Search

Background: Agricultural safety and health researchers have used a variety of classification and coding schemes to identify and categorize injury, illness, and disease associated with agricultural hazards. This paper demonstrates how the Farm and Agricultural Injury Classification (FAIC) and Occupational Injury and Illness Classification System (OIICS) coding schemes can be used in tandem for all agricultural injuries., Methods: Specific cases from the AgInjuryNews.org database were selected to illustrate how the FAIC and OIICS codes can be applied to agricultural injury, and to illustrate limitations that still exist with each coding scheme., Results: Using the FAIC and OIICS together provided a clearer picture of an injury incident by combining more explicit occupational and non-occupational exposures with incident type details in a single table., Conclusions: Both FAIC and OIICS coding schemes depend on sufficient information being available from injury reports. This paper concluded with suggestions for improving coding of agricultural injury., (© 2018 Wiley Periodicals, Inc.)

Background: This preliminary study evaluates a real-time syndromic surveillance system to track occupationally-related emergency room visits throughout New Jersey., Methods: Emergency Department (ED) chief complaint fields were evaluated from 79 of 80 hospitals in NJ in 2014, using work-related keywords and ICD-9 E-codes, to determine its ability to capture non-fatal work-related injuries. Sensitivity analysis and descriptive statistics, were used to evaluate and summarize the occupational injuries identified., Results: Overall, 11 919 (0.3%) possible work-related ED visits were identified from all ED visits. Events with the greatest number of ED visits were slips, trips, and falls (1679, 14%). Nature of injury included cuts, lacerations (1041, 9%). The part of the body most affected was the back (1414, 12%). This work-related classifier achieved a sensitivity of 5.4%, a specificity of 99.8%, and a PPV of 2.8%., Conclusions: This evaluation demonstrated that the syndromic surveillance reporting system can yield real-time knowledge of work-related injuries., (© 2017 Wiley Periodicals, Inc.)

Aim: To examine the quality of manual job coding carried out by occupational health teams with access to a software application that provides assistance in job and business sector coding (CAPS). Methods: Data from a study conducted in an Occupational Health Service were used to examine the first-level coding of 1,495 jobs by occupational health teams according to the French job classification entitled “PSC- Professions and socio-professional categories” (INSEE, 2003 version). A second level of coding was also performed by an experienced coder and the first and second level codes were compared. Agreement between the two coding systems was studied using the kappa coefficient (κ) and frequencies were compared by Chi2 tests. Results: Missing data or incorrect codes were observed for 14.5% of social groups (1 digit) and 25.7% of job codes (4 digits). While agreement between the first two levels of PCS 2003 appeared to be satisfactory (κ=0.73 and κ=0.75), imbalances in reassignment flows were effectively noted. The divergent job code rate was 48.2%. Variation in the frequency of socio-occupational variables was as high as 8.6% after correcting for missing data and divergent codes. Conclusions: Compared with other studies, the use of the CAPS tool appeared to provide effective coding assistance. However, our results indicate that job coding based on PSC 2003 should be conducted using ancillary data by personnel trained in the use of this tool.

Introduction: Studies on autocoding injury data have found that machine learning algorithms perform well for categories that occur frequently but often struggle with rare categories. Therefore, manual coding, although resource-intensive, cannot be eliminated. We propose a Bayesian decision support system to autocode a large portion of the data, filter cases for manual review, and assist human coders by presenting them top k prediction choices and a confusion matrix of predictions from Bayesian models., Method: We studied the prediction performance of Single-Word (SW) and Two-Word-Sequence (TW) Naïve Bayes models on a sample of data from the 2011 Survey of Occupational Injury and Illness (SOII). We used the agreement in prediction results of SW and TW models, and various prediction strength thresholds for autocoding and filtering cases for manual review. We also studied the sensitivity of the top k predictions of the SW model, TW model, and SW-TW combination, and then compared the accuracy of the manually assigned codes to SOII data with that of the proposed system., Results: The accuracy of the proposed system, assuming well-trained coders reviewing a subset of only 26% of cases flagged for review, was estimated to be comparable (86.5%) to the accuracy of the original coding of the data set (range: 73%-86.8%). Overall, the TW model had higher sensitivity than the SW model, and the accuracy of the prediction results increased when the two models agreed, and for higher prediction strength thresholds. The sensitivity of the top five predictions was 93%., Conclusions: The proposed system seems promising for coding injury data as it offers comparable accuracy and less manual coding., Practical Applications: Accurate and timely coded occupational injury data is useful for surveillance as well as prevention activities that aim to make workplaces safer., (Copyright © 2016 Elsevier Ltd and National Safety Council. All rights reserved.)

Background: Estimates of select occupational injuries and illnesses often differ across data sources. We explored agreement in injury classifications and the impact of differences on case estimates among records reported to multiple data sources., Methods: We linked cases reported in the Bureau of Labor Statistics (BLS) annual Survey of Occupational Injuries and Illnesses (SOII) to Washington State workers' compensation (WC) claims and evaluated agreement in injury characteristics coded in each data source according to the same occupational injury and illness classification system., Results: Agreement between data sources was greatest for body part and lowest for event or exposure. Agreement on nature of injury varied by condition. WC-assigned injury codes estimated 94% more amputations than SOII-assigned codes while SOII-assigned codes estimated 34% more work-related MSD cases., Conclusions: Accounting for classification differences may improve case ascertainment within individual data sources and help align injury and illness estimates derived from different data sources., (© 2013 The Authors. American Journal of Industrial Medicine Published by Wiley Periodicals, Inc.)