Comparing On-Site Wet Bulb Globe Temperature Measurements with Phone Application Estimates and their Influence on High School Sports Activity Modification

Context: Measuring WBGT, and modifying sports activity as environmental conditions increase, is an effective exertional heat illness prevention strategy. Many states mandate this strategy and experts recommend WBGT be measured on-site. Phone applications (PA) exist to estimate WBGT but have not been systematically evaluated in public research compared to on-site device (OS) measurements. The purpose of our study was to determine accuracy of PA measures compared to OS WBGT measures. The secondary purpose was to determine if the different measures impacted activity modification categories.

Methods: An observational research design was utilized. The independent variable was measurement type (OS, PA). A convenience sample of high schools in possession of an OS WBGT thermometer (Kestrel 5400) across the United States was used. Environmental data were collected daily by athletic trainers, for a minimum of 2 weeks (August 1-September 30). All OS devices were set-up > ten minutes before the session; positioned on the same surface as the sport activity site. WBGT was recorded a minimum of two readings per activity session. WBGT was measured simultaneously by a PA (Zelus Sports-basic). A google sheet gathered date, time, surface type, and the dependent variables of WBGT and its components (dry bulb, natural wet bulb, globe temperatures). Activity modification categories, based on the NATA position statement, were determined ex post facto using WBGT (category 1 [<27.77°] through category 5 [>33.33°C]). Descriptive statistics were calculated for each dependent variable. The mean absolute error (MAE) and root mean square error (RMSE) were used to quantify differences between paired observations. The probability of detection, or hit rate (hit/(miss+hit); 0 worst, 1 perfect), was calculated to determine how closely PA matched the activity modification category identified by OS measurements.

Results: Schools recorded 1056 paired measurements (OS vs PA), representing 30 high schools in 11 states, across multiple surfaces (artificial turf 53%, natural grass 44%, and other 3%). The mean OS WBGT was 27.92±4.14°C (range: 13.72-37.11°C). The mean PA was 26.78±3.66°C (range: 10.27-35.27°C). MAE and RMSE were calculated for overall WBGT (1.76°C; 2.16°C), dry bulb (1.48°C; 2.04°C), wet bulb (1.83°C; 2.27°C), and globe temperature (3.49°C; 4.38°C). Hit rate was 0.95 for category 1 (observations n=489), 0.46 for category 2 (n=259), 0.26 for category 3 (n=118), 0.10 for category 4 (n=91), and 0.13 for category 5 (n=99).

Conclusions: RMSE and MAE were lowest for dry bulb and highest for globe temperature, possibly explaining a source of error in the PA WBGT estimate. The PA did a good job matching the activity modification category identified by OS devices at the coolest category but did progressively worse at hotter categories. Caution should be applied when using the PA in hot conditions when activity modifications are the most important to implement for heat safety.