Appropriateness of Testing for Serious Bacterial Infection in Children Hospitalized With Bronchiolitis

Discussion

The results of the current study reveal a low incidence of SBI in children hospitalized with bronchiolitis. The only SBI we identified was UTI (2.3% of patients). There were no cases of bacteremia or meningitis. Due to the fact our study was designed to evaluate provider practices and excluded patients admitted to the PICU, the incidence of SBI reported in this article may not be a true reflection of the overall incidence of SBI in patients with bronchiolitis. Nevertheless, our findings are similar to previous studies that have shown a low incidence of concurrent SBI in patients with bronchiolitis.^3–14 A recent literature review reported on occult SBI in young infants with bronchiolitis.¹¹ In the review of 11 included studies, the weighted rate of UTI was 3.3% (95% confidence interval: 1.9–5.7). There were no cases of bacteremia in 8 of the 11 studies and no cases of meningitis in any of the studies. This evidence further supports our findings of the potential low yield of SBI testing in patients with bronchiolitis.

Concern for SBI in the setting of clinical bronchiolitis may be higher for those patients in high-risk age groups such as younger children or children with a history of prematurity, relative immune immaturity, and susceptibility to perinatal pathogens. We found that a significantly larger percentage of the total patient population undergoing SBI testing were aged <60 days, and in an adjusted multivariate analysis, age <60 days was associated with SBI testing. Interestingly, we found that of those patients undergoing SBI evaluation, only 12.8% and 23.9% were in the <30-day and 31- to 60-day age groups, respectively; thus, ∼63% of patients with bronchiolitis undergoing testing were >2 months old.

Despite knowledge that UTI is the most common concurrent SBI in bronchiolitis, we found that a larger percentage of patients undergoing SBI evaluation were evaluated with a blood culture. Many of these patients were only tested for bacteremia without any urine testing. The fact that all of the blood cultures in this study were contaminated further suggests blood culture testing has low utility. In addition, increased false-positive findings have the potential to lead to other adverse consequences, which have been demonstrated in other studies. Antonow et al¹² found that 142 of 282 patients aged 0 to 60 days with bronchiolitis underwent a sepsis evaluation, with a resulting 31 positive culture results. Twenty-seven of those cultures were found to be contaminants (4 CSF, 5 blood, and 18 urine). This outcome was associated with increased antibiotic use, LOS, and costs. Purcell et al⁵ also found that of the 2396 children they studied who were hospitalized with RSV, 12 had positive blood culture results, all of which were found to be contaminants. Although determining the reasoning behind provider practices in diagnostic testing is beyond the scope of the current study, we speculate that when blood is collected at intravenous placement or other diagnostic testing is being sent, providers often order a blood culture without fully thinking through the ramifications of a false-positive finding.

In addition, in evaluating provider practices in SBI testing, we found that approximately one-quarter of patients had only a CBC obtained. As stated earlier, CBC testing was included in our definition of SBI testing because it is most likely that this diagnostic test was used for the evaluation of white blood cell count as a marker of potential bacterial infection. It is unlikely that an appreciable rate of provider concern exists for anemia and/or thrombocytopenia in patients presenting with clinical bronchiolitis. CBC testing alone would not be a thorough evaluation for SBI, therefore it could be considered low utility when done in isolation. This finding again highlights additional low-yield diagnostic testing being performed for patients with bronchiolitis.

Even without positive results, SBI testing is associated with increased antibiotic use. In our study, more than one-half of patients undergoing any SBI testing received antibiotics at some point during their hospitalization compared with 19.1% of those not undergoing testing. Although we did not quantify duration of antibiotic use and included antibiotic doses administered in the emergency department, a previous study by Purcell et al⁵ found that 97% of antibiotics started on admission in patients hospitalized with RSV were continued until discharge. Antibiotics are not a benign treatment; they have potential adverse effects and risk of potentiating resistance patterns.

Our study also found that children undergoing SBI evaluation had significantly longer LOS than those who did not. Because of our retrospective study design, we are unable to determine if there was a causal nature between SBI evaluation and LOS. The increase in LOS, however, may be partially explained by a prolongation of a patient’s hospitalization while microbiologic cultures are observed for growth. Given the average difference in LOS of 1.1 days and a daily hospital cost of ∼$2900 (per hospital billing data), SBI testing may result in as much as $3190 in excess costs per hospitalization. This translates to $1.8 million in costs for the cohort in adjusted 2013 dollars. Reduction of unnecessary SBI testing in this group may therefore result in significant cost savings.

A limitation of our study includes the retrospective nature of the analysis. Although we used both primary and secondary International Classification of Diseases, Ninth Revision, discharge codes for inclusion criteria, we run the risk of underreporting the incidence of SBI because they may have been coded differently once the diagnosis was made. It is also possible that SBI testing may have been overestimated. One potential scenario includes infants aged 0 to 60 days who may have been admitted initially for fever, who underwent neonatal fever evaluation (including SBI testing), and were subsequently found to be RSV-positive; these infants thus carried that discharge diagnosis. A large percentage of the patients in the study were also seen by an outside provider before admission. In these cases, we did not have those outside medical records to review. In addition, we cannot speak to the role of severity of illness as a confounder in the association between SBI evaluation and management of these patients. Strengths of the study include the large patient population and the multicenter study design.