OBJECTIVE. The aim of this study was to evaluate the validity of parentally reported weight and height values for preschool-aged children (3–7 years of age) in Belgium and the accuracy of BMI classifications using those data.
METHODS. The subjects were 297 preschool-aged children (157 boys and 140 girls). A questionnaire with questions about height and weight was distributed to the children and completed by their parents. Nurses of the Centres for Pupils Counselling conducted anthropometric measurements by following standardized procedures. National and international age- and gender-specific BMI cutoff values were used to determine BMI categories.
RESULTS. Mean differences between reported and measured values for weight (−0.57 kg) and BMI (−0.51 kg/m2) were statistically significant, although small. For height, the mean difference (0.14 cm) was not statistically significant. When national and international BMI cutoff values for overweight were used, sensitivity values were 47% and 44%, respectively, and specificity values were 94% and 95%. Among all children requiring nutritional advice on the basis of being overweight or obese, more than one half of the overweight children and >75% of the obese children would be missed with the use of parentally reported weight and height values. Conversely, 70% of the children being classified as underweight with the use of these reported data could be encouraged wrongly to gain weight. The bias of parent-reported BMI values was significantly greater when weight and height were both guessed, rather than being measured at home.
CONCLUSIONS. Our results show the inaccuracy of parentally reported weight and height values in Belgium for classifying preschool-aged children into BMI categories. Therefore, accurate measurements of weight and height should be encouraged in studies in which BMI of children is a variable of interest.
Weight status, particularly overweight and obesity, and weight change are related to the development of many chronic diseases, including type 2 diabetes mellitus, cardiovascular diseases, and some cancers.1–4 Therefore, valid measures of weight and height are of great importance in many surveys. However, because of the logistic difficulties and financial costs involved in measuring directly the weight and height of participants in a survey, these data are often self-reported or proxy-reported. The validity of self-reported weight and height data was studied comprehensively in adults, and a strong correlation between self-reported and measured weight and height was demonstrated.5,6 However, use of BMI based on self-reported weight and height values to define obesity as a categorical variable is considered unreliable in adults.5–7 Similar studies focusing on the validity of parentally reported weight, height, and BMI values for children are scarce, and the conclusions are contradictory.8–11
Because of a growing interest in childhood obesity as a determinant of adult diseases,12 valid measures of recalled childhood weight and height should be of interest to many researchers. Because children are growing quickly, it might be more difficult to report their weight and height accurately, in comparison with adults. Therefore, the aim of the present study was to examine the validity of parentally reported values for current anthropometric factors (weight and height) for children and the factors related to misreporting. We also evaluated the accuracy of BMI categories created from reported values when using national and international BMI cutoff values for underweight, overweight, and obesity. In addition, we examined the validity of parentally reported weight and height measured by the parents at home, in comparison with guessing.
Subjects were residents in the region of Ghent, a medium-sized city in Belgium. A sample of 3- to 7-year-old children was drawn on the basis of a multistage cluster-sampling technique. Three school committees (3 main schools and 4 local residences) in the region of Ghent were selected randomly, and they all agreed to participate. All 29 preschool classes of these schools were selected as final cluster units. All of the children from these 29 selected classes were invited into the study.
Questionnaire and Self-Reported Anthropometric Measures
The questionnaire used in this validation study was used in a large-scale epidemiologic nutrition survey among preschool-aged children in Flanders in 2003.13 In this questionnaire, parents were asked to report the weight and height of their child. In addition, they had to report whether they actually measured their child's weight and height or guessed at these values (these 2 questions were added to the original questionnaire used in 2003). In this questionnaire, information about the child (eg, gender, physical activity level, and consumption of lunch at school), the parents (eg, age, parental education levels, and country of birth), and the household composition was also requested. Children's BMI values (kilograms per meter of height squared) were calculated by using these parentally reported weight and height values. Parents were categorized as having “low parental educational level” when the highest degree of both parents was secondary education. When 1 or both parents smoked, “parental smoking” was considered positive. Children's activity level was categorized, according to the parentally reported leisure time activities, as passive (eg, predominantly reading, watching television, or playing computer games), active (eg, walking or active games), or very active (running, cycling, or swimming).
This validation study of parentally reported weight and height was conducted in collaboration with Centres for Pupils Counselling (CLBs). These CLBs are active in different domains (learning and studying, work attitude, the school career, preventive health care, and social and emotional development) of children and adolescents.14 For preventive health care, certain medical examinations are performed, including weight and height measurements. The staff has been trained to weigh and to measure children in a correct standardized way and recently these CLBs participated in a study to produce new growth charts for Flemish children.15
All of the children participating in this study were examined and measured by a CLB nurse, in a correct standardized way (according to a published protocol15). For these measurements, children were wearing only underwear. Weight was recorded to the nearest 0.1 kg with an electronic weighing scale (Seca 841; Lamoral Medical, Waregem, Belgium), and height was measured to the nearest 0.1 cm with a rigid stadiometer (Seca 220; Lamoral Medical). The stadiometer was checked for accuracy and the scale was calibrated before examination. The actual BMI was calculated from these measured weights and heights.
The directors of the schools and/or the teachers of the classes that participated in the study were given detailed information and instructions about the study. Children of the first and second preschool classes routinely undergo an obligatory medical examination during that school year, conducted by a CLB nurse and doctor. The teachers of these classes were asked to distribute the questionnaire among the parents of the children ∼14 days before this medical examination. An informed-consent form was attached, in which parents were informed about and invited to participate in the study, without being aware that validation of anthropometric measurements was at stake. The completed questionnaires and the signed informed-consent forms were handed in at the school, in a sealed envelope. Nurses of the CLBs were not allowed to open the sealed envelopes, to ensure that they were not influenced by the parentally reported weight and height values.
For children of the third preschool classes, no obligatory medical examination was foreseen during that school year. Therefore, the permission of the parents to measure and to weigh their children in the CLB needed to be obtained. The teachers of these third classes were also asked to hand out the questionnaire and informed consent form to the parents of the children. When the children in these third classes returned the completed questionnaire to the school, a second letter (informed consent) asking the permission of the parents to measure and to weigh their children in the CLB was given. If permission was granted, then a CLB nurse measured the children ∼14 days after completion of the questionnaires. The study was conducted between January and June 2005. The ethics committee of the Ghent University Hospital granted ethical approval for the study.
BMI was calculated with parentally reported and measured height and weight values. Overweight and obesity were identified by using age- and gender-appropriate national and international cutoff points on curves that pass through BMI values of 25 kg/m2 and 30 kg/m2, respectively, at age 18 years.15,16 For identification of underweight, only national BMI cutoff points, passing through 18.5 kg/m2 at age 18 years, were available. The cutoff values are given in the Appendix. The significance of mean differences and prevalences between measured and reported values was assessed by using the Wilcoxon signed-ranks test and McNemar's test. Confidence intervals (CIs) for the differences in prevalences of underweight, overweight, and obesity were obtained according to the Agresti method for comparing dependent proportions.
Spearman correlation coefficients for measured and reported values were given as a measure of associations between measured and reported values. Bland-Altman plots were constructed to study the agreement between parentally reported and measured data on an individual level.17
In identification of underweight, normal weight, overweight, and obesity, misclassification was defined as discordance between BMI categories determined with parentally reported versus measured BMI. The percentage of children classified into the correct and adjacent BMI categories and the percentage grossly misclassified (in opposite BMI categories; reported by parents as being of normal weight but actually being obese) were calculated. The κ statistic was calculated to determine agreement between parentally reported and measured BMI status (eg, obese versus not obese). This κ statistic provides a chance-adjusted measure of reported versus measured agreement.18 Because the κ statistic does not take into account the degree of disagreement (all disagreement is treated equally as total disagreement), it is preferable to use weighted κ values when there are >2 ordered categories. The weighted κ statistic, calculated with a linear set of weights, was used to assess agreement between reported and measured BMI categories.19 The κ values range between −1.00 (perfect disagreement) and 1.00 (perfect agreement), with a value of 0 suggesting no agreement beyond chance alone; κ values of <0.20 are considered to indicate poor agreement, between 0.21 and 0.40 fair agreement, between 0.41 and 0.60 moderate agreement, between 0.61 and 0.80 good agreement, and between 0.81 and 1.00 excellent agreement.19
Sensitivity was defined as the probability that a child categorized in a certain BMI category (eg, overweight) on the basis of CLB data would also be categorized in that BMI category when parentally reported weight and height values were used (true-positive rate). Specificity was defined as the probability that a child assigned as not having a certain BMI status (eg, overweight) with CLB data would also not be assigned to that BMI category with the parentally reported data (true-negative rate). The positive predictive value for obesity was defined as the probability that a child would be declared obese with CLB data if the child was assigned as being obese on the basis of parentally reported data. The negative predictive value for obesity was defined as the probability that a child would not be declared obese with CLB data if the child was not assigned as being obese on the basis of parentally reported data.
Factors associated with bias in parentally reported weight and height were studied by comparing the proportions of subjects for whom the difference between reported and measured BMI values was >1 kg/m2, according to characteristics of children and their parents. SPSS 12.0 for Windows (SPSS Inc, Chicago, IL) was used for data management and all statistical analyses. Unless reported differently, a P value of .05 was used as the threshold for significance. Two-sided significance levels are quoted.
The described sampling procedure yielded a sample of 474 preschool-aged children who were registered officially in the 29 sampled classes. However, 26 children were not eligible for the study because they were on sickness leave on the medical examination day in the CLB. Eventually, only 448 children were considered eligible. Of these, 329 children (73%) actually participated (filled in a questionnaire). In total, 32 children needed to be excluded from the analyses because the parents did not report the weight and/or height of the child in the questionnaire. Subsequently, 297 children were useful for the validity analyses of parentally reported weight, height, and BMI.
Characteristics of the children and their parents are presented in Table 1. Subjects included in the validation study for weight, height, and BMI values for children (n = 297) had a mean age of 5 years (SD: 1 year) and an age range of 3 to 7 years. The 2 genders were almost equally represented in the study. Nearly three fourths of the parents measured their child's weight, and two thirds measured their child's height (Table 1). In total, 89% of the questionnaires were answered by the mother of the child. The mean ± SD weight, height, and BMI were 18.9 ± 3.5 kg, 109.1 ± 7.5 cm, and 15.8 ± 1.6 kg/m2, respectively, when measured by a CLB nurse and 18.4 ± 3.5 kg, 109.2 ± 8.0 cm, and 15.3 ± 1.8 kg/m2 when reported by the parents.
The mean difference in weight between the 2 methods was −0.57 ± 1.47 kg, which demonstrates that the parents, as a group, underestimated slightly the weight of their child, in comparison with the weight measured by the CLB nurse. The mean difference in height between the 2 methods was 0.14 ± 3.58 cm, which demonstrates that the parents, as a group, overestimated slightly the height of their child, in comparison with the height measured by the CLB nurse. The mean difference in BMI between the 2 methods was −0.51 ± 1.60 kg/m2, which demonstrates that the BMI of the child was somewhat underestimated by the parents, in comparison with the BMI calculated from the CLB data. The mean differences between reported and measured values for weight and BMI were, although small, statistically significantly different from 0 (P < .001). For height, the mean difference between the 2 methods was not statistically significant (Table 2).
The numbers of children categorized as underweight, overweight, or obese, by using national and international cutoff values, are presented in Table 2. More children were identified as being overweight or obese when measured weight and height values were used for BMI calculations (overweight: 13.1%–15.2%; obese: 3.0%–4.4%) than when parentally reported weight and height values were used (overweight: 10.1%–12.5%; obese: 2.4%–3.0%). However, these differences in prevalences of overweight and obesity with the use of reported versus measured weight and height values were not statistically significant. The prevalence of underweight when parentally reported weight and height values were used (24.9%) was significantly higher than that when actual weight and height values were used (11.8%).
The Spearman correlation coefficients for correlations between actual and reported weight, height, and BMI values were 0.91, 0.90, and 0.59, respectively, which indicates that the correlations were greater for weights and heights than for BMI values. Bland-Altman plots (Fig 1) showed that parents assessed their child's weight from 3.52 kg below to 2.37 kg above the actual weight.17 As shown graphically in Fig 2, parents assessed their child's height from 7.01 cm below to 7.30 cm above the actual height. BMI values calculated from the parentally reported data were 3.71 kg/m2 below to 2.69 kg/m2 above the actual BMI (Fig 3).
With the national BMI cutoff values, misclassification analysis indicated that 14 children (4.7%) were grossly misclassified, whereas 190 children (64.0%) were classified correctly and 93 (31.3%) were classified in the adjacent category. With international BMI cutoff values, misclassification analysis indicated that 8 children (2.7%) were grossly misclassified, whereas 258 children (86.9%) were classified correctly and 31 (10.4%) were classified in the adjacent category. The weighted κ statistics for BMI categories were 0.34 (95% CI: 0.23–0.45) and 0.39 (95% CI: 0.17–0.61) for national and international BMI cutoff values, respectively, showing fair agreement between BMI classifications based on parent-reported and measured BMI values (Table 3). From Table 3, it could also be concluded that agreement between reported and measured BMI categories was much greater for the group of parents who measured weight and height, in comparison with parents who guessed at both parameters.
The diagnostic tests for classifying underweight, overweight, and obesity from the parentally reported weight and height are shown in Table 4. The sensitivity values for predicting the presence of overweight and obesity status on the basis of parentally reported BMI, compared with measured BMI, were 43.6% and 22.2% for overweight and obesity, respectively, with the international BMI cutoff values for children.16 Specificity values were 95.0% and 98.3% for overweight and obesity with the international cutoff values. When the national BMI cutoff values for estimating underweight, overweight, and obesity were used,15 the sensitivity values for parentally reported BMI values were 62.9%, 46.7%, and 23.1%, respectively. The specificity values for predicting the absence of actual underweight, overweight, and obesity from reported weight and height values were 80.2%, 93.6%, and 97.9%, respectively, when the national cutoff values were used. When national and international cutoff values were used, 53.3% and 56.4%, respectively, of the children who actually were overweight were classified as not being overweight. For children who actually were obese, these figures were 76.9% and 77.8%, respectively. Conversely, 70% of the children who were classified as being underweight when parentally reported data were used were not found to be underweight when CLB data were used. As shown in Table 4, κ values indicated only fair agreement for obesity and underweight but moderate agreement for overweight.
In comparisons of characteristics of children and their parents for underestimation of the BMI by ≥1 kg/m2, the person who completed the questionnaire and the method used to report the weight and height in the questionnaire were associated significantly with bias in reported BMI values. The prevalence of BMI underestimations was significantly greater for questionnaires completed by mothers, in comparison with fathers or other proxies. The prevalence of BMI underestimations in the group of parents who estimated weight and height was significantly greater than the prevalence in the group of parents who measured weight and height (Table 5).
The mean weight and BMI reported by the parents were significantly lower than the actual mean weight and BMI measured by a CLB nurse, whereas the mean height reported by the parents was not significantly greater than the actual height. More than 60% of the parents measured the weight and height of their child at home before recording weight and height values in the questionnaire.
Although correlations between measured and parentally reported weight and height indices were very strong, correlations between BMI values derived from these reported and measured data were much lower. However, it is noteworthy that the relative bias for BMI induced by weight is the same as the bias of weight, whereas the relative bias induced by height is approximately twice the bias of height. The effect on the total bias of the BMI is much smaller when weight and height are biased in the same direction (both overestimated or both underestimated) than when weight and height are biased in opposite directions. In the case of opposite bias, the resulting bias for BMI can become very high even when the biases for weight and height separately are only limited.20 In our study, nearly 50% of the children's weight and height values were biased in opposite directions (data not shown).
The number of children identified as overweight or obese depended on the cutoff values used. More children were categorized as overweight or obese when national cutoff values were used than when international cutoff values were used. More children were identified as being overweight or obese when measured weight and height values were used for BMI calculations than when parentally reported weight and height values were used. When the national cutoff values for underweight were used, significantly more children were classified as being underweight when parentally reported weight and height values were used, compared with measured weight and height values. The use of parentally reported weight and height values for determining the prevalence of underweight in children could suggest incorrectly that one fourth of Belgian preschool-aged children are classified as underweight. This could indicate that parents genuinely do not know what constitutes underweight.
The specificity of predicting the presence of overweight and obesity status on the basis of parentally reported BMI, compared with measured BMI, was high, whereas sensitivity was very low. More than one half of the children who should be targeted for nutritional advice would be missed for intervention purposes with the use of these parentally reported weight and height values. Conversely, 70% of the children classified as underweight with the use of parentally reported data could be encouraged wrongly to gain weight.
An important factor associated with bias in BMI values derived from parentally reported weight and height was whether these parameters were measured at home or were guessed (bias was greater when weight and height were both guessed). The person who completed the questionnaire was also associated with bias in reported BMI (bias was greater when mothers completed the questionnaire). As shown in Table 5, no difference in reporting bias of the parents was found for preschool-aged boys and girls. This could indicate that parents' ability to estimate or to measure their child's weight and height was independent of the gender of the child (at least for children only 3–7 years of age).
Methodologic Issues and Limitations
Some limitations of this study are worth noting. Data were available only for children whose parents completed the questionnaire. Children who were measured by a CLB nurse but whose parents did not complete the questionnaire were excluded from the analyses. It is possible that respondents were more willing, or more able, than nonrespondents to provide accurate assessments of their children's weight and height. Therefore, the errors between parentally reported and measured weight and height in this sample may be underestimates of the true errors, because almost 30% of the parents refused to complete the questionnaire. However, to avoid underestimation of the true errors, the subjects were not aware of the future comparison between reported and measured values. In addition, nonrespondent analyses (results not shown), based on the measured weight and height data of the CLBs and the remaining questions in the questionnaire for the 32 children whose parents did not report weight and height in the questionnaire, indicated no significant differences between the parents who reported weight and height in the questionnaire and those who did not.
In this study, the examination by the CLB nurses, during which weight and height were measured, was performed ∼2 weeks after completion of the questionnaire. Because there might have been up to 2 weeks between the 2 assessments, the true weight and height might have changed. However, large changes, which might explain the present results, are unlikely to have occurred during that time. Questionnaires with parentally reported weight and height values were all distributed and collected before the medical examination, because parents receive a report on the medical examination (including measured weight and height values) of their child shortly after this examination. Parentally reported weight and height values registered after the medical examination could possibly be influenced by this information given to the parents.11
In our study, a comparison was made between parents measuring their child at home before completing the questionnaire and parents guessing their child's weight and height. From these results, it may be assumed justifiably that the biases in parentally reported weight and height observed in our study result not only from guessing but also from inexact measuring or reporting at home. Another hypothesis for the incorrect BMI estimation by the parents could be that their estimation was based on previous weight/height measurements (perhaps performed several months earlier), and their child could have grown during the interval.
However, in our study bias was found to be greater in the group of parents who reported weight and height based on guessing than in the group of parents who measured weight and height at home. Therefore, we conclude that motivating the parents to measure their child's weight and height at home when completing the questionnaire could increase the accuracy of parentally reported weight and height values for children. It is also noteworthy that the results of this study are restricted to Belgian children, because the attention that parents pay to their children's weight and height can be different in different parts of the world and can be influenced by the prevalence of underweight or overweight.
Comparison With Previous Studies
As mentioned before, studies concerning the validity of parentally reported weight and height for preschool-aged children are scarce, which limited comparison of our results with other studies. To our knowledge, this is the first study to compare the accuracy of parentally reported weight, height, and BMI values based on guessing and those based on parental measurements of the child at home.
The strong correlations observed in our study between parentally reported and actual values for weight and height confirm the strong correlations between these 2 indices reported in other studies.11,21 However, the correlations of weight and height found in our study were greater than those found in a study by Davis et al,8 and correlation coefficients for BMI found in our study were more than twice as high as those found by Davis et al.8
Other studies investigating the sensitivity of predicting obesity on the basis of parentally reported weight and height found much higher sensitivity (>70%),9,11 compared with the sensitivity found in our study (<25%). However, according to Sekine et al,11 one of the major problems in their study was the fact that the children had been measured and weighed by the school nurse 2 months earlier and the results had been communicated to the parents.11 Our specificity values were comparable to those found by Sekine et al11 and Garcia-Marcos et al9 Our results regarding the validity of parentally reported weight and height values for classifying children as overweight or obese do compare well with studies investigating the validity of self-reported weight and height values for identifying overweight and obesity in adolescents and adults.7,22
Our results indicate the inaccuracy of parentally reported weight and height values in Belgium in classifying preschool-aged children as being underweight, overweight, or obese. Therefore, accurate measurements of weight and height should be encouraged in studies in which BMI of children is a variable of interest. However, because standardized measurements of weight and height are costly and time-consuming, large surveys of childhood populations are likely to continue to use parentally reported values. In such circumstances, the authors should motivate the parents to measure their children at home. Nevertheless, they must still be aware of the potential bias of these data and should be encouraged to evaluate this bias, at least for a small sample of their population.
We thank the schools and parents who participated in this project and generously volunteered their time and knowledge. We are grateful to the nurses and workers of the CLBs, in particular Lieve Van Neck, Joke Vander Vekens, and Mieke Van Driessen, who made this study possible. We also thank Prof Tim Cole for his expert advice on the interpretation of the results.
- Accepted July 7, 2006.
- Address correspondence to Inge Huybrechts, MS, Department of Public Health, Faculty of Medicine and Health Sciences, Ghent University, UZ-2 Blok A, De Pintelaan 185, B-9000 Ghent, Belgium. E-mail:
The authors have indicated they have no financial relationships relevant to this article to disclose.
- ↵Oh SW, Yoon YS, Shin SA. Effects of excess weight on cancer incidences depending on cancer sites and histologic findings among men: Korea National Health Insurance Corporation Study. J Clin Oncol.2005;23 :4742– 4754
- Poirier P, Giles TD, Bray GA, et al. Obesity and cardiovascular disease: pathophysiology, evaluation, and effect of weight loss: an update of the 1997 American Heart Association Scientific Statement on Obesity and Heart Disease from the Obesity Committee of the Council on Nutrition, Physical Activity, and Metabolism. Circulation.2006;113 :898– 918
- Goodman E, Hinden BR, Khandelwal S. Accuracy of teen and parental reports of obesity and body mass index. Pediatrics.2000;106 :52– 58
- ↵Ministry of Flanders, Education Department. CLB: Centrum voor Leerlingenbegeleiding, 2006. Available at: www.ond.vlaanderen.be/clb/documenten/engels.pdf. Accessed September 13, 2006
- ↵Vrije Universiteit Brussel, Laboratorium voor Antropogenetica. Growth charts, Flanders 2004. Available at: www.vub.ac.be/groeicurven/english.html
- ↵Cole TJ, Bellizzi MC, Flegal KM, et al. Establishing a standard definition for child overweight and obesity worldwide: international survey. BMJ.2000;320 :1240– 1243
- ↵Altman DG. Practical Statistics for Medical Research. London, United Kingdom: Chapman & Hall; 1991
- ↵Molarius A, Kuulasmaa K, Sans S, et al. Quality assessment of weight and height measurements in the WHO MONICA project. Geneva, Switzerland: World Health Organization; 1998. Available at: www.ktl.fi/publications/monica/bmi/bmiqa20.htm. Accessed September 13, 2006
- Copyright © 2006 by the American Academy of Pediatrics