We studied 74 patients whose temperature was normal according to nurses' temperature charts and who were not on antibiotic treatment. The subjects were inpatients whose condition had deteriorated on the ward, or patients admitted the previous day in whom no diagnosis had been established. One simultaneous set of measurements was made of sublingual, rectal, axillary and proximal auditory canal temperatures. A fever was recorded in 63 of 74 patients (85%); 54 febrile patients had a raised rectal temperature, and 54 had a raised proximal auditory canal temperature; 60 patients were febrile at one or both of these sites. A further three patients had raised sublingual temperatures alone. All patients who were regarded as being definitely or probably infected were febrile at one or more sites. Eighty-one per cent of those considered to be possibly infected, and 71% of those with no clinical evidence of infection were also febrile. Rectal and proximal auditory canal temperatures can each detect fever in approximately 86% of febrile patients, sublingual temperature in 66%, and axillary temperature in 32%. Rectal temperature is clinically the most useful temperature measurement in elderly patients. We conclude that significant infections in patients in a warm environment result in a fever which often remains undetected when only sublingual temperature is measured.

In 76 unselected patients aged 70 years or over, the mean increase in rectal temperature in the 24 hours following admission to hospital was 0.4 degrees C. In those who did not receive antibiotics on admission, the mean increase in rectal temperature was 0.6 degrees C, with increases of up to 2.3 degrees C recorded. There were no significant changes in C-reactive protein, white cell count or erythrocyte sedimentation rate over that period, suggesting that the changes were due to passive warming rather than to progression of the underlying disease. Infected patients may have low or normal body temperatures on admission. Within 24 hours, nearly all infected patients (excluding a few with low or normal temperatures on admission, who receive antibiotics) have a raised body temperature. The most sensitive test for a raised body temperature is the rectal temperature measured at least 24 hours after admission. A patient who has a low or normal body temperature on admission has a 61% chance of having a raised body temperature the next day. At least 55% of patients admitted with a febrile illness have low or normal body temperatures on admission.

Twenty-five consecutive elderly patients with hypothermia were studied. Data were gathered regarding their home conditions, the circumstances in which they had been found, and their recent medical history. Clinical and laboratory examinations were performed to establish accurate diagnoses of underlying illnesses present at the time of arrival in hospital. Patients were followed up until the completion of the study. Evidence of an underlying cause was found in all cases. Twenty-two patients had evidence of definite or probable infection at the time of admission. Drugs may have contributed in seven cases. There were multiple significant causes for hypothermia in nine cases. Only 12 patients survived the index admission, and six of these had previous or subsequent admissions with hypothermia.