Jared Egbert

Project title: Evaluation of Buller Estimated Core Body Temperature Algorithm Accuracy and Application in Agricultural Workers

Degree: MPH | Program: Occupational and Environmental Medicine (OEM) | Project type: Thesis/Dissertation
Completed in: 2021 | Faculty advisor: June T. Spector


Background: Adverse health effects of extreme heat in occupational settings are substantial, particularly among outdoor workers who perform physical labor. Core body temperature (CT) is a critical indicator of heat strain. Excessive increase in CT negatively affects physical and cognitive performance and can lead to heat exhaustion or heat stroke. The most accurate locations for measuring CT are the rectum or the esophagus, but the invasiveness of these measurements makes it impractical in field settings. Ingestible pills make it possible to monitor deep body temperatures in field settings. However, these are costly to use on a regular basis, are sensitive to food and fluid intake, and may not be acceptable to all people. Thus, there is a need to be able to non-invasively measure, in real time, accurate CT. Buller et al. developed an algorithm to estimate CT based on heart rate and baseline temperature. Objective: There has been little study of the accuracy of the Buller algorithm among working populations such as agricultural workers in field settings or among females and older workers. The overall goal of this project was to assess and apply the Buller algorithm in a field setting among agricultural workers. Aim 1 evaluated the accuracy of the Buller algorithm for estimating CTs and Physiological Strain Indices (PSIs), compared to ‘gold standard’ CT sensor data from an ingestible pill, in a field setting among agricultural workers. Aim 2 examined the effectiveness of a heat prevention intervention by comparing CT and PSI outcomes estimated using the Buller algorithm between the intervention and comparison groups. We expect the CTs and PSIs in the intervention group to be lower than in the comparison group. Methods: The first aim of this project leveraged ‘gold standard’ ingestible CT sensor data and Buller algorithm-derived CT estimates that have both been collected across one work shift in 2015 among 35 Washington farmworkers. The Bland-Altman method was used to assess how well the observed CT from the ingestible pill and the estimated CT from the Buller algorithm agreed. Analyses were performed both using 37.1°C and also measured aural temperature +0.27°C as the baseline temperature for the Buller algorithm. A similar analysis for PSI was performed. The second aim of this project built upon data collected as part of a 2019 randomized heat intervention study among 75 Washington farmworkers. In this parent study, the intervention group was trained on heat safety and health precautions and supervisors were provided with a decision support application aimed at heat illness prevention. Workers in the intervention group and in the comparison group, which did not receive heat prevention training or the heat application, were evaluated approximately monthly during the summer harvest season to measure heat strain, and the maximum work shift PSI was calculated using the estimated CT from the Buller algorithm. The association between max work shift PSI and group status was assessed using linear mixed effects models with random effects for workers. Results: For Aim 1, the overall CT bias was -0.14°C with LoA of ±0.76 when 37.1°C was used as the baseline temperature in the Buller algorithm. When measured aural temperature +0.27°C was used as the baseline temperature, the overall bias was -0.085°C with LoA of ±0.90. The PSI had a bias and LoA of -0.29 ±1.59 when the baseline temperature was 37.1°C. When aural temperature +0.27°C was used as the baseline temperature, the overall bias for PSI was -0.15 with LoA of ±1.42. For the heat intervention study in Aim 2, the mean (standard deviation) of the maximum shift PSI for all participants was 4.61 (1.49) and 4.30 (1.53) for the comparison and intervention groups, respectively. The unadjusted linear mixed effects model effect estimate of group status (intervention vs. comparison) with max PSI was -0.26 (95% confidence interval [CI]: -0.84, 0.31). After adjustment for farm and ambient maximum shift heat index, the effect estimate was -0.13 (95% CI: -0.50, 0.25). Conclusion: Agricultural workers work under heat stress and are at risk for heat-related illness. The Buller algorithm, based only on heart rate and a baseline core temperature, was independently validated. When the Buller algorithm was applied to evaluate the effectiveness of the heat intervention study, intervention versus comparison group status was associated with a lower max PSI, but this was not statistically significant. Further analyses are needed to assess for potential effect modification, including by task type and exertion. The Buller algorithm may be a promising method for estimating CT and PSI in field conditions for research purposes.

URI: http://hdl.handle.net/1773/47483