Student University of Utah School of Medicine Salt Lake City, Utah, United States
Background: Previous studies of children with neurologic impairment (NI) utilizing paid claims data suggest that most expert-recommended pneumonia prevention strategies are not associated with decreased risk of pneumonia hospitalization. Objective: This study assesses the validity of prior findings by adapting those models to a clinical dataset. Design/Methods: Population: Patients aged ≤21 years with NI and ≥1 pneumonia hospitalization seen at Primary Children's Hospital between June 1, 2008, and May 31, 2023. Primary Outcome: Subsequent hospitalization for community-acquired or aspiration pneumonia at least 30 days after discharge from index pneumonia hospitalization. Exposure variables: Receipt of preventive pneumonia treatment: routine dental care, gastric acid suppression, gastrostomy tube placement, chest physiotherapy, surrogate markers of a medical home (outpatient antibiotics before index hospitalization, clinic visit before or after index hospitalization), and flu vaccine. Analysis: Logistic regression model with propensity scored-based inverse probability treatment weighting. Model adjusted for sociodemographics, medical complexity, and characteristics of the index hospitalization. Results: Among n=3395 children with NI and index pneumonia hospitalization, 2540 (75%) had subsequent pneumonia hospitalization. Mean age was 5.4 (SD 5.7), 45% female, 41% with public insurance, and 83% White. Children with a subsequent hospitalization were younger (p=0.002) and more likely to be on public insurance (p=0.001). The most commonly utilized pneumonia prevention strategies were chest physiotherapy (31%) and g-tube placement (27%). Children with a subsequent hospitalization were more likely to have documented dental care (p < 0.001), gastric acid suppression (p < 0.001), g-tube placement (p < 0.001), chest physiotherapy (p < 0.001), and seasonal flu vaccine (p < 0.001). Adjusted analyses showed significantly increased odds of pneumonia rehospitalization for those who received gastric acid suppression (aOR of 2.32, 95% CI 1.53-3.52) and clinic visit ≤ 30 d after index hospitalization (aOR of 1.79, 95% CI ,1.16-2.76), while g-tube placement (aOR of 0.36, 95% CI 0.20-0.65) showed significantly decreased odds of pneumonia rehospitalization.
Conclusion(s): Our findings were inconsistent with prior studies, requiring further examination of the efficacy of recommended pneumonia prevention strategies. The inconsistency with prior studies and increase in treatments associated with increased risk of pneumonia prevention strategy suggests the model needs further refinement for additional potential confounders by indication.