Almost all paediatric disease states exert effects on body composition. Many diseases impact on weight and growth, but often effects are exerted on more specific outcomes (eg, body fat level, bone mineral density) or regions (eg, central adiposity, limb muscle mass). In some cases, such effects are transient and might be of little clinical importance, but many diseases impose stronger and persistent effects. Despite its sensitivity to disease, and growing awareness of its importance as an adult health outcome, body composition currently receives little attention in paediatric clinical practice.

We believe this situation can be associated historically with three factors, none of which remains justified. First, body composition has traditionally been considered difficult to measure. While weight-for-height (or more recently body mass index, BMI) and skinfold thicknesses have been measured routinely for decades, these measurements provide little information on lean mass. Until recently, many two-component techniques, which divide body weight into fat and lean components, were either unsuitable for routine paediatric application or rarely available. Clinicians interested in body composition as an outcome have therefore been hindered by the lack of appropriate methodologies, and typically still rely primarily on BMI. However, we have recently reviewed the techniques available for measurement of body composition in paediatric clinical practice, and argued that a variety of widely-available techniques can now fulfil different functions in diverse circumstances.1

Second, body composition has not been considered a useful predictor of clinical outcome. We suggest that this perception can be attributed to a lack of studies addressing this issue rather than the fact that no such relationships exist. As we review below, evidence from both adults and children increasingly demonstrates important relationships between baseline body composition and the risk/progress of diseases and their treatment.

Third, some areas of clinical practice assume that knowledge of adult physiology can simply be scaled down according to body size. For example, many treatments use dosages designed for adults but given on a per kilogram body weight basis to children.2 This approach may be inappropriate, given that many physiological processes scale allometrically with body size. Size adjustment is addressed in some branches of paediatrics, but in most cases nutritional management and treatment remain adjusted for body weight rather than the functional components of that weight.

In this article, we consider three areas in which there is increasing evidence that body composition is important, and where its measurement could aid clinical practice, and one further area where we believe that further research is justified (fig 1).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1 Potential benefits of measuring body composition in paediatric clinical practice.

Monitoring

In some disease states, alterations in body composition are one of the primary symptoms, for example obesity or eating disorders involving severe weight loss. These conditions are currently diagnosed on the basis of indices of weight relative to height, such as BMI. However, the same indices are a poor option for monitoring disease progression and response to treatment. There is increasing evidence both of the poor ability of BMI to predict body fat in individuals3 and of the value of information about lean mass.4

At the present time, the criteria for diagnosing obesity comprise standard deviation scores for BMI. However, although excess weight itself can exert adverse effects on some health outcomes, the health risks of obesity stem primarily from excess abdominal fat, in particular visceral fat. There is a two-fold variability in body fat for a given BMI value in children.3 Within those children categorised as obese, there is again wide variability in the relative level of fat for a given BMI value.5 For this reason, measurement of waist circumference is increasingly advocated as an additional tool for identifying those at greatest risk of excess weight.6

In adults, BMI has a U-shaped association with mortality, whereas body fat has a linear association.7 This relationship can be attributed to different relationships between mortality and fat mass as opposed to lean mass – both high levels of fat mass and low levels of lean mass are independently associated with poorer adult health. Thus, it is apparent that when monitoring obese patients, more detailed information about body composition would improve evaluation of treatment efficacy.

Children with eating disorders are likewise diagnosed using a combination of BMI and psychological criteria.8 Historically, due to the high prevalence of such disorders in adolescent girls, much attention during treatment has focused on the putative role of body fat stores in restoring menstrual function, and much less attention has been directed to other outcomes. However, recent data on young patients suggest that loss of lean mass is as marked as loss of fat mass.9 The long-term implications of deficits in lean mass are not known, but are likely to contribute to suboptimal bone health since lean mass is an important determinant of bone mass accretion.4 In the elderly, sarcopenia is associated with an increased risk of type 2 diabetes,10 hence low levels of lean mass in earlier life might likewise predispose to an increased subsequent risk of this disease. This hypothesis requires exploration.

HIV infection represents another condition where BMI is too crude to differentiate changes in fat versus lean mass, or redistribution of fat stores through lipodystrophy. In adults, serial body composition measurements have been used to identify marked loss of lean mass, allowing the derivation of appropriate management strategies.11

BMI, therefore, has significant limitations for monitoring patients with unhealthy nutritional status over time. Longitudinal measurements of BMI cannot discriminate relative changes in fat and lean mass, and in obese children, increases in activity may promote both gains in lean mass and loss of fat stores, resulting in weight maintenance. In eating disorders, gains in functional lean mass are preferable to regaining fat mass alone. A study of 92 adolescents with anorexia nervosa showed that measures of body cell mass by potassium scanning provided a substantially better basis on which to base the management of refeeding programmes than BMI.12

Even simple measurements may prove beneficial in this context. Waist circumference has been shown to have a good relationship with abdominal fat in children, and also shows an association directly with visceral fat.13 Decreases in waist circumference in children have been linked to decreases in cardiovascular risk profile.14 We have recently shown that DXA suffers from variable bias in its measurement of total body fat mass.15 Nevertheless, DXA measurements could discern whether exercise interventions were increasing limb lean mass and hence increasing energy demand. For eating disorder patients, increases in limb lean mass could likewise be assessed using DXA, and increases in fatness by skinfold thicknesses and waist circumference. Some hospitals already use whole body potassium scanning to assess recovery of body cell mass16; however DXA could play a similar role for lean mass and is now widely available in the UK.

In other disease states, body composition may not be the primary outcome, but monitoring of it may still indicate relative treatment success. Examples include cystic fibrosis, renal disease and type 1 diabetes, where monitoring the composition of weight gain or loss could indicate the efficacy of treatment.

Body composition and the incubation of cardiovascular disease

While childhood obesity impacts adversely on short-term health,17 there is also increasing evidence that cardiovascular disease has its origins in childhood and adolescence, such that regardless of body composition in adult life, increased levels of body fat in childhood may themselves increase the risk of later disease.18 19 This area is controversial,20 21 not least because few cohort studies have measured childhood body composition with accuracy, thereby limiting the capacity to discriminate longitudinal effects of high body fatness. Recent studies using data on childhood BMI have questioned the long-term importance of childhood obesity for cardiovascular health.21 However, these observations may have reduced relevance for contemporary health policies since contemporary children appear to have a higher body fat content for a given BMI value compared to those in previous generations.22 Prospective studies with high-quality data on children’s body composition are required to address this issue.

Nevertheless, it has been known for over two decades that adverse cardiovascular outcome in adults is predicted more successfully by prior than current risk factors. This led to the concept of the incubation of cardiovascular disease.23 Within this conceptual framework, many studies are now detecting evidence of the atherosclerotic process in childhood.24 25 While visceral fat mass is most strongly associated with cardiovascular risk, lean mass is also important because, as the primary site of insulin-mediated glucose uptake, it comprises the key determinant of total body insulin sensitivity.26 Variations in lean mass, for example induced by childhood diet, exercise or growth patterns, therefore impact on insulin resistance, which plays a central role in the aetiology of cardiovascular disease.

In view of these findings, monitoring of body composition in paediatric patients could prove important in reducing the burden of later disease. For example, since management of acute lymphoblastic leukaemia with glucocorticoids is associated with excess weight gain and truncal fat accumulation, the treatment regime has been recommended to be as short as possible.27 28

Similarly, we have shown recently that young patients receiving artificial ventilation have lower than expected levels of lean mass, attributable to their reduced activity levels.29 However, weight tends to be within the normal range, such that the patients have high levels of fat. This scenario may apply to many categories of hospital patient in whom normal weight is often assumed to indicate good health. Similar findings for example have been reported in patients with cerebral palsy given a gastrostomy.30 In an extreme case, severely underweight patients with congenital myofibromatosis (−3 z-scores BMI) were found to be 40% fat.31 In such patients, it is clear that energy intake per se is not the limiting factor for healthy growth. It may be that increased protein or micronutrients are required, or that growth in lean mass is reduced due to insufficient stimulus from physical activity. In either case, dietary energy should be adjusted to avoid excess fat deposition; however, this can only be undertaken on the basis of evidence of high fatness. Even simple measurements such as the combination of BMI and waist circumference or skinfold thickness data would prove informative in such contexts.

Body composition as a predictor of clinical outcome

Various aspects of nutritional status and body composition, including quality and quantity of lean mass and adiposity, have been linked to clinical outcome in adults. The current obesity epidemic has focused attention on the harmful effects of excess body fat. For example, truncal fat mass was found to influence inflammation in end-stage renal disease,32 while obesity is a well-established risk factor for many other diseases. However, concern over adiposity represents a recent shift from a historical preoccupation with poor nutritional status, and the relationship between lean mass and outcome is now receiving increased attention once more. In elderly patients, lean mass at baseline predicted length of hospital stay,33 while in patients with chronic obstructive pulmonary disease, lean mass predicted mortality whereas fat mass did not.34 Body composition techniques also allow functional investigations. For example, indices of cellular catabolism obtained from bioelectrical impedance analysis have been shown to predict clinical progression and survival in patients with HIV/AIDS35 or suspected bacteraemia.36

Fewer studies have addressed such issues in children, but again evidence is increasing. Lean mass deposition in childhood is linked with bone deposition,4 which in turn is related to the risk of osteoporosis in later life.37 Increased lean body mass in cystic fibrosis patients is associated with improved respiratory function and outcome,38 while reduced weight-for-height is associated with post-operative growth failure in liver transplant patients,39 and with increased mortality risk in HIV patients.40 Changes in body composition over a 6-week period identified good responders to growth hormone therapy 1 year later.41 In the general population, baseline body fat is positively correlated with subsequent gains in fatness,42 and hence provides information about obesity risk, while in treatment of obese children, greater loss of lean mass during the intervention was associated with greater subsequent increase in body weight.43

The low quantity of evidence associating paediatric body composition with clinical outcomes can be attributed largely to the historical difficulty of measuring body composition in paediatric patients. It is now appropriate to seek such evidence and exploit the capacity of body composition techniques to improve clinical management.

The fact that body composition at baseline often predicts clinical progression does not imply that interventions to alter body composition necessarily improve outcome, hence further studies are required to establish the potential benefits of such interventions.

Treatment requirements in patients

In many scenarios, it is preferable to tailor treatment to body size. Examples include nutritional and fluid intake, drug dosages, radiation dosages and dialysis. Energy intake is perhaps most obviously detrimental if insufficient or excessive, however many pharmacological substances are likewise ineffective at low concentrations but toxic at high concentrations, while dialysis requirements also depend on body fluid content.

Most such treatments are by tradition based on body weight,2 even if the relevant physiological parameter at which the treatment is being aimed is more specific. The higher metabolic rate of infants and young children is well recognised, as is the problem of high body fat in obese children.2 However, solutions to these dilemmas remain simplistic, for example the use of ideal rather than actual body weight for calculating treatment requirements in obese children. Once again, therefore, we suggest body composition measurements could make a significant contribution to clinical management. Body composition measurements can also play an important role in the interpretation of metabolic data, for example assessing leptin resistance relative to body fat level or energy expenditure relative to lean mass.

Many studies have shown that energy requirements of paediatric patients are predicted from body weight with poor accuracy.44 45 This inaccuracy may be attributed to two principal factors: first, weight may provide a poor proxy for lean mass, and second, diseases may alter energy utilisation. Measurement of body composition can address the first but not the second of these problems. In the general population, a change has occurred in the relationship between weight and lean mass in recent decades. Equations generated in the 1960s and 1970s for the prediction of total body water systematically over-estimate this parameter in contemporary children due to the increasing contribution of fat to weight.46 Even if such secular trends are taken into account, the relationship between weight and such outcomes as lean mass and total body water is not tight, and use of weight to predict body water results in inaccurate estimations in many children, particularly in patients with altered body composition due to obesity.47

For estimating energy requirements, one option is to measure resting metabolic rate using indirect calorimetry. While this may prove beneficial in bed-bound patients, in whom resting energy utilisation is a good proxy for total energy requirements, it is less useful in more active patients and requires a metabolic monitor and at least half an hour per patient. An alternative way to improve the prediction of energy utilisation would be to base the prediction on lean mass, the component of body weight that uses energy, possibly in combination with data on physical activity levels. In adults with HIV, for example, measurements of body composition improved the accuracy of resting energy expenditure prediction.48 In children, we have shown that the energy utilisation of artificially-ventilated children is low compared to that of healthy children if expressed relative to body weight, but similar if expressed relative to fat-free mass.29 These studies indicate a potential role of body composition measurements in the nutritional management of paediatric patients; however, appropriate prediction equations relating energy utilisation to lean mass remain to be derived. They would need to be disease-specific to address the differential effect of diseases on metabolism.

For calculation of dialysis dosages, anthropometric predictive equations developed in healthy children overestimate body water in dialysis patients, and new equations derived from patients receiving chronic peritoneal dialysis have been derived.49 However, these equations still have a standard error of 2.2 litres, and in a study of 14 paediatric dialysis patients, measurements of lean mass by DXA allowed estimation of body water with a substantially smaller error of 1 litre.47 Given its widespread availability, the utility of DXA for such a clinical role merits greater attention.

In some cases, the dosage of paediatric drugs is calculated on the basis of surface area rather than weight.2 This approach has been justified on the grounds that surface area has a strong relationship with glomerular filtration rate, and hence the rate of drug metabolism.2 Due to the lack of appropriate technology, surface area is never measured in clinical practice and is predicted instead using a variety of published equations. The validity of these equations in contemporary children is unknown. Children’s equations exist,50 and produce significantly different values from adult equations in children with high or low body weight,51 but are rarely used.

However, lean mass is increasingly considered a more appropriate criterion than surface area on which to base pharmacological dosages. Over 99% of metabolic processes occur in lean mass, making it an appropriate basis for the loading dosage of hydrophilic drugs.52 In adults there is a good correlation between lean mass and systemic clearance, supporting lean mass as the basis for maintenance dosages of liver-eliminated drugs; however, little is known about the specific association between lean mass and hepatic clearance.53 Unlike surface area, direct measurement of lean mass avoids confounding by body fat level. It is likely that lean mass could be measured with sufficient accuracy for this purpose using techniques such as DXA or densitometry, however, this requires empirical confirmation.

Body fatness may also be relevant in some contexts, such as anaesthetics and obesity. Some anaesthetic drugs are fat-soluble and are absorbed into adipose tissue by diffusion. In obese individuals the subsequent release of the drug from adipose tissue can delay recovery.54 Whether incorporation of body composition data in the calculation of anaesthetic doses can improve clinical outcome is thus a further hypothesis meriting investigation. More generally, the effect of obesity on drug distribution volume is predicted to differ between hydrophilic and lipophilic drugs,53 accordingly requiring different strategies. However, this area remains poorly understood.53

In summary, measurements of lean mass have the potential to be used widely in treatment. Techniques such as regional anthropometry and DXA are highly informative and now widely available, while other techniques such as photonic scanning continue to be developed. However, there is a clear need for greater understanding of the relationship between body composition and energy utilisation, drug metabolism and fluid distribution. A disease-specific evidence base is required in order to obtain the full benefits of measuring body composition in routine practice.

CONCLUSIONS

Measurement of body composition can address a variety of requirements in clinical paediatric practice. At present, attention is directed primarily to BMI; however, as discussed above, lean mass and fat mass have different implications for health status and clinical management, and measurements allowing their differentiation are predicted to be beneficial in the management of a wide variety of disease states. Clearly, the effort and cost of body composition measurements can only be justified in paediatric practice on the basis of evidence for their value. We acknowledge that such evidence is still accruing, but hope this article will stimulate further research on this issue. Furthermore, existing studies are greatly improving our understanding of disease processes. In practical terms, access to measurement facilities also remains difficult, and reference data are lacking. However, we believe that even simple approaches can provide valuable information,1 and we will shortly publish UK paediatric reference data for a range of body composition techniques, including DXA, isotope dilution, bio-electrical impedance and anthropometry, which should greatly aid the interpretation of clinical data.