Associations of sleep and circadian phenotypes with COVID-19 susceptibility and hospitalization: an observational cohort study based on the UK Biobank and a two-sample Mendelian randomization study

Study Objectives Sleep and circadian phenotypes are associated with several diseases. The present study aimed to investigate whether sleep and circadian phenotypes were causally linked with coronavirus disease 2019 (COVID-19)-related outcomes. Methods Habitual sleep duration, insomnia, excessive daytime sleepiness, daytime napping, and chronotype were selected as exposures. Key outcomes included positivity and hospitalization for COVID-19. In the observation cohort study, multivariable risk ratios (RRs) and their 95% confidence intervals (CIs) were calculated. Two-sample Mendelian randomization (MR) analyses were conducted to estimate the causal effects of the significant findings in the observation analyses. Beta values and the corresponding 95% CIs were calculated and compared using the inverse variance weighting, weighted median, and MR-Egger methods. Results In the UK Biobank cohort study, both often excessive daytime sleepiness and sometimes daytime napping were associated with hospitalized COVID-19 (excessive daytime sleepiness [often vs. never]: RR=1.24, 95% CI=1.02-1.5; daytime napping [sometimes vs. never]: RR=1.12, 95% CI=1.02-1.22). In addition, sometimes daytime napping was also associated with an increased risk of COVID-19 susceptibility (sometimes vs. never: RR= 1.04, 95% CI=1.01-1.28). In the MR analyses, excessive daytime sleepiness was found to increase the risk of hospitalized COVID-19 (MR IVW method: OR = 4.53, 95% CI = 1.04-19.82), whereas little evidence supported a causal link between daytime napping and COVID-19 outcomes. Conclusions Observational and genetic evidence supports a potential causal link between excessive daytime sleepiness and an increased risk of COVID-19 hospitalization, suggesting that interventions targeting excessive daytime sleepiness symptoms might decrease severe COVID-19 rate.

Addressing selection bias in the UK Biobank neurological imaging cohort

The UK Biobank is a national prospective study of half a million participants between the ages of 40 and 69 at the time of recruitment between 2006 and 2010, established to facilitate research on diseases of aging. The imaging cohort is a subset of UK Biobank participants who have agreed to undergo extensive additional imaging assessments. However, Fry et al (2017) find evidence of "healthy volunteer bias" in the UK Biobank -- participants are less likely to smoke, be obese, consume alcohol daily than the target population of UK adults. Here we examine selection bias in the UK Biobank imaging cohort. We address two common misconceptions: first, that study size can compensate for bias in data collection, and second that selection bias does not affect estimates of associations, which are the primary interest of the UK Biobank. We introduce inverse probability weighting (IPW) as an approach commonly used in survey research that can be used to address selection bias in volunteer health studies like the UK Biobank. We discuss 6 such methods -- five existing and one novel --, assess relative performance in simulation studies, and apply them to the UK Biobank imaging cohort. We find that our novel method, BART for predicting the probability of selection combined with raking, performs well relative to existing methods, and helps alleviate selection bias in the UK Biobank imaging cohort.

Identifying the potential role of insomnia on multimorbidity: A Mendelian randomization phenome-wide association study in UK Biobank

Objectives: To identify the breadth of potential causal effects of insomnia on health outcomes and hence its possible role in multimorbidity. Design: Mendelian randomisation (MR) Phenome-wide association study (MR-PheWAS) with two-sample Mendelian randomisation follow-up. Setting: Individual data from UK Biobank and summary data from a number of genome-wide association studies. Participants: 336,975 unrelated white-British UK Biobank participants. Exposures: Standardised genetic risk of insomnia for the MR-PheWAS and genetically predicted insomnia for the two-sample MR follow-up, with insomnia instrumented by a genetic risk score (GRS) created from 129 single-nucleotide polymorphisms (SNPs). Main outcomes measures: 11,409 outcomes from UK Biobank extracted and processed by an automated pipeline (PHESANT). Potential causal effects (i.e., those passing a Bonferroni-corrected significance threshold) were followed up with two-sample MR in MR-Base, where possible. Results: 437 potential causal effects of insomnia were observed for a number of traits, including anxiety, stress, depression, mania, addiction, pain, body composition, immune, respiratory, endocrine, dental, musculoskeletal, cardiovascular and reproductive traits, as well as socioeconomic and behavioural traits. We were able to undertake two-sample MR for 71 of these 437 and found evidence of causal effects (with directionally concordant effect estimates across all analyses) for 25 of these. These included, for example, risk of anxiety disorders (OR=1.55 [95% confidence interval (CI): 1.30, 1.86] per category increase in insomnia), diseases of the oesophagus/stomach/duodenum (OR=1.32 [95% CI: 1.14, 1.53]) and spondylosis (OR=1.57 [95% CI: 1.22, 2.01]). Conclusion: Insomnia potentially causes a wide range of adverse health outcomes and behaviours. This has implications for developing interventions to prevent and treat a number of diseases in order to reduce multimorbidity and associated polypharmacy.

Polygenic transcriptome risk scores (PTRS) can improve portability of polygenic risk scores across ancestries

Background Polygenic risk scores (PRS) are valuable to translate the results of genome-wide association studies (GWAS) into clinical practice. To date, most GWAS have been based on individuals of European-ancestry leading to poor performance in populations of non-European ancestry. Results We introduce the polygenic transcriptome risk score (PTRS), which is based on predicted transcript levels (rather than SNPs), and explore the portability of PTRS across populations using UK Biobank data. Conclusions We show that PTRS has a significantly higher portability (Wilcoxon p=0.013) in the African-descent samples where the loss of performance is most acute with better performance than PRS when used in combination.

Association of serum 25-hydroxyvitamin D concentrations with risk of dementia among individuals with type 2 diabetes: A cohort study in the UK Biobank

Background Several epidemiological studies have suggested that vitamin D status is associated with risk of dementia in general populations. However, due to the synergistic effect between diabetic pathology and neuroinflammation, and the prothrombotic profile in patients with diabetes, whether vitamin D is associated with risk of dementia among patients with diabetes is unclear. This study aimed to investigate the associations of circulating vitamin D levels with risks of all-cause dementia, Alzheimer disease (AD), and vascular dementia (VD) among adults with type 2 diabetes (T2D). Methods and findings This study included 13,486 individuals (≥60 years) with T2D and free of dementia at recruitment (2006–2010) from the UK Biobank study. Serum 25-hydroxyvitamin D (25[OH]D) concentrations were measured using the chemiluminescent immunoassay method at recruitment. Serum 25(OH)D ≥ 75 nmol/L was considered sufficient, according to the Endocrine Society Clinical Practice Guidelines. Incidence of all-cause dementia, AD, and VD cases was ascertained using electronic health records (EHRs). Each participant’s person-years at risk were calculated from the date of recruitment to the date that dementia was reported, date of death, date of loss to follow-up, or 28 February 2018, whichever occurred first. Among the 13,486 individuals with T2D (mean age, 64.6 years; men, 64.3%), 38.3% had vitamin D ≥ 50 nmol/L and only 9.1% had vitamin D ≥ 75 nmol/L. During a mean follow-up of 8.5 years, we observed 283 cases of all-cause dementia, including 101 AD and 97 VD cases. Restricted cubic spline analysis demonstrated a nonlinear relationship between serum 25(OH)D and risk of all-cause dementia (Pnonlinearity < 0.001) and VD (Pnonlinearity = 0.007), and the nonlinear association reached borderline significance for AD (Pnonlinearity = 0.06), with a threshold at around a serum 25(OH)D value of 50 nmol/L for all the outcomes. Higher serum levels of 25(OH)D were significantly associated with a lower risk of all-cause dementia, AD, and VD. The multivariate hazard ratios and 95% confidence intervals for participants who had serum 25(OH)D ≥ 50 nmol/L, compared with those who were severely deficient (25[OH]D < 25 nmol/L), were 0.41 (0.29–0.60) for all-cause dementia (Ptrend < 0.001), 0.50 (0.27–0.92) for AD (Ptrend = 0.06), and 0.41 (0.22–0.77) for VD (Ptrend = 0.01). The main limitation of the current analysis was the potential underreporting of dementia cases, as the cases were identified via EHRs. Conclusions In this study, we observed that higher concentrations of serum 25(OH)D were significantly associated with a lower risk of all-cause dementia, AD, and VD among individuals with T2D. Our findings, if confirmed by replication, may have relevance for dementia prevention strategies that target improving or maintaining serum vitamin D concentrations among patients with T2D.

Polynomial Mendelian Radomization reveals widespread non-linear causal effects in the UK Biobank

Causal inference is a critical step in improving our understanding of biological processes and Mendelian randomisation (MR) has emerged as one of the foremost methods to efficiently interrogate diverse hypotheses using large-scale, observational data from biobanks. Although many extensions have been developed to address the three core assumptions of MR-based causal inference (relevance, exclusion restriction, and exchangeability), most approaches implicitly assume that any putative causal effect is linear. Here we propose PolyMR, an MR-based method which provides a polynomial approximation of an (arbitrary) causal function between an exposure and an outcome. We show that this method provides accurate inference of the shape and magnitude of causal functions with greater accuracy than existing methods. We applied this method to data from the UK Biobank, testing for effects between anthropometric traits and continuous health-related phenotypes and found most of these (84%) to have causal effects which deviate significantly from linear. These deviations ranged from slight attenuation at the extremes of the exposure distribution, to large changes in the magnitude of the effect across the range of the exposure (e.g. a 1 kg/m2 change in BMI having stronger effects on glucose levels if the initial BMI was higher), to non-monotonic causal relationships (e.g. the effects of BMI on cholesterol forming an inverted U shape). Finally, we show that the linearity assumption of the causal effect may lead to the misinterpretation of health risks at the individual level or heterogeneous effect estimates when using cohorts with differing average exposure levels.