct delayed prescribing. GPs reported using interpersonal skills to explore patient expectations regarding antibiotic prescription, although they seldom explicitly ask patients about their expectations.
"[I do not ask them about their expectations]… in such a straight forward waylike, I normally ask them what they have been doing to cope before they come to me for help." (GP19) All pharmacists reported using interpersonal skills in their interactions with the public to identify customers' needs. Pharmacists did not believe they need additional skills to conduct delayed prescribing because talking to the public about prescriptions is a skill they use frequently.
"…we are really big on coming down and talking to everyone about their medication, this would be another thing that we already do." (Ph10) The public didn't think they need any skills for delaying taking an antibiotic.

Memory, attention and decision-making
GPs who said they use delayed prescribing use it in highly specific situations: for patients whom they believe are well educated and sensible, and whom they trust will delay filling the prescription. Further, they identified that the only reasons they would use delayed prescribing are if patients have difficulties accessing a GP, are going on holiday, are suffering from otitis media, or demand antibiotics. Some GPs stated they would never use delayed prescribing because they prefer to bring patients back for review.
"…if they perceive the condition isn't resolving and to come back if they want." (GP10) Pharmacists had no comment on this domain. The public stated they would wait for worsening symptoms before taking antibiotics but, if a previous infection had progressed to a severe one or they were in pain, they would be more likely to want to take antibiotics immediately.

Optimism (and pessimism)
The minority of GPs who said they already use delayed prescribing regularly were optimistic about using this strategy. Most GPs believed that it would be difficult to implement delayed prescribing: they do not trust patients to delay and do not believe there is enough evidence to refute this perception. Pharmacists and the public were optimistic about using delayed prescribing.
"Some people are just going to go out and fill the prescription anyway." (GP19) "I like the idea that I might be able to make my own decision about when to take antibiotics, this doesn't frighten me at all" (Pt13)

Beliefs about consequences
Participants (GPs, pharmacists and the public) believed antibiotic resistance is a serious issue influenced by antibiotic use. They identified advantages of delayed prescribing as reduced antibiotic use and resistance, Giving up control "Some GPs have this idea that it' s better that you do it (prescribe antibiotics), and then at least you are in control of the situation" (GP07) n/a "If the GP said 'Here is a prescription for antibiotics' I would take them." (Pt12) '-' , Not reported; n/a not applicable providing an opportunity for patient education, and giving professional satisfaction.
"I think it would be goodit would work, education would be for the people using it" (GP16) "Um, I use it as it means less reviews, less people coming in, I guess giving patients control over their condition." (GP09) "Um, I guess it makes me feel I am just playing a small part in reducing antibiotic resistance." (GP15) But they were also concerned that the public would misuse the delayed prescription by either using it immediately or keeping it for a subsequent infection. GPs were specifically concerned about missing serious infections, that delayed prescribing could be more time-consuming and that it could potentially lose them business.
"…but it is easier to give the script because the patient leaves the room immediately. So most of that is time pressure." (GP06)

Intentions
GPs who regularly use delayed prescribing said they do not have to consciously think about itthey believe it is an effective strategy and intend to use it as often as they can. GPs who use delayed prescribing selectively said they need to think carefully if patients fit certain criteria. Most GPs stated they have no intention of using delayed prescribing because they can bring patients back for review. All pharmacists stated that, if they have the time, they intend to talk to patients who request an antibiotic prescription be filled. The public said they intend to do what the GP asks of them.

Goals
Most GPs did not mention that they had a goal of using delayed prescribing. One GP expressed a strong goal of using delayed prescribing if patients were able to make an informed decision on their care. All groups identified the main goal of sick patients is to feel better.
"If I was bedridden then I would take anything to get better." (Pt08)

Emotion
This domain overlaps with the 'beliefs about consequences' domain -GPs and the public expressed a fear of missing a serious diagnosis -one GP stated he is a regular prescriber of antibiotics because he had had some patients experience severe complications from not starting antibiotics sooner. GPs believe delayed prescribing reduces patient anxiety about disease progression because they have a management plan in place.

Social influences
Several conflicting social influences were evident. These influences centre on perceived expectations of antibiotic treatment and on trust between clinicians and patients. All GPs said they do not feel pressured when patients demand antibiotics, but they reported mixed views on whether patients expect antibiotics. The time and day of the week, patients' social situation (e.g. getting back to work, financial status, age and comorbidities) and an existing relationship with a patient influence their antibiotic prescribing. GPs who use delayed prescribing reported the practice encourages trust between GPs and patients. Other GPs reported they don't trust their patients; however, they contradicted this statement by saying they believe that their patients listen to them. "… if we take the time to educate patients here, when they are sitting in the chair, they go home, they don't use the antibiotic and then they get better, if we are talking about the best case scenario, then they learn for the rest of their livesthey will get better without antibiotics but not just thatthey also learn that we have trusted them and it all worked out well." (GP19) Pharmacists acknowledged many people do not want antibiotics to treat their illness and would rather have natural remedies; however, pharmacists also believe some people go to the GP expecting antibiotics. Many members of the public stated they do not want antibiotics if they don't need them; all identified good relationships with both GPs and pharmacists as important because they often want reassurance they are doing the right thing. The public reported trusting GPs, saying they would do what the GP asks of them and that delayed prescribing indicates that the GP trusts them to make decisions about their own health.
"Even if it was also just re-assurance that I knew what I was talking aboutand then he wouldn't need to hand over that antibiotic prescription straight over, and then I wouldn't need to start them on it straight away." (Pt04)

Main findings
To implement delayed prescribing in Australian general practice, GPs need to change from immediate antibiotic prescribing to delayed prescribing and educate patients about how to manage ARI symptoms and when to start antibiotics. This education can take place during any consultation with a person presenting with an ARI in a general practice setting. Nine TDF domains influence this behaviour: knowledge; skills; memory, attention and decision-making; optimism; beliefs about consequences; intentions; goals; emotion; and social influences. Pharmacists and members of the public are broadly supportive of using delayed prescribing; whilst most had not heard of delayed prescribing, they were optimistic that it could be a useful strategy to reduce antibiotic use.
Although effective, uptake of delayed prescribing by general practitioners (GPs) appears to be low: one study found that 15% of patients prescribed antibiotics in the past year had been given the option of a delayed prescription [16]. These findings may be helpful to researchers trying to implement an intervention to support delayed prescribing in other fee-for-service environments.

Strengths and limitations
The main strengths of this study are the systematic and structured approach to defining and specifying, in behavioral terms, the problem of implementing delayed prescribing, and the use of a theory-driven approach to identify facilitators and barriers to changing this behaviour. Interviewing pharmacists and members of the public ensured that views from these two key groups of stakeholder can be used to inform the development of a future intervention.
The main limitation of this study is that the TDF was not used to design the topic guide for Round 1 interviews as the BCW work book was not published and the TDF was not identified as a method for analysis until after the interviews had commenced. However, the TDF was used for the interim analysis and to identify gaps in the emerging data that were subsequently addressed by modifying the interview questions. More than one-third of GPs in this sample reported using delayed prescribing higher than the uptake rates estimated earlier, [16] which could indicate selection bias, in that GPs who were supportive of delayed prescribing were more likely to agree to be interviewed. There could also have been social desirability bias in all interviews where participants answered the questions in a manner that would be viewed as favourable by the researcher. Snowball sampling of GPs on the Gold Coast and convenience sampling of participants on the Sunshine Coast may also have introduced selection bias. Further, most of the pharmacists interviewed were female. Participants were encouraged to arrange interviews at a time and place most convenient to them, with most interviews of GPs and pharmacists occurring at workplaces. This factor may have affected participants' responses due to time pressure, and may have led to participants' feeling less able to disclose certain information. The decision about whether to prescribe antibiotics for patients with ARIs is potentially a sensitive issue. Hence, GPs may be reticent to divulge overprescribingonly one GP admitted to being a high prescriber. We did not collect data on GPs' current rate of antibiotic prescription.

Comparison with existing literature
Similar to findings in other qualitative studies of delayed prescribing, [28][29][30][31][32][33], we found that Australian GPs were concerned about the misuse of prescriptions by patients, diagnostic uncertainty and patient hoarding of medication. We also found that Australian GPs were often highly selective regarding the patients for which they used delayed prescribing. Some of these concerns are not consistent with the evidence: 32% of patients fill a delayed prescription whilst 93% fill an immediate prescription [11,20,34,35]. Hoarding could be a valid concern. When asked about repeat prescriptions (for any indication), one-third of Australian respondents stated they would keep prescriptions for future use [36]. To our knowledge, this study is the first to explore facilitators and barriers to delayed prescribing using the TDF and the BCW. However, this approach has been used successfully for developing other interventions [27,[37][38][39].

Implications for clinicians and policy makers
There is adequate evidence that delayed prescribing reduces antibiotic use, [11] and the concept has been previously introduced into the Australian GP environment [40]. However, our interviews indicate the practice is not consistently used, with many GPs expressing concerns about its implementation. GPs can use the results of this study to identify their individual facilitators and barriers to using this strategy in their own practice. Policymakers and researchers will need to consider the facilitators and barriers to delayed prescribing identified in this study before designing an acceptable implementation intervention.

Unanswered questions and future research
We have identified theory-derived facilitators and barriers to the implementation of delayed prescribing in Australian general practice. The next step is to use the BCW to identify which behaviour-change techniques could be used to target the TDF domains that influence the use of delayed prescribing by Australian GPs, and then to test the feasibility of this theory-based intervention in Australian general practice.