The optimal oral body temperature cutoff and other factors predictive of sepsis diagnosis in elderly patients.

Abstract

Introduction: The aim of this study was to identify the optimal oral temperature cut-off value and other factors predictive of sepsis in elderly patients presenting to emergency department.

Methods: A hospital-based retrospective study was performed on all elderly patients who presented to the Adult Emergency Department at King Abdulaziz Medical City in Riyadh (January to December 31, 2018).

Results: Of total of 13,856 patients, 2170 (15.7%) were diagnosed with sepsis. The associated area under the curve estimate was 0.73, 95% confidence interval (CI) 0.72-0.74. Body temperature ≥37.3 was found as optimal cut-point with sensitivity = 50.97% and specificity = 87.22% and 82.39% of patients with sepsis will be correctly classified using this cut-off. An increase of 1° in body temperature was associated with an odds ratio of 9.95 (95% CI 8.95-11.06, P < 0.0001). Those aged ≥100 years having 11.12 (95% CI 2.29-20.88, P < 0.0001) times the likelihood for sepsis diagnosis compared with those aged 60-69 years. People admitted in weather such as winter, spring, or autumn were more likely to develop sepsis than people admitted in summer.

Conclusion: The risk factors of sepsis such as age, temperature, and seasonal variation inform important evidence-based decisions. The hospitals dealing with sepsis patients should assess older patients for other severe illnesses or co-morbid that might lead to sepsis if left untreated. Therefore, older patients need to be prioritized over younger patients. The body temperature of patients admitted to hospitals needs to be monitored critically and it is important to consider seasonal fluctuations while managing cases of sepsis and allocating resources. Our findings suggest that clinicians should explore the possibility of sepsis in elderly patients admitted to emergency units with oral temperature ≥37.3°C. Risk factors for sepsis reported in this study could inform evidence-based decisions.