# paper.pdf ## Metadata - PDFFormatVersion=1.6 - Language=en-US - IsLinearized=true - IsAcroFormPresent=false - IsXFAPresent=false - IsCollectionPresent=false - IsSignaturesPresent=false - CreationDate=D:20260529161255+05'30' - Creator=Adobe InDesign 17.4 (Windows) - ModDate=D:20260529162901+05'30' - Producer=Adobe PDF Library 16.0.7 - Trapped=False - xmpmm:instanceid=uuid:e2450fdf-ef83-42ff-a85d-b6d532daa94e - xmpmm:originaldocumentid=xmp.did:A33916180F2068118083ACD1B377E157 - xmpmm:documentid=xmp.id:5a4b4d76-d699-5e41-b36b-236e91d93e0e - xmpmm:renditionclass=proof:pdf - xmpmm:derivedfrom=xmp.iid:b1b88a03-2167-5b44-a060-367fcc96d427xmp.did:b1b88a03-2167-5b44-a060-367fcc96d427xmp.did:A33916180F2068118083ACD1B377E157default - xmpmm:history=convertedfrom application/x-indesign to application/pdfAdobe InDesign 17.4 (Windows)/2026-05-29T16:12:55+05:30 - xmp:createdate=2026-05-29T16:12:55+05:30 - xmp:modifydate=2026-05-29T16:29:01+05:30 - xmp:metadatadate=2026-05-29T16:29:01+05:30 - xmp:creatortool=Adobe InDesign 17.4 (Windows) - xmprights:marked=False - photoshop:modifydate=2019-01-31T16:16:25 - photoshop:createdate=2019-01-24T11:43:46 - dc:format=application/pdf - pdf:producer=Adobe PDF Library 16.0.7 - pdf:trapped=False - extensisfontsense:slug=3023672065GraphikOpenType - PS3023672065--0Graphik-Bold1.0011904271029GraphikOpenType - PS1904271029--0Graphik-Regular1.0012428528031News Gothic MT OTOpenType - PS2428528031Monotype Typography0NewsGothicMTOT-Regular1.000543760319News Gothic MT OTOpenType - PS543760319Monotype Typography0NewsGothicMTOT-Bold1.0002785939051KnockoutOpenType - PS2785939051--0Knockout-68FullFeatherwt1.200423856041Graphik X CondOpenType - PS423856041--0GraphikXCond-Super1.0001526035435Graphik X CondOpenType - PS1526035435--0GraphikXCond-Bold1.0003093728306Graphik CondOpenType - PS3093728306--0GraphikCond-Black1.0011341118539Graphik CondOpenType - PS1341118539--0GraphikCond-Regular1.0011385893531GraphikOpenType - PS1385893531--0Graphik-Semibold1.0012986479693Graphik CompactOpenType - PS2986479693--0GraphikCompact-Regular1.0024223138361Graphik CompactOpenType - PS4223138361--0GraphikCompact-Semibold1.0023673047546Harding SevenOpenType - PS3673047546--0HardingSeven-Extrabold0.000103712566Harding SevenOpenType - PS103712566--0HardingSeven-Roman0.0002232452979Harding SevenOpenType - PS2232452979--0HardingSeven-Medium0.000631916388Harding SevenOpenType - PS631916388--0HardingSeven-Bold0.000157629484Publico BannerOpenType - PS157629484--0PublicoBanner-Black1.005 ## Contents ### Page 1 nature human behaviourhttps://doi.org/10.1038/s41562-026-02477-6ArticleSocial determinants of health and epigenetic clocks: a systematic review and meta- analysis of 140 studiesIn the format provided by the authors and uneditedSupplementary information ### Page 2 1Table of ContentsSupplemental Figure 1. World map showing the countries included in this meta-analysis...........................................2Supplemental Figure 2. Funnel plots to assess publication bias in the relationship between socioeconomic statusand epigenetic-clock-measured biological aging. .......................................................................................................3Supplemental Figure 3. Meta-analytic associations between socioeconomic status (SES) and epigenetic clocks,excluding effect sizes from the Health and Retirement Study (HRS). ..........................................................................4Supplemental Figure 4. Meta-analytic associations between socioeconomic status (SES) and epigenetic clocks,excluding effect sizes from meta-analyses. ................................................................................................................5Supplemental Figure 5. Meta-analytic associations between socioeconomic status (SES) and epigenetic clocks,excluding effect sizes from preprints. .........................................................................................................................6Supplemental Figure 6. Meta-analytic associations between socioeconomic status (SES) and epigenetic clocks,excluding effect sizes with small sample sizes (n <100). .............................................................................................7Supplemental Figure 7. Meta-analytic associations between intergenerational socioeconomic status (SES) andepigenetic clocks. ......................................................................................................................................................8Supplemental Figure 8. Associations between repeated measures of socioeconomic status (SES) and epigeneticclocks. .......................................................................................................................................................................9Supplemental Figure 9. Funnel plots to assess publication bias in the relationship between Black-White comparisonsand epigenetic-clock-measured biological aging. .....................................................................................................11Supplemental Figure 10. Funnel plots to assess publication bias in the relationship between Latinx-Whitecomparisons and epigenetic-clock-measured biological aging. ................................................................................12Supplemental Figure 11. Meta-analytic associations between race/ethnicity and epigenetic clocks, excluding effectsizes from preprints. ................................................................................................................................................13Supplemental Figure 12. Meta-analytic associations between race/ethnicity and epigenetic clocks, excluding effectsizes with small sample sizes (n<100). .....................................................................................................................15Supplemental Figure 13. Meta-analytic associations between race/ethnicity and epigenetic clocks, excluding effectsizes from the Health and Retirement Study.............................................................................................................17Supplemental Figure 14. Meta-analytic associations between socioeconomic status (SES) and epigenetic clocks,separately for EPIC and 450k arrays. .......................................................................................................................19Supplemental Figure 15. Heatmap showing the number of effect sizes for epigenetic-clock-measured biologicalaging (x-axis) and social determinants of health (y-axis). ...........................................................................................21Supplemental Figure 16. Meta-analytic associations between socioeconomic status (SES) and epigenetic clocks –biweight midcorrelation (bicor) estimates..................................................................................................................22Supplemental Figure 17. Leave-One-Cohort-Out Analyses for the meta-analytic associations between SES andepigenetic clocks. ....................................................................................................................................................33Supplemental Figure 18. Leave-One-Cohort-Out Analyses for the meta-analytic associations between Black- andWhite-identifying groups and epigenetic clocks. .......................................................................................................35Supplemental Figure 19. Leave-One-Cohort-Out Analyses for the meta-analytic associations between Latinx- andWhite-identifying groups and epigenetic clocks. .......................................................................................................38Supplemental Figure 20. Pearson’s r values for the associations between SES/race/ethnicity and DNAm generationsand measures of biological aging ……………….………………………………………………………………………….39 ### Page 3 2Supplemental Figure 1. World map showing the countries included in thismeta-analysis. Red color indicates that there is at least one effect size from thatcountry (including contributions to meta-analyses). Countries colored in gray do notcontribute any effect sizes. Map data from Natural Earth (public domain),visualisation created by authors in R. ### Page 4 3Supplemental Figure 2. Funnel plots to assess publication bias in therelationship between socioeconomic status and epigenetic-clock-measuredbiological aging.These plots display individual effect sizes on the x-axis and their standard errors onthe y-axis. According to the results of Egger’s regression test, there is no evidence ofpublication bias for any of the epigenetic-clock-measured biological aging (all ps <0.05; Supplemental Table 5A). ### Page 5 4Supplemental Figure 3. Meta-analytic associations between socioeconomicstatus (SES) and epigenetic clocks, excluding effect sizes from the Healthand Retirement Study (HRS). Forest plot depicts the meta-analytic associationsbetween SES and generations of epigenetic clocks, after 174 effect sizes contributedby HRS are removed. Green panels indicate the meta-analyzed effect size estimatesfor first-, second-, and third-generation clocks, followed by individual clocks withineach category. Analyses were conducted using two-sided Fisher-Z transformationand later transformed to Pearson’s r to ease interpretation. Effect size estimates arebounded by 95% confidence intervals (CIs).* denotes significant difference from 0 at p< 0.05 (not corrected for multiple comparisons). The size of the boxes in the middlepanel represents the precision of the effect size estimate (i.e., the smaller the CI, themore precise the estimate). ### Page 6 5Supplemental Figure 4. Meta-analytic associations between socioeconomicstatus (SES) and epigenetic clocks, excluding effect sizes from meta-analyses. Forest plot depicts the meta-analytic associations between SES andgenerations of epigenetic clocks, after 4 effect sizes contributed by meta-analyses areremoved. Green panels indicate the meta-analyzed effect size estimates for first-,second-, and third-generation clocks, followed by individual clocks within eachcategory. Analyses were conducted using two-sided Fisher-Z transformation and latertransformed to Pearson’s r to ease interpretation. Effect size estimates are boundedby 95% confidence intervals (CIs).* denotes significant difference from 0 at p < 0.05(not corrected for multiple comparisons). The size of the boxes in the middle panelrepresents the precision of the effect size estimate (i.e., the smaller the CI, the moreprecise the estimate). ### Page 7 6Supplemental Figure 5. Meta-analytic associations between socioeconomicstatus (SES) and epigenetic clocks, excluding effect sizes from preprints.Forest plot depicts the meta-analytic associations between SES and generations ofepigenetic clocks, after 27 effect sizes contributed by preprints are removed. Greenpanels indicate the meta-analyzed effect size estimates for first-, second-, and third-generation clocks, followed by individual clocks within each category. Analyses wereconducted using two-sided Fisher-Z transformation and later transformed toPearson’s r to ease interpretation. Effect size estimates are bounded by 95%confidence intervals (CIs).* denotes significant difference from 0 at p < 0.05 (notcorrected for multiple comparisons). The size of the boxes in the middle panelrepresents the precision of the effect size estimate (i.e., the smaller the CI, the moreprecise the estimate). ### Page 8 7Supplemental Figure 6. Meta-analytic associations between socioeconomicstatus (SES) and epigenetic clocks, excluding effect sizes with small samplesizes (n <100).Forest plot depicts the meta-analytic associations between SES and generations ofepigenetic clocks, after 85 effect sizes with small sample sizes (n<100) are removed.Green panels indicate the meta-analyzed effect size estimates for first-, second-, andthird-generation clocks, followed by individual clocks within each category. Analyseswere conducted using two-sided Fisher-Z transformation and later transformed toPearson’s r to ease interpretation. Effect size estimates are bounded by 95%confidence intervals (CIs).* denotes significant difference from 0 at p < 0.05 (notcorrected for multiple comparisons). The size of the boxes in the middle panelrepresents the precision of the effect size estimate (i.e., the smaller the CI, the moreprecise the estimate). ### Page 9 8Supplemental Figure 7. Meta-analytic associations between intergenerationalsocioeconomic status (SES) and epigenetic clocks. Forest plot depicts themeta-analytic associations between intergenerational SES and epigenetic clocks.DNAm was measured at one time point. Intergenerational SES is measured by theintergenerational stability in SES, comparing parental occupation with currentoccupation, where the reference group is stably low SES and the target group isstably high SES. Yellow panels indicate the meta-analyzed effect size estimates forfirst- and second-generation clocks, followed by individual clocks within eachcategory. Results are meta-analyzed as unstandardized regression coefficients usingtwo-sided hypothesis testing. Effect size estimates are bounded by 95% confidenceintervals (CIs).* denotes significant difference from 0 at p < 0.05 (not corrected formultiple comparisons). The size of the boxes in the middle panel represents theprecision of the effect size estimate (i.e., the smaller the CI, the more precise theestimate). ### Page 10 9A. Associations between socioeconomic status (SES) and repeated measures ofepigenetic clocks within individualsB. Associations between repeated measures of SES and repeated measures ofepigenetic clocks within individuals ### Page 11 10Supplemental Figure 8. Associations between repeated measures ofsocioeconomic status (SES) and epigenetic clocks. Forest plot in section Adepicts the meta-analytic associations between SES and repeated measures ofepigenetic clocks within individuals. Forest plot in section B depicts the meta-analyticassociations of repeated measures of SES and repeated measures of epigeneticclocks within individuals. Yellow panels indicate the meta-analyzed effect sizeestimates for first-, second-, and third-generation clocks, followed by individual clockswithin each category. As effect sizes for change data are mainly reported asunstandardized regression coefficients, and SDs for change in age acceleration andSES are not reported in any of the studies, change results are meta-analyzed asunstandardized regression coefficients separately using two-sided hypothesis testing.Effect size estimates are bounded by 95% confidence intervals (CIs).* denotessignificant difference from 0 at p < 0.05 (not corrected for multiple comparisons). Thesize of the boxes in the middle panel represents the precision of the effect sizeestimate (i.e., the smaller the CI, the more precise the estimate). ### Page 12 11Supplemental Figure 9. Funnel plots to assess publication bias in therelationship between Black-White comparisons and epigenetic-clock-measured biological aging.These plots display individual effect sizes on the x-axis and their standard errors onthe y-axis. Tests could not be run for Zhang clock due to model constraints. Resultsof Egger’s regression tests can be found in Supplemental Table 5C. ### Page 13 12Supplemental Figure 10. Funnel plots to assess publication bias in therelationship between Latinx-White comparisons and epigenetic-clock-measured biological aging.These plots display individual effect sizes on the x-axis and their standard errors onthe y-axis. Results of Egger’s regression tests can be found in Supplemental Table5E. ### Page 14 13A. Meta-analytic associations between Black-White comparisons and epigeneticclocks (excluding preprints)B. Meta-analytic associations between Latinx-White comparisons and epigeneticclocks (excluding preprints) ### Page 15 14Supplemental Figure 11. Meta-analytic associations between race/ethnicityand epigenetic clocks, excluding effect sizes from preprints. Forest plot depictsthe meta-analytic associations between (A) Black-White comparisons and generationsof epigenetic clocks and (B) Latinx-White comparisons and generations of epigeneticclocks, after 3 effect sizes contributed by preprints are removed from both datasets.Blue panels indicate the meta-analyzed effect size estimates for first-, second-, andthird-generation clocks, followed by individual clocks within each category. Analyseswere conducted using two-sided Fisher-Z transformation and later transformed toCohen’s d to ease interpretation. Effect size estimates are bounded by 95% confidenceintervals (CIs).* denotes significant difference from 0 at p < 0.05 (not corrected formultiple comparisons). The size of the boxes in the middle panel represents theprecision of the effect size estimate (i.e., the smaller the CI, the more precise theestimate). ### Page 16 15A. Meta-analytic associations between Black-White comparisons and epigeneticclocks (excluding effect sizes with small sample sizes)B. Meta-analytic associations between Latinx-White comparisons and epigeneticclocks (excluding effect sizes with small sample sizes)Supplemental Figure 12. Meta-analytic associations between race/ethnicityand epigenetic clocks, excluding effect sizes with small sample sizes (n<100).Forest plot depicts the meta-analytic associations between (A) Black-Whitecomparisons and generations of epigenetic clocks and (B) Latinx-White comparisonsand generations of epigenetic clocks, after 5 and 2 effect sizes with small sample sizes ### Page 17 16are removed, respectively. Blue panels indicate the meta-analyzed effect size estimatesfor first-, second-, and third-generation clocks, followed by individual clocks within eachcategory. Analyses were conducted using two-sided Fisher-Z transformation and latertransformed to Cohen’s d to ease interpretation. Effect size estimates are bounded by95% confidence intervals (CIs).* denotes significant difference from 0 at p < 0.05 (notcorrected for multiple comparisons). The size of the boxes in the middle panelrepresents the precision of the effect size estimate (i.e., the smaller the CI, the moreprecise the estimate). ### Page 18 17A. Meta-analytic associations between Black-White comparisons and epigeneticclocks (excluding effect sizes from the Health and Retirement Study)B. Meta-analytic associations between Latinx-White comparisons and epigeneticclocks (excluding effect sizes from the Health and Retirement Study) ### Page 19 18Supplemental Figure 13. Meta-analytic associations between race/ethnicityand epigenetic clocks, excluding effect sizes from the Health and RetirementStudy. Forest plot depicts the meta-analytic associations between (A) Black-Whitecomparisons and generations of epigenetic clocks and (B) Latinx-White comparisonsand generations of epigenetic clocks, after 30 and 23 effect sizes contributed by theHealth and Retirement Study are removed, respectively. Blue panels indicate the meta-analyzed effect size estimates for first-, second-, and third-generation clocks, followedby individual clocks within each category. Analyses were conducted using two-sidedFisher-Z transformation and later transformed to Cohen’s d to ease interpretation.Effect size estimates are bounded by 95% confidence intervals (CIs).* denotessignificant difference from 0 at p < 0.05 (not corrected for multiple comparisons). Thesize of the boxes in the middle panel represents the precision of the effect size estimate(i.e., the smaller the CI, the more precise the estimate). ### Page 20 19A. Meta-analytic associations between socioeconomic status (SES) andepigenetic clocks (EPIC array)B. Meta-analytic associations between SES and epigenetic clocks (450k array) ### Page 21 20Supplemental Figure 14. Meta-analytic associations between socioeconomicstatus (SES) and epigenetic clocks, separately for EPIC and 450k arrays.Forest plot depicts the meta-analytic associations of SES and epigenetic clocks forEPIC array (section A) and 450k array (section B). Purple panels indicate the meta-analyzed effect size estimates for first-, second-, and third-generation clocks, followedby individual clocks within each category. Analyses were conducted using two-sidedFisher-Z transformation and later transformed to Cohen’s d to ease interpretation.Effect size estimates are bounded by 95% confidence intervals (CIs).* denotessignificant difference from 0 at p < 0.05 (not corrected for multiple comparisons). Thesize of the boxes in the middle panel represents the precision of the effect sizeestimate (i.e., the smaller the CI, the more precise the estimate). ### Page 22 21Supplemental Figure 15. Heatmap showing the number of effect sizes forepigenetic-clock-measured biological aging (x-axis) and social determinants ofhealth (y-axis). The number of effect sizes per category ranges from 60 (dark blue) to 1(light blue). White squares indicate no effect sizes. ### Page 23 22Supplemental Figure 16. Meta-analytic associations between socioeconomicstatus (SES) and epigenetic clocks – biweight midcorrelation (bicor)estimates. Forest plot depicts the meta-analytic associations between SES andepigenetic clocks. As a reliable conversion from bicor to Pearson’s r could not befound, bicor effect sizes are meta-analyzed separately using two-sided hypothesistesting. Orange panels indicate the meta-analyzed effect size estimates for first- andsecond-generation clocks, followed by individual clocks within each category. Effectsize estimates are bounded by 95% confidence intervals (CIs). Effect size estimatesare bounded by 95% confidence intervals (CIs).* denotes significant difference from 0at p < 0.05 (not corrected for multiple comparisons). The size of the boxes in themiddle panel represents the precision of the effect size estimate (i.e., the smaller theCI, the more precise the estimate). ### Page 24 23A. Meta-analytic associations between SES and epigenetic clocks (excludingeffect sizes from Avon Longitudinal Study of Parents and Children)B. Meta-analytic associations between SES and epigenetic clocks (excludingeffect sizes from Birmingham Youth Violence Study) ### Page 25 24C. Meta-analytic associations between SES and epigenetic clocks (excludingeffect sizes from Canadian Longitudinal Study on Aging)D. Meta-analytic associations between SES and epigenetic clocks (excludingeffect sizes from Coronary Artery Risk Development in Young Adults) ### Page 26 25E. Meta-analytic associations between SES and epigenetic clocks (excludingeffect sizes from E-Risk)F. Meta-analytic associations between SES and epigenetic clocks (excludingeffect sizes from Family and Community Health Study) ### Page 27 26G. Meta-analytic associations between SES and epigenetic clocks (excludingeffect sizes from Female Growth and Development Study)H. Meta-analytic associations between SES and epigenetic clocks (excludingeffect sizes from Finnish Twin Cohort) ### Page 28 27I. Meta-analytic associations between SES and epigenetic clocks (excludingeffect sizes from Generation Scotland)J. Meta-analytic associations between SES and epigenetic clocks (excludingeffect sizes from Lothian Birth Cohort 1936) ### Page 29 28K. Meta-analytic associations between SES and epigenetic clocks (excludingeffect sizes from Multi-Ethnic Study of Atherosclerosis)L. Meta-analytic associations between SES and epigenetic clocks (excludingeffect sizes from MRC National Survey of Health and Development) ### Page 30 29M. Meta-analytic associations between SES and epigenetic clocks (excludingeffect sizes from My Body My Story)N. Meta-analytic associations between SES and epigenetic clocks (excludingeffect sizes from Project Viva) ### Page 31 30O. Meta-analytic associations between SES and epigenetic clocks (excludingeffect sizes from Swiss Kidney Project on Genes in Hypertension)P. Meta-analytic associations between SES and epigenetic clocks (excludingeffect sizes from Texas Twin Project) ### Page 32 31Q. Meta-analytic associations between SES and epigenetic clocks (excludingeffect sizes from The Irish Longitudinal Study on Ageing)R. Meta-analytic associations between SES and epigenetic clocks (excludingeffect sizes from Understanding Society) ### Page 33 32S. Meta-analytic associations between SES and epigenetic clocks (excludingeffect sizes from Women’s Health Initiative)Supplemental Figure 17. Leave-One-Cohort-Out Analyses for the meta-analyticassociations between SES and epigenetic clocks. Forest plot depicts the meta-analytic associations between SES and generations of epigenetic clocks excludingeffect sizes from 19 cohorts that contributed at least 3 studies or at least 25 effectsizes. Green panels indicate the meta-analyzed effect size estimates for first-, second-,and third-generation clocks, followed by individual clocks within each category.Analyses were conducted using two-sided Fisher-Z transformation and latertransformed to Pearson’s r to ease interpretation. Effect size estimates are bounded by95% confidence intervals (CIs).* denotes significant difference from 0 at p < 0.05 (notcorrected for multiple comparisons). The size of the boxes in the middle panelrepresents the precision of the effect size estimate (i.e., the smaller the CI, the moreprecise the estimate). ### Page 34 33A. Meta-analytic associations between Black- and White-identifying groups andepigenetic clocks excluding the Birmingham Youth StudyB. Meta-analytic associations between Black- and White-identifying groups andepigenetic clocks excluding Project Viva ### Page 35 34C. Meta-analytic associations between Black- and White-identifying groups andepigenetic clocks excluding the Texas Twin ProjectD. Meta-analytic associations between Black- and White-identifying groups andepigenetic clocks excluding the Women’s Health Initiative ### Page 36 35Supplemental Figure 18. Leave-One-Cohort-Out Analyses for the meta-analytic associations between Black- and White-identifying groups andepigenetic clocks.Forest plots depict the meta-analytic Black-White comparisons and epigenetic clocksexcluding effect sizes from cohorts that contributed at least 3 studies or at least 25effect sizes. Excluding (A) Birmingham Youth Study, (B) Project Viva, (C) Texas TwinProject, (D) Women’s Health Initiative. Blue panels indicate the meta-analyzed effectsize estimates for first-, second-, and third-generation clocks, followed by individualclocks within each category. Analyses were conducted using two-sided Fisher-Ztransformation and later transformed to Cohen’s d to ease interpretation. Effect sizeestimates are bounded by 95% confidence intervals (CIs).* denotes significantdifference from 0 at p < 0.05 (not corrected for multiple comparisons). The size of theboxes in the middle panel represents the precision of the effect size estimate (i.e., thesmaller the CI, the more precise the estimate). ### Page 37 36A. Meta-analytic associations between Latinx- and White-identifying groups andepigenetic clocks excluding Project VivaB. Meta-analytic associations between Latinx- and White-identifying groups andepigenetic clocks excluding the Texas Twin Project ### Page 38 37C. Meta-analytic associations between Black- and White-identifying groups andepigenetic clocks excluding the Women’s Health InitiativeSupplemental Figure 19. Leave-One-Cohort-Out Analyses for the meta-analytic associations between Latinx- and White-identifying groups andepigenetic clocks.Forest plots depict the meta-analytic Latinx-White comparisons and epigenetic clocksexcluding effect sizes from cohorts that contributed at least 3 studies or at least 25effect sizes. Excluding (A) Project Viva, (B) Texas Twin Project, (C) Women’s HealthInitiative. Blue panels indicate the meta-analyzed effect size estimates for first-,second-, and third-generation clocks, followed by individual clocks within eachcategory. Analyses were conducted using two-sided Fisher-Z transformation and latertransformed to Cohen’s d to ease interpretation. Effect size estimates are bounded by95% confidence intervals (CIs).* denotes significant difference from 0 at p < 0.05 (notcorrected for multiple comparisons). The size of the boxes in the middle panelrepresents the precision of the effect size estimate (i.e., the smaller the CI, the moreprecise the estimate). ### Page 39 38Supplemental Figure 20. Pearson’s r values for the associations betweenSES/race/ethnicity and DNAm generations and measures of biological aging. Note:Estimates reflect Pearson’s r to compare estimates across social determinants ofhealth. Estimate’s in Cohen’s d can be found in Figure 4.