15 April 2010
A large observational study found a statistically significant association between all-cause mortality and exposure to insulin in people with type 2 diabetes, with nearly a three times greater risk in the highest exposure group compared with never-users. This dose-response relationship was also seen with the secondary outcomes of cardiovascular and non-vascular mortality.
Level of evidence:
Level 2 (limited quality patient-oriented evidence) according to the SORT criteria.
NICE guidance on the management of type 2 diabetes recommends that patients should be involved in setting their individual glycated haemoglobin (HbA1c) target, which may be above the general target of 6.5% (48mmol/mol). NICE recommends adding a second oral drug where patients are unable to reach their agreed target by lifestyle modification or monotherapy with an oral hypoglycaemic agent (usually metformin). Insulin therapy is suggested by NICE as an option where the HbA1c does not fall below 7.5% (59mmol/mol), or does not reach an agreed higher target. Health professionals may wish to consider the implications of the present study, and others we have blogged, in their discussion with patients about risks and benefits of intensifying drug treatment, setting of individual HbA1c targets and use of insulin in type 2 diabetes mellitus.
What is the background to this?
The ACCORD, ADVANCE and VADT randomised controlled trials (RCTs) did not identify a consistent significant benefit of intensive glycaemic control in the treatment of type 2 diabetes with regard to cardiovascular (CV) outcomes and mortality. Indeed, the ACCORD study, which compared treatments that aimed to achieve HbA1c levels of less than 6.0% (42mmol/mol) with those of between 7.0% (53mmol/mol) and 7.9% (63mmol/mol) was stopped early because of an increased risk of death in the intensive treatment arm. It is unclear whether insulin use was associated with the increased mortality risk seen in the intensive treatment arm of the ACCORD study. (See Blog No.1021 on the post-hoc analysis of ACCORD).
This cohort study compared rates of all cause mortality in more than 12,000 newly treated people with type 2 diabetes exposed to various levels of insulin. Time at risk in patients who received an average of fewer than three insulin prescriptions per year was classed as low exposure, 3 to 12 prescriptions per year (<1 vial/month) was classed as moderate exposure and 12 prescriptions or more per year (> 1 vial/month) was classed as high exposure. Secondary outcomes were CV mortality and non-vascular-related mortality.
What does this study claim?
The study found a statistically significant dose-related association between all cause mortality and insulin exposure level. After adjusting for confounders, the risk of dying varied from about 75% higher in the lower exposure group to 179% higher in the highest exposure group, compared to the non-exposed group (low exposure hazard ratio [HR] 1.75, 95%CI 1.24 to 2.47; moderate exposure HR 2.18, 95%CI 1.82 to 2.60; high exposure HR 2.79, 95%CI 2.36 to 3.30). A similar dose-response relationship was seen for the secondary outcomes: CV mortality and non-vascular mortality increased with increasing exposure to insulin.
Although, the study suggests that increasing levels of insulin exposure are associated with higher levels of mortality in patients with type 2 diabetes, the study has several limitations which mean that its findings should be interpreted with caution. Observational data such as these are subject to bias. Unlike an RCT, in ‘real life’ treatment plans are chosen, changed, or actively not chosen in the light of individual patients’ risk factors, preferences and tolerability or response to other drugs. Therefore, observed differences in outcomes may well be due to differences among the patients, not only the different treatments. Although adjustments were made for many potential confounders in this study, no adjustments could be made for clinical variables such as HbA1c, body mass index and smoking status. In addition, some variables were estimated using proxy markers and may not have been accurate. For example, number of insulin units per day prescribed was used as a marker for insulin dosage per day, and presence of micro- and macrovascular complications was considered as a marker for disease severity. It is possible that some residual bias or confounding could provide an alternate explanation for the results.
The authors of the paper make these points themselves in the paper. However, they also point out that the magnitude of the excess mortality with insulin is so great that according to their calculations this could, for example, only be explained by an imbalance in baseline HbA1c of 10% – a value which is improbable.
The findings of another recent large retrospective cohort study of patients with type 2 diabetes receiving intensive glucose control treatment suggested that the risk of all-cause mortality increases above and below an HbA1c level of about 7.5% (59mmol/mol). In that study, intensifying treatment with insulin was associated with a greater risk of these events than intensifying treatment with oral hypoglycaemic agents (see Blog No. 1017).
NICE guidelines recommend that further intensification of therapy, usually with insulin, should be considered for those patients who are unable to achieve an HbA1c of 7.5% on dual oral hypoglycaemic therapy. Importantly however, NICE also advises that patients should be involved in setting their individual target. Health professionals may wish to consider the implications of these two observational studies in their discussion with patients regarding the risks and benefits of further intensifying therapy, especially if insulin therapy is being considered.
Gamble J –M, et al. Insulin use and increased risk of mortality in type 2 diabetes: a cohort study. Diabetes, obesity and metabolism 2010;12:47–53
Design: Population-based cohort study using administrative databases in Saskatchewan, Canada.
Patients: 12,272 new users of oral hypoglycaemic therapy between 1991–96, grouped according to cumulative insulin exposure based on total insulin dispensations per year: no exposure (reference group); low exposure (0 to <3); moderate exposure (3 to <12) and high exposure (12 or more).
Intervention and comparison: The relationship between insulin exposure was compared with all-cause mortality (primary outcome), CV-related mortality and non-vascular mortality (secondary outcomes) after adjusting for demographics, medications and co-morbidities.
Results: Average age was 65 years, 45% were female and mean follow up was 5.1 years. In total, 1,443 (12%) subjects started insulin, and 2,681 (22%) deaths occurred. After adjustment, a graded risk of mortality associated with increasing exposure to insulin was observed: low exposure adjusted HR 1.75, 95% CI 1.24 to 2.47; moderate exposure adjusted HR 2.18, 1.82 to 2.60; high exposure adjusted HR 2.79, 2.36 to 3.30; P=0.005 for trend. Analyses restricted to CV-related (P=0.042 for trend) and non-vascular (P=0.004 for trend) mortality showed a similar relationship. Adjusted HRs for CV-related mortality were 1.65 (95%CI 0.82 to 3.32), 1.71 (95%CI 1.17 to 2.50) and 2.51 (1.81 to 3.50) respectively. For non-vascular mortality they were 1.78 (95%CI 1.19 to 2.65), 2.34 (1.91 to 2.87) and 2.90 (2.39 to 3.52) respectively.
Sponsorship: Canadian Institute for Health Research (CIHR) Team Grant to the Alliance for Canadian Health Outcomes Research in Diabetes, sponsored by the CIHR Institute of Nutrition, Metabolism and Diabetes.
Make sure you are signed up to NPC Email updates — the free email alerting system that keeps you up to date with the NPC news and outputs relevant to you