An explanatory study into the relationship between socioeconomic status and the risk of type 2 diabetes across the life course in a UK population

Abstract

Background

Low socioeconomic status (SES) across thelife course is associated with type 2 diabetes (T2DM), with potential for adose-dependent effect. However, the extent to which LC models explain thisrelationship, the effect of SES trajectories, and mediating effects, has notbeen investigated in the UK ‘baby boomer’ generation. This study evaluatedthese aspects, and mediators, in the pathway between childhood SES and T2DM, inthe 1958 Birth Cohort Study.

Methodology

SES (low/high) was derived from RG socialclass at 7, 23 and 42 years. A diagnosis of T2DM by 44–45 years was based onglycated haemoglobin (adjusted for anti-hyperglycaemic medication). Theexplanatory power of the accumulation, critical periods, and social mobilityhypotheses was evaluated by comparing nested models for each to a saturatedmodel using a χ² test. The effect of SES trajectories on the risk ofT2DM was compared with the referent stable high trajectory using logisticregression. Pathways between sensitive lifecourse stages for low SES and T2DMwere evaluated via generalised linear modelling with decomposition analysis. Theeffect of SES on glycaemic control in the subpopulation with T2DM was evaluatedusing descriptive statistics and the Kolmogorov-Smirnov test for equality of distribution. A sensitivity analysis with a stricterthreshold for low SES evaluated the effect of exposure intensity. Missing datawere corrected using multiple imputation.

Results

The strict accumulation model had the closestfit to the saturated model in the primary (adjusted χ² (6 df) =8.65,p=0.38) and sensitivity analyses (adjusted χ² (6 df) =15.02, p=0.07),based on p-value. Using an alternative approach based on fit indices, thechildhood critical period model (low SES at 7 years) also demonstrated acomparable fit in the primary analysis based on fit indices (loglikelihood=-1,080.30 versus -1,075.91).

The risk of T2DM was strongest in the L-L-Htrajectory (unadjusted odds ratio [OR]=2.49±0.78, p<0.001; OR adjusted foroverweight or obesity at 23 and 33 years=2.16±0.68, p<0.05). Effect sizeswere larger in the sensitivity analysis (OR=3.28±0.77, p<0.001, andOR=3.09±0.73, p<0.001, respectively), demonstrating a dose-response effect.

Overweight/obesity at 23 and 33 years had asignificant indirect effect on the association between low SES at 7 years andT2DM (OR=1.16, 98% CI, 1.11–1.22 and OR=1.14, 98% CI, 1.09–1.19, both p<0.001).The indirect effect accounted for by obesity was 30% in model that includedoverweight/obesity at 23 years, 24% at 33 years, and 33% for both timepoints,respectively. SES was not associated with glycaemic control in T2DM.

Conclusion

This study indicates that low SES inchildhood is associated with incident T2DM by 44–45 years of age, and that partof the association is mediated by overweight/obesity at 23 and 33 years. Suchevidence can be used to develop targeted strategies to reduce this inequality.