FILES PROVIDED. September analyses to be carried out and what is needed from each cohort. covariates, and study population in each cohort.

Transcription

1 Proposal for Join Analysis: Assessing causality in the association between maternal pre-pregnancy obesity and child neurodevelopment: observational and mendelian randomization analyses Coordinators: Maribel Casas Debbie Lawlor Martine Vrijheid Centre for Research in Environmental Epidemiology (CREAL), Barcelona MRC Integrative Epidemiology Unit and School of Social and Community Medicine, University of Bristol, Bristol CREAL, Barcelona With the collaboration of: Sílvia Alemany CREAL, Barcelona Mariona Bustamante Centre for Genomic Regulation (CRG), Barcelona Mònica Guxens CREAL, Barcelona; Erasmus University Medical Center, Rotterdam - Content 1. Background Objective Methods Design Study population Maternal and paternal BMI Child neuropsychological development Covariates Maternal BMI genetic score Statistical analysis Population sample Treatment of missing values Observational analysis Mendelian randomization analysis Organization, publications, and ethical issues Organization Publications Permissions and ethical issues What is expected from each cohort References

2 FILES PROVIDED File MatBMI & Neuro_Protocol.docx MatBMI & Neuro_Basic information.xlsx MatBMI & Neuro_Results Observational.xlsx MatBMI & Neuro_Results MR.xlsx Content This document, which explains in detail aims and analyses to be carried out and what is needed from each cohort. Description of the neuropsychological tests, covariates, and study population in each cohort. What is expected from cohorts: -Check the neuropsychological tests -Complete covariates sheet with additional information regarding the created variables -Define your study population Results report of the observational study. What is expected from cohorts: -Complete tables ONLY COHORTS THAT ARE DOING THE ANALYSIS BY THEMSELVES Results report of the mendelian randomization study. What is expected from cohorts: -Complete tables ONLY COHORTS THAT ARE DOING THE ANALYSIS BY THEMSELVES AND HAVE GENETIC DATA MatBMI & Neuro_Scripts.docx MatBMI & Neuro_Annex I.docx IMPORTANT: cohorts with genetic data available need to first perform the observational analysis for all cohort participants and then do it for those participants with genetic data. So, for this subsample complete spreadsheets: <MatBMI & Neuro Results Observational.xlsx> <MatBMI & Neuro Results MR.xlsx> Detailed scripts in STATA and R for all the analyses (under development) ONLY FOR COHORTS THAT ARE DOING THE ANALYSIS BY THEMSELVES Explanation on how to generate maternal BMI genetic score. 2

3 1. Background Obesity is increasing worldwide and in Southern Europe the prevalence can reach 30% 1. Obesity is associated with many pregnancy complications such as preeclampsia or gestational diabetes and is characterized as a systemic inflammatory condition 2,3. Chronic low-grade inflammation is an important characteristic of obesity and can be transferred to the offspring 4 resulting in cognitive and behavioural impairment 5,6. During the last decade a number of epidemiological studies have assessed whether maternal pre-pregnancy overweight and obesity can reduce neurodevelopment of the offspring, but results are still inconsistent Some studies suggest no association or associations that can be explained by shared family and socioeconomic contexts, and other studies suggest associations that are likely to reflect intrauterine mechanisms. Confounding by family and socioeconomic contexts can be partially evaluated by comparing the strength of associations of paternal vs. maternal BMI with childhood neuropsychological development: a direct maternal intrauterine mechanism would produce stronger associations than an indirect paternal effect 22. Challenges for future studies include having adequate sample sizes and comparing results in different population settings. On the other hand, inferring an intrauterine effect from conventional epidemiological analyses is difficult because obesity and neurodevelopment are strongly socio-demographically patterned. Mendelian randomization (MR), using genetic variants associated with BMI that serve as a proxy of maternal prepregnancy obesity, has been suggested as an alternative approach. Since genetic variants are assorted randomly during gamete formation and conception, they should be unrelated to confounding factors and can therefore be used to estimate causal associations free from confounding 22. In recent years, large scale genome wide association (GWAS) studies have identified several single nucleotide polymorphisms (SNPs) that are robustly associated with increased BMI 23. Therefore, if there is an intrauterine effect of maternal pre-pregnancy obesity on offspring neurodevelopment, we would expect to see an association between maternal BMI genetic score and impaired offspring neurodevelopment. 2. Objective In this study we aim to assess causality in the association between maternal pre-pregnancy body mass index (BMI) and child neurodevelopment by performing an observational analysis of birth cohorts and a MR analysis in those cohorts with genetic data available. Further, we will use paternal BMI as a negative control exposure and test the role of the child BMI genetic score (Figure 1). Remark: this study will be divided in different study populations depending on the data available in each cohort: Maternal pre-pregnancy BMI and child neurodevelopment basic information + paternal BMI + maternal BMI genetic score + child BMI genetic score 3

4 Figure 1 Addressing causality in the association between maternal BMI and child neurodevelopment: (A) Observational analysis to investigate the association between maternal pre-pregnancy BMI and child neurodevelopment; we will also use paternal BMI as negative control exposure. (B) MR analysis to investigate the causal effect of maternal BMI genetic score on child neurodevelopment; we will also assess the influence of child BMI genetic score. Principle of MR: if maternal pre-pregnancy BMI (X) causally influences child neurodevelopment (Y), then a maternal pre-pregnancy BMI genetic score (Z) will also be associated with child neurodevelopment (Y). This association should remain significant after adjusting for child BMI genetic score (W). A C Potential confounders X Maternal pre-pregnancy BMI Y Child neurodevelopment development B C Potential confounders Z Maternal BMI genetic score X Maternal pre-pregnancy BMI Y Child neurodevelopment development W Child BMI genetic score 3. Methods 3.1. Design Observational and MR analyses of population-based birth cohorts. 4

5 3.2. Study population Selection criteria for cohort inclusion: Basic data: - Maternal pre-pregnancy BMI - Data on at least one neuropsychological or behavioural domain assessed during childhood Additional data: - Paternal BMI - Maternal BMI genetic score - Child BMI genetic score Remark: although for conducting a MR analysis the exposure, in our case maternal prepregnancy BMI, is not mandatory, we would like to compare the results from the MR and the observational ones in the same cohort Maternal and paternal BMI Maternal and paternal BMI (kg/m 2 ) will be used in continuous and categorical: underweight (<18.50 kg/m 2 ), normal ( kg/m 2 ), overweight ( kg/m 2 ), and obese ( 30 kg/m 2 ). If maternal pre-pregnancy BMI was measured multiple times in your cohort, please use the closer value to conception. Please, see sheet 1 of spreadsheet <MatBMI & Neuro_Covariates & Neuro tests.xlsx> for further information about variable names and codings Child neuropsychological development The following neurodevelopmental outcomes are planned to be assessed. We have restricted the age of the evaluation up to 10 years. - Cognition (divided in early cognition from 1 to 5 years and neurocognition from 5 to 10 years). Includes: general cognition and language (including verbal performance) - Psychomotor development (divided in 1 to 5 years and 5 to 10 years). Includes: global psychomotor, fine psychomotor, and gross psychomotor. - Behaviour (from 4 to 10 years). Includes: behavioural/emotional problems, attentiondeficit hyperactivity disorder (ADHD) symptoms, diagnosis and medication, and autism/asperger traits. To homogenize the scales between cohorts, continuous raw scores will be converted into standard deviation units (z-score equals raw score subtracted from mean and divided by the standard deviation) and then standardized to a mean of 100 and a standard deviation of 15 (new score = (15 x z)). Remark: in the spreadsheet <MatBMI & Neuro_Covariates & Neuro tests.xlsx>, sheet 2, we have detailed the child s ages and neurodevelopment tests for each cohort that we would 5

6 like to assess in the present study. We know that there are some errors; for this reason we ask you to: 1. Complete and confirm the information of your cohort in the excel file. If we have missed one test, please add it. If your cohort does not appear in the excel, we will add it after receiving the information we request (below). 2. Provide us the correct age, instrument, evaluator, and number of items for each test. You can fill in the excel file directly or send us an Provide us the specific questions/items/booklets that participants filled in. After having the definitive neuropsychological tests and ages that will be included, we will ask you to provide a short description of each test including the mean child s age and range of evaluation, domain evaluated, number of items, evaluators, and a reference (example: Bayley Scales of Infant Development I-first edition (BSID I): Children s cognitive and psychomotor development was assessed at around 1.2 years (range years) using the Bayley Scales of Infant Development first edition (BSID I). 1 The mental development scale consists of 163 items that assess age-appropriate cognitive development, including performance abilities, memory, and early language skills. The psychomotor development scale consists of 81 items that assess fine and gross psychomotor development. All testing was done in the health care center in the presence of the mother, by 12 specially trained psychologists. Psychologists were not aware of any exposure information.) Covariates A list of covariates, names, and codings is shown in sheet 1 of the spreadsheet <MatBMI & Neuro_Covariates & Neuro tests.xlsx>, sheet 1. Please respect scrupulously the defined names and codings and use the Cohort s comments column to give additional information regarding the created variables and specify the origin of each variable (registers, selfreported questionnaires, measurements, etc) Maternal BMI genetic score Remark: For those cohorts with genetic data available that have already participated in the EGG-EAGLE-mothers MR study aimed to investigate causality between maternal traits and reproductive outcomes (birth weight and ponderal index), you can use the generated score for pre-pregnancy maternal BMI already generated for that study. For those cohorts that have not created maternal BMI scores, please see ANNEX I where we describe all the steps to create this variable Statistical analysis A centralized approach for combined analyses will be followed, where possible. Each cohort will prepare the necessary information and variables dataset and send it to CREAL (with a previous data transfer agreement signed), where relevant variables will be harmonized. If a 6

7 cohort cannot send the dataset to CREAL (decentralized approach), we provide a uniform script in STATA and R in the document <MatBMI & Neuro_Scripts.docx> so you can run your own analysis. Results of the statistical models will be then sent to CREAL where metaanalysis will be performed Population sample Inclusion criteria: - Live-born children ( 500 g or 22 weeks of gestation) - Singleton infants (in case of twin births, both twins and their mother - should be excluded) - Only one child per mother included (in case of siblings, take out randomly one of them) - AND with available data on maternal pre-pregnancy BMI and at least one neuropsychological development test. Please, exclude those children with congenital anomalies (e.g. Down s syndrome, trisomy 18, etc.). Remark: after these inclusions/exclusions, the final dataset/s will include: Observational analysis: - subjects with available data on maternal pre-pregnancy BMI and at least one neuropsychological development test - subjects with available data on maternal and paternal BMI and at least one neuropsychological development test. MR analysis We ask cohorts with genetic data to first perform the observational analysis for all the study population and then do it again for those subjects with maternal BMI genetic score available. - subjects with available data on maternal BMI genetic score, maternal pre-pregnancy BMI, and at least one neuropsychological development test OR - subjects with available data on maternal BMI genetic score, maternal and paternal BMI, and at least one neuropsychological development test OR - subjects with available data on maternal and child BMI genetic score, maternal prepregnancy BMI, and at least one neuropsychological development test - subjects with available data on maternal and child BMI genetic score, maternal and paternal BMI, and at least one neuropsychological development test Please, indicate your corresponding study population in the spreadsheet <MatBMI & Neuro_Covariates & Neuro tests.xlsx>, sheet Treatment of missing values Centralized approach: missing values do not need to be replaced by the birth cohort data managers. A global approach to handle missing values will be developed centrally (CREAL). 7

8 Decentralized approach: if a cohort cannot send the dataset to CREAL, the same procedure to handle missing values should be performed. Only covariates will be imputed. Please, inform us which procedure you usually apply and complete Table 4a of spreadsheet <MatBMI & Neuro Results Observational.xlsx>. Remark: the number of datasets to be imputed will depend on the number of study populations in each cohort (indicated in <MatBMI & Neuro_Basic information tests.xlsx>, sheet 3) Observational analysis Descriptive analysis Analyses will start with descriptive tables of all variables (exposure, outcomes, and confounders) across cohorts. Centralized approach: will be developed at CREAL. Decentralized approach: please, see in <MatBMI & Neuro_Scripts.docx> where we provide the codes for STATA and R to perform the descriptive statistics, and complete Table 1 of spreadsheet <MatBMI & Neuro_Results Observational.xlsx> Bivariate analysis Bivariate associations of maternal pre-pregnancy BMI and each neuropsychological outcome with all covariates will be performed. Covariates tested are: parity, maternal age, paternal age, maternal country of birth, maternal ethnicity, maternal education, paternal education, maternal socioeconomic status, paternal socioeconomic status, maternal active smoking during pregnancy, maternal intelligence quotient, gestational diabetes, and gestational hypertension. Centralized approach: will be developed at CREAL. Decentralized approach: please, see <MatBMI & Neuro_Scripts.docx> document where we provide examples of codes for STATA and R to perform the bivariate analysis, and complete Tables 2 and 3 of spreadsheet <MatBMI & Neuro Results Observational.xlsx> Association of maternal pre-pregnancy BMI with child neuropsychological development First, we will study the association of maternal pre-pregnancy obesity with child neuropsychological development in each individual cohort. For this, individual analysis stratified by outcome (general cognition, language, global psychomotor, gross psychomotor, fine psychomotor, behavioural/emotional problems, ADHD, and austism), neuropsychological test, and age at neuropsychological assessment will be performed in 8

9 each cohort. Four different models will be performed to assess the influence of the different covariates and paternal BMI (ONLY in cohorts where available). - Model 1: Adjusted for sex of the child, child age at time of test administration, evaluator of the test, and quality of the assessment. - Model 2: Adjusted for variables in Model 1 + socioeconomic variables (maternal country of birth, maternal education, paternal education, maternal socioeconomic status, and paternal socioeconomic status) - Model 3: Adjusted for variables in Model 2 + other covariates (parity, maternal age, paternal age, maternal smoking during pregnancy, and maternal intelligence quotient) - Model 4 (ONLY for cohorts with paternal BMI data): Adjusted for variables in Model 3 + paternal BMI Models 1 to 4 will be performed for: - Maternal pre-pregnancy BMI in continuous - adjusted for paternal BMI in continuous in Model 4 - Maternal pre-pregnancy BMI in categories (reference group: normal mothers) - adjusted for paternal BMI in categorical in Model 4 - Paternal BMI in continuous - adjusted for maternal BMI in continuous in Model 4 - Paternal BMI in categories (reference group: normal fathers) - adjusted for maternal BMI in categorical in Model 4 The following sensitivity analyses will be performed in each cohort: - exclude preterm babies - exclude mothers with gestational diabetes - exclude mothers with gestational hypertension - exclude non-caucasian mothers Centralized approach: all these analyses will be performed at CREAL. Decentralized approach: please, see <MatBMI & Neuro_Scripts.docx> document for where we provide examples of codes for STATA and R to perform all these analyses, and complete Tables 5 and 6 of spreadsheet <MatBMI & Neuro Results Observational.xlsx>. Remark: although we will only present the results based on imputed datasets, we ask cohorts to repeat all models with complete case datasets and complete Tables 8 and 9 of spreadsheet <MatBMI & Neuro Results Observational.xlsx> Meta-analysis Second, at CREAL we will combine observational cohort-specific effect estimates using random-effects meta-analysis: - First, stratified by outcome and neuropsychological test (only the most similar across cohorts if the same outcome is being measured by different tests) 9

10 - Second, stratified by outcome and age at neuropsychological assessment (only the oldest age in the case that a cohort has different ages) We will also perform the following analyses to test the robustness of our results: - leave one population out at the time to determine the influence of a particular cohort. - psychomotor development meta-analyses will be repeated evaluating early ( 2 years) versus later ages to determine the influence of child s age of walking on early psychomotor tests. - stratified analysis for those cohorts where maternal BMI was measured and those where it was reported. - stratified analysis for those cohorts where neuropsychological assessment was administered by psychologist and those based on maternal reports Mendelian randomization analysis Centralized approach: all the following analyses will be performed at CREAL. Decentralized approach: Remark: cohorts with genetic data available need to first perform the observational analysis for all cohort participants and then do it for those participants with genetic data. Please, create another version of the spreadsheet <MatBMI & Neuro_Results Observational.xlsx> called <MatBMI & Neuro_Results Observational_genetic data.xlsx> and complete all tables, numbered from 1b to 9b, considering only those subjects with genetic data Quality control We will examine the association of maternal BMI genetic score with maternal pre-pregnancy BMI in continuous. Decentralized approach: please, see <MatBMI & Neuro_Scripts.docx> document for where we provide examples of codes for STATA and R to perform all these analyses, and complete Table 10 of spreadsheet <MatBMI & Neuro_Results MR.xlsx> Bivariate association of maternal BMI genetic score with covariates Bivariate association of maternal BMI genetic score with all covariates will be performed. Maternal BMI genetic score will be categorized in tertiles in order to observe differences among these categories: low, medium, and high. Decentralized approach: please, see <MatBMI & Neuro_Scripts.docx> document for where we provide examples of codes for STATA and R to perform all these analyses, and complete Table 11 of the spreadsheet <MatBMI & Neuro_Results MR.xlsx>. 10

11 MR analysis of maternal BMI genetic score with child neuropsychological development An individual MR analysis stratified by outcome (general cognition, language, global psychomotor, gross psychomotor, fine psychomotor, behavioural/emotional problems, ADHD, and austism), neuropsychological test, and age at neuropsychological assessment will be performed in each cohort. Decentralized approach: please, see <MatBMI & Neuro_Scripts.docx> document for where we provide examples of codes for STATA and R to perform all these models, and complete Table 12 of the spreadsheet <MatBMI & Neuro_Results MR.xlsx> MR Egger analysis Since some of the BMI genetic variants are strongly associated with the nervous system 23 we will conduct Egger regression that allows to test for bias from pleiotropy 24. This analysis will be performed for each of the MR analysis previously conducted. Decentralized approach: please, see <MatBMI & Neuro_Scripts.docx> document for where we provide examples of codes for STATA and R to perform all these models, and complete Table 13 of the spreadsheet <MatBMI & Neuro_Results MR.xlsx> Adjustment for child BMI genetic score Remark: this analysis is only required for cohorts with child genotype available. In that case, please, remember to restrict the study population for the MR analysis to those mothers who also have child BMI genetic score. We will repeat the genotype extraction and recoding steps for the child genotypes as indicated in Annex I for maternal BMI genetic score. The child genotypes will not be combined to make allele scores; instead they will be included in the regression model as individual SNP covariates. Decentralized approach: please, see <MatBMI & Neuro_Scripts.docx> document for where we provide examples of codes for STATA and R to perform all these models, and complete Table 14 of the spreadsheet <MatBMI & Neuro_Results MR.xlsx> Meta-analysis At CREAL we will combine cohort-specific observational and MR effect estimates using random-effects meta-analysis, following the same procedure as described for the observational analysis. 11

12 4. Organization, publications, and ethical issues 4.1. Organization Maribel Casas will coordinate the work and Debbie Lawlor and Martine Vrijheid will supervise it. Each birth cohort principal investigator will assign the researchers who will join the study. Each cohort team will prepare the dataset or develop the analyses according to the uniform protocol. Each cohort team will actively participate in the analysis protocol, TC meetings, interpretation of data and results, and drafting of scientific paper/s Publications At least one publication is planned. Depending on the maximum number of authors allowed by the journal of choice, all working group members will be co-authors. If the maximum is exceeded, there will be a discussion with all cohorts to form a writing group (maximum authors depending on the journal), and a collaborators group Permissions and ethical issues All original cohort studies should be approved by their local institutional review boards, and should have provided written informed consent for using their data. The principal investigators obtain the necessary permissions to perform data analyses at the international level. The principal investigator keeps all personal identifiers according to national guidelines. Data analyses and presentation of papers should adhere to good epidemiological practice and the STROBE guidelines. Aggregated data sets with individual anonymous observations, if any, are to be deleted, when the data analysis is finished. 5. What is expected from each cohort First, in case you have not confirmed your participation, please send an to Maribel Casas (mcasas@creal.cat), informing us if you are willing to participate. ALL COHORTS: inform if you are going to send the dataset to CREAL (centralized approach) or are going to perform the analysis by yourself (decentralized approach). IN CASE OF A CENTRALIZED APPROACH: 1) To fill in and sign the Data Transfer Agreement and send it to us. 2) To provide us the information for the neuropsychological outcomes: a. Complete and confirm the information of your cohort in the excel file <MatBMI & Neuro_Basic information.xlsx>. If we have missed one test, please add it. b. Provide us the correct age, instrument, evaluator, and number of items for each test. You can fill in the excel file directly or send us an . c. Provide us the specific questions/items/booklets that participants filled in. d. Provide us a short description of each test (this could be done in a later stage). 12

13 After this, we will inform you about the final neuropsychological tests and ages that should be included in the dataset so you can proceed creating it. 3) To define your study population and complete spreadsheet <MatBMI & Neuro_Basic information.xlsx> sheet 3. 4) To create in the dataset the variables defined in the spreadsheet <MatBMI & Neuro_Basic information.xlsx>. Please respect scrupulously the defined names and codings and use the Cohort s comments column to give us additional information regarding the created variables. 5) To create a variable named cohort corresponding to the cohort name (identical for all observations). In case your cohort has different regions, create a variable named region. Also, create a unique identifier for each woman-child pair in a variable named id. Missing values should be coded with a dot (. ). If it is not the case, please indicate the coding of missing values used in the dataset that you send to us. 6) Convert the dataset that you created, either in SAS, STATA, or SPSS format. Name it with the name of the cohort, the format, and the date, adding MatBMI & Neuro. Example (using INMA cohort): <MatBMI & Neuro_INMA_STATA_ dta> (if STATA format). Please, transfer the following list of documents: a. the data file created (<MatBMI & Neuro_INMA_STATA_ dta>). b. the spreadsheet <MatBMI & Neuro_Basic information_inma.xlsx> giving additional information regarding the created variables. We will inform you how to transfer them using a SFTP (Secure File Transfer Protocol). 7) Check the descriptive statistics that we will send to you after receiving your dataset. 8) Actively participate in the analysis protocol, TC meetings, interpretation of data and results, and drafting of scientific paper/s. IN CASE OF A DECENTRALIZED APPROACH: 1) To provide us the information for the neuropsychological outcomes: a. Complete and confirm the information of your cohort in the excel file <MatBMI & Neuro_Basic information.xlsx>. If we have missed one test, please add it. b. Provide us the correct age, instrument, evaluator, and number of items for each test. You can fill in the excel file directly or send us an . c. Provide us the specific questions/items/booklets that participants filled in. d. Provide us a short description of each test (this could be done in a later stage). After this, we will inform you about the final neuropsychological tests and ages that should be included in the dataset so you can proceed creating it. 2) To define your study population and complete spreadsheet <MatBMI & Neuro_Basic information.xlsx> sheet 3. 3) Inform about the procedure usually followed in your cohort to handle missing values in covariates. 13

14 4) To create the dataset and perform observational and MR analyses if applicable. Indicate the name of the cohort. Example (using INMA cohort): <MatBMI & Neuro_Results Observational_INMA.xlsx>. Please, send us the following spreadsheets: a. <MatBMI & Neuro_Results Observational.xlsx> in case cohort has no genetic data. b. <MatBMI & Neuro_Results Observational_genetic data.xlsx> and <MatBMI & Neuro Results MR.xlsx> in case cohort has only genetic data. c. <MatBMI & Neuro_Results Observational.xlsx>, <MatBMI & Neuro Results Observational_genetic data.xlsx> and <MatBMI & Neuro Results MR.xlsx> in case cohort can perform the analysis in the total population and in the subsample with genetic data. 5) Actively participate in the analysis protocol, TC meetings, interpretation of data and results, and drafting of scientific paper/s. Please do not hesitate to contact us if anything is not clear or if you need assistance at any step: Maribel Casas 6. References 1. Berghofer A, Pischon T, Reinhold T, Apovian CM, Sharma AM, Willich SN. Obesity prevalence from a European perspective: a systematic review. BMC Public Health ; Castanon N, Lasselin J, Capuron L. Neuropsychiatric comorbidity in obesity: role of inflammatory processes. Front Endocrinol (Lausanne) ; Das UN. Is obesity an inflammatory condition? Nutrition ; van der Burg JW, Allred EN, McElrath TF, et al. Is maternal obesity associated with sustained inflammation in extremely low gestational age newborns? Early Hum Dev ; Bilbo SD, Tsang V. Enduring consequences of maternal obesity for brain inflammation and behavior of offspring. FASEB J ; Tozuka Y, Kumon M, Wada E, Onodera M, Mochizuki H, Wada K. Maternal obesity impairs hippocampal BDNF production and spatial learning performance in young mouse offspring. Neurochem Int ; Basatemur E, Gardiner J, Williams C, Melhuish E, Barnes J, Sutcliffe A. Maternal Prepregnancy BMI and Child Cognition: A Longitudinal Cohort Study. Pediatrics Bliddal M, Olsen J, Stovring H, et al. Maternal pre-pregnancy BMI and intelligence quotient (IQ) in 5-year-old children: a cohort based study. PLoS One ; e Brion MJ, Zeegers M, Jaddoe V, et al. Intrauterine effects of maternal prepregnancy overweight on child cognition and behavior in 2 cohorts. Pediatrics ; e202-e

15 10. Buss C, Entringer S, Davis EP, et al. Impaired executive function mediates the association between maternal pre-pregnancy body mass index and child ADHD symptoms. PLoS One ; e Casas M, Chatzi L, Carsin AE, et al. Maternal pre-pregnancy overweight and obesity, and child neuropsychological development: two Southern European birth cohort studies. Int J Epidemiol ; Chen Q, Sjolander A, Langstrom N, et al. Maternal pre-pregnancy body mass index and offspring attention deficit hyperactivity disorder: a population-based cohort study using a sibling-comparison design. Int J Epidemiol ; Gage SH, Lawlor DA, Tilling K, Fraser A. Associations of Maternal Weight Gain in Pregnancy With Offspring Cognition in Childhood and Adolescence: Findings From the Avon Longitudinal Study of Parents and Children. Am J Epidemiol Heikura U, Taanila A, Hartikainen AL, et al. Variations in prenatal sociodemographic factors associated with intellectual disability: a study of the 20-year interval between two birth cohorts in northern Finland. Am J Epidemiol ; Hinkle SN, Schieve LA, Stein AD, Swan DW, Ramakrishnan U, Sharma AJ. Associations between maternal prepregnancy body mass index and child neurodevelopment at 2 years of age 1. Int J Obes (Lond) ; Huang L, Yu X, Keim S, Li L, Zhang L, Zhang J. Maternal prepregnancy obesity and child neurodevelopment in the Collaborative Perinatal Project. Int J Epidemiol ; Neggers YH, Goldenberg RL, Ramey SL, Cliver SP. Maternal prepregnancy body mass index and psychomotor development in children. Acta Obstet Gynecol Scand ; Rodriguez A, Miettunen J, Henriksen TB, et al. Maternal adiposity prior to pregnancy is associated with ADHD symptoms in offspring: evidence from three prospective pregnancy cohorts. Int J Obes (Lond) ; Rodriguez A. Maternal pre-pregnancy obesity and risk for inattention and negative emotionality in children. J Child Psychol Psychiatry ; Suren P, Gunnes N, Roth C, et al. Parental Obesity and Risk of Autism Spectrum Disorder. Pediatrics Tanda R, Salsberry PJ, Reagan PB, Fang MZ. The Impact of Prepregnancy Obesity on Children's Cognitive Test Scores. Matern Child Health J Smith GD. Assessing intrauterine influences on offspring health outcomes: can epidemiological studies yield robust findings? Basic Clin Pharmacol Toxicol ; Locke AE, Kahali B, Berndt SI, et al. Genetic studies of body mass index yield new insights for obesity biology. Nature ; Bowden J, Davey SG, Burgess S. Mendelian randomization with invalid instruments: effect estimation and bias detection through Egger regression. Int J Epidemiol ;