Identifying Neutropenic Fever Earlier: An Application of a Skin Patch for Continuous Temperature Monitoring

Febrile neutropenia is a medical emergency in patients (pts) undergoing chemotherapy or stem cell transplant (sct). Starting broad-spectrum antibiotics within the first hour of clinical signs of infection is associated with increased survival in a retrospective study of greater than 2000 pts. Each hour of delay in treatment was associated with a 7.6% decrease in survival Pts with leukemia who undergo SCT or intensive chemotherapy suffer from worse outcomes. A device that can continuously monitor body temperature can act as a clinical support tool - allowing for prompt infectious work up and initiation of broad-spectrum antibiotics leading to improve clinical outcomes and decrease incidents of sepsis and ICU transfers. In this study, we compare standard-of-care (SOC) intermittent temperature measurements with a continuous temperature monitoring device on an inpatient SCT unit.

Methods: Seventeen pts admitted for high dose chemotherapy or SCT were asked to participate. The Temp Traq Temperature Skin Patch (TSP) is an FDA Class II Device developed by Blue Spark Technologies. It transmits data to an IPAD or Smart Phone via Blue Tooth Technology in real time. The patches were applied under the axilla every 24 hours. Continuous skin temperature measurements began upon placement and were transmitted to the project IPAD in 10 minute intervals. Patches were changed every 24 hours after pts' daily shower. Pts continued to have routine thermometry monitoring according to SOC. TSP readings were not used for clinical decision making. A questionnaire was designed to capture pts' experience. The correlation of temperature measurements by TSP and SOC was assessed by simple linear regression model. The temperatures detected by TSP between each SOC measurements (i.e., q4hrs) were divided into four intervals based on the time lag from next SOC, 0-30 min(m), 30m-1hr, 1-2hr and 2-4hr. Each interval counted as a separate binary variable, i.e., positive or negative based on presence of at least one value more than 100.2 F in a given interval. Receiver-operator characteristic (ROC) curves of the 4 time intervals were generated and area-under the curve (AUC) determined, assuming SOC as the gold standard test. The diagnostic effectiveness of TSP for a single patient was measured by Youden's Index and Accuracy Index. To identify the potential clinical impact of this device we analyzed the first neutropenic fever throughout the hospital course.

Results: Median age was 58 (range: 29-78). Eleven (65%) were male and the rest female. Eleven pts received SCT and the rest had high dose chemotherapy. Length of stay was 27 days (range: 4-77). All pts received viral and fungal prophylaxis; 15 pts (88%) had bacterial prophylaxis. Five pts (29%) had ongoing corticosteroid use. TSP data of 5,856 continuous hours was studied. The TSP data was successfully transmitted and displayed on the study iPad invariably. The diagnostic effectiveness of the TSP was not influenced by pts' room distance from the IPad receiver (Figure-1). All pts were able to self-wear the patch through the hospital admission and the majority reported it was comfortable to wear and are interested in wearing it in future admissions or at discharge. (Table 1). The TSP in four intervals correlated significantly with SOC temperatures; 0-30m r: 0.29, p<0.001, 30m-1hr r: 0.27, p<0.001, 1-2hr r: 0.24, p<0.001; 2-4 hr r: 0.21, p<00.1 (Fig 2A). AUC of binary TSP in all four time intervals was significantly higher than SOC; 0-30 m AUC: 0.766 (CI: 0.708-0.824, p<0.001); 30m-1hr AUC: 0.755 (CI: 0.701-0.809, p<0.001); 1-2hr AUC: 0.718 (CI: 0.663-0.773, p<0.001); 2-4 hr. AUC: 0.702 (CI: 0.646-0.757, p<0.001) (Fig 2B). Total of 13 pts had fever. The average time from admission to first fever was 9 days (Range 1-17). 8 pts had documented infections. 6 of the 8 fevers preceding infectious work up were detected by both SOC and TSP. These fevers were detected 22.4 hours (mean) and 11.4 hours (median) earlier by TSP than SOC. One initial febrile episode was detected by TSP but not SOC.

Conclusions: Taken together, our data suggest that remote and continuous skin temperature measurement is feasible in the inpatient setting and can provide the opportunity to detect neutropenic fever earlier. Further studies are warranted to understand the clinical implication of this temporal relationship on identifying infectious source and antibiotic sensitivities, as well as rates of sepsis and ICU transfer.

View largeDownload slide

View largeDownload slide

Close modal