- Email: [email protected]
Comparison of Oral and Tympanic Temperatures in a Veterans Administration Outpatient Clinic
Comparison of Oral and Tympanic Temperatures in a Veterans Administration Outpatient Clinic IGOR Z. ABOLNIK, MD; PHILIP A. KITHAS, MD, PHD; JEROME B. SOLLER, PHD; YURII A. IZRAILEVSKY, BSe;
ABSTRACT Background: To examine the accuracy of noncontact tympanic (NCT) temperatures in outpatients, we conducted a prospective study comparing NCT temperature with temperatures obtained by oral mercury thermometers. Methods: The study included 410 patients in whom oral and NCT temperatures were obtained. Results: Mean oral temperature was 36.47 ± 0.44°C and mean NCT temperature was 36.36 ± 0.49°C. On paired-sample, two-sided t-testing, oral temperature differed significantly from NCT temperature, with a P-value < 0.0001. The difference between simultaneous oral and NCT temperatures was <:!:: 1°F; in 63 cases, oral temperature was higher than NCT temperature. Conclusion: We conclude that NCT temperature measurement is not reliable in an internal medicine outpatient clinic setting. KEY INDEXING TERMS: Tympanic temperatures; Outpatients [Am J Med Sci 1999;317(5):301-3.]
O
ral and rectal temperatures are time-proven and verified indicators of disease. 1 However, use of noncontact tympanic (NCT) thermometers is becoming more prevalent in many clinical settings because of its convenience and speed. In emergency room and ICU settings, however, NCT temperature has varying degrees of correlation with core, oral, and rectal temperatures. 2- 4 So far, several studies addressing the accuracy and clinical usefulness of From the Salt Lake City VA Medical Center, University of Utah, Salt Lake City, Utah. Submitted May 30, 1997; accepted in revised form August 14, 1998.
Correspondence: Igor Z. Abolnik, M.D., Infectious Diseases Division, Duluth Clinic, 400 E. 3rd St., Duluth, MN 55805. THE AMERICAN JOURNAL OF THE MEDICAL SCIENCES
JEFF J. MeDONNALD, MD; DONALD L. GRANGER, MD
NCT temperatures have been done, and the conclusions vary from very complimentary5-8 to unreliable. 2,4,9,10 In one study, NCT thermometers were much less sensitive then oral thermometers in the detection of fever.2 Also, because NCT measurements rely on radiant energy from the tympanic membrane, cerumen occlusion may further impede the accuracy of this method. l l Thus, despite convenience, the clinical usefulness of this technique may be questionable. Because the presence of fever changes the management of numerous conditions in ambulatory care, the accuracy of a measurement technique is not trivial. In our clinical practice, NCT temperature measurements seemed to be unreliable. To evaluate this relatively new technique, we conducted a study in patients presenting to an ambulatory internal medicine clinic. We prospectively evaluated the accuracy of NCT temperature measurements compared with oral temperature measurements with mercury thermometers. Methods
The study site was a primary-care internal medicine clinic at the Salt Lake City Veterans Administration Hospital. This clinic sees patients for both routine internal medicine care as well as urgent problems. Subjects that agreed to participate in our study were selected prospectively via convenience sampling between January 8 and May 15, 1996. When a patient arrived for a clinic visit, oral and tympanic temperatures were measured simultaneously by a nurse. Oral temperature was measured by a mercury thermometer covered by a disposable plastic sheath (Baxter), the only one available at our hospital, and positioned in the sublingual pocket for 3 minutes. Oral thermometers used in this study were validated against Fisher scientific thermometer calibrated against National Institute of Standards and Technology traceable equipment for accuracy, as measured in the water bath set at 98.6°F. NCT temperature was taken with one of two digital 301
Tympanic and Oral Temperature
Table 1. Comparison between Oral and NCT Temperatures
Group
Number of Patients
Difference between Mean Oral and NCT Temp
Age 25-65 years Age 66-88 years Total Cerumen plug No cerumen plug
187 223 410 25 179
0.130 (0.061-0.200)* 0.101 (0.038-0.163) 0.114 (0.068-0.160) 0.082 (-0.151-0.315) 0.129 (0.061-0.197)
P-Value
Correlation Coefficient (r)
°C
0.0003** 0.0016 <0.0001 0.47 0.0003
0.4814 0.4717 0.4773 0.3901 0.5394
* Value
(95% confidence intervals) ** P-values are results of paired-sample, two-sided t-test.
NCT thermometers, randomly selected by clinic nurses, calibrated to core mode temperature (model 2090; IVAC Corp.). Calibration to core mode temperature was checked monthly by a hospital technician, using the recommended method. A physician, unblinded to the temperature results, later examined the patient's external ear canals. If cerumen occluded the canal on the side where the NCT temperature was taken, and prevented direct visualization of the tympanic membrane, this was defined as cerumen occlusion. Data were analyzed using the paired-sample, twosided t-test to compare means and correlation coefficients.1 2 ,13 Data is shown as mean ± 1 standard deviation unless otherwise specified. Results
A total of 410 patients were entered into the study between January 8 and May 15, 1996. In all patients, oral and NCT temperatures were obtained simultaneously. The average age of our patients was 64.52 ± 10.83 years (range, 25 to 88 years). Only eight of our patients were women. Measurements obtained by either one of the two NCT thermometers used in this study differed from
oral temperature measurements to a similar extent. Mean oral temperature was 36.47 ± 0.44°C (97.65 ± 0.79°F), with a range of35.56 to 38.78°C. Mean NCT temperature was 36.36 ± 0.49°C (97.44 ± 0.88°F), with a range of 34.44 to 37.94 °C. On paired-sample, two-sided T-testing, oral temperature differed significantly from NCT temperature, with a P-value < 0.0001. On correlation testing, the correlation coefficient was low (r = 0.4773) (Table 1; Figure 1). On 85 occasions (20.73%), the difference between simultaneous oral and NCT temperatures was 2': 1°F; in 63 of those cases, oral temperature was higher then NCT temperature. The distribution of differences between oral and NCT temperatures is shown on Figure 2. Of 13 oral mercury thermometers used in our study checked against the Fisher thermometer, 10 overestimated the temperature of the water bath fixed at 98.6°F by 0.2°F and three overestimated the temperature by 0.3 degrees F. In 204 patients, external ear canals were checked with an otoscope. In 25 of those patients, the cerumen was blocking the view of tympanic membrane in the ear where NCT temperature was measured.
30%
101 100
25%
99 l!l 20% c:
~
~
t
98
~
c.
E
~
....
'0 15%
97
E
~
.~
~ ~
~
Q.
E
10%
~
95
5%
94
'0.... Regression 96.5
97.5
98.5
99.5
100.5
101.5
102.5
95% confid.
Diff. = Oral- Tympanic Temparature
Oral Temperature (OF)
Figure 1. Scatter plot of oral and NCT temperatures. NCT 45.927 + 0.53755 xOral. Correlation: r = 0.47734.
302
O%L..-=~~ <= -2.5 (-2,-1.5] (-1,-.5] (0,.5] (1,1.5] (2,2.5] (3,3.5] (-.5,0] (.5,1] (1.5,2] (2.5,3] > 3.5 (-2.5,-2] (-1.5,-1] (oF)
=
Figure 2. Distribution of differences between oral and NCT temperatures. May 1999 Volume 317 Number 5
Abolnik et 01
On subpopulation testing, presence of cerumen makes NCT temperature measurements even less reliable in our clinic population. We also examined the influence of age on difference between oral and NCT temperatures by subdividing our patients to groups of66 years and older or 65 years and younger. The average age was 72.54 ± 4.02 years in the older group and 54.96 ± 8.33 years in the younger group. Both groups had a very similar correlation coefficient and temperature differences between oral and tympanic temperatures. Details are shown on Table 1.
where more then 20% of measurements differed ;::::l°F in both directions (Figure 2). We believe that our conclusions relating to the inaccuracy of the NCT thermometers in our clinic populations are still valid, despite the differences between the clinical thermometers used in our study and the Fisher scientific thermometer. In conclusion, NCT temperature measurement is not sufficiently reliable in an internal medicine outpatient clinic setting. Its use in this setting should be reconsidered. Acknowledgments
Discussion
Most of the previous studies were done in an intensive care 5 ,7,S setting or on pediatric patients. 4 ,9,lo In previous studies comparing NCT thermometers and oral and/or rectal thermometers in adult outpatient populations, the results have been mixed. Some authors found tympanic temperature unreliable,2 but others considered it an efficient and reliable method. 6 However, in the study by Green et al,6 patients with perforation of the tympanic membrane, cerumen impaction, and otitis media were excluded. In our study, we tried to approximate the conditions of community clinical practice for a general internal medicine outpatient clinic. Measurements were obtained by nurses before physician contact. Of the patients that agreed to participate, none were excluded. In our study, correlation between NCT and oral temperatures was poor with r=0.4773 (Figure 1). This is much lower than the correlations reported previously (r = 0.60-0.93) between those two measurements in other patient populations. 2,6,7 Age did not seem to affect the difference between oral and NCT temperatures to a significant extent; however, the average difference between younger and older subgroups in our population was only 17.6 years. Cerumen occluding the tympanic membrane seems to make the NCT temperature measurement even less reliable; because only 25 patients had that finding, a definitive conclusion can not be derived from our study on this subject. A possible explanation for such poor correlation between oral and NCT measurements in our study is the excessive dependence of NCT temperature reading on measurement technique and ear canal configuration. The difference observed between the oral mercury thermometers used in our study and the Fisher scientific thermometer used as a criterion standard was unexpected. However, this difference was fairly uniform and unidirectional, in contrast to the difference observed between oral and NCT thermometers,
THE AMERICAN JOURNAL OF THE MEDICAL SCIENCES
We thank Sharon K. Carlson, R.N., Faye J. Anderson, R.N., Linda A. Ingles, L.P.N., Kathleen S. Coleman, R.N., Julie A. Regis, N.P., and other members of Blue Primary Care Team at Salt Lake City VA Medical Center, without whom this project could not have been carried out. References 1. Mackowiak PA, Wasserman SS, Levine MM. A critical appraisal of 98.6°F, the upper limit of the normal body temperature, and other legacies of Carl Reinhold August Wunderlich. JAMA 1992;268:1578-80. 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. Rabinowitz RP, Cookson ST, Wasserman SS, et al. Effects of anatomic site, oral stimulation, and body position on estimates of body temperature. Arch Intern Med 1996;156: 777-80. 4. Ros SP. Evaluation of a tympanic membrane thermometer in an outpatient clinical setting. Ann Emerg Med 1989;18: 1004-6. 5. Shinozaki T, Deane R, Perkins FM. Infrared tympanic thermometer: evaluation of a new clinical thermometer. Crit Care Med 1988;16:148-50. 6. Green MM, Danzi DF, Praszkier H. Infrared tympanic thermography in the emergency department. J Emerg Med 1989;7:437-40. 7. Erickson RS, Yount ST. Comparison of tympanic and oral temperatures in surgical patients. Nurs Res 1991;40:90-3. 8. Schmitz T, Bair N, Falk M, et al. A comparison of five methods of temperature measurement in febrile intensive care patients. Am J Crit Care 1995;4:286-92. 9. Freed GL, Fraley JK. Lack of agreement of tympanic membrane temperature assessments with conventional methods in a private practice setting. Pediatrics 1992;89:384-6. 10. Petersen-Smith A, Barber N, Coody DK, et al. Comparison of aural infrared with traditional rectal temperatures in children from birth to age three years. J Pediatr 1994;125: 83-5. 11. Doezema D, Lunt M, Tandberg D. Cerumen occlusion lowers infrared tympanic membrane temperature measurement. Acad Emerg Med 1995;2:17-9. 12. Statistical Analysis in S·Plus. Ver. 3.1 Statistical Sciences. Seattle: Mathsoft; 1993. 13. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986;1:307-10.
303
ABSTRACT Background: To examine the accuracy of noncontact tympanic (NCT) temperatures in outpatients, we conducted a prospective study comparing NCT temperature with temperatures obtained by oral mercury thermometers. Methods: The study included 410 patients in whom oral and NCT temperatures were obtained. Results: Mean oral temperature was 36.47 ± 0.44°C and mean NCT temperature was 36.36 ± 0.49°C. On paired-sample, two-sided t-testing, oral temperature differed significantly from NCT temperature, with a P-value < 0.0001. The difference between simultaneous oral and NCT temperatures was <:!:: 1°F; in 63 cases, oral temperature was higher than NCT temperature. Conclusion: We conclude that NCT temperature measurement is not reliable in an internal medicine outpatient clinic setting. KEY INDEXING TERMS: Tympanic temperatures; Outpatients [Am J Med Sci 1999;317(5):301-3.]
O
ral and rectal temperatures are time-proven and verified indicators of disease. 1 However, use of noncontact tympanic (NCT) thermometers is becoming more prevalent in many clinical settings because of its convenience and speed. In emergency room and ICU settings, however, NCT temperature has varying degrees of correlation with core, oral, and rectal temperatures. 2- 4 So far, several studies addressing the accuracy and clinical usefulness of From the Salt Lake City VA Medical Center, University of Utah, Salt Lake City, Utah. Submitted May 30, 1997; accepted in revised form August 14, 1998.
Correspondence: Igor Z. Abolnik, M.D., Infectious Diseases Division, Duluth Clinic, 400 E. 3rd St., Duluth, MN 55805. THE AMERICAN JOURNAL OF THE MEDICAL SCIENCES
JEFF J. MeDONNALD, MD; DONALD L. GRANGER, MD
NCT temperatures have been done, and the conclusions vary from very complimentary5-8 to unreliable. 2,4,9,10 In one study, NCT thermometers were much less sensitive then oral thermometers in the detection of fever.2 Also, because NCT measurements rely on radiant energy from the tympanic membrane, cerumen occlusion may further impede the accuracy of this method. l l Thus, despite convenience, the clinical usefulness of this technique may be questionable. Because the presence of fever changes the management of numerous conditions in ambulatory care, the accuracy of a measurement technique is not trivial. In our clinical practice, NCT temperature measurements seemed to be unreliable. To evaluate this relatively new technique, we conducted a study in patients presenting to an ambulatory internal medicine clinic. We prospectively evaluated the accuracy of NCT temperature measurements compared with oral temperature measurements with mercury thermometers. Methods
The study site was a primary-care internal medicine clinic at the Salt Lake City Veterans Administration Hospital. This clinic sees patients for both routine internal medicine care as well as urgent problems. Subjects that agreed to participate in our study were selected prospectively via convenience sampling between January 8 and May 15, 1996. When a patient arrived for a clinic visit, oral and tympanic temperatures were measured simultaneously by a nurse. Oral temperature was measured by a mercury thermometer covered by a disposable plastic sheath (Baxter), the only one available at our hospital, and positioned in the sublingual pocket for 3 minutes. Oral thermometers used in this study were validated against Fisher scientific thermometer calibrated against National Institute of Standards and Technology traceable equipment for accuracy, as measured in the water bath set at 98.6°F. NCT temperature was taken with one of two digital 301
Tympanic and Oral Temperature
Table 1. Comparison between Oral and NCT Temperatures
Group
Number of Patients
Difference between Mean Oral and NCT Temp
Age 25-65 years Age 66-88 years Total Cerumen plug No cerumen plug
187 223 410 25 179
0.130 (0.061-0.200)* 0.101 (0.038-0.163) 0.114 (0.068-0.160) 0.082 (-0.151-0.315) 0.129 (0.061-0.197)
P-Value
Correlation Coefficient (r)
°C
0.0003** 0.0016 <0.0001 0.47 0.0003
0.4814 0.4717 0.4773 0.3901 0.5394
* Value
(95% confidence intervals) ** P-values are results of paired-sample, two-sided t-test.
NCT thermometers, randomly selected by clinic nurses, calibrated to core mode temperature (model 2090; IVAC Corp.). Calibration to core mode temperature was checked monthly by a hospital technician, using the recommended method. A physician, unblinded to the temperature results, later examined the patient's external ear canals. If cerumen occluded the canal on the side where the NCT temperature was taken, and prevented direct visualization of the tympanic membrane, this was defined as cerumen occlusion. Data were analyzed using the paired-sample, twosided t-test to compare means and correlation coefficients.1 2 ,13 Data is shown as mean ± 1 standard deviation unless otherwise specified. Results
A total of 410 patients were entered into the study between January 8 and May 15, 1996. In all patients, oral and NCT temperatures were obtained simultaneously. The average age of our patients was 64.52 ± 10.83 years (range, 25 to 88 years). Only eight of our patients were women. Measurements obtained by either one of the two NCT thermometers used in this study differed from
oral temperature measurements to a similar extent. Mean oral temperature was 36.47 ± 0.44°C (97.65 ± 0.79°F), with a range of35.56 to 38.78°C. Mean NCT temperature was 36.36 ± 0.49°C (97.44 ± 0.88°F), with a range of 34.44 to 37.94 °C. On paired-sample, two-sided T-testing, oral temperature differed significantly from NCT temperature, with a P-value < 0.0001. On correlation testing, the correlation coefficient was low (r = 0.4773) (Table 1; Figure 1). On 85 occasions (20.73%), the difference between simultaneous oral and NCT temperatures was 2': 1°F; in 63 of those cases, oral temperature was higher then NCT temperature. The distribution of differences between oral and NCT temperatures is shown on Figure 2. Of 13 oral mercury thermometers used in our study checked against the Fisher thermometer, 10 overestimated the temperature of the water bath fixed at 98.6°F by 0.2°F and three overestimated the temperature by 0.3 degrees F. In 204 patients, external ear canals were checked with an otoscope. In 25 of those patients, the cerumen was blocking the view of tympanic membrane in the ear where NCT temperature was measured.
30%
101 100
25%
99 l!l 20% c:
~
~
t
98
~
c.
E
~
....
'0 15%
97
E
~
.~
~ ~
~
Q.
E
10%
~
95
5%
94
'0.... Regression 96.5
97.5
98.5
99.5
100.5
101.5
102.5
95% confid.
Diff. = Oral- Tympanic Temparature
Oral Temperature (OF)
Figure 1. Scatter plot of oral and NCT temperatures. NCT 45.927 + 0.53755 xOral. Correlation: r = 0.47734.
302
O%L..-=~~ <= -2.5 (-2,-1.5] (-1,-.5] (0,.5] (1,1.5] (2,2.5] (3,3.5] (-.5,0] (.5,1] (1.5,2] (2.5,3] > 3.5 (-2.5,-2] (-1.5,-1] (oF)
=
Figure 2. Distribution of differences between oral and NCT temperatures. May 1999 Volume 317 Number 5
Abolnik et 01
On subpopulation testing, presence of cerumen makes NCT temperature measurements even less reliable in our clinic population. We also examined the influence of age on difference between oral and NCT temperatures by subdividing our patients to groups of66 years and older or 65 years and younger. The average age was 72.54 ± 4.02 years in the older group and 54.96 ± 8.33 years in the younger group. Both groups had a very similar correlation coefficient and temperature differences between oral and tympanic temperatures. Details are shown on Table 1.
where more then 20% of measurements differed ;::::l°F in both directions (Figure 2). We believe that our conclusions relating to the inaccuracy of the NCT thermometers in our clinic populations are still valid, despite the differences between the clinical thermometers used in our study and the Fisher scientific thermometer. In conclusion, NCT temperature measurement is not sufficiently reliable in an internal medicine outpatient clinic setting. Its use in this setting should be reconsidered. Acknowledgments
Discussion
Most of the previous studies were done in an intensive care 5 ,7,S setting or on pediatric patients. 4 ,9,lo In previous studies comparing NCT thermometers and oral and/or rectal thermometers in adult outpatient populations, the results have been mixed. Some authors found tympanic temperature unreliable,2 but others considered it an efficient and reliable method. 6 However, in the study by Green et al,6 patients with perforation of the tympanic membrane, cerumen impaction, and otitis media were excluded. In our study, we tried to approximate the conditions of community clinical practice for a general internal medicine outpatient clinic. Measurements were obtained by nurses before physician contact. Of the patients that agreed to participate, none were excluded. In our study, correlation between NCT and oral temperatures was poor with r=0.4773 (Figure 1). This is much lower than the correlations reported previously (r = 0.60-0.93) between those two measurements in other patient populations. 2,6,7 Age did not seem to affect the difference between oral and NCT temperatures to a significant extent; however, the average difference between younger and older subgroups in our population was only 17.6 years. Cerumen occluding the tympanic membrane seems to make the NCT temperature measurement even less reliable; because only 25 patients had that finding, a definitive conclusion can not be derived from our study on this subject. A possible explanation for such poor correlation between oral and NCT measurements in our study is the excessive dependence of NCT temperature reading on measurement technique and ear canal configuration. The difference observed between the oral mercury thermometers used in our study and the Fisher scientific thermometer used as a criterion standard was unexpected. However, this difference was fairly uniform and unidirectional, in contrast to the difference observed between oral and NCT thermometers,
THE AMERICAN JOURNAL OF THE MEDICAL SCIENCES
We thank Sharon K. Carlson, R.N., Faye J. Anderson, R.N., Linda A. Ingles, L.P.N., Kathleen S. Coleman, R.N., Julie A. Regis, N.P., and other members of Blue Primary Care Team at Salt Lake City VA Medical Center, without whom this project could not have been carried out. References 1. Mackowiak PA, Wasserman SS, Levine MM. A critical appraisal of 98.6°F, the upper limit of the normal body temperature, and other legacies of Carl Reinhold August Wunderlich. JAMA 1992;268:1578-80. 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. Rabinowitz RP, Cookson ST, Wasserman SS, et al. Effects of anatomic site, oral stimulation, and body position on estimates of body temperature. Arch Intern Med 1996;156: 777-80. 4. Ros SP. Evaluation of a tympanic membrane thermometer in an outpatient clinical setting. Ann Emerg Med 1989;18: 1004-6. 5. Shinozaki T, Deane R, Perkins FM. Infrared tympanic thermometer: evaluation of a new clinical thermometer. Crit Care Med 1988;16:148-50. 6. Green MM, Danzi DF, Praszkier H. Infrared tympanic thermography in the emergency department. J Emerg Med 1989;7:437-40. 7. Erickson RS, Yount ST. Comparison of tympanic and oral temperatures in surgical patients. Nurs Res 1991;40:90-3. 8. Schmitz T, Bair N, Falk M, et al. A comparison of five methods of temperature measurement in febrile intensive care patients. Am J Crit Care 1995;4:286-92. 9. Freed GL, Fraley JK. Lack of agreement of tympanic membrane temperature assessments with conventional methods in a private practice setting. Pediatrics 1992;89:384-6. 10. Petersen-Smith A, Barber N, Coody DK, et al. Comparison of aural infrared with traditional rectal temperatures in children from birth to age three years. J Pediatr 1994;125: 83-5. 11. Doezema D, Lunt M, Tandberg D. Cerumen occlusion lowers infrared tympanic membrane temperature measurement. Acad Emerg Med 1995;2:17-9. 12. Statistical Analysis in S·Plus. Ver. 3.1 Statistical Sciences. Seattle: Mathsoft; 1993. 13. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986;1:307-10.
303