WJWilliam Johnson

William Johnson

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


William Johnson has a total of 219 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Heterogeneity in the association of BMI with type 2 diabetes across lifestyle behaviour combinations: A MAIHDA analysis of 293,825 UK Biobank participants

    by Ansuman Swain, Scott Willis, George Leckie, William Johnson, Natalie Pearson

    Background Higher body mass index (BMI) is strongly associated with greater risk of type 2 diabetes (T2D), but it remains unclear how lifestyle behaviour combinations modify the BMI-T2D relationship. We applied multilevel analysis of individual heterogeneity and discriminatory accuracy (MAIHDA) to examine heterogeneity in the BMI-T2D association across lifestyle profiles. Methods The sample comprised 148,363 males and 145,462 females. Participants were classified into 48 strata combining five lifestyle behaviours: fruit and vegetable intake, physical activity, sleep duration, alcohol intake, and smoking. MAIHDA models were fitted with individuals nested within strata. Sex-stratified binary logistic models examined how lifestyle behaviours and BMI accounted for between-strata variation in T2D, including whether BMI associations varied across strata, and adjusted for sociodemographic confounders. Variance partition coefficient (VPC), proportional change in variance (PCV), and strata-specific BMI associations with T2D were estimated. Results T2D prevalence was 11.7% in males and 6.3% in females. Each 1 kg/m² higher BMI was associated with higher T2D odds in males (OR 1.18; 95% CrI 1.17–1.19) and females (OR 1.15; 95% CrI 1.15–1.16). Between-stratum variation was modest in the null models (VPC: males 6.7%; females 8.0%) and was largely explained by additive lifestyle effects (PCV: 96.8% and 97.2%, respectively). The random BMI slope SD was 0.01 (95% CrI 0.00–0.02) in both sexes, with broadly similar positive BMI-T2D associations across lifestyle strata. Conclusion Higher BMI was associated with greater T2D odds in both sexes, with little heterogeneity across lifestyle strata, highlighting the importance of weight management across lifestyle profiles.

  • Open Access

    Impact of Prolonged Obesity Across the Lifecourse on Myocardial Stress and Injury in Older Age: Birth Cohort Study

    by William Johnson, Tom Norris, Alun D. Hughes, Mark Hamer, Steven Bell, G. David Batty

    Background: The consequences of prolonged exposure to obesity on biomarkers of myocardial stress and injury remain unclear. We examined associations of duration and cumulative severity of obesity from childhood to later adulthood with concentrations of N-terminal pro-B-type natriuretic peptide (NT-proBNP), high-sensitivity cardiac troponin I (hs-cTnI), and growth differentiation factor-15 (GDF-15) at age 62 years. Methods: The sample comprised 4,901 participants in the 1958 National Child Development Study who had measurements of NT-proBNP, hs-cTnI, or GDF-15 at age 62 years and at least four BMI assessments between 7 and 62 years of age. Individual BMI trajectories were modelled, from which total duration of obesity and mean excess BMI (area between the trajectory and the age-specific obesity threshold divided by duration) were derived. Associations of ever versus never having been obese, obesity duration, and mean excess BMI with biomarker concentrations were examined using general linear regression models with progressive adjustment for early-life characteristics; contemporaneous cardiometabolic factors; and health behaviours, socioeconomic position, and psychosocial factors. Nonlinear associations of obesity duration were estimated using restricted cubic splines and subsequently approximated using linear splines. Results: Overall, 26.3% of participants had ever experienced obesity. There was no evidence of an association between ever versus never having been obese and NT-proBNP. Among men, ever having been obese was associated with 11.00% (95% CI, 4.62–17.37) higher hs-cTnI in the fully adjusted model, whereas there was no evidence of an association among women. Ever having been obese was associated with higher GDF-15, although the association attenuated substantially after adjustment for contemporaneous cardiometabolic factors. Longer obesity duration was associated with higher concentrations of all three biomarkers in fully adjusted models, but the pattern differed by biomarker: hs-cTnI increased approximately linearly with duration, whereas NT-proBNP and GDF-15 remained relatively stable up to approximately 25 years of obesity before increasing with longer duration. In fully adjusted models, each additional year of obesity beyond 25 years was associated with 1.80% (95% CI, 0.54–3.06) higher NT-proBNP and 0.81% (95% CI, 0.25–1.36) higher GDF-15, while each additional year across the full range of obesity duration was associated with 0.41% (95% CI, 0.07–0.76) higher hs-cTnI. Among participants who had ever experienced obesity, each 1 kg/m² higher mean excess BMI was associated with 4.02% (95% CI, 1.79–6.24) higher NT-proBNP and 2.00% (95% CI, 0.49–3.51) higher hs-cTnI in fully adjusted models; there was no evidence of an association with GDF-15. Conclusions: Longer duration of obesity was associated with higher concentrations of biomarkers reflecting myocardial stress, injury, and cellular stress at age 62 years, with associations particularly evident after approximately 25 years of obesity for NT-proBNP and GDF-15. These findings suggest that duration of obesity may provide clinically relevant information beyond current obesity status, supporting its consideration as an additional dimension of cumulative obesity burden in cardiovascular risk assessment.

  • Are associations of childhood body mass index trajectories with early-onset coronary heart disease modified by parental history of cardiovascular disease?

    by Julie Aarestrup, Elisabeth W. Andersen, Gilad Twig, Rikke K. Jacobsen, William Johnson, Lise G. Bjerregaard, Jennifer L. Baker

  • Open Access

    Early Formal Childcare and Adolescent Psychological Wellbeing in the Millennium Cohort Study: The Role of Socio-economic Position

    by K. Burdenski, E. Petherick, S. Costa, W. Johnson

    Little research has analysed the relation between childcare and positive psychological wellbeing, e.g., self-esteem and life satisfaction. Using data from the UK Millennium Cohort Study, this paper investigates the association of the age of starting and intensity of formal childcare with self-esteem trajectories between 11 and 17 years and life satisfaction at 11 and 14 years ( N = 5,484). Proportional odds for having greater psychological wellbeing and odds ratios for having good compared to poor psychological wellbeing (lowest 25% scores) were estimated. After controlling for children’s socio-economic position, childcare was not associated with self-esteem or life satisfaction. Changes in estimates after adjustment show that children from a lower socio-economic position attended less childcare and had lower odds of later psychological wellbeing.

  • Open Access

    Do minimum acceptable diet and family caregiving mediate the associations of maternal education and household wealth with childhood stunting and wasting in Sri Lanka?

    by Damith Chandrasenage, Paula Griffiths, William Johnson

    Objective: To investigate the extent to which the associations of socio-economic position (SEP) with stunting and wasting are mediated by minimum acceptable diet (MAD) and a family care indicator (FCI) in Sri Lanka. Design: Secondary data analysis of children from the 2016 Sri Lanka Demographic and Health Survey. The outcomes were stunting and wasting, the exposure was a composite measure combining maternal education and household wealth, and the mediators were binary MAD and FCI variables (adequate v . inadequate). Analyses were performed using counterfactual mediation models adjusted for age, sex and place of residence. Setting: A nationally representative sample of children from Sri Lanka. Participants: Mothers/caregivers of children under 36 months (4325). Results: Twenty per cent of children were stunted, and 14 % were wasted. Lower SEP was associated with higher odds of stunting and wasting and inadequate MAD and FCI. Inadequate FCI was associated with higher odds of stunting (OR = 1·47, 95 % CI = 1·24, 1·74) but not wasting (OR = 1·14, 95 % CI = 0·94, 1·38), whereas MAD was not associated with stunting or wasting. Neither MAD nor FCI significantly mediated the relationship between SEP and stunting and wasting. All mediation estimates were statistically non-significant at the 5 % level. For example, the proportion mediated by FCI on the association between the lowest composite SEP and stunting was 13 % (mean difference = 0·13, 95 % CI = < 0·00, 0·26). Conclusion: We did not find consistent or strong evidence that the associations of SEP with childhood stunting and wasting in Sri Lanka are mediated by MAD and FCI. Research with larger samples is needed for more precise estimates.


  • Open Access

    Multilevel analysis of individual heterogeneity and discriminatory accuracy (MAIHDA) to understand how obesity risk varies according to multiple lifestyle behavior recommendations

    by Ansuman Swain, Natalie Pearson, Scott A. Willis, William Johnson

    Background: The combined and interactive effects of multiple lifestyle behaviours on obesity risk are not well understood. We used Multilevel Analysis of Individual Heterogeneity and Discriminatory Accuracy (MAIHDA) to examine how adherence to public health recommendations for five lifestyle behaviours affects BMI and obesity risk. Methods: The sample included 139,540 men and 125,455 women from the UK Biobank. We categorized fruit and vegetable intake, physical activity, sleep duration and alcohol intake as binary variables (meeting vs. not meeting guidelines), and smoking status into three categories (previous, current, never). These categories were combined to form 48 unique strata, representing all possible combinations of the five behaviours. Linear and binary logistic MAIHDA models were used, with individuals nested within strata, and BMI and obesity status (obesity vs. normal weight) as outcomes. Three models were employed: Model 1 (null), Model 2 (with fixed effects for lifestyle behaviours), and Model 3 (with confounders and fixed effects). Variance Partition Coefficient (VPC), Proportional Change in Variance (PCV), and predicted BMI and obesity risk were estimated. Results: For both sexes, strata with the lowest obesity risk were associated with meeting most recommendations, while strata with the highest risk were linked to meeting few. Logistic Model 1 VPCs revealed 7% of variance in obesity risk among males and 5% among females was explained by between-strata differences. In Model 3, VPCs attenuated to 0.5% among males and 0.1% among females, suggesting differences in obesity risk were largely additive effects. PCVs from Model 3 also indicated primarily additive rather than interactive effects. Results were similar for BMI in the linear models. Conclusions: Using a novel statistical approach, this study shows that additive effects of multiple lifestyle behaviours predominantly explain differences in BMI and obesity risk. Meeting more public health lifestyle recommendations is important in mitigating obesity risk.


  • Open Access

    Intersectional inequality in general and central obesity: cross-sectional UK Biobank study

    by Joseph Hutchinson, Natalie Darko, Rebecca Hardy, David Webb, Francesco Zaccardi, William Johnson

    Background: People exist at a combination of different individual and neighbourhood deprivations. Each of these combinations may have a unique impact on health. However, little is known about the intersectional inequality of these combinations on general and central obesity, including when considering their demographics. This study aims to answer these questions. Methods: The sample comprised 452,339 participants from the UK Biobank study. Individuals were grouped into 320 intersectional strata according to their household income, neighbourhood deprivation, sex, ethnicity and age. Linear and logistic multilevel analysis of individual heterogeneity and discriminatory accuracy was used to establish the total, additive and interactive inequality of body mass index (BMI), fat mass index (FMI), and waist to height ratio (WHtR), as well as the associated obesity classifications. Results: 6.5%, 25.2% and 9.1% of the total variation in BMI, FMI and WHtR, respectively, was due to inequality between the strata. Of this, 26.5%, 3.5% and 22.0% is interactive. 79, 58 and 93 strata for BMI, FMI and WHtR demonstrate a significant interactive effect. We found some patterns; for example, affluent white women have an advantaged interactive effect, whilst deprived black women have a disadvantaged effect. Meanwhile men experience the inverse relationship. The relationship between individual and neighbourhood deprivation is not universally experienced by all strata. For example, black men living in areas of high deprivation have higher BMIs as their household income increases. Conclusions: A large proportion of variation in general and central obesity is due to intersectional inequality, with up to 26.5% being interactive. It is important that these intersectional effects are considered when designing policy interventions to avoid policy failure, such as by focussing on groups with high total and interactive risk.


  • Open Access

    Stunting and inequality in Sri Lanka compared with other low- and middle-income South Asian countries

    by Damith Chandrasenage, William Johnson, Paula L. Griffiths

    Objective: This study investigates and measures whether the association of childhood stunting with household socio-economic position (SEP) differs in Sri Lanka compared with other South Asian countries. Design: Secondary analysis of data of children from the latest available Demographic and Health Surveys data (survey years, 2016–2018). The exposures (SEP) were maternal education and wealth. The outcome was stunting. Binary logistic regression models incorporated SEP, country and SEP-by-country interaction terms. Setting: A nationally representative sample of children from Bangladesh, India, Nepal, Pakistan and Sri Lanka. Participants: Mothers/caregivers of children under 36 months (133 491). Results: The prevalence of stunting in Sri Lanka of 19 % was much lower than that observed for all the other low- to low–middle income South Asian countries (37 % in Bangladesh, 36 % in India, 31 % in Nepal and 30 % in Pakistan). The association of SEP with odds of stunting was similar in Sri Lanka compared with other South Asian countries. The only exception was weaker associations of wealth with stunting in Sri Lanka compared with Bangladesh. For example, in Sri Lanka, the poorest group had 2·75 (2·06, 3·67) times higher odds of stunting compared with the richest group, but in Bangladesh, this estimate was 4·20 (3·24, 5·44); the difference between these two estimates being 0·65 (0·44, 0·96) on the OR scale. Conclusions: The lower prevalence of stunting in Sri Lanka is unlikely to be due to less inequality. It is more likely that the lower prevalence of stunting in Sri Lanka is related to there being fewer mothers belonging to the lowest SEP groups.


  • Open Access

    Are associations of adulthood overweight and obesity with all-cause mortality, cardiovascular disease, and obesity-related cancer modified by comparative body weight at age 10 years in the UK Biobank study?

    by William Johnson, Tom Norris, Natalie Pearson, Emily S. Petherick, James A. King, Scott A. Willis, Rebecca Hardy, Susan Paudel, Emma Haycraft, Jennifer L. Baker, et al.

    Objective: Adults living with overweight or obesity do not represent a single homogenous group in terms of mortality and disease risks. The aim of our study was to evaluate how the associations of adulthood overweight and obesity with mortality and incident disease are modified by (i.e., differ according to) self-reported childhood body weight categories. Methods: The sample comprised 191,181 men and 242,806 women aged 40–69 years (in 2006–2010) in the UK Biobank. The outcomes were all-cause mortality, incident cardiovascular disease (CVD), and incident obesity-related cancer. Cox proportional hazards regression models were used to estimate how the associations with the outcomes of adulthood weight status (normal weight, overweight, obesity) differed according to perceived body weight at age 10 years (about average, thinner, plumper). To triangulate results using an approach that better accounts for confounding, analyses were repeated using previously developed and validated polygenic risk scores (PRSs) for childhood body weight and adulthood BMI, categorised into three-tier variables using the same proportions as in the observational variables. Results: In both sexes, adulthood obesity was associated with higher hazards of all outcomes. However, the associations of obesity with all-cause mortality and incident CVD were stronger in adults who reported being thinner at 10 years. For example, obesity was associated with a 1.28 (1.21, 1.35) times higher hazard of all-cause mortality in men who reported being an average weight child, but among men who reported being a thinner child this estimate was 1.63 (1.53, 1.75). The ratio between these two estimates was 1.28 (1.17, 1.40). There was also some evidence that the associations of obesity with all-cause mortality and incident CVD were stronger in adults who reported being plumper at 10 years. In genetic analyses, however, there was no evidence that the association of obesity (according to the adult PRS) with mortality or incident CVD differed according to childhood body size (according to the child PRS). For incident obesity-related cancer, the evidence for effect modification was limited and inconsistent between the observational and genetic analyses. Conclusions: Greater risks for all-cause mortality and incident CVD in adults with obesity who perceive themselves to have been a thinner or plumper than average child may be due to confounding and/or recall bias.