Identification of hospitalized tracheostomy and ventilator dependent patients in administrative data

BACKGROUND: Patients with established tracheostomy and ventilator dependence are hospitalized more frequently and use more healthcare resources than other patients with complex chronic conditions. However, data to compare variation in hospitalization and resource use among patients in this population across the United States is deficient, partly due to the lack of structured methods to query national databases. AIM: Determine the best method for identifying the subset of children with established tracheostomy and ventilator dependence in Pediatric Health Information System (PHIS). HYPOTHESIS: A combination of identifiable characteristics coded in the PHIS database can be used to identify the population of patients with established tracheostomy and ventilator dependence who are admitted to the hospital. METHODS: This cross-sectional, retrospective cohort study used established methods to extract data from PHIS and assessed the sensitivity and specificity of an algorithm to identify patients with established tracheostomy and ventilator dependence as compared with a local registry of ventilator dependent patients. RESULTS: A newly created algorithm identified >90% of the 157 patients with established tracheostomy and ventilator dependence hospitalized at our organization during 2014. The sensitivity and specificity of the algorithm to identify these patients was 91% and 99%, respectively. CONCLUSIONS: This new algorithm can be used to reliably identify and further study healthcare utilization by this population of patients with established tracheostomy and ventilator dependence. In addition, future work can determine the applicability of this algorithm to other administrative datasets.