Maternal Perceptions of Weight Status of Children

Maternal Perceptions of Weight Status of Children

Maynard et al. 2003

You asked for an outline to put on a PowerPoint presentation. You can bullet most of these points and put them right in a presentation:

Previous studies have found that a significant percentage of mothers with overweight preschool children failed to see the child as being overweight. These findings appear dependent on education with mothers with a high school education or less being more likely to misperceive their children's weight than those with a college education.

Such investigations are believed to be directly relevant to the success or prevention of overweight prevention or weight intervention programs as mothers who do not perceive a problem are less likely to take action for their children

(Good discussion point here is that mothers who misclassify children's weight use less negative descriptions of children such as "thick" or "solid")

Researchers wanted to better understand the extent to which mothers misclassify children's weight. Goals of the study were to:

A. Quantify extent to which mothers misclassify children's weight

B. Examine the demographic characteristics for both mother and children regarding misclassifications

C. Look at what factors may determine maternal misclassifications of at risk children

Methods

I. Data came from CDC study Third National Health and Nutrition Study

A. household surveys with caregivers (mothers) from 0-3 months after examination of child

B. Final analyses included 5500 children age 2-11 from a pool of 8309 (inclusion criteria for the study are on the third paragraph of page 1227 if you want to discuss this)

C. Mothers were asked if they thought child was under-over- or about the right weight (this is an ordinal variable given the way it is asked). Responses to the question were compared to child's Body Mass Index (BMI) that was recorded at the time of the examination of child (BMI would be an interval variable). BMI for age was converted to a z-score.

D. Weight classifications for study based on BMI:

1. Children with a BMI > 95th percentile were classified as overweight

2. > 85-95 percentile classified as at risk

3. 5-85 percentile normal

4. < 5 percentile underweight

E. Another important factor to mention is that BMI for age Z-scores were used (standard normal distribution)

F. Other variables and classifications: child's age (interval or ratio), height (interval or ratio), child's weight (interval or ratio), sex (nominal), race (nominal- non-Hispanic white, Hispanic, none-Hispanic Black, Mexican-American), region (nominal-northeast, Midwest, south, west), household poverty income ratio (interval), and mother's BMI (interval).

Variables

Measures of central tendency used in the study: Mean values for maternal BMI; child for age BMI- converted to Z scores, age (interval or ratio), and stature (height, interval or ratio)

Measures of dispersion used in study: standard errors for above variables; standard error of the percentage of mothers classifying their children's weight as either under-over- or about right.

Results

There are two tables on page 1228; the best one use is the Table 2-

I've reproduced what you need here- perhaps you can paste some of it in PowerPoint:

Weight Status of Children*

Underweight Normal Weight At Risk Overweight

Mother's Perception

Boys

Girls

Boys

Girls

Boys

Girls

Boys

Girls

Underweight

55.5

(7.5)

66.6

(8.1)

12.2

(1.3)

9.6

(1.3)

1.3

(1.3)

0.6

(0.4)

0.3

(0.2)

2.2

(1.3)

About Right Weight

44.5

(7.5)

33.4

(8.1)

86.1

(1.4)

88.6

(1.4)

84.7

(4.1)

70.4

(3.5)

35.0

(6.2)

29.2

(4.7)

Overweight

0.0

0.0

1.8

(0.6)

1.9

(0.5)

14.0

(3.8)

29.0

(3.5)

64.7

(6.2)

68.7

(4.4)

Figures are in percentages; (standard errors)

A. Relevant points for results:

1. 32.1% of mothers misclassified overweight as child as about right weight

2. For at risk 84.7% for boys and 72.4% for girls perceived them to be about the right weight

3. Mothers were more likely to perceive normal as underweight

4. Logistic Regression analysis indicated that the younger the child and lower child's BMI for age Z score the more likely they would classify an overweight child as about right. This means that younger overweight children were perceived as about right and the less obese that children were the more likely moms overlooked their weight problem.

5. For at risk children the logistic regression indicated that mothers more likely misclassified them as overweight due to their sex (females consider over more often than males), their age (older were more often thought overweight), higher BMI for age z scores, and mother's BMI (the lower the mother's BMI the more likely she considered at risk child being overweight

Overall conclusions

1. Almost 1/3 of mothers failed to classify overweight children as overweight

2. Girls at risk for being overweight were almost three times as likely to be classified by mom as overweight than were boys who were at risk.

3. Ethnic background did not predict misclassification rates

4. Younger children less likely to be classified as overweight (perhaps parents think younger children will grow out of it)

5. AS BMI for age Z score increased, overweight children less likely to be misclassified (As kids get fatter it is harder to lie-)

Questions you were asked regarding the statistics:

1. How were measures of central tendency and dispersion used?

Measures of central tendency and dispersion were used to convert the children's weight and height into BMI for age standard scores (Z scores). This allowed for the accurate classification of children as overweight, normal, or underweight (based on Z-score) in the analysis and maintained the variable as interval for use in regression analysis. The Z scores, which are standardized measures of dispersion, allow the convenient comparison of disparate measurements or variables with one another. The standard deviations in this data were used to calculate standard errors. Means and standard errors were also used to develop confidence intervals for the regression analysis results.

2. Were measures appropriate?

The mean is the most appropriate measure for the variables used as they are at least interval level variables. The standard error allows for great precision as it is a measure of dispersion like the standard deviation, but it tells us how the sample means are spread around the population mean (it is also useful in determining confidence intervals). The use of the standard error in this study is appropriate as it indicates that the current sample means are relatively good representations of the population means. Recall that the study is from a large population-based data set.

3. You are asked to compare data with the standard normal distribution, but the relevant data is converted to a standard normal distribution (BMI for age is converted to Z scores). Converting raw scores to scores on a standard normal distribution (converting to z scores) allows you to tell the position of any score in a distribution relative to the mean (positive z scores are above the mean, negative below) and also give you the ability to classify according to percentile rank. In addition, you can compare different scores from different variables when they are converted to z scores (for example you could compare one's weight to their IQ-that was not done here). In this study any child's status as obese, normal, or underweight can be determined by looking at their z-score. One issue that you might bring up is that we might not expect the weights of children to be normally distributed in the population and therefore z-scores may not adequately reflect the positions of children's weights or BMI's.