Heat-related illness (HRI) is not well-defined nationwide, is not legally reportable in most states, and most active surveillance is specific to extreme heat events, rather than assessing sustained exposure to high temperatures or heat indices. In addition, limited spatial weather observations make properly assessing heat and humidity exposure challenging. However, many individuals, especially those in vulnerable populations (e.g., outdoor workers, children, older adults), are at risk of HRI. Leveraging technology and supplementing traditional public health surveillance and meteorological observations with individual-level exposure observations can provide a more complete picture of HRI risk.
iButtons are small, wireless temperature and relative humidity sensors that record observations in timed intervals. By outfitting individuals in at-risk subpopulations with iButtons, individual exposure can be assessed. However, the placement (i.e., wear location) of the iButton on the body is critical to accurate observations.
In order to determine the optimal wear location for studying individual heat and humidity exposure, a pilot field test was conducted. Analysis of intraperson and interperson variation and comparison of iButton observations to weather station observations were conducted. The specific iButton model utilized was the DS1923, which have an accuracy range of +/- 0.5°C and record both temperature and relative humidity.
This test was conducted from 9:00 a.m. to 3:30 p.m. in Tallahassee, FL on May 5, 2015 at two different locations, a park and a neighborhood 1.14 miles apart. Two investigators were identically outfitted with iButtons: one on top of each shoe, one on both sides of waist, one inside the pocket of athletic shorts, one attached outside the front collar of shirt (neck outside shirt), and one hanging on lanyard under front of shirt (neck inside shirt). The iButtons recorded observations in 3-minute intervals throughout the day. Three different activities were completed in each location to represent a range of outdoor activity: heavy (jogging), moderate (walking), resting (no activity). Cool-off and data wash-out periods took place between activities.
The nearest National Weather Service (NWS) observation station was an Automated Surface Observing System (ASOS) station located at the Tallahassee Regional Airport, 5 miles southwest of central Tallahassee.
Following the test, iButton and weather observations were analyzed with paired t-tests and ANOVA models in SAS 9.3. Data cleaning included time alignment and assignment of wash-out periods. iButtons placed on each investigator's shoes (shoe mean) and outside waist (waist mean) were averaged together due to the insignificant variation between the paired observations.
An assessment of intraperson variation was conducted to determine if all temperature and relative humidity observations from an individual investigator were consistent at each measured time point. Overall, the most consistent (i.e., no statistically significant differences) locations were the waist mean, neck outside shirt, and shoe mean (relative humidity only). The least consistent (i.e., statistically significant differences) locations were inside the pocket of shorts and neck inside shirt for both temperature and relative humidity. When analyzing intraperson variation by place and activity, some deviations from the strong relationship between the waist mean, neck outside shirt, and shoe mean were observed. At the park (AM), the shoe mean and neck inside shirt observations were consistent (humidity) during both heavy activity and resting periods, perhaps due to dew surface moisture and limited perspiration. At the neighborhood (PM), the shoe mean and neck inside shirt observations were consistent (temperature) during heavy activity, perhaps due to hot pavement and physical exertion, and the neck inside shirt and pocket observations were consistent (temperature) during moderate activity, perhaps due to physical exertion.
An assessment of interperson variation was conducted to determine if iButtons in the same wear location were consistent between investigators. Based on the intraperson results, the observations of iButtons placed under the clothing (i.e., neck inside shirt and pocket) appeared to be heavily influenced by body temperature and perspiration. Therefore, the final interperson analysis only included the shoe mean, waist mean, and neck outside shirt wear locations. Overall, shoe mean showed the most consistency between investigators. When analyzing interperson variation by place, a deviation from the overall mean was found with a strong consistency between all wear locations (relative humidity) at the park location.
When compared to observations from the local NWS station, maximum observations from iButtons worn on the waist, neck outside shirt, and shoe were no more than 5°C and 9% relative humidity above the daily maximum.
The results of the pilot test indicate that body placement of iButtons does in fact influence intraperson and interperson variation of temperature and relative humidity observations. In comparison to local NWS data, iButton temperatures were higher, but small scale influences, such as the urban heat island effect, could be playing a role.
When conducting individual heat and humidity exposure surveillance, participants should wear iButtons on the outside of clothing, with the sensor face-out, and avoid touching or holding the device to limit the transfer of body heat. Foot, waist, and neck outside shirt observations were relatively consistent for both temperature and relative humidity, and the optional wear location will likely depend on other considerations, such as safety, activity, and environment. Attachment of an iButton on the body must not pose a safety threat, such as being caught on foreign objects or restricting movement. The participants' activities could compromise the iButton itself and risk being lost or damaged. For participants who are engaged in strenuous activity, perspiration must be accounted for in order to obtain accurate humidity observations. Exposure to surfaces such as asphalt or concrete can radiate heat and influence temperature observations; therefore measuring ambient temperature based on foot placement may not represent the most accurate compared to the relative surrounding environment. Additionally, surface water may affect relative humidity observations if worn on the shoe, either from close proximity to water or through possible evaporation.
This was a very limited pilot, and further testing of optimal wear locations, with additional participants and in different conditions, would contribute to best practices for individual level surveillance using iButtons or similar sensors.