A Tale of Two Challenges Conducting Longitudinal Studies in Children and Adolescents: Accurately Measuring Diet and Body Composition in ALSPAC P. K. Newby,

A Tale of Two Challenges Conducting Longitudinal Studies in Children and Adolescents:

Accurately Measuring Diet and Body Composition in ALSPAC

P. K. Newby, ScD, MPH, MSAssociate Professor of Pediatrics, Epidemiology, Nutrition, and Gastronomy & Research Scientist

Boston University

pknewby@post.harvard.edu

http://www.pknewby.com

http://blog.pknewby.com

University of Bristol, UK

19 October 2011

Acknowledgments

Sabrina E. Noel, PhD, MS, RDSherman Bigornia, MAMichael LaValley, PhD

Lynn Moore, DScCarine Lenders, MD, ScD

Kate Northstone, PhD, MSPauline Emmett, PhD

Andy Ness, PhD, DPH (etc.)Li Benfield, PhD

Calum Mattocks, PhDChris Riddoch, PhD

Funding SourcesAmerican Diabetes Association

The UK Medical Research Council, Wellcome Trust, and the University of Bristol provide core support for ALSPAC.

Challenge 1: Measuring Diet– How to quantify dietary measurement errors?– Research example: Flavored milk and body fat

Challenge 2: Measuring Body Composition– Are we measuring what we think we’re measuring? – Research example: SSBs and body fat

Challenge 1: Dietary Reporting Errors

Significant misreporting of dietary intakes has been reported among children – Especially with increasing body weight and body fatness

Bandini et al, AJCN, 1990

r = -0.48, p < 0.001

Accounting for reporting errors is key for understanding diet-obesity relationships (but it is often overlooked)

Methods Used for Capturing Implausible Energy Reporters

• Premise: reported energy intake = energy expenditure under weight-stable conditions

• Direct measure of energy expenditure using doubly labeled water (DLW)– Compare reported intake to energy expenditure– Not feasible for large population

studies

• Equations to estimate implausible and plausible reporting– Compare reported intake to estimates of energy requirements

Goldberg et al, Eur J Clin Nutr, 1991; McCrory et al, Public Health Nutr, 2002; Huang et al, Obes Res 2004 & 2005DLW figure: http://www.iaea.org/newscenter/features/nutrition/energyintake.html

Capturing Implausible Reporters

• Age- and sex-specific cut-off for the ratio of reported energy intake to predicted energy requirements

• Predicted energy requirement equation (IOM)– Includes coefficients for age, physical activity (PA),

and weight and constants for sex and energy deposition during growth

Huang et al, 2004

Study Objective: Include objective measures of physical activity in equations used to predict energy requirements and quantify dietary reporting errors

- 2 methods used physical activity data from accelerometers- 1 assumed a low-active level

Three Variations of the PA Coefficient

IOM PA Category

IOM Description of Categories

IOM PA Coefficient

Categories based on mins of MVPA

Boys Girls

Sedentary Typical daily living activities

1.00 1.00 <30 minutes of MVPA

Low-active Sedentary + 30-60 min moderate activity

1.13 1.16 30 to <60 minutes of MVPA

Active Sedentary + 60 min moderate activity

1.26 1.31 >60 minutes to <120 minutes of MVPA

Very active Sedentary + 60 min moderate + 60 min vigorous or 120 min moderate activity

1.42 1.56 >120 minutes of MVPA

Percent Agreement between Methods

2. PAL Value Method 3. MVPA Method

1. Low-active Method

UR 51.8%

PR 37.9%

OR 10.3%

UR 37.1%

PR 42.4%

OR 20.4%

UR, 51.5% 88.0 15.5 0 97.4 36.1 0

PR, 40.8% 12.0 78.8 45.8 2.6 63.5 63.0

OR, 7.7% 0 5.7 54.2 0 0.4 37.0

к = 0.66 between the low-active and PAL value method; к = 0.53 between the low-active and MVPA method

Body Fatness Across Dietary Reporting Categories

Body Fat (%)

Method for Capturing Reporting Errors

a

bc

a

bc

a

bc

Comparison of Methods

0

10

20

30

40

50

60

UR

OR

DLW Studies 11- 15 y

Prediction Equations

% C

lass

ified

Our Methods

Conclusions and Next Steps

• All three methods were associated with sociodemographic and body composition measures as expected

• Inclusion of objectively measured physical activity as MVPA may have resulted in more reasonable estimates of plausible and implausible reporters

• Improving measurement of dietary reporting errors will improve precision and accuracy of results

• Future: Better quantification of MVPA using accelerometer data and direct comparisons with EE using DLW

Research example 1: Chocolate Milk, Body Fat, and Body Weight

Serving Size 1 cup (240mL)Amount per Serving

Calories 170Calories from Fat 25

% Daily ValuesTotal Fat 3g 4%  Saturated Fat  2g 9%  Trans Fat  0g  Cholesterol 15mg 4%Sodium 170mg 7%Total Carbohydrate 28g

9%

Fiber <1g 3%Sugar 26g  Protein 9g 17%

Vitamin A 10% Vitamin C 0%Calcium 50% Iron 4%Vitamin D 25%  

http://www.hood.com/Products/prodDetail.aspx?id=639

Flavored milk consumers had less favorable changes in body fat

Means were adjusted for pubertal status, maternal BMI and educational attainment, changes in age, height, height squared, physical activity, and intakes of total fat, ready-to-eat cereal, 100% fruit juice, sugar-sweetened beverage, and plain milk. Plausible reporters only.

Conclusions and Next Steps

• Less favorable changes in body fat and weight were seen for overweight children consuming flavored milk compared with non-consumers over a 2 year period

• Associations were strengthened when reporting errors were considered.

• These results limit recommendations that promote flavored milk consumption among children, especially those who are overweight or obese

• Future: Repeating study with greater variability in intakes and conducting an analysis looking at total dairy

Challenge 2: How to Measure Body Fat

• Central adiposity is an important chronic disease risk factor in adults

• Studies in children suggest correlations between central and total adiposity are high due to limited accrual of visceral fat

• Little is known how these relationships change as children move through puberty.

Study Objectives:1. Examine relationships between central and total adiposity

assessed by anthropometry, DXA and MRI (11 and 13 y only) at 9, 11, 13, and 15 y of age

2. Compare how measures of central and total adiposity were associated with SSBs and systolic blood pressure

Methods

Body composition• Total adiposity: BMI (kg/m2) and total body fat mass

(TBFM, g) by DXA• Central adiposity: waist circumference (WC, cm),

trunk fat mass (TFM, g) by DXA, and intra-abdominal adipose tissue (IAAT, cm3) by MRI

Sexual Maturity • Self-reported tanner stage (5 levels) collapsed to

pre (1), early (2-3), and late (4-5).

Relationships between central and total adiposity measures among children at ages 9, 11, 13, and 15 y.*

*WC, waist circumference; TBFM, total body fat mass, TFM, trunk fat mass† Values are the partial variances (%) accounted by select adiposity measures by multivariate linear regression with adjustment for age, height , and pubertal stage (pre-, early, and late).

n=2031 n=1816 n=1616 n=962 n=437 n=505 n=370 n=192

n=672 n=646 n=486 n=228 n=2183 n=2079 n=1824 n=1173

Relationships between adiposity measures and intra-abdominal adipose tissue volume at ages 11 and 13 y*

*Data are Pearson’s partial correlation coefficients adjusted for age and height. P < 0.05 for all values.

†MRI data were collected at 11 and 13 on a subset of ALSPAC participants.

Conclusions

• Central and total fat measures were strongly correlated at all ages and modestly attenuated at age13 and 15 years.

• BMI, WC, TBFM, and TFM correlations with IAAT were comparable.

• Similar associations were observed with SBP (data not shown).

• Our findings have implications for the interpretation of epidemiological studies examining central adiposity on metabolic outcomes in late childhood and early adolescence, highlighting the need to also consider associations with total adiposity as they explain a large amount of variation in central adiposity

Research Example 2: SSBs and Body Composition

1) Examine the effect of change in SSB intake from 10 to 13 y (∆SSB) on total adiposity (BMI and total body fat) at 13 y

1) Determine whether SSB consumption has similar and additional effects on measures of total and central adiposity (waist circumference)

2) Adjust for dietary reporting errors

Methods

Diet • 3 day diet records at 10 and 13 y • Sugar-sweetened beverages (SSB): fruit squashes,

cordials and fizzy drinks (i.e. soda) with added sugar. 140 g water assumed for every 40 g of concentrate. 180 g = 1 serving

• Change in SSB (∆SSB) = SSB 13 – SSB 11

Adiposity • BMI, waist circumference (WC), and total body fat

mass (TBFM) at 13 y as previously described

∆SSBs (servings/d) and central and total adiposity at 13 y (n=2,455)

Adiposity at 13

Model1

Change in adiposity per

∆SSB (servings/d)2

Standardized Beta

P value

BMI, kg/m2 1 0.07 (0.03) 0.028 0.0252 0.09 (0.03) 0.039 0.0023 0.16 (0.04) 0.074 <0.001

Waist, cm 1 0.13 (0.10) 0.020 0.1882 0.22 (0.10) 0.034 0.0253 0.55 (0.14) 0.097 <0.001

Total body fat, kg

10.10 (0.08) 0.017 0.203

2 0.19 (0.08) 0.033 0.0113 0.33 (0.11) 0.065 0.003

∆SSBs (servings/d) and central adiposity at 13 y (n=2,455)

General adiposity at 13

adjustmentModel

Change in adiposity per ∆SSB

(servings/d)2

Standardized Beta

P value

Waist, cm BMI, kg/m2

1 0.07 (0.07) 0.011 0.292 0.06 (0.07) 0.010 0.373 0.24 (0.10) 0.042 0.02

Waist, cm

Total body fat, kg 1 0.11 (0.07) 0.018 0.10

2 0.08 (0.07) 0.013 0.223 0.27 (0.11) 0.048 0.01

Conclusions

• Increased SSB intakes over 3 y was associated with higher BMI and fat mass at 13 y supporting recommendations to limit SSB consumption to combat excess weight gain

• SSBs have somewhat stronger and additional effects on WC independent of total adiposity but these are likely not clinically meaningful

• Accounting for dietary reporting errors uniformly strengthened effect estimates, highlighting the importance of measuring and accounting for these errors.

Publications (Published and In Progress)

Noel SE, Ness AR, Northstone K, Emmett PE, Newby PK. Flavored milk consumption and changes in body fat in children: a prospective study. Journal of Nutrition. Submitted.

Bigornia SJ, Noel SE, LaValley MP, Moore LL, Ness AR, Newby PK. Sugar-sweetened beverage intake among children from 10 to 13 years of age and central and total adiposity: a prospective population based cohort study. International Journal of Obesity. Submitted.

Bigornia SJ, LaValley MP, Benfield LL, Ness AR, Newby PK. Relationships between direct and indirect measures of central and total adiposity in children at 9, 11, 13, and 15 years of age. American Journal of Clinical Nutrition. Submitted.  

Noel SE, Ness AR, Northstone K, Emmett P, Newby PK. Milk intakes are not associated with percent body fat in children from ages 10 to 13 years. Journal of Nutrition 2011; Sept 21. [Epub ahead of print]

Noel SA, Mattocks C, Riddoch C, Emmett PE, Ness AR, Newby PK. Use of accelerometer data in prediction equations for capturing implausible dietary intakes among adolescents. American Journal of Clinical Nutrition 2010;92(6):1436-45.

Thank you for your attention!

P. K. Newby, ScD, MPH, MSAssociate Professor of Pediatrics, Epidemiology, Nutrition, and Gastronomy & Research Scientist

Boston University

pknewby@post.harvard.edu

http://www.pknewby.com

http://blog.pknewby.com

University of Bristol, UK

19 October 2011

Supplemental Slides

Sample characteristics by flavored milk consumption

Sample Characteristics

Flavored milk non-consumers,

age 10 y

Flavored milk consumers, age

10 y

P value

Girls, % 55.8 49.0 0.01

Body fat, %

11 y 25.5 ± 9.1 25.5 ± 9.3 0.98

13 y 24.4 ± 10.1 24.8 ± 10.7 0.50

Physical activity

11 y 587.8 ± 171.9 585.5 ± 165.6 0.80

13 y 536.0 ± 193.5 534.5 ± 177.9 0.89

Dieting at age 13 y, %

25.7 19.4 0.02

Maternal body mass index, kg/m2

24.5 ± 4.4 24.6 ± 4.8 0.73

Table 2. Adjusted means of daily total energy and selected nutrient & food intakes

Energy, nutrient and food group intake

Flavored milk non-consumers, age 10

y (n=1890)

Flavored milk consumers, age 10 y

(n=380)

P value

Total energy, kcal 1917 ± 11 2064 ± 24 <0.001

Fat, g 75.6 ± 0.32 77.5 ± 0.71 0.01

Saturated fat, g 29.2 ± 0.16 30.6 ± 0.37 <0.001

Carbohydrate, g 251.0 ± .86 258.0 ± 1.9 0.001

Fiber, g 11.8 ± 0.07 11.2 ± 0.16 0.002

Added sugars, g 89.1 ± 0.67 85.9 ± 1.5 0.05

Dietary calcium, g 796.1 ± 5.7 917.4 ± 12.8 <0.001

Sugar-sweetened beverages3, g

106.8 ± 3.31 92.6 ± 7.39 0.08

Means for total energy intake were adjusted for sex only. Means for all other nutrients and food groups were adjusted for sex and total energy intake.

Flavored milk non-consumers, age 10

(n=1890)

Flavored milk consumers, age 10

(n=380)

P value

Mean 95% CI Mean 95% CI

Normal weight childrenChange in % body

fat, (n=1,715) Model 1 -0.83 -1.42, -0.24 -0.63 -1.37, 0.12 0.48

Model 2 -0.86 -1.44, -0.27 -0.60 -1.35, 0.14 0.40

Overweight/obese childrenChange in % body

fat, (n=449)Model 1 -2.64 -3.82, -1.45 -1.09 -2.60, 0.41 0.01

Model 2 -2.64 -3.83, -1.45 -1.11 -2.62, 0.40 0.01

Model 1 was adjusted for change in counts per minute, pubertal status, maternal BMI and educational attainment, change in total fat intake, and change in ready-to-eat cereal, 100% fruit juice and SSB intake. Model 2 also included change in total milk intake.

Pearson’s partial correlations between systolic blood pressure and BMI, WC, TBFM and TFM from 9 to 15 y adjusted for age and height