Antibiotic Prescription in Febrile Children: A Cohort Study during Out-of-Hours Primary Care

Discussion

In the present study, approximately one in 4 children with fever, whose parents contacted the out-of-hours service, and who were not referred or already used antibiotics at initial presentation, received an antibiotic prescription. All antibiotics were prescribed during face-to-face contact with the family physician: approximately 36% of the children with this type of contact were prescribed antibiotics. In the multivariate analysis, several signs and symptoms were significantly related to antibiotic prescription, providing insight into the family physicians' decision-making process when assessing children with fever. Signs and symptoms, however, explained only a small proportion of the antibiotic prescriptions.

In the present study, the amount of prescribed antibiotics was high, especially when you consider that we excluded 44 children of our analysis who were already using antibiotics, so the total amount of antibiotic use is even higher (ie, 161 of 506 [31.8%]). In comparison with the United States, and other European countries, Dutch family physicians have one of the lowest overall rates of antibiotic prescriptions.^21,27 In The Netherlands, Otters et al¹⁹ investigated the antibiotic prescription in children per family physician contact and found that (in our age-group of ≤6 years) approximately 12% was prescribed an antibiotic. However, their percentage was based on all consultations, not solely on those concerning children with fever, and also included regular office hours. Other studies also described antibiotic prescription rates, but differences in the characteristics of the study population (eg, only children with AOM, not solely febrile children)^{27⇓⇓–30} complicate comparison.

An explanation for the considerable amount of antibiotic prescriptions in the present study could be that only children who contacted an out-of hours service were included. These children may be more seriously ill than those seen during regular office hours and therefore more eligible for antibiotic treatment. However, it seems unlikely that this selection of patients leads to more infections caused by bacteria, because the severity of illness is not solely determined by the causative agent. In addition, because only a small proportion of the antibiotic prescription is explained by the reported signs and symptoms, other considerations may have contributed substantially.

Also, in The Netherlands, children and their families are all registered at one family physician, who usually knows the child from previous visits. However, as a result of the organization of out-of-hours primary care, family physicians in The Netherlands are generally not familiar with the patients they see at the out-of-hours service. Therefore, assessing the patient's expectations and providing adequate safety netting is more difficult. This could make managing the child's disease during out-of-hours service more difficult, leading to a more defensive treatment and, thus, to more antibiotic prescription.

The signs and symptoms that were multivariately related to antibiotic prescription can be clearly related to distinct disease profiles (ie, rhinosinusitis, tonsillitis, AOM). It was not our objective to judge whether the prescribed antibiotics were legitimate for individual cases, but rather to identify and discuss which factors may play a role in the family physicians' decision-making process.

In the multivariate analysis, earache resulting in altered behavioral/sleeping patterns and signs of throat infection are significantly related to antibiotic prescription (OR, 2.58; 95% CI, 1.05—6.33 and OR, 2.53; 95% CI, 1.44–4.43, respectively). These can be related to the disease profiles for AOM and tonsillitis, for which the Dutch guidelines for family physicians have clear recommendations for antibiotic prescription under certain conditions.^6,9 In these guidelines, the rationale for giving antibiotic treatment is mainly based on the possible reduction of duration and severity of illness.^14,15 Therefore, this seems to play an important role in the decision-making process of the family physicians.

In the bivariate and multivariate analyses, both ill appearance and parental concern as assessed by the research nurse were significantly associated with increased antibiotic prescription. Respectively, 47.6%, and 36.1% of the children had received antibiotics. Ill appearance is similar, although not identical, to the physician's opinion that “something is wrong.” This and parental concern have been shown to be important red flags in identifying children with serious infections in primary care.³¹ Therefore, it seems rational to prescribe antibiotics for febrile children who appear to be ill or have concerned parents. Because the home visit was performed after the family physician's consultation, the research nurse might have assessed the child with antibiotics as more severely ill simply because the family physician had prescribed antibiotics. Also, she might have judged the parents as more concerned. However, the research nurses were not aware of the research question; thus, it seems unlikely that this has led to important bias.

Decreased urine production was bivariately and multivariately associated with antibiotic prescription. Although this symptom indicates dehydration, none of the Dutch (or international) guidelines describes dehydration in children as an indication for antibiotic treatment.^24,32 Therefore, it seems that, in at least a subgroup of our patients, antibiotics were not prescribed appropriately.

Children in the age of 3 to 6 months and children with increased rectal temperature were associated with decreased antibiotic prescription. For logistic reasons, we did not include the children who were admitted to the hospital directly after their visit to the out-of-hours service (n = 19). This seems to have introduced some selection bias, leading to these surprising findings. For instance, the children admitted to hospital were younger (median age, 15 months; range, 3–66 months) versus 20 months in the children included in our analysis (range, 3–70 months), and it could be that they also had a higher rectal temperature (data not known) and may have received antibiotics but were not included in our analyses.

A further limitation is that the research nurse noted the patient's history and made the physical examination the day after the patient had contact with the out-of-hours service. We chose this study design, because we did not want to interfere with regular care of the out-of-hours service (especially in case of telephone advice without face-to-face contact). However, the research nurse specifically asked for the signs that were present at time of consultation of the out-of-hours service. In addition, the median time elapsed between time of consultation of the out-of-hours service and our home visit was only 14.5 hours. Given this short delay, it is unlikely that the antibiotic treatment started after the family physician consultation influenced our findings by physical examination, because it generally takes longer to show an effect than the interval we allowed between the family physician consultation and our data collection.³³

Only a small proportion of the antibiotic prescriptions is explained by the related signs and symptoms. Although our study does not allow further exploration of these unknown factors, it does indicate that other, nonmedical factors may influence the family physicians in their decision to prescribe antibiotics. This is in clear contrast with the national and international guidelines that solely base their recommendations to prescribe antibiotics on medical considerations.^{4⇓⇓–7,18} Previous studies have also shown that other (not medically based) considerations may play a role in the family physician's decision to prescribe antibiotics (eg, assuming that the patient or the parents expect antibiotics).^10⇓–12 In our study, we show that a substantial amount of the antibiotic prescriptions is prescribed on the basis of these considerations.