Article Text

Download PDFPDF

Validation of weight estimation by age and length based methods in the Western Cape, South Africa population
  1. Heike Geduld,
  2. Peter W Hodkinson,
  3. Lee A Wallis
  1. Division of Emergency Medicine, University of Cape Town and Stellenbosch, Cape Town, South Africa
  1. Correspondence to Dr P W Hodkinson, PO Box 38227, Eshowe 3894, South Africa; pwhodkinson{at}gmail.com

Abstract

Objective To evaluate four paediatric weight estimation methods (APLS, Luscombe and Owens, Best Guess and Broselow tape) in order to determine which are accurate for weight estimation in South African children.

Method From a database of 2832 children aged 1–10 years seen at Red Cross Hospital in Cape Town, measured weight was compared to estimated weights from all four methods.

Results APLS formula and the Broselow Tape showed the best correlation with measured weight. Mean error was 3.3% for APLS (for 1–10-year olds) and 0.9% for Broselow tape (children <145 cm length and <35 kg). Both the Best Guess and Luscombe and Owens formulae tended to overestimate weight (15.4% and 12.4%, respectively).

Conclusion The Broselow tape and APLS estimation methods are most accurate in estimating weight in the Western Cape paediatric population, even though they have a small tendency to underestimate weight. Clinicians need to bear in mind that none of the formulae are infallible and constant reassessment and clinical judgement should be used, as well as a measured weight as soon as possible in an emergency situation.

  • South Africa
  • weight
  • age
  • estimation
  • child
  • paediatrics
  • epidemiology
  • paediatric resuscitation

Statistics from Altmetric.com

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Introduction

Resuscitation of a seriously ill or injured child is stressful and difficult. It is impractical to weigh the child before any intervention is instituted—access to weighing equipment may be limited, or the child may be in pain, acutely ill or injured which precludes weighing. Healthcare workers (and parents) are notoriously bad at estimating the weight of children.1 2 Accurate weight assessment is crucial in implementing effective resuscitation interventions: most medication dosages in children are calculated per kilogram of body weight; equipment size is determined by patient size (and therefore weight), and resuscitative interventions (such as defibrillation energy or fluid bolus volumes), are also calculated by weight in kilograms.

There are multiple formulae available for estimating children's weight, but the accuracy of the different formulae varies in terms of the populations in which they were developed.3–5They are often geographically or ethnically specific. The parameters used to estimate weight may also vary in their appropriateness and not be consistent over all ages.6

Age based calculation methods

The oldest and most commonly used formulae are based on age to weight ratios. These are simple to use, requiring nothing from the healthcare worker other than to remember the formula and undertake a simple calculation. They can be used in any environment and do not require resources such as measuring tapes or tables. The most well known formula is that taught on the Advanced Paediatric Life Support (APLS) course7:

Weight(kg)=(Age+4)×2

This formula was derived from National Centre for Health Statistics (NCHS) population data from 1977 in the United States.8 However, subsequent studies from the USA and the UK have suggested that the APLS formula underestimates children's weight by up to 20% in developed world studies.3 9 10

Luscombe and Owens developed a new formula (specifically for the UK) in 2007.11 Their premise was that the APLS formula underestimates weight, and that a new age based formula was required. They were able to derive a new formula:

Weight(kg)=(3× Age)+7

The Best Guess formula was recently derived in Australia.12 Formulae were developed in three age categories corresponding to the different growth velocities at different ages:

Infants<12 months: Weight(kg)=(Age in months+9)/2Children1-4 years: Weight(kg)=2×(Age+5)Children5-14 years: Weight(kg)=4×(Age)

Length-to-weight based methods

There has been a shift to using length based calculations which show less variation with body habitus; they are thought to be more accurate and may show less variability with age.10 However, they do rely on physical measurement which needs to be accurate, often involve training, and need accurate measuring tools. Measurement may also be time-consuming in emergency situations.

The Broselow–Luten tape is a colour-coded tape which is placed next to the patient.13 The child is measured from head to toe, and by length is assigned to a specific weight class. The usefulness of the tape extends beyond weight estimation, as the colour coded weight classes also list dosages of commonly used resuscitation drugs and the sizes of resuscitation equipment (such as endotracheal tubes and suction catheters). This decreases the cognitive workload placed on the clinician. The Broselow tape was also derived from the 1977 NCHS data.13 Several studies have shown good correlation between Broselow weights and measured weights (in US children), with a trend towards underestimation.14–16

In the developing world, the tape still has good correlation with measured weights: two Indian studies showed it to be an accurate method of weight estimation, especially for children under 10 kg.17 18 In Hong Kong, Kun found a good statistical correlation with measured weight between 10 and 25 kg, but above 25 kg the tape tended to underestimate.19

Developing world weight estimation

It is hard to predict whether the traditional formulae hold true in South Africa (where there is a mix of developed and developing world populations).20 21 This study sought to compare the different methods of weight estimation and assess their accuracy in South African children, and consequently to make recommendations for which method to recommend and teach in the South African setting.

Methods

Study location and data collection

The Red Cross War Memorial Children's Hospital is located in Cape Town, South Africa and is the tertiary referral centre for paediatric emergencies in the Western Cape Province. We developed a prospective database of all emergency patients aged 1–10 years attending the hospital over a 9-month period. In developing the database, standardisation of measurement of weight and height was achieved by staff education, use of regularly calibrated scales and laminated measuring tapes. Age was rounded down to the last completed year; height was rounded off to the nearest 1 cm, and weight rounded off to the nearest 0.1 kg. Only children from age 1 year up to and including 10 years of age, who had both weight and height data recorded, were included in this study. Of note, the Broselow tape estimates are only valid for weight <35 kg and height <145 cm, so patients above these limits were excluded for the Broselow estimates only (115 patients excluded).

Data analysis

Statistical analysis of the data was performed using standard statistical methods, Microsoft Excel 2007 and Intercooled STATA V.10.0. A p value <0.05 was considered significant for statistical tests.

A Bland–Altman analysis was performed to compare the different estimates and evaluate their bias and precision, and to maintain consistency with previous literature on the subject.22

An analysis of how many of the estimates gave a predicted weight within 10% of the measured weight was performed.

Ethical considerations

The study was approved by the Human Research Ethics Committee at the University of Cape Town. Patient confidentiality was maintained as no identifiers to individual patients were recorded or stored in the database.

Outcomes

The primary outcome was the relationship between measured weight and estimated weights as calculated from the four tools (table 1).

Table 1

Description of formulae and limitations

Mean percentage difference (or error) between measured and estimated weights was calculated as a means of expressing the relationship between the two values (measured weight − estimated weight/measured weight×100).

Results

A total of 2832 patients were included in the study; of these, 1779 (62.8%) were male. Combined male and female results are presented as separate gender analysis yielded similar conclusions. The mean (SD) age of the patients was 4.7 (2.9) years. Although ethnic distribution was not taken into account in the evaluation of weights, the same population has been described elsewhere as composed of the following ethnic groups: black 56%, coloured 30%, Asian 8%, white 4%, and other 2%.21 Table 2 presents the patient numbers as well as the mean measured weights for each age.

Table 2

Mean frequency and weights for each year of age (kg)

Figure 1 shows the mean measured weights at each age, and the estimated weights using the four different formulae, as well as recent UK measured weight data for comparison.11 Table 3 shows the mean percentage error, as a measure of deviation from measured weight, for each estimation method, and is subdivided to allow comparison of ages 1–5 and 6–10; as well as the percentage of estimates correctly predicting the weight to within 10% of the measured weight. The Broselow tape method appears to best estimate the weight in all ages of children as the mean percentage error is the smallest at 0.89%; it also has 64.2% of estimates within 10% of the measured weight.

Figure 1

Measured and derived weights for each of the estimation methods for each age. SA MEASURED, from this database; APLS (Advanced Paediatric Life Support formula), derived; LO (Luscombe and Owens), derived; BG (Best Guess), derived; BLT (Broselow Luten tape), derived; UK MEASURED, from Luscombe and Owens 2007.

Table 3

Mean percentage error and accuracy for estimated weights (negative values suggest overestimation)

Figure 2 shows the four plots for the difference between estimated and measured weight against measured weight for each estimation method. The closer the scatter grouping in each plot, the better the agreement and thus accuracy of the method, while deviation from the zero difference line shows the bias. Table 4 shows the Bland–Altman limits of agreement (LOA) to compare the reliability and precision of each method (a small LOA reflects good reliability). The LOA are similar for all estimates and reflect a similar picture to that given by the mean percentage error analysis, with very small CIs due to the large sample.

Figure 2

Observed weight versus difference between observed and expected for: (A) APLS (Advanced Paediatric Life Support); (B) Luscombe and Owens; (C) Best Guess; (D) Broselow tape.

Table 4

Bland–Altman comparison of bias and precision for each estimation method

Discussion

There is no validated universal tool for the estimation of paediatric weight, with particular concern over the applicability of developed world tools in developing world populations. The South African population has a mixture of ethnicities and different socioeconomic levels and is regarded as a rapidly developing country.20 It is thus unclear and unstudied which, if any, of the paediatric weight estimation tools are most appropriate for use in South Africa. This large database represents a sample paediatric population from South Africa, with an age and sex distribution typical of childhood emergency presentations.23

The Broselow tape gave the best estimated weight within this population when used within its weight and height limitations. The mean error is an underestimation of 0.89%. The Broselow tape has advantages other than a reasonably accurate weight estimate: it reduces cognitive load by calculating weight, listing appropriate sized equipment and the doses of commonly used drugs for resuscitation, thus reducing the potential for medication error. It can be used with a system of prepackaged equipment. Broselow tapes are a useful and cost effective tool that provides an accurate assessment of weight.

The APLS formula, which is easy to remember and already widely used throughout the Western Cape, correlates well with measured weights. The mean underestimation of 3.27% increases with increasing age. It is most accurate in the younger age groups. It is clearly accurate in this population; it is familiar to many practitioners in this country, is easy to remember and allows for simple mental arithmetic. In higher socioeconomic groups it is probable that the APLS formula will underestimate weight more significantly as it has been shown to do in developed world populations.16

Luscombe and Owens and the Best Guess formula consistently overestimate the weights of children in our population: 12.36% for Luscombe and Owens, and 15.41% for Best Guess. Thus it would seem that the best weight estimation tools for this population are those based on the 1977 US data; the two more recent formulae which use UK and Australian children's data in 2007 and 2005, respectively, would seem to overestimate weights for this population. This is perhaps an expected finding as it has been demonstrated that most developed world children are becoming heavier for their ages (possibly through increased body mass index (BMI)).24

Does incorrect weight estimation have an effect on morbidity and mortality in paediatric resuscitation? Although statistical differences between measured weight and estimated weights exist for all four of the age–weight estimation techniques, the true differences are perhaps small, but at best only 64% (Broselow tape) and at worst 31% (Best Guess) of the estimates are within 10% of the measured weight. Whether the differences are clinically significant when used by a clinician who is constantly evaluating the effects of therapy is unlikely.23 There is little clarity as to what dose of resuscitation drugs is ideal: much is dependent on the child's metabolism and premorbid state. It is logical though that the smaller the child, the more likely it is that incorrect doses will have an effect. Physiologically, fluid and energy expenditure have been calculated from lean body mass, therefore drug dosages and fluid volumes correlate best with lean body mass. The increase in size in populations, particularly in the developed world, is an increase in adiposity and BMI rather than an increase in height and lean body mass.24 Nevertheless, at the present time the gold standard is still a measured weight: as soon as the patient is stable, every effort should be made to weigh the child as this allows for accurate management during the course of their hospital stay—trolley or bed scales are useful tools for this. None of these formulae should be used in isolation—it remains the responsibility of the physician to use clinical judgement in conjunction with any weight estimation method.

There is a need for a prospective multicentre study to validate the Broselow tape and APLS formula for the South African population across all socioeconomic and ethnic groups. It may be necessary to repeat this research every few decades as growth patterns in the country change.

Limitations

One of the limitations of this study is that the sample was taken at one hospital in the Western Cape. Although Red Cross Hospital receives patients from all over the Western Cape and across all social groups, the sample may only be reflective of this population and not be translatable to the rest of the country. The sample numbers were large however and this would mitigate these effects to some extent.

The study population is currently lighter than its developed world counterparts,20 24 but in a few years may also outgrow the current weight estimation tools.

This study only looked at children aged 10 years and less, which ignores the perhaps more difficult prepubescent ages, with greater variation dependent on the onset of puberty and perhaps more influence of genetic factors. In addition, the exclusions of the Broselow tape method which specifies an upper weight and height limit, makes direct comparison between this and the other methods more difficult and will tend to make the Broselow tape estimation appear more accurate.

Conclusion

The Broselow tape and APLS methods are both accurate in the Western Cape population for children aged 10 years and younger at present and should be advocated for all personnel treating children in emergency situations. Other estimation tools such as the Best Guess and Luscombe and Owens formulae, which are based on recent developed country data, tend to overestimate weight in all age groups and may introduce unacceptable error to the emergency treatment. Further studies may be necessary to allow extrapolation to the greater South African population, and to validate these tools as the population changes with time.25

Acknowledgments

Special thanks are extended to Dr Landon Myers for help with the data analysis.

References

Footnotes

  • Funding This study was presented in partial fulfilment of the MMed (Emergency Medicine) at the University of Cape Town (HG).

  • Competing interests None.

  • Ethics approval This study was conducted with the approval of the University of Cape Town.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Linked Articles

  • Primary survey
    Geoff Hughes