{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,10,8]],"date-time":"2026-10-08T20:28:15Z","timestamp":1791491295734,"version":"4.3.4"},"update-to":[{"DOI":"10.12688\/wellcomeopenres.15087.1","type":"new_version","label":"New version","source":"publisher","updated":{"date-parts":[[2019,12,16]],"date-time":"2019-12-16T00:00:00Z","timestamp":1576454400000}}],"reference-count":40,"publisher":"F1000 Research Ltd","license":[{"start":{"date-parts":[[2019,12,16]],"date-time":"2019-12-16T00:00:00Z","timestamp":1576454400000},"content-version":"tdm","delay-in-days":0,"URL":"http:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"funder":[{"DOI":"10.13039\/501100000781","name":"European Research Council","doi-asserted-by":"publisher","award":["669545"],"award-info":[{"award-number":["669545"]}],"id":[{"id":"10.13039\/501100000781","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/501100000781","name":"European Research Council","doi-asserted-by":"publisher","award":["758813"],"award-info":[{"award-number":["758813"]}],"id":[{"id":"10.13039\/501100000781","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/501100000265","name":"Medical Research Council","doi-asserted-by":"publisher","award":["MC_UU_00011\/6"],"award-info":[{"award-number":["MC_UU_00011\/6"]}],"id":[{"id":"10.13039\/501100000265","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/100000002","name":"National Institutes of Health","doi-asserted-by":"publisher","award":["DK10324"],"award-info":[{"award-number":["DK10324"]}],"id":[{"id":"10.13039\/100000002","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/501100000272","name":"National Institute for Health Research","doi-asserted-by":"publisher","award":["NF-SI-0611-10196"],"award-info":[{"award-number":["NF-SI-0611-10196"]}],"id":[{"id":"10.13039\/501100000272","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/100004440","name":"Wellcome Trust","doi-asserted-by":"publisher","award":["102215"],"award-info":[{"award-number":["102215"]}],"id":[{"id":"10.13039\/100004440","id-type":"DOI","asserted-by":"publisher"}]},{"name":"Medical Research Council","award":["MC_UU_00011\/3"],"award-info":[{"award-number":["MC_UU_00011\/3"]}]}],"content-domain":{"domain":["wellcomeopenresearch.org"],"crossmark-restriction":false},"short-container-title":["Wellcome Open Res"],"abstract":"<ns4:p>\n                    <ns4:bold>Background:<\/ns4:bold>\n                    The Avon Longitudinal Study of Parents and Children-Generation 2 (ALSPAC-G2) was set up to provide a unique multi-generational cohort. It builds on the existing ALSPAC resource, which recruited 14,541 pregnancies to women resident in the South West of England who were expected to deliver between 01\/04\/1991 and 31\/12\/1992. Those women and their partners (Generation 0; ALSPAC-G0) and their offspring (ALSPAC-G1) have been followed for the last 27 years. This profile describes recruitment and data collection on the next generation (ALSPAC-G2)\u2014the grandchildren of ALSPAC-G0 and children of ALSPAC-G1.\n                  <\/ns4:p>\n                  <ns4:p>\n                    <ns4:bold>Recruitment:<\/ns4:bold>\n                    Recruitment began on the 6\n                    <ns4:sup>th<\/ns4:sup>\n                    of June 2012 and we present details of recruitment and participants up to 30\n                    <ns4:sup>th<\/ns4:sup>\n                    June 2018 (~6 years). We knew at the start of recruitment that some ALSPAC-G1 participants had already become parents and ALSPAC-G2 is an open cohort; we recruit at any age. We hope to continue recruiting until all ALSPAC-G1 participants have completed their families. Up to 30\n                    <ns4:sup>th<\/ns4:sup>\n                    June 2018 we recruited 810 ALSPAC-G2 participants from 548 families. Of these 810, 389 (48%) were recruited during their mother\u2019s pregnancy, 287 (35%) before age 3 years, 104 (13%) between 3-6 years and 30 (4%) after 6 years. Over 70% of those invited to early pregnancy, late pregnancy, second week of life, 6-, 12- and 24-month assessments (whether for their recruitment, or a follow-up, visit) have attended, with attendance being over 60% for subsequent visits up to 7 years (too few are eligible for the 9- and 11-year assessments to analyse).\n                  <\/ns4:p>\n                  <ns4:p>\n                    <ns4:bold>Data collection:<\/ns4:bold>\n                    We collect a wide-range of socioeconomic, lifestyle, clinical, anthropometric and biological data on all family members repeatedly. Biological samples include blood (including cord-blood), urine, meconium and faeces, and placental tissue. In subgroups detailed data collection, such as continuous glucose monitoring and videos of parent-child interactions, are being collected.\n                  <\/ns4:p>","DOI":"10.12688\/wellcomeopenres.15087.2","type":"journal-article","created":{"date-parts":[[2019,12,16]],"date-time":"2019-12-16T10:15:13Z","timestamp":1576491313000},"page":"36","update-policy":"https:\/\/doi.org\/10.12688\/wellcomeopenres.crossmark-policy","source":"Crossref","is-referenced-by-count":47,"title":["The second generation of The Avon Longitudinal Study of Parents and Children (ALSPAC-G2): a cohort profile"],"prefix":"10.12688","volume":"4","author":[{"ORCID":"https:\/\/orcid.org\/0000-0002-6793-2262","authenticated-orcid":false,"given":"Deborah A.","family":"Lawlor","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Melanie","family":"Lewcock","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Louise","family":"Rena-Jones","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Claire","family":"Rollings","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Vikki","family":"Yip","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-6467-2023","authenticated-orcid":false,"given":"Daniel","family":"Smith","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Rebecca M.","family":"Pearson","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-8985-3154","authenticated-orcid":false,"given":"Laura","family":"Johnson","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-4787-8411","authenticated-orcid":false,"given":"Louise A. C.","family":"Millard","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Nashita","family":"Patel","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Andy","family":"Skinner","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Kate","family":"Tilling","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"name":"ALSPAC Executive","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"2560","published-online":{"date-parts":[[2019,12,16]]},"reference":[{"key":"ref-1","doi-asserted-by":"publisher","first-page":"111-27","DOI":"10.1093\/ije\/dys064","article-title":"Cohort Profile: the 'children of the 90s'--the index offspring of the Avon Longitudinal Study of Parents and Children.","volume":"42","author":"A Boyd","year":"2013","journal-title":"Int J Epidemiol."},{"key":"ref-2","doi-asserted-by":"publisher","first-page":"97-110","DOI":"10.1093\/ije\/dys066","article-title":"Cohort Profile: The Avon Longitudinal Study of Parents and Children: ALSPAC mothers cohort.","volume":"42","author":"A Fraser","year":"2013","journal-title":"Int J Epidemiol."},{"key":"ref-3","doi-asserted-by":"publisher","DOI":"10.1093\/hropen\/hoy001","article-title":"The Environment and Reproductive Health (EARTH) Study: A Prospective Preconception Cohort.","volume":"2018","author":"C Messerlian","year":"2018","journal-title":"Hum Reprod Open."},{"key":"ref-4","doi-asserted-by":"publisher","first-page":"20-27","DOI":"10.1016\/j.socscimed.2018.07.035","article-title":"It's the mother!: How assumptions about the causal primacy of maternal effects influence research on the developmental origins of health and disease.","volume":"213","author":"G Sharp","year":"2018","journal-title":"Soc Sci Med."},{"key":"ref-5","doi-asserted-by":"publisher","first-page":"1866-1886","DOI":"10.1093\/ije\/dyw314","article-title":"Triangulation in aetiological epidemiology.","volume":"45","author":"D Lawlor","year":"2016","journal-title":"Int J Epidemiol."},{"key":"ref-6","doi-asserted-by":"publisher","first-page":"804-814","DOI":"10.1038\/s41588-019-0403-1","article-title":"Maternal and fetal genetic effects on birth weight and their relevance to cardio-metabolic risk factors.","volume":"51","author":"N Warrington","year":"2019","journal-title":"Nat Genet."},{"key":"ref-7","doi-asserted-by":"publisher","first-page":"1129-1240","DOI":"10.1001\/jama.2016.1975","article-title":"Genetic Evidence for Causal Relationships Between Maternal Obesity-Related Traits and Birth Weight.","volume":"315","author":"J Tyrrell","year":"2016","journal-title":"JAMA."},{"key":"ref-8","doi-asserted-by":"publisher","first-page":"227-41","DOI":"10.1177\/1536867X0400400301","article-title":"Multiple imputation of missing values.","volume":"4","author":"P Royston","year":"2004","journal-title":"Stata J."},{"key":"ref-9","article-title":"ALSPAC-G2 cohort","author":"D Lawlor","year":"2019"},{"key":"ref-10","doi-asserted-by":"publisher","DOI":"10.1101\/500256","article-title":"GLU: A tool for analysing continuously measured glucose in epidemiology.","author":"L Millard","year":"2018","journal-title":"bioRxiv."},{"key":"ref-11","doi-asserted-by":"publisher","first-page":"62-71","DOI":"10.1016\/j.infbeh.2017.02.006","article-title":"Through babies' eyes: Practical and theoretical considerations of using wearable technology to measure parent-infant behaviour from the mothers' and infants' view points.","volume":"47","author":"R Lee","year":"2017","journal-title":"Infant Behav Dev."},{"key":"ref-12","doi-asserted-by":"publisher","DOI":"10.1017\/S0029665116002573","article-title":"Use of the Remote Food Photography Method by young pregnant women: a feasibility study.","volume":"75","author":"C England","year":"2016","journal-title":"Proc Nutr Soc."},{"key":"ref-13","article-title":"Prevalence of prenatal depression symptoms among 2 generations of pregnant women.","author":"R Pearson","year":"2018","journal-title":"JAMA Network."},{"key":"ref-14","doi-asserted-by":"publisher","first-page":"2267-75","DOI":"10.1016\/j.csda.2010.04.005","article-title":"Avoiding bias due to perfect prediction in multiple imputation of incomplete categorical variables.","volume":"54","author":"I White","year":"2010","journal-title":"Comput Stat Data Anal."},{"key":"ref-15","doi-asserted-by":"publisher","first-page":"6","DOI":"10.1109\/WCNCW.2018.8368985","article-title":"An Accelerometer Lossless Compression Algorithm and Energy Analysis for IoT Devices.","author":"J Pope","year":"2018","journal-title":"IEEE Wireless Communications and Networking Conference Workshops (WCNCW)."},{"key":"ref-16","doi-asserted-by":"publisher","first-page":"72-81","DOI":"10.1111\/jhn.12014","article-title":"Measuring food intake with digital photography.","volume":"27 Suppl 1","author":"C Martin","year":"2014","journal-title":"J Hum Nutr Diet."},{"key":"ref-17","doi-asserted-by":"publisher","DOI":"10.1017\/S0029665116002342","article-title":"FoodFinder: developing a rapid low-cost crowdsourcing approach for obtaining data on meal size from meal photos.","volume":"75","author":"L Johnson","year":"2016","journal-title":"Proc Nutr Soc."},{"key":"ref-18","doi-asserted-by":"publisher","first-page":"1631-40","DOI":"10.2337\/dc17-1600","article-title":"International Consensus on Use of Continuous Glucose Monitoring.","volume":"40","author":"T Danne","year":"2017","journal-title":"Diabetes Care."},{"key":"ref-19","article-title":"Births by parents\u2019 characteristics in England and Wales: 2016","author":"N Haines","year":"2017"},{"key":"ref-20","doi-asserted-by":"publisher","first-page":"226-35","DOI":"10.1093\/ije\/dyx206","article-title":"Collider scope: when selection bias can substantially influence observed associations.","volume":"47","author":"M Munaf\u00f2","year":"2018","journal-title":"Int J Epidemiol."},{"key":"ref-21","doi-asserted-by":"publisher","first-page":"1895-1903","DOI":"10.1093\/ije\/dyw328","article-title":"Causal inference-so much more than statistics.","volume":"45","author":"N Pearce","year":"2016","journal-title":"Int J Epidemiol."},{"key":"ref-22","doi-asserted-by":"publisher","first-page":"1-9","DOI":"10.1097\/EDE.0b013e31827623b1","article-title":"Loss to follow-up in cohort studies: bias in estimates of socioeconomic inequalities.","volume":"24","author":"L Howe","year":"2013","journal-title":"Epidemiology."},{"key":"ref-23","doi-asserted-by":"publisher","first-page":"1294-1304","DOI":"10.1093\/ije\/dyz032","article-title":"Accounting for missing data in statistical analyses: multiple imputation is not always the answer.","volume":"48","author":"R Hughes","year":"2019","journal-title":"Int J Epidemiol."},{"key":"ref-24","doi-asserted-by":"publisher","first-page":"937-45","DOI":"10.1093\/ije\/dyv035","article-title":"Using linked educational attainment data to reduce bias due to missing outcome data in estimates of the association between the duration of breastfeeding and IQ at 15 years.","volume":"44","author":"R Cornish","year":"2015","journal-title":"Int J Epidemiol."},{"key":"ref-25","doi-asserted-by":"publisher","first-page":"1038-1039k","DOI":"10.1093\/ije\/dyz063","article-title":"Data Resource Profile: The ALSPAC birth cohort as a platform to study the relationship of environment and health and social factors.","volume":"48","author":"A Boyd","year":"2019","journal-title":"Int J Epidemiol."},{"key":"ref-26","doi-asserted-by":"publisher","first-page":"1207-1216","DOI":"10.1093\/ije\/dyy060","article-title":"Exploring the association of genetic factors with participation in the Avon Longitudinal Study of Parents and Children.","volume":"47","author":"A Taylor","year":"2018","journal-title":"Int J Epidemiol."},{"key":"ref-27","doi-asserted-by":"publisher","first-page":"249-256","DOI":"10.1192\/bjp.bp.108.053751","article-title":"Selective drop-out in longitudinal studies and non-biased prediction of behaviour disorders.","volume":"195","author":"D Wolke","year":"2009","journal-title":"Br J Psychiatry."},{"key":"ref-28","doi-asserted-by":"publisher","first-page":"1012-1014","DOI":"10.1093\/ije\/dys223","article-title":"Why representativeness should be avoided.","volume":"42","author":"K Rothman","year":"2013","journal-title":"Int J Epidemiol."},{"key":"ref-29","doi-asserted-by":"publisher","first-page":"1014-1015","DOI":"10.1093\/ije\/dyt101","article-title":"Commentary: On representativeness.","volume":"42","author":"J Elwood","year":"2013","journal-title":"Int J Epidemiol."},{"key":"ref-30","doi-asserted-by":"publisher","first-page":"1016-1017","DOI":"10.1093\/ije\/dyt102","article-title":"Commentary: Epidemiologists have debated representativeness for more than 40 years--has the time come to move on?","volume":"42","author":"E Nohr","year":"2013","journal-title":"Int J Epidemiol."},{"key":"ref-31","doi-asserted-by":"publisher","first-page":"1018-1022","DOI":"10.1093\/ije\/dyt103","article-title":"Commentary: Representativeness is usually not necessary and often should be avoided.","volume":"42","author":"L Richiardi","year":"2013","journal-title":"Int J Epidemiol."},{"key":"ref-32","doi-asserted-by":"publisher","first-page":"1026-1028","DOI":"10.1093\/ije\/dyt124","article-title":"Rebuttal: When it comes to scientific inference, sometimes a cigar is just a cigar.","volume":"42","author":"K Rothman","year":"2013","journal-title":"Int J Epidemiol."},{"key":"ref-33","doi-asserted-by":"publisher","first-page":"1022-1026","DOI":"10.1093\/ije\/dyt105","article-title":"Commentary: Should we always deliberately be non-representative?","volume":"42","author":"S Ebrahim","year":"2013","journal-title":"Int J Epidemiol."},{"key":"ref-34","doi-asserted-by":"publisher","first-page":"2089-2097","DOI":"10.1016\/S0140-6736(15)00131-2","article-title":"Screening for fetal growth restriction with universal third trimester ultrasonography in nulliparous women in the Pregnancy Outcome Prediction (POP) study: a prospective cohort study.","volume":"386","author":"U Sovio","year":"2013","journal-title":"Lancet."},{"key":"ref-35","doi-asserted-by":"publisher","first-page":"368-737","DOI":"10.1093\/ije\/dyl284","article-title":"Teenage pregnancy and adverse birth outcomes: a large population based retrospective cohort study.","volume":"36","author":"C Xi-Kuan","year":"2007","journal-title":"Int J Epidemiol."},{"key":"ref-36","doi-asserted-by":"publisher","first-page":"1084-1096","DOI":"10.1093\/aje\/kwx016","article-title":"Quantitative Serum Nuclear Magnetic Resonance Metabolomics in Large-Scale Epidemiology: A Primer on -Omic Technologies.","volume":"186","author":"P Wurtz","year":"2017","journal-title":"Am J Epidemiol."},{"key":"ref-37","doi-asserted-by":"publisher","DOI":"10.1093\/ije\/dyz098","article-title":"4-Hydroxyglutamate is a novel predictor of pre-eclampsia.","author":"U Sovio","year":"2019","journal-title":"Int J Epidemiol."},{"key":"ref-38","doi-asserted-by":"publisher","first-page":"1893","DOI":"10.1038\/s41467-019-09671-3","article-title":"Meta-analysis of epigenome-wide association studies in neonates reveals widespread differential DNA methylation associated with birthweight.","volume":"10","author":"L K\u00fcpers","year":"2019","journal-title":"Nat Commun."},{"key":"ref-39","doi-asserted-by":"publisher","first-page":"257-261","DOI":"10.1093\/ntr\/nty008","article-title":"StopWatch: The Preliminary Evaluation of a Smartwatch-Based System for Passive Detection of Cigarette Smoking.","volume":"21","author":"A Skinner","year":"2019","journal-title":"Nicotine Tob Res."},{"key":"ref-40","doi-asserted-by":"publisher","first-page":"543-70","DOI":"10.1080\/13854046.2012.680913","article-title":"Scoring higher the second time around: meta-analyses of practice effects in neuropsychological assessment.","volume":"26","author":"M Calamia","year":"2012","journal-title":"Clin Neuropsychol."}],"container-title":["Wellcome Open Research"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/wellcomeopenresearch.org\/articles\/4-36\/v2\/xml","content-type":"application\/xml","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/wellcomeopenresearch.org\/articles\/4-36\/v2\/pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/wellcomeopenresearch.org\/articles\/4-36\/v2\/iparadigms","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2020,10,27]],"date-time":"2020-10-27T08:11:46Z","timestamp":1603786306000},"score":1,"resource":{"primary":{"URL":"https:\/\/wellcomeopenresearch.org\/articles\/4-36\/v2"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2019,12,16]]},"references-count":40,"URL":"https:\/\/doi.org\/10.12688\/wellcomeopenres.15087.2","relation":{"has-review":[{"id-type":"doi","id":"10.21956\/wellcomeopenres.17079.r37442","asserted-by":"subject"},{"id-type":"doi","id":"10.21956\/wellcomeopenres.17079.r37442","asserted-by":"object"}]},"ISSN":["2398-502X"],"issn-type":[{"value":"2398-502X","type":"electronic"}],"subject":[],"published":{"date-parts":[[2019,12,16]]},"assertion":[{"value":"Indexed","URL":"https:\/\/wellcomeopenresearch.org\/articles\/4-36\/v2#article-reports","order":0,"name":"referee-status","label":"Referee status","group":{"name":"current-referee-status","label":"Current Referee Status"}},{"value":"10.21956\/wellcomeopenres.16460.r35257, Anna Vignoles, Department of Education, University of Cambridge, Cambridge, UK, 18 Apr 2019, version 1, 1 approved, 1 approved with reservations","URL":"https:\/\/wellcomeopenresearch.org\/articles\/4-36\/v1#referee-response-35257","order":0,"name":"referee-response-35257","label":"Referee Report","group":{"name":"article-reports","label":"Article Reports"}},{"value":"<b>Deborah Lawlor<\/b>; \n<i>Posted: 14 Nov 2019<\/i>; This is a very informative description of a particular cohort of children called ALSPAC-G2. These children are the grandchildren of the original ALSPAC-G0 cohort of mothers who were recruited into the ALSPAC study if they were due to give birth in the Avon area in 1991\/1992. The children of the ALSPAC-G0 cohort (ALSPAC-G1, known to the public as Children of the 90s) have been followed into early adulthood and are now having children themselves. Their children (ALSPAC-G2 or Children of the Children of the 90s) are in the process of being recruited into the study, providing an intergenerational study spanning three generations. This paper describes ALSPAC-G2, the motivation for collecting data on these children and generations. This paper describes ALSPAC-G2, the motivation for collecting data on these children and the key hypotheses that can be addressed with the data. It provides a vitally important resource guide for those considering using the data.The paper also reports some very interesting work that has already been done with the data andpublished elsewhere. This nicely illustrates the strengths of ALSPAC-G0\/1\/2 and certainly enhances the paper. However, it does mean that some of the findings reported cannot be verified with the data provided in the paper (though full references are given). This is why I have indicated that the detailed account of methods of analysis is only partly available in the paper. I do not see this as a problem however, and it is common in this type of paper to summarise others\u2019 work in this way.The study is scientifically sound without further changes. However, I would note the following issueswhich the authors may wish to consider or comment on.&nbsp;\n<b>Significant issues<\/b>  The summary of the strengths and weaknesses of the data is very clear and comprehensive. In particular, the authors acknowledge the increasing selection bias that is likely to arise as ALSPAC recruits subsequent generations. It also discusses some of the potential risks associated with that. In particular, the data from ALSPAC-G2 will not be as representative of the Avon area as the original cohort. That said, the linkage with administrative data offers a potential opportunity not just to draw in additional phenotype data etc. but also to provide a source of external data that can be used to assess the extent of the representativeness of the ALSPAC data. This might be noted.  \n<b><u>Response<\/u><\/b>&nbsp;We thank the reviewer for these comments. In response to reviewer 1, in the revised paper we have extended the section on strengths and limitations and use subheadings covering three key areas \u2013 selection bias, representativeness, and measurements across generations. Below we have copied the now extended discussion of the first two of these, which we hope provide sufficient information to inform future users of the data. Within this we have also tried to encourage future users to also share methodological insights into selection bias\/representativeness within this resource. These will also have potential value for other studies with complex family data.&nbsp;\n<b><i>\u201cSelection bias <\/i><\/b>\n<i>There are a number of sources of potential selection bias in ALSPAC-G2 <\/i>\n<i><sup>&nbsp;20<\/sup><\/i> \n<i>. Some of these, such as incomplete recruitment and loss to follow-up, are common to most prospective cohort studies, whereas others, such as the current recruitment at a relatively young age and the focus of analyses in cross-generational effects, are more specific to ALSPAC-G2. Below we describe some of the key potential sources of selection bias, general methods that might be used to mitigate against some of the selection mechanisms, and data and analytical approaches that could be used to explore and control for potential selection bias. It is important to note that appropriate methods for exploring and controlling for selection bias are specific to the analyses being undertaken. Furthermore, ALSPAC is a resource for the global scientific community to use and we do not dictate how scientists using this resource complete their analyses. Whilst we make suggestions here that might be used to explore selection bias, researchers must decide how they want to approach this for the specific questions they are exploring with these data.<\/i>\n<i>Our response is 56% of those invited (53% overall) and respondents are more likely to be those already engaged with ALSPAC as indicated by being more likely to have participated in the two most recent clinic assessments for all ALSPAC-G1 participants. Once recruited a high proportion remain actively engaged with follow-up, though the extent to which this will continue as the cohort ages is unknown. Currently, ALSPAC-G2 are children being born to relatively young parents (mean 22 years (range 17-26)); though the age range of parents will increase with continued follow-up. Whilst cross-generational comparisons that have already been published and\/or presented in this paper are age matched, it is possible that the factors that influence starting a family at a relatively young age differ across generations. If these factors are unknown or unmeasured (and therefore cannot be controlled for) there may be bias in the cross generational comparisons. This may be further complicated by secular trends in some of these factors. For example, preliminary analyses presented here suggest that age matched ALSPAC-G1 mothers with an ALSPAC-G2 child are more educated than were their ALSPAC-G0 mother\u2019s generation when recruited during their pregnancy ~30years ago. This difference may be due to true differences between UK pregnant women across generations. For example, there may have been a reduction in the stigma attached to becoming pregnant at a young age and hence a reduction in the association of education\/socioeconomic position with it. ALSPAC-G1 mothers may also be truly more educated than their parent\u2019s generation as a result of changes in education policy and secular trends related to that policy. Over the last 20-30 years there has been a doubling of the proportion of people obtaining a University degree in the UK, stimulated by policy changes, including previous further education establishments becoming Universities in the 1990s. Alternatively these cross-generational differences in education could be due to selection bias. This is supported by our finding that those ALSPAC-G1 parents recruited as pregnant women or parents to ALSPAC-G2 are more engaged with the study than those not recruited (Table 2), and that ALSPAC-G1 participants who are more engaged with the study are on average more educated than those less engaged<sup>1,2<\/sup>. <\/i>\n<i>At the time we began recruitment, ALSPAC-G1 participants were aged 19\u201321 years and some had already become parents. Thus, whilst the cohort is currently of relatively young parents, we do not have detailed pregnancy data on those who became pregnant before 19-years. To maximise participation of all ALSPAC-G2 we will continue to recruit participants at different ages. Forthose ALSPAC-G2 participants recruited after birth, either because they were born prior to when we began recruitment or because their parents only decided to join ALSPAC-G2 when their child\/children were older, their mother\u2019s pregnancy information and their birth\/infancy information will be limited to what we can obtain from record linkage. With the extensive linkage that is now possible, including primary-, secondary- and community health care, education and environmental linkage we will still have a lot of information on these pregnancies and the resulting infants. Furthermore, the proportion recruited in pregnancy has increased over time and once a family have joined ALSPAC-G2, subsequent children are mostly recruited during their mother\u2019s pregnancy.<\/i>\n<i>To mitigate against some of these sources of bias we aim to minimise participant burden by using record linkage and remote data collection as much as possible. The ability to do this is supported by the existence of a National Health Service in the UK and large-scale initiatives across UK research funders for supporting ethical and appropriate access to anonymised health, environmental, and socioeconomic data. Linked data can also be used for describing the likely nature of any selection bias and informing sensitivity analyses. <\/i>\n<i>Both existing research and record linkage data, including on grandparents and parents (ALSPAC-G0 and -G1), can be used to explore potential selection bias from any of the sources described above, and to inform sensitivity analyses to explore the extent of bias that might occur for any given analyses <\/i>\n<i><sup>20<\/sup><\/i> \n<i>. In this section we describe some of these approaches in general terms, as it is important to note that which methods are most appropriate to use will depend on the specific research questions being addressed and the likely source of selection (missing data).<sup>21,22<\/sup> Furthermore, ALSPAC data are available for scientists across the world to use. Whilst we highlight potential sources of selection bias in this cohort profile and will do so through conference presentations and the study website, we cannot dictate how other scientists approach these issues when using the study data. We will encourage others to explore issues of selection bias when using ALSPAC-G2 data and make details of their methods open access (including via linked Wellcome Open Research publications).<\/i>\n<i>In our work we have found directed acyclic graphs<sup>21<\/sup> useful to illustrate specific sources of selection bias and their assumptions of any methods used to explore and control for these (see for example references<sup>20-24<\/sup>). Selection bias results from missing data (for example, missing data due to not being recruited, being lost to follow-up or missing data on factors influencing different fertility practices across generations) thus methods for dealing with missing data are appropriate approaches to exploring and controlling for selection bias. Publication reporting guidelines, such as Strengthening The Reporting of Observational studies in Epidemiology (STROBE; <\/i>\n<i>https:\/\/strobe-statement.org\/index.php?id=strobe-home<\/i>\n<i>) require authors to describe the extent of missing data and how potential bias from that has been explored and we strongly encourage users of ALSPAC to adhere to those reporting guidelines. Such reporting guidelines do not recommend specific approaches, e.g. for dealing with missing data, just that there is clarity about the issue. A common approach to dealing with missing data is complete case analyses (i.e. in which only participants with complete data on exposure, outcome and all covariables included in the main analyses are included in analyses). This is in general expected to be unbiased if missingness is not related to the outcome in the main analysis model conditioned on exposure and all covariables included in the main analyses.<sup>23<\/sup> Other commonly used methods for exploring and mitigating against selection bias include multiple imputation and inverse probability weighting, both of which assume that data are missing at random (MAR, missingness depends on observed data only<sup>23<\/sup>). When analysing repeat measurements, e.g. trajectories of change in BMI or cognitive function, all participants with at least one value (of the variable that has been assessed repeatedly) can be included in the main analyses under a MAR assumption. Ideally, one would want to use more than one method, with differing assumptions, for exploring selection bias (bias due to missing data). In the preliminary cross-generational analyses presented here we compared results from complete case analyses to those using multiple imputation (including using two different approaches to the latter) and found similar results across all three approaches which increases confidence in these not being biased.<\/i>\n<i>In general, research data that have been specifically collected as part of a study and linked data have complementary characteristics for exploring selection bias under the MAR assumption. Specific research data might predict missing values more strongly because it will have been collected at a similar time to variables with missing data (e.g. in sweeps of data collection) and will include some repeat collections of data obtained using the same methods across time. Such data area also likely to be obtained more extensively (e.g. several domains of socioeconomic position) and more precisely (e.g. DEXA scan total and body region fat and lean mass). However, specific research data may have more missingness due to loss to follow-up or some measures not being completed by some participants. Linked data is often available on all (or a large proportion) of participants.<\/i>\n<i> All generations of ALSPAC have both repeatedly assessed research and record linked data on many variables (see<sup>1,2<\/sup> and Figure 3), which considerably enhance the possibility for exploring and controlling for potential selection bias. Currently, ALSPAC-G0 and ALSPAC-G1 have been linked to obstetric-, primary-, secondary- and community health data, mortality and migration records and education data. Recent geocoding has also provided linkage to detailed environmental and area socioeconomic data.<sup>25<\/sup> Similar linkage for ALSPAC-G2 and the new (to ALSPAC) partners of existing ALSPAC-G1 is on-going and will be regularly updated. These data can be informative for methods assuming data are MAR (e.g.<sup>22<\/sup>) and also where selection (missing information) is assumed to be missing not at random (i.e. where bias can remain even after taking into account observed predictors of missingness).<sup>24<\/sup> For example, using linked education data as proxy for IQ in ALSPAC-G1 it has been shown that methods assuming data were missing at random underestimated the strength of the association in comparison to the method assuming missing not at random.<sup>24<\/sup> In addition to linked data, genome-wide genetic data, which is available on a high proportion of ALSPAC-G0 and ALSPAC-G1 participants and will be on ALSPAC-G2 and the newly recruited partners of ALSPAC-G1, can be useful in exploring the nature of missing data.<sup>26<\/sup> <\/i>\n<i>Whilst we,<sup>22-24<\/sup> and others,(e.g.<sup>27<\/sup>) have used the methods described above extensively in ALSPAC-G0 and ALSPAC-G1, their use for cross-generational analyses, in particular where the inclusion of siblings varies considerably across generations, has not yet been extensively explored. As demonstrated with ALSPAC-G2, such cross-generational analyses by necessity either restrict analyses to a narrow age range of parents in one generation (e.g. if we were to only continue to recruit ALSPAC-G2 for the next 5-years) or have participants recruited over a long period of years in one generation (as we hope to do with ALSPAC-G2).<\/i>\n<i> In the cross-generational analyses presented here we have been careful to describe the results as preliminary and illustrative. We undertook complete case analyses and used multiple imputation to explore potential selection due to missing data in the cross-generational comparisons presented here (Table 3). We used two approaches to multiple imputation, one including both generations in the same imputation analyses and one that separated the G0\/G1 from the G1\/G2 generations for the initial imputation. We found results were very similar across all three approaches. However, for simplicity, in these illustrative analyses we restricted the maternal age range in both generations to that of the ALSPAC-G1\/G2 participants and to just one pregnancy per mother in each generation. This resulted in more of the G1\/G2 pregnancies being excluded from analyses than G0\/G1 pregnancies, which could have introduced bias that is not addressed in our multiple imputation analyses. In future research we plan to develop and use methods that can more fully explore the sources of selection bias in cross-generational analyses. This includes methods that combine multiple imputation and other missing data methods with&nbsp; methods that appropriately take account of the complex multi-level structure of the data (i.e. repeat pregnancies within women across generations and potentially including complex family structures of siblings, cousins, nieces, nephews and so on). These methods and related analysis code will be disseminated widely via the ALSPAC website and via links to this cohort profile. We hope that other researchers will also use the unique multi-generational data of ALSPAC for exploring the likely impact of selection bias on this and other human multigeneration cohorts.<\/i>\n<b><i>Representativeness of different generations to the target population<\/i><\/b>\n<i>The original women recruited to ALSPAC (i.e. ALSPAC-G0) had similar socioeconomic characteristics and ethnic backgrounds to those of the population of the former county of Avon (South West region of England).<sup>2<\/sup> They were more affluent and more likely to be White European than the whole of the UK in the early 1990s, which reflects the difference between the population in the South West of England where the ALSAPC resource is based and the rest of the UK.<sup>2<\/sup> At age 16-years their children (ALSPAC-G1 \u2013 i.e. the parents of ALSPAC-G2) had somewhat higher school test scores than the national average, with this difference increasing in sub-samples based on increasing loss to follow-up.<sup>1 <\/sup>As recruitment of ALSPAC-G2 has been higher for those whose ALSPAC-G1 parents were most engaged with the study it is possible that over time ALSPAC-G2 will show greater differences to national averages. We will monitor that through the extensive record linkage, including data on education. Whilst selection bias and representativeness are sometimes treated as if they reflect the same concept, here we consider that lack of representativeness to the target population may bias estimates of prevalence but does not necessarily influence association or causal effect estimates. There have been considerable debates on this topic (see for example<sup>28-33<\/sup>), with evidence that representativeness is not necessary to obtain unbiased estimates of association or effect,<sup>28-32<\/sup> though this will depend on the extent of selection into a study and the availability of data to predict this.<sup>20,33<\/sup> In this regard a key strength of ALSPAC-G2 is the existence of extensive data on previous generations and linkage to national data across all generations.<sup> 33 <\/sup><\/i>\n<i>With respect to analyses that might only be done within the ALSPAC-G2 cohort currently, we would consider association or causal effect estimates to generalise to couples becoming pregnant at a relatively young age. This would be consistent with other pregnancy studies that recruit at younger ages and\/or only recruit nulliparous women (such as<sup>34,35<\/sup>). Once we have had the opportunity to invite all of ALSPAC-G2 we will have a unique study in which to explore the relationships of age at which couples become pregnant\/parents across a wide age range within a relatively homogeneous cohort in terms of date and place of birth.\u201d (<\/i>\n<b>Discussion section of the paper page 24-28 of clean version<\/b>\n<i>)<\/i>  As discussed, ALSPAC is a local cohort study but of course participants move away. Participants are offered travel costs to the Bristol based clinics for the assessments so technically even if a cohort member of ALSPAC-G1 has moved away, they can participate. However, it would be useful to note the proportion of ALSPAC-G1 parents who are no longer living in Bristol and response rates to ALSPAC-G2 across those who have and have not moved away. This would help quantify any potential selection bias on this particular issue.  \n<b><u>Response<\/u><\/b>We agree this is a really useful suggestion and have now added to Table 2 in the revision a row that shows the proportion of people living in the Bristol area (BS postcode) at the time that they were invited to join the study by whether they participated or did not. There is no difference between the two, with 82% of the respondents and 83% of the non-respondents being resident in Bristol. Before doing these additional analyses (i.e. comparing Bristol residency between respondents and non-respondents) we \n<i>a priori<\/i> agreed that this was valuable analyses to undertake and that residence in Bristol at the time of recruitment could be a predictor of missingness and therefore \n<i>a priori<\/i> we would include it in the prediction models for the multiple imputation analyses that are presented in both Table 2 and Table 3. As expected given the similarity between respondents and non-respondents with respect to living in Bristol none of the multiple imputation results changed in any substantial way but there are small changes that have been made to the results in these tables in the revised version.  The comparisons of young mothers across the two generations do not discuss the fact that selection into early motherhood may be driven by different (unobserved) factors across the two generations. If, for example, attitudes have changed and early childbirth is now viewed more negatively by women in the more recent generation (even when comparing women with similar characteristics), this may lead to selection bias into young parenthood. This may be particularly relevant for some outcomes, such as maternal postnatal depression. There is no way to account for this empirically but it might be noted.  \n<b><u>Response<\/u><\/b>We agree, and in the extended section on selection bias (see above) this is now noted more clearly.  Table 3. The comparison of age at first pregnancy between the two cohorts (ALSPAC-G0\/G1, and ALSPAC-G1\/G2) depends crucially on restricting the sample to age 19-26. Otherwise the mean age at first birth statistic will be misleading because, given that ALSPAC-G1 is a cohort of roughly the same age, we will not know the true mean age of mother at birth of first child until some more years have passed and others in the ALSPAC-G1 generation have had their children. This is carefully footnoted but I wonder whether the age restriction might usefully appear in the title of the table.  \n<b><u>Response<\/u><\/b>Thank you, we have now made this change to the title of Table 3&nbsp;\n<b>Minor issues<\/b>  The abstract mentions data collection and notes that ALSPAC-G2 collects a wide range of \u201csocial\u201d data. In the main body of the text this is sometimes changed to \u201csocioeconomic\u201d. The latter is more accurate and might replace the word \u201csocial\u201d throughout.  \n<b><u>Response<\/u><\/b>We have changed most of the times we use \u2018social\u2019 to socioeconomic; in particular we have done this when we are referring to socioeconomic data, such as occupational social class or education. We still use social for e.g. \u2018social life\u2019 and \u2018social media\u2019.  Bullet 4 on page 3 is somewhat strangely worded. The intention of ALSPAC-G2 is to undertakeintergenerational analyses of the factors influencing decisions by ALSPAC-G0 compared to ALSPAC-G1 about whether, and when, to start a family, parenting patterns, child development, parental and child mental and physical health and \u2018work-family\/outside of work\u2019 balance. This main aim might come earlier in the bullet before discussing the variables that are in the data?  \n<b><u>Response<\/u><\/b>We have made this change and reworded this section in the revised paper so that we hope it is now clearer. This now reads as follows:\n<i>\u201c3.<\/i>\n<i> Understanding the impact of major changes that have occurred over the last 20-25 years and how these have influenced differences between ALSPAC-G0 and ALSPAC-G1 in relation to whether, and when, to start a family, parenting patterns, and \u2018work-family\/outside of work\u2019 balance and of these on differences in health and development of ALSPAC-G1 and ALSPAC-G2. These changes have been unprecedented and include changes in environmental (e.g. climate change, air-pollution), societal (e.g. different methods of communication, changing patterns of between and within country migration), political (e.g. distrust of experts, departures from traditional political parties, changes in gender politics, retirement age and types of employment) technological (e.g. widespread use of information technology and the emergence of artificial intelligence) and lifestyle (e.g. sedentary behaviour, reduced smoking and alcohol consumption, increased vaping) factors. For example, ALSPAC-G0 participants mostly finished their education and started their careers before the internet was invented. In its current recognisable form, widespread availability of the internet began in 1990 just before ALSPAC-G1 were born but its reach into primary schools will have been limited for this generation. In contrast, ALSPAC-G2 are likely to spend their whole lives extensively using the internet for their education, work and social life.\u201d<\/i> \n<b>(introduction section page 3)<\/b> &nbsp;  There is a typographical error on page 3 under the section \u201cWho is in the cohort?\u201d. \u201cWe aim is to recruit..\u201d  \n<b><u>Response<\/u><\/b>This has now been corrected, thank you.  References to the Bristol IVF and congenital heart disease cohorts might be provided to help the reader.  \n<b><u>Response<\/u><\/b>We have now added these for the cleft lip and palate and the IVF cohort and note that currently there is no website for the congenital heart disease cohort.&nbsp;","URL":"https:\/\/wellcomeopenresearch.org\/articles\/4-36\/v2#referee-comment-4005","order":1,"name":"referee-comment-4005","label":"Referee Comment","group":{"name":"article-reports","label":"Article Reports"}},{"value":"10.21956\/wellcomeopenres.17079.r37442, Susan M. Morton, Growing Up in New Zealand, School of Population Health, \u00a0University of Auckland, Auckland, New Zealand, 20 Dec 2019, version 2, indexed","URL":"https:\/\/wellcomeopenresearch.org\/articles\/4-36\/v2#referee-response-37442","order":2,"name":"referee-response-37442","label":"Referee Report","group":{"name":"article-reports","label":"Article Reports"}},{"value":"ALSPAC receives core-funding from the University of Bristol, Wellcome Trust and MRC (102215\/2\/13\/2). This study was supported by the National Institute for Health Research NIHR Biomedical Research Centre at the University Hospitals Bristol NHS National Health Service Foundation Trust and the University of Bristol, the US National Institute for Health (grant R01 DK10324), and the European Research Council under the European Union\u2019s Seventh Framework Programme (grant FP\/2007-2013) \/ European Research Council ERC Grant Agreements (Grant grants refs 758813 MHINT and 669545 DevelopObese). Medtronic Limited provided support for the pilot study of continuous glucose monitoring. The Wellcome Trust Institutional Strategic Support Fund and Elizabeth Blackwell Institute at the University of Bristol provided support for the novel diet methods pilot study. LACM is funded by a University of Bristol Vice Chancellor\u2019s Fellowship. DAL, RP, LACM, AS and KT work in a Unit that is supported by the UK Medical Research Council (MC_UU_00011\/6 and MC_UU_00011\/3) and University of Bristol, and DAL is a National Institute for Health Research Senior Investigator (NF-SI-0611-10196). The funders had no role in designing the study, collecting or analysing data or contributing to writing the paper. The views expressed in this paper are those of the authors and not necessarily any funding body. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.","order":3,"name":"grant-information","label":"Grant Information"},{"value":"This is an open access article distributed under the terms of the Creative Commons Attribution Licence, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.","order":0,"name":"copyright-info","label":"Copyright"}]}}