Electronically delivered interventions to reduce antibiotic prescribing for respiratory infections in primary care: cluster RCT using electronic health records and cohort study

Article history

The research reported in this issue of the journal was funded by the HTA programme as project number 13/88/10. The contractual start date was in January 2015. The draft report began editorial review in January 2018 and was accepted for publication in June 2018. The authors have been wholly responsible for all data collection, analysis and interpretation, and for writing up their work. The HTA editors and publisher have tried to ensure the accuracy of the authors’ report and would like to thank the reviewers for their constructive comments on the draft document. However, they do not accept liability for damages or losses arising from material published in this report.

Declared competing interests of authors

Martin C Gulliford is a member of the Health Services and Delivery Research Board. A Toby Prevost is an employee of Imperial College London and is a member of the Public Health Research (PHR) Funding Board. Alastair D Hay is a member of the Health Technology Assessment (HTA) Clinical Trials Board. Paul Little is a member of the HTA Pandemic Influenza Board, the National Institute for Health Research Journals Library Board and the Programme Grants for Applied Research Expression of Interests – HTA Projects remit. Lucy Yardley is a member of the HTA Antimicrobial Resistance Themed Call Board, the HTA Efficient Study Designs Board and the PHR Funding Board. Michael Moore is a member of the government’s Advisory Committee on Antimicrobial Prescribing, Resistance and Healthcare Associated Infection.

Copyright statement

© Queen’s Printer and Controller of HMSO 2019. This work was produced by Gulliford et al. under the terms of a commissioning contract issued by the Secretary of State for Health and Social Care. This issue may be freely reproduced for the purposes of private research and study and extracts (or indeed, the full report) may be included in professional journals provided that suitable acknowledgement is made and the reproduction is not associated with any form of advertising. Applications for commercial reproduction should be addressed to: NIHR Journals Library, National Institute for Health Research, Evaluation, Trials and Studies Coordinating Centre, Alpha House, University of Southampton Science Park, Southampton SO16 7NS, UK.

2019 Queen’s Printer and Controller of HMSO

Background and rationale

This research was proposed in response to the National Institute for Health Research (NIHR)’s call for proposals on antimicrobial drug resistance (AMR). There is growing concern that the widespread, and sometimes unnecessary, use of antibiotics is leading to the development of AMR. 1,2 In economic terms, the antimicrobial effect of drugs may be viewed as a natural resource. Antimicrobial resistance represents a negative externality of antibiotic use. If resistant organisms persist over time, then the resource of antibiotic drugs will be exhaustible. 3,4 Reducing inappropriate use of antibiotics, as well as ensuring that antibiotics can be used when they are needed, represents an important component of a strategy to control infectious diseases. 5

Governments of all countries need to adopt a stewardship role so as to ensure that effective antimicrobial drugs are available to future generations. 1 This should include responding to the requirement to improve governance and standards of clinical practice with respect to antimicrobial drug utilisation. The present research addresses a subject of great public health importance because overutilisation of antibiotics contributes to the emergence of AMR and, consequently, infections that may be very difficult to treat. The review on antimicrobial resistance,2 chaired by Jim O’Neill, identified education and training to reduce inappropriate and unnecessary antibiotic use as key measures to fight AMR. This research specifically aims to address the problem of inappropriate and unnecessary prescribing of antibiotics to patients with self-limiting respiratory tract infections (RTIs) in primary care.

The eCRT

The systematic review and recent trials are important in identifying strategies that may be effective at changing prescribers’ behaviour. However, previous trials required resource-intensive interventions and these intervention techniques have not yet been translated on a wide and sustainable scale into the NHS. For example, the trial by Gonzales et al. 23 required clinicians to participate in a half-day training session, triage nurses to provide patients with education leaflets to read before their consultation, a specially designed structured template to be programmed into the practice system to provide an algorithm-based probability of the patient having pneumonia, and ‘order sets’ to be created for group diagnosis and treatment options for different types of RTIs. The challenge now is to take the components of intervention that have been shown to be effective and to find methods to deploy these efficiently into routine practice settings.

The methods of deploying the components of the intervention into routine practice settings have been investigated in the recently completed eCRT, which involved cluster randomisation of general practices that contribute EHRs to a national primary care database, the CPRD. 16 The study included 104 general practices in England and Scotland. Decision support tools (DSTs) were delivered remotely to general practices. The effectiveness of the intervention was evaluated by analysing EHRs that are routinely collected into the database. Data were analysed for > 600,000 individual participants, with a financial cost of about 27 pence per participant. Even with a very simple intervention, the trial showed a near 2% reduction in antibiotic prescribing. This eCRT showed that it was feasible to use the CPRD to evaluate interventions that may be readily scaled up to the population level. Feedback received in the eCRT process evaluation,30 together with evidence from other trials cited above, identifies ways to increase engagement in the intervention and increase effect sizes.

Safety outcomes

Material in this section is reproduced from Gulliford et al. 31 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 3.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: http://creativecommons.org/licenses/by/3.0/.

The present research also addresses the safety of reducing antibiotic prescribing in primary care. There may be well-grounded clinical concern that reducing antibiotic use might increase the risk of complications following RTIs. Clinical trial evidence suggests that antibiotics may reduce the risk of suppurative complications of RTIs,13 but the more serious complications are generally too rare to evaluate precisely in randomised studies. Petersen et al. 32 conducted a cohort study in 162 general practices in the General Practice Research Database (the predecessor of the CPRD) from 1991 to 2001 to evaluate the effect of antibiotic treatment on the incidence of pneumonia after ‘chest infection’, peritonsillar abscess after sore throat, and mastoiditis after otitis media. The results suggested that antibiotic treatment was associated with lower odds of each these complications, but the overall risk of complications was generally small and the number of patients who would have to be treated to avoid one complication was estimated to be in excess of 4000. However, pneumonia following chest infection was more frequent. In people aged < 65 years, it was estimated that there might be one case for every 100 antibiotic prescriptions for chest infection avoided, and one case for every 39 prescriptions avoided in people aged > 65 years. Infections of the middle ear or sinuses may rarely be complicated by intracranial abscess. 33 Lemierre syndrome,34 from thrombophlebitis of the internal jugular vein associated with Fusobacterium necrophorum infection, is a rare complication of sore throat,34 but F. necrophorum infections may be frequently detectable in patients with symptoms of sore throat. 35,36 The annual number of cases of Lemierre syndrome in England was reported to have increased from 19 in 1997 to 34 in 1999, prompting a reminder from the Chief Medical Officer that some sore throat symptoms may require antibiotic treatment. 37 In addition to concerns about complications, medical practitioners may be concerned about the potential consequences of diagnostic misclassification. The initial symptoms of meningitis may sometimes resemble an influenza-like illness. 38 Awareness of the possibility of a more serious diagnosis might prompt GPs to issue an antibiotic prescription for conditions in which this is not usually indicated.

These observations raise important questions for a policy to reduce antibiotic prescribing for RTIs in primary care: is there a safe level of antibiotic prescribing for RTIs? What target can general practices safely adopt in reducing the proportion of RTI consultations with antibiotics prescribed? Is there a threshold for antibiotic prescribing below which complications may increase? An additional aim was added to the present research to evaluate the safety of a policy to reduce antibiotic prescribing for RTIs in primary care. The incidence of pneumonia, peritonsillar abscess, mastoiditis, empyema, meningitis, intracranial abscess and Lemierre syndrome was evaluated. The study aimed to determine whether or not these complications were more frequent at general practices that prescribe fewer antibiotics for self-limiting RTIs than at higher-prescribing practices. The use of this information was aimed to quantify the potential clinical and public health impact of changes in antibiotic-prescribing practices.

Evidence explaining why this research is needed now

The recent systematic review, together with the additional, more recent, trials, show that interventions to modify prescribing behaviour in primary care can be effective. However, there is a block in the translational pathway because it has not been possible to roll-out this evidence into routine practice; antibiotic prescribing for RTIs remains high outside trial settings. There is a lack of effective interventions that can easily be translated, in a sustained way, into routine practice settings. This research aimed to use the strengths of EHRs to inform, deliver and evaluate an intervention. This will be achieved with a high degree of efficiency by employing at the research environment a database of EHRs accessed through the CPRD.

This research is at a later stage of translation than previous trials. In order to overcome the block in the translational pathway, the study aimed to develop and evaluate more effective complex multicomponent interventions that can be implemented and delivered remotely. Development of the interventions was informed by evidence from recent trials, as well the process evaluation of the eCRT study. 30 The research focused on deployment of interventions with potential to be readily scaled up, through remote delivery using electronic media, to large samples of unselected practices. The present research built on previous experience of implementing the eCRT within the CPRD. In the eCRT, the intervention was an educational and DST39 that aimed to support evidence-based antibiotic prescribing for respiratory illness in primary care. The intervention was installed remotely at practices and utilisation of the intervention was monitored.

The approach of utilising the EHRs of the CPRD to provide the environment for delivering and testing the interventions has several advantages:

1. both the interventions and a cluster randomised trial of the interventions can be implemented at a very low cost

2. the sample available for the study is nationally representative for the UK and large sample sizes are expected

3. the sustainability of the effect of the intervention may be evaluated after the end of the trial, because data continue to be collected from trial practices

4. utilisation of the intervention can be routinely monitored through electronic information routinely collected into EHRs

5. a cost-effectiveness analysis may be implemented using data on health-care utilisation, which are collected for all patients in CPRD40

6. translation of the trial results is readily feasible because the interventions are delivered using the practice systems that are employed in delivering routine care within the NHS.

Aim and objectives

Intervention development and delivery

Cluster randomised trial

The trial was conducted in general practices contributing to the UK’s CPRD. The CPRD is one of the world’s largest databases of primary care EHRs; it includes data from general practices throughout the UK. CPRD data have been extensively evaluated and employed for epidemiological research. 49 Recently, the CPRD has begun to be used as a resource for interventional research. 16,50–52

Population-based cohort study of safety outcomes

Material in this section is reproduced from Gulliford et al. 31 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 3.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: http://creativecommons.org/licenses/by/3.0/.

The data source for the cohort study was the UK’s CPRD. 49 For the present study, data were included for the 10-year period, from 2005 to 2014, before the start of the cluster trial. The CPRD included data for an open cohort of about 4.5 million registered patients during this period.

Participants

Data collection for the intervention study was completed between March 2015 and September 2015. There were 28 GPs and three NPs who took part in the interviews and 35 interviews were conducted in total (three GPs and one nurse were interviewed on two occasions). There were 31 interviews conducted face to face in general practices and four were conducted over the telephone. Participants were recruited from 21 non-trial general practices covering several areas of England (London, Oxfordshire and Yorkshire) and from general practices of different sizes – ranging in the number of registered patients from 5891 to 17,797 (on average, 10,968 registered patients). Interviews lasted between 20 and 76 minutes.

Intervention development: emerging contextual themes

The interviews for intervention development provided contextual information, which contributed to the understanding of influences on practitioners’ interaction with the intervention materials. Table 4 summarises the main themes relevant to the implementation of the trial intervention.

Intervention development: feedback from interviews

Table 5 provides a summary of themes relevant to intervention design.

Trial interventions

A summary of the main features of the final versions of trial interventions is given in Table 6. There were no modifications to the interventions following the start of the trial.

Webinar

Antibiotic-prescribing reports

Antibiotic-prescribing reports were presented on professionally designed templates, including the logos of the NIHR and King’s College London. The main features of the antibiotic-prescribing reports were as follows:

• The intervention trial arm practices were sent antibiotic-prescribing reports at the start of the trial, and after each completed month of the 12-month intervention period.

• At the end of the trial, practices received a final report accompanied by a certificate of completion of the trial.

• Prescribing reports were portable document format (PDF) files each presented as a single side of A4 paper.

• The reports presented data in a table, including a count of the number of RTI consultations and number of antibiotic prescriptions, together with the proportion of RTI consultations with antibiotics prescribed for each month of the intervention period.

• Data were compared with the preceding 12 months at the same practice for reference.

• A brief accompanying narrative explained the numerical data included in the table and summarised any change in antibiotic utilisation based on the proportion of RTI consultations with antibiotics prescribed.

• The narrative also included a reminder on how to access the REDUCE trial DSTs.

• A bar chart illustrated changes in the proportion of RTI consultations with antibiotics prescribed in comparison with the preceding 12 months at the same practice.

• Based on interview feedback, it was decided not to include an external comparator (such as antibiotic-prescribing data for the whole of the CPRD) because this might offer little incentive for those practices already prescribing fewer antibiotics than average.

• Data reported were aggregated across all ages and for men and women.

• No changes were made to the overall design of the prescribing reports during the course of the trial.

• Data analysis and report writing were performed using bespoke code written in the R program.

Decision support tools

The REDUCE trial DSTs (Figure 1) were delivered into trial practices using proprietary software known as DXS Point-of-Care, which is integrated with the Vision practice system. The DSTs were triggered when a Read code for a RTI was entered into the practice system. A ‘banner’ then appeared in the DXS Point-of-Care tool bar. Practitioners gained access to the DSTs by clicking on the toolbar. The process was explained in the introductory webinar.

FIGURE 1.

Map of DSTs.

Figure 1 provides a map of the DSTs. After clicking on the DXS Point-of-Care toolbar, practitioners were presented with a ‘landing page’ that offered advice for the five groups of self-limiting RTI, including common cold, sore throat, sinusitis, otitis media and cough/bronchitis. Advice for the last two conditions was specifically tailored for either children or adults, based on clinical advice from GP members of the study team. For each group of conditions, practitioners were offered the alternative of accessing a patient information leaflet (Figure 2) or a prescription indication page (Figure 3).

FIGURE 2.

Part of one of the REDUCE trial patient information leaflets.

FIGURE 3.

Screenshot showing the REDUCE trial prescription indication page.

The REDUCE trial patient information leaflets included the following items of information:

• the expected duration of symptoms from the specified class of self-limiting RTI

• evidence concerning antibiotic utilisation including:

  • lack of clinical effectiveness of antibiotics at reducing the duration or severity of symptoms

  • increased risk of common drug side effects

  • risks to individuals and families from AMR

• advice on self-management of symptoms including:

  • need for rest

  • fluid intake

  • management of fever with paracetamol or ibuprofen

  • obtaining advice on symptomatic treatments from a pharmacist

• advice on seeking help when there are possible warning signs including:

  • possible signs of meningitis (headache or vomiting)

  • respiratory difficulties (rapid breathing, cyanosis)

  • possible signs of septicaemia (cold skin, discoloration, rash)

  • local complications (difficulty swallowing, drooling)

  • ‘feeling a lot worse’.

The patient information leaflets were professionally designed and included the NIHR and King’s College London logos. The leaflets were presented as a single side of A4 in a PDF, which could be viewed on a computer screen or printed and handed to patients.

The REDUCE trial prescription indication page drew on the criteria for antibiotic prescribing in the NICE guidelines (see paragraph 1.7);6 these guidelines identify patients at risk of complications using criteria that include how unwell the patient is, whether or not there are clinical features suggestive of complications if comorbidity is present, or if the patient is of an older age and having features that suggest a higher level of risk. The prescription-indication page enabled practitioners to evaluate with patients whether or not an indication for an immediate antibiotic prescription was present.

Recruitment and baseline characteristics

The trial recruited 80 (25%) general practices to the trial, of which one withdrew from the CPRD before the start of the intervention and the remaining 79 were included in the intention-to-treat analysis. Figure 4 provides a flow chart showing the progress of practices and participants through the trial.

FIGURE 4.

A flow chart showing trial practices and registered populations.

The trial included general practices from throughout the UK, including 19 practices from Scotland, 15 from Wales, nine from Northern Ireland and the remaining 36 from regions in England (Table 7). There were 18 practices allocated in November 2015, 31 in January and February 2016 and 30 in June–August 2016. The registered population included participants of all ages, including approximately 17% aged under 15 years and approximately 8% aged ≥ 75 years. General practice list sizes were generally slightly higher in intervention trial arm practices, but the range of list sizes was similar across trial arms. There were 17.1% of participants with comorbidity in intervention trial arm practices and 14.5% at control trial arm practices. In the baseline period, the point estimates for RTI consultation rate and antibiotic-prescribing rate were slightly higher at control trial arm practices, but in the context of the wide range of variation among practices (Table 7).

Main results

Initial analyses were conducted using cluster-level statistics. Figure 5 presents a plot of the age-standardised estimates for each general practice for the intervention phase of the trial. The baseline antibiotic-prescribing rate is plotted on the x-axis and the change in antibiotic-prescribing rate (trial period – baseline) on the y-axis. Linear models, weighted for practice size, are plotted by trial arm, together with 95% CIs. There was a wide range of antibiotic-prescribing rates. Both intervention and control trial arm practices showed lower mean antibiotic-prescribing rates during the trial than with baseline, with no clear difference between trial arms. However, changes in antibiotic-prescribing rates showed wide variation at any level of baseline antibiotic prescribing. The data appear to be overdispersed, with several extreme results possibly having undue influence in this representation of the data. In an analysis of covariance, weighted for practice size, the adjusted mean difference between the intervention and control trial arms for the antibiotic-prescribing rate for RTI was –0.5 per 1000 patient-years (95% CI –8.2 to 7.2 per 1000 patient-years).

FIGURE 5.

Scatterplot showing change (trial period – baseline) in age-standardised antibiotic-prescribing rate for each trial practice plotted against the baseline antibiotic-prescribing rate.

Figure 6 presents a density plot for the distribution of changes in antibiotic-prescribing rate from baseline for the intervention and control trial arms separately. Although the trial analysis was not based on changes from baseline, in a cautious interpretation of Figure 6 there appears to be substantial overlap in the distributions for the two trial arms, although the curve for intervention trial arm practices in green is generally shifted to the left in comparison with control trial arm practices in black. Both Figures 5 and 6 illustrate wide variation among trial general practices, which might make it difficult to draw clear conclusions from the trial.

FIGURE 6.

Density plot showing distribution of change from baseline in antibiotic prescribing at ages for intervention (green) and control (black) trial arms.

Table 8 presents the results of the primary analysis. There were 31,907 antibiotic prescriptions for RTIs during 323,155 patient-years at 41 intervention trial arm practices and 27,923 antibiotic prescriptions during 259,520 person-years at 38 control trial arm practices. There were 98.7 antibiotic prescriptions per 1000 patient-years in the intervention trial arm and 107.6 per 1000 in the control arm, giving an unadjusted RR of 0.92. The RR for antibiotic prescribing for RTIs was 0.88 (95% CI 0.78 to 0.99; p = 0.040). Point estimates for the RR for RTI consultations (adjusted RR 0.94, 0.86 to 1.03; p = 0.186) and the proportion of antibiotic consultations with antibiotics prescribed (adjusted RR 0.96, 0.89 to 1.03; p = 0.253) were both lower than unity. However, the CIs were wide and the p-values were large, which was consistent with an insufficiently precise estimate being obtained given the size of the trial and the substantial level of variation in these measures.

Intervention effect by the subgroup of age

Respiratory tract infection consultations and antibiotic prescribing were strongly associated with age. In the control trial arm, children aged 0–14 years accounted for 18% of all person-time, but 31% of all RTI consultations and 23% of antibiotic prescriptions for RTIs. People aged 45–54 years accounted for 15% of all person-time, but 11% of all RTI consultations and 12% of all antibiotic prescriptions. People aged 65–74 years accounted for 9% of all person-time, but 12% of antibiotic prescriptions. These figures are presented as per cent of the total person-time in each trial arm and an effect of intervention is not intended to be demonstrated.

There was strong evidence that the effect of the intervention varied by age group. A Wald test of the trial arm by age–group interaction terms gave a p-value of < 0.001. The results of a prespecified subgroup analysis, by age, are shown in Figure 7. There was no evidence of an effect of intervention in children aged < 15 years (adjusted RR 0.96, 0.82 to 1.12) or in people aged ≥ 85 years (adjusted RR 0.97, 95% CI 0.79 to 1.18). At intermediate ages, antibiotic prescribing was lower in the intervention trial arm. Adjusted RRs were 0.84 (95% CI 0.73 to 0.96) for people aged 45–54 years; 0.79 (95% CI 0.69 to 0.91) for 55- to 64-year-olds; 0.80 (95% CI 0.70 to 0.91) for 65- to 74-year-olds; and 0.79 (95% CI 0.69 to 0.91) for 75- to 84-year-olds. Antibiotic prescribing was generally similar across categories of gender and comorbidity.

FIGURE 7.

Effect of intervention for primary outcome, by age group. (a) All ages; and (b) 15- to 84-year-olds. RRs were adjusted for the random effect of general practice and fixed effects of gender, comorbidity, region, quarter in study, practice-specific baseline rate and interaction with period of randomisation. LL, lower limit of CI; UL, upper limit of CI.

Effect modification was summarised by age by comparing effect measures in children, adults aged 15–84 years and senior elderly aged ≥ 85 years (Table 9). The intervention was associated with lower antibiotic prescribing for RTIs in adults aged 15–84 years (adjusted RR 0.84, 95% CI 0.75 to 0.95). During the intervention period the mean rate of antibiotic prescribing for RTIs in adults aged 15–84 years at control trial arm practices was 100.2 per 1000 patient-years. The relative risk reduction was used to estimate an absolute risk reduction of 16.0 (95% CI 5.0 to 25.1) antibiotic prescriptions per 1000 patient-years. In adults, one antibiotic prescription was avoided for every 62 (95% CI 40 to 200) registered patients aged 15–84 years per year. Figure 8 presents a density plot of changes in antibiotic prescribing for adults aged 15–84 years. This reveals a higher proportion of general practices showing reductions in antibiotic prescribing in the intervention trial arm, providing visual confirmation of the estimates provided from statistical modelling.

FIGURE 8.

Density plot showing distribution of change from baseline in antibiotic prescribing at ages 15–84 years for the intervention (green) and control (black) trial arms.

For adults aged 15–84 years, the relative risk estimate was lower than unity for the proportion of RTI consultations with antibiotics prescribed (adjusted RR 0.95, 95% CI 0.89 to 1.03; p = 0.209), but the estimate was imprecise and the p-value was large, which is consistent with a lack of evidence of an effect on this measure (Table 9). There was evidence that, for adults aged 15–84 years, the rate of RTI consultations was lower in the intervention trial arm (adjusted RR 0.90, 95% CI 0.82 to 0.98). These findings suggest that the effect observed for the primary outcome might be more strongly associated with differences in respiratory consultation rates rather than changes in the proportion of RTI consultations with antibiotics prescribed. Figure 9 shows a series of density plots comparing changes in age-standardised general practice-specific values for RTI consultation rates, antibiotic-prescribing rates and proportion of consultations with antibiotics prescribed, for all ages and for the age range 15–84 years. This series shows that RTI consultation rates were generally lower during the trial period than in the baseline period, with similar reductions being observed in both trial arms. Antibiotic-prescribing rates, and the proportion of RTI consultations with antibiotics prescribed, showed substantial overlap, but any reductions tended to be slightly greater in the intervention trial arm. There were several outlying values. This provided visual confirmation of the statistical modelling results.

FIGURE 9.

Density plots showing distribution of change from baseline for RTI consultation rates, antibiotic-prescribing rates and proportion of RTI consultations with antibiotics prescribed at all ages and for ages 15–84 years for the intervention (green) and control (black) trial arms.

Additional subgroup analyses

The intervention effects for the primary outcome (Figure 10) were generally similar for men and women and for comorbidity categories. Estimated effects varied widely among regions, but caution must be observed when drawing firm conclusions because of the small numbers of general practices included in each region. When the primary outcome was analysed by subgroups of baseline antibiotic-prescribing rate quartile, there was no evidence for any consistent pattern of effect. Adjusted RRs were 0.94 (95% CI 0.66 to 1.34) for the lowest quartile, 0.79 (95% CI 0.57 to 1.09) for the second quartile, 0.99 (95% CI 0.81 to 1.20) for the third quartile and 0.99 (95% CI 0.89 to 1.10) for the highest quartile.

FIGURE 10.

Forest plot showing RRs for antibiotic prescribing, by the subgroups of gender, age, comorbidity and region. Estimates were adjusted for the random effect of general practice and the fixed effects of quarter in study, practice-specific baseline rate and interaction with period, as well as each of the variables shown. LL, lower limit of CI; UL, upper limit of CI.

Estimated effects for the proportion of RTI consultations with antibiotics prescribed (Figure 11) showed consistent results for men and women and for comorbidity categories. Estimates by age group were generally consistent with the overall effect, but there was no evidence of a clear pattern of variation by age, though the estimate for people aged ≥ 85 years was consistent with absence of effect. It is relevant to note that the analyses for the antibiotic-prescribing proportion uses a different denominator, that is, the number of RTI consultations, as compared with the analyses for antibiotic-prescribing rate and RTI consultation rate, which both incorporate person-time as denominator.

FIGURE 11.

Forest plot showing RRs for the antibiotic-prescribing proportion, by the subgroups of gender, age, comorbidity and region. Estimates were adjusted for the random effect of general practice and the fixed effects of quarter in study, practice-specific baseline rate and interaction with period, as well as each of the variables shown. LL, lower limit of CI; UL, upper limit of CI.

Sensitivity analyses

Sensitivity analyses were conducted to evaluate whether or not the use of potentially better-fitting models might influence conclusions (Table 10). The rate of respiratory consultations was adjusted for in the trial period. In addition, interactions of covariates with age group was allowed for. Estimates and p-values from these models were generally consistent with the findings from the base-case model. DHGLMs were fitted to address possible overdispersion of the general practice-derived data. In DHGLMs, random effects may be fitted in the dispersion part of the model, as well as in the mean model for the fixed effects. In the DHGLM framework, an overdispersed Poisson model yielded a similar estimate to the base case. In addition, results were compared using EQL and EQL1 estimation, as the former is known to give biased estimates in some data configurations, but the same estimates were obtained with either option.

Total antibiotic prescribing

Table 11 presents data for antibiotic prescribing for all indications during the trial. There were 185,924 antibiotic prescriptions in intervention trial arm practices and 150,539 in control trial arm practices. Crude rates were 575.3 and 581.0 per 1000 patient-years, respectively. The adjusted RR was 0.93 (95% CI 0.83 to 1.04; p = 0.184). There was no evidence that total antibiotic prescribing was reduced by this intervention.

Deferred prescriptions

Table 13 shows the number of deferred prescriptions by trial arm and period. During the intervention period of the trial, 20 control practices and 22 intervention trial arm practices recorded no deferred antibiotic prescriptions. There were three control practices and six intervention trial arm practices that recorded > 50 deferred antibiotic prescriptions. Across all practices, the rate of deferred antibiotic prescriptions was similar in the intervention and control trial arms and did not increase after the introduction of the intervention.

Safety outcomes

The safety outcomes for the trial were those agreed by the Data Monitoring Committee and the study team. Table 14 presents the number of safety outcomes and rates per 100,000 patient-years by trial arm. Figure 12 presents a forest plot showing relative incidence rates for each safety outcome. There was no evidence to suggest that any of these outcomes were more frequent in the intervention trial arm. Estimates were imprecise for rare outcome events. Figure 13 shows the distribution of counts of safety events by age group and trial arm. Scarlet fever was the most frequently recorded outcome in children; pyelonephritis and peritonsillar abscess were most frequent at intermediate ages, whereas pneumonia was the most frequent outcome at older ages. It may be noted that the denominator population is slightly larger for the intervention trial arm. There was weak evidence that scarlet fever might have been more frequent in the control trial arm, but this might have been a chance finding.

TABLE 14 - Numbers of safety outcomes and rates per 100,000 patient-years, by trial arm

Condition	Trial arm
	Intervention		Control
	Number	Rate (per 100,000 patient-year)	Number	Rate (per 100,000 patient-year)
Pneumonia	367	113.6	299	115.2
Pyelonephritis	115	35.6	84	32.4
Scarlet fever	60	18.6	72	27.7
Peritonsillar abscess	49	15.2	42	16.2
Septic arthritis	13	4.0	14	5.4
Osteomyelitis	21	6.5	11	4.2
Mastoiditis	7	2.2	7	2.7
Meningitis	9	2.8	6	2.3
Empyema	13	4.0	4	1.5
Intracranial abscess	1	0.3	3	1.2
Toxic shock/septicaemia	7	2.2	4	1.5
Lemierre syndrome	0	0	1	0.4

FIGURE 12.

Forest plot showing RRs (95% CI) of safety outcomes in the intervention trial arm compared with the control trial arm as reference. Estimates were from a Poisson model adjusted for age group, gender and comorbidity. Analyses for pneumonia were adjusted for the random effect of general practice. LL, lower limit of CI; UL, upper limit of CI.

FIGURE 13.

The number of safety outcomes, by trial arm and age group.

Process evaluation

The trial outcome measures were evaluated in relation to the level of utilisation of DSTs at the intervention trial arm practices. In the lowest quartile of utilisation, DSTs were viewed at < 1% of RTI consultations. In the highest quartile, up to 28% of RTI consultations were associated with DST viewing (Table 15). In adults aged 15–84 years, there was evidence of a linear trend across quartiles of DST utilisation (adjusted RR 0.96, 95% CI 0.93 to 0.99). This association was not evident for children (adjusted RR 0.98, 95% CI 0.94 to 1.03), the senior elderly (adjusted RR 0.99, 95% CI 0.94 to 1.05) and, only weakly, for the sample as a whole (adjusted RR 0.97, 95% CI 0.93 to 1.00; p = 0.043).

Table 16 shows the responses of GPs to the process evaluation questionnaire items. There were 51 respondents from 31 out of 41 (76%) intervention trial arm practices. Respondents generally gave positive responses to items concerning the monthly antibiotic-prescribing reports. Respondents were pleased to receive the reports and found that they provided credible data, which were was easy to read and understand. Respondents discussed these with colleagues and found that these data were useful for their practice. However, items concerning the possible impact on antibiotic prescribing were generally slightly less favourably addressed, with < 80% of respondents agreeing that the reports encouraged them to reduce antibiotic prescribing or had impact on the prescribing of antibiotics at the practice. The webinar was generally favourably received. The DSTs tended to be slightly less favourably received than the prescribing reports, with nearly one-third of respondents not affirming that the tools might support their reduction in antibiotic prescribing.

Main findings of the intervention development study

The intervention development study led to the refinement of the intervention materials for the trial, including electronically delivered antibiotic-prescribing reports and DSTs, supported by a webinar. The development study also provided insights into the preparedness of health professionals to engage in this antimicrobial stewardship intervention. The analysis revealed two main themes: the perception that the problem of overprescribing is not individually applicable and the perception that the researchers responsible for planning interventions in primary care do not understand how general practices operate.

Participants considered that they were not personally responsible for high levels of antibiotic prescribing, sometimes taking the view that the problem of overprescribing did not apply to them. This is inconsistent with analysis of data from 568 UK general practices contributing to the CPRD, which showed an overall antibiotic-prescribing proportion of between 50% and 60% for RTIs. 8 Such rates of prescribing suggest that most UK GPs and NPs prescribe at rates in excess of the current recommendations. 17 This perception may result from the ‘better than average effect’72 – a phenomenon in which people perceive themselves as special and feel that a special set of psychological rules apply to them. McKay and Dennett73 have argued that the illusion of superiority is an evolved human trait, serving an adaptive function by motivating adaptive behaviours. This illusion of superiority might be resistant to modification and should be recognised as part of normal human functioning.

A second overarching theme was the perception that researchers do not understand the demands of general practice. Most of the clinicians interviewed considered that clinical uncertainty concerning RTIs was low and were confident in diagnosing RTIs. However, convincing patients that antibiotics are not needed for a RTI episode was time-consuming and frustrating, and this perception was strengthened by the high-pressure working environment. Many participating GPs expressed the view that they have been de-professionalised in the process of achieving targets and pursing financial incentives, to the detriment of patient-centred care. Some GPs perceived certain elements of the intervention as being potentially unhelpful, with frequent audits being inconsistent with patient demands. The small risk of adverse complications representing an important concern in the context of performance review. These factors combined undermine GPs’ sense of professional competency and create anxiety. Therefore, it is possible that the intervention was not perceived as a possible source of support for GPs, resulting in less likelihood of using these tools. Similar barriers have been reported in the implementation literature,74,75 literature on GP morale76,77 and studies exploring the reasons why GPs leave practice early. 78

Although participants expressed concerns about the target-driven NHS culture and de-professionalised role of a GP, at the same time a majority of interviewed clinicians considered that funding to incentivise antibiotic reduction initiatives should be provided, as this would demonstrate the importance of reducing antibiotic prescribing to the NHS and would promote engagement. 79,80 Although the introduction of the pay-for-performance scheme in the NHS in 2004/2005 was associated with some improvements in quality of care for chronic diseases, evidence also suggest that practices divert their attention to activities associated with incentives. 81 It is possible that the scheme has contributed to the paradox reported by clinicians in this study, when they recognise the negative aspects of the target paradigm (i.e. improvements are achieved at the expense of those aspects of care that are not incentivised) and when clinicians are focused on ‘ticking the boxes’ rather than demonstrating empathy towards the patient.

Previous reviews of qualitative data suggest that GPs would welcome interventions that would bring benefits once implemented into general practice. 82 One such benefit in the REDUCE trial intervention is decreased workload, as patients who are not prescribed antibiotics for RTIs are less likely to consult in the future. 83 However, interviewed clinicians were concerned that involvement would have negative consequences in terms of workload and time. It is possible, therefore, that the tools developed for this study were not successful at conveying the message that the intervention would not disrupt usual daily practice and might bring future benefits in terms of a reduced number of consultations for RTIs.

The results demonstrate the potential limitations of implementing individual-level interventions, even when informed by a substantial understanding of user perspective and grounded in well-established behaviour change theory. A review of interventions to reduce unnecessary antibiotic prescribing found only a modest reduction of 9.7% (interquartile range 6.6–13.7%) in the proportion of patients receiving antibiotics. 20 More recent studies have achieved similar reductions, although the GRACE [Genomics to combat Resistance against Antibiotics for Community-acquired lower respiratory tract infection (LRTI) in Europe] trial reported a reduction of 15%. 21 Factors identified in this study, such as lack of perception of personal responsibility for changing antibiotic prescribing, might be less amenable to individual-level interventions. Attempts to curtail the clinical autonomy of GPs are likely to be counterproductive and the findings suggest that approaches aligned to the professional values of GPs are likely to be more acceptable.

Limitations of the intervention development study

The intervention development study included a convenience sample of GPs and NPs from non-trial practices. It is possible that those who volunteered for the study could be more motivated by, for example, vocational interest or a sense of professional duty. It is also not known whether the GPs recruited were comparable either with the national GP profile (e.g. representing group vs. single handed practices) or with those who later took part in the trial. The study cannot be sure whether or not intervention development respondents represented a mix of low and high antibiotic prescribers. The results of the interview study might overestimate the importance of individual-level factors and underestimate the importance of patient-level factors, organisational factors and the cultural and social context. Each interview was conducted using the same interview guide, but the differences in interviewing styles might have elicited different responses. Finally, this study asked clinicians about hypothetical, rather than the actual, use of the intervention, reducing the generalisability of our findings. Previous research suggests that initial scepticism and belief that the intervention would not be beneficial diminished over time and with experience in using the intervention that may not have been captured in our study. 84

Main findings of the cluster randomised trial

The cluster randomised trial evaluated the effectiveness of a complex electronically delivered intervention including antibiotic-prescribing reports and DSTs, supported by a webinar, at reducing antibiotic prescribing for self-limiting RTIs to people of all ages in primary care. The study provided evidence that these interventions might be associated with lower overall antibiotic prescribing for RTIs in primary care, but, given the size of the trial, this evidence may not be viewed as strong. This might be accounted for, in part, by the wide variations in practice that exist. It was observed that antibiotic utilisation for RTIs is patterned by age, with children accounting for a high proportion of all RTI consultations and antibiotic prescriptions. 9 There was no evidence that the intervention was effective at reducing antibiotic prescribing to children or the very old. In the adult population aged 15–84 years, there was evidence that antibiotic utilisation for RTIs was 16% lower in the intervention trial arm. The reduction in antibiotic utilisation was greater at practices that used the trial DSTs more frequently, which suggests a causal association. The study did not find evidence that the secondary outcome of proportion of RTI consultations with antibiotics prescribed was reduced. There was no evidence that the costs of health-care utilisation differed between trial arms over the course of the trial.

The agreed aim of the study, following discussion between the study team and the funders, was to address antibiotic prescribing in people of all ages. The trial general practice prescribing reports presented data aggregated across age groups. This was justified because any effective strategy must be shown to reduce antibiotic utilisation across the whole population. The results of the trial suggest that future interventions should stratify feedback and advice by age group. The trial DSTs specifically addressed common diagnostic concerns in children, including otitis media and cough and bronchitis. Prescribing to the youngest and oldest age groups may be more difficult to modify because safety concerns may be more salient at these ages. More research is needed to evaluate safety outcomes for more vulnerable subgroups of the population.

Strengths and limitations of the cluster randomised trial

The trial was conducted in the context of system-wide efforts to reduce unnecessary antibiotic prescribing in primary care. The NHS introduced an incentive known as the ‘Quality Premium’ to encourage reduced antibiotic utilisation. Antimicrobial stewardship interventions became available from a number of national and local sources. Perhaps as a result of some of these interventions, the English Surveillance Programme for Antimicrobial Utilisation and Resistance (ESPAUR) found that there was a 13% decrease in the number of antibiotic prescriptions dispensed from GP settings between 2012 and 2016. These system-wide interventions will have ‘contaminated’ both trial arms of the study. It is possible that a greater intervention effect might have been observed in their absence.

The intervention materials were successfully delivered into intervention trial arm practices. The DSTs were triggered when a Read code was entered into the general practice system. Thus, there was no control over the exact point during the consultation at which the intervention would be triggered. Practitioners had discretion over whether or not to view the intervention materials; viewing of the materials was not ‘forced’ through active-alerting or ‘pop-ups’ because these have been associated with negative feedback in previous studies. The study included a large sample of general practices. These were drawn from all parts of the UK including England, Scotland, Wales and Northern Ireland, including all registered patients in trial analyses. These features of the study enhanced the external validity of the study. However, it is possible that general practices that agreed to take part in the study might be more motivated to reduce antibiotic prescribing. At the time of the study, data could only be used from CPRD general practices employing the Vision practice system, many of which, during the course of the study, migrated to other practice systems and thus could not provide data to the REDUCE trial. There was a smaller pool of eligible practices than anticipated, the final number of practices included was smaller than originally intended and several practices were unable to continue with the trial before the end of the intervention period because they transferred to a different practice system. It should be noted that since this study was completed, the CPRD has increased its coverage to almost 1000 practices [the majority of which use the EMIS GP software system (EMIS Health, London, UK)] and the CPRD general practice coverage continues to grow. Data from EMIS practices could not be used in this study, but data are planned to be made available for future studies with the CPRD. Additionally, there was very wide variation in antibiotic prescribing, with different general practices prescribing antibiotics at as few as 12% or as many as 78% of RTI consultations. The sample size calculation did not allow for this overdispersion. Interventions might offer little benefit to those practices that had low prescribing rates. Consequently, the effect of intervention was not estimated precisely and there may be a risk of overlooking small effects that could be of clinical or public importance. Based on prespecified subgroup analysis, the study identified effects in adults aged 15–84 years. Interpretation of effects for subgroups carries a risk of false-positive interpretation. However, it was found that utilisation of intervention DSTs was associated with effect size, which points to a causal association. The study analysed data for antibiotic prescriptions issued by trial general practices. Patients may have received antibiotic prescriptions at consultations with walk-in centres and out-of-hours or emergency services; some antibiotic prescriptions may have been issued as delayed or deferred prescriptions but not recorded as such. It is possible that these alternative patterns of antibiotic utilisation differed between the trial arms, potentially contributing to biased assessment of intervention effects. It is also possible that altered diagnostic code selection may have occurred in order to justify antibiotic prescriptions. 85 There was necessarily no blinding of general practice staff to the intervention, which included feedback of counts of antibiotic prescriptions that were the outcome for the study. The study employed a hierarchical random-effects model for analysis. This provided estimates that can be interpreted as being conditional on cluster/general practice membership. Marginal population-averaged estimates are sometimes recommended in randomised trials, but Lee and Nelder86 showed that marginal predictions may also be obtained from a conditional model and this will usually have advantages compared with a marginal model. The data obtained from general practices showed evidence of overdispersion, but models were fitted that allow for overdispersion and these provided similar estimates. There was some evidence for regional variation in intervention effects and these could be investigated further, perhaps through qualitative research.

At the design stage of the project, it was concluded, through discussion with a health economist, that attempting a cost-effectiveness analysis would not be appropriate for this project. The study focused on the costs of health-care utilisation on the grounds that decision-makers need to be aware of whether or not an intervention leads to overall increased costs. The study did not identify any difference in health-care utilisation between general practices in the intervention and control trial arms. However, the study did not have access to linked Hospital Episode Statistics data for all trial practices, which would have strengthened the assessment of hospital utilisation. The research was conducted within the set of general practices contributing to the CPRD, which represents an existing research infrastructure. The CPRD draws on a single general practice information system; practices were further selected for this study that used, or were prepared to use, the DXS system. The immediate costs of delivering the intervention comprised the fees paid to the CPRD and DXS for work on the trial and for EHR data, as well as costs of staff time and materials for developing the intervention materials. The resources required to roll out a similar intervention into routine practice in the NHS might differ. On the one hand, economies of scale might reduce the marginal cost of delivering the intervention to a single general practice. On the other hand, there will be significant logistical challenges associated with working across general practices that employ different practice systems, sometimes with no established systems for data collection and analysis. Consequently, further feasibility work to evaluate resource requirements and logistical issues will be required before any possible scaling up of these interventions. Caution is also needed, as the intervention effects estimated in this 12-month trial might not be sustained in the longer term.

Comparison of trial results with other studies

Previous studies of audit and feedback interventions show that these often have only small effects. 29 Roshanov et al. 29 found that feedback interventions that provide advice to patients as well as physicians are associated with a greater chance of success. This was exemplified in the REDUCE trial DSTs, which offered patient information leaflets that could be viewed online or printed, as well as offering advice to physicians on the recognised indications for giving an antibiotic prescription. A recent trial reported on the outcome of quarterly feedback on antibiotic utilisation over 2 years among 2900 Swiss physicians. 87 Over the first and second years of the trial, there was no difference in antibiotic prescribing to all patients. The study found that there was nearly a 5% lower antibiotic utilisation to adults aged 19–65 years in the second year of the study, but not in the first year. The study appeared to identify age-related changes in prescribing that were not consistent over time. However, there are important differences from the present study because the feedback employed by Hemkens et al. 87 was less immediate (this study provided monthly rather than quarterly feedback) and because Switzerland already has very low antibiotic prescribing rates. 88 A study of dental practices in Scotland, which might account for 10% of community antibiotic prescribing, found that feedback of past antibiotic prescription data were associated with a 5.7% relative reduction in antibiotic prescribing over 12 months. 89 Audit and feedback have also been employed successfully to reduce other forms of high-risk prescribing in primary care. 90 Antibiotic prescribing to children was investigated by Cabral et al. ;91 the authors reported an analysis of four qualitative studies. Their analysis suggested that both parents and clinicians view children as being more vulnerable to the risks associated with RTIs. For parents and for doctors, giving an antibiotic prescription are viewed as safe courses of action. ‘Failing’ to intervene appropriately was anticipated to incur future disapproval from peers. 91 Similar considerations may influence the management of very old people. The development of clearer criteria for the prescribing of antibiotics to these more vulnerable groups is desirable. 92

Main findings of the cohort study of safety outcomes

Material in this section is reproduced from Gulliford et al. 31 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 3.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: http://creativecommons.org/licenses/by/3.0/.

The study used a large data set of EHRs to investigate the safety of reducing unnecessary antibiotic prescribing for RTIs in primary care. The results show that non-trial general practices that prescribe fewer antibiotics for respiratory infections may expect to have a higher incidence of pneumonia and peritonsillar abscess than higher-prescribing general practices. If a general practice, with an average list size of 7000 patients, reduced the proportion of RTI consultations with antibiotics prescribed by 10%, it might encounter about one additional case of pneumonia each year and one additional case of peritonsillar abscess each decade. Changes will be proportionately greater for larger reductions in antibiotic prescribing. These estimates represent averages across general practice populations, but complications might be fewer than expected if GPs are able to effectively stratify antibiotic prescribing on the basis of the level of risk. There was no evidence that diagnoses of mastoiditis, empyema, bacterial meningitis or intracranial abscess might increase. Lemierre syndrome was rare, with about one case per 2 million person-years, but there was no evidence that this was more frequent at low-prescribing practices. This is reassuring in view of recent suggestions that F. necrophorum may often be present in patients with sore throat. 36 These estimates must be viewed in the context of quantitatively important declining secular trends in incidence for several infective complications of RTIs, including peritonsillar abscess, mastoiditis and meningitis. Bacterial meningitis from pneumococcal, meningococcal or Haemophilus infection has declined following the introduction of vaccination programmes. 93 However, the incidence of pneumonia showed a slight increase over time, consistent with previous studies based on hospital admissions. 94,95

Reducing the proportion of RTI consultations with antibiotics prescribed by 10% is expected to be accompanied by some 2000 fewer antibiotic prescriptions per practice over a 10-year period. Benefits to individual patients from avoiding antibiotics include reductions in common adverse reactions to antibiotics (including rashes, vomiting and diarrhoea), which may affect 10% of patients,12 as well as less common side-effects, such as anaphylaxis. Benefits to general practices may include a demedicalisation of RTIs followed by a decline in the consultation rate, as previous observational studies show higher-prescribing general practices attract more consultations for RTIs. 83 Trial evidence shows that even one antibiotic prescription increases the likelihood of re-consultation with a new episode of RTI. 11 Most of the ‘complications’ identified do not require hospital admission and, currently, respond well to antibiotics, so simply the occurrence of an uncommon complication rates is not in itself a reason to justify more widespread prescribing of antibiotics for initially uncomplicated presentations. There was no evidence for a threshold between ‘safe’ or ‘unsafe’ prescribing levels. Inspection of forest plots suggested some departure from linearity, but the addition of non-linear terms did not improve the goodness of fit of regression models.

Strengths and weaknesses of the cohort study in relation to other studies

Previous studies have consistently demonstrated high levels of unnecessary prescribing of antibiotics for RTIs in primary care. 96 Although there has been a declining trend in the consultation rate for RTIs,7 there has been little change in the proportion of consultations with antibiotics prescribed in the UK7 in spite of the efforts of researchers, clinicians and policy-makers to bring about changes. This is in contrast to Sweden, where there has been a long-term decline in antibiotic prescribing for respiratory indications without evidence of increased risk of bacterial complications. 97 General practitioners may often be concerned to meet patients’ expectations for antibiotic prescriptions,98 but both patients and prescribers may also have concerns about the safety of non-prescribing strategies. 98 Petersen et al. 32 provided evidence that antibiotics reduced the risk of pneumonia, mastoiditis and peritonsillar abscess (quinsy), but did not quantify the potential population impact of these complications.

The present results represent averages across general practice populations. There is considerable diversity among the population of patients at risk of RTIs. Current management guidelines for RTIs recommend that specific groups of patients should be considered to have positive indications for antibiotic treatment. 12 An immediate antibiotic prescription is recommended if patients are very unwell, or have clinical features suggestive of serious illness or complications,99 or if comorbidity is present, or the patient is very young or very old. Further research is needed to evaluate whether or not the present results will be confirmed when subgroups that may be at higher risk, including older adults, are analysed separately. It is possible that general practices with the same overall level of antibiotic prescribing may differ in the appropriateness of their management of patients with defined markers of vulnerability and this could influence the rate of complications. However, the clinical features of a RTI episode may have only limited predictive value for the future occurrence of complications, and a high proportion of complications may occur in patients who appeared to be at low risk. 100 A delayed antibiotic-prescribing strategy, in which a prescription is issued but used only if symptoms fail to improve, is sometimes recommended as a method for reducing antibiotic utilisation in the management of RTIs. 27 Delayed antibiotic prescribing may be as effective as immediate use of antibiotics in the prevention of complications of sore throat. 101 The development and application of point-of-care testing to guide antibiotic prescribing may have a future role in identifying individuals who may benefit from antibiotic treatment. 102,103

Strengths and weaknesses of the cohort study

This study comprised > 600 general practices, with a registered population of > 4 million patients and 45 million person-years of observation. Consequently, the study provided very precise estimates for the more frequent outcomes evaluated. It is acknowledged that there was lower power to evaluate potential changes in less frequent outcomes. It can be concluded that the absolute risks of mastoiditis, empyema, intracranial abscess or Lemierre syndrome remain small even in low-prescribing practices. This study adopted a population perspective, aiming to quantify the outcomes of either high- or low-prescribing strategies in the management of RTIs. Consequently, the study evaluated changes in infective complications at the level of the general practice population. The research did not address variation in prescribing at the level of the individual physician. The research did not show whether or not individual patients who experienced complications received antibiotics. Conclusions might differ if individual-level analyses showed that complications arise in patients who were treated with antibiotics. This study did not evaluate the outcomes of individual patients who were diagnosed with complications. Further research is required to evaluate the severity of complications, such as pneumonia, and their outcomes, including mortality. Future studies might also make use of linked hospital episode data, which are available for selected CPRD practices in England in more recent years, in order to evaluate hospitalised cases in more detail. The risk of complications associated with different classes of antibiotics also merits study. It is acknowledged that there may be other complications, such as a proportion of all cases of septicaemia diagnosed in primary care, which might follow from RTIs. The study acknowledges several sources of misclassification: the study used a sample of the CPRD to estimate consultation and prescribing rates; there is variation among general practices in the use of diagnostic categories;85 general practice populations may vary in their use of out-of-hours and emergency services, whose generally higher antibiotic prescribing may not be captured in the CPRD; and some general practices may use delayed antibiotic-prescribing strategies,101 but these were not distinguished in the analysis of prescriptions issued. It is also possible that use of near-patient testing may have contributed to better diagnosis during the period. These forms of misclassification will generally tend to diminish estimated associations, but might cause bias if effects are differentially distributed across prescribing categories. Diagnostic coding may have a subjective element104 and bias might arise if low-prescribing practices are more likely than high-prescribing practices to code ‘pneumonia’ in order to justify the prescription of an antibiotic. The research utilised non-randomised data, and the data were adjusted for age, gender, region, deprivation category and general practice, but it is possible that unmeasured confounders might have biased the reported associations. Caution is required in the analysis of large data sets; ‘significant’ results must be judged in relation to their clinical importance. Antibiotic prescribing in the UK is high compared with some international comparators, and the study cannot be sure that the associations reported here would also hold at very low antibiotic-prescribing levels.

Conclusions

The study identified a range of views that may influence GPs’ acceptance of an intervention addressing the problem of inappropriate antibiotic prescribing to patients with RTIs. Primary care prescribers may tend to deny personal responsibility for overprescribing and may not engage with the intervention as having relevance for them. Such individual attitudinal barriers might suggest that solutions need to be developed and implemented at an organisational level as well as, or instead of, at the individual practitioner level. It appears likely that barriers to active engagement in research interventions identified in this study might go beyond antibiotic prescribing and this might suggest a need to shift the focus of research towards less individual-based approaches, as well as addressing the possible negative attitudes of individual health professional. These findings might be useful for researchers, professional organisations and guideline developers who engage in designing interventions for primary care.

This research successfully conducted an efficient trial to evaluate the effectiveness of electronically delivered antimicrobial stewardship interventions in a large population. The study included 79 general practices, with a registered population of > 600,000 people. The study found evidence that, overall, general practice antibiotic prescribing for RTI is reduced by these electronically delivered interventions. The imprecise result reflected very wide variations in clinical practice. The study found evidence that antibiotic prescribing might be reduced by 16% in adults aged 15–84 years, but not in children or very old people. This finding is consistent with previous research that shows that parents and doctors may be cautious in their management of children. It can be concluded that future intervention strategies for antimicrobial stewardship should employ stratified interventions that are tailored to the needs and concerns of specific age groups, with reporting of antibiotic utilisation disaggregated by age.

There was no evidence from trial data that any safety outcomes might be increased through interaction with the intervention. The larger cohort study in non-trial practices provides evidence that general practices that prescribe antibiotics less frequently at consultations for RTIs may experience a slight increase in the incidence of pneumonia and peritonsillar abscess, which would be expected to respond to treatment while bacterial pathogens remain sensitive to antibiotics. No increase in mastoiditis, empyema, meningitis, intracranial abscess or Lemierre syndrome is likely. Even a large reduction in antibiotic prescribing was predicted to be associated with only a small increase in numbers of cases observed over a 10-year period, and this would be expected to reduce the risks of antibiotic resistance, the side effects of antibiotics and, the medicalisation of largely self-limiting illnesses. The safety outcomes of no antibiotic-prescribing strategies for RTIs are an important aspect for communication to patients and the public in the context of wider communication strategies to support antimicrobial stewardship. 105

Data sources

The study is based, in part, on data from the CPRD obtained under licence from the UK Medicines and Healthcare products Regulatory Agency. However, the interpretation and conclusions contained in this report are those of the authors alone.

Funding

The trial is funded by the NIHR Health Technology Assessment Programme (reference number 13/88/10). Martin C Gulliford was supported by the NIHR Biomedical Research Centre at Guy’s and St Thomas’ Hospitals.

Trial registration

The trial was registered as ISRCTN95232781. The trial was registered on 18 November 2014.

Trial Steering Committee

Jackie Cassell (chairperson), Susan Hopkins (Public Health England), Tim Chadborn (Public Health England) and Nanik Pursani (Lay member).

Data Monitoring Committee

Christine A’Court (chairperson), Helen Strongman (CPRD/MHRA), Derek Cook (St George’s, University of London) and Jason Oke (University of Oxford).

Contributions of authors

Martin C Gulliford (Professor of Public Health) contributed to trial design, conduct and analysis and drafting of the report.

Dorota Juszczyk (Research Associate) led the intervention study and the design of the intervention. She analysed the qualitative data and reported on the intervention development study.

A Toby Prevost (Professor of Medical Statistics and Clinical Trials) contributed statistical advice on the design and conduct of the study. He drafted the statistical analysis plan and advised on the conduct and reporting of the statistical analyses.

Jamie Soames (Senior Clinical Project Manager) was responsible for recruiting general practices to the study and liaising with practices for intervention delivery and process evaluation.

Lisa McDermott (Research Associate) contributed to the design of the trial, the development of the intervention and the design of the process evaluation.

Kirin Sultana (Senior Clinical Project Manager) was responsible for recruiting general practices to the study and liaising with practices for intervention delivery and process evaluation.

Mark Wright (Director of Interventional Research) was responsible for overseeing the design and conduct of the CPRD contribution to the study.

Robin Fox (General Practitioner) contributed to the design of the study and provided clinical advice on intervention development and interpretation of results.

Alastair D Hay (Professor of Primary Care) contributed clinical advice to the study, including intervention design and development and interpretation of trial and safety outcomes.

Paul Little (Professor of Primary Care Research) contributed clinical advice to the design of the study, including intervention design and development and analysis of trial and safety outcomes.

Michael Moore (Professor of Primary Care) contributed clinical advice to the study, including intervention design and development and analysis of trial and safety outcomes.

Lucy Yardley (Professor of Health Psychology) contributed advice on the intervention development study, the design of the intervention, and the design and analysis of the process evaluation.

Mark Ashworth (Reader in General Practice) contributed clinical advice to the study, including intervention design and development and interpretation of results.

Judith Charlton (Research Associate) analysed all the CPRD data for the prescribing reports for the study, wrote a program to prepare the reports on an automated basis, and programmed the main trial analyses from the statistical analysis plan.

All authors contributed to, and approved, the final manuscript.

Publications

Juszczyk D, Charlton J, McDermott L, Soames J, Sultana K, Ashworth M, et al. Electronically delivered, multicomponent intervention to reduce unnecessary antibiotic prescribing for respiratory infections in primary care: a cluster randomised trial using electronic health records-REDUCE Trial study original protocol. BMJ Open 2016;6:e010892.

Gulliford MC, Moore MV, Little P, Hay AD, Fox R, Prevost AT, et al. Safety of reduced antibiotic prescribing for self limiting respiratory tract infections in primary care: cohort study using electronic health records. BMJ 2016;354:i3410.

Gulliford MC, Prevost AT, Charlton J, Juszczyk D, Soames J, McDermott L, et al. Effectiveness and safety of electronically delivered prescribing feedback and decision support on antibiotic use for respiratory illness in primary care: REDUCE cluster randomised trial. BMJ 2019;364:l236.

Data-sharing statement

The data analysed in this study were obtained from the CPRD (www.cprd.com) under licence. The agreements in place for these data do not permit further distribution or sharing. Further information can be obtained from the corresponding author.

Patient data

This work uses data provided by patients and collected by the NHS as part of their care and support. Using patient data is vital to improve health and care for everyone. There is huge potential to make better use of information from people’s patient records, to understand more about disease, develop new treatments, monitor safety, and plan NHS services. Patient data should be kept safe and secure, to protect everyone’s privacy, and it’s important that there are safeguards to make sure that it is stored and used responsibly. Everyone should be able to find out about how patient data are used. #datasaveslives You can find out more about the background to this citation here: https://understandingpatientdata.org.uk/data-citation.

Disclaimers

This report presents independent research funded by the National Institute for Health Research (NIHR). The views and opinions expressed by authors in this publication are those of the authors and do not necessarily reflect those of the NHS, the NIHR, NETSCC, the HTA programme or the Department of Health and Social Care. If there are verbatim quotations included in this publication the views and opinions expressed by the interviewees are those of the interviewees and do not necessarily reflect those of the authors, those of the NHS, the NIHR, NETSCC, the HTA programme or the Department of Health and Social Care.