This study presents additional support for the use of one approximation method to screen for potential fever in young children. There was a close agreement between the pacifier and rectal temperature measurement when the 6-minute pacifier temperature was adjusted upward 0.5°F; the difference between the two measures was not statistically significant. The findings from the paired samples t test correspond closely with that found in Press and Quinn (1997), where the mean ± SD difference was -0.01°F ± 0.42°F.
Correlations can be compared with other similar studies. These have ranged from 0.62 (Beckstrand et al., 1996) to 0.78 (Hensley et al., 1999) to 0.95 (Press & Quinn, 1997). Interpretation of what constitutes an acceptable correlation varies among temperature approximation studies (Erickson & Woo, 1994; Sganga et al., 2000) despite published recommendations regarding standards of reliability (Nunnally, 1967). Such recommendations expect a correlation above .90 using applied research instruments that determine care interventions. Although statistically significant, the correlation between rectal and pacifier thermometer measurements fell short of this target; only a moderate (r=.772) relationship was determined.
The use of the correlation in temperature studies, however, must be interpreted with caution. A significant issue with temperature approximation studies is the use of correlation techniques to determine accuracy of the select measure (Shann & Mackenzie, 1996). Correlations indicate the strength of the relationship and do not necessarily provide information as to the agreement between measures (Howell, 1997). Two points on a grid, however random, constitute a perfect correlation. The correlation provides some assurance that the measures are moving in the same direction. Variability is expected when measuring temperature at two divergent locations. Based on this, the correlation between the measures does not provide a convincing argument to replace rectal temperature measurement with supralingual temperature measurement. In rare in stances, when an exact rectal temperature measurement is required, the supralingual temperature cannot be substituted. The same argument is applied to other biophysical measures such as weight, skin folds using calipers, and blood pressure. Consistency with measures, particularly site and instrument, provide more useful information to evaluate trends over time. This modest correlation therefore highlights the need to consistently measure temperature at one site to make adequate comparisons of temperature fluctuations.
For temperature screening purposes, however, these data do provide compelling and useful evidence of the relationship between the two measures as demonstrated by the paired samples t test. There was not a statistically significant difference between these measures. Also, the correlation between 3-and 6-minute pacifier measurements (0.913) provides useful evidence that this measure is stable and accurately reflects supralingual temperature. Clinical utility was also demonstrated as 88% of the rectal and pacifier temperatures were within 1.0°F - very near the goal of 90%. These data provide support to the assertion that pacifier thermometry is an acceptable method of temperature approximation in young children, particularly in the home setting, in those children who are accepting and capable of appropriately using a pacifier thermometer. The pacifier thermometers are reusable and, similar to digital rectal thermometers, can be cleaned easily in the home by wiping all surfaces with an alcohol prep pad.
Interpretations must be viewed within the limitations of this study, which involve the sample, potential researcher measurement error, and potential equipment measurement error. A convenience sample was utilized, although the within-subjects design strengthens the outcome of the data and allows for increased external validity of the data. The sample size met the requirements of the power analysis but was small and did not allow adequate comparisons between temperature levels or age groups. This was problematic when determining clinical utility based on febrile status. More specifically, even though 4% of the total sample was determined to be febrile by the rectal route and found afebrile by the pacifier route, this comprised 25% of the subjects who actually had a fever. This means that the pacifier thermometer missed 1 in 4 fevers. Obviously, there were too few subjects in the febrile group. If this trend continues in a larger febrile sample, the clinical utility of the pacifier thermometer would be questioned. Another issue with the febrile subjects was the number of subjects detected as febrile by the pacifier but not by the rectal route. If this trend continued in a larger sample, this could certainly lead to inappropriate sepsis evaluations for those patients (14%) who do not truly have a fever. Additional research is suggested to determine the efficacy with detecting fever and to compare accuracy among various age groups.
Measurement error is always an issue with any type of research. The study findings were based on one type of pacifier thermometer and cannot be generalized to all types of pacifier thermometers. Using rectal temperature measurement as the criterion standard depends upon completely accurate measurement of rectal temperatures. Although the researchers were consistently trained and followed a specific protocol, rectal temperature readings can vary with depth of insertion (Blackburn & Loper, 1992). The greatest variation is found in the first 4 to 5 cm; the temperature reading increases with greater depth of insertion. An insertion at 5cm can be up to 1.5 degrees C higher than that inserted at 1cm (Karlberg, 1949). The study protocol required insertion of 1.5 cm, the standard depth of insertion for infants (Potter & Perry, 2001). Greater depths of insertion with rigid instruments, for purposes of this study, were considered unsafe. This potential source of measurement error may have explained the large percentage of adjusted pacifier temperatures that exceeded rectal temperature measurements. Further research is needed to distinguish the potential sources of error, including depth of rectal thermometer insertion and how this impacts judgments of pacifier thermometer utility.
Findings of this research are also limited by the potential equipment errors. The rectal thermometers did undergo calibration testing prior to and after use. All calibrations indicated a functional thermometer. The pacifiers did not have a calibration feature. The pacifiers do undergo a "self-check" when activated but there was no other method to determine if the unit was functioning adequately. Of the 25 pacifiers used, one did not work properly (no digital display appeared) and was discarded and replaced. The product also required subjects to suck consistently for 6 minutes. Accurate measurement may be impeded by inability or unwillingness to use the pacifier as directed in other settings. For example, the interference of nasal congestion would force the child to breath around the device for 6 minutes. This could certainly impact the supralingual temperature reading. Also, a medium size pacifier was selected for this research. This size may not have been ideal for some of the subjects in this study and an optimal temperature measurement may only be achieved with a smaller or larger nipple. Also, if a certain size is required, it may be impractical for families to replace the thermometer with stages of growth of the child. Future research should compare pacifier thermometer sizes as a source of measurement error.
The practicality of pacifier thermometer use appears highest in the home. Cost and time are likely to prohibit use in the hospital setting. The cost of a pacifier thermometer is approximately $12-$15. Each patient would need his or her own pacifier thermometer and the measurement requires six minutes. Also, certain conditions that elicit fever, such as an upper respiratory infection, may prohibit sucking consistently for this length of time due to nasal congestion and breathing obstruction. Certainly, digital rectal thermometers are quicker and considerably less expensive. Also, a new digital rectal disposable sheath can be used to cover the thermometer probe between patients to prevent the transmission of disease. Data regarding use by parents in the home setting, comparing parental measurement of rectal versus pacifier temperature, would be valuable in further explicating the actual utility of this measurement.
Researchers, not surprisingly, have been unwilling to recommend substituting any approximation method for a rectal or other core temperature measurement when an exact temperature is required. Neither the Society of Pediatric Nurses (SPN, 2004) nor the Ameri can Academy of Pediatrics (AAP, 2004) has published position statements regarding temperature measurement in children. Although the sample size of febrile children was too small to be conclusive, the pacifier thermometer does appear to be an adequate instrument for screening for fever and gives a reasonable approximation to (but cannot re place) rectal temperature measurement. The pacifier thermometer does require the user to add 0.5°F to obtain a rectal equivalent. This may not be clear to those who do not read the package insert. A manufacturer's recalibration may be more appropriate. Other modes of temperature measurement do have a "core mode" that automatically completes the needed equivalency calculations (Erickson & Woo, 1994); this is recommended for future development of the pacifier thermometer as well.
Pediatr Nurs. 2006;32(5):413-418. © 2006 Jannetti Publications, Inc.
Cite this: Accuracy of Pacifier Thermometers in Young Children - Medscape - Sep 01, 2006.