- Email: [email protected]
Lack of agreement between tympanic and oral temperature measurements in adult hospitalized patients
BRIEF REPORT Lack of agreement between tympanic and oral temperature measurements in adult hospitalized patients Farrin A. Manian, MD, MPH Sandy Griesenauer, RN, MSN, CIC St Louis, Missouri
Background: The purpose of this study was to compare temperature measurements obtained by tympanic thermometers with those obtained by oral electronic or mercuryglass thermometers in adult hospitalized patients. Methods: A prospective study of 406 nonintensive care unit adult patients hospitalized during an 8-month period in a tertiary care community medical center. Results: Poor agreement was observed between tympanic versus electronic thermometer readings, with 95% limits of agreement of –2.11°F to +2.81°F. Similarly, poor agreement was observed between tympanic versus oral mercury-glass temperatures, with 95% limits of agreement of –1.72°F and +2.64°F. Of 27 oral electronic temperatures 100°F (37.7°C) or higher, 10 (37%, 95% confidence interval 19% to 58%) readings were 99.5°F (37.5°C) or lower, and six (22%, 95% confidence interval 9% to 42%) measured lower than 98.6°F by tympanic thermometers. Conclusions: Temperatures measured by tympanic thermometers generally have poor agreement with those measured by oral electronic or mercury-glass thermometers in adult hospitalized patients. We recommend that tympanic thermometers not be used for routine screening for fever in this patient population. (AJIC Am J Infect Control 1998;26:428-30)
Although tympanic thermometers have been widely used in health care facilities for many years, the degree of agreement between temperatures measured by these devices compared with those measured by other types of thermometers remains controversial.1,2 To date, the majority of studies have involved patients in pediatric units, ambulatory clinics, intensive care units, or emergency departments. We studied the agreement between temperatures recorded by tympanic thermometers and those recorded by oral electronic From the Department of Infection Control, St John’s Mercy Medical Center. Reprint requests: Farrin A. Manian, MD, MPH, Chief, Division of Infectious Diseases, St John’s Mercy Medical Center, 621 S New Ballas, St Louis, MO 63141. Copyright © 1998 by the Association for Professionals in Infection Control and Epidemiology, Inc. 0196-6553/98 $5.00 + 0
428
17/47/90130
or mercury-glass thermometers in adult hospitalized patients outside the intensive care unit. METHODS
The study was performed from January through August 1997 at St. John’s Mercy Medical Center, a tertiary care community hospital. A convenient patient sample from the medical and surgical wards (outside the intensive care unit) was selected, with no more than 3 measurements per patient during the study period. Temperatures were measured by a registered nurse throughout the study. Patient temperatures were simultaneously (at a single setting) measured by tympanic, oral electronic, and mercury-glass thermometers. The tympanic thermometer (IVAC Core-Check, model 2090, IVAC Corporation, San Diego, Calif) was used according to the manufacturer’s instructions, placing the probe tip with the disposable
AJIC
Manian and Griesenauer 429
Volume 26, Number 4
Table 1. Agreement between various methods for measuring temperature (°F) in adult hospitalized patients 95% limits of agreement Methods* compared
Electronic oral vs tympanic Mercury oral vs tympanic Electronic oral vs mercury oral Right vs left tympanic
N†
Mean difference
SD
Lower
Upper
465 462 481 398
0.35 0.46 –0.01 –0.05
1.23 1.09 0.63 0.69
–2.11 –1.72 –1.36 –1.43
+2.81 +2.64 +1.16 +1.33
*Right ear tympanic measurement was used for comparison with electronic and mercury thermometers. †N is not the same for all comparisons because of the lack of paired measurements in all cases (eg, because of patient refusal or inability to measure tympanic temperatures in some neurosurgical cases).
single-use cover in the opening of the ear canal, and applying gentle but firm pressure to seal the canal. Before temperature measurement, the thermometer lens was checked for debris and fingerprints. Each temperature reading took less than 3 seconds. Several different tympanic thermometers used routinely by the nursing staff on the ward were used on the basis of availability. Whenever possible, temperatures measured from both ears were recorded. Oral temperatures were measured in the sublingual pocket with 1 of 2 IVAC Temp Plus II thermometers (Model 2080A, IVAC Corporation). The oral thermometer probe was left in the sublingual pocket until the thermometer beeped. Patients who had cold or hot drinks, chewing gum, or hard candy less than 30 minutes before the temperature measurement were not included in the study. Temperatures measured with oral mercury-glass thermometers were obtained by placing the distal half of the thermometer in the sublingual pocket and leaving it in place for a minimum of 3 minutes. Six mercury-glass thermometers were used on a rotating basis, with a new plastic sheath placed on the device before each use. Although no attempt was made at calibration of mercury-glass thermometers, all were new at the start of the study. Because this project was part of continuous quality improvement activities of the hospital and oral electronic and mercury-glass thermometers were used widely before the hospital’s adoption of tympanic thermometers, no formal approval of this project was obtained from the Human Research Committee of the hospital. However, patients reserved the right to refuse temperature measurement by 1 or more of the thermometers. Statistical analysis was performed with the method proposed by Bland and Altman3 for
assessing agreement between 2 methods of clinical measurement. Ninety-five percent limits of agreement between 2 methods were calculated with the mean difference ± 2 SDs. RESULTS
Of a total of 406 hospitalized patients undergoing temperature measurements by 2 or more of the techniques described, 224 (55%) were females. Patient age range was 15 to 97 years with a median age of 62 years. Table 1 shows the mean difference, SD of the difference, and 95% lower and upper limits of agreement between the various methods of temperature measurement, and between the right and left ear by tympanic thermometry. The highest mean differences and the greatest range of differences between 2 methods were observed when tympanic temperatures were compared with those measured by oral electronic and mercuryglass thermometers. The mean differences between electronic and mercury-glass thermometer-derived temperatures, and between right and left tympanic temperatures, were near zero. Of 27 electronic oral temperatures 100°F or higher, 10 (37%, 95% confidence interval [CI] 19% to 58%) measured 99.5°F (37.5°C) or lower, and six (22%, 95% CI 9% to 42%) measured lower than 98.6°F (37°C) by tympanic thermometers. Of 10 electronic oral temperatures 101°F (38.3°C) or higher, two (20%, 95% CI 3% to 56%) measured 99.5°F or lower, and one (10%, 95% CI 0.25% to 44.5%) measured lower than 98.6°F by tympanic thermometers. DISCUSSION
The apparent ease of use, lack of potential exposure to mucous membranes, and almost instantaneous temperature readout of the tympanic ther-
AJIC
430 Manian and Griesenauer
mometers have led to their widespread use in a variety of settings, including hospitals. However, a recent review of the literature comparing tympanic thermometry with rectal thermometry concluded that tympanic thermometry was poorly sensitive for detection of fever with nearly one half of the studies reporting less than 70% sensitivity for fever when tympanic thermometers were used.2 Unfortunately, the great majority of these studies have focused on pediatric patients, and the studies involving adults have taken place primarily in emergency department settings. Because a variety of factors, such as ambient temperature, patient cooperation, and age, may potentially affect temperature measurements by tympanometry,4,5 it is important to study the utility of tympanic thermometers in relatively ill adult hospitalized patients. We found relatively poor agreement between tympanic temperatures and oral temperatures measured by electronic or mercury-glass methods in adult hospitalized patients. Furthermore, of those patients with electronic oral temperatures of 100°F or higher, 37% were found to have tympanic temperatures of 99.5°F or lower (usually not considered a significant temperature elevation), resulting in a sensitivity of
August 1998
only 63% when temperatures of 100°F or higher are considered fever. In conclusion, there is poor agreement between temperatures measured by tympanic thermometry and those measured by oral electronic or mercury-glass thermometers in adult patients hospitalized outside the intensive care unit. These findings have led us to recommend that tympanic thermometers not be generally used for exclusion of fever in this patient population. We thank Tammy Gillman and C. J. Manian for their diligent assistance in data processing.
References 1. Terndrup TE. An appraisal of temperature assessment by infrared emission detection tympanic thermometry. Ann Emerg Med 1992;21:1483-92. 2. Hooker EA, Houston H. Screening for fever in an adult emergency department: oral vs tympanic thermometry. South Med J 1996;89:230-4. 3. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986;1:307-10. 4. Selfridge J, Shea SS. The accuracy of the tympanic membrane thermometer in detecting fever in infants aged 3 months and younger in the emergency department setting. J Emerg Nurs 1993;19:127-30. 5. Terndrup TE. Tympanic thermometers. South Med J 1994;87:1059.
Correction
In the article entitled “Guideline for infection control in health care personnel, 1998” (Bolyard EA, Tablan OC, Williams WW, Pearson ML, Shapiro CN, Deitchman SD. AJIC Am J Infect Control 1998;26:289-354 and Infect Control Hosp Epidemiol 1998;19:407-63), the length of time that susceptible health care personnel exposed to the varicella-zoster virus should be restricted from duty was misstated. Susceptible health care personnel exposed to the varicella-zoster virus should be restricted from duty from the 10th day after the first exposure through the 21st day after the last exposure.
Background: The purpose of this study was to compare temperature measurements obtained by tympanic thermometers with those obtained by oral electronic or mercuryglass thermometers in adult hospitalized patients. Methods: A prospective study of 406 nonintensive care unit adult patients hospitalized during an 8-month period in a tertiary care community medical center. Results: Poor agreement was observed between tympanic versus electronic thermometer readings, with 95% limits of agreement of –2.11°F to +2.81°F. Similarly, poor agreement was observed between tympanic versus oral mercury-glass temperatures, with 95% limits of agreement of –1.72°F and +2.64°F. Of 27 oral electronic temperatures 100°F (37.7°C) or higher, 10 (37%, 95% confidence interval 19% to 58%) readings were 99.5°F (37.5°C) or lower, and six (22%, 95% confidence interval 9% to 42%) measured lower than 98.6°F by tympanic thermometers. Conclusions: Temperatures measured by tympanic thermometers generally have poor agreement with those measured by oral electronic or mercury-glass thermometers in adult hospitalized patients. We recommend that tympanic thermometers not be used for routine screening for fever in this patient population. (AJIC Am J Infect Control 1998;26:428-30)
Although tympanic thermometers have been widely used in health care facilities for many years, the degree of agreement between temperatures measured by these devices compared with those measured by other types of thermometers remains controversial.1,2 To date, the majority of studies have involved patients in pediatric units, ambulatory clinics, intensive care units, or emergency departments. We studied the agreement between temperatures recorded by tympanic thermometers and those recorded by oral electronic From the Department of Infection Control, St John’s Mercy Medical Center. Reprint requests: Farrin A. Manian, MD, MPH, Chief, Division of Infectious Diseases, St John’s Mercy Medical Center, 621 S New Ballas, St Louis, MO 63141. Copyright © 1998 by the Association for Professionals in Infection Control and Epidemiology, Inc. 0196-6553/98 $5.00 + 0
428
17/47/90130
or mercury-glass thermometers in adult hospitalized patients outside the intensive care unit. METHODS
The study was performed from January through August 1997 at St. John’s Mercy Medical Center, a tertiary care community hospital. A convenient patient sample from the medical and surgical wards (outside the intensive care unit) was selected, with no more than 3 measurements per patient during the study period. Temperatures were measured by a registered nurse throughout the study. Patient temperatures were simultaneously (at a single setting) measured by tympanic, oral electronic, and mercury-glass thermometers. The tympanic thermometer (IVAC Core-Check, model 2090, IVAC Corporation, San Diego, Calif) was used according to the manufacturer’s instructions, placing the probe tip with the disposable
AJIC
Manian and Griesenauer 429
Volume 26, Number 4
Table 1. Agreement between various methods for measuring temperature (°F) in adult hospitalized patients 95% limits of agreement Methods* compared
Electronic oral vs tympanic Mercury oral vs tympanic Electronic oral vs mercury oral Right vs left tympanic
N†
Mean difference
SD
Lower
Upper
465 462 481 398
0.35 0.46 –0.01 –0.05
1.23 1.09 0.63 0.69
–2.11 –1.72 –1.36 –1.43
+2.81 +2.64 +1.16 +1.33
*Right ear tympanic measurement was used for comparison with electronic and mercury thermometers. †N is not the same for all comparisons because of the lack of paired measurements in all cases (eg, because of patient refusal or inability to measure tympanic temperatures in some neurosurgical cases).
single-use cover in the opening of the ear canal, and applying gentle but firm pressure to seal the canal. Before temperature measurement, the thermometer lens was checked for debris and fingerprints. Each temperature reading took less than 3 seconds. Several different tympanic thermometers used routinely by the nursing staff on the ward were used on the basis of availability. Whenever possible, temperatures measured from both ears were recorded. Oral temperatures were measured in the sublingual pocket with 1 of 2 IVAC Temp Plus II thermometers (Model 2080A, IVAC Corporation). The oral thermometer probe was left in the sublingual pocket until the thermometer beeped. Patients who had cold or hot drinks, chewing gum, or hard candy less than 30 minutes before the temperature measurement were not included in the study. Temperatures measured with oral mercury-glass thermometers were obtained by placing the distal half of the thermometer in the sublingual pocket and leaving it in place for a minimum of 3 minutes. Six mercury-glass thermometers were used on a rotating basis, with a new plastic sheath placed on the device before each use. Although no attempt was made at calibration of mercury-glass thermometers, all were new at the start of the study. Because this project was part of continuous quality improvement activities of the hospital and oral electronic and mercury-glass thermometers were used widely before the hospital’s adoption of tympanic thermometers, no formal approval of this project was obtained from the Human Research Committee of the hospital. However, patients reserved the right to refuse temperature measurement by 1 or more of the thermometers. Statistical analysis was performed with the method proposed by Bland and Altman3 for
assessing agreement between 2 methods of clinical measurement. Ninety-five percent limits of agreement between 2 methods were calculated with the mean difference ± 2 SDs. RESULTS
Of a total of 406 hospitalized patients undergoing temperature measurements by 2 or more of the techniques described, 224 (55%) were females. Patient age range was 15 to 97 years with a median age of 62 years. Table 1 shows the mean difference, SD of the difference, and 95% lower and upper limits of agreement between the various methods of temperature measurement, and between the right and left ear by tympanic thermometry. The highest mean differences and the greatest range of differences between 2 methods were observed when tympanic temperatures were compared with those measured by oral electronic and mercuryglass thermometers. The mean differences between electronic and mercury-glass thermometer-derived temperatures, and between right and left tympanic temperatures, were near zero. Of 27 electronic oral temperatures 100°F or higher, 10 (37%, 95% confidence interval [CI] 19% to 58%) measured 99.5°F (37.5°C) or lower, and six (22%, 95% CI 9% to 42%) measured lower than 98.6°F (37°C) by tympanic thermometers. Of 10 electronic oral temperatures 101°F (38.3°C) or higher, two (20%, 95% CI 3% to 56%) measured 99.5°F or lower, and one (10%, 95% CI 0.25% to 44.5%) measured lower than 98.6°F by tympanic thermometers. DISCUSSION
The apparent ease of use, lack of potential exposure to mucous membranes, and almost instantaneous temperature readout of the tympanic ther-
AJIC
430 Manian and Griesenauer
mometers have led to their widespread use in a variety of settings, including hospitals. However, a recent review of the literature comparing tympanic thermometry with rectal thermometry concluded that tympanic thermometry was poorly sensitive for detection of fever with nearly one half of the studies reporting less than 70% sensitivity for fever when tympanic thermometers were used.2 Unfortunately, the great majority of these studies have focused on pediatric patients, and the studies involving adults have taken place primarily in emergency department settings. Because a variety of factors, such as ambient temperature, patient cooperation, and age, may potentially affect temperature measurements by tympanometry,4,5 it is important to study the utility of tympanic thermometers in relatively ill adult hospitalized patients. We found relatively poor agreement between tympanic temperatures and oral temperatures measured by electronic or mercury-glass methods in adult hospitalized patients. Furthermore, of those patients with electronic oral temperatures of 100°F or higher, 37% were found to have tympanic temperatures of 99.5°F or lower (usually not considered a significant temperature elevation), resulting in a sensitivity of
August 1998
only 63% when temperatures of 100°F or higher are considered fever. In conclusion, there is poor agreement between temperatures measured by tympanic thermometry and those measured by oral electronic or mercury-glass thermometers in adult patients hospitalized outside the intensive care unit. These findings have led us to recommend that tympanic thermometers not be generally used for exclusion of fever in this patient population. We thank Tammy Gillman and C. J. Manian for their diligent assistance in data processing.
References 1. Terndrup TE. An appraisal of temperature assessment by infrared emission detection tympanic thermometry. Ann Emerg Med 1992;21:1483-92. 2. Hooker EA, Houston H. Screening for fever in an adult emergency department: oral vs tympanic thermometry. South Med J 1996;89:230-4. 3. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986;1:307-10. 4. Selfridge J, Shea SS. The accuracy of the tympanic membrane thermometer in detecting fever in infants aged 3 months and younger in the emergency department setting. J Emerg Nurs 1993;19:127-30. 5. Terndrup TE. Tympanic thermometers. South Med J 1994;87:1059.
Correction
In the article entitled “Guideline for infection control in health care personnel, 1998” (Bolyard EA, Tablan OC, Williams WW, Pearson ML, Shapiro CN, Deitchman SD. AJIC Am J Infect Control 1998;26:289-354 and Infect Control Hosp Epidemiol 1998;19:407-63), the length of time that susceptible health care personnel exposed to the varicella-zoster virus should be restricted from duty was misstated. Susceptible health care personnel exposed to the varicella-zoster virus should be restricted from duty from the 10th day after the first exposure through the 21st day after the last exposure.