- Email: [email protected]
Mid-arm circumference can be used to estimate children's weights
Resuscitation 81 (2010) 1105–1110
Contents lists available at ScienceDirect
Resuscitation journal homepage: www.elsevier.com/locate/resuscitation
Clinical paper
Mid-arm circumference can be used to estimate children’s weights夽 G.N. Cattermole a , P.Y.M. Leung b , P.S.K. Mak a , C.A. Graham a , T.H. Rainer a,∗ a b
Accident and Emergency Medicine Academic Unit, Chinese University of Hong Kong, Hong Kong Faculty of Medicine, University of Melbourne, Melbourne, Australia
a r t i c l e
i n f o
Article history: Received 5 January 2010 Received in revised form 20 April 2010 Accepted 22 May 2010 Keywords: (MeSH) Pediatrics Resuscitation Body weights and measures
a b s t r a c t Introduction: Accurate measurement of children’s weight is rarely possible in paediatric resuscitation, and rapid estimates are made to ensure appropriate drug and fluid doses and equipment selection. Weight is commonly estimated from formulae based on children’s age, or from their height using the Broselow tape. Foot-length and mid-arm circumference have also been suggested as the basis of weight-estimation formulae. Objectives: To determine which of age, height, foot-length or mid-arm circumference had the strongest relationship with weight in healthy children, to derive a simple weight-estimation formula from the strongest correlate, and to compare its performance with existing weight-estimation tools. Methods: This was a population-based prospective observational study of Hong Kong Chinese children aged 1–11 years old last birthday. Weight was measured to the nearest 0.2 kg; height, foot-length and midarm circumference to the nearest 0.1 cm. Multiple regression analysis was used to determine the strongest independent relationships with weight, and linear regression analysis derived a weight-estimation formula. Accuracy and precision of this formula were compared with standard age-based and height-based weight-estimation methods. Results: Mid-arm circumference had the strongest relationship with weight, and this relationship grew stronger with age. The formula, weight [kg] = (mid-arm circumference [cm] − 10) × 3, was at least as accurate and precise as the Broselow method and outperformed the age-based rule in school-age children, but was inadequate in pre-school children. Conclusion: This weight-estimation formula based on mid-arm circumference is reliable for use in schoolage children, and an arm-tape could be considered as an alternative to the Broselow tape in this population. © 2010 Elsevier Ireland Ltd. All rights reserved.
1. Introduction In paediatric resuscitation, it is necessary to know the child’s weight in order to provide appropriate drug and fluid doses, equipment selection and ventilator settings. Because measurement of weight itself is rarely possible in time-critical situations, and because there is often no one available who knows the child’s weight, rapid and accurate methods of estimation need to be
Abbreviations: kg, kilogramme; cm, centimetre; MAC, mid-arm circumference; PEM, protein energy malnutrition; APLS, Advanced Paediatric Life Support. 夽 A Spanish translated version of the abstract of this article appears as Appendix in the final online version at doi:10.1016/j.resuscitation.2010.05.015. ∗ Corresponding author at: Accident and Emergency Medicine Academic Unit, Chinese University of Hong Kong, Rooms 107/113, Trauma and Emergency Centre, Prince of Wales Hospital, Shatin, New Territories, Hong Kong. Tel.: +852 2632 1033; fax: +852 2648 1469. E-mail addresses: [email protected] (G.N. Cattermole), [email protected] (P.Y.M. Leung), [email protected] (P.S.K. Mak), [email protected] (C.A. Graham), [email protected] (T.H. Rainer). 0300-9572/$ – see front matter © 2010 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.resuscitation.2010.05.015
applied. Most commonly, weight can be estimated from formulae based on the child’s age,1 or from the child’s height using the Broselow tape.2 Other suggested methods include weight estimation according to foot-length, or mid-arm circumference (MAC).3–5 Neither is currently used in clinical practice for weight estimation in paediatric resuscitation. However, a formula including MAC and knee height has recently been validated in geriatric patients in the emergency department.6 MAC has also been used for many years in the assessment of malnutrition in the developing world.7 One of the most widely used age-based estimation method is that recommended in the UK-based Advanced Paediatric Life Support course1 : weight in kg = 2 × (age in years + 4). It has been criticised for under-estimating children’s weight, and new formulae have been derived.8,9 The Broselow tape has consistently been found to outperform age-based formulae in estimating weight,8,10,11 but both the Broselow tape and age-based weight estimation methods are less precise in older and heavier children.2,8,10,12,13 Systematic underestimation of childhood weight is likely to be due to the increasing actual weights, especially
1106
G.N. Cattermole et al. / Resuscitation 81 (2010) 1105–1110
of older, Western children. The US National Health and Nutrition Examination survey has demonstrated a steady increase in childhood obesity: data from 1999 to 2002 found that 16% of 6–19 year olds, and 10.2% of 2–5 year olds, were overweight (≥95th centile body mass index).14 In 6–18 year olds in Hong Kong, the proportion increased from 7.1% to 10.1% between 1993 and 2006.15 It is not known whether a weight-estimation method based on foot-length or MAC might be more appropriate than either the Broselow tape or age-based methods, especially in older children. The objectives of this study were to determine which of age, height, foot-length or MAC had the strongest relationship with weight in healthy children, to derive a simple weight-estimation formula from the strongest correlate, and to compare its performance with existing weight-estimation tools. 2. Methods This was a population-based observational study, part of the prospective “Healthy Children’s Vital Signs and USCOM study”, which also included physiological and ultrasound cardiac output monitor (USCOM) measurements. It was conducted in primary schools and kindergartens in Hong Kong, and recruited healthy Chinese children aged 1–11 years on their last birthday. Head teachers of all relevant institutions in the Shatin area of Hong Kong were asked for permission to have their schools participate in the study. Classes were selected by schools to provide a representative distribution of age, and parents of children in participating classes were sent an information pack and consent form. Children with any chronic or current illness, or taking any medications, were excluded. Pragmatic language and cultural barriers precluded gathering information on parental reasons for exclusion or failure to consent. Weight was measured to the nearest 0.2 kg using electronic scales (Compact Precision Scale C200H, Conair Far East Ltd.). Height was measured with to the nearest 0.1 cm with a stadiometer (Harpenden Portable Stadiometer, Holtain, UK). Foot-length was measured from the right heel to the tip of the hallux, with the child supine and the ankle at 90◦ with the foot flat against the base of a rigid box, the heel resting on the inside wall of the box. MAC was measured with the child’s right arm relaxed in 90◦ of flexion at the elbow. The olecranon and acromion were identified, and an inelastic tape-measure was extended along the arm between these two points. The mid-point was marked and the tape was wrapped around the arm, taking care to ensure that the tape lay flat against the arm without pinching the underlying skin. Children were measured without shoes, wearing the lightest school uniform including socks. School uniforms were weighed separately and subsequent adjustments made to the measured weight of the child. 2.1. Statistical analysis LMS Chartmaker Pro v2.3 software (Cole and Pan, Medical Research Council UK, 2006) was used to describe the data in centile curves (2.5, 10, 50, 90, and 97.5). The relationship between weight and each of the four variables (age, height, foot-length and MAC) were modeled by the LMS method of Cole and Green.16 Briefly, the relationship is described by three age-specific cubic spline curves known as L, M and S. M represents the median, S is the coefficient of variation, and L is the Box–Cox transformation that renders the data to follow a normal distribution, conditional on age. Combination of these three functions generates centile values for each parameter. Analysis was weighted according to the size of gender groups. MedCalc v10.4 (Schoonjans, 2009) was used for all subsequent statistical analysis.
Pearson’s correlation coefficients were calculated for each of the four variables with weight, and stepwise multiple regression analysis was performed to determine the variable with the strongest independent relationship. Linear regression analysis was used to define a simple linear equation relating this variable with weight. Three methods of weight estimation were then compared: the newly derived linear formula, the APLS age-based formula, and height-related weights as defined on the Broselow tape. The Broselow tape itself was not used: we used the height measurements obtained with the stadiometer and cross-referenced these with the markings on the 1998 version of the tape. It was expected that a large proportion of older children would exceed the height limit of the Broselow tape (143.6 cm). For these children, we assumed a weight equal to that of a medium-sized adult, which in 18-year-old Hong Kong Chinese is 55 kg.17 As an alternative approach in the oldest group of children, we also compared the new formula with the Broselow tape only in the group of children who were within the height limits of the tape. Three methods of comparison were used to assess the accuracy and precision of each method of estimation. Firstly, coefficient of variation was calculated as the standard deviation of the differences between estimated and actual weights, divided by the overall mean of estimated and actual weights.18 Secondly, Bland Altman plots19 determined the mean bias and the 95% limits of agreement. The bias indicates the mean percentage difference between estimated and actual weight, and the limits of agreement indicate in what range 95% of the differences between estimated and actual weights will fall. Thirdly, we determined the proportion of estimates that were within 10%, 20% and 30%, or more than 30% different from, the actual weights. Chi-squared tests were used to compare these proportions between the weight-estimation methods. Three age-groups were defined based on those in the Advanced Trauma Life Support course20 in order to create approximately equal subgroups. Toddlers (1–2 years old last birthday) and preschool children (3–5 years old last birthday) were grouped together as “1–5”. School age children (6–11 years old last birthday) were split into two groups, “6–8” and “9–11”. 2.2. Ethics Ethical approval was obtained from the Clinical Research Ethics Committee of the Chinese University of Hong Kong. Written parental consent was obtained for all subjects more than a week prior to the school visit. Children who were unwilling to participate on the day of the study were not included. 3. Results 1391 Chinese subjects aged 1–11 years on their last birthday were recruited from 14 schools and kindergartens. In the six institutions which recorded how many consent letters were distributed, 48% of invited parents consented. The time constraints of school timetables prevented us from obtaining all four anthropometric measurements in 21 children, who were excluded from further analysis. 1370 (98.5% of the eligible sample) were therefore included, of whom 55% were boys. There were 448 children in the 9–11 year-group, 473 in the 6–8 year-group, and 449 in the 1–5 year-group. The LMS-derived centile curves for age, height, foot, MAC are shown in Fig. 1. In contrast to the other three variables, weight-forMAC is near linear with less spread. Correlation coefficients (r) for each variable with weight are shown in Table 1 with 95% confidence intervals. All individual correlations were statistically significant (p < 0.0001). p-Values are given in the table for each variable after multiple regression. The
G.N. Cattermole et al. / Resuscitation 81 (2010) 1105–1110
1107
Fig. 1. Centile curves of weight for age, height, foot-length and MAC (10th, 25th, 50th, 75th, and 90th). .
coefficient of determination (R2 ) is given for each multiple regression model, with p < 0.001 for all models. In all age-groups, weight is more strongly correlated with MAC than with any of the other three variables, although this is not statistically significant in the 1–5 year-group. The strength of the correlation of weight with MAC increases with increasing age, but the strengths of the correlations of weight with age, height and footlength decrease with increasing age. After multiple regression, age is an important correlate with weight only in the 1–5 year-group, and only height and MAC remain important in the oldest group. The strongest correlate with weight, both on independent correlation and after multiple regression, is MAC.
By linear regression the following formula was obtained for weight (kg) and MAC (cm): weight = (2.94 × MAC) − 29.14. For pragmatic simplicity, this was approximated and rearranged to: Weight = (MAC − 10) × 3 This weight-estimation formula was then compared with the Broselow height and the APLS age-rule methods (Table 2). The heights of 171 children were outside the limits of the Broselow tape, all but two of whom were in the oldest group. By any method of comparison, the APLS age-rule was significantly worse than either of the other weight-estimation methods, except in
Table 1 Correlation and multiple regression. Year-group
n
Overall
1370
1–5 years last birthday
6–8 years last birthday
9–11 years last birthday
449
473
448
Pearson’s correlation, r (95%CI) Multiple regression coefficient (p-value) Pearson’s correlation, r (95%CI) Multiple regression coefficient (p-value) Pearson’s correlation, r (95%CI) Multiple regression coefficient (p-value) Pearson’s correlation, r (95%CI) Multiple regression coefficient (p-value)
Age
Height
Foot-length
MAC
0.80 (0.78–82)
0.88 (0.87–0.89)
0.87 (0.85–0.88)
0.91 (0.90–0.92)
ns
0.25 (p < 0.0001)
0.23 (p = 0.0032)
1.87 (p < 0.0001)
0.70 (0.65-0.74)
0.78 (0.74–0.81)
0.8 (0.76–0.83)
0.84 (0.81–0.86)
0.41 (p < 0.0001)
0.08 (p < 0.0001)
0.62 (p < 0.0001)
1.29 (p < 0.0001)
0.43 (0.35–0.50)
0.77 (0.73–0.80)
0.71 (0.66–0.75)
0.90 (0.88–0.92)
ns
0.29 (p < 0.0001)
0.26 (p = 0.022)
1.62 (p < 0.0001)
0.38 (0.30–0.46)
0.75 (0.70–0.79)
0.69 (0.64–0.74)
0.91 (0.90–0.93)
ns
0.41 (p < 0.001)
ns
2.01 (p < 0.0001)
R2
0.96
0.91
0.93
0.95
1108
G.N. Cattermole et al. / Resuscitation 81 (2010) 1105–1110
Table 2 Comparison of weight-estimation methods. MAC formula
Broselow height
APLS formula
p
MAC formula
Broselow height
p
(Subjects outside the height limit of the Broselow tape assumed to be of medium adult weight, 55 kg)
(Only subjects within the height limit of the Broselow tape)
Overall
n = 1370
n = 1199
COV
16.1
23.8
28.7
15.6
16.7
Bias Upper LOA Lower LOA
2.9 37.2 −31.4
1.5 35.4 −32.4
−13.4 27.2 −54.0
5.1 38.7 −28.5
−1.9 25.7 −29.5
≤10% >10% >20% >30%
44.2 30.4 14.9 10.5
53.0 25.9 11.8 9.3
39.0 27.7 16.9 16.4
44.1 29.4 14.6 11.8
58.4 27.5 11.0 3.1
1–5 years last birthday
n = 449
COV
16.3
15.0
19.0
Bias Upper LOA Lower LOA
17.2 47.0 −12.5
−1.3 24.7 −27.2
−7.4 24.1 −39.0
≤10% >10% >20% >30%
26.5 25.4 20.3 27.8
65.3 22.9 8.5 3.3
49.2 31.6 12.5 6.7
6–8 years last birthday
n = 473
<0.0001 0.0087 ns
<0.0001 <0.0001 <0.0001
<0.0001 0.0032 0.032
COV
13.6
16.3
24.0
Bias Upper LOA Lower LOA
−0.5 27.1 −28.0
−1.9 25.9 −29.7
−10.6 28.7 −49.9
≤10% >10% >20% >30%
53.9 31.7 11.4 3.0
56.7 30.0 9.9 3.4
42.9 26.6 15.6 14.8
9–11 years last birthday
n = 448
COV
11.7
24.4
27.5
11.6
16.8
Bias Upper LOA Lower LOA
−7.9 17.0 −32.9
8.0 50.3 −34.4
−22.2 22.0 −66.4
−5.1 20.1 −30.4
−2.9 27.1 −32.8
≤10% >10% >20% >30%
51.8 33.9 13.2 1.1
36.8 24.6 17.0 21.7
24.6 24.8 22.8 27.9
56.3 31.9 10.8 1.1
50.2 30.5 16.8 2.5
ns ns ns n = 279
<0.0001 <0.0001 <0.0001
ns 0.02 ns
Notes: All figures are percentages. COV refers to coefficient of variation. Bias, upper and lower LOA (limits of agreement) refer to Bland Altman analysis. ≤10%, >10%, >20%, >30% refer to the proportions of weight estimates within those limits of deviation from actual weight. P values are for chi-squared comparison of proportions within that limit of weight estimation error versus all those outside that limit, for the best result against the next best.
the youngest age-group, in whom the MAC formula performed most poorly. The Broselow height method performed best in the youngest age-group, and the MAC formula in the oldest agegroup. There was no significant difference between the two in the 6–8 year-group. 4. Discussion We have shown that MAC is the strongest correlate with children’s weight, in all age-groups, and that this relationship strengthens with age. Height, foot-length and age are also correlated with weight, but this relationship weakens with age, and only height remains an important correlate in older children, alongside MAC. We have derived a weight-estimation formula based on mid-arm circumference: weight = (MAC − 10) × 3. This rule outperforms the Broselow tape and the APLS age-based formula in older children, is at least as good as the Broselow tape and outperforms the APLS formula in the 6–8 year-group, but is less suitable in children under 6 years old.
This study has several strengths. It is the largest to have studied MAC as a weight-estimation method, and the first to have assessed the accuracy and precision of a MAC-based formula. We have used several accepted methods of comparison18 and our results for Broselow and APLS methods are consistent with previous studies. MAC was first described in 1960 as a simple and reliable method to assess nutritional status in pre-school tropical children.21 The same authors have published tables of MAC-for-age, from infancy to old age, but not of weight-for-MAC.7,22 MAC and weight-forheight are now the two standard anthropometric measurements for a “Rapid Health Assessment” in developing world and catastrophe situations.23 MAC has also been used in all age groups and in more sophisticated settings than the developing world.7 MAC has been shown to correlate strongly with weight in adults. Two adult studies have found coefficients between 0.87 and 0.88.4,5 Both of these papers suggested that a table or formula relating weight with MAC, for use in emergency or critical care situations, could be derived in a further study, but none has been published. Knee height and MAC are included in a weight-estimation tool for
G.N. Cattermole et al. / Resuscitation 81 (2010) 1105–1110
elderly adults authored by Ross Laboratories, involving gender and ethnicity specific formulae,6 which have been adapted for use in Hong Kong.24 However these are complex, include variables other than MAC, and are not designed for children. Studies from the 1960s found coefficients of correlation between 0.5 and 0.82 in children.4,25 In a study of 169 children in a UK outpatient clinic, Carroll found a strong linear correlation of weight with MAC identical to that in our study, r = 0.91, and derived a formula, which can be written as: weight = (MAC × 3.6) − 40.3 This was published in abstract form only, and has not been validated nor widely used. However, to our knowledge, this is the only simple linear formula relating weight to MAC in children in the previous literature. Weight-estimation tools for emergency situations should be both simple and reliable. For a formula, simplicity implies linearity. Reliability involves both accuracy and precision. Accuracy is a measure of the average deviation of an estimation rule from reality, and is reflected by the Bland Altman bias. Precision is a measure of the scatter, and is reflected by the limits of agreement, and coefficient of variation. The proportions of estimates outside 10%, 20%, or 30% limits reflect both accuracy and precision. For non-linear relationships such as weight-for-age, linear formulae are intrinsically less accurate, but are easy to calculate rapidly and correctly, and are therefore commonly accepted in clinical practice.1 MAC however does appear to be related to weight in a far more linear fashion than age or height, and our linear formula is simple to calculate. The accuracy of a formula or table is shown by the proximity of its predictions to the average real values. The more complex the tool, the more accurately it can be “fine-tuned”: perhaps the most accurate would be to use a centile curve or table of values. However, aside from the loss of simplicity and the requirement for access to a physical chart, it may not produce greater precision. Precision is dependent on the inherent scatter of the relationship involved. Weight-for-age has a higher scatter than weight-forheight; age-based rules will therefore always be more imprecise than height-based. And for clinical practice, the precision is probably more important than accuracy. A rule with zero bias (i.e., as accurate as possible) but with very wide limits of agreement will be less safe than a rule with 5% bias but very narrow limits of agreement. The former will result in many more children being administered doses outside the therapeutic range than in the latter. Our results for the Broselow and age-based methods of weight estimation are consistent with previous findings. Both are more accurate and precise in younger children than older,2,8,10,12,13 and Broselow is superior to the APLS formula.8,10,11 In the original description of the Broselow tape, 59.7% of estimates for children within the height limits of the tape, aged one week to 12 years, were within 10% of true weight.2 Other studies have found values of 55.3–65%.26,27 In our study the figure was 58.4% for Broselow and 39% for the APLS rules. Krieser published respective values of 61% and 34% in a similar subject group.12 In our study, the figures for children aged 6–8 years last birthday were 53.9% for the MAC-rule and 56.7% for the Broselow method. In the 9–11 yeargroup, this small difference was inverted: for children within the Broselow height limits, the values were 56.3% for MAC and 50.2% for Broselow. The difference was even more marked if children too tall for Broselow were included and assumed to be medium adults: 51.8% for MAC and 36.8% for Broselow. In school-age children, the MAC formula is at least as good as the Broselow tape and much better than the APLS rule. The trends of the coefficient of variation and Bland Altman limits of agreement imply the same conclusions. If the Broselow tape and APLS rules are considered acceptable practice in school-age children, then so should the MAC rule. In younger children, the imprecision of the MAC formula suggests that no MAC-derived rule
1109
Fig. 2. Insert tape (Lasso-o).
is likely to outperform Broselow or even age-based rules in this age-group. However, the accuracy of the rule could be refined further in school-age children. This study has derived a simple linear formula, but a more complex tool could be derived and validated using an arm-tape similar in principle to the Broselow tape, and colour-coded appropriately according to the same weight-groups. Colour-coded arm tapes have been used previously for malnutrition assessment,28 and recommended because the arm is accessible and usually requires no undressing.7 In resuscitation, the arm is usually already exposed for blood pressure measurement and venous access. An arm-tape would be more portable than a body-length tape, cheaper to produce, and could be used in children who are sitting rather than prone, such as those with respiratory distress. There are some potential limitations. Firstly, we used a standard tape measure rather than the specialised UNICEF insert-tape,7 because these were unavailable in Hong Kong at the time. This might have reduced the accuracy and precision of our measurements. However, the Lasso-o insert tape (Fig. 2), used by the UK Department of Health’s Nutrition survey, has been shown to have an inter-observer intra-class coefficient (ICC) of 0.979,29 which is consistent with an ICC of 0.981 in our study. For clinical use we would recommend a colour-coded insert tape. Secondly, the Broselow tape in this study was virtual: we used precise measurements of children’s height cross-referenced with the weight-for-height on the Broselow tape. However, it is probable that use of the Broselow tape in practice would be less precise than the heights obtained with the stadiometer. This would be unlikely to change the bias of the Broselow method, but it would be likely to increase the scatter, leading to wider Bland Altman limits of agreement and greater proportions of estimates more than a given percentage different from true weight. Together with the fact that we did not use an insert tape for MAC, it is therefore possible that our results under-estimate the precision of the MAC formula, and over-estimate that of the Broselow tape. Despite this, we have
1110
G.N. Cattermole et al. / Resuscitation 81 (2010) 1105–1110
still shown a clear advantage of a MAC derived formula for weight estimation in older children. Thirdly, we included only healthy children. For obvious reasons, we did not want to include children with an illness that might affect their weight. Although many illnesses and medications have no effect on weight, language barriers and cultural differences of the concepts of disease led us to make the pragmatic decision to exclude children with any illness. We do not think that our results will be significantly skewed by excluding children with illnesses that do not affect weight. Fourthly, our study included only Chinese children, aged 1–11 years old last birthday. Because of the increasing precision of the rule with increasing age, a MAC formula or arm-tape could be useful in older people. Further studies will be needed to demonstrate the reliability of the MAC formula in other ethnic groups, and in adolescents and adults. Fifthly, the use of the proportion of estimates within 10% as a measure of reliability is arbitrary. It is likely that most resuscitation drugs and fluids can safely be given with a wider margin of dosage error, perhaps as much as 20–30%.30 Certainly this is the case in adults, in whom standard doses are given regardless of weight. It might be more relevant to consider the proportions of estimates greater than 30%, although this would not affect the conclusions of this study. Lastly, we have only considered MAC as a predictor of weight, and therefore of weight-related doses of drugs and fluids. Both the Broselow tape, and age-based formulae,1 also provide estimates of appropriate airway sizes. We were unable to assess this, and the use of MAC to predict airway size could be the subject of future studies. 5. Conclusion Weight correlates more strongly with mid-arm circumference in children, than with age, height or foot-length. We have derived a weight-estimation formula based on mid-arm circumference: weight = (MAC − 10) × 3. This rule out-performs the Broselow tape and the APLS age-based formula in older children, but is less suitable in children under 6 years old. A colour-coded arm-tape could be developed for rapid assessment of paediatric weight in the resuscitation room for use in older children. Conflict of interest statement None of the authors has any competing interests to declare. Funding We received a Direct Grant, ref. 4450252, of HK$72,000 (approximately US$10,000) from the Chinese University of Hong Kong to conduct the “Healthy Children’s Vital Signs and USCOM Study”. We also received a grant of HK$100,000 (approximately US13,000$) from the Hong Kong College of Emergency Medicine. References 1. Mackway-Jones K, Molyneux E, Phillips B, Wieteska S, editors. Advanced paediatric life support—the practical approach. 4th ed. Oxford: Blackwell; 2005.
2. Lubitz DS, Seidel JS, Chameides L, Luten RC, Zaritsky AL, Campbell FW. A rapid method for estimating weight and resuscitation drug dosages from length in the pediatric age group. Ann Emerg Med 1988;17:576–81. 3. Carroll W, Jan Y, Alexander J. Towards better weight estimation in the seriously ill child—a comparison of methods. Arch Dis Child 2001;84:A12. 4. Larson LL. Relationship of upper arm circumference and body weight. J Emerg Nurs 1985;11:246–8. 5. Utley R. Mid-arm circumference: estimating patients’ weight. Dimens Crit Care Nurs 1990;9:75–81. 6. Lin BW, Yoshida D, Quinn J, Strehlow M. A better way to estimate adult patients’ weights. Am J Emerg Med 2009;27:1060–4. 7. Jelliffe DB, Jelliffe EFP. Community nutritional assessment. Oxford: Oxford University Press; 1989. 8. Argall JAW, Wright N, Mackway-Jones K, Jackson R. A comparison of two commonly used methods of weight estimation. Arch Dis Child 2003;88:789–90. 9. Luscombe M, Owens B. Weight estimation in resuscitation: is the current formula still valid? Arch Dis Child 2007;92:412–5. 10. Black K, Barnett P, Wolfe R, Young S. Are methods used to estimate weight in children accurate? Emerg Med (Fremantle) 2002;14:160–5. 11. Krieser D, Nguyen K, Kerr D, Jolley D, Clooney M, Kelly A-M. Parental weight estimation of their child’s weight is more accurate than other weight estimation methods for determining children’s weight in an emergency department? Emerg Med J 2007;24:756–9. 12. So TY, Farrington E, Absher RK. Evaluation of the accuracy of different methods used to estimate weights in the pediatric population. Pediatrics 2009;123:e1045–51. 13. Thompson MT, Reading MJ, Acworth JP. Best guess method for age-based weight estimation in paediatric emergencies: validation and comparison with current methods. Emerg Med Australas 2007;19:535–42. 14. DuBois D, Baldwin S, King WD. Accuracy of weight estimation methods for children. Pediatr Emerg Care 2007;23:227–30. 15. So H-K, Nelson EAS, Li AM, et al. Secular changes in height, weight and body mass index in Hong Kong Children. BMC Public Health 2008;8:320. 16. Cole TJ, Green PJ. Smoothing reference centile curves: the LMS method and penalized likelihood. Stat Med 1992;11:1305–19. 17. Sung RYT, Choi KC, So H-K, et al. Oscillometrically measured blood pressure in Hong Kong Chinese children and associations with anthropometric parameters. J Hypertens 2008;26:678–84. 18. Bailey SM, Sarmandal P, Grant JM. A comparison of three methods of assessing inter-observer variation applied to measurement of the symphysis-fundal height. Br J Osbtet Gynaecol 1989;96:1266–71. 19. Altman DG, Bland JM. Measurement in medicine: the analysis of method comparison studies. Statistician 1983;32:307–17. 20. American College of Surgeons Committee on Trauma. Advanced trauma life support student course manual. 8th ed. Chicago: ACS; 2008. 21. Jelliffe DB, Jelliffe EPP. Prevalence of protein-calorie malnutrition in Haitian preschool children. Am J Public Health Nations Health 1960;50:1355–66. 22. Jelliffe DB. The assessment of the nutritional status of the community. Geneva: World Health Organisation; 1966. 23. Chaloner E, England M, Nicol A, Hawley A. Care of refugees: medical aspects. In: Lumley JSP, Ryan JM, Baxter PJ, Kirby N, editors. Handbook of the medical care of catastrophes. London: Royal Society of Medicine Press Ltd.; 1996. p. 163–72. 24. Jung MY, Chan MS, Chow VSF, et al. Estimating geriatric patients’ body weight using the knee height caliper and mid-arm circumference in Hong Kong Chinese. Asia Pac J Clin Nutr 2004;13:261–4. 25. Rutishauser IHE. The arm circumference as a public health index of proteincalorie malnutrition of early childhood. (V) Correlations of the circumference of the mid-upper arm with weight and weight for height in three groups in Uganda. J Trop Pediatr 1969;15:196–7. 26. Hofer CK, Ganter M, Tucci M, Klaghoer R, Zollinger A. How reliable is lengthbased determination of body weight and tracheal tube size in the paediatric age group? The Broselow tape reconsidered. Br J Anaesth 2002;88:283–5. 27. Nieman CT, Manacci CF, Super DM, Mancusco C, Fallon WF. Use of the Broselow tape may result in the underresuscitation of children. Acad Emerg Med 2006;13:1011–9. 28. Shakir A, Morley D. Measuring malnutrition. Lancet 1974;1:758–9. 29. Bartram JL, Rigby AS, Baxter PS. The “Lasso-o” tape: stretchability and observer variability in head circumference measurement. Arch Dis Child 2005;90:820–1. 30. Lack JA, Stuart-Taylor ME. Calculation of drug dosage and body surface area in children. Br J Anaesth 1997;78:601–5.
Contents lists available at ScienceDirect
Resuscitation journal homepage: www.elsevier.com/locate/resuscitation
Clinical paper
Mid-arm circumference can be used to estimate children’s weights夽 G.N. Cattermole a , P.Y.M. Leung b , P.S.K. Mak a , C.A. Graham a , T.H. Rainer a,∗ a b
Accident and Emergency Medicine Academic Unit, Chinese University of Hong Kong, Hong Kong Faculty of Medicine, University of Melbourne, Melbourne, Australia
a r t i c l e
i n f o
Article history: Received 5 January 2010 Received in revised form 20 April 2010 Accepted 22 May 2010 Keywords: (MeSH) Pediatrics Resuscitation Body weights and measures
a b s t r a c t Introduction: Accurate measurement of children’s weight is rarely possible in paediatric resuscitation, and rapid estimates are made to ensure appropriate drug and fluid doses and equipment selection. Weight is commonly estimated from formulae based on children’s age, or from their height using the Broselow tape. Foot-length and mid-arm circumference have also been suggested as the basis of weight-estimation formulae. Objectives: To determine which of age, height, foot-length or mid-arm circumference had the strongest relationship with weight in healthy children, to derive a simple weight-estimation formula from the strongest correlate, and to compare its performance with existing weight-estimation tools. Methods: This was a population-based prospective observational study of Hong Kong Chinese children aged 1–11 years old last birthday. Weight was measured to the nearest 0.2 kg; height, foot-length and midarm circumference to the nearest 0.1 cm. Multiple regression analysis was used to determine the strongest independent relationships with weight, and linear regression analysis derived a weight-estimation formula. Accuracy and precision of this formula were compared with standard age-based and height-based weight-estimation methods. Results: Mid-arm circumference had the strongest relationship with weight, and this relationship grew stronger with age. The formula, weight [kg] = (mid-arm circumference [cm] − 10) × 3, was at least as accurate and precise as the Broselow method and outperformed the age-based rule in school-age children, but was inadequate in pre-school children. Conclusion: This weight-estimation formula based on mid-arm circumference is reliable for use in schoolage children, and an arm-tape could be considered as an alternative to the Broselow tape in this population. © 2010 Elsevier Ireland Ltd. All rights reserved.
1. Introduction In paediatric resuscitation, it is necessary to know the child’s weight in order to provide appropriate drug and fluid doses, equipment selection and ventilator settings. Because measurement of weight itself is rarely possible in time-critical situations, and because there is often no one available who knows the child’s weight, rapid and accurate methods of estimation need to be
Abbreviations: kg, kilogramme; cm, centimetre; MAC, mid-arm circumference; PEM, protein energy malnutrition; APLS, Advanced Paediatric Life Support. 夽 A Spanish translated version of the abstract of this article appears as Appendix in the final online version at doi:10.1016/j.resuscitation.2010.05.015. ∗ Corresponding author at: Accident and Emergency Medicine Academic Unit, Chinese University of Hong Kong, Rooms 107/113, Trauma and Emergency Centre, Prince of Wales Hospital, Shatin, New Territories, Hong Kong. Tel.: +852 2632 1033; fax: +852 2648 1469. E-mail addresses: [email protected] (G.N. Cattermole), [email protected] (P.Y.M. Leung), [email protected] (P.S.K. Mak), [email protected] (C.A. Graham), [email protected] (T.H. Rainer). 0300-9572/$ – see front matter © 2010 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.resuscitation.2010.05.015
applied. Most commonly, weight can be estimated from formulae based on the child’s age,1 or from the child’s height using the Broselow tape.2 Other suggested methods include weight estimation according to foot-length, or mid-arm circumference (MAC).3–5 Neither is currently used in clinical practice for weight estimation in paediatric resuscitation. However, a formula including MAC and knee height has recently been validated in geriatric patients in the emergency department.6 MAC has also been used for many years in the assessment of malnutrition in the developing world.7 One of the most widely used age-based estimation method is that recommended in the UK-based Advanced Paediatric Life Support course1 : weight in kg = 2 × (age in years + 4). It has been criticised for under-estimating children’s weight, and new formulae have been derived.8,9 The Broselow tape has consistently been found to outperform age-based formulae in estimating weight,8,10,11 but both the Broselow tape and age-based weight estimation methods are less precise in older and heavier children.2,8,10,12,13 Systematic underestimation of childhood weight is likely to be due to the increasing actual weights, especially
1106
G.N. Cattermole et al. / Resuscitation 81 (2010) 1105–1110
of older, Western children. The US National Health and Nutrition Examination survey has demonstrated a steady increase in childhood obesity: data from 1999 to 2002 found that 16% of 6–19 year olds, and 10.2% of 2–5 year olds, were overweight (≥95th centile body mass index).14 In 6–18 year olds in Hong Kong, the proportion increased from 7.1% to 10.1% between 1993 and 2006.15 It is not known whether a weight-estimation method based on foot-length or MAC might be more appropriate than either the Broselow tape or age-based methods, especially in older children. The objectives of this study were to determine which of age, height, foot-length or MAC had the strongest relationship with weight in healthy children, to derive a simple weight-estimation formula from the strongest correlate, and to compare its performance with existing weight-estimation tools. 2. Methods This was a population-based observational study, part of the prospective “Healthy Children’s Vital Signs and USCOM study”, which also included physiological and ultrasound cardiac output monitor (USCOM) measurements. It was conducted in primary schools and kindergartens in Hong Kong, and recruited healthy Chinese children aged 1–11 years on their last birthday. Head teachers of all relevant institutions in the Shatin area of Hong Kong were asked for permission to have their schools participate in the study. Classes were selected by schools to provide a representative distribution of age, and parents of children in participating classes were sent an information pack and consent form. Children with any chronic or current illness, or taking any medications, were excluded. Pragmatic language and cultural barriers precluded gathering information on parental reasons for exclusion or failure to consent. Weight was measured to the nearest 0.2 kg using electronic scales (Compact Precision Scale C200H, Conair Far East Ltd.). Height was measured with to the nearest 0.1 cm with a stadiometer (Harpenden Portable Stadiometer, Holtain, UK). Foot-length was measured from the right heel to the tip of the hallux, with the child supine and the ankle at 90◦ with the foot flat against the base of a rigid box, the heel resting on the inside wall of the box. MAC was measured with the child’s right arm relaxed in 90◦ of flexion at the elbow. The olecranon and acromion were identified, and an inelastic tape-measure was extended along the arm between these two points. The mid-point was marked and the tape was wrapped around the arm, taking care to ensure that the tape lay flat against the arm without pinching the underlying skin. Children were measured without shoes, wearing the lightest school uniform including socks. School uniforms were weighed separately and subsequent adjustments made to the measured weight of the child. 2.1. Statistical analysis LMS Chartmaker Pro v2.3 software (Cole and Pan, Medical Research Council UK, 2006) was used to describe the data in centile curves (2.5, 10, 50, 90, and 97.5). The relationship between weight and each of the four variables (age, height, foot-length and MAC) were modeled by the LMS method of Cole and Green.16 Briefly, the relationship is described by three age-specific cubic spline curves known as L, M and S. M represents the median, S is the coefficient of variation, and L is the Box–Cox transformation that renders the data to follow a normal distribution, conditional on age. Combination of these three functions generates centile values for each parameter. Analysis was weighted according to the size of gender groups. MedCalc v10.4 (Schoonjans, 2009) was used for all subsequent statistical analysis.
Pearson’s correlation coefficients were calculated for each of the four variables with weight, and stepwise multiple regression analysis was performed to determine the variable with the strongest independent relationship. Linear regression analysis was used to define a simple linear equation relating this variable with weight. Three methods of weight estimation were then compared: the newly derived linear formula, the APLS age-based formula, and height-related weights as defined on the Broselow tape. The Broselow tape itself was not used: we used the height measurements obtained with the stadiometer and cross-referenced these with the markings on the 1998 version of the tape. It was expected that a large proportion of older children would exceed the height limit of the Broselow tape (143.6 cm). For these children, we assumed a weight equal to that of a medium-sized adult, which in 18-year-old Hong Kong Chinese is 55 kg.17 As an alternative approach in the oldest group of children, we also compared the new formula with the Broselow tape only in the group of children who were within the height limits of the tape. Three methods of comparison were used to assess the accuracy and precision of each method of estimation. Firstly, coefficient of variation was calculated as the standard deviation of the differences between estimated and actual weights, divided by the overall mean of estimated and actual weights.18 Secondly, Bland Altman plots19 determined the mean bias and the 95% limits of agreement. The bias indicates the mean percentage difference between estimated and actual weight, and the limits of agreement indicate in what range 95% of the differences between estimated and actual weights will fall. Thirdly, we determined the proportion of estimates that were within 10%, 20% and 30%, or more than 30% different from, the actual weights. Chi-squared tests were used to compare these proportions between the weight-estimation methods. Three age-groups were defined based on those in the Advanced Trauma Life Support course20 in order to create approximately equal subgroups. Toddlers (1–2 years old last birthday) and preschool children (3–5 years old last birthday) were grouped together as “1–5”. School age children (6–11 years old last birthday) were split into two groups, “6–8” and “9–11”. 2.2. Ethics Ethical approval was obtained from the Clinical Research Ethics Committee of the Chinese University of Hong Kong. Written parental consent was obtained for all subjects more than a week prior to the school visit. Children who were unwilling to participate on the day of the study were not included. 3. Results 1391 Chinese subjects aged 1–11 years on their last birthday were recruited from 14 schools and kindergartens. In the six institutions which recorded how many consent letters were distributed, 48% of invited parents consented. The time constraints of school timetables prevented us from obtaining all four anthropometric measurements in 21 children, who were excluded from further analysis. 1370 (98.5% of the eligible sample) were therefore included, of whom 55% were boys. There were 448 children in the 9–11 year-group, 473 in the 6–8 year-group, and 449 in the 1–5 year-group. The LMS-derived centile curves for age, height, foot, MAC are shown in Fig. 1. In contrast to the other three variables, weight-forMAC is near linear with less spread. Correlation coefficients (r) for each variable with weight are shown in Table 1 with 95% confidence intervals. All individual correlations were statistically significant (p < 0.0001). p-Values are given in the table for each variable after multiple regression. The
G.N. Cattermole et al. / Resuscitation 81 (2010) 1105–1110
1107
Fig. 1. Centile curves of weight for age, height, foot-length and MAC (10th, 25th, 50th, 75th, and 90th). .
coefficient of determination (R2 ) is given for each multiple regression model, with p < 0.001 for all models. In all age-groups, weight is more strongly correlated with MAC than with any of the other three variables, although this is not statistically significant in the 1–5 year-group. The strength of the correlation of weight with MAC increases with increasing age, but the strengths of the correlations of weight with age, height and footlength decrease with increasing age. After multiple regression, age is an important correlate with weight only in the 1–5 year-group, and only height and MAC remain important in the oldest group. The strongest correlate with weight, both on independent correlation and after multiple regression, is MAC.
By linear regression the following formula was obtained for weight (kg) and MAC (cm): weight = (2.94 × MAC) − 29.14. For pragmatic simplicity, this was approximated and rearranged to: Weight = (MAC − 10) × 3 This weight-estimation formula was then compared with the Broselow height and the APLS age-rule methods (Table 2). The heights of 171 children were outside the limits of the Broselow tape, all but two of whom were in the oldest group. By any method of comparison, the APLS age-rule was significantly worse than either of the other weight-estimation methods, except in
Table 1 Correlation and multiple regression. Year-group
n
Overall
1370
1–5 years last birthday
6–8 years last birthday
9–11 years last birthday
449
473
448
Pearson’s correlation, r (95%CI) Multiple regression coefficient (p-value) Pearson’s correlation, r (95%CI) Multiple regression coefficient (p-value) Pearson’s correlation, r (95%CI) Multiple regression coefficient (p-value) Pearson’s correlation, r (95%CI) Multiple regression coefficient (p-value)
Age
Height
Foot-length
MAC
0.80 (0.78–82)
0.88 (0.87–0.89)
0.87 (0.85–0.88)
0.91 (0.90–0.92)
ns
0.25 (p < 0.0001)
0.23 (p = 0.0032)
1.87 (p < 0.0001)
0.70 (0.65-0.74)
0.78 (0.74–0.81)
0.8 (0.76–0.83)
0.84 (0.81–0.86)
0.41 (p < 0.0001)
0.08 (p < 0.0001)
0.62 (p < 0.0001)
1.29 (p < 0.0001)
0.43 (0.35–0.50)
0.77 (0.73–0.80)
0.71 (0.66–0.75)
0.90 (0.88–0.92)
ns
0.29 (p < 0.0001)
0.26 (p = 0.022)
1.62 (p < 0.0001)
0.38 (0.30–0.46)
0.75 (0.70–0.79)
0.69 (0.64–0.74)
0.91 (0.90–0.93)
ns
0.41 (p < 0.001)
ns
2.01 (p < 0.0001)
R2
0.96
0.91
0.93
0.95
1108
G.N. Cattermole et al. / Resuscitation 81 (2010) 1105–1110
Table 2 Comparison of weight-estimation methods. MAC formula
Broselow height
APLS formula
p
MAC formula
Broselow height
p
(Subjects outside the height limit of the Broselow tape assumed to be of medium adult weight, 55 kg)
(Only subjects within the height limit of the Broselow tape)
Overall
n = 1370
n = 1199
COV
16.1
23.8
28.7
15.6
16.7
Bias Upper LOA Lower LOA
2.9 37.2 −31.4
1.5 35.4 −32.4
−13.4 27.2 −54.0
5.1 38.7 −28.5
−1.9 25.7 −29.5
≤10% >10% >20% >30%
44.2 30.4 14.9 10.5
53.0 25.9 11.8 9.3
39.0 27.7 16.9 16.4
44.1 29.4 14.6 11.8
58.4 27.5 11.0 3.1
1–5 years last birthday
n = 449
COV
16.3
15.0
19.0
Bias Upper LOA Lower LOA
17.2 47.0 −12.5
−1.3 24.7 −27.2
−7.4 24.1 −39.0
≤10% >10% >20% >30%
26.5 25.4 20.3 27.8
65.3 22.9 8.5 3.3
49.2 31.6 12.5 6.7
6–8 years last birthday
n = 473
<0.0001 0.0087 ns
<0.0001 <0.0001 <0.0001
<0.0001 0.0032 0.032
COV
13.6
16.3
24.0
Bias Upper LOA Lower LOA
−0.5 27.1 −28.0
−1.9 25.9 −29.7
−10.6 28.7 −49.9
≤10% >10% >20% >30%
53.9 31.7 11.4 3.0
56.7 30.0 9.9 3.4
42.9 26.6 15.6 14.8
9–11 years last birthday
n = 448
COV
11.7
24.4
27.5
11.6
16.8
Bias Upper LOA Lower LOA
−7.9 17.0 −32.9
8.0 50.3 −34.4
−22.2 22.0 −66.4
−5.1 20.1 −30.4
−2.9 27.1 −32.8
≤10% >10% >20% >30%
51.8 33.9 13.2 1.1
36.8 24.6 17.0 21.7
24.6 24.8 22.8 27.9
56.3 31.9 10.8 1.1
50.2 30.5 16.8 2.5
ns ns ns n = 279
<0.0001 <0.0001 <0.0001
ns 0.02 ns
Notes: All figures are percentages. COV refers to coefficient of variation. Bias, upper and lower LOA (limits of agreement) refer to Bland Altman analysis. ≤10%, >10%, >20%, >30% refer to the proportions of weight estimates within those limits of deviation from actual weight. P values are for chi-squared comparison of proportions within that limit of weight estimation error versus all those outside that limit, for the best result against the next best.
the youngest age-group, in whom the MAC formula performed most poorly. The Broselow height method performed best in the youngest age-group, and the MAC formula in the oldest agegroup. There was no significant difference between the two in the 6–8 year-group. 4. Discussion We have shown that MAC is the strongest correlate with children’s weight, in all age-groups, and that this relationship strengthens with age. Height, foot-length and age are also correlated with weight, but this relationship weakens with age, and only height remains an important correlate in older children, alongside MAC. We have derived a weight-estimation formula based on mid-arm circumference: weight = (MAC − 10) × 3. This rule outperforms the Broselow tape and the APLS age-based formula in older children, is at least as good as the Broselow tape and outperforms the APLS formula in the 6–8 year-group, but is less suitable in children under 6 years old.
This study has several strengths. It is the largest to have studied MAC as a weight-estimation method, and the first to have assessed the accuracy and precision of a MAC-based formula. We have used several accepted methods of comparison18 and our results for Broselow and APLS methods are consistent with previous studies. MAC was first described in 1960 as a simple and reliable method to assess nutritional status in pre-school tropical children.21 The same authors have published tables of MAC-for-age, from infancy to old age, but not of weight-for-MAC.7,22 MAC and weight-forheight are now the two standard anthropometric measurements for a “Rapid Health Assessment” in developing world and catastrophe situations.23 MAC has also been used in all age groups and in more sophisticated settings than the developing world.7 MAC has been shown to correlate strongly with weight in adults. Two adult studies have found coefficients between 0.87 and 0.88.4,5 Both of these papers suggested that a table or formula relating weight with MAC, for use in emergency or critical care situations, could be derived in a further study, but none has been published. Knee height and MAC are included in a weight-estimation tool for
G.N. Cattermole et al. / Resuscitation 81 (2010) 1105–1110
elderly adults authored by Ross Laboratories, involving gender and ethnicity specific formulae,6 which have been adapted for use in Hong Kong.24 However these are complex, include variables other than MAC, and are not designed for children. Studies from the 1960s found coefficients of correlation between 0.5 and 0.82 in children.4,25 In a study of 169 children in a UK outpatient clinic, Carroll found a strong linear correlation of weight with MAC identical to that in our study, r = 0.91, and derived a formula, which can be written as: weight = (MAC × 3.6) − 40.3 This was published in abstract form only, and has not been validated nor widely used. However, to our knowledge, this is the only simple linear formula relating weight to MAC in children in the previous literature. Weight-estimation tools for emergency situations should be both simple and reliable. For a formula, simplicity implies linearity. Reliability involves both accuracy and precision. Accuracy is a measure of the average deviation of an estimation rule from reality, and is reflected by the Bland Altman bias. Precision is a measure of the scatter, and is reflected by the limits of agreement, and coefficient of variation. The proportions of estimates outside 10%, 20%, or 30% limits reflect both accuracy and precision. For non-linear relationships such as weight-for-age, linear formulae are intrinsically less accurate, but are easy to calculate rapidly and correctly, and are therefore commonly accepted in clinical practice.1 MAC however does appear to be related to weight in a far more linear fashion than age or height, and our linear formula is simple to calculate. The accuracy of a formula or table is shown by the proximity of its predictions to the average real values. The more complex the tool, the more accurately it can be “fine-tuned”: perhaps the most accurate would be to use a centile curve or table of values. However, aside from the loss of simplicity and the requirement for access to a physical chart, it may not produce greater precision. Precision is dependent on the inherent scatter of the relationship involved. Weight-for-age has a higher scatter than weight-forheight; age-based rules will therefore always be more imprecise than height-based. And for clinical practice, the precision is probably more important than accuracy. A rule with zero bias (i.e., as accurate as possible) but with very wide limits of agreement will be less safe than a rule with 5% bias but very narrow limits of agreement. The former will result in many more children being administered doses outside the therapeutic range than in the latter. Our results for the Broselow and age-based methods of weight estimation are consistent with previous findings. Both are more accurate and precise in younger children than older,2,8,10,12,13 and Broselow is superior to the APLS formula.8,10,11 In the original description of the Broselow tape, 59.7% of estimates for children within the height limits of the tape, aged one week to 12 years, were within 10% of true weight.2 Other studies have found values of 55.3–65%.26,27 In our study the figure was 58.4% for Broselow and 39% for the APLS rules. Krieser published respective values of 61% and 34% in a similar subject group.12 In our study, the figures for children aged 6–8 years last birthday were 53.9% for the MAC-rule and 56.7% for the Broselow method. In the 9–11 yeargroup, this small difference was inverted: for children within the Broselow height limits, the values were 56.3% for MAC and 50.2% for Broselow. The difference was even more marked if children too tall for Broselow were included and assumed to be medium adults: 51.8% for MAC and 36.8% for Broselow. In school-age children, the MAC formula is at least as good as the Broselow tape and much better than the APLS rule. The trends of the coefficient of variation and Bland Altman limits of agreement imply the same conclusions. If the Broselow tape and APLS rules are considered acceptable practice in school-age children, then so should the MAC rule. In younger children, the imprecision of the MAC formula suggests that no MAC-derived rule
1109
Fig. 2. Insert tape (Lasso-o).
is likely to outperform Broselow or even age-based rules in this age-group. However, the accuracy of the rule could be refined further in school-age children. This study has derived a simple linear formula, but a more complex tool could be derived and validated using an arm-tape similar in principle to the Broselow tape, and colour-coded appropriately according to the same weight-groups. Colour-coded arm tapes have been used previously for malnutrition assessment,28 and recommended because the arm is accessible and usually requires no undressing.7 In resuscitation, the arm is usually already exposed for blood pressure measurement and venous access. An arm-tape would be more portable than a body-length tape, cheaper to produce, and could be used in children who are sitting rather than prone, such as those with respiratory distress. There are some potential limitations. Firstly, we used a standard tape measure rather than the specialised UNICEF insert-tape,7 because these were unavailable in Hong Kong at the time. This might have reduced the accuracy and precision of our measurements. However, the Lasso-o insert tape (Fig. 2), used by the UK Department of Health’s Nutrition survey, has been shown to have an inter-observer intra-class coefficient (ICC) of 0.979,29 which is consistent with an ICC of 0.981 in our study. For clinical use we would recommend a colour-coded insert tape. Secondly, the Broselow tape in this study was virtual: we used precise measurements of children’s height cross-referenced with the weight-for-height on the Broselow tape. However, it is probable that use of the Broselow tape in practice would be less precise than the heights obtained with the stadiometer. This would be unlikely to change the bias of the Broselow method, but it would be likely to increase the scatter, leading to wider Bland Altman limits of agreement and greater proportions of estimates more than a given percentage different from true weight. Together with the fact that we did not use an insert tape for MAC, it is therefore possible that our results under-estimate the precision of the MAC formula, and over-estimate that of the Broselow tape. Despite this, we have
1110
G.N. Cattermole et al. / Resuscitation 81 (2010) 1105–1110
still shown a clear advantage of a MAC derived formula for weight estimation in older children. Thirdly, we included only healthy children. For obvious reasons, we did not want to include children with an illness that might affect their weight. Although many illnesses and medications have no effect on weight, language barriers and cultural differences of the concepts of disease led us to make the pragmatic decision to exclude children with any illness. We do not think that our results will be significantly skewed by excluding children with illnesses that do not affect weight. Fourthly, our study included only Chinese children, aged 1–11 years old last birthday. Because of the increasing precision of the rule with increasing age, a MAC formula or arm-tape could be useful in older people. Further studies will be needed to demonstrate the reliability of the MAC formula in other ethnic groups, and in adolescents and adults. Fifthly, the use of the proportion of estimates within 10% as a measure of reliability is arbitrary. It is likely that most resuscitation drugs and fluids can safely be given with a wider margin of dosage error, perhaps as much as 20–30%.30 Certainly this is the case in adults, in whom standard doses are given regardless of weight. It might be more relevant to consider the proportions of estimates greater than 30%, although this would not affect the conclusions of this study. Lastly, we have only considered MAC as a predictor of weight, and therefore of weight-related doses of drugs and fluids. Both the Broselow tape, and age-based formulae,1 also provide estimates of appropriate airway sizes. We were unable to assess this, and the use of MAC to predict airway size could be the subject of future studies. 5. Conclusion Weight correlates more strongly with mid-arm circumference in children, than with age, height or foot-length. We have derived a weight-estimation formula based on mid-arm circumference: weight = (MAC − 10) × 3. This rule out-performs the Broselow tape and the APLS age-based formula in older children, but is less suitable in children under 6 years old. A colour-coded arm-tape could be developed for rapid assessment of paediatric weight in the resuscitation room for use in older children. Conflict of interest statement None of the authors has any competing interests to declare. Funding We received a Direct Grant, ref. 4450252, of HK$72,000 (approximately US$10,000) from the Chinese University of Hong Kong to conduct the “Healthy Children’s Vital Signs and USCOM Study”. We also received a grant of HK$100,000 (approximately US13,000$) from the Hong Kong College of Emergency Medicine. References 1. Mackway-Jones K, Molyneux E, Phillips B, Wieteska S, editors. Advanced paediatric life support—the practical approach. 4th ed. Oxford: Blackwell; 2005.
2. Lubitz DS, Seidel JS, Chameides L, Luten RC, Zaritsky AL, Campbell FW. A rapid method for estimating weight and resuscitation drug dosages from length in the pediatric age group. Ann Emerg Med 1988;17:576–81. 3. Carroll W, Jan Y, Alexander J. Towards better weight estimation in the seriously ill child—a comparison of methods. Arch Dis Child 2001;84:A12. 4. Larson LL. Relationship of upper arm circumference and body weight. J Emerg Nurs 1985;11:246–8. 5. Utley R. Mid-arm circumference: estimating patients’ weight. Dimens Crit Care Nurs 1990;9:75–81. 6. Lin BW, Yoshida D, Quinn J, Strehlow M. A better way to estimate adult patients’ weights. Am J Emerg Med 2009;27:1060–4. 7. Jelliffe DB, Jelliffe EFP. Community nutritional assessment. Oxford: Oxford University Press; 1989. 8. Argall JAW, Wright N, Mackway-Jones K, Jackson R. A comparison of two commonly used methods of weight estimation. Arch Dis Child 2003;88:789–90. 9. Luscombe M, Owens B. Weight estimation in resuscitation: is the current formula still valid? Arch Dis Child 2007;92:412–5. 10. Black K, Barnett P, Wolfe R, Young S. Are methods used to estimate weight in children accurate? Emerg Med (Fremantle) 2002;14:160–5. 11. Krieser D, Nguyen K, Kerr D, Jolley D, Clooney M, Kelly A-M. Parental weight estimation of their child’s weight is more accurate than other weight estimation methods for determining children’s weight in an emergency department? Emerg Med J 2007;24:756–9. 12. So TY, Farrington E, Absher RK. Evaluation of the accuracy of different methods used to estimate weights in the pediatric population. Pediatrics 2009;123:e1045–51. 13. Thompson MT, Reading MJ, Acworth JP. Best guess method for age-based weight estimation in paediatric emergencies: validation and comparison with current methods. Emerg Med Australas 2007;19:535–42. 14. DuBois D, Baldwin S, King WD. Accuracy of weight estimation methods for children. Pediatr Emerg Care 2007;23:227–30. 15. So H-K, Nelson EAS, Li AM, et al. Secular changes in height, weight and body mass index in Hong Kong Children. BMC Public Health 2008;8:320. 16. Cole TJ, Green PJ. Smoothing reference centile curves: the LMS method and penalized likelihood. Stat Med 1992;11:1305–19. 17. Sung RYT, Choi KC, So H-K, et al. Oscillometrically measured blood pressure in Hong Kong Chinese children and associations with anthropometric parameters. J Hypertens 2008;26:678–84. 18. Bailey SM, Sarmandal P, Grant JM. A comparison of three methods of assessing inter-observer variation applied to measurement of the symphysis-fundal height. Br J Osbtet Gynaecol 1989;96:1266–71. 19. Altman DG, Bland JM. Measurement in medicine: the analysis of method comparison studies. Statistician 1983;32:307–17. 20. American College of Surgeons Committee on Trauma. Advanced trauma life support student course manual. 8th ed. Chicago: ACS; 2008. 21. Jelliffe DB, Jelliffe EPP. Prevalence of protein-calorie malnutrition in Haitian preschool children. Am J Public Health Nations Health 1960;50:1355–66. 22. Jelliffe DB. The assessment of the nutritional status of the community. Geneva: World Health Organisation; 1966. 23. Chaloner E, England M, Nicol A, Hawley A. Care of refugees: medical aspects. In: Lumley JSP, Ryan JM, Baxter PJ, Kirby N, editors. Handbook of the medical care of catastrophes. London: Royal Society of Medicine Press Ltd.; 1996. p. 163–72. 24. Jung MY, Chan MS, Chow VSF, et al. Estimating geriatric patients’ body weight using the knee height caliper and mid-arm circumference in Hong Kong Chinese. Asia Pac J Clin Nutr 2004;13:261–4. 25. Rutishauser IHE. The arm circumference as a public health index of proteincalorie malnutrition of early childhood. (V) Correlations of the circumference of the mid-upper arm with weight and weight for height in three groups in Uganda. J Trop Pediatr 1969;15:196–7. 26. Hofer CK, Ganter M, Tucci M, Klaghoer R, Zollinger A. How reliable is lengthbased determination of body weight and tracheal tube size in the paediatric age group? The Broselow tape reconsidered. Br J Anaesth 2002;88:283–5. 27. Nieman CT, Manacci CF, Super DM, Mancusco C, Fallon WF. Use of the Broselow tape may result in the underresuscitation of children. Acad Emerg Med 2006;13:1011–9. 28. Shakir A, Morley D. Measuring malnutrition. Lancet 1974;1:758–9. 29. Bartram JL, Rigby AS, Baxter PS. The “Lasso-o” tape: stretchability and observer variability in head circumference measurement. Arch Dis Child 2005;90:820–1. 30. Lack JA, Stuart-Taylor ME. Calculation of drug dosage and body surface area in children. Br J Anaesth 1997;78:601–5.