Understanding the role of socioeconomic factors on antidiabetic medication use patterns and long-term consequences of type 2 diabetes via novel causal inference methods

Background A recent study on Swedish data showed that the incidence of type 2 diabetes (T2D) is strongly related to occupational groups. Therefore, socioeconomic determinants of T2D are becoming a growing concern.

After disease onset, correct use of antidiabetic medication (ADM) is a key factor for proper disease control and prevention of long-term severe consequences.

A recent study from our team on a population of pregnant women showed associations between socioeconomic factors such as income, educational level, and country of origin, and ADM interruption.

Currently, no study has yet investigated ADM use and socioeconomic inequalities in the larger population of T2D patients.Aim First, to identify socioeconomic differences in T2D patients with different ADM use patterns including treatment interruption and provide insights for policy makers.

Second, to investigate associations between ADM use patterns and long-term complications of T2D, such as retinopathy and renal failure, in relation to specific socioeconomic disadvantaged groups.

Ultimately, ADM use patterns and socioeconomic disparities will be investigated in relationship to comedication patterns used as proxies to evaluate the impact of comorbidities.Methods Patients’ ADM use is difficult to evaluate in a clinical trial setting since therapies for chronic diseases (such as T2D) may be life-long, but Swedish national registers offer a valuable opportunity.

In the proposed project, Swedish national and diabetes quality register data will be used to identify individuals with T2D.

The study population will be followed over time to provide a life course assessment of ADM use patterns in relation to socioeconomic factors, T2D complications, and comedication patterns.

To properly adjust for confounding and allow for causality assessment in observational data, advanced causal inference methods such as inverse probability weighting will be used.