Tuesday, June 6, 2017: 11:10 AM
420B, Boise Centre
BACKGROUND:
Inpatient Hospital Discharge Database (HDD) is a population-based data source used for occupational health surveillance. Current Council of State and Territorial Epidemiologist (CSTE) occupational health indicators (OHI) include inpatient discharges with workers' compensation (WC) as the primary payer source. In states that allow employers to "opt-out" of WC, identifying work-related hospitalizations based on WC may lead to an undercount. Because Texas is an opt-out state, we hypothesized that using auxiliary data fields and ICD-9-CM codes, in addition to primary payer source would significantly increase the number of work-related hospitalizations than just using a primary payer source.METHODS:
We extracted five years (2010-2014) of records from the Texas inpatient HDD for individuals 15 years and older. Work-related hospitalizations were identified using three different methods: method 1 included hospitalizations with primary payer ="WC"; method 2 included hospitalizations meeting method 1 criteria and/or any of following: secondary payer= "WC", occurence code (event relating to claim) or condition code (condition relating to claim) = "employment-related", or value code ="WC"; and method 3 included hospitalizations meeting method 2 criteria and/or having potential work-related ICD-9-CM codes present [E000.0-civilian activity for pay; work-related transportation E-codes; location E-codes (farm, mine/quarry, industrial); or work-related V-codes]. Descriptive statistics were calculated, and McNemar’s test was used to determine if there was a statistically significant change in the number of cases detected by newer methods. We also examined the agreement between method 1 vs. individual auxiliary variables and case-identification methods 2 and 3 using Cohen’s Kappa.RESULTS:
During 2010-14, method 1 identified 41,459 work-related hospitalizations. Seventy-three percent of these were males, and patients were mainly in age groups 45-49 (11.6%), 50-54 (14.0%), and 55-59 years (12.4%). Method 2 identified 49,491 work-related cases (7,952 or 19.2% additional cases vs. method 1), whereas method 3 identified 50,834 cases (9,375 or 22.6% additional cases vs. method 1). The increase in the number of work-related hospitalizations ascertained by newer methods was statistically significant (McNemar’s p= <0.001 for both methods 2 and 3).CONCLUSIONS:
The use of auxiliary inpatient HDD variables in addition to primary payer source significantly increased the number of work-related hospitalizations ascertained. Using a combination of variables allowed us to capture work-related cases, which were missed when only a single data field was used for case-identification.