What depression adds to the outcome of diabetes

Design

systematic review and random-effects meta-analysis of 26 observational studies from nine regions, searched to December 2024

Population

adults with diagnosed depressive disorder and diabetes compared with adults with diabetes alone

Primary outcome

all-cause and cause-specific mortality, and occurrence of specific diabetes complications

Effect

all-cause mortality RR 1.30 (1.21-1.39), cardiovascular mortality 1.15 (1.02-1.29), any complication 1.28 (1.18-1.40), metabolic complications 1.63 (1.33-1.99), retinopathy 0.84 (0.76-0.94)

Earlier work on depression in diabetes has been muddied by using symptom questionnaires, which pick up diabetes distress and subclinical low mood as well as depressive disorder. This meta-analysis restricted itself to studies comparing people with diagnosed depression and diabetes against people with diabetes alone, pooling 26 studies from nine regions to December 2024.

All-cause mortality was 30% higher in the depression-diabetes group (RR 1.30, 95% CI 1.21-1.39) and cardiovascular mortality 15% higher (1.15, 1.02-1.29). Complications overall were 28% commoner (1.28, 1.18-1.40), driven by metabolic complications (1.63, 1.33-1.99) and cardiovascular complications (1.20, 1.11-1.29). Cerebrovascular events, nephropathy and peripheral vascular complications did not differ. Retinopathy was less common (0.84, 0.76-0.94), which the authors do not explain and which most likely reflects differential screening rather than protection.

The pattern points somewhere specific. Metabolic complications — the acute, glycaemia-driven ones — carried the largest excess, which fits worse glycaemic control rather than accelerated vascular disease. That makes depression a treatable contributor to diabetes outcomes rather than a comorbidity to note and move past. Heterogeneity was substantial and not explained by the subgroup analyses, so the individual numbers should be held loosely; the direction is consistent across regions and over time.

• Screen for depression at annual review, using a validated tool rather than an impression
• Distinguish depressive disorder from diabetes distress — they need different responses
• Where control has deteriorated without an obvious cause, ask about mood before intensifying therapy
• Do not read the lower retinopathy rate as protection; it more likely reflects less screening attendance
• Coordinate care: treating depression is part of treating the diabetes, not a separate referral

The statistics, in plain English

A relative risk of 1.30 means 30% more deaths in relative terms; how many extra deaths that represents depends on the baseline rate, which varies widely across these 26 studies. The cardiovascular mortality interval (1.02-1.29) barely clears 1.0, so that estimate is fragile. Substantial unexplained heterogeneity means the studies disagree by more than chance, so a single pooled number understates the uncertainty. The apparently protective retinopathy result is the clearest example of why: a real protective effect of depression on the retina is implausible, and detection bias is the likelier explanation.

Read the rest in the app

You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.

Get Daily Doctor — free

Scan to keep reading on your phone. No account needed to start.