Abstract
Background Respiratory tract infections (RTIs) in children are common and often result in antibiotic prescription despite their typically self-limiting course.
Aim To assess the effectiveness of primary care based interventions to reduce antibiotic prescribing for children with RTIs.
Design and setting Systematic review.
Method MEDLINE®, Embase, CINAHL®, PsycINFO, and the Cochrane library were searched for randomised, cluster randomised, and non-randomised studies testing educational and/or behavioural interventions to change antibiotic prescribing for children (<18 years) with RTIs. Main outcomes included change in proportion of total antibiotic prescribing or change in ‘appropriate’ prescribing for RTIs. Narrative analysis of included studies was used to identify components of effective interventions.
Results Of 6301 references identified through database searching, 17 studies were included. Interventions that combined parent education with clinician behaviour change decreased antibiotic prescribing rates by between 6–21%; structuring the parent–clinician interaction during the consultation may further increase the effectiveness of these interventions. Automatic computerised prescribing prompts increased prescribing appropriateness, while passive information, in the form of waiting room educational materials, yielded no benefit.
Conclusion Conflicting evidence from the included studies found that interventions directed towards parents and/or clinicians can reduce rates of antibiotic prescribing. The most effective interventions target both parents and clinicians during consultations, provide automatic prescribing prompts, and promote clinician leadership in the intervention design.
INTRODUCTION
Respiratory tract infections (RTIs) in children are common and costly conditions for families, healthcare providers, and health systems.1 Clinicians frequently prescribe antibiotics for RTIs,2 despite the fact that most are self-limiting and use of antibiotics for most RTIs is of uncertain value.3 Overuse of antibiotics is associated with development of antimicrobial resistance,4 increased care-seeking behaviour,5 and adverse effects.6 Perhaps the greatest threat to public health comes from the continuing emergence of antimicrobial resistance. This leads to increased use of second/third generation antibiotics, costlier treatment, and further bacterial resistance.
Efforts to reduce antibiotic prescribing have been ongoing for decades,7 and have included a wide range of strategies and campaigns targeted at patients, clinicians, practices, and whole populations. To some extent these efforts have been successful, leading to reductions in the UK of 24%, from 572 antibiotic prescriptions per 1000 child-years in 1996 to 435 prescriptions per 1000 child-years in 2000.8 However, US data indicates that while overall antibiotic prescription rates decreased during the 1990s, prescribing rates of broad spectrum antibiotics for children with RTIs actually increased.9 Prescribing rates for non-specific RTIs in the UK have increased by 10% since 2002.8 This upward trend is concerning in light of evidence-based practice recommendations that propose a ‘wait and see’ approach for the majority of RTIs.
A recent review of interventions to modify parental help-seeking behaviour for RTIs in children found that interventions that engaged children in addition to parents and provided specific symptom guidance were effective at influencing consulting behaviour.10 However, strategies are also needed to help clinicians determine which children are most in need of antibiotics,11 and reduce inappropriate antibiotic prescribing.12,13 Given the importance of the parent–clinician interaction in guiding antibiotic use, this study aimed to systematically review the effectiveness of educational or behavioural interventions directed to parents, clinicians, or both, to reduce antibiotic prescribing for children with RTIs in primary care.
METHODS
MEDLINE®/PubMed, CINAHL®, Embase, PsycINFO and the Cochrane Library (from inception through June 2012) were searched using terms for RTIs, children, parents, education, antibiotic prescription, and consultation (Table 1). One author screened titles and abstracts based on predefined inclusion criteria to identify relevant studies and reviewed reference lists and related citations of selected studies to identify additional references. Two authors reviewed the full-text of selected studies to determine inclusion. Disagreements were settled through discussions with a third author.
Characteristics of included studies
How this fits in
Prescribing rates of antibiotics for RTI in children have declined, but are still high and largely unnecessary. Reducing unnecessary prescriptions is a priority in order to reduce inappropriate antibiotic use in primary care. Based on a systematic review of 17 studies it was found that the most effective interventions target both parents and clinicians during consultation, provide automatic prescribing prompts, and promote clinician leadership in the intervention design. These can produce significant reductions in antibiotic prescribing for children with RTIs.
Controlled studies were included that used a randomised, cluster randomised, non-randomised or one-group pre- and post- test design to assess the effectiveness of educational or behavioural interventions to change clinicians’ antibiotic prescribing for acute RTIs in children (birth to 18 years) in primary care settings (family practice, emergency, or paediatric primary care). Outcomes of interest were change in proportion of antibiotic prescriptions issued for RTIs in children, or change in ‘appropriate’ antibiotic prescribing. Comparisons included no-treatment or alternate treatment controls. Studies were excluded if they were: from in-patient settings; evaluations of treatment guidelines, public health interventions, diagnostic tests; studies of children with chronic illnesses or serious comorbidities; or studies from countries not classified as high-income by the Organisation for Economic Co-operation and Development.
Two reviewers used an extraction form developed for a previous systematic review10 to independently extract data for study design, setting, patient population, intervention, comparison, outcome(s), and assessment method. Disagreements were resolved by discussion with a third author. Reviewers were not blinded to any aspect of the studies. Data from a French language study were extracted following translation.
Two reviewers independently assessed study quality using a framework adapted from the Cochrane handbook.14 Randomised or cluster randomised trials were assessed based on randomisation, blinding, description of intervention, exposure to intervention, and generalisability. Non-randomised controlled trials were assessed on the basis of comparability of groups, intervention description, exposure to intervention, and generalisability. One-group designs were assessed based on intervention description, exposure to intervention, and generalisability. A judgement of ‘low’, ‘high’, or ‘unclear’ was made regarding the risk of bias for each criterion; based on this, each study was then given an overall judgement of ‘low’, ‘moderate’, or ‘high’ risk of bias (Table 2). Overall quality assessments were used to interpret the findings.
Effects of interventions targeting clinicians and parents to reduce antibiotic prescribing for respiratory tract infections in children
Mean differences were calculated with 95% confidence intervals (CI) for changes in mean numbers of prescriptions, and odds ratios (OR) with 95% CI for changes in prescribing rates, using Yates’s correction and Fisher’s exact test where an expected cell was below five (EpiInfo version 3.4.3). Where raw data were unavailable, proportional or mean differences were presented. Considerable statistical and clinical heterogeneity prevented pooling of outcomes; therefore results of each study are presented individually and interpreted using narrative analysis.
RESULTS
Of the 6301 references returned in the search, 17 studies met inclusion criteria (Figure 1).15–31 One study21 included three different interventions (targeting parents, clinicians, or both) for a total of 19 interventions among the 17 studies. Thirteen studies involved 228 practices or clinics (four studies15,18,21,31 did not report number of included practices). The studies varied in design, paediatric population, and length of follow-up (Table 1). The majority of studies used a randomised design (n = 12), with the remaining studies using pre- and post-test (n = 3) or non-randomised designs (n = 2). Most were conducted in the US (n = 10), followed by Israel (n = 3), Europe (n = 3), and Australia (n = 1). The interventions were delivered in family practice or paediatric care settings, except for one set in an after-hours clinic.23 The majority of interventions (n = 10) were directed toward clinicians and parents.15–24 Six interventions were directed toward clinicians only;21,25–29 three interventions targeted parents only.21,30–31
Flow of included studies
Effects of interventions targeting clinicians and parents
Eight of the 10 interventions which targeted both clinicians and parents reported significantly decreased prescribing rates,15,17–20,22–24 with reductions ranging from 6–21% at follow-up from 1 week15 to 2 years24 (Table 2). The largest effect was observed in a study which used a combined parent–clinician ‘interactive book’ during the consultation, resulting in a lower prescribing rate of 19.5% (versus 40.8%, P<0.001) at 2 weeks.19 One intervention, a combination of academic detailing and written parent education, showed no effect.16 The remaining study by Mainous et al reported increased rates of antibiotic prescribing in both intervention (15.3%) and control groups (22.5%) during the 5-month study period.21
Most studies had a moderate risk of bias due to poor reporting of methods or uncertain participant exposure to the intervention.16–18,20,21,23,24 The method of randomisation was not reported in either study where there was no intervention effect.16,21 Indeed, Doyne et al noted that the lack of effect could in part be due to dissimilar prescribing rates between groups at baseline and to concurrent media campaigns.16
Three studies reported adverse events or re-consultation rates, and reported no difference in rates of mastoiditis23 or re-consultations.19,20 Cohen et al found no significant difference in symptom duration between groups despite a significantly lower rate of antibiotic prescription (26.7% versus 37.1%, P<0.001) in the intervention group.15
Effects of interventions targeting clinicians only
Of the six interventions targeted only to clinicians, one29 reported a significant reduction in antibiotic prescribing, and a further two27,28 reported significant reductions in inappropriate prescribing (Table 3). The remaining three studies found either no significant reduction or an increase in antibiotic prescribing.21,25,26
Effects of interventions targeting clinicians to reduce antibiotic prescribing for respiratory tract infections in children
Razon et al studied the effect of a 1-day educational seminar for clinicians and found significant reductions in antibiotic prescriptions per diagnosis at 4 months for acute otitis media (OR 0.52 [95% CI = 0.42 to 0.66]), but not for pharyngitis/tonsillitis, sinusitis or undefined upper respiratory infection.29 Using a computer decision support system (CDSS) that automatically began each time a clinician wrote an antibiotic prescription, Christakis et al reported a 34% reduction in the frequency of inappropriate prescribing (prescriptions for >10 days); however, the prescribing rate increased overall during the 7-month follow-up, though to a lesser extent in the intervention group (4.3% versus 16.8%, non-significant).27 Margolis et al found significant decreases in rates of ‘incorrect’ antibiotic prescribing for otitis media (34%) and pharyngitis (29%) but not upper respiratory infections among clinicians using a computerised algorithm, however, the study was stopped prematurely due to low participation.28
Prescribing feedback reports did not reduce prescribing, in fact rates increased by 15.2% at 5 months (versus 22.5% among controls).21 Interventions in two other studies that did not find significant reductions in prescribing included an optional CDSS26 (in contrast to the automatic system in the Christakis study), and prescribing feedback reports coupled with group education sessions.25
All studies in this group presented a moderate risk of bias due to unclear methods,21,25,26,28 uncertain or low exposure to the intervention,21,25,26,28 lack of detail in the intervention description,21,28 or lack of control group.29
Effects of interventions targeting parents only
None of the interventions directed only at parents significantly reduced antibiotic prescribing (Table 4).21,30–31 Ashe et al reported that a waiting room poster with information on judicious antibiotic use did not produce a significant difference in prescribing rates between intervention and control groups (41.9% versus 48.6%, P = 0.09).30 Testing the effect of patient education pamphlets, Mainous et al observed an overall 12.6% increase in prescribing over 5 months, although this was lower than the increase seen in the control group (22.5%).21 In the study by Taylor et al, parents viewed a brief videotape message and received a pamphlet 6 weeks and 6 months after initial randomisation (materials included education about the judicious use of antibiotics); at 12 months there was no significant difference in the number of RTI consultations resulting in antibiotic prescription between groups.31 The studies by Ashe and Taylor had a low risk of bias, whereas the Mainous study had a moderate risk of bias due to unclear methods and intervention description. No study assessed extent of exposure to the intervention.
Effects of interventions targeting parents to reduce antibiotic prescribing for respiratory tract infections in children
DISCUSSION
Summary
Conflicting evidence from the 17 studies found that interventions directed towards parents and/or clinicians can reduce rates of antibiotic prescribing for children with RTIs. The most effective interventions involved targeting both parents and clinicians during a consultation,19 providing automatic computer prompts for evidence-based prescribing,27 and promoting clinician leadership or participation in the design of treatment guidelines and/or peer education.22,24 There was moderately strong evidence that interventions were more effective in reducing antibiotic prescribing when delivered to clinicians in collaboration with parents.15,17–19,22–24 In contrast, based on limited evidence, passive strategies targeting only parents, such as waiting room posters or pamphlets, do not appear to alter prescribing rates significantly.30–31 Moreover, interventions involving printed materials for parents varied in effect; those with actionable information (such as self-care advice and signs to re-consult)18 were more effective in reducing rates of antibiotic prescription than materials with generic information on the appropriate use of antibiotics.30 The findings suggest computer-based interventions are only successful when integrated into routine clinical processes (for example, writing prescriptions) and less so when clinicians must manually employ the application.
Strengths and limitations
Only published studies were included therefore unpublished studies of relevant interventions may have been missed. To address risk of publication bias multiple databases were searched, the search by language was not limited, and reference lists and related citations of included studies were also searched. The focus was on studies from high income countries, which may limit the generalisability of the findings to low/middle income settings. Overall methodological quality of the included studies was highly variable and generally moderate. Most studies did not report the extent of parent and/or clinician completion or participation in intervention activities; this risk of bias may further limit the robustness of conclusions that can be drawn from the reported findings. Studies which assess ‘appropriate’ prescribing could be subject to changes in diagnostic labelling by participating clinicians, which would bias the results toward a positive intervention effect.32 Also, diagnostic criteria for eligible RTIs were not clearly described, and it is unclear how generalisable the spectrum of illness was in study populations at enrolment. Only three studies reported complications or re-consultations; results from these studies did not indicate increased risk of adverse events related to decreased prescribing but would not have been adequately powered to identify effects on less common adverse outcomes (for example, hospital admission).19,20,23
Comparison with existing literature
Previous reviews have explored the effectiveness of interventions to change antibiotic prescribing behaviour of clinicians for various types of infection in adults and children.13,33–35 These reviews and prior research similarly concluded that the most effective interventions involve clinicians and patients,35–37 as well as the general public.7,32 In a systematic review of antimicrobial control programmes in paediatric outpatient and hospital settings (of which four studies overlap with this review) Patel et al concluded that provider-targeted interventions which featured diagnosis-specific education were more likely to change prescribing for childhood infections.34 Two systematic reviews (Arnold33 and Ranji35) examined effectiveness of clinician and/or parent strategies to reduce antibiotic prescribing for adults and children for all conditions in outpatient settings. Although only a small number of the studies in this review overlapped with these (six out of 30 studies in Ranji; two out of 39 studies in Arnold), the findings broadly concur with their conclusions that effective interventions to reduce antibiotic prescribing involve multifaceted approaches targeting clinicians and patients, and that printed materials or audit and feedback had limited effect.
Finally, a systematic review13 of interventions to change health professional’s behaviour (including prescribing, referral, clinician knowledge, and guideline compliance) in management of children with upper RTIs in any type of setting identified 10 studies (six of which are also included in this review20,23–25,27,28). It concluded that computer interventions, educational sessions, collaboratively developed guidelines and training videos were effective in changing practice, and that multifaceted and computer interventions worked best. This review identified an additional 11 studies specific to antibiotic prescribing for children with RTI in primary care, and included interventions directed towards both clinicians and parents, which more realistically reflects actual practice. Only mixed evidence was found to support CDSS to change clinician behaviour (partly due to the inclusion of a newer study26) and noted that the more effective CDSS provided: recommendations rather than just assessments; and automatic decision support at the time and location of decision-making.38 We found two studies including consultation skills training,19,22 which has been shown to be effective in reducing antibiotic prescribing for adults.39–40
Implications for practice and research
For policymakers, the findings of this study suggest that more appropriate prescribing for RTIs in children may be achieved when interventions are designed in consultation with participants, incorporate changes into everyday prescribing processes, and address the needs of parents and clinicians. Passive approaches such as waiting room posters and written materials in isolation have limited effects. However, the cost-effectiveness of these interventions and effects on other health service outcomes such as repeat attendance or risk of complications need to be determined. In addition, clinicians and parents need evidence for the effectiveness of alternatives to antibiotic therapy for symptomatic relief of RTIs.41 Qualitative research, involving parents and clinicians, of the reasons why some interventions are more effective than others could improve the understanding of effective interventions. Ongoing studies involving multi-component interventions, HAPPY AUDIT,42 DECISION+,43 and TARGET (http://targetstudy.org.uk/), will likely contribute new data to these research gaps.
Acknowledgments
We gratefully acknowledge the contribution of Dr Jan Verbakel for assistance extracting data from a non-English language study.
Notes
Funding
This article presents independent research funded by the National Institute for Health Research (NIHR) under its Programme Grants for Applied Research Programme (RP-PG-0608-10018). The views expressed in this publication are those of the authors and not necessarily those of the NHS, the NIHR or the Department of Health.
Provenance
Freely submitted; externally peer reviewed.
Competing interests
The authors have declared no competing interests.
Discuss this article
Contribute and read comments about this article on the Discussion Forum: http://www.rcgp.org.uk/bjgp-discuss
- Received November 6, 2012.
- Revision received November 26, 2012.
- Accepted January 8, 2013.
- © British Journal of General Practice 2013
REFERENCES
In this issue
Jump to section
More in this TOC Section
Related Articles
Cited By...
