Diabetes mellitus, fasting blood glucose concentration, and risk of vascular disease: a collaborative meta-analysis of 102 prospective studies.
Level 1 - systematic review of randomized trials
Individual participant data meta-analysis of 102 prospective cohort studies
PubMed 20609967 · doi:10.1016/S0140-6736(10)60484-9
What was done
An individual participant data meta-analysis was conducted using records from 102 prospective studies in the Emerging Risk Factors Collaboration, covering 698,782 participants without baseline vascular disease (8.49 million person-years, 52,765 non-fatal or fatal vascular outcomes). Within-study regressions were adjusted for age, sex, smoking, systolic blood pressure, and body-mass index to calculate hazard ratios (HRs) for vascular disease associated with diabetes and fasting blood glucose concentrations.
What was found
Adjusted HRs associated with diabetes were 2.00 (95% CI 1.83–2.19) for coronary heart disease (CHD), 2.27 (95% CI 1.95–2.65) for ischaemic stroke, 1.56 (95% CI 1.19–2.05) for haemorrhagic stroke, 1.84 (95% CI 1.59–2.13) for unclassified stroke, and 1.73 (95% CI 1.51–1.98) for other vascular deaths. Additional adjustment for lipid, inflammatory, or renal markers did not appreciably alter HRs. CHD HRs were higher in women than men, higher at ages 40–59 years than ≥70 years, and higher for fatal than non-fatal events. At 10% adult diabetes prevalence, diabetes was estimated to account for 11% (95% CI 10–12%) of vascular deaths. Fasting blood glucose had a non-linear relationship with CHD: relative to 3.90–5.59 mmol/L, HRs were 1.07 (95% CI 0.97–1.18) for <3.90 mmol/L, 1.11 (95% CI 1.04–1.18) for 5.60–6.09 mmol/L, and 1.17 (95% CI 1.08–1.26) for 6.10–6.99 mmol/L. In non-diabetic individuals, fasting glucose did not significantly improve vascular risk prediction over conventional risk factors.
Why it matters
This study robustly quantifies the independent twofold vascular risk of diabetes across specific stroke and coronary subtypes, while showing that fasting glucose offers minimal incremental predictive value in non-diabetic populations.
Limits
The observational nature of the included cohorts cannot entirely rule out residual confounding. The abstract does not report distinctions between type 1 and type 2 diabetes, data on diabetes duration or treatment, or alternative glycemic measures such as HbA1c.
Cited by
- contradicts People with full-blown type 2 diabetes have a 5 to 8 times increased risk of cardiovascular disease compared to individuals with high cholesterol alone.