A comparison of oral, tympanic, and rectal temperature measurement in the elderly1, 2

The Journal of Emergency Medicine, Vol. 22, No. 2, pp. 153–157, 2002 Copyright © 2002 Elsevier Science Inc. Printed in the USA. All rights reserved 0736-4679/02 $–see front matter

PII S0736-4679(01)00457-7

Original Contributions

A COMPARISON OF ORAL, TYMPANIC, AND RECTAL TEMPERATURE MEASUREMENT IN THE ELDERLY Shawn M. Varney,

MD,*

David E. Manthey, MD, FAAEM,† Victoria E. Culpepper, Joseph F. Creedon, Jr., PhD, PA-C§

RN,‡

and

*San Antonio Uniformed Services Health Education Consortium, Emergency Medicine Residency Program, San Antonio, Texas; †Wake Forest University Baptist Medical Center, Winston-Salem, North Carolina; ‡Brooke Army Medical Center, Ft. Sam Houston, Texas;, and §US Army Research Institute of Environmental Medicine, Natick, Massachusetts Reprint Address: Shawn M. Varney, MD, 2705 Vergils Ct, Crofton, MD 21114-3165

e Abstract—This cross-sectional study enrolled 95 elderly Emergency Department patients aged 60 years or more to determine if rectal temperatures identify fevers more often than oral or tympanic temperatures when the chief complaint suggests an infection. A fever was defined as a temperature greater than 38°C (100.4°F). Discordance was defined as any patient with an oral or tympanic temperature of 38°C or less but manifesting a rectal temperature greater than 38°C and 0.5°C (1°F) greater than the oral or tympanic temperature. Rectal thermometry identified a fever in 14 of 95 (14.7%) patients who were afebrile orally and in 11 of 90 (12.2%) patients who were afebrile tympanically. Five of 90 (5.6%) patients were febrile rectally but were afebrile by both oral and tympanic thermometry. Thus, rectal thermometry identified fevers missed orally

and tympanically in elderly patients whose presentation suggested infection. To identify these febrile patients with possible infection, clinicians must be attentive to elderly patients’ vague clinical presentation. © 2002 Elsevier Science Inc. e Keywords—fever; rectal thermometry; elderly; core temperature

INTRODUCTION Elderly patients with infections often present to the Emergency Department (ED) with vague chief complaints such as dizziness, altered mental status, headache, shortness of breath, cough, nausea, vomiting, abdominal pain, weakness, or edema. Infection is often the cause of these complaints, but may not be sought due to the lack of fever. Many physicians have noted anecdotally that patients with “normal” oral temperatures subsequently have a fever by rectal thermometry. Many EDs use oral or tympanic thermometry to measure temperatures. We sought to determine if rectal temperatures will identify fevers (⬎38°C or 100.4°F) in patients over 60 years of age more reliably than oral or tympanic temperatures when the chief complaint suggests an infection.

This research was performed at Brooke Army Medical Center in San Antonio, Texas. This article represents the views of the authors and is not to be interpreted as official or as representing the US Army, US Air Force, or the Department of Defense. The abstract of this paper was presented in poster format at the ACEP Scientific Assembly in October 1999 in Las Vegas, NV. It was also presented in oral format at the Joint Services Symposium: Emergency Medicine on the Riverwalk 2000 in February 2000 in San Antonio, TX, and at the Society of Air Force Clinical Surgeons in April 2000 in Biloxi, MS.

RECEIVED: 29 December 2000; FINAL ACCEPTED: 10 August 2001

SUBMISSION RECEIVED:

17 July 2001;

153

154

S. M. Varney et al.

A search of the English language medical literature from 1965 to the present revealed no studies comparing elevated oral or tympanic thermometry to rectal temperature measurements in the elderly population with suspected infections.

MATERIALS AND METHODS We conducted a cross-sectional study of 95 urban, levelone trauma center ED patients aged 60 years or more whose chief complaint suggested infection. Sixty years of age was arbitrarily chosen based on other studies involving the geriatric population to define the elderly population. Patients were entered into the study if the provider determined that the patient could have an infection based on clinical presentation, past medical history, physical examination, and the provider’s overall impression. This convenience sample was enrolled between December 1998 and March 1999. Patients were excluded for an initial oral temperature ⬎38.9°C (102oF), neutropenia with a documented oral fever ⬎38°C (100.4oF), or absence of a rectum. The Infectious Disease Society of America defines a fever as a single oral temperature higher than 38.5°C (101°F) or a temperature greater than 38°C (100.4°F) over at least 1 h (1). For this study, we defined a fever as any temperature greater than 38°C. Recognizing that rectal temperatures generally register 0.5°C (1°F) higher than oral temperatures, we defined a discrepancy between the two as any temperature measured rectally that was greater than 38°C and 0.5°C over the oral or tympanic temperature (2). A full set of vital signs, including an oral temperature, was taken at triage. When the evaluating provider suspected an infection, informed consent was obtained from the patient or the patient’s legal guardian. Oral, tympanic, and rectal temperatures were obtained in the treatment area. All medics and nurses received instruction on proper use of the instruments for temperature collection. Non-blinded medical personnel obtained the temperatures according to the package insert for the thermometers. The electronic oral/rectal probe used was the IVAC Temp Plus II, model 2080A (IVAC Corporation, San Diego, CA). The tympanic probe was the First Temp Genius Infrared Tympanic Thermometer, model 8300G (Intelligent Medical Systems, Carlsbad, CA). This device reads the infrared emissions from the tympanic membrane without direct contact and operates in only one mode: oral/tympanic. All thermometers were calibrated prior to the study.

Figure 1. Fever detection by rectal versus oral routes (n ⴝ 95).

The temperatures were gathered within a 10-minute period in all but three patients. These three were not febrile by rectal thermometry. The oral probe was placed in the posterior sublingual pocket until the thermometer beeped. For patients who had ingested any beverage or smoked 15 min prior to triage, an oral temperature was repeated 15 min later. Tachypneic and mouth-breathing patients were urged to form the best seal possible around the thermometer. Oxygen by nasal cannula was left in place. The infrared ear probe was placed in the external auditory canal oriented toward the opposite temple with traction applied to the pinna (3). Straightening the auditory canal with the ear tug technique yields a higher and more reliable temperature reflecting infrared emissions from the tympanic membrane instead of the canal walls (4). The rectal probe was lubricated with room temperature gel, inserted in the anus 2– 4 cm, and removed after the beep signaled completion. Using the rectal temperature as the gold standard, we compared the results of oral, tympanic, and rectal temperatures to evaluate how well they correlated in detecting fever in the elderly population (5,6). The data were evaluated by using the frequency, percent, and 95% confidence intervals as well as odds ratios and correlation coefficients. The Institutional Review Board approved this study.

RESULTS Of the 98 enrolled patients, two were excluded for initial oral temperatures above 38.9°C (102°F), and one had no oral or tympanic temperature recorded. Five of the remaining 95 had no tympanic temperature recorded. Figure 1 displays fever detection by rectal versus oral

Rectal Temperature Measurement in the Elderly

155

edema (1/13). The disposition of afebrile patients differed in that 26/67 (39%) were admitted and 41 of 67 (61%) were discharged.

DISCUSSION

Figure 2. Fever detection by rectal versus tympanic routes (n ⴝ 90).

routes, whereas Figure 2 shows rectal versus tympanic routes. As shown in Figures 1 and 2, rectal thermometry identified a fever in 14 of 95 (14.7%) patients who were afebrile orally, and in 11 of 90 (12.2%) patients who were afebrile tympanically. Five of 90 (5.6%) patients were febrile rectally but afebrile by both oral and tympanic routes. A small percentage of patients were febrile rectally and afebrile orally (3 of 95, 3.2%) or tympanically (4 of 90, 4.5%) but by less than the expected difference of 0.5°C (1°F). Rectal thermometry identified all patients who were febrile by oral and tympanic thermometry. Table 1 displays the frequency of agreement and strength of the relationship between oral and rectal, as well as tympanic and rectal, temperatures for both presence and absence of fever, showing that there is a large amount of discordance between the various methods. There were 19 incidences of fever detection by rectal thermometry that were missed by oral (8), tympanic (6), or both (5) methods. Thirteen of these 19 (68%) febrile patients were admitted, whereas the remaining six (32%) were discharged. The diagnoses among the febrile admitted patients were pneumonia (10/13), chronic obstructive pulmonary disease (2/13), and pulmonary

The core temperature reflects the most accurate body temperature. Rectal thermometry provides the closest approximation in an outpatient setting without invasive monitoring (pulmonary artery catheter, esophageal thermometry, or bladder catheter). Our institution uses oral, digital, electronic thermometers in adults unless a rectal temperature is specifically requested. No thermometry route is ideal. Oral temperatures can be inaccurate secondary to mouth breathing, tachypnea (respiratory rate ⬎20 breaths/min), drinking cold or hot beverages, smoking, and positioning of the thermometer (6 –10). Ear probes may provide erroneous values by an inadequate seal and inaccurately measuring the infrared emissions from the tympanic membrane. Cerumen impaction and otitis media have insignificant effects on tympanic readings (2,3). Some investigators feel that rectal temperatures lag behind changes in arterial blood temperatures because of the poor arterial supply of the rectum. Furthermore, they feel that hard stool may prevent adequate placement of the rectal thermometer, that various sites in the rectum may yield different temperatures, and that inflammation around the rectum may falsely elevate the core temperature (11,12). Despite these potential drawbacks, rectal thermometry is less prone to operator error and is not affected by mouth breathing, tachypnea, oral ingestion, or external environment. Hooker et al. studied 332 adult patients in the ED and compared fever detection between oral and tympanic thermometry with the rectal as the gold standard. They found that tympanic thermometers were poorly sensitive for detecting fever compared with rectal thermometers (6).

Table 1. Frequency of Concordance, Percent, 95% Confidence Interval, and Correlation with Rectal Temperature Higher Than 38°C (100.4°F) Between the Methods Rectal vs. oral, Rectal vs. Tympanic, Oral vs. Tympanic Method

Frequency

Percent

95% Confidence Interval*

Correlation**

Rectal vs. oral Rectal vs. tympanic Oral vs. tympanic

79/95 74/90 79/90

83.1 82.2 87.7

74.1–90.1 72.7–89.5 79.2–93.7

0.621 0.764 0.641

* Exact binomial confidence interval. ** Pearson correlation coefficient.

156

We found that rectal thermometry identified fever in patients documented as afebrile by oral and tympanic thermometry. We noted that neither rectal versus oral nor rectal versus tympanic temperatures correlated well (r ⫽ 0.621, r ⫽ 0.764, respectively). Our numbers are similar to previously reported data (5,13). Although this study was not designed to determine a change in management, the discovery of a fever may affect disposition and management. Detection of a fever by rectal thermometry led to a change in management in three of the 13 (23.1%) patients in whom fever was missed orally. Finding fever initiated an investigation for infection. One patient had a chronic obstructive pulmonary disease exacerbation with a fever rectally of 39.5°C (103.3°F). This prompted a chest radiograph that identified pneumonia. Rectally febrile patients who were missed by oral or tympanic routes were about twice as likely to be admitted compared to afebrile patients. Of course, incorporation bias is present if the criteria for admitting the patient included manifesting a fever. However, if fever is an indication for admission in the respective clinical setting, then we should find it. By relying solely on fever to identify infectious processes, we will miss elderly patients with infections who cannot mount a fever regardless of how the temperature is taken. Therefore, clinical evaluation based on a composite of the patient’s presentation and medical problems, along with the provider’s general impression, should guide management and disposition of these patients. Our study has limitations. The conclusions would have carried more significance if the study had involved consecutive patients rather than a convenience sample. We had originally intended to enroll consecutive patients, but because patients were not continuously enrolled, we were forced to describe our data as a convenience sample. Thus, there is the potential for selection bias. The investigators obtaining temperatures were not blinded. The study took place during winter and may not be representative of the year-round incidence of fever and disease. No investigator examined the external auditory canal for infections or cerumen impaction. These conditions have been shown to have negligible effects on temperature in other populations. It is possible that such conditions may adversely impact measurements in elderly patients. Because many different medical personnel gathered the data, there may be increased inter-observer variability. However, this may also make these findings more universally applicable.

S. M. Varney et al.

There is probably some variability inherent in temperature measurement by any method. We did not do repeated measures. Therefore, the small difference of 0.5°C (1°F) could have resulted in an excessive number of positive findings. Defining fever is difficult, depending on which source one reviews, and a precise definition may or may not be helpful. We chose 60 years of age to define “elderly” to capture a broad range of older patients. Age alone is less descriptive of patients than an understanding of their medical problems. The small sample size precludes any conclusions regarding potential differences in temperature response between age groups. External validity of the study may be a concern because only patients suspected of having an infection were enrolled. However, that was the aim of this study, and future studies may include all elderly patients.

CONCLUSION Rectal thermometry identified fevers that were missed by oral and tympanic thermometry in elderly ED patients whose presentation suggested infection. To detect fever in the elderly population, clinicians must appreciate the vague nature of their chief complaints. Finding fever may alter disposition and management. If the provider’s overall impression (based on the chief complaint, clinical history, past medical history, and physical examination) dictates, we recommend obtaining a rectal temperature to help identify an otherwise undetected fever.

REFERENCES 1. Shah SM, Searls L. The febrile adult: Part 2. Differential diagnosis and management of infectious and noninfectious syndromes. Emerg Med Rep 1998;19(18)(suppl):183–90. 2. Petersdorf RG, Root RK. Disturbances of heat regulation. In: Braunwald E, Isselbacher KJ, Petersdorf RG, eds. Harrison’s principles of internal medicine. 11th edn. New York: McGraw-Hill; 1987:44. 3. Chamberlain JM, Grandner J, Rubinoff JL, et al. Comparison of a tympanic thermometer to rectal and oral thermometers in a pediatric emergency department. Clin Pediatr 1991;30(suppl):24 –9. 4. Terndrup TE, Rajk J. Impact of operator technique and device on infrared emission detection tympanic thermometry. J Emerg Med 1992;10:683–7. 5. Stavem K, Saxholm H, Smith-Erichsen N. Accuracy of infrared ear thermometry in adult patients. Intensive Care Med 1997;23:100 –5. 6. Hooker EA, Houston H. Screening for fever in an adult emergency department: oral vs tympanic thermometry. South Med J 1996; 89(2):230 – 4. 7. Woodman EA, Parry SM, Simms L. Sources of unreliability in oral temperatures. Nurs Res 1967;16:276 –9. 8. Erickson R. Thermometer placement for oral temperature measurement in febrile adults. Int J Nurs Stud 1976;13:199 –208.

Rectal Temperature Measurement in the Elderly 9. Durham ML, Swanson B, Paulford N. Effect of tachypnea on oral temperature estimation: a replication. Nurs Res 1986;35:211– 4. 10. Tandberg D, Sklar D. Effect of tachypnea on the estimation of body temperature by an oral thermometer. New Engl J Med 1983; 308(16):945– 6. 11. Rotello LC, Crawford L, Terndrup TE. Comparison of infrared ear thermometer derived and equilibrated rectal temperatures in esti-

157 mating pulmonary artery temperatures. Crit Care Med 1996;24: 1501– 6. 12. Blainey CG. Site selection in taking body temperature. Am J Nursing 1974;10:1859 – 61. 13. Talo H, Macknin ML, VanderBrug Medendorp S. Tympanic membrane temperatures compared to rectal and oral temperatures. Clin Pediatr 1991;(suppl):30 –3.