Centile Charts and Assessing Growth

Centile charts show the position of a measured parameter within a statistical distribution. They do not show if that parameter is normal or abnormal. They merely show how it compares with that measurement in other individuals.

They are called centiles and not per centiles. If a parameter such as height is on the 3rd centile, this means that for every 100 children of that age, 3% would be expected to be shorter and 97 taller. On the 97th centile, 97 would be shorter and 3 taller.

Centile charts are very useful for plotting changing parameters such as assessing a child's height or weight over time or head circumference of an infant or fetus. They may also be used in public health. For example, PCTs may show their index for CHD or cancer as a centile to demonstrate how the local prevalence compares with the country as a whole.

Because centile charts are usually used to assess a parameter over time, they are normally presented graphically. The parameter such as height, weight or head circumference is shown on the y axis and the age or gestation on the x-axis. The graph shows a number of lines representing important centiles. These would usually be the 50th centile (average), 25th and 75th centiles as well as the 3rd and/or 10th centile and the 90th and/or 97th centile. These charts are purely to illustrate that presentation. Note that there are different charts for boys and girls.

It is dangerous to take an arbitrary figure such as the 10th or even 3rd centile and to declare that anything below that is abnormal. Not only does this automatically label 3% or even 10% of the population as "abnormal" but some cases of abnormality will be missed. What matters is not so much an isolated value as changes with time.

Suppose that an ultrasound scan at 32 weeks gestation shows that a fetus has a head and abdominal circumference both on the 50th centile. The scan is repeated at 36 weeks and whilst both circumferences have grown, the head circumference has fallen to the 25th centile and the abdominal circumference is a little above the 10th centile. The obstetrician will not wait for either parameter to fall to the "abnormal" level of the 10th or 3rd centile as there is already compelling evidence of intrauterine growth retardation and the fetus is at risk. Delivery should be expedited.

The baby is delivered and weight is on the 10th centile. Over the next few weeks the baby gains weight but how is he doing? If his weight stays on the 10th centile he is maintaining the status quo. His weight may rise but slower than expected so that it starts to fall towards the 3rd centile. This is cause for concern as there is failure to thrive. Babies with intrauterine growth retardation are usually hungry, gain weight fast and their weight rises through the centiles.

A child may be brought to the consultation with a complaint of being overweight but more often this is noticed during the consultation and it may be difficult to convince the parents and child that a problem exists.

BMI charts should be used with great care and not without interpretation until at least 16 and possible 18 years of age. A BMI of 20 for young teenagers is too heavy and before puberty BMI should be substantially less. The UK 90 charts do have a component for the measurement of BMI in children and this has ben validated.¹ The Royal College of Paediatricians and Child Health recommend that the UK 90 BMI reference be used for assessing weight relative to height but that that weight-for-height should not be used. This presumably means rejection of simple figures of BMI as applicable to adults. They also warn that the UK90 BMI reference may not truly represent the prevailing BMI distribution in contemporary children due to a secular trend to increased childhood fatness.²

There are often excuses and denial such as "He is big for his age", "It's just puppy fat", "He'll grow out of it" and "We are all big in our family" (we all overeat together). Fat children are unhappy children³ and they become fat, unhappy and unhealthy teenagers and adults. Both undernutrition and obesity are more common with social deprivation.⁴

Record height and weight but ignore the calculation of BMI. Use centile charts to find those two parameters for that child's age. There are different charts for boys and girls. Fat children are usually tall for their age and so they will be high in the height centiles but they will be even higher in the weight centiles. Thus a child whose height is on the 75th centile and weight over the 97th centile is obese. If a child is overweight but not tall for age, this is a cause for concern.

The approach to weight loss for children is also different from adults. It may not be necessary or desirable that the child actually loses weight. Instead, the gain in weight should be so slow that the child's weight falls through the centiles and approaches the same centile as height.

A child may be brought with a complaint that he or she is too small. Note any medical history that may suggest a cause. Ask about diet and eating habits and weigh and measure the child. Again, look at the centile charts. If the child has height on the 25th centile and weight between the 3rd and 10th centile, this does suggest poor nutrition. Ask the height of the parents. Tall parents can expect to have tall children and short parents have short children.

It may be that there is insufficient cause for concern to justify immediate referral to a paediatrician and a period of observation is required. Make serial recordings of height and weight and plot them on centile charts. If the child continues along the same centile, this is probably just constitutional smallness. He may catch up and show some rise through the centiles. If he falls further through the centiles, even if he is gaining height and weight, this is a cause for concern and a reason to refer.

A child may be brought with complaints that he or she is too fat or too thin. Ask the child how he or she feels about body size. Does she agree with her mother that she is too thin? Again record height and weight and check centile charts. Height and weight do not have to be on exactly the same centiles but so long as they are in the same order of magnitude, this is reassuring. Parents may be trying to overfeed the child but also eating disorders can start young. Ask the child about any teasing at school about weight. Do not be afraid to express your opinion about whether the child looks too fat, too thin or absolutely fine.

Centile charts are supposed to demonstrate the statistical scatter of the normal population. Therefore, a great many individuals must have the relevant parameter measured so that the sample size is large enough to be statistically rigorous. They are supposed to represent the normal population and so those with obvious pathology should be excluded from the data to produce the charts. This may include children with a syndrome such as Down's syndrome or achondroplasia. It may include those with diseases that lead to malnutrition of recurrent infection including coeliac disease or cystic fibrosis or cyanotic congenital heart disease.

There is also a question of whether or not to include those who are simply too fat. Unless they have a disease, they should probably be included although if their numbers are great they will skew the distribution.

If centile charts are based on the normal population and they are being used mostly for children with disease, who are not part of the normal population, are they valid? The answer if probably that they are the best available tool but their limitations must be remembered. There are specific reference charts for Down's syndrome and Turner's syndrome. Charts for very low birthweight infants are proving problematical.⁵

Are they applicable if based on one race but used in another? Variations within races are greater than variations between races and so they are probably applicable but with the usual caveats, including asking about the height of parents.

The World Health Organisation has suggested that centile charts for babies should be based on those who are breast fed as formula fed babies gain weight quicker. Not only may this cause undue concern amongst breast feeding mothers, but their babies' rate of gain in weight may be more desirable.

Many of the height and weight centile charts still in use today date back to 1956. Children are bigger now then they were then. They grow into bigger adults and so the charts, especially for older children, may not be an accurate reflexion of modern childhood. Although overweight children will be big for their age, modern children of correct proportion grow into significantly taller adults than in the 1950s. Obese children were also rather uncommon in the 1950s. They are much more common today and if 20% of children are overweight or obese, this will skew the upper reaches of any charts based on modern children. A child on the 90th centile for height and weight on a modern chart may not be as ideal as the correlation for height and weight suggests. A study from Liverpool suggested that between 1989 and 1998, the number of overweight children increased from 14.7 to 23.6% and the proportion who were obese rose from 5.4 to 9.2%.⁶

It is not only height and weight charts where old references are a problem. If the Gardiner-Pearson chart is used to assess the head circumference of boys at 6 months, the average modern boy is on the 75th centile.⁷

Puberty is a time of rapid growth but it is variable in onset. Hence it is to be expected that children will cross centile lines around this time. Those with an early puberty will be climbing through them whilst those awaiting puberty will fall through them as their peers reach their growth spurt earlier.

The World Health Organisation is eager that updated charts should be used⁸ but the Royal College of Paediatricians and Child Health seems less than enthusiastic.⁹ The College has previously recommended that the Tanner and Whitehouse references should be discarded in favour of the Buckler or UK90.² Their reticence may relate to recent acceptance of the latest UK90 update and validation.¹⁰ If charts are changed too often it is difficult to map secular changes of height, weight and obesity with time.