{
  "abstract": "Background Obesity can lead to a range of serious health conditions, many of which display large inequalities across socio-economic position (SEP). These inequalities likely result from both differential exposure (exposure prevalence differing across groups) and differential susceptibility (exposure effect differing across groups). Differential exposure exists; obesity prevalence is 36% and 21% in England’s most and least deprived areas respectively. Differential susceptibility (effect modification) has been found for smoking and alcohol, but has not been investigated for obesity. To address this, we explored whether obesity’s effect on a range of obesity-related disease outcomes was modified by SEP, using a longitudinal population-based cohort.Methods UK Biobank recruited 500,000 people between 2006 and 2010. Baseline data were linked to hospital and mortality records to investigate eight disorders where obesity is strongly implicated: diabetes, non-alcoholic fatty liver disease (NAFLD), osteoarthritis, urinary stress incontinence (USI), sleep apnoea, essential hypertension, pancreatitis and gout. Nurse measured BMI was categorised into underweight, normal, overweight or obese using standard cut-offs. Four SEP measures were investigated: Townsend Deprivation Index (TDI), highest educational attainment, household income, and home ownership. Participants with prior disease or underweight were excluded, and follow-up occurred until the outcome, death or end of follow-up (Nov 2022). Cox models were adjusted for sex, age, assessment centre and ethnicity (for household income, household size was an additional covariate). Due to its public health importance our primary interest was differential susceptibility on the additive scale. This we measured using the relative excess risk of interaction (RERI) statistic (replacing risks with hazards).Results For all disorders, except pancreatitis and USI, obesity was more harmful for those of low versus high SEP. The RERI hazard ratios (95% CI) for the additional risk not explained by the main effects of obesity and low SEP (TDI tertile) were: diabetes 3.03 (2.66; 3.4), essential hypertension 0.51 (0.43; 0.58), gout 1.11 (0.7; 1.51), NAFLD 0.85 (0.58; 1.12), osteoarthritis 0.19 (0.11; 0.28), sleep apnoea 4.17 (3.35; 4.99), pancreatitis -0.1 (-0.41; 0.22), and USI 0.02 (-0.31; 0.34). Results were similar across SEP measures. There was little evidence for multiplicative effect modification.Conclusion Obesity appears more harmful for socioeconomically disadvantaged people for several obesity-related diseases. So interventions that reduce obesity in more socioeconomically disadvantaged groups could yield disproportionate population health benefits and reduce absolute inequalities. The work requires external validation as the UK Biobank cohort is healthier, wealthier and more elderly than the UK population.",
  "authors": [
    {
      "affiliations": [
        "MRC/CSO Social and Public Health Sciences Unit, University of Glasgow, Glasgow, UK"
      ],
      "name": "Desmond Campbell"
    },
    {
      "affiliations": [
        "MRC/CSO Social and Public Health Sciences Unit, University of Glasgow, Glasgow, UK"
      ],
      "name": "Rachel Thomson"
    },
    {
      "affiliations": [
        "MRC/CSO Social and Public Health Sciences Unit, University of Glasgow, Glasgow, UK"
      ],
      "name": "Anna Pearce"
    },
    {
      "affiliations": [
        "School of Cardiovascular and Metabolic Health, University of Glasgow, Glasgow, UK"
      ],
      "name": "Jana Anderson"
    },
    {
      "affiliations": [
        "School of Cardiovascular and Metabolic Health, University of Glasgow, Glasgow, UK"
      ],
      "name": "Naveed Sattar"
    },
    {
      "affiliations": [
        "MRC/CSO Social and Public Health Sciences Unit, University of Glasgow, Glasgow, UK"
      ],
      "name": "Vittal Katikireddi"
    }
  ],
  "title": "OP76 Do the harmful effects of obesity differ by socio-economic position? Assessing effect modification using linked UK biobank cohort data",
  "uid": "fa62fe91-5c79-5a8e-9d93-a7735922a15b"
}