Variation in Bacterial Respiratory Culture Results in Children With Neurologic Impairment

This multicenter retrospective cohort study included hospitalizations in Pediatric Health Information System Plus (PHIS+) (Children’s Hospital Association, Lenexa, KS). PHIS+ augmented Pediatric Health Information System data with detailed laboratory, microbiologic, and radiographic results from 6 children’s hospitals between 2007 and 2012.^10,11  One hospital was excluded because of data quality concerns.

Hospitalizations for ARI in which a bacterial respiratory culture was obtained within 2 days of admission for children between 1 and 18 years of age with NI were included. NI was defined as a chronic neurologic diagnosis likely to result in substantial functional impairments.^12–14  Hospitalizations for ARI were identified by primary diagnosis codes within the Clinical Classifications Software (Agency for Healthcare Research and Quality, Rockville, MD) respiratory group indicative of ARI.¹⁵  Hospitalizations for children transferred from other hospitals and for children with HIV, with tuberculosis, or receiving chemotherapy were excluded.

We examined the first bacterial respiratory culture obtained. Cultures included specimens from bronchoalveolar lavage, endotracheal tubes, sputum, and tracheal aspirates. We excluded cultures identified as mouth, nasal, pharyngeal, and throat specimens, including saliva, to decrease likelihood of oral flora contamination. Microbiology laboratory guidelines were applied to all respiratory culture results to standardize across hospitals,¹⁶  which omitted organisms consistent with oral flora (eg, viridans group streptococci) and nonspeciated organisms (eg, Gram-negative rods). Results of bacterial respiratory cultures were classified as negative or positive with organisms identified.

To characterize the study population, patient demographics and medical complexity by using complex chronic conditions (CCCs)^17,18  were examined. We calculated the prevalence of positive respiratory culture results and the prevalence of each organism identified with exact binomial 95% confidence intervals. Using Rao-Scott χ² tests to account for within-hospital and within-patient clustering, we compared the prevalence of positive culture results and of the prevalence each organism across the 5 hospitals and across age groups (1–5, 6–12, and 13–18 years). Subsequently, we examined cultures in subanalyses of children with and without tracheostomies.^18,19  P values <.05 were considered statistically significant. This study was approved by the institutional review board.

A total of 693 of 4900 (14.1%) hospitalizations of 485 children with NI and ARI had a bacterial respiratory culture obtained within 2 days of admission. Most cultures examined were from hospitalizations of children who were male (54.8%) and White (70.4%) and who had ≥3 CCCs (73.2%), had tracheostomies (54.7%), and were in the ICU at the time of culture collection (51.5%). Significant differences in several of these patient characteristics were noted across hospitals (Table 1).

A total of 377 (54.5%) of 693 bacterial respiratory culture results were positive, revealing 19 different organisms. More than one-quarter (28.1%) were polymicrobial. The most common organisms were Pseudomonas aeruginosa (34.2%), Staphylococcus aureus (all strains, 20.6%; methicillin resistant, 5.1%), Serratia marcescens (9.4%), Haemophilus influenzae (8.5%), Streptococcus pneumoniae (7.9%), and Moraxella catarrhalis (6.2%). In the subanalysis of the 379 cultures from hospitalizations of children with tracheostomies, the results of 242 (63.9%) were positive. The most common organisms were P aeruginosa (42.0%), S aureus (all strains, 22.4%; methicillin resistant, 5.3%), S marcescens (14.8%), S pneumoniae (10.8%), H influenzae (9.8%), and Stenotrophomonas maltophilia (9.8%). In the subanalysis of the 314 cultures from hospitalizations of children without tracheostomies, the results of 135 (43%) were positive. The most common organisms were P aeruginosa (24.8%), S aureus (all strains, 18.5%; methicillin resistant, 4.8%), H influenzae (7%), S pneumoniae (4.5%), M catarrhalis (3.5%), and group B Streptococcus (3.2%).

The prevalence of positive respiratory culture results ranged from 18.6% to 83.2% across the 5 hospitals (P < .001). Polymicrobial culture growth ranged from 1.8% to 54.1% (P < .001). In Fig 1A, we illustrate the significant variation in organism prevalence across hospitals, with the exception of S pneumoniae. In the subanalysis of children with tracheostomies, the prevalence of positive respiratory culture results ranged from 21.4% to 87.5% (P < .001); the most common organisms had significant variation across hospitals (Fig 1B). In the subanalysis of children without tracheostomies, the prevalence of positive respiratory culture results ranged from 14.5% to 70.0% (P < .001). There was significant variation across hospitals for the most common organisms, except S pneumoniae and M catarrhalis (Fig 1C).

The prevalence of positive respiratory culture results was similar across ages; prevalence was 60.3% for 6 to 12 years, 54.0% for 1 to 5 years, and 47.5% for 13 to 18 years (P = .141). Polymicrobial growth ranged from 22.8% to 31.4% (P = .203). In Fig 2A, we illustrate no significant variation in prevalence for the most common organisms across age groups, except M catarrhalis. In the subanalysis of children with tracheostomies, the prevalence of positive respiratory culture results ranged from 57.1% to 70.4% (P = .255). The most common organisms identified and lack of variation of prevalence across age groups paralleled that of the entire cohort, except for H influenzae (Fig 2B). In the subanalysis of children without tracheostomies, the prevalence of positive respiratory culture results ranged from 41.4% to 45.4% (P = .906). Of the most common organisms, only M catarrhalis had significant variation in prevalence across ages (P = .016; Fig 2C).

In this multicenter study, the results of just over half of the bacterial respiratory cultures obtained from children with NI who required hospitalization for ARI were positive. P aeruginosa and S aureus were the most commonly identified organisms across hospitals and age groups and in patients with and without tracheostomies. There was significant variation in the prevalence of positive respiratory culture results, polymicrobial growth, and specific organism growth across hospitals but not age groups.

In our study, the 2 predominant pathogens of bacterial ARI in otherwise healthy children, S pneumoniae and H influenzae,^20,21  were among the commonly identified organisms; however, the higher prevalence of P aeruginosa and S aureus identified in this study suggests a different epidemiology of bacterial ARI in children with NI compared with otherwise healthy children. Children with NI experience frequent hospitalizations and frequent antibiotic use and are often dependent on technology, all of which are known to increase antibiotic resistance, colonization, and infection risk with typical and atypical organisms.^19,22–28  As these exposures increase with time in children with NI, we hypothesize that older children and those with tracheostomies will be more likely to have positive culture results. However, positivity rates and organism prevalence remained similar across age groups. Children with NI and tracheostomies had higher rates of positive culture results. The most commonly identified organisms in children with and without tracheostomies were P aeruginosa and S aureus. Although it is possible these organisms could represent chronically colonizing bacteria rather than the bacterial cause of ARI, colonizing bacterial flora can become pathogenic sources of infection^29,30  and are often treated as such given the lack of criteria or testing to distinguish between acute and chronic organisms on respiratory cultures. Hence, although our data support some differences in bacteriology when tracheostomy is present, we believe coverage of P aeruginosa and S aureus should be considered for all children with NI hospitalized with concern for bacterial ARI. Larger studies are necessary to better elucidate both the epidemiology and outcomes of ARI in this population, including treatment strategies and the comparative effectiveness of those antibiotic strategies.

Variation in the rate of positive respiratory culture results and in prevalence of isolated organisms during ARI may be related to differences in populations among the hospitals and local environmental factors (eg, infection control policies, local epidemiology). Similar institutional variability of infectious epidemiology has been described for ventilator-associated pneumonia^31–33  and bacteremia³⁴  in adults. In our study, differences in patient characteristics across hospitals (eg, demographics, degree of illness as noted by the ICU at the time of culture collection, and tracheostomy presence) likely contribute to some of the variation in culture results across hospitals. However, the majority of the most common organisms continued to have significantly different prevalence across hospitals when analyzing the hospitalizations of children both with and without tracheostomies. Furthermore, patient complexity, as measured by CCC count, was similar across hospitals with higher and hospitals with lower positivity rates.

Another factor that may contribute to the respiratory culture variation noted in this study was microbiology reporting practices at the 5 hospital laboratories. Because we did not complete a formal assessment of the reporting practices at the various institutions, we can only speculate on the differences on the basis of the culture results reported and our own institution protocols for respiratory culture reporting. Several hospitals appeared to use stringent laboratory reporting protocols, in which clinically nonsignificant and nonpathogenic results were not reported (eg, P aeruginosa plus the terminology “oral flora”), whereas other hospitals seemed to report most or all organisms isolated (eg, P aeruginosa, viridans group streptococci, and Gram-negative rods), regardless of growth quantity or clinical significance. In this study, we attempted to limit the effects of these noted differences by standardizing results across hospitals with the application of laboratory guidelines. However, reporting variation across institutions remains a challenge in studying respiratory cultures and must be considered in when designing and evaluating results of future multicenter studies.

This study has several limitations. Given the retrospective and descriptive study design using limited data included in PHIS+, residual confounding (including previous cultures or nursing home resident status) likely exists. Inclusion of these variables would unlikely change the absolute prevalence of positive culture results. PHIS+ data are limited to 2007–2012; it is possible that practice change in obtaining respiratory cultures over the last decade could alter the frequency of culture positivity or the frequency of specific organisms identified. However, published national guidelines for the diagnosis of ARIs continue to exclude children with NI.^20,35  Although there is growing awareness of diagnostic stewardship in the context of high-value care and it is possible that individual practitioners have altered their behaviors over the past decade,^36,37  we are unaware of any institutional-level efforts or literature revealing practice change in this population. To limit the inclusion of hospital-acquired infections, we only examined respiratory cultures obtained within the first 2 days of hospitalization. In addition, because of limitations in the database regarding viral studies, we were unable to analyze possible viral coinfection. Thus, our results may not reflect the entirety of ARIs in children with NI.

In this multicenter cohort of hospitalizations for ARI in children with NI, there was significant variation in prevalence of positive culture results and organisms across hospitals. P aeruginosa and S aureus were the most commonly identified organisms during ARI, regardless of a child’s age or tracheostomy dependence. Children with NI may have different bacterial causes of ARI compared with otherwise healthy children; further studies are necessary to better understand the comparative effectiveness of antibiotic therapy strategies in the context of respiratory culture organism growth.