All-Cause and Cause-Specific Mortality among Users of Basal Insulins NPH, Detemir, and Glargine.

BACKGROUND: Insulin therapy in type 2 diabetes may increase mortality and cancer incidence, but the impact of different types of basal insulins on these endpoints is unclear. Compared to the traditional NPH insulin, the newer, longer-acting insulin analogues detemir and glargine have shown benefits in randomized controlled trials. Whether these advantages translate into lower mortality among users in real life is unknown. OBJECTIVE: To estimate the differences in all-cause and cause-specific mortality rates between new users of basal insulins in a population-based study in Finland. METHODS: 23 751 individuals aged ≥40 with type 2 diabetes, who initiated basal insulin therapy in 2006-2009 were identified from national registers, with comprehensive data for mortality, causes of death, and background variables. Propensity score matching was performed on characteristics. Follow-up time was up to 4 years (median 1.7 years). RESULTS: 2078 deaths incurred. With NPH as reference, the adjusted HRs for all-cause mortality were 0.39 (95% CI, 0.30-0.50) for detemir, and 0.55 (95% CI, 0.44-0.69) for glargine. As compared to glargine, the HR was 0.71 (95% CI, 0.54-0.93) among detemir users. Compared to NPH, the mortality risk for both cardiovascular causes as well as cancer were also significantly lower for glargine, and especially for detemir in adjusted analysis. Furthermore, the results were robust in various sensitivity analyses. CONCLUSION: In real clinical practice, mortality was substantially higher among users of NPH insulin as compared to insulins detemir or glargine. Considering the large number of patients who require insulin therapy, this difference in risk may have major clinical and public health implications. Due to limitations of the observational study design, further investigation using an interventional study design is warranted.

A note on estimating the posterior density of a quantitative trait locus from a Markov chain Monte Carlo sample.

We provide an overview of the use of kernel smoothing to summarize the quantitative trait locus posterior distribution from a Markov chain Monte Carlo sample. More traditional distributional summary statistics based on the histogram depend both on the bin width and on the sideway shift of the bin grid used. These factors influence both the overall mapping accuracy and the estimated location of the mode of the distribution. Replacing the histogram by kernel smoothing helps to alleviate these problems. Using simulated data, we performed numerical comparisons between the two approaches. The results clearly illustrate the superiority of the kernel method. The kernel approach is particularly efficient when one needs to point out the best putative quantitative trait locus position on the marker map. In such situations, the smoothness of the posterior estimate is especially important because rough posterior estimates easily produce biased mode estimates. Different kernel implementations are available from Rolf Nevanlinna Institute's web page (http://www.rni.helsinki.fi/;fjh).