Reliability of Three Length Measurement Techniques in Term Infants.
To be accurate, infant length measurements must be reliable. Reliability is defined as the accuracy with which a measurement can be repeated (Rosenberg, Verzo, Engstrom, Kavanaugh, & Meier, 1992). Intraexaminer reliability is the ability of one examiner to obtain similar or identical measurements on separate occasions when measurements are obtained under the same conditions (Rosenberg et al., 1992). Interexaminer reliability is the ability of two or more clinicians to obtain similar or identical measurements when the measurements are obtained under the same conditions (Rosenberg et al., 1992).
Numerous length measurement techniques have been described in the literature, but the reliability of these techniques has not been described and compared in sufficient detail to determine whether one measurement technique is more reliable than the others. The purpose of this study was to describe and compare the intra- and interexaminer reliability of length measurements in term infants, obtained using three different length measurement techniques: supine, paper barrier, and Auto-length[TM] measuring device (Olympic Medical; Seattle, WA). An additional purpose of this study was to determine if there was a significant difference between the length measurements obtained by the three different measurement techniques.
Review of Literature
The reliability of infant length measurements has been studied by numerous investigators and the results of those studies have been summarized elsewhere (Johnson, Engstrom, Warda, Kabat, & Peters, 1998; Johnson et al., 1997). Although the reliability of infant length measurements has been reported in numerous studies, the statistics used to describe the results have often not been reported in sufficient detail for clinicians to determine whether a measurement technique produces measurements sufficiently reliable for clinical practice (Johnson et al., 1997). Additionally, infant length measurement techniques have not been compared adequately to determine whether one technique yields measurements that are more reliable than other measurement techniques.
Two studies have described and compared the reliability of infant length measurements in term infants in sufficient detail to allow for comparison (Johnson et al., 1997; Johnson et al., 1998).
In the first study (Johnson et al., 1997), the reliability of infant length measurements obtained using the supine measurement technique was studied as a part of a larger study that described the reliability of six infant anthropometric measurements: length, weight, head circumference, chest circumference, abdominal circumference, and mid-arm circumference. A convenience sample of 50 term infants was studied. Two examiners obtained each of the six measurements twice and were blind to their own and each other's measurements. The mean absolute intraexaminer differences for length measurements for examiners one and two were 0.92 and 1.18 cm, respectively. The percentage of differences [is less than] 1 cm for examiners one and two were 58% and 60%, respectively. The mean absolute interexaminer differences of length measurements for sets one and two were 1.57 and 1.47 cm, respectively. The percentage of differences [is less than] 1 cm for sets one and two were 36% and 40%, respectively. The findings from this study demonstrated that the differences within and between examiners' length measurements were much larger than anticipated. The reliability of infant length measurements was also remarkably lower than the reliability of the other five infant anthropometric measurements studied.
Based on the results of the above study, our research team designed and completed another study (Johnson et al., 1998) to describe and compare the intra- and interexaminer reliability of four different length measurement techniques: crown-heel, supine, paper barrier, and Neo-infantometer[TM]. A convenience sample of 32 clinically stable, term infants was studied. Two examiners obtained each of the measurements twice using the four measurement techniques. The examiners were blinded to their own and each others' measurements. The mean absolute intraexaminer differences for examiners one and two, respectively, were as follows: crown-heel = 1.22, 1.24 cm; supine = 0.80, 0.53 cm; paper barrier = 0.62, 0.56 cm; and Neo-infantometer = 0.50, 0.71 cm. The mean absolute interexaminer differences for sets one and two, respectively, were as follows: crown-heel = 1.49, 1.38 cm; supine = 0.74, 0.84 cm; paper barrier = 0.68, 0.57 cm; and Neo-infantometer = 0.81, 0.61 cm. The reliability of length measurements obtained by the crown-heel technique was significantly lower than the reliability of measurements obtained by the other three measurement techniques. Although the reliability of measurements obtained by the supine, paper, and Neo-infantometer technique was greater than the reliability of measurements obtained by the crown-heel technique, the amount of error observed in measurements obtained by these techniques was still larger than is optimal for clinical practice and research.
The findings of these studies suggest that further research should be conducted to determine if more reliable methods of measuring infant length in term infants can be identified.
Sample and setting. A convenience sample of 48 term infants was selected from a Level 1I nursery in a medium-sized hospital in a midwestern city in the United States. Criteria for sample selection were as follows: gestational age [is greater than or equal to] 37 weeks, as determined by the mother's last menstrual period or prenatal ultrasound and confirmed by clinical examination (Ballard et al., 1991); postnatal age [is less than] 72 hours; axillary temperature [is greater than or equal to] 97.6 [degrees] F; no supplemental oxygen; no equipment attached to the infant; no obvious physical anomalies; and infant status not classified as "critical." Verbal consent for participation in the study was obtained from the infant's mother.
Each examiner attempted to obtain two complete sets of measurements using each measurement technique for each infant. However, a malfunction in the Auto-length device resulted in some missing measurements. Examiner 1 had four missing sets (three Auto-length measurements for each set) and examiner 2 had two missing sets of measurements. Instruments. Two instruments were used. The first, paper tape measures, were prepared for the supine and paper barrier length measurement techniques. The tape measures were prepared from white, laminated, unmarked paper and were 76.25 cm (30 in) x 1.27 cm (.5 in). Each tape measure was marked with the following information: subject's study number, examiner's initials, set number, and measurement technique.
The second instrument used was the Olympic Auto-length (Olympic Medical; Seattle, WA). This is an electronic length measurement device that is attached to an Olympic Smart[TM] scale (Olympic Medical, Seattle, WA). The infant is positioned in the tray with his or her head touching the end of the tray. First, the examiner pressed the on button and then the measure button. The measuring bar then touched the infant's heel three times gently (with 2 ounces of pressure), three measurements were displayed in 12 seconds, and those measurements are accurate within 2 mm (Olympic Medical product brochure). The longest of the three measurements was used as the infant's length as recommended by the manufacturer (Olympic Medical product brochure).
Procedures. Prior to the onset of data collection, the order of examiners for the length measurements was determined by using a fixed randomization plan, so that each possible combination of the order of techniques and examiners was represented an equal number of times. Data collection envelopes were prepared for each infant with the infant's identification number and the order of the examiners' measurements. The tape measures were marked with the infant's study identification number, examiner's initial, set number, and measurement technique. Four color-coded index cards were prepared to record the measurements obtained with the Auto-length and were placed in the data collection envelope. Institutional Review Board approval was obtained from both the University and participating hospital before data collection was initiated.
At the time data collection was initiated, infant hospital records were reviewed to identify infants that met the criteria for sample selection. When infants met this criteria, the purpose of the study was explained to the infant's mother and her verbal consent for her infant to participate in the study was obtained. The measurements were then obtained in consecutive sets using the predetermined order.
After data collection was completed on the infant, demographic, anthropometric, and obstetric data from the mothers' and infants' hospital records were recorded on an index card and placed in the data collection envelope. This information included (a) gestational age by dates, ultrasound, and clinical examination; and (b) birth weight.
To protect the infants from hypothermia, all length measurements were obtained while the infant was under a Warmth without Light[TM] heat lamp (Olympic Medical, Seattle, WA).
The two length measurements (supine and paper barrier) were chosen to compare to the Auto-length measuring device (a new length measurement device just placed on the market) because they are commonly used in the clinical setting. The procedures used for obtaining infant length measurements were as follows:
Supine length. The first examiner placed the infant supine on top of the tape measure in the tray on the infant scale and gently held the infant's head in a neutral position with the top of the head even with the zero point on the tape. The infant's body was kept in alignment with the tape measure, with the legs extended and the feet perpendicular to the legs (Fletcher, 1994). The first examiner creased the tape measure at a point even with the base of the infant's heel and then handed the tape measure to the research assistant.
Paper barrier length. The first examiner !aid the infant on a piece of scale paper. The infant's head was held gently in a neutral position with the body supine, the legs extended, and the feet perpendicular to the legs (Fletcher, 1994). After the infant was positioned, the first examiner marked the point at the uppermost part of the infant's head and at the base of the infant's right heel using a pen held at a 90 [degrees] angle. Then, after the paper barrier was removed, the first examiner used a blank tape to measure the distance between the head and heel marks (Rosenberg et al., 1992; Wong, 1995). The first examiner creased and marked the tape at a point even with the mark from the base of the infant's heel, then handed the tape measure to the research assistant.
Olympic Auto-length measuring device (Olympic Medical; Seattle, WA). The digital display on the device was covered with a notecard. The examiner placed the infant supine in the tray, holding the infant gently in a midline position with the body aligned, head in a neutral position (neither extended or flexed), legs extended, and feet perpendicular to the legs. The head was held in contact with the end of the tray. The examiner pressed the on/off button and then the measure button. While the examiner held the infant in the position described above, the measuring bar moved towards the infant's heel, touched the infant's heel three times, and then displayed the three measurements. After the measurement process was completed, the research assistant lifted the note card from the digital display and recorded the three measurements. The examiner was blinded to the digital display throughout the measuring process.
Only one examines positioned the infant and obtained the length measurement. The reason for this is twofold. First, the Auto-length was designed so that one person could obtain length measurements. Second, although several authors recommend that two people obtain length measurements (Doull, McCaughey, Bailey, & Betts, 1995; Wong, 1995), this study was designed to replicate clinical practice. In clinical practice, in nurseries, labor and delivery settings, and busy clinics, only one person is available to take infant length measurements.
After all measurements were obtained twice from each infant, the research assistant placed the tape measures and index cards in the data collection envelope, and the data collection envelopes were sealed until data collection from all 48 infants was completed. Then the tapes were measured using a standard metal ruler. Each tape was measured separately by two researchers who were blind to the birth measurements and to each other's measurements. If the measurements recorded by the two researchers differed by more than 1.0 mm, the tape was to be remeasured by a third researcher experienced with anthropometric measurements. The measurement closest to the third measurement served as the final measurement. This occurred only a few times.
Analysis. The statistical analysis included statistics appropriate for describing the reliability of physical measures such as mean absolute differences, standard deviations of the net differences, technical error of measurement, minimum and maximum differences, percentage of differences within a critical value (Engstrom, 1988, Utermohle, Zegura, & Heathcote, 1983), and the percentage of error (Johnson et al., 1997). A Type I error of 5% was used for all tests of statistical significance.
Intraexaminer reliability. The differences between the individual nurses' length measurements obtained using the three length measurement techniques are described in Table 1. RN-1 had four sets (3 measurements each) of measurements missing for the Auto-length and RN-2 had two sets missing for the Auto-length. Thus, RN-1 had four Auto-length measurements missing, and RN-2 had two Auto-length measurements missing in the analysis.
Table 1. Intraexaminer Reliability of Three Infant Length Measurement Techniques (in cm) (N=48) Paper Auto- Statistic Supine barrier Length Mean absolute difference RN-1 0.92 0.84 0.60(a) RN-2 0.74 0.81 0.84(b) Standard deviation RN-1 1.17 1.01 0.85 RN-2 0.99 1.28 1.06 Technical error of measurement RN-1 0.82 0.73 0.63 RN-2 0.69 0.89 0.75 Maximum difference RN-1 3.4 2.6 3.0 RN-2 2.6 6.5 3.1 Percentage of differences [+ or -] .5 cm RN-1 39.6 39.6 59.1 RN-2 52.1 47.9 34.8 Percentage of differences [+ or -] 1.0 cm RN-1 64.6 66.7 84.1 RN-2 77.1 75.0 78.3 Percentage of error RN-1 1.81 1.68 1.20 RN-2 1.46 1.61 1.73
(a) n = 44
(b) n = 46
RN-1 had a smaller mean absolute difference for measurements obtained using the Auto-length technique. RN2 had a slightly smaller mean absolute difference for measurements obtained using the supine technique. The mean absolute differences between each nurse's first and second measurements obtained with the three length measurement techniques were compared using a Friedman's chi square tests. The differences were statistically significant for RN-1 ([X.sup.2] = 8.33, df = 2, p = .0155) but not for RN-2 ([X.sup.2] = 0.62, df = 2, p. = .7336). For RN-1's differences, posteriori comparisons were performed using Wilcoxon matched-pairs, signed-ranks tests corrected for multiple comparisons using Bonferroni's correction. These analyses demonstrated statistically significant differences between the absolute differences of measurements obtained with the paper barrier and the Auto-length measurement techniques (Z = -2.59, p = .0095) and between the supine and Auto-length measurement techniques (Z = -2.41, p = .0161), but not between the supine and paper barrier techniques (Z = -.04, p = .9711). Similarly, the number of differences less than or equal to 1 cm for the three measurement techniques were significantly different for RN-1 (Cochran Q = 6.95, df = 2, p = .0309) but not for RN-2 (Cochran Q = .23, df = 2, p = .89 ] 0).
Interexaminer reliability. The differences between the pairs of length measurements obtained by the two nurses using the three length measurement techniques are described in Table 2. The mean absolute differences for the first set of measurements were similar, but were smallest for the Auto-length technique. The mean absolute differences for the second set of measurements were largest for the supine length measurement technique.
Table 2. Interexaminer Reliability of Three Infant Length Measurement Techniques (in cm) (N=48) Paper Auto- Statistic Supine barrier Length Mean absolute difference Set-1 1.13 1.18 1.02(a) Set-2 1.39 0.77 0.82(b) Standard deviation Set-1 1.09 1.66 1.02 Set-2 1.36 1.04 1.05 Technical error of measurement Set-1 1.01 1.20 0.87 Set-2 1.19 0.74 0.77 Maximum difference Set-1 4.3 7.1 3.0 Set-2 3.6 4.0 3.7 Percentage of differences [+ or -] .5 cm Set-1 31.3 35.4 26.7 Set-2 18.8 50.0 45.5 Percentage of differences [+ or -] 1.0 cm Set-1 56.3 58.3 55.6 Set-2 50.0 72.9 70.5 Percentage of error Set-1 2.24 2.33 2.05 Set-2 2.74 1.54 1.66
(a) n = 45
(b) n = 44
The mean absolute differences between the pairs of nurses' measurements obtained using the three length measurement techniques were compared using Friedman's chi square tests. The differences were not statistically significant for the first set of measurements ([X.sup.2] = .14, df = 2, p = .9303), but were statistically significant for the second set of measurements ([X.sup.2] = 14.56, df = 2, p = .0007). For the second set of measurements, a posteriori comparisons were performed using Wilcoxon matched-pairs, signed-ranks tests corrected for multiple comparisons using Bonferroni's correction. These analyses demonstrated statistically significant differences between the supine and paper barrier techniques (Z = -3.64, p = .0003) and the supine and auto- length measurement techniques (Z = -2.94, p = .0033), but not between the paper barrier and Auto-length techniques (Z = .72, p = .4688). The number of differences less than or equal to I cm for the three measurement techniques was not significantly different for the first (Cochran Q = .27, df = 2, p = .8752) or second (Cochran Q = 4.56, df = 2, p = .1023) set of measurements.
Comparison of measurements obtained using the three measurement techniques. The means for each technique was calculated to determine if different length measurement techniques yielded significantly different determinations. The mean results for supine, paper-barrier, and Auto-length measurements were 50.88, 50.33, and 49.67 cm, respectively. The measurements were compared using a nonparametric analysis of variance for dependent measures (Friedman two-way ANOVA). The results demonstrated that the mean measurements were significantly different ([X.sup.2] = 56.56, df = 2, p = .00). Posteriori comparisons were performed using Wilcoxon matched-pairs, signed-ranks tests corrected for multiple comparisons using Bonferroni's correction. These analyses demonstrated that all of the comparisons were statistically significant (paper barrier vs. supine: Z = -4.93, p = .00; paper barrier vs. Auto-length: Z = -5.11, p = .00; supine vs. Auto-length: Z = -5.62, p = .00).
The results of this study demonstrated that the mean absolute differences between the length measurements of individual examiners and pairs of examiners are relatively large, with a remarkable number of the differences exceeding 1.0 cm. These findings are consistent with the results of previous studies evaluating the reliability of length measurements (Johnson et al., 1997; Johnson et al., 1998) and suggest that it is difficult to obtain reliable length measurements in term infants. This may be due in part to the natural flexion of the newborn infant and the cranial molding that takes place during the labor and delivery process. These findings suggest that clinicians should be aware of the magnitude of error in length measurements and interpret length measurements with caution.
One promising finding from this study was that the length measurements obtained with the Auto-length resulted in significantly more reliable measurements for one nurse. This is consistent with the previous study (Johnson et al., 1998) where one nurse had smaller mean absolute differences with the paper-barrier measurement technique. This finding suggests that individuals may be able to modify or perform the procedures in a particular length measurement technique in some way to make their measurements more reliable. These procedures should be studied further to determine if measurement procedures or techniques can be used to obtain more reliable length measurements. Future research should also attempt to identify factors that may interfere with obtaining reliable length measurements.
The findings from this study also demonstrate that length measurements obtained using different measurement techniques produce significantly different measurements. This finding is consistent with those of a previous study that compared measurements obtained using four different length measurement techniques: (a) crown-heel, (b) supine, (c) paper barrier, and (d) Neo-infantometer (Johnson et al., 1998). Thus, clinicians should consistently use one measurement technique to obtain length measurements, and all clinicians in the same setting should use the same length measurement technique.
Ballard, J.L., Khoury, J.C., Wedig, K., Wang, L., Eilers-Walsmann, B.L., & Lipp, R. (1991). New Ballard score expanded to include extremely premature infants. Journal of Pediatrics, 119, 417-423.
Doull, J.M., McCaughey, E.S., Bailey, B.J.R., & Betts, P.R. (1995). Reliability of infant length measurement. Archives of Diseases in Children, 72, 520-521.
Engstrom, J.L. (1988). Assessment of the reliability of physical measures. Research in Nursing and Health, 11, 383-389.
Fletcher, M.A. (1994). Physical assessment and classification. In G.B. Avery, M.A. Fletcher, & M.G. MacDonald (Eds.), Neonatology: Pathophysiology and management of the newborn (4th ed.) (pp. 269-288). Philadelphia: J.B. Lippincott Company.
Johnson, T.S., Engstrom, J.L., & Gelhar, D. (1997). Intra- and interexaminer reliability of anthropometric measurements of term infants. Journal of Pediatric Gastroenterology and Nutrition, 24, 497-505.
Johnson, T.S., Engstrom, J.L., Warda, J.A., Kabat, M., & Peters. B. (1998). Reliability of length measurements in term infants. Journal of Obstetric, Gynecologic, & Neonatal Nursing, 27, 270-276.
Rosenberg, S.N., Verzo, B., Engstrom, J.L., Kavanaugh, K., & Meier, P.P. (1992). Reliability of length measurements for preterm infants. Neonatal Network, 11, 23-27.
Utermohle, C.J., Zegura, S.L., & Heathcote, G.M. (1983). Multiple observers, humidity, and choice of precision statistics: Factors influencing craniometric data quality. American Journal of Physical Anthropology, 61, 85-95.
Wong, D.L. (1995). Nursing care of infants and children (5th ed.). St. Louis: Mosby-Year Book, Inc.
Teresa S. Johnson, PhD, RN, is Associate Professor, School of Nursing, University of Wisconsin-Milwaukee, Milwaukee, WI.
Janet L. Engstrom, PhD, CNM, is Associate Professor, Department of Maternal Child Nursing, University of Illinois at Chicago, Chicago, IL.
Sharon L. Haney, MS, CNM, is a Certified Nurse Midwife, Advanced Health Care for Women, Inc., Downers Grove, IL.
Sandra L. Mulcrone, MS, CNM, is Certified Nurse Midwife, University of Illinois at Chicago, Chicago, IL.
Note: This study was partially funded by a grant from Olympic Medical, Seattle, WA.
|Printer friendly Cite/link Email Feedback|
|Author:||Johnson, Teresa S.; Engstrom, Janet L.; Haney, Sharon L.; Mulcrone, Sandra L.|
|Date:||Jan 1, 1999|
|Previous Article:||Exploring Language About Families.|
|Next Article:||Weight Change of Infants, Age Birth to 12 Months, Born to Abused Women.|