Community Paediatrics

Appropriateness of antibiotic prescribing in paediatrics: retrospective controlled study assessing a multifaceted intervention in Northern Italy in a 7-year period

Abstract

Background Academic detailing, audit and feedback, and peer comparison have been advocated as effective ways to promote appropriateness of prescribing and antimicrobial stewardship (AMS). This study explored the effectiveness of a multifaceted intervention aimed at supporting the appropriateness of antibiotic prescribing in paediatrics.

Methods Over the course of 7 years, all 89 paediatricians of the Local Health Authority (LHA) of Reggio Emilia (530 000 residents) were provided with scientific literature focused on antimicrobial resistance and the appropriateness of use of specific antibiotics, together with local data on antimicrobial resistance and prescribing reports comparing each paediatrician with colleagues in the same district and with local averages. Prescribing rates of specific target antibiotics/classes of antibiotics were evaluated by comparing Reggio-Emilia with the other seven LHAs of the Emilia-Romagna Region (control area), adjusting for prescriptions during a 2-year baseline period.

Results A significant increase in the rate of amoxicillin prescriptions (91 more per 1000 children/year) was observed in the intervention area compared with the control area along with a significant reduction in the rate of amoxicillin+clavulanate prescriptions (70 fewer per 1000 children/year) and a significant increase in the ratio of their prescription rates. No differences were observed in cephalosporin and macrolide prescription rates and overall antibiotic prescriptions.

Conclusions Improvements in prescribing appropriateness were observed. This study confirms the importance of an audit and feedback approach through small group meetings supported by scientific literature, local resistance data and prescribing reports. Such approach should always be considered as part of multifaceted interventions to promote AMS.

What is already known on this topic

  • Academic detailing, prescribing audit and feedback, and peer comparison have all been advocated as effective ways to promote appropriateness of prescribing and antimicrobial stewardship.

What this study adds

  • A multifaceted intervention consisting of ad hoc prescription reports, data on antimicrobial resistance, scientific literature and peer comparison, can be effective in improving the appropriateness of antibiotic prescribing

How this study might affect research, practice or policy

  • Such integrated approach is easily applicable in outpatient settings and should be considered as part of antimicrobial stewardship strategies

Introduction

The excessive use of antibiotics promotes antimicrobial resistance, which represents a growing threat to global health. Antibiotics are often prescribed unnecessarily and inappropriately, especially for respiratory tract infections. They are the most prescribed drugs in the paediatric field.1 It has been shown that 20%–50% of these prescriptions are potentially inappropriate2; most children receive broad-spectrum antibiotics for viral upper respiratory infections, or courses of antibiotics that are significantly longer than necessary.3 To mitigate the emergence of antimicrobial resistance, it is necessary to promote prescribing appropriateness, avoiding antibiotic use in case of mild and self-limiting infections. In this regard, the WHO classified antibiotics into three categories (‘Access’, ‘Watch’ and ‘Reserve’), considering their impact on antimicrobial resistance, in order to highlight the importance of their proper use.4

Cultural factors seem more important than epidemiological contexts, to explain the threefold difference in antibiotic prescribing between European countries with the highest and the lowest use (usually Southern and Northern European countries, respectively).5 Italy is among the European countries with the highest consumption and with the highest rates of resistance and multiresistance (resistance of a bacterium to at least four antibiotics of different classes).6 Italy is also among the European countries with the lowest consumption of antibiotics in the Access group, that is, first choice ones, whereas it is one of the countries with the highest consumption of Watch group drugs7—antibiotics which should be used with caution due to the greater risk of inducing resistance. As regards territorial variability, there is more than twofold variation in the prescription of antibiotics between the regions of Southern and Northern Italy. Approximately, 90% of antibiotic consumption in the National Health System (NHS) is the result of prescriptions from general practitioners or paediatricians.7 In Italy, Local Health Authorities (LHAs) are responsible for the organisation and provision of healthcare services and the management of public health within specific geographical areas. Among these services is the provision of comprehensive care for children from birth to 14 years old. Primary care paediatricians work as independent practitioners under a contractual agreement with the LHA.

Educational interventions have been associated with more appropriate antibiotic prescribing.8–10 The objective of this study was to assess whether a multifaceted intervention could improve quality of antibiotic prescriptions in primary care paediatrics through a reduction in the prescription of broad-spectrum antibiotics.

Methods

Between 2013 and 2019, an antimicrobial stewardship (AMS) programme was promoted at the LHA of Reggio Emilia, which serves a province in the Emilia-Romagna Region (in Northern Italy) with approximately 530 000 residents (approximately 15% <14 years old). Historically, Emilia Romagna has been characterised by a low prevalence of antibiotic prescriptions compared with other Italian regions, also thanks to the implementation of recommendations on the appropriate prescription of antibiotics in primary care. Our intervention was therefore implemented in an area where the attention on this issue was already high.

More specifically, district meetings (one for each of the six districts of the LHA of Reggio Emilia) were held in Autumn of each of the monitored years. These meetings involved a total of approximately 80 paediatricians each year, pharmacists from the National Health Service, and managers of primary care services,9 and had the aim of promoting the exchange of information and encouraging discussion among peers. Before the district meetings, paediatricians were emailed updated scientific literature and guidelines related to antimicrobial resistance and the effectiveness/safety of antibiotics, for example, about the sufficiency of amoxicillin in most bacterial infections of the upper respiratory tract and about the importance of avoiding its association with clavulanate when it is not necessary.11 Such literature was also presented during each meeting, together with local data on antimicrobial resistance and annual reports on antibiotic prescriptions, registered from supply data from community pharmacies. For each health district, the reports showed prescribing variability by cross-comparing NHS prescriptions of each paediatrician in that district and by comparing them with prescribing data aggregated by health district and LHA. Figure 1 shows an example of a prescription report of a health district: prescribing data were represented by histograms relating to various classes of antibiotics, specifically amoxicillin (ATC J01CA04); amoxicillin with clavulanic acid (ATC J01CR02); macrolides (ATC J01FA); cephalosporins (ATC J01DB, J01DC, J01DD, J01DE); overall antibiotics prescription. It was therefore possible to visualise the variability of prescriptions of individual paediatricians and to support discussion among peers, in order to investigate whether the observed variability of prescriptions could be consistent (or not) with the epidemiological context and scientific evidence.

Figure 1
Figure 1

Example of a prescribing report comparing 11 paediatricians (P) of a health district (no. of prescriptions × 1000 assisted children).

All these data and information were then openly presented during district meetings, coordinated by RM and a district pharmacist, explaining some key concepts on the appropriate use of antibiotics in light of scientific evidence and local resistance data, showing prescribing variability and temporal trends observed in the health district for antibiotics prescriptions, and stimulating discussion among peers. Further contacts or face-to-face meetings were carried out to support individual paediatricians, if necessary (eg, in case of emerging prescribing problems). Figure 2 summarises the conceptual framework of the adopted strategy, a type of audit cycle12 nested in theoretical models of behaviour change.13

Figure 2
Figure 2

Framework of the adopted strategy to promote modification of prescribing behaviour for antimicrobial stewardship.

To analyse the impact of this intervention, we retrospectively retrieved data on outpatient antibiotic prescriptions from regional prescription databases, showing supply data associated with NHS prescriptions. We evaluated antibiotic prescriptions within the intervention area (health districts of the LHA of Reggio Emilia) and the corresponding control areas (all the other health districts of LHAs of the Emilia Romagna Region, sharing geographical proximity, a similar epidemiological context and organisation of health services) in the period from January 2013 to December 2019. The units of analyses were the health districts. Differences in prescribing between intervention and control areas were assessed through generalised linear mixed models tailored for longitudinal data analysis. Data were adjusted to take into account prescribing levels during the preintervention period from 2010 to 2012. When evaluating the prescribing data, we assessed the number of prescriptions per 1000 children per year relating to the four antibiotic groups of interest. Furthermore, we examined the ratio of prescriptions of amoxicillin/amoxicillin-clavulanate and amoxicillin/all the other antibiotics considered.

In addition to assessing the absolute impact of the intervention in terms of number of prescriptions, we explored prescribing trends before and after the intervention using a comparative interrupted time series analysis (ITSA). This statistical approach involved fitting models using linear regression with Newey-West standard errors, addressing potential autocorrelation up to a certain lag.14 The preimplementation period covered antibiotic prescriptions from January 2010 through December 2012, while the postimplementation period included data from January 2013 through December 2019. To perform the statistical analyses, we used Stata V.17.0 (StataCorp. 2021. Stata Statistical Software: Release 17. College Station, TX: StataCorp LLC), conducting two-sided tests with a significance level of 0.05 or less.

Patient and public involvement

Not applicable: this was a retrospective assessment of prescription of antibiotics in a 7-year period, aimed at assessing whether information provided to paediatricians could improve prescribing appropriateness.

Results

Figure 3 shows the temporal trends of antibiotic prescriptions in both the intervention and control areas, starting from the baseline period and up to 2019 (the dotted lines indicate the initiation of the intervention period). Figure 4 shows the temporal trends of the ratios of prescriptions of amoxicillin/amoxicillin-clavulanate and amoxicillin/all the other antibiotics considered.

Figure 3
Figure 3

Rate of prescription for 1000 children/year of different antibiotics in the intervention and control areas from 2010 to 2019. Dashed lines indicate the start of the intervention period.

Figure 4
Figure 4

Ratio of prescriptions of amoxicillin/amoxiclavulanate and amoxicillin/all other antibiotics in the intervention and control areas from 2010 to 2019. Dashed lines indicate the start of the intervention period.

At baseline, there was a significant difference in the average prescription levels of amoxicillin between intervention and control areas (+89.66 per 1000 children, p<0.001). Both areas showed a decreasing prescription trend before the intervention, without significant differences in the magnitude of the slopes.

The results of the generalised linear mixed models adjusted for baseline differences, as detailed in table 1, reveal a notable postintervention increase (especially in the immediate postintervention period, mostly maintained in subsequent years) in the rate of prescriptions of amoxicillin in the intervention area vs control area, with an average of 91 more prescriptions per 1000 children per year; and, at the same time, a significant decrease in the rate of amoxicillin+clavulanate prescriptions, with an average of 70 fewer prescriptions per 1000 children per year. These results also led to a significant increase in the ratios between the prescription rates of amoxicillin/amoxicillin+clavulanate (1.4) and amoxicillin/other antibiotics (0.6) in the comparison between the LHA of Reggio Emilia and the rest of the Region. No differences were observed in prescriptions of cephalosporins, macrolides and overall antibiotic prescriptions.

Table 1
|
Differences in prescriptions per 1000 children per year between the intervention and the control areas

Regarding prescribing trends, ITSA demonstrated a consistent decline in overall antibiotic prescriptions after the intervention period in both areas. This trend was consistent for all antibiotics analysed, with the exception of an increasing trend for amoxicillin in Reggio Emilia, with no further differences between the two areas (table 2). Regarding antibiotic prescription ratios, both areas saw an increase in the second analysis period for amoxicillin/amoxicillin-clavulanate and amoxicillin/amoxicillin-clavulanate+cephalosporins + macrolides. However, Reggio Emilia showed a significantly higher rate of increase, compared with the rest of the Region (table 2).

Table 2
|
Differences in linear postintervention trends for prescription of antibiotics between the intervention and the control areas

Discussion

This study adds to existing evidence (from Italian and international studies8–10) confirming the importance of providing prescribing feedback to physicians during group meetings, specifically supporting AMS programmes, together with information on benefits and risks of antibiotic use in specific clinical circumstances and information on local antimicrobial resistance. Our intervention was particularly aimed at promoting the use of amoxicillin in upper respiratory tract infections, due to its excellent profile in terms of effectiveness and antimicrobial resistance rates. In line with this aim, the quality rather than quantity of prescriptions was affected, as amoxicillin use was significantly higher and amoxiclavulanate use lower in the intervention compared with the control area. This effect occurred early, as it was concentrated during the first year after the intervention and was mostly maintained in the following years. In line with this result, there was an increase in the ratios of prescription rates of amoxicillin/amoxicillin+clavulanate and amoxicillin/other antibiotics in the comparison between the LHA of Reggio Emilia and the rest of the Region.

Since the objective of our study was to assess whether our intervention could improve the quality of paediatricians’ prescribing behaviour, we did not have patient data to assess clinical outcomes associated with antibiotic use. In terms of quantity of antibiotic prescribed, a general reduction in antibiotic prescriptions was observed in both the intervention and control areas. However, Di Mario et al15 showed that such a decrease (that they observed in a period between 2005 and 2019) was not associated with a higher rate of complications in the Emilia-Romagna Region. Anyway, we did not observe any difference between the two areas in the overall prescription of antibiotics, so the increased prescription of amoxicillin in the Reggio Emilia LHA was at the expense of broader spectrum antibiotics (in particular amoxiclavulanate).

The observed effects were consistent with available evidence on the potential effectiveness and relevance of audit and feedback interventions to improve prescribing appropriateness.16 Peer discussion allows for a more in-depth analysis of prescribing dynamics than receiving individual feedback, especially in an educational context where busy physicians can receive evidence-based information from scientific literature and laboratory sources, independent of commercial interests, underlining the relevance for public health of careful use of antibiotics.

The visual comparison of the individual prescriptions of the paediatricians using, for each of them, clusters of histograms which reported each type of antibiotic prescribed together with the number of prescriptions, was a key feature of the prescription reports: everyone could compare their own prescriptions with those of each of their colleagues, so that prescribing variability could be assessed within the health district. This approach was crucial in promoting peer discussion within the meetings and should be particularly considered in small-group meetings with prescribers, as one of the features that can help strengthen the impact of audit and feedback interventions.12 If privacy were of concern, data could be anonymised so that each doctor could simply recognise their own data, keeping the interpeer comparison feature which—once again—is essential to appreciate one’s own prescribing behaviour by looking at variability within the health district, therefore providing more informative feedback.

Discussion of the selected scientific literature was, of course, another key element of our approach: academic detailing has long been shown to be effective in promoting appropriate prescribing through peer discussions, comparing one’s approach with evidence from the available literature.8 17 18 Key features of our intervention included targeted clinician education offering support for clinical decision-making and tailoring messages to the local context,19 also making use of local opinion leaders.

Finally, providing laboratory data on local antibiotic resistance was part of the scientific information provided, to offer better contextualisation on how prescribing is related to resistance and how prescribing less-effective antibiotics may affect the health of individual patients. This is also fundamental in AMS programmes,20 as also highlighted by the WHO.21

The described intervention represents an example of using behavioural science to promote AMS.22 The inappropriate prescription of broad-spectrum antibiotics in epidemiological contexts not requiring the use of β-lactamase inhibitors could be referred as the behavioural diagnosis that requires our intervention; the goal of our AMS strategy was to improve paediatrician prescriptions. The mere availability of guidelines promoting AMS does not in itself lead to changes in prescription practices, and such changes may require additional support through academic detailing. Raising awareness of how one’s prescription habits diverge from these guidelines could be a critical step towards improving prescribing appropriateness. Then, peer comparison within an audit and feedback framework offers fertile ground to discuss possible determinants of prescribing variability and its reasonableness, prompting each peer to analyse their own prescribing habits in light of those of their peers, of scientific evidence and of local resistance data, ultimately shaping ‘social norms’ that drive improvements in prescribing behaviour.23 Evidence supporting behavioural science to address antimicrobial resistance is growing,24 and global and national action plans recommend considering it as part of multifaceted strategies to reduce unnecessary antibiotic prescribing.25 Bearing in mind that AMS strategies should be planned to be long-lasting, as their effects may not be maintained in the long term.26

The intervention was implemented during the district meetings and participation of paediatricians was compulsory. This was a key element to ensure its optimal implementation, together with the careful selection of scientific literature and the prompt availability of prescription reports and local resistance data. However, we are aware that our study has methodological limitations (eg, the retrospective and non-randomised design, the use of administrative data and the possibility that similar interventions could have been conducted in the control area) and cannot bring neither ‘hard’ evidence on change on antibiotic prescribing associated with the multifaceted intervention described, nor evidence on its effects on patient outcomes (which was outside the scope of the study). As baseline data show, our intervention area already had a more favourable prescribing profile than the control area before implementing the intervention. However, the use of a geographically related control area sharing a similar epidemiological context, and statistical adjustments for the baseline data, can strengthen our findings: in both areas, overall prescription of antibiotics in paediatrics had decreased over the study period whereas, in the same period, in the rest of Italy no differences can be appreciated.27 The fact that a favourable impact was observed in a region where attention on the appropriateness of antibiotic prescriptions was already higher than in other Italian contexts may suggest that, if implemented in other regions, our intervention could have an even greater impact.

In conclusion, a multifaceted intervention conducted through small group meetings that promoted peer discussions with the support of scientific data reports that highlight prescribing variability and, speaking of antibiotics, local resistance data, improved prescribing appropriateness of antibiotics at district level. Our study adds to existing evidence on the potential relevance of audit and feedback approaches including academic detailing. The elements of our multifaceted intervention are generally applicable in different outpatient settings and we believe that this type of approach should be considered as part of interventions that should be sustained over time to promote rational prescribing and AMS.