Available online 28 April 2023

Author links open overlay panel, , AbstractPurpose

A strength of SuperLearner is that it may accommodate key interactions between model variables without a priori specification. In prior research, protective associations between fruit intake and preeclampsia were stronger when estimated using SuperLearner with targeted maximum likelihood estimation (TMLE) compared with multivariable logistic regression without any interaction terms. We explored whether heterogeneity (i.e., differences in the effect estimate due to interactions between fruit intake and covariates) may partly explain differences in estimates from these two models.

Methods

Using a US prospective pregnancy cohort (2010-2013, n=7781), we estimated preeclampsia risk differences (RDs) for higher versus lower fruit density using multivariable logistic regression and included 2-way statistical interactions between fruit density and each of the 25 model covariates. We compared the RDs with those from SuperLearner with TMLE (gold standard) and logistic regression with no interaction.

Results

From the logistic regression models with 2-way statistical interactions, 48% of the preeclampsia RDs were ≤-0.02 (closer to SuperLearner with TMLE estimate); 40% equaled -0.01 (same as logistic regression with no interaction estimate); the minority were at or crossed the null.

Conclusions

Our exploratory analysis provided preliminary evidence that heterogeneity may partly explain differences in estimates from logistic regression versus SuperLearner with TMLE.

Section snippetsINTRODUCTION

Interactions among the bioactive compounds in food may partly explain the benefits of whole-diet interventions, like the Mediterranean diet and Dietary Approaches to Stop Hypertension (DASH) [1], [2], [3], [4]. Fruits and vegetables, for instance, contain antioxidants with antibacterial and antiviral effects to counteract damage (e.g., aging, chronic diseases) from oxidative stress [5], [6], [7], and concurrent intake of dietary fiber and certain fats may enhance antioxidant bioavailability [8]

MATERIAL AND METHODS

Data were from the Nulliparous Pregnancy Outcomes Study: monitoring mothers-to-be, a prospective cohort in 8 US medical centers (2010–2013) that has been described previously [17]. In brief, ~10,000 pregnant individuals were enrolled during their first trimester. Individuals were eligible if they had a viable singleton gestation, ≤3 prior miscarriages, and no prior pregnancy lasting ≥20 weeks’ gestation. Medical chart abstraction ascertained medical history and pregnancy details. Participants

RESULTS

Participants with higher fruit density tended to be older, have higher educational attainment and physical activity level, and were more likely to have private health insurance and a planned pregnancy (eTable 2). Individuals with higher fruit density also tended to have higher dietary densities for total vegetables, seafood and plant proteins, and whole grains and lower densities for refined grains and empty calories.

We replicated the previous findings [15] that demonstrated protective

DISCUSSION

Our results provide preliminary support that dietary heterogeneity may partly explain the differences in the associations between fruit density and preeclampsia risk estimated from logistic regression versus TMLE. When 2-way statistical interaction terms were added to logistic regression, almost half of the models yielded RD point estimates that were more protective than those obtained from the model with no interaction terms, and half of these models included an interaction with another

CONCLUSIONS

A logical next step in this research is to identify which dietary and participant characteristics interact to result in differential effects of a given dietary exposure across groups of individuals. A potentially useful method is causal forests [30], which apply random forests to identify and quantify which covariates explain the largest degree of effect heterogeneity. Knowledge of important interactions can be used to further epidemiologic discovery and develop clinical dietary guidelines and

Sources of Funding

The results reported herein correspond to specific aims of grant R01 HD102313 to Bodnar LM and Naimi AI from the Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD). This work was also supported by grants U10 HD063036 to RTI International; U10 HD063072 to Case Western Reserve University; U10 HD063047 to Columbia University; U10 HD063037 to Indiana University; U10 HD063041 to University of Pittsburgh; U10 HD063020 to Northwestern University; U10 HD063046 to

CRediT authorship contribution statement

All authors made substantial contributions to all of the following: (1) the conception and design of the study, or acquisition of data, or analysis and interpretation of data, (2) drafting the article or revising it critically for important intellectual content, (3) final approval of the version to be submitted. Julie M. Petersen: Conceptualization, Methodology, Formal analysis, Writing – original draft. Ashley I. Naimi: Conceptualization, Funding acquisition, Supervision, Writing - review &

Declaration of Competing Interest

The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: Lisa Bodnar and Ashley Naimi report financial support was provided by National Institute of Child Health and Human Development.

ACKNOWLEDGEMENTS

We thank Sara Parisi for her contribution to data management, data cleaning, and derived variable creation.

The following institutions and researchers compose the Nulliparous Pregnancy Outcomes Study: Monitoring Mothers-to-be (nuMoM2b) Network: Case Western Reserve University / Ohio State University - Brian M. Mercer, MD, Jay Iams, MD, Wendy Dalton, RN, Cheryl Latimer, RN, LuAnn Polito, RN, JD; Columbia University / Christiana Care - Matthew K. Hoffman, MD, MPH, Ronald Wapner, MD, Karin Fuchs,

References (30)X. Chen et al.Interaction between Flavonoids and Carotenoids on Ameliorating Oxidative Stress and Cellular Uptake in Different Cells

Foods

(2021)

R.M. Han et al.Reaction dynamics of flavonoids and carotenoids as antioxidants

Molecules

(2012)

L.M. Bodnar et al.Use of a Doubly Robust Machine-Learning-Based Approach to Evaluate Body Mass Index as a Modifier of the Association Between Fruit and Vegetable Intake and Preeclampsia

Am J Epidemiol

(2022)

S.H. ZeiselPrecision (Personalized) Nutrition: Understanding Metabolic Heterogeneity

Annu Rev Food Sci Technol

(2020)

L.M. Bodnar et al.Machine learning can improve the development of evidence-based dietary guidelines

Public Health Nutr

(2022)

View full text

© 2023 Elsevier Inc. All rights reserved.