Tool to predict babies likely to become obese

“Parents can learn whether their newborn is at risk of becoming fat using a simple online calculator,” The Daily Telegraph has reported.

The story is based on a study that looked at whether a baby’s chances of becoming obese in childhood can be accurately modelled. Researchers hope that identifying ‘high-risk’ babies will prompt parents and health professionals to take action to reduce the chances of their child being obese later in life.

There are several recognised risk factors for childhood obesity, including:

parental body mass index (BMI)
infant birth weight
the rate at which a mother puts on weight in pregnancy
maternal smoking habits – mothers who smoke during pregnancy are more likely to give birth to children who become obese
the size of the household – children growing up in one-parent families are more likely to become obese
the mother’s professional status – children born to unskilled or semi-skilled women are more likely to become obese than children born to skilled or professional women

The researchers found that, when combined, these factors could be used at birth to predict the future childhood risk of obesity, with the parents’ BMI being the most important risk factor.

They also tested whether genetic factors associated with obesity could be used to predict the risk, but found these made little difference to childhood obesity risk.

It is important to stress that this study does seem to confirm that while there are obesity risk factors, there is no such thing as a child who is ‘destined to be obese’.

Promoting healthy eating habits and regular physical activity at an early age should help offset the risk of children becoming obese later in life.

Where did the story come from?

The study was carried out by researchers from a number of institutions in Europe and North America including Imperial College London. It was funded by several organisations including the Academy of Finland, the European Commission, the Medical Research Council and the US National Institutes of Health.

The study was published in the peer-reviewed open-access journal Public Library of Science (PLoS) ONE.

While the main body and methods of the research were reported reasonably accurately in the media, readers could have come away with the mistaken impression that the researchers devised a foolproof test for predicting childhood obesity. To be fair to the researchers, they make it very clear that this is not the case.

The BBC helpfully included comments from an independent childhood obesity specialist, Professor Paul Gately, who highlights that using targeted methods like this could help save the NHS money.

What kind of research was this?

The researchers point out that childhood and adolescent overweight and obesity have become major public health problems and are leading causes of early type 2 diabetes and cardiovascular disease.

Since studies have shown a strong correlation between early infant weight and childhood body weight, prevention of obesity should start as soon as possible after birth, they argue.

Assessing the risk for future overweight or obesity in newborns means that those at risk can be targeted for preventative treatment during the first few months of life.

The researchers say that several factors have been linked to later obesity, including genetic variants but, as yet, no study has looked at whether these factors might be combined to predict which newborns are at risk of childhood obesity.

Using these factors, they aimed to build and test a “predictive algorithm” for identifying newborns at risk of childhood obesity.

To test the accuracy of certain risk factors in predicting childhood obesity, the researchers used data from a large Finnish birth cohort.

They repeated tests of risk factors in two further cohort studies undertaken in Italy and the US.

What did the research involve?

The researchers used data from 4,032 participants in a Finnish birth cohort set up in 1986, who have been followed since the 12th week of their mothers’ pregnancies.

The study has systematically recorded several well-known risk factors for childhood obesity.

For this study, the researchers used data from these 4,032 participants who had their height and weight recorded at 7 and 16 years of age.

Drawing on previous research they selected factors associated with childhood obesity.

These were:

gender – young girls are more likely to develop childhood obesity than boys
pre-pregnancy parental BMI
parental professional status
single parenthood
maternal weight gain during pregnancy
smoking during pregnancy
number of household members
the baby’s birthweight

Using genetic profiling, they also selected 44 common genetic variants that have been associated with being overweight or obese.

They analysed whether, in this cohort, childhood obesity could be predicted using:

traditional risk factors alone, or
genetic profiling alone, or
risk factors combined with genetic profiling

They looked separately at whether these three factors could be used to predict:

childhood obesity (obesity at 7 years of age)
childhood overweight or obesity (overweight or obesity at 7 years of age)
adolescent obesity (obesity at 16 years of age)
adolescent overweight or obesity (overweight or obesity at 16 years of age)
severe sub-types of childhood obesity that persist into adolescence (obesity at 7 and 16 years of age)
childhood overweight or obesity persistent into adolescence (overweight or obesity at 7 and 16 years of age)

Overweight and obesity were defined by internationally agreed standards (a BMI of between 25 and 29 was considered to be overweight and a BMI of 30 or above was considered to be obese).

They then tested the model for childhood obesity they developed in two further studies that included children from different countries and cultural backgrounds. They did this to see if their prediction model could accurately predict overweight and obesity in children from other backgrounds.

The first of these was a study of obesity among 1,503 children aged 4-12 from Italy, published in 1993, who had similar rates of obesity to the children in the Finnish cohort.

The study was retrospective, which meant researchers had to go back and collect past information from around the time of the birth of the children about risk factors for obesity.

The second study was done on a more recent sample of 1,032 US children aged 7 who had higher rates of obesity than those seen in the Finnish study.

The researchers say that for these two studies they only tested whether their model worked to predict childhood obesity (the first of the classifications above).

This was because the model for predicting childhood overweight or obesity (the second category) was not considered accurate enough to be clinically useful. Also, neither of these two additional studies provided information on older cohorts that would allow any meaningful insight into models of adolescent obesity.

Additionally, information regarding genetic variants was not available for these two studies.

The researchers used data from these two studies to build the new obesity prediction models, and tested these additional prediction equations. They also combined the three equations predicting childhood obesity and used this to develop an electronic risk calculator. This was linked to by some media sites.

What were the basic results?

The researchers say that parental BMI, birth weight, maternal weight gain in pregnancy, number of household members, professional status of mother and smoking habits during pregnancy were all independent risk factors for obesity in all or most of the six outcomes.

When they looked at the combined accuracy of these risk factors they found that the cumulative accuracy of traditional risk factors predicting childhood obesity, adolescent obesity and childhood obesity persistent into adolescence was reasonably accurate.

In particular:

parental BMI was the most important factor in determining childhood obesity
adding the genetic score made little difference to the prediction

When testing the model on the Italian and American data sets, they found that the equation for childhood obesity remained “acceptably accurate”.

The two additional equations for childhood obesity, newly drawn from the Italian and the US data sets, showed good accuracy at predicting childhood obesity in those groups.

Researchers converted the three equations for childhood obesity into simple Excel risk calculators for potential clinical use.

How did the researchers interpret the results?

The researchers say their study provides the first example of a “handy tool” for predicting childhood obesity in newborns, by means of easily recorded information.

It also shows that currently known genetic variants associated with an increased risk of obesity have very little usefulness for such predictions.

Conclusion

This is an interesting study, but it is premature to conclude that the researchers’ model should be used to make instant calculations about a newborn’s risk of future obesity.

The results of this study are more mixed and less conclusive than the media have implied. For example, the researchers concede that when the US study was taken alone, the model proved less accurate at predicting risk.

It’s also worth noting that, in the Finnish study, the formula could not be used to predict which newborns would go on to be overweight during childhood. Also, that predictions of adolescent obesity could not be validated in the further two studies due to differences in the data sets.

The Italian study was retrospective, which meant researchers had to go back and collect information from around the time of the birth of the children in the 1980s. This could have affected the study’s reliability.

The researchers selected certain risk factors for obesity, but it is possible that other important risk factors may have been omitted, such as diet and levels of physical activity.

Developing a predictive tool for obesity, which enables health professionals to focus on those most at risk at an early stage in life, is a valid area of research.

It is possible that new parents could be prompted by such predictions to follow the advice given by health professionals on how best to ensure their baby is a healthy weight. Research has found that, in many cases, parents who set the right example for their children from an early age, in terms of diet and exercise, are less likely to have children who become obese.

However, as the researchers point out, this kind of predictive tool needs to satisfy several requirements before it can be used routinely, especially if it were to underpin a national obesity prevention strategy.

Currently, there is little evidence of any effective preventive strategy involving babies. Trials that prove the effectiveness of preventive strategies among babies and their families are needed before such a tool could be usefully used by doctors.

It is tempting for prospective and new parents to use the online calculator but it is important to bear in mind that it offers no explanation as to how the statistical risk it calculates should be interpreted and no advice on how to prevent obesity if the risk appears to be high. At this stage, the calculator should be approached with caution.