BACKGROUND: For nearly 20 years, the occupational supplement to the National Electronic Injury Surveillance System (NEISS-Work) has captured nonfatal, occupational injuries and illnesses treated in a sample of approximately 67 hospital emergency departments (EDs). Cases are captured by hospital coders who review all ED records at the sampled hospitals and abstract data from those identified as work-related. Similar to other occupational injury and illness surveillance systems, there has been a concern about the accuracy of NEISS-Work case capture. The National Institute of Occupational Safety and Health (NIOSH) conducted an assessment to better understand the data collection process and the accuracy of the data captured.
METHODS: Assessments were conducted at 20 NEISS-Work hospitals by NIOSH staff. At each hospital, NIOSH staff reviewed a minimum of 1,000 ED records to identify work-related cases. These records had been previously reviewed by the hospital coders. The cases identified as work-related by NIOSH were compared to those captured by the hospital coder. NIOSH staff also collected qualitative data from interviews with the coders and observations made during chart reviews. In addition to the hospital assessments, NIOSH conducted separate telephone interviews with workers identified from the NEISS-Work data. This allowed post-audit evaluation of certain problematic variables identified during the audit, namely payer and employment status.
RESULTS: The comparison of coder-identified cases with those identified by NIOSH staff indicated that hospital coders correctly identified approximately 69% of the work cases. Work-related illnesses were one of the most commonly missed case types. Hospital coders incorrectly classified about 1% of non-work cases as work. Qualitative findings provided insight into reasons for false positive and false negative case identification as well as factors that facilitated case identification and data quality. Assessments of payer and employment status indicated that payer was incorrectly captured about 30% of the time and employment status was incorrectly captured about 19% of the time.
CONCLUSIONS: Findings from this assessment have been used by NIOSH to guide changes in the NEISS-Work data to improve case capture and data quality. Illnesses that occur at work but have no indication of a work cause or contribution stated in the ED record are no longer captured. Based on the qualitative findings, case identification was clarified for the coders during their annual training and the NEISS-Work coder manual was updated. Finally, collection of the payer variable was discontinued.