Background Cross country comparison facilitates examination of health-system and country-level similarities and differences on disease burden. Diabetes is a leading global health issue with outcomes influenced by psychological comorbidity. Diabetes is consistently associated with depression in different populations but any association is influenced by variations in prevalence, study design and measure used. We compared depressive symptom prevalence in older adults with and without diabetes across three health systems using a validated measure to examine the causal association between diabetes and depression.

Methods We used data of adults aged 50 years and older, from three nationally representative ageing datasets; The Irish Longitudinal Study on Ageing (TILDA), the English Longitudinal Study on Ageing (ELSA) and the Health and Retirement Study (HRS). Data were collected during 2009–2011 (TILDA), and 2010 (ELSA, HRS). Variables; country, gender, age, education level, marital status and smoking status, were selected a priori using literature and a directed acyclic graph. Variables were self-report, with the exception of ‘country’, which was linked to the original dataset. The 20-item (cut-off-score>16;TILDA) and 8-item CESD (cut-off-score>3;ELSA and HRS) were used to categorise depression. Prevalence was presented as a percentage with corresponding 95% confidence intervals and group-specific differences presented using Pearson’s chi-square test for categorical data and Student’s t-test for continuous data. Binomial logistic regression examined the odds of depression by diabetes status, adjusting for pre-selected variables. Analyses of country-specific factors related to depression is ongoing. Data were analysed using Stata v15.

Results Diabetes prevalence varied across the countries (Ireland; 8%[(95%CI:7.5–8.6), England; 11%(95%CI:10.6–12.0), USA; 22%[(95%CI:21.7–22.8); p<0.001]) and was consistently higher among males than females (Ireland; 10%(95%CI:8.9–10.8) vs. 6%(95%CI:5.3–6.7), England; 13%(95%CI:12.4–14.5) vs. 10%(95%CI:8.7–10.3), USA; 24%(95%CI:22.9–24.7) vs. 21%[(95%CI:20.3–21.8); p<0.001]). Depression prevalence was significantly higher among people with diabetes in all countries (Ireland; 12%(95%CI:9.8–15.1) vs. 10%[(95%CI:9.0–10.4), p=0.035], England; 34%(95%CI:31.0–37.0) vs. 22%[(95%CI:20.8–22.6):p<0.001], USA; 20%(95%CI:19.0–21.4) vs. 14%[(95%CI:13.6–14.7); p<0.001], in females (Ireland; 19%(95%CI:14.6–23.9) vs. 12%[(95%CI:11.0–12.9); p=0.001], England; 39%(95%CI:34.2–43.4) vs. 25%[(95%CI:24.0–26.6); p<0.001], USA; 24%(95%CI:21.9–25.3) vs. 16%[(15.4–16.9); p<0.001]), and in males (Ireland; 8%[(95%CI:5.3–10.8) vs. 7%[(95%CI:6.1–7.9); p=0.644], England; 30%(95%CI:25.8–33.9) vs. 17%(95%CI:15.7–18.3); p<0.001], USA; 16%(95%CI:14.4–17.7) vs. 11%[(95%CI:10.6–12.2); p<0.001]. After adjusting for all model variables, people with diabetes had 1.5 increased odds of depression compared to people without diabetes [(95%CI:1.4–1.6), p<0.001].

Conclusion In older people in three countries with different health systems, depressive symptom prevalence was (1) consistently higher among men and women with diabetes than non-diabetes counterparts and (2) varied across health systems. While use of self-report data may limit the accuracy of the results, use of the CESD to categorise depression and of large nationally representative datasets strengthens the study. Interrogation of additional country-level factors associated with depression will further explain variation in depressive symptom prevalence across health systems.