|Year : 2020 | Volume
| Issue : 9 | Page : 1229-1236
Comparison of age-based weight estimation with actual measured weight in children aged one to twelve years in Enugu
BO Edelu1, KK Iloh1, OO Igbokwe1, CI D Osuorah2, ON Iloh1, IK Ndu3, JN Eze1, IN Obumneme-Anyim1, OC Nduagubam3, UC Akubilo1
1 Department of Paediatrics, University of Nigeria Teaching Hospital, Enugu, Enugu State, Nigeria
2 Child Survival Unit, Medical Research Council UK, The Gambia Unit, Fajara, The Gambia, West Africa
3 Department of Paediatrics, Enugu State University Teaching Hospital, Enugu, Enugu State, Nigeria
|Date of Submission||28-Jan-2020|
|Date of Acceptance||09-Jun-2020|
|Date of Web Publication||10-Sep-2020|
Dr. K K Iloh
Department of Paediatrics, College of Medicine, University of Nigeria, Enugu
Source of Support: None, Conflict of Interest: None
| Abstract|| |
Aims: This study was carried out to evaluate the degree of accuracy of age-based weight estimation methods in assessing the weight of the Nigerian child. Method: The weights of one thousand, four hundred and fifty-six (1,456) children were measured and compared with the updated Advanced Paediatric Life Support (APLS), Best guess, Nelson and Luscombe & Owen methods. Result: The updated APLS, Nelson and Luscombe & Owen methods underestimated the weights in younger children while overestimating in older ones. Best guess overestimated the weights across all ages. The Nelson formula had the best agreement within 10% and 20% of the measured weights among all methods. A linear regression analysis produced an equation for weight estimation: weight (W) = 2.058 Y + 9.925, where W is weight in kilogram and Y is the age in years. Conclusion: None of the weight estimation formulae assessed was entirely accurate in our study, though the Nelson method showed superior agreement.
Keywords: Accuracy, children, estimated weight, measured weight
|How to cite this article:|
Edelu B O, Iloh K K, Igbokwe O O, D Osuorah C I, Iloh O N, Ndu I K, Eze J N, Obumneme-Anyim I N, Nduagubam O C, Akubilo U C. Comparison of age-based weight estimation with actual measured weight in children aged one to twelve years in Enugu. Niger J Clin Pract 2020;23:1229-36
|How to cite this URL:|
Edelu B O, Iloh K K, Igbokwe O O, D Osuorah C I, Iloh O N, Ndu I K, Eze J N, Obumneme-Anyim I N, Nduagubam O C, Akubilo U C. Comparison of age-based weight estimation with actual measured weight in children aged one to twelve years in Enugu. Niger J Clin Pract [serial online] 2020 [cited 2020 Sep 27];23:1229-36. Available from: http://www.njcponline.com/text.asp?2020/23/9/1229/294682
| Introduction|| |
Weight determination is an essential part of pediatric practice all over the world. It is necessary in making both diagnostic and treatment decisions. Whether it's an emergency or routine clinical practice, the weight of the patient is a vital information which the health care provider requires to prescribe drugs and intravenous fluids at the right dosages as well as provide the appropriate sizes of equipment and energy levels for defibrillation.,,
The most accurate method of determining a child's weight is to weigh the child on a standard machine with calibrated scales. However, in emergency situations, getting the actual weight of a child may be an arduous task and valuable time might be lost in addressing the immediate need of the critically ill child.
Consequently, different weight estimation methods have been developed by different researchers in order to provide a simplified way of calculating the weight of a paediatric patient. One way of doing that is to calculate the weight from the child's age using a formula. Some of these methods which are based on the age of the child include the original and updated Advanced Paediatric Life Support (APLS) formulae,, the Australian formula known as Best Guess, the Nelson formula, and Luscombe & Owen formula. The limitation of the age based methods is that the actual age of the child must be known in order to get an estimate that is as close as possible to the actual weight of the child.
There is need therefore to determine which of these weight estimation methods correlates best with the actual body weight. The aim of this study was to determine which methods among the updated APLS, Best guess, Nelson formulae and the Luscombe and Owen formula is closest to the measured weight. We secondarily evaluated the linear regression equation that best predicts the weight of surveyed children based on their age at last birthday in a developing community.
| Methodology|| |
This cross-sectional multi-center study was conducted concurrently over a 2½ year period between January 2017 and May 2019 in two different tertiary health facilities namely the University of Nigeria Teaching Hospital (UNTH) and the Enugu State University Teaching Hospital (ESUTH), both in Enugu state, a predominantly Igbo speaking state. Each of these hospitals offers specialized medical services and serves as a referral center to primary, secondary and private health facilities mostly from within, but also outside Enugu state.
Ethical clearance was sought for and obtained from the Health Research and Ethics Committee of the hospitals prior to the commencement of the study (NHREC/05/01/2008B-FWA00002459-1 RB00002323).
Participants were children aged 1-12 years old presenting to the emergency room and out-patient clinic. They were consecutively enrolled into the study. Excluded were children whose parents/caregiver refused to give consent. Also excluded from this study were children with serious medical conditions such that an actual weight cannot be measured and children with any medical condition (such as amputation, dehydration, edema, growth hormone deficiency, severe joint contractures, or neurologic defects) which could substantially affect their weight and/or height. Participation was entirely voluntary without any form of inducement.
The children's actual weight was taken using an Omron digital scale (HN-289-EB) for children two years old and above, and for those less than two years old, SECA infant weighing scale was used. The study participants were weighed wearing only light clothings without footwears. Weights were rounded off to nearest 0.1 kg. At the beginning of each measurement day, the accuracy of the weighting scales was checked by using a known standardized weight placed on the scale. Before each measurement, the scale is usually turned to 'zero” to correct for zero error.
Height was measured to the nearest 0.1 cm using a stadiometer [SECA213, Hamburg August 2014] for children two years and above, while the length of children less than two years were measures using SECA headboard.
The weights of the enrollees were estimated using the following formulae:
- The updated APLS weight for ages 0-12 years:
- Infants 0-12 months: weight (kg) = (0.5 × age in months) + 4
- Children 1-5 years: weight (kg) = (2 × age in years) + 8
- Children 6-12 years: weight (kg) = (3 × age in years) + 7
- Best Guess weight as follows:
- Infants 1-11 months: weight (kg) = (age in months + 9)/2
- Children 1-4 years: weight (kg) = 2 × (age + 5)
- Children 5-14 years: weight (kg) = 4 × age
- Nelson's Formulae:
- 3-12 mo: (age in months + 9)/2;
- 1-6 yr: age in years × 2 + 8;
- 7-12 yr: (age in years × 7-5)/2]
- Luscombe and Owen:
- Weight (kg) = (3 × age) + 7
The weight estimates were done with electronic calculator.
Information on children's basic socio-demographic characteristics was also obtained. These included i) Age, categorized into 1-12 years ii) Gender, categorized as male and female iii) Socioeconomic status of child's family was calculated using Oyedeji's formula. This was further re-categorized as high, middle and low and iv) Body Mass Index (BMI), was calculated using their actual weight in kilogram and height in meters. This was categorized as underweight (<5th percentile), normal (5-85th percentile), overweight (85th-95th percentile) and obese (>95th percentile) based on their BMI for age and sex.
Sample size calculation
The number of respondents enrolled in this study was calculated using the Cochran formula, N = f(α,β).2s2/(δ)2 for calculation of sample size based on a confidence interval of 95% which is equivalent to a confidence coefficient of 1.96. The minimum sample size will be 84 in each age group, giving a minimum sample size of 1008 children.
Data collection and statistical analysis
Data collection was done using questionnaires administered by one of the researchers and/or trained research assistants. Information were inputted into the relevant sections of the questionnaire and subsequently transferred into a Microsoft Excel Sheet. Statistical analysis was performed using SPSS (version 21; SPSS Inc., Chicago, IL USA) software considering 95% confidence interval in demographic data analyses. The difference between the actual weight and the estimated weights by updated APLS, Best guess, Nelson, Luscombe & Owen formulae were calculated and displayed using cross tabulation and Chi square test. Mean percentage errors were calculated to compare the actual and estimated weights.
A Bland-Altman plot was displayed to graphically present the bias and 95% limits of agreement. The percentage differences (errors) between estimated and measured weights were plotted on the y-axis while the averages of the two were on the x-axis. The dotted lines represent the limits of agreement (confidence interval) showing the degree of reliability while the spread of the scattered points depict the extent of agreement.
| Result|| |
Demographics of children enrolled in study
[Table 1] shows the characteristics of children surveyed. In all, 1456 consented and were enrolled with a modal age distribution of 7 years. The mean age was 6.89 ± 2.89 years (range 1 to 12 years).
There were a total of 669 males with a male to female ratio of 1:1.2. The actual weight of surveyed children ranged from 5 to 56 kg with a mean weight of 24.01 ± 7.27 kilograms. One thousand, two hundred and twenty six (84.2%) of the children had normal body mass index, 1.2% had severe thinness, while 4.5% were obese.
Paired comparisons of measured and estimated weight by age
[Table 2] shows a summary of mean weight estimation using the updated Advanced Pediatrics Life support (APLS), Best Guess, Nelson Luscombe & Owen Formulae. The differences between means of measured and estimated weights of children in each age category are as shown in [Table 3]. Weights estimated using the updated APLS significantly underestimated weight in the 1-5 years and significantly overestimated weight in the 6-12 years age groups. Unlike the updated APLS, the best guess method of weight approximation overestimated weight in all age category except in the 4 and 5 years age groups were it underestimated weight by -0.52 ± 2.77 kg (P = 0.124) and -0.55 ± 3.67 (P = 0.986) respectively and both of which were not significantly different from the actual mean weight.
|Table 2: Stratification of measured and estimated weights using the update APLS, Best guess, Nelson formula and Luscombe & Owen weight estimation methods|
Click here to view
|Table 3: Mean difference between measured weight and estimated weights using the updated APLS, Best Guess, Nelson formulae and Luscombe & Owen methods|
Click here to view
Furthermore, the estimated weight using the Nelson formulae was not significantly different from the actual mean weight in age groups 1 (-0.21 ± 1.94, P = 0.361) and 9 year (0.15 ± 5.36, P = 0.679). However, it significantly underestimated and overestimated weight in age groups 2-8 and 10-12 years, respectively [Table 3]. Finally, the Luscombe and Owen weight estimation method underestimated weight in 1 (-0.53 ± 1.81, P = 0.007) and 2 years old (-0.56 ± 1.98, P = 0.004) and significantly overestimated in all other age categories except 3 (0.39 ± 2.41, P = 0.155) 4 (0.40 ± 2.78, P = 0.256) and 7-years old (2.90 ± 4.57, P = 0.523) ± 4.57, P = 0.523).
Comparisons of estimated and measured weight
The degree of deviation of the estimated weight from the actual weight also referred to as the mean percentage error (or bias estimation) was computed for the four weight estimation methods considered in this study and the results presented in [Table 4]. It was noted that the updated APLS underestimated weight in 1 year-5 years old children by 2.43%, 8.77%, 7.77%, 11.99% and 11.01%, respectively. Conversely, the APLS overestimated weight in 6 to 12 years old children by 11.18%, 16.73%, 18.47%, 20.47%, 27.25, 35.63 and 38.69%, respectively.
|Table 4: Mean percentage error for estimated weights using updated APLS, Best guess, Nelson formula and Luscombe & Owen methods|
Click here to view
The Best guess technique mostly overestimated weights by wide margins in all other age categories except 4- and 5-year-olds. The overestimation ranged from 5.45% in 2-year group to 55.19% in 12-years. However, it slightly underestimated weight in the four and 5-year old groups by 0.55% and 0.32% respectively.
Unlike the Best guess technique, the Nelson formula equally underestimated and overestimated weights across all the age groups, but with narrower margins compared to the other methods. Its weight underestimation was in 2 to 8-year groups and ranged from 2.58% in 8-year to 11.25% in 4-year old children. In contrast, it overestimated weight by 1.1% in 1-years, 2.89% in 9-years, 11.38% in 10-years, 22.07% in 11-years and 27.52% in 12 year old children.
The Luscombe and Owen method of weight estimation like the best guess method mostly overestimated weight but unlike the best guess method, its mean percentage error is narrower. Apart from the age groups 1 and 2 year where it significantly underestimated weights by 2.04% and 2.23%, respectively, the Luscombe and Owen methods increasingly overestimated weight with older age ranging from a mean percentage error of 4.79% in 3-year-old to 38.81% in 12-year-old respondents [Table 4].
Agreements between measured and estimated weights
Of the 4 weight estimation methods assessed, it was noted that the Nelson formula had the best agreement with the actual weight measure in surveyed children. The agreement analysis summarized in [Table 5] shows that overall, the agreement of all 4 methods of weight estimation with the actual measured was better in the lower age groups. The Nelson formula showed a superior agreement within 10-20% measured weight in most age categories with an overall agreement of 42.2% (614/1456) within 10% of measured weight and 75.1% (1093/1456) weight estimated within 20% of measured weight [See [Table 5].
|Table 5: Agreement within 10 and 20% measured weight of estimated weight using the updated APLS, Best guess, Nelson formula and Luscombe & Owen methods|
Click here to view
Figures I-IV represent the Bland-Altman plot for graphic representation of the estimated weights with measured weight. In the APLS method [Figure I], it was noted that in the younger age categories, the estimated weights were well above the lines of agreement, implying overestimation in this age categories, while in the older age categories, the formula mostly underestimated weights in surveyed children. Like the APLS, the Bland-Altman plot for best Guess method [Figure II] showed similar pattern of weight overestimation in the younger and under estimation in the older children. The plot for Nelson formula [Figure III] unlike the APLS and Best Guess formulae showed that the formula both under and overestimated weights across all age categories. Finally, it was noted that the Luscombe formula [Figure IV] mostly underestimated weights of children in all age categories of surveyed children. It is worth noting that in all 4 weight estimation formulae, the degree of weight underestimation was more as age or weight increased.
|Figures 1: I-IV: Bland-Altman plot of difference and average of measured and estimated weight using (I) updated APLS, (II) Best guess (III) Nelson formula and (IV) Luscombe and Owen weight estimation method for children aged 1-12 years|
Click here to view
Correlation of measured weight with age in surveyed children
Linear regression analysis showed that for every increase in age by 1 year in children surveyed, the weight increased by a factor of 2.058 giving a regression equation of weight (W) = 2.058 Y + 9.925, where W is weight in kilogram and Y is the age in years at the last birth day.
| Discussion|| |
Weight determination is an important tool in the practise of paediatrics. In emergency situations or where actual weight cannot be measured, weight estimation is commonly done and it is expected that the estimated weight does not vary significantly from the actual weight. In this study, wide variations were observed between the actual weight and estimated weight using the age based formulae: updated APLS, Best Guess, Nelson Formula, and the Luscombe and Owen formula. It was observed that the age-based formulae tended to overestimate the weights of older children and underestimate those of younger children.
Of the four methods of weight estimation compared in the current study, the Nelson formula had the closest mean and agreement to the actual weight of the study participants. Although, the said formula underestimated the weights in children up to 8 years and overestimated in 9 to 12 years, it did so with a narrower margin than the Best guess and APLS, as well as the Luscombe and Owen methods. In addition, it had a more uniform pattern of agreement with equal underestimation and overestimation across all age groups. This, however, does not negate the fact that a significant difference exists between the Nelson formula and the actual weight in all but 2 age groups, making it still inaccurate for use. Omisanjo et al. and Eke and colleagues in South-west and South-east Nigeria respectively also documented significant variations in the Nelson formula weight estimates compared to the actual weight. This finding was corroborated by other researchers in Kenya, Cameroun and India.,, This suggests that the Nelson formula which was devised many years ago, may have questionable validity for weight estimation in children from low to middle income countries as it was primarily developed in Western countries using the weights of mostly Caucasian children.
The updated APLS formula was noted to significantly underestimate the weights of younger children (1-5 years) while overestimating the weight in the older age group (6-12 years). This varies slightly with the findings by some other authors while maintaining the same general inaccuracy. In the work of Luscombe et al. in the United Kingdom (UK), APLS underestimated the weights of children across all ages while Loo et al. in Singapore found that the APLS significantly underestimated the weights of children 1-10 years. On the other hand, Graveset al. in Australia observed that the APLS formula was more accurate in children 6-10 years.
In our study, the “Best Guess” method overestimated the weight of children across all ages except for the 4 and 5 year olds where there was marginal underestimation. In the same vein, Omisanjo et al. among South-west Nigerian children demonstrated that the “Best Guess” formula consistently overestimated the weights of the children aged 3 months to 11 years by as much as 10.11 to 30.67%.
The formula by Luscombe and Owens, underestimated children 1 to 2 years in our study while overestimating the older children. Geduld and colleagues in South Africa also documented that the formula overestimated the weights of children by as much as 12.4%, similar to the best guess formula that overestimated by 15.4%. On the contrary, Kelly et al. in Australia validated the Luscombe formula as being accurate. The variations in the estimated weights using this Luscombes formula could probably be due to the fact that this formula was derived from a population that comprised children from more developed countries with probably higher weights and body mass indices than those of African children.
This study has demonstrated that the existing formulas for age based weight estimation do not yield reliable weight estimates in all age groups and in emergency situations, this could have dire consequences. In the course of our study it was observed that for every increase in age by 1 year in children surveyed, the weight increased by a factor of 2.058, and hence a regression equation of weight (W) = 2.058 Y + 9.925, as a possible formula for weight estimation in Nigerian children where W is weight in kilogram and Y is the age in years at the last birth day. Large scale studies will however be needed to validate the accuracy of this formula.
| Conclusion|| |
None of the weight estimation formulae assessed was entirely accurate in our study, though the Nelson method showed superior agreement. Efforts should therefore be made to obtain the actual weight of any child presenting to the emergency department as much as possible and where this is impossible, the Nelson formula may be used as it had the closest correlation with actual weight in our study population.
Declaration of patient consent
The authors certify that they have obtained all appropriate patient consent forms. In the form the patient (s) has/have given his/her/their consent for his/her/their images and other clinical information to be reported in the journal. The patients understand that their names and initials will not be published and due efforts will be made to conceal their identity, but anonymity cannot be guaranteed.
Financial support and sponsorship
Conflicts of interest
There are no conflicts of interest.
| References|| |
Omisanjo AO, Orimadegun AE and Akinbami FO. Accuracy of nelson and best guess formulae in estimation of weights in Nigerian children population. Ann Ib Postgard Med 2014;12:80-8.
Black K, Barnett P, Young S. Are methods used to estimate weights in children accurate&? Emerg Med (Fremantle) 2002;14:160-5.
Iloh ON, Edelu B, Iloh KK, Igbokwe OO, Ndu IK, Nduagubam OC, et al.
Weight estimation in Paediatrics: How accurate is the Broselow-tape weight estimation in the Nigerian child. Ital J Pediatr 2019;45:146-52.
Advanced Life Support Group. Advanced Paediatric Life Support: The Practical Approach. 5th
ed.. UK, Chichester: John Wiley & Sons Ltd.; 2011.
Graves L, Chayen G, Peat J, O'Leary F. A comparison of actual to estimated weights in Australian children attending a tertiary children's' hospital, using the original and updated APLS, Luscombe and Owens, Best Guess formulae and the Broselow tape. Resuscitation 2014;85:392-6.
Tinning K, Acworth J. Make your Best Guess: An updated method for paediatric weight estimation in emergencies. Emerg Med Australas 2007;19:528-34.
Keane V. Assessment of growth. In: Kliegman RM, Behrman RE, Jenson HB, editors. Nelson Textbook of Pediatrics. 18th ed.. Philadelphia: Saunders Elsevier; 2007: pg 31.
Luscombe M, Owens B. Weight estimation in resuscitation: Is the current formula still valid? Arch Dis Childhood 2007;92:412-5.
Oyedeji GA. Socioeconomic and cultural background of hospitalized children in Ilesha. Nig J Paediatr 1985;12:111-7.
Eke CB, Ubesie AC, Ibe BC. Comparison of actual (measured) weights and heights with the standard formula methods of estimation among children in Enugu. Nig J Paed 2014;41:307-11.
House DR, Ngetich E, Vreeman RC, Rusyniak DE. Estimating the weight of children in Kenya. Do the Broselow tape and age-based formula measure up&? Ann Emerg Med 2012:61:1-8.
Mbanga CM, Agbor VN, Chobufo D, Louise DT, Tsi N. Validation of Paediatric weight estimation formulae in a suburban Camerounian Population; A cross sectional study. J Trop Paed 2019;65:90-7.
Varghese A, Vasudevan VK, Lewin S, Indumathi CK, Dinakar C, Rao SD. Do the length based (Broselow) Tape, APLS, Argall and Nelson's formulae accurately estimate weight of Indian children. Ind Pediatr 2006;43:889-94.
Luscombe MD, Owens BD, Weight estimation in paediatrics: A comparism of the APLS formula and the formula weight=3(age+7). Emerg Med J 2011;28:590-3.
Loo PY, Chong SL, Lek N, Bautista D, Ng KC. Evaluation of three paediatric weight estimation methods in Singapore. J Paediatr Child Health 2013;49:E311-6.
Geduld H, Hodkinson PW, Wallis LA. Validation of weight estimation by age and length based methods in the Western Cape, South Africa population. Emerg Med J 2011;28:856-60.
Kelly AM, Nguyen K, Krieser D. Validation of the Luscombe weight formula for estimating children's weight. Emer Med Australas 2011;3:59-62.
[Table 1], [Table 2], [Table 3], [Table 4], [Table 5]