The main outcome of our study has been a significant reduction in antibiotics prescription rates among children who undertook a rapid nasopharyngeal test for IAV, IBV and RSV, which allowed for a correct differential diagnosis with respect to bacterial RTIs. In general, the prescription of antibiotics increased with increasing number of symptoms, and was more common in the presence of pharyngitis, lower respiratory tract symptoms, vomiting, diarrhea and hospitalization. It also increased in case of loss of appetite, but this latter result is likely to stem from a disproportion of this symptom between the two groups. As often is very difficult to distinguish between bacterial and viral RTIs, swab molecular testing has been extremely useful in limiting the improper use of antibiotics by distinguishing among the two etiologies. Considering the reality our hospital deals with, the probability of treating an “uncertain” RTI case with antibiotics was 9 times greater when swab molecular testing was not yet available than when it became available. This finding is strengthened by considering that we strictly used the same symptoms and selection criteria to define an infant or child with “non-obvious causes of RTIs” among our study population and that symptoms were equally represented in the two Groups (Table 1). As indicated by the univariable and multivariable logistic regression (Table 2), given the same symptoms, non-swabbed infants were more likely to receive antibiotics.
The respiratory system, despite many specific and non-specific defense mechanisms, is particularly vulnerable to infections, reason why respiratory diseases are among the most commonly encountered in pediatrics. RTIs represent one of the major reasons for seeking care in the pediatric age [1]. The management of these issues touches upon several controversial aspects, and despite the numerous guideline indications, the approach of pediatricians to care is still heterogeneous. The etiological agents involved in RTIs are numerous and often cause reinfections within the same winter season. Among bacteria, commonly encountered pathogens are Streptococcus pneumoniae, Haemophilus influenzae, Mycoplasma pneumoniae and Chlamydia pneumoniae. Among viruses, the most frequent etiological agents are represented by IAV and IBV and, especially in infants and young children, by RSV. Despite the etiopathology of the infection remains often uncertain, and although antibiotic treatment is not indicated except in high-risk subjects, their use is extremely common in medical practice [1,2,3]. Penicillin combinations represent the most used class (21.6%), followed by macrolides (12.1%) and cephalosporins (11.8%), in accordance with the guidelines for the treatment of common pediatric infections. It has been hypothesized that as much as 50% of antibiotic prescriptions in children are not actually needed [4]. Consequently, in the recent years, the excessive and sometimes improper use of antibiotics has resulted in the rise of resistant bacterial strains. The issue of antibiotic resistance is being increasingly considered worldwide and requires, according to the WHO, a shift towards a more rational and congruous use of these drugs. This warning holds especially true considering that current research in antibiotic resistance has essentially reached a standstill in the last years [5, 6]. Antibiotic resistance is a growing threat, carrying high costs and serious consequences for the community, such as failure to respond to treatments, prolonged illness and an increased risk of complications [7]. In the European rankings describing the distribution of bacterial resistance across the continent, Italy solidly ranks first along with Greece [5, 8].
Four different types of Influenza viruses have been identified. Of these zoonotic pathogens, only IVA and IVB can spread to people and are hence responsible for seasonal influenza epidemics [9]. One of the characteristics of these viruses is genetic instability. Minor mutations (genetic drift) consist of a gradual variation in the sequence of amino acids making up their surface proteins, thus allowing the virus to evade humoral immunity (i.e., antibodies) in previously exposed subjects. These mutations are responsible for seasonal epidemics [10,11,12,13]. Bronchiolitis is a seasonal respiratory disease, too. According to the estimates provided by the WHO, it is the most frequent lower RTI in children aged < 1 year and one of the most common causes of hospitalization [14, 15].
It is not often possible to clinically distinguish viral pneumonia (especially if caused by Adenovirus) from the disease caused by Mycoplasma pneumoniae and other bacterial pathogens [16], yet the differential diagnosis, based on the identification of the possible etiological agent, is of fundamental importance in order to plan a correct therapeutic strategy.
In fact, despite the different etiologies, the phenotypical similarities brought by viral and bacterial infections can make a differential diagnosis quite challenging if that is solely based upon symptomatology. This holds especially true in case of community-acquired pneumonia, where current serological biomarkers are inefficient to specify the etiology of the infection. Indeed, various biomarkers, including C-Reactive Protein, procalcitonin, lipocalin-2 and tumor necrosis factor-related apoptosis-inducing ligand have been evaluated for their ability to differentiate these pneumonia etiologies. For many of these biomarkers, values differ in children with bacterial compared with viral causes of pneumonia, but the reliability of these tests is not sufficiently high to justify routine use. Studies of these biomarkers have also been hampered by the lack of a gold standard to determine pneumonia etiology and the relatively frequent occurrence of viral-bacterial co-infections [16].
Therefore, the detection of viral nucleic acids gives a major contribution to the diagnostic evaluation of children with respiratory infections [1]. Currently, several molecular tests are available for the detection of viral RNA, either relying on conventional RT-PCR (Reverse Transcriptase-Polymerase Chain Reaction) or Real Time RT-PCR methods [17,18,19,20,21,22]. Those methods are sensitive and specific; however, they are not indicated in every clinical context. On the contrary, the methodology we used of in our study, ID NOW Influenza A and B and ID NOW RSV tests, ensures the differential and qualitative detection of IAV, IBV and RSV from nasal and nasopharyngeal swabs, with many advantages. First, the speed (approximately 15 min) and simplicity of execution is remarkable. Second, the nucleic acid detection can be requested by medical practitioners at any time, during both morning and night shifts, enabling its essential application in emergencies. Third, by not requiring specific technical skills, this test is within reach for virtually any laboratory. Lastly, the sensitivity and specificity of the tests are high. Sensitivity of 96.3% and specificity of 97.4% for IVA, sensitivity of 100% and specificity of 97.1% for IVB and sensitivity of 98.6% and specificity of 98% for VRS [23, 24].
The rapid test is undoubtedly very useful for diagnostics and therapy, but also is in the event that an epidemic outbreaks in confined areas, such as in hospitals, nursing homes and schools. Therefore, it can also serve the purpose of providing epidemiological information to national and international health surveillance authorities.
Our results showed a marked and statistically significant reduction in the number of antibiotic therapy prescriptions in the group of children getting a swab compared to the group not getting any. This result demonstrates how useful can the test be in allowing for a certain diagnosis, and how much, in the absence of the latter, defensive medicine may influence the conduct of a doctor.
From a cost-effectiveness point of view the following considerations can be made. The overall antibiotic prescription rate was 85.0% (148) in Group 1 and 38.2% (81) in Group 2. According to unpublished data taken from our Hospital Farmacy service, the average cost of an oral antibiotic therapy for 7 days is 1.5€ for a child ≤6 years old and 3€ for a child > 6 years old, independently on the type of antibiotic; the average cost of an intramuscular/intravenous antibiotic therapy for 7 days ranges from 3.2€ for aminoglycosides to 3.78€ for cephalosporins and up to 21€ for amoxicillin with clavulanate in children ≤6 years old and from 6.02€ for cephalosporins to 18€ for aminoglycosides and up to 31.5€ for amoxicillin with clavulanate in children > 6 years old. We calculated a total antibiotic cost of 1011€ for Group 1 and of 634€ for Group 2 based upon the age of included patients, antibiotics used and their way of administration. The cost of one rapid test (of the type considered in our study) is 22€, for a total amount of 9.328 € for Group 2.
Concerning the effectiveness of the proposed tests, we could not carry out a thorough benefit analysis in terms of long-term reduction of antibiotic resistance, as it requires complex statistical models and goes beyond the purpose of our manuscript. Yet, it is plausible to assume that a judicious administration of antibiotics can only translate into important economic and health advantages. Economic advantages come from the fact that as resistance develops and more specific antibiotics are needed, their price tend to rise accordingly. Health advantages come from the overall attenuation of antibiotic resistance and the consequent reduction in infections caused by hard-to-treat multi-resistant “bugs”.
All in all, although medical expenses were higher in Group 2, the cost-effectiveness of the rapid tests is high, as the benefits related to the reduction of antibiotic prescriptions greatly outweighs the higher cost of the laboratory tests when compared to the antibiotic therapy itself.
Finally, in a retrospective study like ours, some limitations need to be highlighted. We recognize that the percentage of antibiotic therapy in Group 1 is pretty high and consequently it might be supposed that there were more bacterial infections in this Group. However, as indicated in the Methods section, baseline symptoms were equally distributed in the 2 Groups. In addition, study population selection criteria were strictly similar, making this etiological imbalance between Groups unlikely. We believe that the high antibiotic prescription rate in Group 1 is rather attributable to the subjectivity of the operator, a lack of written protocols at the time of study, and the characteristics of the selected population. The intrinsic diversity of health professionals who have examined the children without following specific protocols and prescribed the therapies evaluated in the study may affect the results: the modus operandi of the training schools of origin, the experience and the personal beliefs of every single doctor can inevitably influence the final outcome. Moreover, taking into account the characteristics of our ED population with more than 50% of users displaying social fragility (e.g. foreigners, gypsies, low income families, etc.), a “generous” administration of antibiotics is justified by the difficulties related to children follow-up and the generally low compliance of the families. The same considerations justify the high percentage of hospitalization in both groups. In addition, it should be considered that the population seeking care in our ED generally has a poor cultural background. Lack of general knowledge in medical matter may induce families to reach out to the ED only when their children are already in more compromised conditions.