Many countries are currently experiencing large increases in the prevalence of childhood obesity (1, 2). This is a concern, because childhood obesity frequently tracks into adult life, when co-morbidities pose serious health problems. Until parents recognize that they have overweight children, it is unlikely that they will adopt strategies to limit excessive fat gain. Thus, if effective public health strategies to prevent overweight children from becoming overweight adults are to be put in place, it is a critically important first step for parents to correctly identify children who have high fat mass, early in life.
Unfortunately, recent reports indicate that parents frequently under-rate obesity (as established by high BMI values) in young children and adolescents (3, 4). Overweight teenagers (5) and adults (6, 7) also frequently underestimate their own true weight. Although high BMI values generally indicate elevated weight for height, BMI does not distinguish the contributions of lean mass from those of fat mass to body weight. Some children with high BMI values may not have high fat mass, but instead may show a high muscle mass. None of the previous studies of parental perceptions of child weight has measured body composition. It is, therefore, of interest to determine whether a group of young children with elevated BMI values have higher fat mass than those of normal weight, and whether children whose weight status is underestimated by their parents have higher lean mass or a different fat mass than those whose weight status is correctly assessed by their parents.
The present study had three aims: 1) to compare parental assessments of child body weight status with measurements of BMI; 2) to determine whether children who are incorrectly assessed differ in body composition from those whose parents correctly rate body fatness; and 3) because several studies suggest that obese parents have both more overweight children (8, 9, 10) and possibly a lower recognition of childhood obesity than parents of healthier weight, to find out whether children of obese parents differ from those of non-obese parents in body composition or parentally perceived body weight.
Research Methods and Procedures
Pre-pubertal children ages 3 to 8 years were enrolled for a nutritional study of dental and bone health that was approved by the University of Otago Ethics Committee. They were recruited irrespective of BMI values, and parents were not aware that one focus of the study was perception of body weight in their children relative to measured fatness. Signed informed consent was obtained for every participant, and each child attended the research unit for one visit accompanied by a parent or caregiver. Waist circumferences were measured as described previously (11). The children were carefully weighed (Seca electronic scale; Seca Corp., Birmingham, United Kingdom), and their stature was measured (Harpenden stadiometer; Holtain, Ltd., Crymych, Pembs., United Kingdom) in light clothing without shoes. BMI was calculated as weight (kg) divided by height (m) squared. The height and weight of the accompanying parent were also measured, and values for the other parent were taken from the report of the partner/spouse. Parents with BMI values >30 were considered obese. Information concerning gestational age and birth weight was collected from each child's medical records. Ethnicity was self-identified by parents, according to questions used in the last New Zealand census. To measure the child's body composition, a total body DXA scan was performed and analyzed by one experienced operator using a Lunar DPX-L scanner (GE Medical Systems Lunar, Madison WI); standard scanning procedures were used (11, 12), and scans were analyzed using the Lunar software package, version 4.7. In our laboratory, the coefficients of variation for repeated in vivo scans on 10 adults were 2.6% for fat mass, 2.5% for fat percentage, and 1.1% for bone-free lean mass.
Parents completed a health questionnaire for every child and were asked to classify the weight of their child: score l = underweight, score 2 = normal-weight, score 3 = slightly overweight, score 4 = overweight. The Centers for Disease Control and Prevention's BMI age- and sex-specific ratings selected as equivalent to these four categories were: BMI <5th percentile, “underweight”; BMI ≥5th to <85th percentile, “normal-weight”; BMI ≥85th to <95th percentile, “at-risk-of-overweight”; and BMI ≥95th percentile, “overweight” (13).
Statistical analysis was performed using Stata software (StataCorp., College Station, TX). Raw data are expressed as means and standard deviation. Data were log-transformed, and adjustments for age and sex were used. Regression analysis using log-transformed data and adjusting for age and sex was used to compare groups of interest. Results are presented as ratios of geometric means with 95% confidence intervals (CIs)1.
One hundred one children (51 boys, 50 girls) were enrolled; 89 were white, 10 were New Zealand Maori, and 2 were of Pacific Island ethnicity. None had any known health disorders influencing weight, and none had commenced puberty. Parental BMI values were not available in 2 children who were adopted. Because DXA measurements were not taken in 5 children who were too shy to be scanned, we are reporting results for only the 96 children with complete DXA data. The 5 excluded children were all rated parentally as of healthy weight, despite 2 having BMI values above the 85th percentile.
Table l demonstrates that only 7 parents of the 96 children with DXA measurements rated the weight of their child above normal weight, although 31 children actually had elevated BMI values (at or above the 85th percentile). Parents failed to perceive any increased body weight in 20 (83%) of the 24 children having BMI values lying between the 85th and 95th percentiles. They also under-rated the weight status of 6 (86%) of the 7 children with BMI values above the 95th percentile.
|Parental score of child weight status|
|Measured BMI status*||1 Underweight||2 Normal weight||3 Slightly overweight||4 Overweight||Total (n)|
|Healthy weight (BMI <85)||14||51†||0||0||65|
|At risk of overweight (BMI ≥85 to <95)||2||18||4†||0||24|
|Overweight (BMI ≥95)||0||4||2||1†||7|
Although no parent overestimated child weight status, 40 parents underestimated the true weight status of their child (Table l). Sixteen children were classified as underweight, although no child had a BMI below the 5th percentile; 14 of these children had a BMI value in the healthy range, and 2 had BMI values between the 85th and 95th percentiles. Yet these children were not unduly lean, and of the l4 healthy-weight children who were misclassified as underweight, only two had BMI values below the 25th percentile (observed range for BMI percentiles, 11.7 to 83.1).
The DXA measurements (raw data) displayed in Table 2 confirmed that children with BMI values at or above the 85th percentile truly were heavier, with higher BMI values, greater fat mass, greater fat percentage, and larger waist girths than those with healthy BMI values. The birth weights of these groups also differed, with the higher-BMI children having heavier birth weights. Moreover, the seven children who were perceived as too heavy by their parents tended to be the children in the sample with the highest fat mass (Figure l).
|BMI <85th percentile (n = 65)||BMI ≥85th percentile (n = 31)||p|
|Age (years)||6.20 (1.47)||6.16 (1.48)||NS|
|Height (cm)||116.2 (10.7)||119.6 (11.8)||NS|
|Weight (kg)||21.94 (4.65)||27.31 (8.54)||0.0001|
|BMI (kg/m2)||16.07 (0.93)||18.66 (2.34)||0.0001|
|Lean mass (kg)||16.73 (3.61)||18.69 (4.09)||0.0097|
|Fat mass (kg)||3.38 (1.27)||6.74 (4.87)||0.0001|
|Fat percentage||16.06 (3.66)||23.89 (7.66)||0.0001|
|Waist girth (cm)||55.0 (3.5)||60.9 (7.4)||0.0001|
|Birth weight (kg)||3.26 (0.78)||3.58 (0.55)||0.02|
|Gestational age (weeks)||38.7 (2.98)||39.3 (1.64)||NS|
|Maternal BMI (kg/m2)||27.18 (5.59)||27.16 (5.08)||NS|
|Paternal BMI (kg/m2)||27.45 (4.12)||27.16 (4.23)||NS|
To examine the magnitude of group differences in anthropometry and measures of body composition in data adjusted for age and sex, we then calculated results from the overweight children (BMI values at or above the 85th percentile, n = 31) as ratios of those obtained in children of healthy weight (BMI below the 85th percentile, n = 65). The mean ratios of overweight to healthy-weight children and 95% CIs were: weight (1.23, 95% CI, 1.17 to 1.29), fat mass (1.83, 95% CI, 1.62 to 2.06), fat percentage (1.46, 95% CI, 1.34 to 1.59), lean mass (l.12, 95% CI, 1.07 to 1.17), waist girth (1.11, 95% CI, 1.08 to 1.14), and height (1.03, 95% CI, 1.01 to 1.05).
When the children whose weight status was correctly classified by their parents were compared with those whose weight was incorrectly classified, no differences were seen in the raw data (data not shown). Moreover, children of different body composition were incorrectly graded throughout the age range (Figure 2). However, after adjusting for age and sex, the incorrectly classified group was shown to be lower in BMI, fat mass, and fat percentage than those correctly classified by the parents, although the groups did not differ in height, lean mass, or waist girth (Table 3).
|Ratio of incorrect to correct parental score||95% confidence interval|
|Height (cm)||1.00||0.98 to 1.02|
|Weight (kg)||0.96||0.90 to 1.02|
|BMI (kg/m2)||0.96*||0.92 to 0.99*|
|Fat mass (kg)||0.81*||0.69 to 0.95*|
|Lean mass (kg)||0.99||0.95 to 1.04|
|Fat percentage||0.85*||0.77 to 0.95*|
|Waist girth (cm)||0.97||0.94 to 1.00|
We also compared the anthropometry and body composition of children having no obese parents (n = 54) with body composition values in children having one or more obese parent (n = 40) but found no significant differences. Finally, we investigated the ability of obese compared with non-obese parents to classify child weight but found no significant differences in these parental assessments of child weight status.
DXA provides accurate, simple, and objective measurements of both body fat content and lean mass in young children (14, 15). Our study appears to be the first to use this technique to examine the body composition of a sample of young children with comparative information regarding parental perception of the weight status of their children and measured BMI percentile ratings. Using DXA, we were able to establish that the group of children having BMI values above the 85th percentile were, indeed, significantly heavier than children with healthy BMI values, predominantly because they had higher body fat mass than the group of children having BMI values below the 85th percentile. Observed increases in height (3%), lean mass (12%), and waist girth (11%) in the group having BMI values above the 85th percentile were of much smaller magnitude than increases in fat mass (83%) or fat percentage (46%), showing that the elevated BMI values were associated predominantly with accumulation of excessive body fat, rather than high muscularity. The magnitude of the group difference in waist girth that we observed is small, perhaps explaining why parents frequently fail to notice high adiposity in children of this age. We also noted that the seven parents who graded their children as having above-average body weight had children with fat mass at the upper end of the distribution. This observation suggests that parents detect over-fatness when it is more extreme considerably more easily than when it is of smaller magnitude. Others are in agreement with this view and also suggest that overweight is easier to detect in older children than in young ones (16).
This study confirms our earlier impressions (17) that contemporary New Zealand parents assess high adiposity in their young children extremely poorly. Weight status was rated as above normal by parents in only 7 (23%) of the 31 children having BMI values above the 85th percentile. These findings support the view that parents today do not distinguish high fatness in children easily. The present findings are consistent with recent results obtained in other countries by researchers who have studied the ability of parents to assess childhood obesity (7, 18, 19, 20, 21, 22). Indeed, even medical professionals classify high fatness poorly in young children (23).
We were concerned to note that although no parent overestimated the weight status of their child, there was a strong tendency for parents not only to underestimate high body weight status, but also to classify children with healthy BMI values as underweight. When we examined the body composition of children whose weight status was underestimated by their parents, we found that those children whose weight had been underestimated had a lower BMI and a lower fat mass than the children whose parents correctly assessed weight status. Nevertheless, their body fat content was healthy, with children incorrectly rated as underweight having measured fat percentages of 10 to 23%. Importantly, these groups did not differ in lean mass or waist circumference. Maynard et al. (16) similarly noted that children with lower BMI-for-age z-scores had greater odds of underclassification of overweight by their parents than children with high BMI z-scores. These results are consistent with the view that contemporary parents do not perceive children as too heavy until fat mass is excessively high. This is worrying, as the fat mass of children with a higher fat content usually continues to track higher as the children grow older (17, 24). Perhaps, because numerous contemporary children are overweight, many parents consider leaner children, who actually have quite healthy weight, as “too thin.” As obesity becomes more common, it may become even harder for parents to recognize children with BMI values “at-risk-of-overweight” as having too much fat. Excess weight will not be noticed because being overweight will be the norm (7).
We did not find any evidence to show that the children having one or more obese parent were heavier or fatter than those having non-obese parents, although others have reported this (25, 26). In agreement with Jeffery et al. (7), we found that obese and non-obese parents rated the weight of their children similarly. In our study, both groups of parents were equally poor in assessing child weight status accurately.
Because participating parents were unaware that an important objective of the present study was to examine perceptions of child weight status, and no emphasis was placed by the researchers on child weight when administering the questionnaire, we think that no bias was present in the parental classifications of child weight status that we collected. Parental denial of high child weight, therefore, was not an issue. However, it was interesting to note that children with BMI values above the 85th percentile had heavier birth weights than children with healthy BMI values. This finding supports the view that there is a positive relationship between high birth weight and overweight in growing children (10, 27).
Strengths of the present study include objective DXA measurements of both fat mass and lean mass, use of standard BMI cut-off points (13) to define elevated body weight for age and sex during growth, and collection of information concerning child birth weight and parental BMI. We collected no dietary information on energy intake and, therefore, do not know to what extent high intakes of energy, or an imbalance of energy intake to energy expenditure, contributed to high body fat. Although our study sample was not large, it allowed comprehensive assessment of child adiposity, and the proportions of overweight and obesity observed are similar to levels seen in a recent representative sample of over 4000 New Zealand children ages 5 to l4 years (28). We, therefore, consider our findings applicable to the wider New Zealand population.
Prevention of obesity is considered to be a more effective public health strategy than reducing excess fat mass. A recent consensus statement on childhood obesity concluded that interventions to prevent excessive weight gain were justified in children with BMI values above the 85th percentile (29). Modifications of diet and decreased inactivity were recommended, with more aggressive treatment being offered to children with BMI values above the 95th percentile. The view was expressed that lifestyle interventions such as these to prevent or reverse obesity in its early stages may be more successful than those instigated later in adolescence or adulthood (29).
It is evident that major changes in parents’ perceptions of healthy childhood weight will be required if they are to put in place effective strategies to limit excessive weight gain in their growing children earlier in life. Until they recognize that their child is heavier than he or she should be, parents are unlikely to take steps to limit excessive weight gain by adopting strategies to improve the balance between energy intake and energy expenditure in their offspring. In the present study, children with BMI values above the 85th percentile were shown to have a higher fat mass and a greater fat percentage than those of normal weight, confirming the value of this cut-off point in identifying young children “at-risk-of-overweight.” We, therefore, suggest that providing parents with information regarding the BMI weight status (13) of their children at a young age, so that they can appreciate whether their child ranks as underweight, normal-weight, at-risk-of-overweight, or overweight, would considerably improve the parental awareness of true child weight status. Until parents recognize that they have overweight children, it is unlikely that they will adopt appropriate strategies to limit excessive fat gain. We suggest that this information could be provided inexpensively and appropriately at the time each child starts school, a time of life when parents are likely to be responsible for providing healthy food and encouraging good exercise habits in their children (17).