Childhood nutrient intakes are differentially associated with hepatic and abdominal fats in adolescence: The EPOCH study

OBJECTIVE

The aim of this study was to examine whether nutrient intakes in childhood are associated with abdominal and hepatic fat depots later in adolescence.

METHODS

Using data from 302 participants in the longitudinal Exploring Perinatal Outcomes among CHildren (EPOCH) study, energy partition and nutrient density models were constructed to examine associations of nutrient intakes in childhood (~10 years of age), assessed by food frequency questionnaire, with abdominal subcutaneous adipose tissue (SAT), visceral adipose tissue (VAT), and hepatic fat in adolescence (~16 years of age).

RESULTS

In energy partition models (energy intake not held constant), total, monounsaturated, and polyunsaturated fat intakes in childhood were associated with higher SAT in adolescence (β [95% CI]: 8.5 [0.1-17.1], 25.1 [2.1-48.1], and 59.7 [16.1-103.3] mm² per 100 kcal/d), higher starch intake was associated with log-hepatic fat (back-transformed β [95% CI]: 1.07 [1.01-1.15] per 100 kcal/d), and, in boys only, higher animal protein intake was associated with VAT (β [95% CI]: 5.3 [0.3-10.3] mm² per 100 kcal/d). Most associations were unchanged when adjusted for energy intake in nutrient density models.

CONCLUSIONS

Childhood nutrient intakes were differentially associated with adolescent body fats; specifically, unsaturated fat intake predicted abdominal SAT, animal protein intake predicted VAT, and starch intake predicted hepatic fat. These nutrient intakes may, therefore, be targets for intervention studies aiming to modify adolescent body fat distribution.

The maternal diet index in pregnancy is associated with offspring allergic diseases: the Healthy Start study

BACKGROUND

A systematic review showed limited associations between pregnancy diet and offspring allergy. We developed a maternal diet index during pregnancy that was associated with offspring allergy outcomes.

METHODS

Data came from Healthy Start, a Colorado pre-birth cohort of mother/offspring dyads. Food propensity questionnaires were completed during pregnancy. Offspring allergic rhinitis, atopic dermatitis, asthma, wheeze, and food allergy diagnosis up to age four were verified from electronic medical records. Data were randomized into test and replication sets. The index included the weighted combination of variables that best predicted a combined outcome of any allergy in the test set. Index utility was verified in the replication set. Separate adjusted and unadjusted logistic models estimated associations between the index and each offspring allergy diagnosis in the full sample.

RESULTS

The index included weighted measures of intake of vegetables, yogurt, fried potatoes, rice/grains, red meats, pure fruit juice, and cold cereals. Vegetables and yogurt were associated with the prevention of any allergy, while other components were associated with increased disease. In adjusted models, a one-unit increase in the index was significantly associated with reduced odds of offspring allergic rhinitis (odds ratio (CI) 0.82 [0.72-0.94]), atopic dermatitis (0.77 [0.69-0.86]), asthma (0.84 [0.74-0.96]), and wheeze (0.80 [0.71-0.90]), but not food allergy (0.84 [0.66-1.08]).

CONCLUSIONS

This is the first study that has shown associations between an index of maternal dietary intake during pregnancy and multiple offspring allergic diseases. The results give hope for prevention of allergic diseases in utero.

Disparities in Risks of Inadequate and Excessive Intake of Micronutrients during Pregnancy

BACKGROUND

Inadequate or excessive intake of micronutrients in pregnancy has potential to negatively impact maternal/offspring health outcomes.

OBJECTIVE

The aim was to compare risks of inadequate or excessive micronutrient intake in diverse females with singleton pregnancies by strata of maternal age, race/ethnicity, education, and prepregnancy BMI.

METHODS

Fifteen observational cohorts in the US Environmental influences on Child Health Outcomes (ECHO) Consortium assessed participant dietary intake with 24-h dietary recalls (n = 1910) or food-frequency questionnaires (n = 7891) from 1999-2019. We compared the distributions of usual intake of 19 micronutrients from food alone (15 cohorts; n = 9801) and food plus dietary supplements (10 cohorts with supplement data; n = 7082) to estimate the proportion with usual daily intakes below their age-specific daily Estimated Average Requirement (EAR), above their Adequate Intake (AI), and above their Tolerable Upper Intake Level (UL), overall and within sociodemographic and anthropometric subgroups.

RESULTS

Risk of inadequate intake from food alone ranged from 0% to 87%, depending on the micronutrient and assessment methodology. When dietary supplements were included, some women were below the EAR for vitamin D (20-38%), vitamin E (17-22%), and magnesium (39-41%); some women were above the AI for vitamin K (63-75%), choline (7%), and potassium (37-53%); and some were above the UL for folic acid (32-51%), iron (39-40%), and zinc (19-20%). Highest risks for inadequate intakes were observed among participants with age 14-18 y (6 nutrients), non-White race or Hispanic ethnicity (10 nutrients), less than a high school education (9 nutrients), or obesity (9 nutrients).

CONCLUSIONS

Improved diet quality is needed for most pregnant females. Even with dietary supplement use, >20% of participants were at risk of inadequate intake of ≥1 micronutrients, especially in some population subgroups. Pregnancy may be a window of opportunity to address disparities in micronutrient intake that could contribute to intergenerational health inequalities.

Adherence to index-based dietary patterns in childhood and BMI trajectory during the transition to adolescence: the EPOCH study

BACKGROUND

In the United States, one in five adolescents are obese. Index-based dietary patterns are measures of the overall diet that have the potential to serve as valuable obesity risk stratification tools. However, little is known about the association between adherence to index-based dietary patterns in childhood and BMI during the transition from childhood to adolescence.

OBJECTIVE

To prospectively examine the relationship between adherence to three index-based dietary patterns in childhood and BMI trajectory during the transition to adolescence.

METHODS

The study included 581 children enrolled in a Colorado prospective cohort study conducted between 2006 and 2015. Dietary intake was assessed with the Block Kids Food Frequency Questionnaire at age 10 years. Scores were calculated for the Healthy Eating Index-2010 (HEI-2010), the alternate Mediterranean (aMED) diet, and the Dietary Approaches to Stop Hypertension (DASH) diet. Weight and height were assessed via anthropometry at two research visits (ages 10 and 16 years), with interim clinical measurements extracted from Kaiser Permanente medical records. Separate mixed models were used to assess the association between each diet index score and BMI over a 6-year period. Models were stratified by sex and adjusted for age, race/ethnicity, income, and exposure to gestational diabetes.

RESULTS

Median (IQR) number of BMI assessments was 14 (10-18). Among girls, for every ten-unit increase in HEI-2010 score, there was an average 0.64 kg/m² decrease (p = 0.007) in BMI over time, after adjustment for covariates. Among girls, there was no association between BMI and aMED (β = -0.19, p = 0.24) or DASH (β = 0.28, p = 0.38). Among boys, there was no statistically significant association between BMI and HEI-2010 (0.06, p = 0.83), aMED (0.07, p = 0.70), or DASH (0.42, p = 0.06).

CONCLUSIONS

Efforts to prevent adolescent obesity could benefit from considering the degree of adherence to federal dietary guidance, as assessed by the HEI, in the period preceding adolescence, especially among girls.

Associations of Nutrient Intake Changes During Childhood with Adolescent Hepatic Fat: The Exploring Perinatal Outcomes Among CHildren Study

OBJECTIVE

To examine associations of dietary changes from childhood to adolescence with adolescent hepatic fat and whether the PNPLA3 rs738409 risk allele, a strong genetic risk factor for hepatic fat, modifies associations.

STUDY DESIGN

Data were from 358 participants in the Exploring Perinatal Outcomes among CHildren (EPOCH) study, a longitudinal cohort in Colorado. Diet was assessed by food frequency questionnaire in childhood (approximately 10 years of age) and adolescence (approximately 16 years of age) and converted to nutrient densities. Hepatic fat was assessed in adolescence by magnetic resonance imaging. Linear regression was used to test associations of dietary changes from childhood to adolescence with adolescent hepatic fat.

RESULTS

Increases in fiber, vegetable protein, and polyunsaturated fat intake from childhood to adolescence were associated with lower adolescent hepatic fat, and increases in animal protein were associated with higher hepatic fat (β per 5-unit increase on log-hepatic fat: -0.12 [95% CI, -0.21 to -0.02] for ▵fiber; -0.26 [95% CI, -0.45 to -0.07] for ▵vegetable protein; -0.18 [95% CI, -0.35 to -0.02] for ▵polyunsaturated fat; 0.13 [95% CI, 0.04-0.22] for ▵animal protein). There was evidence of effect modification by PNPLA3 variant, whereby inverse associations of ▵fiber and ▵vegetable protein and positive associations of ▵saturated fat with adolescent hepatic fat were stronger in risk allele carriers. Most conclusions were similar after adjusting for obesity in adolescence, but associations of ▵saturated fat with hepatic fat were attenuated toward the null.

CONCLUSIONS

Our results suggest that nutrient intake changes between childhood and adolescence, particularly decreases in fiber and vegetable protein and increases in saturated fat intake, interact with the PNPLA3 variant to predict higher hepatic fat in adolescence, and may be targets for reducing hepatic fat in high-risk youth.

A power approximation for the Kenward and Roger Wald test in the linear mixed model

We derive a noncentral F power approximation for the Kenward and Roger test. We use a method of moments approach to form an approximate distribution for the Kenward and Roger scaled Wald statistic, under the alternative. The result depends on the approximate moments of the unscaled Wald statistic. Via Monte Carlo simulation, we demonstrate that the new power approximation is accurate for cluster randomized trials and longitudinal study designs. The method retains accuracy for small sample sizes, even in the presence of missing data. We illustrate the method with a power calculation for an unbalanced group-randomized trial in oral cancer prevention.

A Prudent dietary pattern is inversely associated with liver fat content among multi-ethnic youth

OBJECTIVES

To identify dietary patterns associated with hepatic fat fraction (HFF), a measure of liver fat content and risk factor for non-alcoholic fatty liver disease, in a prospective study of 397 multi-ethnic youth.

METHODS

We obtained information on habitual dietary intake via the Block Kids Food Frequency Questionnaire at age 6 to 15 years ('T1') and 12 to 19 years ('T2'), and measured HFF using magnetic resonance imaging at T2. We derived dietary patterns via principal components analysis and examined associations with ln-transformed HFF using linear regression models that accounted for maternal education, gestational diabetes exposure and smoking habits; and child pubertal status, BMI and physical activity.

RESULTS

At T1, none of the dietary patterns identified were associated with HFF measured at T2. At T2, a Prudent dietary pattern characterized by high fruit and vegetable intake was inversely associated with HFF (-0.08 [95% CI: -0.16, -0.00]). Similarly, increased adherence to the Prudent pattern across T1 and T2 corresponded with lower ln-HFF (-0.11 [-0.18, -0.04] units). On the other hand, adherence to a Western pattern comprising fried foods and refined carbohydrates at T2 correlated with higher HFF among non-Hispanic White participants (0.16 [0.06, 0.26]). These findings persisted after accounting for child BMI.

CONCLUSIONS

Even in healthy youth, a diet high in fruits and vegetables is associated with lower HFF, whereas a diet high in fried foods and refined carbohydrates is related to higher HFF. Dietary changes may serve as an early preventive measure to mitigate liver fat accrual.

Power for balanced linear mixed models with complex missing data processes

When designing repeated measures studies, both the amount and the pattern of missing outcome data can affect power. The chance that an observation is missing may vary across measurements, and missingness may be correlated across measurements. For example, in a physiotherapy study of patients with Parkinson's disease, increasing intermittent dropout over time yielded missing measurements of physical function. In this example, we assume data are missing completely at random, since the chance that a data point was missing appears to be unrelated to either outcomes or covariates. For data missing completely at random, we propose noncentral F power approximations for the Wald test for balanced linear mixed models with Gaussian responses. The power approximations are based on moments of missing data summary statistics. The moments were derived assuming a conditional linear missingness process. The approach provides approximate power for both complete-case analyses, which include independent sampling units where all measurements are present, and observed-case analyses, which include all independent sampling units with at least one measurement. Monte Carlo simulations demonstrate the accuracy of the method in small samples. We illustrate the utility of the method by computing power for proposed replications of the Parkinson's study.

Examining Associations Between Dietary Inflammatory Index in Pregnancy, Pro-inflammatory Cytokine and Chemokine Levels at Birth, and Offspring Asthma and/or Wheeze by Age 4 Years

BACKGROUND

Few studies have demonstrated associations between maternal dietary inflammatory index (DII) during pregnancy and offspring asthma and/or wheeze.

OBJECTIVE

The study aimed to assess associations between maternal DII during pregnancy and 1) offspring cord sera pro-inflammatory cytokines (interleukin [IL]-1β, IL-4, IL-6, IL-10, tumor necrosis factor-α) and chemokines (IL-8, monocyte chemoattractant protein-1) at birth and 2) offspring asthma and/or wheeze at age 4 years.

DESIGN

The Healthy Start study is a prospective prebirth longitudinal study that recruited pregnant women in Denver, Colorado and tracked their offspring.

PARTICIPANTS AND SETTING

This study used data from 1228 mother-child dyads enrolled in the Healthy Start study. Pregnant women were recruited in Denver, Colorado, between 2009 and 2014, and offspring tracked until age 4 years.

MAIN OUTCOME MEASURES

Cord sera cytokines and chemokines were analyzed with multiplex panel immunoassays. Offspring diagnosis of asthma and/or wheeze by age 4 years was extracted from electronic medical records.

STATISTICAL ANALYSES PERFORMED

Unadjusted and adjusted linear and logistic regression models were used to assess associations. Covariates included factors such as nulliparity, race/ethnicity, gestational smoking, and maternal history of asthma.

RESULTS

Unadjusted analysis showed that increasing maternal DII scores were associated with increased odds of child asthma and/or wheeze by 4 years (odds ratio = 1.17; 95% CI: 1.07-1.27), but the association was attenuated and no longer statistically significant in the adjusted model (odds ratio = 1.15; 95% CI: 0.99-1.33). There were no significant associations between DII scores and cord sera cytokine or chemokine levels.

CONCLUSIONS

The study showed that the inflammatory profile of the maternal diet was not associated with cytokines and chemokine levels at birth. The results suggested that a more inflammatory maternal diet was associated with increased odds of offspring asthma and/or wheeze by age 4 years, which could be considered of clinical relevance but the finding was not statistically significant at the .05 level.

Maternal blood glucose level and offspring glucose-insulin homeostasis: what is the role of offspring adiposity?

AIMS/HYPOTHESIS

The aim of this work was to investigate the association of maternal HbA_1c during mid-pregnancy with biomarkers of glucose-insulin homeostasis during early childhood (4-7 years of age) and to assess whether and how offspring adiposity at birth and at age 4-7 years mediates this relationship among 345 mother-child pairs in the Healthy Start Study.

METHODS

The exposure was maternal HbA_1c (mmol/mol) measured at 20-34 gestational weeks and categorised into tertiles. The outcomes were offspring fasting glucose, 1/insulin, HOMA2-IR, and HOMA2-B at age 4-7 years. The mediators were per cent fat mass (%FM) at birth, %FM at age 4-7 years, and the sum of the two as a metric of cumulative adiposity. Mediation analyses were conducted via a counterfactual-based approach. All models accounted for maternal race/ethnicity, offspring age and sex.

RESULTS

There was a significant total effect of maternal HbA_1c on offspring glucose and 1/insulin. Specifically, we observed a positive trend across tertiles of HbA_1c and offspring glucose (p trend <0.001), and an inverse trend across tertiles of HbA_1c and offspring 1/insulin (p trend = 0.04). For instance, compared with offspring of women in the lowest tertile of HbA_1c, those whose mothers were in the second and third tertiles had 0.04 mmol/l (95% CI -0.05, 0.13) and 0.17 mmol/l (95% CI 0.08, 0.26) higher fasting glucose concentrations at age 4-7 years, respectively. Adjustment for pre-pregnancy BMI did not appreciably change the results. We found no evidence of mediation by offspring adiposity at any life stage.

CONCLUSIONS/INTERPRETATION

Offspring of women with higher HbA_1c during pregnancy had higher fasting glucose and lower insulin sensitivity by early childhood. These relationships were largely unaffected by the child's own adiposity. Graphical abstract.

Cord Blood Vitamin D Levels and Early Childhood Blood Pressure: The Healthy Start Study

Background Vitamin D deficiency is associated with cardiovascular events among adults, but it is unclear whether early-life vitamin D deficiency influences cardiovascular risk factors in children. Methods and Results We measured total and bioavailable 25-dihydroxyvitamin D (25OHD) in cord blood and in blood from 4- to 6-year-old children, and we assessed cardiovascular risk factors (blood pressure, arterial stiffness, body size, and adiposity) at 4 to 6 years. We tested for racial/ethnic differences in total and bioavailable 25OHD (n=715) and modeled the adjusted association between cord blood 25OHD and childhood cardiovascular risk factors (n=171). We observed racial/ethnic differences in total and bioavailable 25OHD levels in both cord and child blood samples (all P<0.05). Each 25-nmol/L increase in cord blood total 25OHD was associated with a 2.5-mm Hg (SE 0.8) decrease in systolic blood pressure ( P=0.002) and a 1.7-mm Hg (SE 0.6) decrease in diastolic blood pressure ( P=0.01), independent of childhood 25OHD levels, race/ethnicity, and other covariates. There was no association between cord blood total 25OHD and any other cardiovascular risk factors. Cord blood levels of bioavailable and free 25OHD were not associated with any cardiovascular risk factor in childhood. Conclusions In this diverse prebirth cohort, we observed lower systolic and diastolic blood pressure among children with higher total 25OHD levels at birth. Our findings suggest that intrauterine exposure to vitamin D may contribute to early-life programming of offspring blood pressure. Intervention studies are needed to determine whether increasing fetal vitamin D exposure can reduce the risk of elevated blood pressure in childhood.

A prospective study of associations between in utero exposure to gestational diabetes mellitus and metabolomic profiles during late childhood and adolescence

AIMS/HYPOTHESIS

This study aimed to: (1) identify metabolite patterns during late childhood that differ with respect to exposure to maternal gestational diabetes mellitus (GDM); (2) examine the persistence of GDM/metabolite associations 5 years later, during adolescence; and (3) investigate the associations of metabolite patterns with adiposity and metabolic biomarkers from childhood through adolescence.

METHODS

This study included 592 mother-child pairs with information on GDM exposure (n = 92 exposed), untargeted metabolomics data at age 6-14 years (T1) and at 12-19 years (T2), and information on adiposity and metabolic risk biomarkers at T1 and T2. We first consolidated 767 metabolites at T1 into factors (metabolite patterns) via principal component analysis (PCA) and used multivariable regression to identify factors that differed by GDM exposure, at α = 0.05. We then examined associations of GDM with individual metabolites within factors of interest at T1 and T2, and investigated associations of GDM-related factors at T1 with adiposity and metabolic risk throughout T1 and T2 using mixed-effects linear regression models.

RESULTS

Of the six factors retained from PCA, GDM exposure was associated with greater odds of being in quartile (Q)4 (vs Q1-3) of 'Factor 4' at T1 after accounting for age, sex, race/ethnicity, maternal smoking habits during pregnancy, Tanner stage, physical activity and total energy intake, at α = 0.05 (OR 1.78 [95% CI 1.04, 3.04]; p = 0.04). This metabolite pattern comprised phosphatidylcholines, diacylglycerols and phosphatidylethanolamines. GDM was consistently associated with elevations in a subset of individual compounds within this pattern at T1 and T2. While this metabolite pattern was not related to the health outcomes in boys, it corresponded with greater adiposity and a worse metabolic profile among girls throughout the follow-up period. Each 1-unit increment in Factor 4 corresponded with 0.17 (0.08, 0.25) units higher BMI z score, 8.83 (5.07, 12.59) pmol/l higher fasting insulin, 0.28 (0.13, 0.43) units higher HOMA-IR, and 4.73 (2.15, 7.31) nmol/l higher leptin.

CONCLUSIONS/INTERPRETATION

Exposure to maternal GDM was nominally associated with a metabolite pattern characterised by elevated serum phospholipids in late childhood and adolescence at α = 0.05. This metabolite pattern was associated with greater adiposity and metabolic risk among female offspring throughout the late childhood-to-adolescence transition. Future studies are warranted to confirm our findings.

Sex differences in infant body composition emerge in the first 5 months of life

Background Sex differences in body composition are appreciated throughout the lifespan with probable contributions from sex steroids: testosterone and estrogen. The purpose of this longitudinal observational study was to determine if sex differences in body composition emerge during the first months of life in healthy infants, corresponding to the age at which male infants produce endogenous testosterone. Methods Linear growth and body composition parameters using air displacement plethysmography were obtained from 602 healthy infants after birth and again at 5 months of age. Rate of change in body composition parameters were compared between sexes. Results Sex differences in length, total mass, fat free mass (FFM), and percent fat mass (%FM) were present both at birth and at 5 months (p < 0.001 for all), with males having greater total mass and FFM but lower %FM. Gain in %FM over the first 5 months was significantly lower in males (p = 0.0004). This difference was secondary to a gain of 17 g/week more in FFM in males compared to females. Conclusions Sex differences in body composition emerge in the first months of life, with lower adiposity accumulation in males. Endogenous testosterone production in males ~1-4 months of age may account for findings and may have lifelong implications for sex differences in body composition.

Persistent effects of in utero overnutrition on offspring adiposity: the Exploring Perinatal Outcomes among Children (EPOCH) study

AIMS/HYPOTHESIS

We previously showed that intrauterine exposure to gestational diabetes mellitus (GDM) increases selected markers of adiposity in pre-pubertal adolescents. In the present study, we examined these associations in adolescence, and explored whether they are strengthened as the participants transition through puberty.

METHODS

Data from 597 individuals (505 unexposed, 92 exposed) participating in the longitudinal Exploring Perinatal Outcomes among Children (EPOCH) study in Colorado were collected at two research visits when the participants were, on average, 10.4 and 16.7 years old. Adiposity measures included BMI, waist/height ratio, and visceral and subcutaneous adipose tissue (as determined by MRI). Separate general linear mixed models were used to assess the longitudinal relationships between exposure to maternal GDM and each adiposity outcome. We tested whether the effect changed over time by including an interaction term between exposure and age in our models, and whether the associations were explained by postnatal behaviours.

RESULTS

Compared with unexposed participants, those exposed to maternal GDM had higher BMI (β = 1.28; 95% CI 0.35, 2.21; p < 0.007), waist/height ratio (β = 0.03; 95% CI 0.01, 0.04; p = 0.0004), visceral adipose tissue (β = 4.81; 95% CI 1.08, 8.54; p = 0.01) and subcutaneous adipose tissue (β = 35.15; 95% CI 12.43, 57.87; p < 0.003). The magnitude of these differences did not change over time and the associations did not appear to be explained by postnatal behaviours.

CONCLUSIONS/INTERPRETATION

Our data provide further evidence that intrauterine exposure to maternal GDM is associated with increased offspring adiposity, an effect that appears early in life and tracks throughout adolescence. Efforts to prevent childhood obesity following intrauterine exposure to maternal GDM should target the prenatal or early life periods.

Calculating Power for the General Linear Multivariate Model With One or More Gaussian Covariates

We describe a noncentral ℱ power approximation for hypotheses about fixed predictors in general linear multivariate models with one or more Gaussian covariates. The results apply to both single and multiple parameter hypotheses. The approach extends power approximations for models with only fixed predictors, and for models with a single Gaussian covariate. The new method approximates the noncentrality parameter under the alternative hypothesis using a Taylor series expansion for the matrix-variate beta distribution of type I. We used a Monte Carlo simulation to evaluate the accuracy of both the novel power approximation, and published power approximations. The simulation study accounted for randomness in both the predictors and the errors. We varied the number of outcomes, the number of parameters in the hypothesis, the per-treatment sample size, and the correlations between the random predictors and the outcomes. We demonstrate that our approximation is more accurate than published methods, both in small and large samples. We show that the run time for a single power calculation with the new method is on the order of milliseconds, compared to an average empirical simulation time of roughly three minutes. Approximate and simulated power can be calculated using the free, open-source rPowerlib package. (http://github.com/SampleSizeShop/rPowerlib).

Improving the diagnostic accuracy of a stratified screening strategy by identifying the optimal risk cutoff

BACKGROUND

The American Cancer Society (ACS) suggests using a stratified strategy for breast cancer screening. The strategy includes assessing risk of breast cancer, screening women at high risk with both MRI and mammography, and screening women at low risk with mammography alone. The ACS chose their cutoff for high risk using expert consensus.

METHODS

We propose instead an analytic approach that maximizes the diagnostic accuracy (AUC/ROC) of a risk-based stratified screening strategy in a population. The inputs are the joint distribution of screening test scores, and the odds of disease, for the given risk score. Using the approach for breast cancer screening, we estimated the optimal risk cutoff for two different risk models: the Breast Cancer Screening Consortium (BCSC) model and a hypothetical model with much better discriminatory accuracy. Data on mammography and MRI test score distributions were drawn from the Magnetic Resonance Imaging Screening Study Group.

RESULTS

A risk model with an excellent discriminatory accuracy (c-statistic [Formula: see text]) yielded a reasonable cutoff where only about 20% of women had dual screening. However, the BCSC risk model (c-statistic [Formula: see text]) lacked the discriminatory accuracy to differentiate between women who needed dual screening, and women who needed only mammography.

CONCLUSION

Our research provides a general approach to optimize the diagnostic accuracy of a stratified screening strategy in a population, and to assess whether risk models are sufficiently accurate to guide stratified screening. For breast cancer, most risk models lack enough discriminatory accuracy to make stratified screening a reasonable recommendation.

Childhood Metabolic Biomarkers Are Associated with Performance on Cognitive Tasks in Young Children

OBJECTIVE

To evaluate the hypothesis that metabolic measures (fasting glucose, insulin, and Homeostatic Model of Assessment for Insulin Resistance [HOMA-IR] levels) are inversely associated with performance on cognitive tasks using data from young (4- to 6-year-old), typically developing, healthy children.

STUDY DESIGN

Data were obtained from children participating in the Healthy Start study, a pre-birth cohort in Colorado. HOMA-IR, glucose, and insulin values were centered and scaled using the study sample means and SD. Thus, they are reported in number of SD units from the mean. Fully corrected T scores for inhibitory control (Flanker task), cognitive flexibility (Dimensional Change Card Sort test), and receptive language (Picture Vocabulary test) were obtained via the National Institutes of Health Toolbox cognition battery.

RESULTS

Children included in this analysis (n = 137) were 4.6 years old, on average. Per 1-SD unit, fasting glucose (B = -2.0, 95% CI -3.5, -0.5), insulin (B = -1.7, 95% CI -3.0, -0.4), and HOMA-IR values (B = -1.8, 95% CI -3.1, -0.5) were each significantly and inversely associated with inhibitory control (P < .05 for all, respectively). Fasting glucose levels were also inversely associated with cognitive flexibility (B = -2.0, 95% CI -3.7, -0.2, P = .03).

CONCLUSIONS

Our data suggest that metabolic health may impact fluid cognitive function in healthy, young children.

Combined environmental and social exposures during pregnancy and associations with neonatal size and body composition: the Healthy Start study

BACKGROUND

Prenatal environmental and social exposures have been associated with decreased birth weight. However, the effects of combined exposures in these domains are not fully understood. Here we assessed multi-domain exposures for participants in the Healthy Start study (Denver, CO) and tested associations with neonatal size and body composition.

METHODS

In separate linear regression models, we tested associations between neonatal outcomes and three indices for exposures. Two indices were developed to describe exposures to environmental hazards (ENV) and social determinants of health (SOC). A third index combined exposures in both domains (CE = ENV/10 × SOC/10). Index scores were assigned to mothers based on address at enrollment. Birth weight and length were measured at delivery, and weight-for-length z-scores were calculated using a reference distribution. Percent fat mass was obtained by air displacement plethysmography.

RESULTS

Complete data were available for 897 (64%) participants. Median (range) ENV, SOC, and CE values were 31.9 (7.1-63.2), 36.0 (2.8-75.0), and 10.9 (0.4-45.7), respectively. After adjusting for potential confounders, 10-point increases in SOC and CE were associated with 27.7 g (95%CI: 12.4 - 42.9 g) and 56.3 g (19.4 - 93.2 g) decreases in birth weight, respectively. SOC and CE were also associated with decreases in % fat mass.

CONCLUSIONS

Combined exposures during pregnancy were associated with lower birth weight and % fat mass. Evidence of a potential synergistic effect between ENV and SOC suggests a need to more fully consider neighborhood exposures when assessing neonatal outcomes.

Infant Feeding Practices In a Diverse Group of Women: The Healthy Start Study

Background

To describe infant feeding practices among a diverse group of mother-offspring pairs and identify factors associated with adherence to the American Academy of Pediatrics (AAP) recommendations.

Methods

Data were analyzed from 835 mother-offspring dyads in The Healthy Start Study, an ongoing longitudinal prebirth cohort in Denver, Colorado. Maternal report of infant feeding practices was obtained at 4 to 6 months and 18 to 24 months postnatally. Practices were classified according to the following AAP recommendations: exclusive breastfeeding for first 6 months, continued breastfeeding through 12 months, and introduction of solid foods around 6 months of age. Participants who met all 3 recommendations were categorized as "adherent." All others were categorized as "not adherent."

Results

About 77% of dyads did not adhere fully to the AAP recommendations. Women who worked ⩾35 hours/week or had a higher prepregnancy body mass index were more likely to be nonadherent. Women who were older, college educated, or had offspring with greater weight for gestational age at birth were less likely to be nonadherent.

Conclusions

Most of the women in a large contemporary cohort are not adhering to AAP infant feeding recommendations. Our results highlight the specific subgroups of women who may need additional support to optimize infant feeding practices.

Proinflammatory Diets during Pregnancy and Neonatal Adiposity in the Healthy Start Study

OBJECTIVE

To evaluate the association between dietary inflammatory index (DII) scores during pregnancy and neonatal adiposity.

STUDY DESIGN

The analysis included 1078 mother-neonate pairs in Healthy Start, a prospective prebirth cohort. Diet was assessed using repeated 24-hour dietary recalls. DII scores were obtained by summing nutrient intakes, which were standardized to global means and multiplied by inflammatory effect scores. Air displacement plethysmography measured fat mass and fat-free mass within 72 hours of birth. Linear and logistic models evaluated the associations of DII scores with birth weight, fat mass, fat-free mass, and percent fat mass, and with categorical outcomes of small- and large-for-gestational age. We tested for interactions with prepregnancy BMI and gestational weight gain.

RESULTS

The interaction between prepregnancy BMI and DII was statistically significant for birth weight, neonatal fat mass, and neonatal percent fat mass. Among neonates born to obese women, each 1-unit increase in DII was associated with increased birth weight (53 g; 95% CI, 20, 87), fat mass (20 g; 95% CI, 7-33), and percent fat mass (0.5%; 95% CI, 0.2-0.8). No interaction was detected for small- and large-for-gestational age. Each 1-unit increase in DII score was associated a 40% increase in odds of a large-for-gestational age neonate (1.4; 95% CI, 1.0-2.0; P = .04), but not a small-for-gestational age neonate (1.0; 95% CI, 0.8-1.2; P = .80). There was no evidence of an interaction with gestational weight gain.

CONCLUSIONS

Our findings support the hypothesis that an increased inflammatory milieu during pregnancy may be a risk factor for neonatal adiposity.

TRIAL REGISTRATION

Clinicaltrials.gov: NCT02273297.

Infant Adiposity is Independently Associated with a Maternal High Fat Diet but not Related to Niacin Intake: The Healthy Start Study

Objectives Over-nutrition during pregnancy resulting from maternal obesity or an unhealthy diet can lead to excess infant adiposity at birth. Specific dietary macro- and micronutrients have been shown to increase fat cell development in both in-vitro and in-vivo models and may therefore link maternal diet to increased infant adiposity. We hypothesized that high maternal dietary niacin intake during pregnancy, especially in combination with a high-fat diet (HFD) would increase infant adiposity. Methods We included 1040 participants from a pre-birth cohort of mother-infant pairs. Maternal diet was assessed using multiple 24-hour dietary recalls. HFD was defined as ≥30% of calories from fat and ≥12% of fat calories from saturated fat. Neonatal body composition (% fat mass [%FM], fat mass [FM], fat-free mass [FFM]) was measured by PEAPOD. We used multivariate regression to assess the joint effect of maternal dietary niacin and maternal HFD on neonatal body composition. Results Dietary niacin was not associated with neonatal body composition, and maternal HFD did not modify this finding. However, maternal HFD was independently associated with %FM (β = 0.8 [0.1, 1.4]%, p < 0.01] and FM (β = 32.4 [6.7, 58.0] g, p < 0.01). Conclusions for Practice Our results suggest that a HFD during pregnancy may increase infant adiposity, therefore supporting the need for improved diet counseling of pregnant women at both the clinical and community levels.

Fetal overnutrition and offspring insulin resistance and β-cell function: the Exploring Perinatal Outcomes among Children (EPOCH) study

AIMS

To examine the associations of intrauterine exposure to maternal diabetes and obesity with offspring insulin resistance, β-cell function and oral disposition index in a longitudinal observational study of ethnically diverse offspring.

METHODS

A total of 445 offspring who were exposed (n=81) or not exposed (n=364) to maternal diabetes in utero completed two fasting blood measurements at mean (sd) ages of 10.5 (1.5) and 16.5 (1.2) years, respectively, and an oral glucose tolerance test at the second visit. We used linear mixed models and general linear univariate models to evaluate the associations of maternal diabetes and pre-pregnancy BMI with offspring outcomes.

RESULTS

Maternal diabetes in utero predicted increased insulin resistance [18% higher updated homeostatic model assessment of insulin resistance (HOMA2-IR), P=0.01; 19% lower Matsuda index, P=0.01 and 9% greater updated homeostatic model assessment of β-cell function (HOMA2-β), P=0.04]. Each 5-kg/m² increase in pre-pregnancy BMI predicted increased insulin resistance (11% greater HOMA2-IR, P<0.001; 10% lower Matsuda index, P<0.001; 6% greater HOMA2-β, P<0.001). Similar results were obtained in a combined model with both exposures. After adjustment for offspring BMI, only maternal diabetes was associated with higher HOMA2-IR (β=1.12, P=0.03) and lower Matsuda index (β=0.83, P=0.01). Neither exposure was associated with early insulin response or oral disposition index.

CONCLUSIONS

Intrauterine exposure to diabetes or obesity is associated with greater offspring insulin resistance than non-exposure, supporting the hypothesis that fetal overnutrition results in metabolic abnormalities during childhood and adolescence.

Exposure to secondhand smoke, exclusive breastfeeding and infant adiposity at age 5 months in the Healthy Start study

BACKGROUND

Infant adiposity may be influenced by several environmental risk factors, but few studies have explored these interactions.

OBJECTIVE

To examine the interaction between exposure to secondhand smoke and breastfeeding exclusivity on adiposity at age 5 months.

METHODS

We studied 813 mother-offspring pairs from the longitudinal Healthy Start study. Fat mass and fat-free mass were measured by air displacement plethysmography. Linear regression analyses were used to estimate the association between household smokers (none, any) with fat mass, fat-free mass, percent fat mass, weight-for-age z-score, weight-for-length z-score and BMI-for-age z-score as separate outcomes. Interaction terms between household smokers and breastfeeding exclusivity (<5 months, ≥5 months) were added to separate models.

RESULTS

The combination of exposure to secondhand smoke and a lack of exclusive breastfeeding was associated with increased adiposity at age 5 months. For example, within the not exclusively breastfed strata, exposure to secondhand smoke was associated with increased fat mass (0.1 kg; 95% CI: 0.0-0.2; P = 0.05). Conversely, within the exclusively breastfed strata, there was virtually no difference in fat mass between exposed and non-exposed infants (coefficient: -0.1; 95% CI: -0.3-0.1; P = 0.25).

CONCLUSIONS

Our findings may inform new public health strategies with potential relevance for both smoking cessation and obesity prevention.

An observational cohort study of weight- and length-derived anthropometric indicators with body composition at birth and 5 mo: the Healthy Start study

Background: Despite widespread use of weight- and length-based anthropometric indexes as proxies for adiposity, little is known regarding the extent to which they correspond with fat mass (FM) or fat-free mass (FFM) during infancy.Objective: This study aimed to examine associations of 3 derived indicators-weight-for-age z score (WFAZ), weight-for-length score (WFLZ), and body mass index z score (BMIZ)-with FM, percentage of FM, and FFM measured by air-displacement plethysmography during the first 5 mo of life.Design: Applying prospectively collected data from 1027 infants in a Colorado prebirth cohort, we used multivariate regression to evaluate associations between the derived indicators and body composition at birth and at 5 mo, and with change (Δ) during follow-up.Results: At birth, all 3 derived indicators were more strongly associated with FFM than with FM. Each unit of WFAZ corresponded with 0.342 kg FFM (95% CI: 0.331, 0.351 kg FFM), compared with 0.121 kg FM (95% CI: 0.114, 0.128 kg FM) (P < 0.0001); similar trends were observed for WFLZ and BMIZ. By 5 mo, WFLZ and BMIZ were more strongly associated with FM than with FFM, whereas WFAZ correlated similarly with the 2 components of body composition. ΔWFLZ and ΔBMIZ were both more strongly related to ΔFM than to ΔFFM; however, a direct comparison of the 2 indexes with respect to change in the percentage of FM indicated that ΔBMIZ was the optimal proxy of adiposity gain (P < 0.0001, pairwise difference).Conclusions: Weight- and length-based indexes are poor surrogates for newborn adiposity. However, at 5 mo, WFLZ and BMIZ are suitable proxies of FM. When assessing adiposity gain, ΔBMIZ is the best indicator of fat accrual during the first 5 postnatal months. This trial was registered at clinicaltrials.gov as NCT02273297.

Prenatal Vitamin D Intake, Cord Blood 25-Hydroxyvitamin D, and Offspring Body Composition: The Healthy Start Study

Vitamin D deficiency in pregnancy may be associated with increased offspring adiposity, but evidence from human studies is inconclusive. We examined associations between prenatal vitamin D intake, 25-hydroxyvitamin D (25(OH)D) in cord blood, and offspring size and body composition at birth and 5 months. Participants included 605 mother-offspring dyads from the Healthy Start study, an ongoing, pre-birth prospective cohort study in Denver, Colorado, USA. Prenatal vitamin D intake was assessed with diet recalls and questionnaires, and offspring body composition was measured via air displacement plethysmography at birth and 5 months. General linear univariate models were used for analysis, adjusting for maternal age, race/ethnicity, pre-pregnancy body mass index (BMI), offspring sex, and gestational age at birth. Non-Hispanic white race, lower pre-pregnancy BMI, higher prenatal vitamin D intake, and summer births were associated with higher cord blood 25(OH)D. Higher 25(OH)D was associated with lower birthweight (β = -6.22, p = 0.02), but as maternal BMI increased, this association became increasingly positive in direction and magnitude (β = 1.05, p = 0.04). Higher 25(OH)D was also associated with lower neonatal adiposity (β = -0.02, p < 0.05) but not after adjustment for maternal BMI (β = -0.01, p = 0.25). Cord blood 25(OH)D was not associated with offspring size or body composition at 5 months. Our data confirm the hypothesis that vitamin D exposure in early life is associated with neonatal body size and composition. Future research is needed to understand the implications of these associations as infants grow.

Exposure to maternal diabetes in utero and offspring eating behavior: The EPOCH study

The risk of becoming overweight among offspring exposed to gestational diabetes (GDM) in utero is two-fold higher than in the general population. The responsible mechanisms are likely multifactorial, with some evidence that GDM exposure alters brain satiety signaling, which may impact eating behavior. To better understand these effects, we investigated the relationship between GDM exposure, eating behavior, and total energy intake in 268 adolescents from the Exploring Perinatal Outcomes among Children cohort, who were exposed (n = 50) or not exposed (n = 217) to GDM in utero. Eating behavior was measured by the Eating in the Absence of Hunger in Children and Adolescents (EAH-C) questionnaire, which included subscale scores for Negative Affect, External Stimuli, and Fatigue/Boredom. Total energy intake (kcal/day) was derived from the Block Kid's Food Questionnaire. The associations between GDM exposure and the outcomes of total score and each EAH-C subscale were evaluated in separate multivariable models. In addition to the main predictor, GDM, the models included a GDM-by-sex interaction term and were adjusted for important covariates. The associations between EAH-C total and subscale scores and the outcome of total energy intake were also tested in separate multivariable models. Female offspring exposed to GDM in utero (vs unexposed males and females) were more likely to continue eating beyond satiation due to feelings of boredom and fatigue (β = 0.47, 95% CI: 0.11, 0.83), and in general (EAH-C total score; β = 4.20, 95% CI: 0.56, 7.86) compared to unexposed males. All EAH-C subscale and total scores were significantly, positively associated with higher energy intake (p < 0.05 for all, respectively). Our findings highlight the need for further investigation into the possible early life programming of eating behaviors by GDM exposure in utero.

Predictors of Infant Body Composition at 5 Months of Age: The Healthy Start Study

OBJECTIVE

To examine associations of demographic, perinatal, and infant feeding characteristics with offspring body composition at approximately 5 months of age.

STUDY DESIGN

We collected data on 640 mother/offspring pairs from early pregnancy through approximately 5 months of age. We assessed offspring body composition with air displacement plethysmography at birth and approximately 5 months of age. Linear regression analyses examined associations between predictors and fat-free mass, fat mass, and percent fat mass (adiposity) at approximately 5 months. Secondary models further adjusted for body composition at birth and rapid infant growth.

RESULTS

Greater prepregnant body mass index and gestational weight gain were associated with greater fat-free mass at approximately 5 months of age, but not after adjustment for fat-free mass at birth. Greater gestational weight gain was also associated with greater fat mass at approximately 5 months of age, independent of fat mass at birth and rapid infant growth, although this did not translate into increased adiposity. Greater percent time of exclusive breastfeeding was associated with lower fat-free mass (-311 g; P < .001), greater fat mass (+224 g; P < .001), and greater adiposity (+3.51%; P < .001). Compared with offspring of non-Hispanic white mothers, offspring of Hispanic mothers had greater adiposity (+2.72%; P < .001) and offspring of non-Hispanic black mothers had lower adiposity (-1.93%; P < .001). Greater adiposity at birth predicted greater adiposity at approximately 5 months of age, independent of infant feeding and rapid infant growth.

CONCLUSIONS

There are clear differences in infant body composition by demographic, perinatal, and infant feeding characteristics, although our data also show that increased adiposity at birth persists through approximately 5 months of age. Our findings warrant further research into implications of differences in infant body composition.

Exploring the association between maternal prenatal multivitamin use and early infant growth: The Healthy Start Study

BACKGROUND

Prenatal multivitamin supplementation is recommended to improve offspring outcomes, but effects on early infant growth are unknown.

OBJECTIVES

We examined whether multivitamin supplementation in the year before delivery predicts offspring mass, body composition and early infant growth.

METHODS

Multivitamin use was assessed longitudinally in 626 women from the Healthy Start Study. Offspring body size and composition was measured with air displacement plethysmography at birth (<3 days) and postnatally (median 5.2 months). Separate multiple linear regressions assessed the relationship of weeks of daily multivitamin use with offspring mass, body composition and postnatal growth, after adjustment for potential confounders (maternal age, race, pre-pregnant body mass index; offspring gestational age at birth, sex; breastfeeding exclusivity).

RESULTS

Maternal multivitamin use was not related to offspring mass or body composition at birth, or rate of change in total or fat-free mass in the first 5 months. Multivitamin use was inversely associated with average monthly growth in offspring percent fat mass (β = -0.009, p = 0.049) between birth and postnatal exam. Offspring of non-users had a monthly increase in percent fat mass of 3.45%, while offspring at the top quartile of multivitamin users had a monthly increase in percent fat mass of 3.06%. This association was not modified by exclusive breastfeeding.

CONCLUSIONS

Increased multivitamin use in the pre-conception and prenatal periods was associated with a slower rate of growth in offspring percent fat mass in the first 5 months of life. This study provides further evidence that in utero nutrient exposures may affect offspring adiposity beyond birth.

Maternal diet quality in pregnancy and neonatal adiposity: the Healthy Start Study

BACKGROUND/OBJECTIVES

Poor maternal diet in pregnancy can influence fetal growth and development. We tested the hypothesis that poor maternal diet quality during pregnancy would increase neonatal adiposity (percent fat mass (%FM)) at birth by increasing the fat mass (FM) component of neonatal body composition.

METHODS

Our analysis was conducted using a prebirth observational cohort of 1079 mother-offspring pairs. Pregnancy diet was assessed via repeated Automated Self-Administered 24-h dietary recalls, from which Healthy Eating Index-2010 (HEI-2010) scores were calculated for each mother. HEI-2010 was dichotomized into scores of ⩽57 and >57, with low scores representing poorer diet quality. Neonatal %FM was assessed within 72 h after birth with air displacement plethysmography. Using univariate and multivariate linear models, we analyzed the relationship between maternal diet quality and neonatal %FM, FM, and fat-free mass (FFM) while adjusting for prepregnancy body mass index (BMI), physical activity, maternal age, smoking, energy intake, preeclampsia, hypertension, infant sex and gestational age.

RESULTS

Total HEI-2010 score ranged between 18.2 and 89.5 (mean: 54.2, s.d.: 13.6). An HEI-2010 score of ⩽57 was significantly associated with higher neonatal %FM (β=0.58, 95% confidence interval (CI) 0.07-1.1, P<0.05) and FM (β=20.74; 95% CI 1.49-40.0; P<0.05) but no difference in FFM.

CONCLUSIONS

Poor diet quality during pregnancy increases neonatal adiposity independent of maternal prepregnancy BMI and total caloric intake. This further implicates maternal diet as a potentially important exposure for fetal adiposity.

Diet, physical activity and mental health status are associated with dysglycaemia in pregnancy: the Healthy Start Study

AIMS

To examine the association between dysglycaemia and multiple modifiable factors measured during pregnancy.

METHODS

The Healthy Start Study collected self-reported data on modifiable factors in early and mid-pregnancy (median 17 and 27 weeks gestation, respectively) from 832 women. Women received one point for each modifiable factor for which they had optimum scores: diet quality (Healthy Eating Index score ≥64), physical activity level (estimated energy expenditure ≥170 metabolic equivalent task-h/week), and mental health status (Perceived Stress Scale score <6 and Edinburgh Postnatal Depression Scale score <13). Dysglycaemia during pregnancy was defined as an abnormal glucose challenge result, ≥1 abnormal results on an oral glucose tolerance test, or a clinical diagnosis of gestational diabetes. Logistic regression models estimated odds ratios for dysglycaemia as a function of each factor and the total score, adjusted for age, race/ethnicity, pre-pregnancy BMI, history of gestational diabetes, and family history of Type 2 diabetes.

RESULTS

In individual analyses, only physical activity was significantly associated with a reduced risk of dysglycaemia (adjusted odds ratio 0.67, 95% CI 0.44-1.00). We observed a significant, dose-response association between increasing numbers of optimal factors and odds of dysglycaemia (adjusted P=0.01). Compared with having no optimal modifiable factors, having all three was associated with a 73% reduced risk of dysglycaemia (adjusted odds ratio 0.27, 95% CI 0.08-0.95).

CONCLUSIONS

An increasing number of positive modifiable factors in pregnancy was associated with a dose-response reduction in risk of dysglycaemia. Our results support the hypothesis that modifiable factors in pregnancy are associated with the risk of prenatal dysglycaemia.

Multivariate test power approximations for balanced linear mixed models in studies with missing data

Multilevel and longitudinal studies are frequently subject to missing data. For example, biomarker studies for oral cancer may involve multiple assays for each participant. Assays may fail, resulting in missing data values that can be assumed to be missing completely at random. Catellier and Muller proposed a data analytic technique to account for data missing at random in multilevel and longitudinal studies. They suggested modifying the degrees of freedom for both the Hotelling-Lawley trace F statistic and its null case reference distribution. We propose parallel adjustments to approximate power for this multivariate test in studies with missing data. The power approximations use a modified non-central F statistic, which is a function of (i) the expected number of complete cases, (ii) the expected number of non-missing pairs of responses, or (iii) the trimmed sample size, which is the planned sample size reduced by the anticipated proportion of missing data. The accuracy of the method is assessed by comparing the theoretical results to the Monte Carlo simulated power for the Catellier and Muller multivariate test. Over all experimental conditions, the closest approximation to the empirical power of the Catellier and Muller multivariate test is obtained by adjusting power calculations with the expected number of complete cases. The utility of the method is demonstrated with a multivariate power analysis for a hypothetical oral cancer biomarkers study. We describe how to implement the method using standard, commercially available software products and give example code. Copyright © 2015 John Wiley & Sons, Ltd.

An internal pilot design for prospective cancer screening trials with unknown disease prevalence

BACKGROUND

For studies that compare the diagnostic accuracy of two screening tests, the sample size depends on the prevalence of disease in the study population, and on the variance of the outcome. Both parameters may be unknown during the design stage, which makes finding an accurate sample size difficult.

METHODS

To solve this problem, we propose adapting an internal pilot design. In this adapted design, researchers will accrue some percentage of the planned sample size, then estimate both the disease prevalence and the variances of the screening tests. The updated estimates of the disease prevalence and variance are used to conduct a more accurate power and sample size calculation.

RESULTS

We demonstrate that in large samples, the adapted internal pilot design produces no Type I inflation. For small samples (N less than 50), we introduce a novel adjustment of the critical value to control the Type I error rate. We apply the method to two proposed prospective cancer screening studies: 1) a small oral cancer screening study in individuals with Fanconi anemia and 2) a large oral cancer screening trial.

CONCLUSION

Conducting an internal pilot study without adjusting the critical value can cause Type I error rate inflation in small samples, but not in large samples. An internal pilot approach usually achieves goal power and, for most studies with sample size greater than 50, requires no Type I error correction. Further, we have provided a flexible and accurate approach to bound Type I error below a goal level for studies with small sample size.

Reducing decision errors in the paired comparison of the diagnostic accuracy of screening tests with Gaussian outcomes

Background

Scientists often use a paired comparison of the areas under the receiver operating characteristic curves to decide which continuous cancer screening test has the best diagnostic accuracy. In the paired design, all participants are screened with both tests. Participants with suspicious results or signs and symptoms of disease receive the reference standard test. The remaining participants are classified as non-cases, even though some may have occult disease. The standard analysis includes all study participants, which can create bias in the estimates of diagnostic accuracy since not all participants receive disease status verification. We propose a weighted maximum likelihood bias correction method to reduce decision errors.

Methods

Using Monte Carlo simulations, we assessed the method’s ability to reduce decision errors across a range of disease prevalences, correlations between screening test scores, rates of interval cases and proportions of participants who received the reference standard test.

Results

The performance of the method depends on characteristics of the screening tests and the disease and on the percentage of participants who receive the reference standard test. In studies with a large amount of bias in the difference in the full areas under the curves, the bias correction method reduces the Type I error rate and improves power for the correct decision. We demonstrate the method with an application to a hypothetical oral cancer screening study.

Conclusion

The bias correction method reduces decision errors for some paired screening trials. In order to determine if bias correction is needed for a specific screening trial, we recommend the investigator conduct a simulation study using our software.

Improving the Diagnostic Accuracy of a Stratified Screening Strategy by Identifying the Optimal Risk Cutoff

We give a novel decision-theoretic approach for finding the optimal risk cutoff for additional breast cancer screening for women who have a high model-predicted risk of breast cancer. Methods: The American Cancer Society (ACS) suggests a stratified screening strategy for breast cancer. The strategy includes assessing women's risk of breast cancer using a risk model, screening women at high risk of breast cancer with both contrast-enhanced breast MRI and mammography, while screening women at low risk of breast cancer with mammography alone. The ACS used expert consensus opinion to choose the risk cutoff for additional screening. Instead, we suggest a risk cutoff chosen to maximize the full area under the receiver operating characteristic curve for the strategy. We use three inputs to find the cutoff, including: 1) the distribution of five-year breast cancer risk scores, 2) the probability of breast cancer given the risk score, and 3) the diagnostic accuracies of digital mammography, and contrast-enhanced screening breast MRI. No data were publicly available for testing the performance of stratified screening strategies based on the ACS suggested risk models. Instead, we used publicly available Breast Cancer Screening Consortium (BCSC) risk data to seek an optimal risk cutoff. A previous study suggests that the BCSC risk model has similar predictive accuracy as the Tyrer-Cuzick and Claus models. For comparison, we used a hypothetical risk model with much better predictive accuracy than the BCSC model. Results: For the BCSC risk model, the strategy with the highest diagnostic accuracy for the entire population is to screen almost all women with both digital mammography and contrast- enhanced breast MRI. Under the strategy based on the hypothetical risk model, only women at 18% or higher five-year risk would receive dual modality screening. Conclusion: The results occur because the BCSC risk model does not accurately predict which women will or will not develop breast cancer. However, if there were a more accurate risk model, a stratified screening strategy would offer benefits for the entire population by balancing the harms of breast cancer screening with the potential for improved breast cancer detection.

: Online Power Computation for Linear Models with and without a Baseline Covariate

GLIMMPSE is a free, web-based software tool that calculates power and sample size for the general linear multivariate model with Gaussian errors (http://glimmpse.SampleSizeShop.org/). GLIMMPSE provides a user-friendly interface for the computation of power and sample size. We consider models with fixed predictors, and models with fixed predictors and a single Gaussian covariate. Validation experiments demonstrate that GLIMMPSE matches the accuracy of previously published results, and performs well against simulations. We provide several online tutorials based on research in head and neck cancer. The tutorials demonstrate the use of GLIMMPSE to calculate power and sample size.

Abstract A06: Reducing decision errors in the paired comparison of the diagnostic accuracy of continuous screening tests

Scientists often use a paired comparison of the areas under the receiver operating characteristic curves to decide which continuous cancer screening test has the best diagnostic accuracy. In the paired design, all participants are screened with both tests. Participants with unremarkable screening results enter a follow-up period. Participants with suspicious screening results and those who show evidence of disease during follow-up receive the gold standard test. The remaining participants are classified as non-cases, even though some may have occult disease. The standard analysis includes all study participants in the analysis, which can create bias in the estimates of diagnostic accuracy. If the bias affects the area under the curve for one screening test more than the other screening test, scientists may make the wrong decision as to which screening test has better diagnostic accuracy. We propose a weighted maximum likelihood bias-correction method to reduce decision errors. The amount of bias in the study and the performance of the bias correction method depend on characteristics of the screening tests and the disease, and on the percentage of study participants who receive the gold standard test. In order to determine if bias correction is needed for a specific screening trial, we recommend the investigator conduct a simulation study using our software. We demonstrate the proposed method with an application to a hypothetical oral cancer screening study.

Citation Format: Brandy M. Ringham, Todd A. Alonzo, John T. Brinton, Keith E. Muller, Deborah H. Glueck. Reducing decision errors in the paired comparison of the diagnostic accuracy of continuous screening tests. [abstract]. In: Proceedings of the Fifth AACR Conference on the Science of Cancer Health Disparities in Racial/Ethnic Minorities and the Medically Underserved; 2012 Oct 27-30; San Diego, CA. Philadelphia (PA): AACR; Cancer Epidemiol Biomarkers Prev 2012;21(10 Suppl):Abstract nr A06.

Bias in estimating accuracy of a binary screening test with differential disease verification

Sensitivity, specificity, positive and negative predictive value are typically used to quantify the accuracy of a binary screening test. In some studies, it may not be ethical or feasible to obtain definitive disease ascertainment for all subjects using a gold standard test. When a gold standard test cannot be used, an imperfect reference test that is less than 100 per cent sensitive and specific may be used instead. In breast cancer screening, for example, follow-up for cancer diagnosis is used as an imperfect reference test for women where it is not possible to obtain gold standard results. This incomplete ascertainment of true disease, or differential disease verification, can result in biased estimates of accuracy. In this paper, we derive the apparent accuracy values for studies subject to differential verification. We determine how the bias is affected by the accuracy of the imperfect reference test, the percent who receive the imperfect reference standard test not receiving the gold standard, the prevalence of the disease, and the correlation between the results for the screening test and the imperfect reference test. It is shown that designs with differential disease verification can yield biased estimates of accuracy. Estimates of sensitivity in cancer screening trials may be substantially biased. However, careful design decisions, including selection of the imperfect reference test, can help to minimize bias. A hypothetical breast cancer screening study is used to illustrate the problem.

Estimates of sensitivity and specificity can be biased when reporting the results of the second test in a screening trial conducted in series

Background

Cancer screening reduces cancer mortality when early detection allows successful treatment of otherwise fatal disease. There are a variety of trial designs used to find the best screening test. In a series screening trial design, the decision to conduct the second test is based on the results of the first test. Thus, the estimates of diagnostic accuracy for the second test are conditional, and may differ from unconditional estimates. The problem is further complicated when some cases are misclassified as non-cases due to incomplete disease status ascertainment.

Methods

For a series design, we assume that the second screening test is conducted only if the first test had negative results. We derive formulae for the conditional sensitivity and specificity of the second test in the presence of differential verification bias. For comparison, we also derive formulae for the sensitivity and specificity for a single test design, both with and without differential verification bias.

Results

Both the series design and differential verification bias have strong effects on estimates of sensitivity and specificity. In both the single test and series designs, differential verification bias inflates estimates of sensitivity and specificity. In general, for the series design, the inflation is smaller than that observed for a single test design.

The degree of bias depends on disease prevalence, the proportion of misclassified cases, and on the correlation between the test results for cases. As disease prevalence increases, the observed conditional sensitivity is unaffected. However, there is an increasing upward bias in observed conditional specificity. As the proportion of correctly classified cases increases, the upward bias in observed conditional sensitivity and specificity decreases. As the agreement between the two screening tests becomes stronger, the upward bias in observed conditional sensitivity decreases, while the specificity bias increases.

Conclusions

In a series design, estimates of sensitivity and specificity for the second test are conditional estimates. These estimates must always be described in context of the design of the trial, and the study population, to prevent misleading comparisons. In addition, these estimates may be biased by incomplete disease status ascertainment.

Dual-energy X-ray absorptiometry modeling to explain the increased resting energy expenditure associated with the HIV lipoatrophy syndrome

BACKGROUND

The HIV lipoatrophy syndrome is characterized by loss of subcutaneous fat and is associated with increased resting energy expenditure (REE). Recently, dual-energy X-ray absorptiometry (DXA) modeling of organ-tissue mass combined with specific organ-tissue metabolic rates has been used to gain further insight into the relation of the lean body mass to REE and to better understand differences in REE between groups.

OBJECTIVE

This study examined the organ-tissue basis of the increased REE shown in HIV lipoatrophy.

DESIGN

REE was measured in 29 HIV-infected patients with lipoatrophy and in 29 HIV-infected and 19 healthy control subjects. Five organ-tissue mass components (brain, bone, skeletal muscle, adipose tissue, and residual mass) were calculated with the use of DXA modeling and body weight.

RESULTS

DXA modeling showed no significant differences in predicted REE between the 3 groups. However, measured REE was significantly greater in subjects with lipoatrophy than in control subjects. Measured REE remained significantly greater in lipoatrophy subjects after routine adjustment for lean body mass and after adjustment for each organ-tissue mass component. Finally, DXA and regression modeling of REE suggests that increased energy expenditure in skeletal muscle may account for the resting hypermetabolism of patients with HIV lipoatrophy.

CONCLUSIONS

Increased REE in subjects with HIV lipoatrophy cannot be explained by differences in organ-tissue mass as modeled by DXA. Instead, DXA and regression modeling of REE suggests that skeletal muscle is hypermetabolic in patients with HIV lipoatrophy. This may be a form of adaptive thermogenesis in response to an inability to store triglyceride fuel in a normal manner.

Bias in trials comparing paired continuous tests can cause researchers to choose the wrong screening modality

Background

To compare the diagnostic accuracy of two continuous screening tests, a common approach is to test the difference between the areas under the receiver operating characteristic (ROC) curves. After study participants are screened with both screening tests, the disease status is determined as accurately as possible, either by an invasive, sensitive and specific secondary test, or by a less invasive, but less sensitive approach. For most participants, disease status is approximated through the less sensitive approach. The invasive test must be limited to the fraction of the participants whose results on either or both screening tests exceed a threshold of suspicion, or who develop signs and symptoms of the disease after the initial screening tests.

The limitations of this study design lead to a bias in the ROC curves we call paired screening trial bias. This bias reflects the synergistic effects of inappropriate reference standard bias, differential verification bias, and partial verification bias. The absence of a gold reference standard leads to inappropriate reference standard bias. When different reference standards are used to ascertain disease status, it creates differential verification bias. When only suspicious screening test scores trigger a sensitive and specific secondary test, the result is a form of partial verification bias.

Methods

For paired screening tests with bivariate normally distributed scores, we give formulae and programs to quantify the effect of paired screening trial bias on a paired comparison of area under the curves. We fix the prevalence of disease, and the chance a diseased subject manifests signs and symptoms. We derive the formulas for true sensitivity and specificity, and those for the sensitivity and specificity observed by the study investigator.

Results

The observed area under the ROC curves is quite different from the true area under the ROC curves. The typical direction of the bias is a strong inflation in sensitivity, paired with a concomitant slight deflation of specificity.

Conclusion

In paired trials of screening tests, when area under the ROC curve is used as the metric, bias may lead researchers to make the wrong decision as to which screening test is better.