We included patients aged 0–16 years discharged home from the PED with a diagnosis matching any of the infectious diseases contemplated in the ACM guideline. For the purpose of the analysis of the appropriateness of prescription, we treated suspected diagnoses as confirmed diagnoses. We excluded children with infectious diseases that were not included in the protocols or guideline, immunosuppressed patients, patients that received a prescription of antibiotherapy before the visit to the PED and patients that were admitted to hospital.

In the sample size calculation, we determined that we would need 600 patients to achieve our primary objective with a precision of 4% and a level of confidence of 95%. Given that a proportion of appropriate antibiotic prescription at discharge from the emergency department had not been published in the past, we calculated the sample size assuming the worst possible statistical scenario, with an expected proportion of appropriate prescription of 50%. To estimate the number of emergency care episodes required to meet the necessary sample size, we performed an exploratory analysis in a sample of emergency visits managed in the PED on the 13th of each month in the 12 months preceding the start of the study period. We found that 704 patients met the inclusion criteria. Since that selection allowed us to prevent biases associated with seasonal patterns, to randomly include both weekday and weekend visits and to assume losses of up to 15% due to exclusion criteria, we decided to use the exploratory sample in the study.

We analysed demographic variables such as age, sex, weight and the presence of allergies; clinical and treatment variables such as diagnostic category (grouped according to the ACM guideline classification: ear–nose–throat [ENT], respiratory, ophthalmologic, oral, genitourinary, cutaneous or gastrointestinal), prescription of antimicrobial therapy (yes/no), prescribed antibiotic agent, route, dose, interval between doses, duration; prescribing physician (adjunct or resident) and speciality of prescriber (paediatrics, FCM, other).

We analysed demographic variables relative to the total number of patients included in the study, and clinical and treatment variables relative to the total number of episodes that met the inclusion criteria and were considered independent and valid for recruitment, so that different episodes could correspond to the same patient.

The data were collected through the review of electronic health records in the hospital health information system (HP-HCIS®, DXC IT Corporate Solutions).

The primary outcome was the appropriateness of antibiotic prescription based on the ACM guideline and the secondary outcome the appropriateness of prescription based on hospital protocols. To determine the appropriateness of prescription, we applied 3 steps sequentially, so that if one step was found to be inappropriate, we did not assess the steps that followed: indication of treatment (disease warranting antibiotherapy), selection of the appropriate antimicrobial, and dosage (taking into account the dose, interval between doses, duration of treatment and route of administration). If a prescription was inappropriate in any of these aspects, we considered it inappropriate overall. If any of these prescription variables had not been documented, we treated the prescription event as missing data and did not take it into account in the analysis of overall appropriateness. To assess the appropriateness of antimicrobial selection, we considered appropriate selection of the first-line drug and also of second-line drugs in case it was justified (allergy to first-line drug or contraindication of first-line drug) (Fig. 1).

Figure 1.

Algorithm for sequential analysis of the appropriateness of antimicrobial prescription practices. Shaded boxes present the different steps in the assessment of appropriateness.

(0.24MB).

A resident in paediatrics and a resident in pharmacy, both in the 4th year of their residencies, were responsible for the initial determination of the appropriateness of prescriptions, and their conclusions were then cross-checked by 2 specialists in hospital pharmacy and 2 specialists in paediatric emergency medicine. In the case of disagreement, the prescription event was discussed between the senior paediatrics and pharmacy specialists.

The data were processed and analysed with the statistics software SPSS® 21.0 (IBM, Armonk, NY, USA).

We have expressed categorical variables as absolute frequencies and percentages and performed the inferential analysis with a 95% level of confidence. We expressed quantitative variables as median and interquartile range (IQR) due to the skewed distribution of the data.

We compared qualitative data with the Pearson χ2 test and considered p-values of less than 0.05 statistically significant.

We assessed the adherence of prescriptions to the ACM guideline and hospital protocols using the Cohen kappa coefficient, considering agreement poor for kappa values under 0.20; weak for kappa values between 0.21 and 0.40; moderate for values of 0.41–0.60; good for values of 0.61–0.80 and very good for values of 0.81–1.00.16

In the final sample, 55.3% of patients were male (n = 328), and the median age was 2.6 years (IQR, 0.9–4.6 years) (Table 1).

The most frequent diagnostic group was respiratory infection (n = 330; 50.9%), with a predominance of upper respiratory tract infections (n = 238; 36.7%) (Table 2).

Of the total care episodes analysed, 466 (71.9%) were managed by physicians of the department of paediatrics: 350 (75.1%) were residents in paediatrics, and 116 (24.9%) paediatricians. Other physicians involved in care delivery were residents in FCM and other departments, such as paediatric surgery.

In 165 care episodes (25.5%; 95% confidence interval [CI], 22.1%–28.8%) at least one antimicrobial agent was prescribed at discharge: antibiotics in 156 cases, antifungals in 8 and an antiparasitic agent in 1. In most cases, the prescription was for administration via the oral route (81.8%). When it came to antibiotic prescription, penicillins were the predominant group, prescribed in 121 of the care episodes (73.3%): amoxicillin was prescribed most frequently, in 100 care episodes (60.6%), followed by amoxicillin–clavulanic acid in 20 (12.1%) and penicillin v in 1 (0.6%) (Appendix B).

Overall, we found adherence to the ACM guideline in every aspect (indication, selection, dose, dose interval, duration of treatment and route of administration) in 550 of the 625 episodes (88.0%; 95% CI, 85.5%–90.6%) (Table 3). We considered another 23 episodes, all treated with antimicrobials, losses to follow-up due to the absence of data required to assess aspects of antibiotic dosage. Fig. 3 presents the stepwise evaluation of antibiotic prescription appropriateness. In 591 episodes (94.6%), adherence to the guideline was considered adequate, in 3 (0.5%) inadequate due to failure to prescribe antibiotherapy in cases in which it was indicated (submandibular lymphadenopathy, conjunctivitis and animal bite) and in 31 (5.0%) inadequate due to unnecessary antibiotic prescription.

Figure 3.

Appropriateness of antimicrobial prescription in reference to the ACM guideline based on the different assessment steps (indication, selection, dosage). Shaded boxes present inappropriate treatment approaches.

*Overall data, without sequential analysis: some prescriptions were inappropriate in more than one aspect of dosage.

aPrescription considered appropriate; n = 550.

(0.45MB).

We were able to assess the appropriateness of prescription in 142 episodes treated with antimicrobials, with prescription being appropriate in 70 episodes (49.3%; 95% CI, 41.1%–57.5%), and erroneous dosage being the most frequent reason that prescription was inappropriate (Table 4).

In the separate analysis of the indication of antibiotherapy, we found an increased probability of unnecessary antimicrobial prescription in the group of patients under 2 years (16/59 episodes [27.1%] compared to 15/106 episodes [14.2%] in older children and adolescents).

We found that overall, prescription practices adhered to the guideline in 85% of cases in paediatricians and residents in paediatrics or FCM, a level of adherence that was significantly greater compared to physicians in other specialities. The adherence was substantially lower in the subset of episodes treated with antimicrobials, without differences based on the type of prescriber (Table 5).

Tables 3 and 4 summarise the results concerning the appropriateness of antimicrobial prescription by diagnosis. We found that inappropriate prescription practices were most frequent in the management of acute otitis media (AOM), especially in the 0–2 years age group (32.4% of appropriate prescription in these patients vs 52.2% in the overall sample), conjunctivitis and urinary tract infection (UTI) (Table 3).

Table 4 presents the distribution of inappropriately treated episodes by reason prescription was inappropriate (indication, selection or dosage). Conjunctivitis and AOM stood out for the substantial proportion of cases treated with unnecessary antibiotherapy based on the ACM guideline (60.0% and 31.7%, respectively). Three cases of viral conjunctivitis were treated with tobramycin in the absence of manifestations suggestive of bacterial superinfection. In the case of AOM, there was a particularly high frequency of inappropriate prescription in patients aged 0–2 years (13/32 cases; 40.6%) compared to patients aged 2 or more years (6/28; 21.4%).

In the 6 cases of AOM with inappropriate prescription due to selection, the error was the prescription of an agent with a spectrum too narrow for empirical therapy in patients aged 6–24 months that required amoxicillin–clavulanic acid due to severe symptoms, recurrent AOM o previous treatment with amoxicillin in the past 30 days. In the case of patients with conjunctivitis, selection errors occurred in 2 infants under 12 months treated with tobramycin, when the guideline recommended prescription of erythromycin or aureomycin, and in cases of UTI, in 2 patients treated with cotrimoxazole instead of fosfomycin.

When it came to dosage, the diseases in which prescription was inappropriate most frequently were UTIs, with 3/7 inappropriate prescriptions (42.9%). In 2 of these cases, fosfomycin sodium, which was indicated, was prescribed to be administered every 6 h, when the recommended dose interval in the guideline is 8 h. There were 6 cases with inappropriate dosage among patients with AOM (22.9%), in 4 cases due to prescription of treatment with amoxicillin–clavulanic acid for too short a duration, and in the rest due to prescription of an inappropriate dose. We did not assess the appropriateness of dosage in cases of conjunctivitis, as the ACM guideline did not specify a recommended duration of treatment for this disease.

We found that hospital protocols applied to 372 of the 648 episodes (57.4%), and were able to analyse the appropriateness of prescription in 353 (19 episodes lost due to missing data). In this subset, we found an overall frequency of appropriate prescription in every considered aspect of 89.0%. Of all these episodes, 236 (66.9%) were not treated with an antimicrobial, which was the correct approach in 232 episodes (98.3%), while antibiotherapy was indicated but not prescribed in another 4 episodes: 3 of conjunctivitis and 1 of pharyngitis.

Analysing only the 117 treated episodes (33.1%) using hospital protocols as reference, we found that 82 (70.1%) had been managed correctly (Table 3). Errors were distributed uniformly between indication, drug selection and dosage (Table 4). We found a marked proportion of inappropriate prescription in the management of UTIs (57.1%), with poorer adherence relative to hospital protocols versus the ACM guideline because the hospital protocol called for administration of more than 1 dose of fosfomycin trometamol. Contrary to this, prescriptions were for a single dose, as recommended by the ACM guideline.

We analysed the agreement between adherence to the ACM guideline and to hospital protocols, and found a good overall agreement (κ = 0.673). The agreement was complete for the most frequent infectious diseases (acute gastroenteritis, laryngitis and bronchiolitis) (κ = 1), and very high in the approach to acute pharyngitis/tonsillitis (κ = 0.915). It was good in the management of pneumonia (κ = 0.765), moderate for AOM (κ = 0.586) and weak for UTIs (κ = 0.308), and there was no agreement when it came to conjunctivitis (κ = 0.000).