Secondary prevention of severe pneumonia in neurologically impaired children: a comparative effectiveness study

Background: Children with neurologic impairment (NI) have increased risk of pneumonia compared to other children. Pneumonia prevention strategies for children with NI are largely based on expert opinion with a limited evidence base. The objective of this study is to conduct a comparative effectiveness study of pneumonia prevention strategies in children with NI.
Methods: The California Children’s Services database contains complete capture of inpatient, outpatient, and pharmacy use for the largest Title V program in the United States. Our study population included children with NI and at least one pneumonia hospitalization. We excluded children with cystic fibrosis and whose only pneumonia hospitalization occurred during the index birth hospitalization. The primary outcome variable was subsequent pneumonia hospitalization. Exposure variables included: dental care, oral secretion management, gastric acid suppression, gastrostomy tube placement, chest physiotherapy, outpatient antibiotic prior to initial pneumonia hospitalization, clinic visit prior to initial pneumonia hospitalization, and clinic visit within 30 days after discharge for initial pneumonia hospitalization. Our primary analysis used a 1:2 propensity score matched model. Covariates for propensity score generation included sociodemographics (i.e. age, gender, race/ethnicity, type of CCS insurance) and characteristics of the initial pneumonia hospitalization (i.e. length of stay, ICU admission, pleurisy, intubation). Secondary analyses included a traditional logistic regression model using propensity score covariates and an inverse probability treatment weighted model.
Results: 3632 children with NI were included in our analysis. 1362 (37.5%) had a subsequent pneumonia hospitalization. Children had increased risk of subsequent pneumonia hospitalization if they had new gastrostomy tube placement (OR 3.09, CI 2.34-4.09), had a clinic visit within 7 days prior to initial hospital admission (OR 1.33, CI 1.10-1.61), or had a clinic visit within 30 days after discharge from initial hospital admission (OR 1.85, CI 1.45-2.35). No exposures were associated with reduced risk of subsequent pneumonia hospitalization. Similar patterns persisted in our traditional logistic regression and inverse probability treatment weighted analyses.
Conclusions: None of the pneumonia prevention strategies we evaluated were associated with decreased risk of subsequent severe pneumonia. Further validation of our findings is needed in other populations due to the risk of residual confounding by indication but a randomized control trial is needed for definitive assessment. The strongest association of gastrostomy tube with increased risk of severe pneumonia warrants particular attention given the common practice of gastrostomy tube placement in children with NI.