Testing a Liquid Crystal Visual Thermometer Device in Newborns and Young Infants

Objective Severe neonatal jaundice is a major cause of death and disability among newborns in low- and middle-income countries (LMICs). Filtered sunlight phototherapy (FSPT) is safe and effective but requires hourly temperature monitoring to detect hyperthermia/hypothermia. This need for close temperature monitoring by family members and community health extension workers is impeding FSPT scale-up in LMICs, where healthcare providers are scarce and nurseries are understaffed. The ability of a caregiver to accurately measure temperature can affect infant health outcomes in many other illnesses as well. This study aims to evaluate the accuracy and usability of a modified Liquid Crystal Thermometer Device (LCTD) strip, modified from the original ThermoSpot™ (Maternova, Providence, RI), for caregivers and healthcare providers in measuring infant temperatures. The modification was attempted to improve monitoring for hyperthermia better than the original ThermoSpot™, which had only one color for hyperthermia but three colors for hypothermia, and changed as the infants got colder. Methods We conducted a cross-sectional study of infants (zero to six months) at the University of Minnesota Masonic Children's Hospital (UMMCH) in Minnesota and Lagos University Teaching Hospital (LUTH) in Nigeria. Analysis was limited to 40 infants enrolled at LUTH following preliminary data review. We compared standard of care (SoC) temperatures using digital thermometers with LCTD readings at five different time points over two hours. Photos of the LCTD strips were taken at each time point and were later reviewed by researchers. Results Forty infants were studied to determine the agreement between LCTD readings and SoC temperatures. These results were then compared between caregivers and researchers. The average agreement between all readers (caregivers and research assistants) was 64%. After a methodological adjustment to remove the initial LCTD temperature reading from each infant's dataset, the overall agreement rate between caregivers and research assistants was 63.5%. We also separately analyzed the accuracy of caregiver and research assistant readings of the LCTD strip when compared to the SoC temperature, finding that caregivers had an overall accuracy of 65%, while the research assistants had an overall accuracy of 63%. Lastly, we conducted a blinded photo analysis among research assistants to determine whether LCTD readings were being interpreted consistently between researchers, and found a slight bias for higher temperatures: for readings at or above 37°C, the interpreted temperature tended to be higher than the measured temperature. Conclusion The modified LCTD did not correlate well with digital thermometer temperature readings. Further modifications are needed to make it useful for managing neonates under FSPT and other conditions/illnesses where temperature monitoring is advised. There remains a gap in providing illiterate caregivers a reliable way to determine if their infant is normothermic and safe under FSPT.