ABSTRACT
Background Antimicrobial resistance is a global patient safety priority and inappropriate antimicrobial use is a key contributing factor. Trials have shown that delayed (back-up) antibiotic prescriptions (DP) are an effective and safe strategy for reducing unnecessary antibiotic use but its uptake is controversial.
Methods We conducted a realist review (a literature review which goes beyond assessing whether an intervention works) to ask why, how, and in what contexts general practitioners (GPs) use DP. The review is focused on those who wish to use DP and not for those who are against using DP. We searched five electronic databases for relevant articles and included DP-related data from interviews with healthcare professionals in a related study. Data were analysed using a realist theory-driven approach – theorising which context(s) influenced (mechanisms) resultant outcome(s) (context-mechanism-outcome-configurations: CMOCs).
Results Data were included from 76 articles and 41 interviews to develop a program theory comprising nine key and 56 related CMOCs. These explain the reasons for GPs’ tolerance of risk to different uncertainties—epistemological (knowledge-orientated); scientific (data-orientated); hermeneutic (interpretation-orientated); practical (structures/processes-orientated); technological (skills/software/equipment-orientated), and existential (world-view-orientated). These interact with GPs’ work environment, self-efficacy and perceived patient concordance to make using DP as a safety-net or social tool more or less likely, at a given time-point. Our program theory explains how DP can be used to mitigate some uncertainties but also provoke or exacerbate others.
Conclusion This review explains how, why and in what contexts GPs are more or less likely to use DP, as well as various uncertainties GPs face which DP may mitigate or provoke. We recommend that efforts to plan and implement interventions to optimise antibiotic prescribing in primary care should consider these uncertainties and the contexts when DP may be (dis)preferred over other interventions to reduce antibiotic prescribing. We also recommend the following and have included example activities for: (i) reducing demand for immediate antibiotics; (ii) framing DP as an ‘active’ prescribing option; (iii) documenting the decision-making process around DP; and (iv) facilitating social and system support.
SUMMARY BOX
What is already known on this topic – Trials have shown that delayed (back-up) antibiotic prescriptions (DP) are an effective and safe strategy for reducing unnecessary antibiotic use but its variable uptake in primary care remains to be understood.
What this study adds – This realist review provides a program theory to explain the complexity and interactivity of influencing factors on general practitioners’ (GPs) antibiotic prescribing decisions. It explains how GPs have a reasoned tolerance of risk to various uncertainties (beyond clinical uncertainty) which interact with GPs’ work environment, self-efficacy and perceived patient concordance to make using DP as a safety-net or a social tool more or less likely, at a given time-point. It applies nuanced concepts from the uncertainty literature - epistemological uncertainty (knowledge-orientated), scientific uncertainty (data-orientated), hermeneutic uncertainty (data interpretation-orientated), practical uncertainty (structures/processes-orientated), technological uncertainty (skills/software/equipment-orientated), and existential uncertainty (world-view and identity-orientated) - to better understand DP clinical decision-making.
How this study might affect research, practice or policy – Policy makers, commissioners, and prescribers who would like to optimise antibiotic prescribing should become familiar with the varieties of uncertainties at play when GPs consult with patients and consider how these different uncertainties are mitigated and/or provoked when developing support interventions to optimise DP or implementation support.
BACKGROUND
Drug-resistant infection associated with antimicrobial resistance (AMR) was attributed to 1.27 million deaths globally in 2019[1] and is a global patient safety priority.[1] Antimicrobial stewardship (AMS) promotes and monitors prudent use of antimicrobials, and is critical in healthcare to maximise patient benefit from immediate antibiotic treatment and to preserve future antimicrobial effectiveness by decelerating the development and spread of AMR.[2] Trials have demonstrated effectiveness of many AMS strategies within primary care.[3] One strategy, with evidence dating back to 1997[4] and advocated in national guidelines[5,6] is delayed (back-up) antibiotic prescriptions (DP). DP has shown to be effective for managing acute cough, acute sore throatClick or tap here to enter text., acute otitis media, sinusitis, and lower urinary tract infections.[7–20] DP is a prescription given to a patient (directly, or by post-dating, recontact, or collection[21]) with advice that antibiotics are not needed immediately and how to use the DP if symptoms worsen or do not improve after a certain time.[6,22,23]
Evidence suggests DP may reduce antibiotic use by patients[21,33] and may mitigate clinical uncertainty by providing a safety-net to minimise the risk of developing severe complications,[15,34–38] whilst maintaining patient satisfaction and not adversely impacting symptom severity or duration and reconsultation rates.[7–9,39] It is considered safe for most patients, including some in higher risk subgroups.[39] DP helps patients understand that antibiotics are not always needed, empowers patients in self-management, and reduces reconsultations.[40]
Those who do not favour DP, argue that the risk of harm from delayed antibiotics to an individual may outweigh potential benefits from reduced antibiotic use[41] and that DP confuses patients through mixed messages about antibiotic appropriateness.[28] Clinicians may be reluctant to transfer the responsibility of treatment decision-making to patients, believing it to unfairly burden the patient[28,37,42–47] or that the patient might use the antibiotic immediately or store for future use.(13) DP may also be less relevant in practices where reconsultation is the preferred safety-net[29,43] or in some rural and dispensing practices.[40]
Around a quarter of a century after the first evidence of DP as an effective AMS strategy,[4] a lack of reliable routine DP data hinders assessment of its uptake[7,24,25] and our understanding of what drives general practitioners’ (GPs) decision to use DPs or not is unclear. This study does not seek to resolve the controversy behind the pros and cons of DP use by GPs. Instead, the aim is to explain how, when and why it is used, therefore enabling the findings to inform design of DP implementation strategies in relevant infection pathways to support shared decision-making and enhance patient care.
METHODS
We have conceptualised DP use by GPs as a complex phenomenon where usage (or not) is likely to be dependent on context. Unlike other forms of systematic reviews, which summarise the effectiveness of an intervention, we have conducted a realist review to understand how context influences the implementation of an intervention. A realist review asks how, when, why, and in what circumstances an intervention works.[51] Our realist review had six iterative overlapping stages[52] conducted in accordance with the Realist And Meta-narrative Evidence Syntheses: Evolving Standards (RAMESES).[53] This study is part of a wider research program (STEP-UP, https://www.expmedndm.ox.ac.uk/step-up). Our stages of the review process are set out in Table 1 and Box 1 highlights the additional concepts of uncertainty in healthcare literature that was also used to develop our program theory.
Concepts of uncertainty in healthcare literature which may be helpful to understanding DP use
Fears and hopes of uncertainty
When uncertainty is present, clinicians will usually be guided by the widely held professional principle of non-maleficence (‘to do no harm’). This is typically associated with risk of adverse events (e.g. from treated or untreated infections) and negative emotions. This may manifest as fear and anxiety about missing diagnosis, feeling vulnerable to complaints or disciplinary action. Clinicians may usually be motivated to avoid, mitigate or prevent uncertainty as much as possible or practical[62,63] (e.g. by prescribing immediate antibiotics).
Uncertainty also pertains to beneficence (‘to do good’) and to positive emotions (e.g. courage, confidence, curiosity) and hope for beneficial outcomes, e.g. that the patient learns that their illness may be self-managed without antibiotics. Clinicians may thus be motivated to ‘embrace’ uncertainty[66,67] when they realise that, paradoxically, some benefits are gained only when risks are taken.[68,69]
Early conceptualisations of uncertainty in healthcare
Uncertainty in healthcare is commonly understood as ‘clinical uncertainty’—where medical knowledge, though dynamically changing with generation of new knowledge, is always limited, as is the mastery of this knowledge by any one individual doctor.[59,64,65]
One aspect of this knowledge is information about the risk (probability) of a beneficial or adverse event (e.g. probability that 20% of patients will benefit or be harmed from treatment). Uncertainty may arise from ambiguity of this risk information regarding its reliability, credibility or adequacy (e.g. confidence interval of 10-30% probability that patients will benefit/be harmed with expert disagreement on benefits/harms, and insufficient scientific evidence). It may also arise from its complexity, when its features make it difficult to understand or apply (e.g. 20% probability of benefit/harm for a certain type of patient with/without a multiplicity of risk or causal factors and interpretive cues).[63]
Towards a more nuanced conceptualisation of uncertainty in healthcare
Several types of uncertainty are already conceptualised in the wider healthcare literature,[62–64,70,71] and these may be important to understanding use of DP as an AMS strategy:
- epistemological (around knowledge)[70] and scientific (around data, diagnosis, prognosis, cause, and treatment)[62,63] (both of which may be akin to clinical uncertainty)
- practical (around structures and processes of care)[62,63]
- technological (around use of equipment or software)[70]
- hermeneutic (around interpreting data, e.g. test results or patients’ narratives)[70]
existential (around issues of personal identity, worldview or meaning,[63,64,70,71] from professional (what it means to be a ‘good’ doctor)[70] and patient[63,71] perspectives).
RESULTS
We included 76 articles (Supplemental file 4), comprising 50 empirical studies, nine literature reviews and 17 commentaries. We also included excerpts from interviews with 22 healthcare professionals involved in local AMS implementation and 19 general practice professionals ( Supplemental file 2)
Program theory: Influence of uncertainties and risk tolerance in context on DP use
Our program theory explains how different antibiotic decision-making contexts may shift during a primary care consultation and that DP decisions are influenced by two important concepts, namely the way GPs respond to different types of uncertainties and their tolerance of risk to these uncertainties.
Broadly, for GPs not ideologically averse to DP, we found that when GPs had a lower tolerance of risk but were uncertain if an immediate antibiotic prescription was needed, then they were more likely to use DP. In this situation, as scientific and epistemological uncertainty increased (see Box 1 for types of uncertainties and Supplemental file 5 for an illustration of the program theory), the GPs tended to increase DP use. However, the behavioural pattern above is influenced by: 1) perceived value of DP as a safety-net; 2) perceived patient concordance (i.e. perceived agreement after a collaboration to incorporate “the hopes, beliefs and actions of prescriber and recipient”[72] and GP self-efficacy to facilitate this; 3) perceived value of DP as a social tool; and 4) work environment (Table 2). Importantly, the context may change during the consultation as more information becomes apparent. Our program theory illustration (Error! Reference source not found.Supplemental file 5) represents snapshots of different antibiotic decision-making contexts, and how these may shift, from a nominal starting point (e.g. 61% immediate; 25% no antibiotic; 13% DP; for acute cough[7) during a primary care consultation.
Program theory comprising four broad areas of influence encompassing nine context-mechanism-outcome-configurations (CMOCs) -key theories explaining general practitioners’ (GPs) decision-making for immediate antibiotic prescription, delayed prescription (DP), or no antibiotic. CMOCs 1.1 -9.10 provide further granularity on specific contexts and mechanisms that when present may increase or decrease use of DP.
Within the wider program theory, we identified nine key[73] interacting context-mechanism-outcome configurations (CMOCs) that explain shifts in a decision between immediate, DP or no antibiotic (Table 2). Each key CMOC was further broken down into two or more related CMOCs (n=56) to provide further granularity on contexts and mechanisms that may increase or decrease the likelihood of DP use. We highlight examples associated with different types of uncertainty under each explanation (Supplemental file 6).
I. The value of DP as a safety-net
The safety-netting value of a DP is apparent when a GP is scientifically or epistemologically uncertain about whether the infection is viral, bacterial, self-limiting and/or whether a patient would deteriorate without timely antibiotics, and believes DP allows patients access to antibiotics if needed (CMOC:1). This is illustrated by an increased likelihood of using DP with increasing scientific and epistemological uncertainty. However, other present contexts (Table 2) may mitigate or provoke other uncertainties experienced by the GP and may influence their decision to use DP to safety-net. For example, when a GP identifies constraints on a patient’s ability to access antibiotics, this may provoke practical or technological uncertainty in addition to their scientific/epistemological uncertainty. This may be mitigated by the GP issuing a DP in an accessible way (CMOC:1.2) or, if their work environment allows, instead offering reconsultation to mitigate scientific uncertainty about illness progression and medico-legal concerns (CMOC:1.7). Another example is when a GP uses clinical scores or diagnostic tests: a clearly high or low score/test result may mitigate scientific uncertainty and lead to an immediate or no antibiotic decision (CMOC 1.5); an intermediary result may provoke hermeneutic uncertainty and lead to DP becoming preferred (CMOC:1.1). Moreover, all these contexts and uncertainties may be influenced by a GP’s existential uncertainty regarding their worldview of what makes a ‘good’ doctor: e.g., beliefs around the appropriateness of transferring antibiotic decision-making to patients (CMOC:1.6); or around the risks and benefits of giving patients potentially unnecessary medication (CMOC:1.9). For patients perceived at high-risk of serious complications, fear of missing a serious diagnosis is the main reason for using immediate antibiotics. DP use was influenced by how GPs’ scientific/epistemological uncertainty was addressed when they interpret evidence/guidelines as supporting DP for vulnerable patients as a safety-net (CMOC:2.1) or cautioning against (CMOC:2.2). Determining high-risk patients may differ among GPs, e.g. patient’s age was associated with different perceptions of risk by different GPs. This scientific/epistemological uncertainty may reflect their confidence or experience with managing perceived high-risk patients. Table 2 also shows further contexts and mechanisms that influence GPs’ use of DP to safety-net.
II. Perceived patient concordance and GP self-efficacy
We found that GPs’ beliefs about a patient’s ability to understand and follow DP instructions (CMOC:3), receiving feedback about patients’ use of DP (CMOC:4), and GPs’ self-efficacy to explain DP (CMOC:5), all influence their confidence that they and their patients will reach a shared understanding of DP and agreement for its use, i.e. patient concordance. When perceived patient concordance is low, e.g. when a GP believes a patient to have a lower level of education (CMOC:3.3) or does not know the patient sufficiently (e.g. when working out-of-hours) to assess whether they will use DP as intended (linked CMOC:1.8), then GPs may use DP less. Different types of uncertainty, arising from different contexts, may contribute to a GP’s perception of potential patient concordance and DP may mitigate (CMOCs:4.1-4.2, 5.1-5.3), provoke or exacerbate (CMOCs:3.4, 4.3, 5.6-5.8) these uncertainties. For example, existential uncertainty around patient concordance may be mitigated with attention to scientific uncertainty and improved data on how individual patients, or patients in general, use DP (CMOC:4.1). If perceived patient concordance is due to practical or technological uncertainty about how the patient will access the DP, then this may be improved by providing the GP with information about different DP formats which may facilitate/hinder access (CMOC:4.2). Practical and technological uncertainty may be further addressed by improving GPs’ self-efficacy in explaining the rationale and instruction for DP (CMOCs:5.1-5.3), and that antibiotics are not needed (CMOC:5.5), or to discover that the patient does not want antibiotics (CMOC:5.4).
III. Using DP as a social tool
In situations of less scientific/epistemological uncertainty and where there is perceived or actual patient demand for antibiotics, other uncertainties may predominate. Rather than using DP as a safety-net, some GPs used DP more as a social tool to address practical and existential uncertainty around their relationship with the patient, including minimising conflict (CMOC:6); educating and empowering patients to make illness management decisions (CMOC:7); and reassuring them their concerns were heard (CMOC:7.4). GPs may want to provide a DP which minimises patients accessing antibiotics another way (CMOC:6.3) and choose a DP format to suit them or their patient (CMOC:6.1) rather than one which patients perceive as obstructive or paternalistic (CMOC:6.4). GPs may believe DP acts as a reassuring ‘psychological safety-net’ for patients (CMOC:7.4), especially useful for those patients wanting to avoid antibiotics (CMOC:7.5). Practical and existential uncertainty is also relevant when GPs perceive patients as receptive to self-management (CMOC:7.1) and that educating them about DP can save GP and patient time by reducing future care-seeking behaviour for similar illnesses (CMOC:7.2). Table 2 shows additional contexts and mechanisms (operating separately or together) that make using DP as a social tool more likely (CMOCs:6.1-6.3,7.1-7.5, 1, 1.4, 3.1-3.2, 8.1-8.4, 9.2) or less likely (CMOCs:6.4-6.6,7.6, 1.6).
IV. DP and the work environment
Two key work environment contexts influenced a GP’s prescribing decision: workload and practice culture. They found practical and technological uncertainty relevant when GPs have a high workload and feel under time pressure and want to take the most efficient approach for them or their patient (CMOC:8). If they perceive DP to take little additional time to explain (CMOC:8.1), or administer (CMOC:8.5), or to save time in the future (CMOC:8.2), they may use DP. Conversely, if GPs perceive DP to be more time-consuming than prescribing immediate antibiotics—either by explaining (CMOC:8.6) or administering DP (CMOCs:8.7-8.8), they may use DP less. Practical uncertainty here is also influenced by when the consultation occurs, e.g., Friday afternoon consultations, public holidays, or before an event important to the patient, may mean GPs use DP to enable patient timely access to antibiotics (CMOCs:8.3, 1.2) or to avoid seeking out-of-hours care (CMOC:8.4). Practical and technological uncertainty also depends on how the practice culture and structure supports routine DP use by peers (CMOC:9). Additional contexts or mechanisms here may be pertinent to scientific/epistemological uncertainty, e.g., incorporating (or not) DP use into local guidelines and the extent to which this is perceived as confusing (CMOCs:9.1, 9.7-9.9), or whether peer monitoring of antibiotic prescribing occurs (CMOCs:9.3, 9.6). This includes contexts/mechanisms that may attend to practical uncertainty by making practice approaches and beliefs more visible such as being clear about the variety of DP formats that GPs can provide to suit them or their patient (CMOC:9.2) and reminder cues (e.g., electronic alerts) to GPs about DP (CMOC:9.4). Practical uncertainty may also be important depending on the remuneration healthcare model in place (CMOCs:9.5, 9.10).
DISCUSSION
Our program theory explains how, when and why GPs use DP in different consultation contexts. It posits that GPs have a reasoned tolerance of risk to a range of uncertainties: epistemological (knowledge-orientated); scientific (data-orientated); hermeneutic (data interpretation-orientated); practical (structures/processes-orientated); technological (skills/software/equipment-orientated), and existential (world-view and identity-orientated) (Box 1). These uncertainties interact (individually or together) with GPs’ work environment, perceived patient concordance and self-efficacy to influence the likelihood of issuing DP as a safety-net or social tool at a given time-point, which may shift as the consultation progresses. Our program theory corroborates and advances knowledge around the complex interacting manifold clinical, social, and moral factors that play a role in a GP’s decision to prescribe antibiotics.[40,41,44,54,116–118]
DP may mitigate, provoke or exacerbate various uncertainties
We propose that use of DP may mitigate, provoke or exacerbate various uncertainties, beyond that typically understood as ‘clinical uncertainty’,[59,119] arising from different contexts in primary care consultations (Table 2). For example, although diagnostic tests are often perceived to help reduce clinical uncertainty[120,121] (more nuancedly, scientific/epistemological uncertainty), other literature (pre-COVID-19) suggests that some GPs consider clinical uncertainty rare[40] and interventions to address it thus less relevant or that such technologies may instead lead to more uncertainty over the meaning of results.[40,65,116] Wider literature identifies this uncertainty as hermeneutic uncertainty.[70] We propose hermeneutic uncertainty occurs when GPs have difficulty interpreting intermediate diagnostic/clinical score results; when results conflict with their clinical judgement; when there is uncertainty around a patient’s level of risk for developing serious complications; and when interpreting patients’ narratives to determine their expectations for antibiotics. We found that use of DP to safety-net may mitigate some uncertainties – hermeneutic, practical, technological, and existential – but, and importantly, it may also provoke them. For example, DP may mitigate practical uncertainty around how patients will access antibiotics if needed but provoke a different type of practical/technological uncertainty around which DP format to use and how.[35]
From an existential uncertainty perspective,[63,64,70,71] the safety or professionalism of transferring antibiotic decision-making responsibility to patients sits within moral philosophy arguments for GPs to consider patients and public outside the immediate consultation and future generations of patients in their antibiotic decision-making.[118] This view of what it means to be a ‘good’ doctor may be important not only for how GPs view their own professional role[40] but also, we show, GPs’ concern about whether patients may view them as ‘incompetent’. It is also relevant within the ‘too much medicine’ paradigm[122,123] where GPs, usually guided by principles of ‘doing good’ and/or ‘doing no harm’,[26,27] may be uncertain around the competing pragmatic benefits of DP and the ideal of no antibiotic, with some believing that DP may be a distraction in achieving this ideal. For example, using DP to help patients pragmatically self-manage illness may sit uneasy with those who believe this is better achieved without using DP[28] and who worry about counterproductive ‘harms’, e.g. iatrogenic effects of unnecessary antibiotics or unintentionally increasing antibiotic use.
Impact of heuristics on DP perceptions and use
We have described how different aspects of the work environment influence GPs’ use of DP and time pressure (amongst others) was a key factor in their decision-making. To help overcome time pressure, some GPs used heuristics to make efficiency-thoroughness trade-offs (ETTO).[124] Heuristics, normal and necessary mental shortcuts to lessen cognitive load, involves seeing aspects of current situations similarly to previously experienced situations and where decision responses are transferred from the previous situation to the present. Example heuristics from our findings include, ‘my explanation has/hasn’t worked before, so it will/won’t work now’ and ‘it seems like the patient is (in)sufficiently “sensible” to understand DP, so DP probably isn’t/is appropriate’.[124] Although this transfer is clearly valuable as it uses previous learning to avoid assessing everything from scratch, it is also potentially problematic with the risk of negative transfer, where learned responses are inappropriate.[124] For example, negative transfer from heuristics around ‘clinical uncertainty’ can lead to action bias, anticipated regret, and risk aversion—all of which, in the context of antibiotic prescribing, tend behaviour towards action rather than inaction and overprescribing immediate antibiotics ‘just-in-case’ to prevent potential harm to patients.[125] From this perspective, DP is considered to be a form of inaction by some GPs and looked upon as a relatively high risk strategy compared with immediate antibiotic. For example, some GPs viewed DP as a type of inaction if the DP format hinders patient access to antibiotics.[41]
Some heuristics around DP are also related to human factors, i.e. “interactions among humans and other elements of a system”[126–128], including how DP is operationalised in different practice settings. Despite clinical trials and studies involving DP scripts, formats, and communication support tools,[21,99] uncertainty remains in clinical practice for how GPs can use DP and how the prescribing system and work environment supports or hinders this decision option. This includes misperceptions such as DP being limited to post-dating prescriptions and that it is not possible to issue DP via electronic transfer to community pharmacies in some health care systems.
To help address the challenges of heuristics, four psycho-social strategies have been proposed by Tarrant and Krockow,[125] derived from a range of health literatures, to mitigate antibiotic overprescribing: 1. Strategic framing of treatment options (e.g. frame immediate antibiotics as ‘low-value’ in some circumstances and DP as an ‘active’ option that provide more optimal benefit vs risk for individual patients); 2. Substitution – replacing an undesired behaviour with an alternative (e.g. provide evidence and implement guidance to support substitution of immediate antibiotic with DP for specific infections and situations); 3. Documentation – enabling and encouraging prescribers to document decision-making process and discussions with the patient to mitigate against accusation of negligence; 4. Social support - this recognises that antibiotic decision-making by an individual prescriber is often influenced by their interactions with colleagues and expert advisors as well as their own knowledge and experience. Interventions that enhance social support through senior or peer review or feedback from patients can therefore help prescribers to fine-tune their tolerance of uncertainty and discourage defensive medicine.[125]
Implications for practice
For policy makers, implementation teams and researchers, Table 2 could be used as a resource for reviewing why DP is working or not for certain patient cohorts, situations or settings to identify what potential interacting mechanisms are at play and what type(s) of uncertainty may be targeted for intervention.
In practice and as evidenced in this review, DP is often seen as a compromise to no antibiotics (and not just as an alternative to immediate antibiotics). This then has the potential unintended consequence of driving up antibiotic use. To mitigate this, we suggest efforts to support DP be explicitly presented as an alternative to immediate antibiotics i.e. when a GP is uncertain about the benefits of immediate antibiotic and DP provides an appropriate safety-netting tool (rather than primarily as a social tool). DP should only be used for certain infections where there is evidence of benefits for specific patient groups/situations and risks of harm can be minimised e.g. through additional safety-netting advice. Ultimately, the goal is to improve confidence, by both prescribers and patients, in how certain infections can be effectively and safely managed without antibiotics, supported by self-care advice. Our recommendations are therefore to consider how DP aligns with no antibiotic options and self-care advice from a broader infection management perspective, and we have framed these using the psycho-social strategies by Tarrant and Krockov (Table 3).
Strengths and limitations
A key strength of this review is incorporation of more nuanced types of uncertainty, beyond clinical uncertainty, on how and why GPs use DP. By using a systematic theory-driven realist synthesis of the evidence, we have been able to identify mechanisms likely to be applicable across different types of interventions to facilitate appropriate DP use. Rather than be prescriptive about when DP should or should not be used over other interventions, our program theory recognises the complexity of shared decision-making and sheds light on what range of contextual factors tend to enable safe and effective DP use. It also highlights how each of these factors may potentially be addressed and that a multi-faceted approach to reducing different types of uncertainty in different settings is required.
Other strengths are using different types and sources of evidence, including empirical data and demonstrating face validity through collaboration with a multi-professional team of researchers, clinicians and subject-matter experts. Unlike conventional systematic literature reviews, a realist review generates a program theory which, if sufficient evidence was analysed appropriately, should withstand the test of time and subsequent evidence would more likely reinforce than dispute the core constructs. However, we mostly identified and used evidence from high-income countries and focused our review on the perspective of GPs and not on patients, or other prescribers, which may limit some transferability. All the evidence included in the review was collected by February 2021 and none referred to the COVID-19 pandemic; however, we discuss below the potential implications of our program theory in the context of a pandemic and increased use of remote consultations.
Future Research
The COVID-19 pandemic has led to significant changes in health service delivery.[129,130] The need to reduce disease spread has meant minimising face-to-face appointments and a dramatic increase in remote-consulting along with reduced ability to examine patients.[129–132] It has also meant various changes in health-seeking behaviour, including reported delays in health-seeking advice by patients.[130] Changing COVID-19 testing requirements, healthcare access and provision challenges due to mandated total triage models have further complicated the consultation process.[129,130] Consequently, actual or perceived demand (and need) for antibiotics from those who do seek healthcare advice may have been higher, e.g. from individuals suffering symptoms longer. Post-pandemic antibiotic prescribing data are now emerging; these should be examined for longer lasting effects of the pandemic on prescribing habits or changing public/patient beliefs about antibiotics.
Here we consider how our program theory can be applied in the context of the COVID-19 pandemic which may have substantially increased various uncertainties. For example, GPs, who would normally see most patients face-to-face, may have felt uncomfortable assessing patients remotely, experiencing increased scientific uncertainty by not being able to physically examine patients or use diagnostic tests in the GP surgery. This may have been confounded early in the pandemic with practical and technological uncertainty when many were less experienced in remote consulting and it was unclear whether issuing a DP was practically or technologically possible. One recent interview study with GPs concurs, revealing mixed views on perceived DP usefulness and actual use since the pandemic.[130] Since the COVID-19 pandemic, widespread self-testing has become standard for many patients and may facilitate uptake and acceptance of other point-of-care testing by patients in the future.
Further research is needed to understand and improve how different uncertainties are self-managed by patients. Where a GP or other healthcare professional (e.g. nurse practitioner, community pharmacy staff) is involved, research is needed to better understand how uncertainties are communicated to patients, and how some of these may be mitigated or provoked by DP. This includes improving how patients, GPs and other healthcare professionals may develop better DP explanations on when and how to use it. Further research may also be carried to support implementation of our recommendations in Table 3.
CONCLUSION
This review provides practical insights for considering how various uncertainties influence antibiotic decision-making and DP-related behaviour. We show how different uncertainties (beyond clinical uncertainty)—epistemological, scientific, hermeneutic, practical, technological, existential— interact with GPs’ work environment, perceived patient concordance and GPs’ self-efficacy to influence use of DP to safety-net or as a social tool in primary care. Efforts by policy makers, commissioners, and prescribers to optimise antibiotics can use our program theory to help identify, design and implement interventions to address these uncertainties. We also recommend activities to: 1) reduce demand for immediate antibiotics by reframing infection management strategy options and emphasising clinical situations where there is relatively little or no benefit from immediate antibiotics; 2) reinforce DP as an ‘active’ prescribing alternative to immediate antibiotic; 3) document the decision-making process around DP; and 4) social and system support to facilitate a practice culture for using and improving infection management (including use of DP).
SUPPLEMENTAL FILES
Supplemental file 1 – Primary literature search strategy
Supplemental file 2 – Document characteristics, source, use and nature of evidence Supplemental file 3 – Categorising contexts, mechanisms and outcomes Supplemental file 4 – Literature selection, inclusion and exclusion
Supplemental file 5 – Program theory on delayed prescription decision-making Supplemental file 6 – Summary program theory CMOCs and uncertainty types
AUTHOR CONTRIBUTION STATEMENT
All authors contributed to the design of the study and its development. MMc and BH developed the search strategy with support from a specialist librarian. Articles were screened, data were extracted, and early themes developed by MMc, BH, AC and AB. The initial program theory was developed by MMc and BH, presented and discussed with wider STEP-UP team, then iteratively refined with all authors and substantial input from AC and GW. MMc and AC conducted the principal analysis and GW, BH, AB, STC contributed to the interpretation and presentation of results to the wider STEP-UP team. Iterative refinement and preparation of draft manuscripts were prepared by MMc and AC with contributions from GW. All authors provided critical feedback and helped shape the research, analysis and manuscript. All authors have read and agreed to the final version of the manuscript.
Data Availability
All data produced in the present study are available upon reasonable request to the authors
FUNDING
The study was funded by the Economic and Social Research Council (ESRC) through the Antimicrobial Resistance Cross Council Initiative supported by the seven research councils in partnership with other funders (grant reference: ES/P008232/1) and supported by the National Institute for Health Research (NIHR) Health Protection Research Unit (HPRU) in Healthcare Associated Infections and Antimicrobial Resistance at the University of Oxford (NIHR200915) and Imperial College London (NIHR200876) in partnership with the UK Health Security Agency (UKHSA), the NIHR Oxford Biomedical Research Centre and the NIHR Imperial Patient Safety Translational Research Centre (PSTRC-2016-004). The support of the funders is gratefully acknowledged. The funders played no role in the design of the study, the collection, analysis, and interpretation of data, or in writing the manuscript. The views expressed are those of the authors and not necessarily those of the NHS, the NIHR, the UKHSA or the UK Department of Health and Social Care.
CONFLICTS OF INTEREST
BH works for eConsult, a provider of asynchronous consultations in primary, secondary, and urgent/emergency care. Other authors declare no conflict of interest.
ACKNOWLEDGMENTS
This review is part of the work of the STEP-UP team (https://www. expmedndm.ox.ac.uk/step-up/step-up) comprising: Philip E Anyanwu, Aleksandra Borek, Nicole Bright, James Buchanan, Christopher Butler, Anne Campbell, Ceire Costelloe, Benedict Hayhoe, Alison Holmes, Susan Hopkins, Azeem Majeed, Monsey McLeod, Michael V Moore, Liz Morrell, Koen B Pouwels, Julie V Robotham, Laurence S J Roope, Sarah Tonkin-Crine, A. Sarah Walker, Sarah Wordsworth, Carla Wright, Sara Yadav, and Anna Zalevski. The team worked with the British Society for Antimicrobial Chemotherapy (BSAC) to deliver an elearning course that includes preliminary findings from the work in this manuscript. A payment was made to BSAC to help develop the course format and support platform administration. We would also like to acknowledge and thank Natasha Lal for additional editing support.
Footnotes
REFERENCES
