Objective: The purpose of our study was to correlate the results of the radionuclide salivagram with the corresponding chest radiography findings on patients being evaluated for salivary aspiration to determine the utility of the salivagram.
Materials and methods: We identified 222 patients younger than 21 years who underwent salivagram and chest radiography within 3 months of each other. Salivagrams were blindly interpreted by two readers and chest radiographs were blindly interpreted by two other readers. The kappa coefficient with 95% CI was used to measure the level of interobserver agreement. Multivariate logistic regression was applied to determine whether age, sex, and neurologic diagnosis were predictors of a positive salivagram, with the odds ratio used to estimate association.
Results: Interobserver agreement on salivagram interpretation was excellent (κ = 0.988; p < 0.0001; 95% CI, 0.968-1.000). Interobserver agreement on chest radiography interpretation was excellent (κ = 0.905; p < 0.0001; 95% CI, 0.845-0.965). The salivagram was positive for aspiration in 55 patients (25%). Chest radiography was positive in 54 patients (24%). When the interpretations of the salivagram (normal or abnormal) were compared with interpretations of the chest radiograph (normal or abnormal), there were 213 agreements and nine disagreements (intermethod agreement κ = 0.891; p < 0.0001; 95% CI, 0.831-0.952). Independent of age (p = 0.80) and sex (p = 0.31), patients with a neurologic diagnosis had odds of a positive salivagram 5.6 times higher than other diagnoses (odds ratio = 5.6; 95% CI, 2.5-13.1; p < 0.0001).
Conclusion: Infants with abnormal findings on salivagrams also had a high rate of abnormal findings on chest radiographs, which may indicate that some of the lung disease may be due to aspirated saliva. Salivagrams may be useful in children at risk of aspiration to identify those in whom intervention may help minimize the consequences of aspiration.