Portal de Periódicos da CAPES

A prospective study of fever in the accident and emergency department

2003; Elsevier BV; Volume: 9; Issue: 8 Linguagem: Inglês

10.1046/j.1469-0691.2003.00665.x

1469-0691

Peter Jan van Laar, J. Cohen,

Sepsis Diagnosis and Treatment

Body temperature is one of the most commonly performed measurements in the accident and emergency department, and is often used as the basis for clinical decisions. Fever is frequently assumed to indicate infection, but there are few data on unselected patients presenting acutely. We studied 101 consecutive patients with fever (temperature (38 °C) among 3991 presentations to two emergency departments in tertiary-care inner city hospitals. The commonest categorical discharge diagnosis, or diagnosis at seven days, was infection (63%). There were no simple clinical tests that could distinguish fever due to infection from other diagnoses. The results suggest that, for patients presenting to an emergency department, pyrexia often indicates infection that may necessitate admission, especially in the elderly. Body temperature is one of the most commonly performed measurements in the accident and emergency department, and is often used as the basis for clinical decisions. Fever is frequently assumed to indicate infection, but there are few data on unselected patients presenting acutely. We studied 101 consecutive patients with fever (temperature (38 °C) among 3991 presentations to two emergency departments in tertiary-care inner city hospitals. The commonest categorical discharge diagnosis, or diagnosis at seven days, was infection (63%). There were no simple clinical tests that could distinguish fever due to infection from other diagnoses. The results suggest that, for patients presenting to an emergency department, pyrexia often indicates infection that may necessitate admission, especially in the elderly. Measurement of body temperature is one of the most commonly performed clinical procedures, and often it is fever that prompts the physician to order diagnostic procedures, commence treatment, and admit patients to hospital. Fever is a cardinal feature of both infective and non-infective processes, and is measured routinely in patients in accident and emergency departments. Given its importance, it is curious that it has been so little studied in this setting. We therefore carried out a prospective evaluation of patients presenting with fever to two inner city tertiary-care hospitals, to determine how often fever was caused by infection, and whether there were any features that would distinguish infection-related fever. This was a prospective study in all patients 16 years of age or older who presented to both accident and emergency departments associated with Hammersmith Hospitals Trust during four weeks in May and June, 2001. The departments accept acute admissions from the surrounding area, and over 75 000 patients/year are seen. All patients had their temperature recorded on admission. Core temperatures were determined with the use of a tympanic membrane thermometer, oral thermometer or axillary thermometer, at the discretion of the admitting nurse. Differences in readings between these methods were not systematically noted. Measurements of white blood cell (WBC) count, C-reactive protein (CRP) level, and erythrocyte sedimentation rate (ESR), and requests for blood, urine and sputum cultures or chest X-ray (CXR) examination, were obtained at the discretion of the duty physician. Fever was defined as a body temperature ≥38 °C. All X-rays were reviewed by consultant radiologists who were not involved in the study. A CXR was defined as abnormal if there were significant abnormalities attributable to the acute presentation. Incidental or unrelated abnormalities were discounted. Patients who were discharged after assessment or treatment were evaluated on the basis of the information available on the chart, and no further follow-up was done. Patients who were admitted to hospital were followed up on the ward at seven days. Descriptive statistics were used to analyze continuous and categorical variables to characterize the population. Spearman's rank correlation was used to identify predictive variables; P-values less than 0.05 were considered as significant. We studied 101 patients with fever among 3991 presentations to both departments. The mean (±SD) age was 48.2 (±21.1) years, and 51 patients were men. Fifty-nine (58%) patients were white, six (6%) were black Caribbean, six (6%) were black African, and 28 (28%) were from other ethnic groups; for two patients, the ethnic group was not given. Ninety-seven (96%) patients had resided for more than one year in the UK. Among the 101 patients with fever at presentation, 28 (28%) were seen and discharged, and 73 (72%) were admitted. The mean duration of admission was 4.6 (±3.0) days. Patients who were admitted had a mean age of 53.0 (±21.3) years, compared to patients who were seen and discharged, whose mean age was 35.7 (±14.8) years (P < 0.001). There was also a significant relationship between the duration of admission and age: older patients with pyrexia were more likely to be in hospital longer than younger ones (P < 0.001). Infection was the commonest categorical diagnosis both at presentation (75%) and by discharge diagnosis or diagnosis on the ward at seven days (63%). Significant causes in the infection group at discharge or at seven days were chest infection (n = 10), tonsillitis (n = 9), urinary tract infection (n = 8), infective exacerbation of chronic obstructive airways disease (n = 6), unspecified viral illness (n = 6), and gastroenteritis (n = 3). Non-infective diagnoses included chronic myeloid leukemia (n = 3), metastatic malignancy (n = 2), heart failure (n = 2), and epilepsy (n = 2). Two patients remained pyrexial without a diagnosis at seven days. The mean CRP was 112 (±100) mg/L (n = 80), the mean WBC value was 15.7 (±33.3) × 109/L (n = 81), and the mean ESR was 45 (±35) mm/h (n = 13). A CXR was obtained in 59 patients, and was abnormal in 21 (36%). Forty-five blood cultures were taken; 11 were positive, of which ten were considered to be clinically significant. There were 26 urine cultures, ten of which were positive; of these, eight were clinically significant. There were five sputum cultures, of which four were positive and clinically significant. None of the laboratory data available in the accident and emergency department (CRP, WBC, ESR, CXR) could predict the need for admission, or the categorical discharge diagnosis or diagnosis at seven days, either when taken alone or taken in combination. Relatively few studies have been carried out to determine the importance of fever in patients presenting to the accident and emergency department. In part, this may be because it is regarded as such a non-specific finding that it is unhelpful; yet, paradoxically, it is often used as the basis for clinical decisions. Our principal findings were that of the 2.5% of patients presenting to the emergency department with fever, the majority (63%) had an infection. A surprisingly high proportion of febrile patients (72%) was admitted, particularly elderly patients, and the majority of these patients (84%) proved to have infection. No single readily available laboratory test could distinguish the patients in whom fever was due to infection. We are aware of two other previous studies that have directly addressed the question of fever in the accident and emergency department. In one study of 188 patients with a pyrexia of ≥37.6 °C, over 90% of the patients had a fever with an infectious cause [1Manning LV Touquet R The relevance of pyrexia in adults as a presenting symptom in the accident and emergency department.Arch Emerg Med. 1988; 5: 86-90Crossref PubMed Scopus (6) Google Scholar]. Thirty-three per cent of the 188 patients were admitted, compared to 7.5% for all adult patients presenting to the department during that time. As in our study, the investigators found that 72% of all patients aged 45 years or older were admitted, compared to 22% in the younger age groups. This is probably a reflection of the lower threshold for admitting elderly patients because of a greater incidence of underlying diseases, and for social reasons such as a requirement for additional social service support. In another study, advancing age was significantly associated with more serious disease [2Keating HJ Klimek JJ Levine DS Kierman FJ Effect of aging on the clinical significance of fever in ambulatory adult patients.J Am Geriatr Soc. 1984; 32: 282-289Crossref PubMed Scopus (61) Google Scholar]. Although the cause of infections in elderly patients differs from that seen in the young, this alone is unlikely to explain the excess in admissions in the elderly. The decision to admit a patient to hospital is complex, and depends on a wide range of issues that include the severity of the illness and also the need for nursing care. It is salutary to note that there is no consensus on the criteria for defining fever, or how it should be measured, and the clinical significance of laboratory tests in the evaluation of febrile patients remains unclear [3Procop GW Hartman JS Sedor F Laboratory tests in evaluation of acute febrile illness in pediatric emergency room patients.Am J Clin Pathol. 1997; 107: 114-121PubMed Google Scholar, 4Ruiz-Laiglesia FJ Torrubia-Perez C Amiguet-Garcia JA Fiteni-Mera I Value of C-reactive protein for detecting bacteremia in febrile patients.Presse Med. 1996; 25: 1105-1108PubMed Google Scholar, 5Wenz B Gennis P Canova C Burns ER The clinical utility of the leukocyte differential in emergency medicine.Am J Clin Pathol. 1986; 86: 298-303PubMed Google Scholar]. Our hope was to identify a simple, readily available diagnostic test that would have a high sensitivity and specificity for infection as the cause of fever. Perhaps unsurprisingly, this was not possible. While it may be possible to construct a more complex algorithm using a combination of tests and multiple logistic regression, this is unlikely to be of much value in everyday clinical practice. Factors such as the characteristics of the population and the time of year will certainly affect the epidemiology. Nevertheless, these data suggest that fever is a valuable clinical sign that is frequently associated with infection, even in an unselected population presenting to an accident and emergency department. Further, larger studies should focus on better ways to identify patients with infection, particularly those who require admission. We thank Professor Konrad Jamrozik and Dr Ian Grace for help with the statistical analysis, Dr Kevin O'Kane for helpful discussions, and Professor Jan Verhoef for his supervision. P. J. van Laar was supported by a scholarship from the Erasmus program.