This study will analyze the narrative text fields in all National Fire Fighter Near-Miss Reporting System (NFFNMRS) reports submitted since the system was created in 2005 (3,695 reports as of September 30, 2009). Near miss reporting systems have made major contributions to safety in such industries as aviation, nuclear power, petrochemical processing, steel production, and military operations, because the same patterns of causes of failure and their relations precede both adverse events and near misses. However, firefighters and researchers lack a scientific system to fully analyze the near miss data collected each year. This innovative effort will advance knowledge in firefighter safety by applying novel Bayesian methods of analysis to the narrative text fields of a new data source that has not yet been rigorously investigated. The proposal has 3 aims: I. to use recently developed auto coding methods to characterize firefighter near miss narratives and classify these narratives into mechanisms of risk/injury. This analysis will apply the International Classification of External Cause of Injuries (ICECI) using Bayesian machine learning techniques to identify the various mechanisms captured in the near miss narratives and their relative prevalence. II. To identify the correlation between each mechanism of risk/injury and each of the """"""""Contributing Factors"""""""" listed on the NFFNMRS reporting form. The results will reveal any patterns and trends in the distribution of the contributing factors among the mechanisms, creating a deeper understanding of near miss circumstances, as well as a basis for improving the quality of future near miss data collection. III. To use manual coding to identify actual injury incidents contained within a random sample of 1,000 near miss narratives and correlate these injuries with the """"""""Loss Potential"""""""" categories on the NFFNMRS reporting form. The results will demonstrate how actual injuries are distributed within the reporting form's """"""""Loss Potential"""""""" categories. This proposed study of the near miss narrative text in combination with coded data has the potential to reveal new insights that can strengthen firefighter safety through primary prevention. This study addresses a major gap in firefighter safety knowledge, i.e. the insufficient understanding of near miss events, and will have a high impact on efforts to improve the occupational health and safety of firefighters.
This study will analyze the narrative text fields in all National Fire Fighter Near-Miss Reporting System (NFFNMRS) reports submitted since the system was created in 2005 (3,695 reports as of September 30, 2009). Near miss reporting systems have made major contributions to safety in such industries as aviation, nuclear power, petrochemical processing, steel production, and military operations, because the same patterns of causes of failure and their relations precede both adverse events and near misses. However, firefighters and researchers lack a scientific system to fully analyze the near miss data collected each year. This innovative effort will advance knowledge in firefighter safety by applying novel Bayesian methods of analysis to the narrative text fields of a new data source, the NFFNMRS that has not yet been rigorously investigated. This proposed study of the near miss narrative text in combination with coded data has the potential to reveal new insights that can strengthen firefighter safety through primary prevention.
