In relation to the significant influence of parenting practices on children’s eating behaviours, which has been well established( Reference Wang, Beydoun and Li 1 , Reference Reicks, Banna and Cluskey 2 ), several studies suggest parental perception of their child’s overweight status is a protective factor against future weight gain during childhood( Reference Duncan 3 ). When parents correctly assessed their children’s weight status, they were more likely to perceive their overweight children as being less active and to overeat more often than other children( Reference Mathieu, Drapeau and Tremblay 4 ). In addition, overweight children with correct weight perceptions increased attempts to lose weight( Reference Yost, Krainovich-Miller and Budin 5 , Reference Edwards, Pettingell and Borowsky 6 ).
However, worldwide, about half of parents underestimate their children’s overweight status as reported by a recent systematic review with mean ages between 2 and 16 years( Reference Lundahl, Kidwell and Nelson 7 ). The National Health and Nutrition Examination Survey in the USA showed that parental underestimation of pre-school children’s overweight status has increased by 30 % in the last 20 years( Reference Duncan, Hansen and Wang 8 ). Due to their cross-sectional designs, these studies have not shown a causal relationship between overweight perception and later weight loss. In contrast, a few longitudinal studies have found that children who are labelled fat by parents, family or peers have a higher risk of subsequent weight gain( Reference Robinson and Sutin 9 ) or obesity( Reference Sutin and Terracciano 10 ), but the underlying reasons are unknown. To our knowledge, children’s age( Reference Lundahl, Kidwell and Nelson 7 ), BMI( Reference Mathieu, Drapeau and Tremblay 4 , Reference Lundahl, Kidwell and Nelson 7 , Reference Yao and Hillemeier 11 , Reference Remmers, van Grieken and Renders 12 ), parental education level( Reference Mathieu, Drapeau and Tremblay 4 , Reference Remmers, van Grieken and Renders 12 ), maternal weight status( Reference Lundahl, Kidwell and Nelson 7 ), maternal concern about their child’s weight gain( Reference Hernandez, Reesor and Machuca 13 ) and cultural influence( Reference Remmers, van Grieken and Renders 12 ) significantly contribute to parental perception of child’s weight status, although the effect of child sex( Reference Lundahl, Kidwell and Nelson 7 , Reference Yao and Hillemeier 11 ) is not significant in all of these studies.
Interestingly, parents in China had higher misperception of their children’s weight status compared with parents in some other countries( Reference Lundahl, Kidwell and Nelson 7 ). It is reported that 56·0 % of parents( Reference Wen and Hui 14 ) and 65·4 % of children( Reference Zhao, Zhang and Zhou 15 ) had inaccurate assessment of children’s weight status. Although these perceptions seem to be associated with children’s adverse health behaviours( Reference Wen and Hui 14 ), the cross-sectional study design did not allow testing how the perception may have affected weight-related parenting practices (e.g. parents might encourage their children to change their physical activity (PA) and eating behaviours to change their weight), while childhood obesity rates have been increasing over the last three decades in China( Reference Wang, Mi and Shan 16 , Reference Sun, Ma and Han 17 ).
Using nationwide longitudinal survey data, the present study examined: (i) the prevalence of maternal and child perceptions of the child’s weight status; (ii) the differences in perceptions by child sex, age, measured weight status, maternal weight status, parental education level and residence area; and (iii) their associations with subsequent weight-related parenting practices, child health behaviours and weight gain during follow-up.
Methods and materials
Study design and study sample
The China Health and Nutrition Survey (CHNS), which began in 1989, is conducted by the University of North Carolina at Chapel Hill and the Chinese Center for Disease Prevention and Control. It is a long-term, prospective, open-cohort study that applies a multistage random clustering sampling design and obtained information from about 4400 households with a total of 26 000 individuals in nine Chinese provinces at its inception. Later, more provinces and municipal cities were added. Thus far, CHNS has collected data in fifteen provinces and municipal cities. CHNS was not designed to be nationally representative, but it covers areas that account for approximately 45 % of China’s total population varying substantially in geography, economic development, public resources and health indicators to examine various socio-economic factors affecting people’s health and nutritional status.
In CHNS, various measures on individuals (e.g. health, diet and PA), their family (e.g. residence, structure and income) and community have been collected. Data on individuals were collected by in-home visit interviews and survey questionnaires. Selected adults and children from the same household were enrolled. During in-home visit interviews, the data collectors asked the primary care provider’s help if needed for the child response to survey questionnaires for children younger than 12 years old. Detailed information about the CHNS is provided elsewhere( 18 ). Informed consent was obtained from all study participants for data collection. The present study was approved by the State University of New York at Buffalo Institutional Review Board.
The present study included only those first-born children aged 6–18 years (n2310) in families surveyed in 2004, 2006, 2009 and 2011, because data on maternal perception of child’s weight status were collected from the mothers only for their first child in CHNS.
Children without information for age, sex, at least two weight and height measurements, and maternal and child perceptions of the child’s weight status were excluded from our analysis, resulting in a final analytic sample of 816 children. For stratified analysis, two age groups were created: 6–11 and 12–18 years old.
Assessment and measures
Child weight status
Child weight and height were measured to the nearest 0·1 kg and 0·1 cm in light, indoor clothing without shoes using a weight scale (Floor Weight Scale No. 877; SECA, UK) and measuring tape (Mechanical Measuring Tape No. 206; SECA) by trained and certified staff during a detailed physical examination. All staff followed standard protocols of the National Center for Health Statistics in the USA and had to take inter-observer reliability tests as part of their training( 18 ). Child BMI (kg/m2) was calculated and transformed to a BMI Z-score according to the sex- and age-specific WHO reference growth charts( 19 ). Change in BMI Z-score was used to show weight gain during follow-up. Child weight status was classified as underweight, normal weight or overweight and obese, based on the International Obesity Task Force BMI cut-offs( Reference Cole, Bellizzi and Flegal 20 ).
Maternal and child perceptions of the child’s weight status
Mothers were asked to describe their first child’s weight status, and children were asked to describe themselves as normal weight, underweight or overweight. Maternal and child perceptions of the child’s weight status were categorized as correct/under-/over-perception by comparing their perceptions with the child’s actual weight status based on the measured and calculated BMI and the International Obesity Task Force BMI cut-offs( Reference Cole, Bellizzi and Flegal 20 ).
Weight-related parenting practices
Mothers were asked whether they encouraged their child to engage in more PA and to lose or gain weight through dieting, which meant modifying their usual eating habits to change weight.
Child health behaviours
Children’s responses on their PA level, diet, attempts to diet and screen time in the CHNS child survey were used in our analysis. Children’s self-reports were classified by whether they were on a diet and whether their PA was too little, just the right amount or too much. PA was defined as sports or activities that increased the heart rate or made them sweat.
Daily recreational screen time was defined as the average number of hours spent watching television, videos, DVD and (online) movies, playing video/computer games, surfing the Internet, and chatting online before or after school and over the weekend.
Children’s food consumption data were collected using a 24 h recall and same-day interview conducted by trained nutritionists over three consecutive days. Total energy intake (kcal/d) and fat consumption (g/d) were calculated from the food consumption data( 18 ). Percentage of energy from fat was calculated as follows: % energy from fat= [fat (g)×9 kcal/g]/[total energy (kcal)].
Parental and household characteristics
To consider the associations between family and household socio-economic status and maternal and child perceptions of child’s weight status, we used parental weight status (overweight or obese, not overweight or obese), highest parental education (less than middle school, middle and high school, advanced degree(s)), household income per capita (tertiles) and residence (rural, urban). Parental weight status was classified using the Chinese BMI cut-points for overweight (24·0≤BMI<28·0 kg/m2) and obese (BMI≥28·0 kg/m2) with measured anthropometrics.
Statistical analysis
First, we described the distributions of maternal and child perceptions of the child’s weight status and examined differences in both perceptions and the mother–child matched perceptions with the child’s actual weight status by demographic characteristics of the child, their families and households using the χ 2 test. Second, we tested agreement between maternal and child perceptions of the child’s weight status using weighted kappa (κ w) and the total percentage of agreement in the overall sample and within subgroups.
To examine the longitudinal associations between maternal/child perceptions of the child’s weight status and subsequent maternal behaviours and child’s health, we used baseline perceptions from pooled data of the CHNS 2004, 2006 and 2009 and measurements of weight-related parenting practices, child’s health behaviours and anthropometrics at least 2 years post-baseline. The interaction term between maternal/child weight perceptions and BMI Z-score at baseline was included in the mixed models to test whether the effect of perceived weight status on weight gain was moderated by the child’s actual weight at baseline.
Mixed models were used, considering the CHNS sampling methods and hierarchical data structure. Our models included fixed and random effects (allowing individuals to have their own intercepts and slopes of their age) and adjusted for sex, age, BMI Z-score at baseline, survey wave, the interaction between maternal/child perceptions and BMI Z-score at baseline, urban/rural residence and income level. In addition, parental education level and maternal weight status were included in the analyses to adjust for the influence of parental characteristics on parental perceptions.
Analyses were conducted using the statistical software packages SAS version 9.3 and Stata version 14. Effect size was presented as β coefficient with se or OR with 95 % CI, and significance was set as P<0·05 for all analyses.
Results
At baseline, half of the 816 children aged 6–18 years were boys (51·5 %) and 34·3 % lived in urban areas. Less than 10 % of their parents had advanced degree(s) (Table 1).
Table 1 Characteristics of Chinese children aged 6–18 years and their parents and families from baseline data collected in the China Health and Nutrition Survey (CHNS) of 2004, 2006 and 2009 (n 816)Footnote *
* Although CHNS started in 1989, in the present study ‘baseline data’ refers to the first observation of subjects from pooled data of CHNS rounds of 2004, 2006 and 2009. Subjects having (i) both maternal and child perceptions of child weight status and (ii) at least 2 years’ time gap between perceptions and the subsequent BMI Z-score were included in data analysis. The sample size was 816 unless indicated otherwise.
† BMI Z-score was calculated based on the sex- and age-specific WHO reference growth charts( 19 ).
‡ Child weight status was classified by the extended International Obesity Task Force BMI cut-offs( Reference Cole, Bellizzi and Flegal 20 ).
§ Highest parental education was defined by the higher achieved degree of either parent.
║ Parental weight status was classified using the Chinese BMI cut-off points for overweight (24≤BMI<28·0 kg/m2) and obese (≥28·0 kg/m2).
Maternal and child perceptions of child’s weight status at baseline
Among 816 children, 13·6 % of children were overweight (Table 1). Table 2 describes the distributions of maternal and child perceptions of child weight status in all and stratified groups by demographics. Overall about 10 % of mothers and children perceived the child as being overweight (mother: 9·6 %, child: 10·9 %; Table 2). More than half of mothers (63·8 %) correctly perceived their child’s weight status, while 22·2 % of mothers under-perceived and 14·0 % of mothers over-perceived compared with the child’s actual weight status (see online supplementary material, Supplemental Table 1). Compared with normal- and underweight children, overweight children were less likely to be perceived correctly by mothers (40·5 %) and by the children themselves (37·8 % of 111; Table 2).
Table 2 Maternal and child perceptions of the child’s weight status and their agreement from baseline data collected in the China Health and Nutrition Survey (CHNS) of 2004, 2006 and 2009 (n 816)Footnote *
κ w, weighted kappa.
* Although CHNS started in 1989, in the present study ‘baseline data’ refers to the first observation of subjects from pooled data of CHNS rounds of 2004, 2006 and 2009. The final sample size in the analysis was 816 unless indicated otherwise.
† Agreement between maternal and child perceptions was described by sociodemographic characteristics. For example, overweight children had 80·2 % agreement between maternal and child perceptions of the child’s weight status, and their κ w=0·65 (P<0·001). All κ w statistics were statistically significant (P<0·001).
‡ Maternal weight status was classified using the Chinese BMI cut-points for overweight (24≤BMI<28·0 kg/m2) and obese (BMI≥28·0 kg/m2).
§ Highest parental education was defined as the higher achieved education degree of either parent.
Overweight mothers and their children and urban children tended to perceive the child as being overweight more than their counterparts, which was not surprising because children with overweight mothers and living in urban areas had higher rates of overweight (Table 2). Mothers and children tended to under-perceive younger children’s weight status compared with the weight status of older children (P<0·05; online supplementary material, Supplemental Table 1). However, mothers’ and children’s weight perceptions of the child did not differ by child sex.
Overall mothers’ and children’s weight perceptions of the child were fairly consistent with each other (κ w=0·56, P<0·001), especially for boys, overweight children, urban children and children with overweight mothers compared with their counterparts (Table 2). Comparing the mother–child matched perceptions with the child’s actual weight status, 52·8 % of children and their mothers correctly assessed the child’s weight status, but 24·3 % of children and their mothers had the same wrong perception (both under- and over-perceptions) of child’s weight status (online supplementary material, Supplemental Fig. 1). This misperception by both children and their mothers was greater for younger children (26·6 % of 552) than older children (19·3 % of 264, P<0·001).
Associations between maternal and child perceptions of child’s weight status and follow-up weight-related parenting practices, child health-related behaviours and child weight gain
Weight-related parenting practices, child health behaviours and child weight gain were significantly affected by prior maternal perception of the child’s weight status in longitudinal analysis after adjusting for individual, parental and household demographic characteristics. Mothers who considered their children overweight were more likely to encourage their children to engage in more PA (OR=1·8; 95 % CI 1·0, 3·3) and to change his/her weight through dieting (OR=4·3; 95 % CI 2·3, 7·8) than mothers who viewed their children as normal weight (Table 3).
Table 3 Longitudinal associations of baseline maternal and child perceptions of the child’s weight status with subsequent (during follow-up) weight-related parenting practices and child’s health-related behaviours and weight gain: the China Health and Nutrition Survey (CHNS) 2004–2011 (n 816)Footnote *
PA, physical activity.
* In total 816 children aged 6–18 years old and having (i) both maternal and child perceptions of child weight status and (ii) at least 2 years’ later follow-up BMI Z-score were included in data analysis from CHNS 2004–2011 (number of data points=1089). Separate mixed models were fit for each perception (maternal v. child’s) and for each outcome, respectively. The models adjusted for child sex, age, BMI Z-score at baseline, survey wave, interaction between maternal/child perceptions and BMI Z-score at baseline, urban/rural residence, family income level, parental education (only included in the analysis of maternal perception effect) and maternal weight status (only included in the analysis of maternal perception effect). Significant results (P<0·05) are shown in bold font.
† All associations between maternal and child’s perception of the child’s weight status and BMI Z-score were statistically significant (P≤0·001).
‡ Children’s self-reported data were used in the analysis.
Children who were perceived as overweight either correctly or incorrectly by mothers believed they had insufficient PA (OR=2·8; 95 % CI 1·6, 4·7) compared with children with normal weight perception by mothers (Table 3); however, there were no significant differences in children’s attempts to diet and their levels of daily energy intake, percentage of energy from fat in total energy intake and screen time by maternal perception of the child’s weight status (data not shown).
Also, children who were perceived as overweight by their mothers had a higher BMI Z-score increase (β=0·95 (se 0·12), P<0·001) and a higher likelihood of becoming overweight at follow-up (OR=21·7; 95 % CI 8·9, 52·7) than those perceived as normal weight (Table 3). There were no significant interactions between child BMI Z-score and maternal weight perception of the child for every outcome variable. These results suggest that the association between identifying a child’s weight status and the child’s weight gain, as well as child health behaviours and weight-related parenting practices, are not dependent on the actual BMI of the child at baseline.
Children previously perceived as underweight by their mothers also were more likely to be encouraged to change their weight through dieting (OR=2·5; 95 % CI 1·6, 4·1) than those perceived as normal weight. These children also had a lower BMI Z-score increase (β=−0·54 (se 0·12), P<0·001) and a lower likelihood of becoming overweight (OR=0·2; 95 % CI 0·1, 0·7) at follow-up. Children’s own weight perceptions had the same associations with follow-up weight-related parenting practices, child health behaviours and weight gain as mothers’ perceptions of their child’s weight status (Table 3).
Discussion
Using nationwide longitudinal survey data collected during 2004–2011 in China, we examined maternal and child perceptions of the child’s weight status, mother–child agreement, and how these perceptions affected follow-up weight-related parenting practices, the child’s health-related behaviours and weight gain. Only 10 % of Chinese mothers and children perceived the child as being overweight or obese, while 59·5 % of mothers perceived their overweight child as being not overweight, compared with 50·7 % of mothers in other countries as reported by a recent systematic review( Reference Lundahl, Kidwell and Nelson 7 ). In particular, overweight children and younger children were less likely to be perceived correctly regarding their weight status by their mothers and themselves, compared with their counterparts.
Maternal perception of her child’s overweight status appears to be a protective factor for weight-related parenting practices. Mothers would encourage their child to pursue weight management through PA and diet if they perceived their child as overweight. However, such maternal perception and encouragement were not linked to the child’s healthy eating and self-rated PA level, despite being associated with increased childhood weight gain during the follow-up. Other factors of the household environment may impede the effect of maternal perception and encouragement on preventing childhood obesity.
Maternal perception of the child’s weight status has been considered a key factor in obesity prevention and management( Reference Duncan 3 ). First, parents are not good at accurately classifying their child’s weight status( Reference Towns and D’Auria 21 ). In a worldwide meta-analysis( Reference Lundahl, Kidwell and Nelson 7 ), about 50·7 % parents underestimated their children’s overweight status. Second, when parents underestimate their children’s weight, they are more likely to report that their overweight children ‘overeat less often’ and are ‘moderately active’ compared with parents with correct classifications( Reference Mathieu, Drapeau and Tremblay 4 ). Third, parents do not allow their children to participate in an obesity prevention intervention if they believe their overweight/obese child does not have a weight problem( Reference Taveras, Hohman and Price 22 ). These findings from cross-sectional studies support the popular belief that being perceived overweight by parents is a protective factor against children’s future weight gain through better weight-related parenting practices( Reference Duncan 3 ).
We found moderate agreement between maternal and child perceptions of the child’s weight status (total agreement=77·1 %; κ w=0·56, P<0·001), which may be partially due to parental impact on children’s self-weight perception. As expected, we also found weight-related parenting practices were significantly different by maternal perception of the child’s weight status. Mothers who perceived their child as overweight were more likely to encourage the child to increase PA and to change his/her weight through dieting. This pattern is similar to previous findings: correct weight perception is helpful for better control over lifestyles such as attempts to do more PA and eat less( Reference Yost, Krainovich-Miller and Budin 5 , Reference Edwards, Pettingell and Borowsky 6 ).
Yet, as indicated in the present study, maternal concerns about their children’s weight status were not sufficient to change child health behaviours or to reduce the risk of overweight during the follow-up years. Children whose mothers had perceived them as overweight thought they still had insufficient PA and they did not try to diet, modify their energy intake or reduce sedentary time more than children perceived as normal weight. Subsequently, children perceived as overweight gained more weight and had a higher risk of being overweight across childhood than did their counterparts. Also, several recent studies have found neither self-perception of overweight status among adults nor parental perception of child’s overweight status is a protective factor for subsequent weight gain, even though they were associated with an increased risk of future weight gain in longitudinal analysis( Reference Robinson and Sutin 9 , Reference Robinson, Hunger and Daly 23 , Reference Liechty and Lee 24 ). The underlying mechanisms remain to be determined.
We speculate these results could be due to the possible struggle between self-motivation from recognizing undesirable weight status (i.e. being overweight) and low self-esteem from weight stigma( Reference Tomiyama 25 ). Tomiyama described the cyclic obesity/weight-based stigma model as a positive feedback loop wherein weight stigma begets weight gain through increased eating behaviour and increased cortisol secretion governed by behavioural, emotional and physiological mechanisms, which are theorized to ultimately result in weight gain and difficulty of weight loss. Indeed, one study found that children who were labelled as fat by a family member or peer in early childhood were more likely to gain more weight into adolescence( Reference Hunger and Tomiyama 26 ). As indicated by weight management programmes, promoting children’s self-motivation through self-esteem is a key factor for successful obesity interventions( Reference Lowry, Sallinen and Janicke 27 ). Encouraging overweight children’s self-esteem may further enhance the moderate effects of the combined obesity interventions delivered in schools and home( Reference Wang, Cai and Wu 28 ).
Our study indicates that the rate of Chinese mothers with correct overweight perception of their children is lower than that of other countries, based on a worldwide meta-analysis( Reference Lundahl, Kidwell and Nelson 7 ). However, those mothers who were overweight or who had overweight children or urban children recognized the child as being overweight more than their counterparts, as these children had a higher risk of becoming overweight. To help these groups at risk, obesity intervention programmes in China should promote children’s self-motivation and improve their health behaviours. This will help reduce the growing epidemic of obesity and chronic diseases in China( Reference Wang, Mi and Shan 16 , Reference Wang, Wang and Qu 29 – Reference Xu, Ware and Leslie 32 ).
The current study had some limitations. Since data on the maternal perception of child’s weight status were collected only for the mother’s first child in CHNS, second and later children may have somewhat different effect size in the association. The level or intensity of children’s PA was not available to analyse as a function of maternal weight perception due to high rates of missing data. Due to data limitations, we could not account for the effects of breast-feeding history, birth weight and infancy growth characteristics. However, the study used a nationwide sample and longitudinal data, and studied the impact of maternal perceptions of children’s overweight status on children’s health behaviours and longitudinal risk of weight gain. Fifteen provinces and municipal cities that vary substantially in geography, economic development and public resources across China were included CHNS. Thus, findings based on the CHNS data could be generalized for the total Chinese population. The results provide some insight for future family-based interventions to fight the rising child obesity epidemic in China and other countries.
Conclusion
In conclusion, maternal perceptions of their children being overweight status led to weight-related parenting practices for child weight management, such as encouraging their child to increase exercise and modify their diet, but this did not improve children’s health behaviours or weight status. In China, obesity intervention programmes need to empower parents on accurate perception of their children’s weight status and healthy parenting practices and to motivate overweight children to maintain a healthy body weight.
Acknowledgements
Acknowledgements: The authors thank Dr Paula Vincent for her assistance in helping improve the manuscript. Financial support: The study was supported by the US National Institutes of Health (NIH; grant number U54HD070725). The U54 project (U54 HD070725) is co-funded by the NIH Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD) and the Office of the Director, National Institutes of Health (OD). Y.W. is the principal investigator. The content of the paper is solely the responsibility of the authors and does not necessarily represent the official views of the funders. Conflict of interest: None. Authorship: J. Min, V.H.C.W. and Y.W. formulated the research. J. Min, V.H.C.W. and Y.W. analysed the data and drafted the article. H.X., J. Mi and Y.W. revised the manuscript with critical comments. Ethics of human subject participation: Informed consent was obtained from all study participants for data collection in the CHNS. The present study was approved by the State University of New York at Buffalo Institutional Review Board.
Supplementary material
To view supplementary material for this article, please visit https://doi.org/10.1017/S1368980017001033
