The Accuracy and Reliability of Tympanic Thermometry Compared to Rectal and Axillary Sites in Young Children

Pediatric Nursing, May, 2000 by Lisa C. Houlder

Routine measurement of temperature in young children is usually invaluable information when assessing their degree of illness. Parents frequently perceive the presence of fever as a determinant in seeking medical attention for their children. However, obtaining a temperature in very young children can be difficult for parents, considering that the most commonly recognized sites include rectal and axillary. Although the rectal site has long been considered the "gold standard" for temperature measurement in young children (Braun, Preston, & Smith, 1998; Loveys, 1998; Rice, 1997), most parents find it invasive and time-consuming as well as anxiety provoking for both them and their children. Obtaining axillary temperature (AT) provides some advantage, but the child must be undressed and the arm held immobile for at least 5 minutes. In addition, studies have shown that ambient temperature and the shunting of blood from the skin's surface during fever results in unreliable axillary readings (Bernardo, Henker, & O'Connor, 1999).

With the development of the tympanic thermometer (TT), many of the problematic issues in obtaining rectal or axillary temperatures are avoided. Although TT offers fast, noninvasive recordings, early studies indicate many contradictory findings about its accuracy and reliability. Hooker (1993) concluded that TT is only 66% sensitive for detecting fever in children, and Ros (1989) found a poor correlation between TT and rectal thermometry (RT). Conversely; Kenney, Fortenberry, Surrat, Ribbeck, and Thomas (1990) concluded that TT provides data consistent with traditional thermometry in the pediatric population.

Clinical Question

Because patients, parents, and nurses appreciate TT for its speed, ease, cleanliness, and safety, it is important to provide them with accurate information regarding its accuracy, reliability, and concerns. Therefore, this article will review and critique the latest evidence to answer the following question: Is TT reliable and accurate compared to axillary and rectal temperature sites in young children? Medline, Cinahl, and PubMed databases were searched using the keywords of pediatric, fever, and thermometry. The statement "comparison of" and pediatrics were chosen to focus the search. Inclusion criteria further limited the search to current English articles specific to the age group 0-18 years. No report on this topic was found in the Cochrane database of systematic reviews.

The evidence. Study #1. Wilshaw, Beckstrand, Waid, and Schaalje (1999) examined the relationship between TT and axillary and rectal temperatures in infants less than 1 year of age, and provided an extensive review of the literature. A convenience sample of 120 infants was obtained: 5 infants less than 90 days old and 54 infants greater than 90 days old with fever, and 34 infants less than 90 days old and 27 infants greater than 90 days old without fever who presented to a western community clinic. Inclusion and exclusion criteria as well as length of the study were not stated. The TT, digital, and mercury thermometers used in the study were calibrated to control for instrument error. The three instruments used were: Ototemp LighTouch Pedi Q tympanic thermometer, a B-D Flexible digital thermometer for axillary temperatures, and a single mercury rectal thermometer. There was one data collector who received comprehensive training in technique and methodology. Temperatures were obtained on each subject in the order of tympanic, axillary, and rectal sites to minimize subject agitation. The children's ears were not checked for cerumen and/or otitis media based on prior research studies that concluded that these variables do not affect temperature readings. Fever was defined as a rectal temperature of greater than 99.6 [degrees] F.

In analyzing the data, a Pearson's correlation coefficient was computed to determine the relationship between temperature readings for all subjects. A Pearson's coefficient also was used to analyze the data for infants less than and greater than 90 days of age. Regression analyses were used to determine if age, fever, right or left ear sites, and gender predicted temperature readings between the thermometers. Sensitivity and specificity were calculated for fever greater than 99.6 [degrees] F axillary and 100.4 [degrees] F rectally.

A moderate correlation existed between the TT/RT (r =.47) and the AT/RT (r = .53) sites, with p values [is less than] .005. Ear side or gender did not affect the relationships. However, age and fever had significant effects on the TT/RT relationship (F = 5.26, p [is less than] .05 for age and F = 11.52, p [is less than] .001 for fever). Age had a significant effect on the AT/RT relationship (F = 10.33, p [is less than] .001). Sensitivity values for TT with RT [is greater than] 99.6 [degrees] F were moderate for all subjects, although poor for infants greater than 90 days of age. Specificity values were moderate for infants greater than 90 days and poor for those subjects less than 90 days of age for TT. The AT sensitivity was moderate and specificity weak for all classes. Sensitivity and specificity for TT when registering a RT [is greater than] 100.4 [degrees] F improved significantly (100% and moderate range respectively). The AT sensitivity remained moderate, except for those under 90 days old when it was 100% (due to one fever reading). Specificity improved from fair to moderate. The measurement of error (standard deviation) indicated that the TT and AT had a lower reliability than RT, and RT reliability was greater in infants greater than 90 days old. The TT recorded fever 96 times (80%) and the AT 86 times (72%), whereas the RT recorded fever 59 times (49%). When the RT detected low-grade fevers ([is less than] 100.2 [degrees] F), the TT did not register a fever seven times (12%) and the AT failed nine times (15%). The TT showed significantly more variation when subjects older than 90 days presented with fever. For febrile children, a one-degree change in rectal temperature reflected a one and one-half degree increase in the tympanic route. Axillary temperatures were lower in infants less than 90 days for any given RT, and the authors believe that the ability of the infant to wiggle was the cause of this result.