Robbee Wedow’s research while affiliated with Massachusetts General Hospital and other places

Data within biobanks capture broad yet detailed indices of human variation, but biobank-wide insights can be difficult to extract due to complexity and scale. Here, using large-scale factor analysis, we distill hundreds of variables (diagnoses, assessments and survey items) into 35 latent constructs, using data from unrelated individuals with predominantly estimated European genetic ancestry in UK Biobank. These factors recapitulate known disease classifications, disentangle elements of socioeconomic status, highlight the relevance of psychiatric constructs to health and improve measurement of pro-health behaviours. We go on to demonstrate the power of this approach to clarify genetic signal, enhance discovery and identify associations between underlying phenotypic structure and health outcomes. In building a deeper understanding of ways in which constructs such as socioeconomic status, trauma, or physical activity are structured in the dataset, we emphasize the importance of considering the interwoven nature of the human phenome when evaluating public health patterns.

Introduction Early mid-life is marked by accumulating risks for cardiometabolic illness linked to health-risk behaviors like nicotine use. Identifying polygenic indices (PGI) has enriched scientific understanding of the cumulative genetic contributions to behavioral and cardiometabolic health, though few studies have assessed these associations alongside socioeconomic (SES) and lifestyle factors. Methods Drawing on data from 2,337 individuals from the United States participating in the National Longitudinal Study of Adolescent to Adult Health, the current study assesses the fraction of variance in five related outcomes – use of conventional and electronic cigarettes, body mass index (BMI), waist circumference, and glycosylated hemoglobin (A1c) – explained by PGI, SES, and lifestyle. Results Regression models on African ancestry (AA) and European ancestry (EA) subsamples reveal that the fraction of variance explained by PGI ranges across outcomes. While adjusting for sex and age, PGI explained 3.5%, 2.2%, and 0% in the AA subsample of variability in BMI, waist circumference, and A1c, respectively (in the EA subsample these figures were 7.7%, 9.4%, and 1.3%). The proportion of variance explained by PGI in nicotine-use outcomes is also variable. Results further indicate that PGI and SES are generally complementary, accounting for more variance in the outcomes when modeled together versus separately. Conclusions PGI are gaining attention in population health surveillance, but polygenic variability might not align clearly with health differences in populations or surpass SES as a fundamental cause of health disparities. We discuss future steps in integrating PGI and SES to refine population health prediction rules. Implications Study findings point to the complementary relationship of polygenic indices (PGI) and socioeconomic indicators in explaining population variance in nicotine outcomes and cardiometabolic wellness. Population health surveillance and prediction rules would benefit from the combination of information from both polygenic and socioeconomic risks. Additionally, the risk for electronic cigarette use among users of conventional cigarettes may have a genetic component tied to the cumulative genetic propensity for heavy smoking. Further research on PGI for vaping is needed.

Response to survey questionnaires is vital for social and behavioural research, and most analyses assume full and accurate response by participants. However, nonresponse is common and impedes proper interpretation and generalizability of results. We examined item nonresponse behaviour across 109 questionnaire items in the UK Biobank (N = 360,628). Phenotypic factor scores for two participant-selected nonresponse answers, ‘Prefer not to answer’ (PNA) and ‘I don’t know’ (IDK), each predicted participant nonresponse in follow-up surveys (incremental pseudo-R² = 0.056), even when controlling for education and self-reported health (incremental pseudo-R² = 0.046). After performing genome-wide association studies of our factors, PNA and IDK were highly genetically correlated with one another (rg = 0.73 (s.e. = 0.03)) and with education (rg,PNA = −0.51 (s.e. = 0.03); rg,IDK = −0.38 (s.e. = 0.02)), health (rg,PNA = 0.51 (s.e. = 0.03); rg,IDK = 0.49 (s.e. = 0.02)) and income (rg,PNA = –0.57 (s.e. = 0.04); rg,IDK = −0.46 (s.e. = 0.02)), with additional unique genetic associations observed for both PNA and IDK (P < 5 × 10⁻⁸). We discuss how these associations may bias studies of traits correlated with item nonresponse and demonstrate how this bias may substantially affect genome-wide association studies. While the UK Biobank data are deidentified, we further protected participant privacy by avoiding exploring non-response behaviour to single questions, assuring that no information can be used to associate results with any particular respondents.

Genetic liability to substance use disorders can be parsed into loci that confer general or substance-specific addiction risk. We report a multivariate genome-wide association meta-analysis that disaggregates general and substance-specific loci for published summary statistics of problematic alcohol use, problematic tobacco use, cannabis use disorder, and opioid use disorder in a sample of 1,025,550 individuals of European descent and 92,630 individuals of African descent. Nineteen independent SNPs were genome-wide significant (P < 5e-8) for the general addiction risk factor (addiction-rf), which showed high polygenicity. Across ancestries, PDE4B was significant (among other genes), suggesting dopamine regulation as a cross-substance vulnerability. An addiction-rf polygenic risk score was associated with substance use disorders, psychopathologies, somatic conditions, and environments associated with the onset of addictions. Substance-specific loci (9 for alcohol, 32 for tobacco, 5 for cannabis, 1 for opioids) included metabolic and receptor genes. These findings provide insight into genetic risk loci for substance use disorders that could be leveraged as treatment targets.

Broad yet detailed data collected in biobanks captures variation reflective of human health and behavior, but insights are hard to extract given their complexity and scale. In the largest factor analysis to date, we distill hundreds of medical record codes, physical assays, and survey items from UK Biobank into 35 understandable latent constructs. The identified factors recapitulate known disease classifications, highlight the relevance of psychiatric constructs, improve measurement of health-related behavior, and disentangle elements of socioeconomic status. We demonstrate the power of this principled data reduction approach to clarify genetic signal, enhance discovery, and identify associations between underlying phenotypic structure and health outcomes such as mortality. We emphasize the importance of considering the interwoven nature of the human phenome when evaluating large-scale patterns relevant to public health.

Response to survey questionnaires is vital for social and behavioral research, and most analyses assume full and accurate response by survey participants. However, nonresponse is common and impedes proper interpretation and generalizability of results. We examined item nonresponse behavior across 109 questionnaire items from the UK Biobank (UKB) (N=360,628). Phenotypic factor scores for two participant-selected nonresponse answers, "Prefer not to answer" (PNA) and "I don't know" (IDK), each predicted participant nonresponse in follow-up surveys, controlling for education and self-reported general health. We performed genome-wide association studies on these factors and identified 39 genome-wide significant loci, and further validated these effects with polygenic scores in an independent study (N=3,414), gaining information that we could not have had from phenotypic data alone. PNA and IDK were highly genetically correlated with one another and with education, health, and income, although unique genetic effects were also observed for both PNA and IDK. We discuss how these effects may bias studies of traits correlated with nonresponse and how genetic analyses can further enhance our understanding of nonresponse behaviors in survey research, for instance by helping to correct for nonresponse bias.

Hamer et al . argue that the variable “ever versus never had a same-sex partner” does not capture the complexity of human sexuality. We agree and said so in our paper. But Hamer et al . neglect to mention that we also reported follow-up analyses showing substantial overlap of the genetic influences on our main variable and on more nuanced measures of sexual behavior, attraction, and identity.