The study aims to assess accuracy of standard practice measurement of neonatal length compared with a gold-standard length-board technique.
The study aims to assess accuracy of standard practice measurement of neonatal length compared with a gold-standard length-board technique.
Data were obtained from a population-based, cross-sectional study of 602 term babies at Royal Prince Alfred Hospital, Sydney, Australia, in 2010. Neonatal length was measured by standard clinical practice and by a length-board (gold standard) and measurements compared. Standard growth curve percentiles were used to plot length measurements. The Bland and Altman method was used to assess agreement, and acceptable levels of agreement were set at ≤1 cm and ≤0.5 cm.
The limits of agreement were between −3.06 cm (95% CI −3.08 to −3.04) and 2.67 cm (95% CI 2.65 to 2.69). Neonates whose standard-practice length fell within 0.5 cm of the gold standard totalled 41% (241 neonates), while 59% (342) were >0.5 cm. The change in length resulted in a change in the percentile range of 53% (309) on a standard growth curve percentile. When examining neonates whose length was plotted at the extremes of percentile regions, the positive predictive value results of the standard practice compared with the gold standard were poor, with positive predictive values of 37.5%, 57.1% and 31.3% for neonates who were measured as <3rd, <10th and ≥90th percentile, respectively.
In current clinical practice, measures of neonatal length are often inaccurate, which has implications for potentially erroneous clinical care. Health-care providers should be educated on the importance of length and trained in how to measure length with the correct technique using a length-board.
Growth in the first years of life is an expression of health, nutritional status and well-being. Length, as well as weight, is a sensitive and readily measurable indicator of malnutrition and other neonatal health problems. Newborn measurement of length and weight reflects fetal nutrition and forms the basis on which future growth measurements are compared and important treatment decisions are made when they are plotted on standard growth curves. Furthermore, growth faltering and undernutrition are associated with long-term consequences including short adult height, cognitive impairments, morbidity and mortality.
Neonatal growth measurements include both weight and length. Yet when monitoring the nutritional state of neonates, emphasis is often placed on weight, while length is viewed as a secondary factor that falters after weight has faltered. In 2006, the new World Health Organization reference standards were introduced based on optimal pregnancies and predominantly breastfed babies, rather than the previously used standards, which included mostly formula-fed populations.[6, 7] As a result, more accurate weight and length measurements were described, and it was shown that rather than a change in length following a change in weight, both length and weight may falter at birth, and their patterns are not as intimately related as once thought.[1, 8-12] This recent finding emphasises the importance of accurate length and weight measurements for the early detection of growth failure and timely interventions. However, length is often measured inaccurately or ignored in clinical practice because of the perceived difficulty of measuring length and a lack of understanding as to its importance.[13, 14]
The length-board measurement has been shown to be the most reliable and accurate measurement of neonatal length,[15-17] and more recent designs have improved ease of use. However, in clinical practice it may be considered inconvenient, time consuming and more expensive than other methods such as the tape measure method.
We hypothesise that the current standard-practice method for measuring neonatal length at birth is inaccurate when compared with the use of a length-board with the correct technique. We analysed the accuracy of current measurement methods and the potential impact inaccuracies may have on clinical outcomes.
The study population was drawn from a population-based, cross-sectional study of 602 term babies born at Royal Prince Alfred Hospital, Sydney, Australia between August 2010 and October 2010. Babies were well term neonates (37–42 weeks gestation) recruited within the first 48 h of life. Ineligible neonates included those with a major congenital anomaly or those admitted to the neonatal intensive care unit for >48 h.
The neonate's length was measured in centimetres using two methods: (i) standard practice; and (ii) the gold-standard length-board method. Standard practice was defined as the measurement of a neonate's length performed by the midwife on the labour ward or in the birthing unit, usually within 6 h of birth. The method was varied; the most commonly described method involved one examiner using a non-standardised supine length-board, although on some occasions the father would hold the neonate's head while the midwife held the legs and obtained the measurement. This method was used in approximately 90% of cases. Alternatively, in 10%, a marked tape measure was used by one midwife, with the neonate placed on the side and measured from the head to the heel of the foot. Given our aim to evaluate usual protocol, the standard-practice method was not altered or monitored in any way. The offset on the length-board enabled accurate measurement of a graduated scale to 0.1 cm.
The gold-standard method was defined as the length-board measurement performed by a trained professional (one of five trained research staff, all of whom had been trained by a neonatologist (HJ) in correct anthropometric techniques) with an assistant, using the manufacturer's stipulated technique. All neonates were measured within 48 h of birth. The infant length-board used was the Easy-Glide Bearing Infantometer (Perspective Enterprises, Portage, MI, USA). The neonate was placed supine and unclothed on the board and held gently with his or her body aligned and head in a neutral position. One person held the head in contact with the headboard while another extended the left leg by placing the hand over the left knee, depressing the knee, straightening the leg and moving the footboard to touch the plantar surface of the foot at a right angle to the leg.
The standard-practice measurements were collected from the neonate's medical records, and the measurers were blinded to the study aim. Any large discrepancy between the gold-standard measurement and the standard-practice one was hand-checked against the original record. All other data were obtained from a questionnaire administered at the time of the length measurements and from the medical records of the mother and neonate. The study was approved by the Human Research Ethics Committee of Royal Prince Alfred Hospital. Informed parental written consent was obtained and participation was voluntary.
Baseline information was expressed as frequencies and percentages for categorical data; continuous data were presented as means and standard errors. Statistical significance was defined as P < 0.05.
The inter-rater reliability of the length-board measurement was tested in 44 neonates who were randomly selected to be measured twice with the length-board by two different trained professionals. The second examiners were blinded to the first measurement, and measurements within 0.5 cm of each other were considered good agreement.
Bland and Altman analysis was used to compare the length of the neonate obtained from the standard-practice method with the length obtained from the gold-standard method. The bias or mean difference between the two methods was used to determine if the standard-practice measurement, compared with the gold-standard method, under- or overestimated length and, if so, by how much.
The limits of agreement (mean difference ± 1.96 times the standard deviation of the difference) were calculated to indicate the possible extent of the variation between the two measurements. We assessed the agreement proportion between the standard practice and the gold standard as acceptable if the two measurements fell within 0.5 cm of each other. We also examined measurements within 1 cm of each other, as other authors have also considered this an acceptable difference.[13, 14, 19-21]
For both the standard-practice measurement and the gold-standard measurement, standard growth curve percentiles were determined by a computer-based percentile calculator using gestational age, neonatal sex and length. A difference in percentile regions between the two measurements was determined. Percentile regions were grouped as follows: <3rd, 3rd–10th, 10th–25th, 25th–50th, 75th–90th, ≥90th. Clinically, it is most important to correctly categorise neonates at the extremes of percentiles, as these neonates, especially at the lower end of the percentile extremes, are more likely to receive intervention. We therefore calculated the positive predictive value (PPV) of standard-practice measures against the gold standard as the reference, at the <3rd, <10th and ≥90th percentile. The PPV is the probability that a neonate who falls within a given percentile region when measured by standard practice is actually in that percentile region according to the gold standard.
Of the 602 eligible term neonates (37–42 weeks gestational age), 97% (583 neonates) had two completed length measurements, while 3% (19) did not have their length measured by standard practice and were excluded. Neonatal and maternal characteristics are presented in Table 1. The intra-class correlation coefficient of the gold-standard technique was 0.968, showing good inter-rater reliability.
|Total n = 583|
|Gestational age (weeks)||39.5 (1.2)|
|Birthweight (g)||3403 (490.1)|
|Length by gold standard (cm)||49.6 (0.2)|
|Head circumference (cm)||34.5 (1.2)|
|Maternal age (years)||32 (5.4)|
The mean difference of the two group measurements was −0.19 cm (95% CI −0.32 to −0.07). The limits of agreement were between −3.06 cm (95% CI −3.08 to −3.04) and 2.67 cm (95% CI 2.65 to 2.69), indicating that standard-practice techniques may underestimate length by 6.2% or overestimate length by 5.4% (Fig. 1). The percentage of neonates whose standard-practice length fell outside of 0.5 cm of the gold-standard measurement was 59% (342). Using a more lenient definition of acceptable agreement (1 cm), we found that 34% (199) of neonatal measurements were either over- or undermeasured. The difference in length between the two measurements resulted in 53% (309) of the 583 neonates being plotted in a different percentile range on a standard growth curve percentile. Of these 309 neonates, 77% (239) changed one percentile region, 18% (54) changed two regions, 4% (13) changed three regions and <1% (3) changed greater than three regions (Fig. 2).
The PPV test results of standard practice compared with the gold standard were poor. The PPV for neonates measured as <3rd percentile was 37.5%, for <10th percentile 57.1% and for ≥90th 31.3%.
This study, set in a large tertiary maternity hospital in Sydney, compared length measured using standard practice with that measured using a gold-standard technique. Our results show considerable inaccuracies in length measurements of neonates in standard practice. We found that 95% of length-measurement discrepancies were between −3.06 cm and 2.67 cm of the gold-standard measurement, which falls considerably out of any range of acceptable agreement. The level of agreement accepted as being accurate in similar studies varies. Some studies have argued for the limits of agreement to be between ±0.5 cm of the mean difference.[13, 19, 20] Others have examined differences of ±1 cm,[14, 21, 24] which may be more clinically appropriate. We used both error limits and found that 59% of neonates were outside the 0.5-cm margin and 34% were outside the 1-cm margin; hence, a large number of length measurements fall outside the acceptable range and would not be considered accurate in clinical practice. The differences in length measurements resulted in more than half of the neonates being plotted in an incorrect percentile range. The results of the PPV tests clearly describe this finding. For neonates identified as <3rd percentile in length, less than half were actually <3rd percentile. For neonates measured as <10th percentile, it is somewhat better at almost 60%; and for longer neonates measured as ≥90th percentile, it was 31%. These errors have significant implications for the management of these neonates immediately following their birth, and potential implications for their future growth. Neonates who should be plotted as <10th percentile but are missed may not receive the appropriate nutritional support and monitoring they need, with the potential for aberrant growth in the future. Neonates who are incorrectly plotted as <10th percentile may erroneously receive extra monitoring and feeds because they are labelled as small for gestational age; breastfeeding may be ceased inappropriately, with potential implications for maternal bonding and the neonate's future burden of disease.
These findings are consistent with the results of other studies evaluating the reliability of length measurements. There is substantial literature examining the inaccuracy of the tape-measure technique. Johnson et al. found that the reliability of measuring neonatal length using a tape measure was much less than that of four other neonatal anthropometric measurement techniques studied. In a small study, they found that when two different people used the tape-measure technique, the percentage of difference <0.5 cm ranged from 20% to 30%, and the percentage of difference <1 cm ranged from 36% to 40%. A further two papers found that differences were significantly larger when the tape-measure technique was compared with the two person length-board technique.[15, 24] A study assessing the accuracy of tape measures compared with the length-board in 25 children of less than 36 months found that inaccuracies in length resulted in a change in the percentile range in almost half of the children.
While the superior accuracy of two people using the length-board over other techniques is well documented, we found little evidence of correct technique in our clinical practice. An American study found that only 12% of practices in Philadelphia used a length-board for measuring infants' length. In our study, despite the majority of neonates being measured by a length-board, the correct two-person procedure was not always followed with subsequent inaccuracies. These inaccuracies demonstrate the importance of both correct equipment and correct technique. The need for correct technique was confirmed by Lipman et al., who showed significant improvement in length accuracy after an intervention involving intensive training.
Importantly, unlike Lipman et al., we found that training does not need to be intensive. Our own measurements, with a near-perfect intra-class correlation, were achieved by 1 h of training. We believe that with simple training in the correct technique with a standardised length-board, midwives could achieve similar accurate results. In addition, we used a SCORPIO method of teaching which incorporates, firstly, ‘tell and show’ on how to measure and then ‘do and feedback’ on correct and incorrect technique. This method has been shown to be effective in changing the behaviour of perinatal health providers and can be readily introduced with few resources.[27, 28]
The importance of length as a measure of growth faltering is now well described in the literature.[8-10] However, we postulate that this has not been well translated into practice. We suggest that many health providers do not recognise the impact of inaccurate length measurements. Lipman et al. argues that primary health-care providers are often cavalier about the need for precision of length measurements. Educating health-care workers on the importance of length measurements could increase their motivation to obtain measurements using correct techniques. The importance of length and thus the need for accuracy in measurement are two messages that should be translated into clinical practice.
The strengths of this study include the large population-based sample, the prospective data collection, the blinding of all measurement examiners to the hypothesis and the high intra-class correlation of the gold-standard measurement. A limitation of this study is that we did not witness the standard practice; rather, we sought a description from midwives of how they measured neonatal length. In doing this, we have an accurate reflection of clinical practice methods, and we can be sure that they are heterogeneous. Our findings can be generalised to health settings with similar heterogeneous standard practices. Less-resourced areas such as regional and rural centres may be less likely to use accurate length-boards and have less access to training.
This study demonstrates that inaccurate length measurements can lead to incorrect plotting of percentiles for neonates. Any errors in length may result in errors in plotted length for weight measurements. Length at birth is an important immediate indicator of post-natal growth faltering,[8-10] and it is also the baseline against which future length measurements are compared and used for clinical decisions. Incorrect length measurement may lead to an inability to detect growth failure and underlying disorders, or the inappropriate referral of a normally growing neonate, with negative consequences for breastfeeding.
We recommend that health-care providers involved in measuring neonatal length should be aware of the importance of length and trained in how to accurately measure length with two people using the length-board. A possible feasibility issue may be the need for two people to obtain a length measurement. However, once the initial training is implemented, using a length-board is cheap, easy and accurate, and we believe that the benefits obtained from an accurate length measurement outweigh the time costs of doing so. To ensure standards are maintained, a systematic approach involving education, a standardised protocol describing the correct technique and re-evaluation should also be implemented. We suggest that future research could audit such protocols and evaluate whether accuracy with training persists over time.
We thank the parents and newborns who were so agreeable to this study and the Sydney Medical students (Elizabeth Hayles, Erin Donnelley, Lucia Wang, Cheryl Au) who so willingly assisted with the project. We would also like to thank the staff of Royal Prince Alfred Women and Babies hospital for their support and Kevin McGeechan for statistical assistance. AC is supported by an APA PhD award. CRG is supported by NHMRC Early Career Fellowship #511481.