Lowering the temperature criterion improves detection of infections in nursing home residents

The blunted fever response to infections in nursing home residents contributes to a delay in diagnosis and high morbidity and mortality. We prospectively compared the sensitivity and specificity of various criteria of fever to see if an improved definition of fever could be derived. Nursing home residents had morning baseline temperatures measured and were prospectively monitored for infections that met established criteria. Then the sensitivity and specificity of different criteria for infection, including a maximum temperature (T(max))≥ 101°F, 100°F, or 99°F and a change from baseline (ΔT) ≥ 2.4 or 1.4°F, were determined. A total of 111 nursing home residents of the Veterans Administration Medical Center (VAMC) West Los Angeles had a mean baseline temperature 97.4°F and a mean age of 76.9 years, and suffered 42 infections over 6 months. Using infection criteria that did not include temperature, the sensitivity was 40% using T(max) > 101°F as the criterion for infection-induced fever versus 70% using T(max) > 100°F. In addition, 7 of 24 infections with T(max) < 101°F had an adequate ΔT from baseline > 2.4°F. In these VAMC nursing home residents, lowering the threshold for a clinically significant fever to 100°F increases the sensitivity of detecting infections. Further studies on the impact of changes in temperature monitoring policies are needed.