Comparative risks of diabetes-related complications of basal insulins: a longitudinal population-based cohort of type 1 diabetes 1999–2013 in Taiwan

Aim: We compared the effects of two types of basal insulin: long-acting insulin analogues vs. intermediate/long-acting human insulin, on diabetes-related complications in type 1 diabetes. Methods: A total of 1188 patients with type 1 diabetes who had recently started on long-acting insulin analogues or intermediate/long-acting human insulin were identified in 2004–2008 and followed until death or the end of 2013. Clinical outcomes included acute (i.e. hyperglycaemia, hypoglycaemia) and chronic (i.e. nephropathy, retinopathy, neuropathy, cardiovascular diseases) complications. Diabetes-related complications were measured as a composite outcome which included acute and chronic complications. Cox proportional hazards models were used to assess the time to event hazard ratio. Three propensity score (PS) methods were applied to adjust for baseline imbalances between basal insulin groups, including the PS-matching approach (as the main analysis), standardized mortality ratio weighting (SMRW) and inverse probability of treatment weighting (IPTW). Results: Long-acting insulin analogues vs. intermediate/long-acting human insulin had a lower risk for a composite of diabetes-related complications {adjusted hazards ratios [aHRs] [95% confidence interval (CI)] 0.782 [0.639, 0.956], 0.743 [0.598, 0.924] and 0.699 [0.577, 0.846] according to the PS-matching approach, SMRW and IPTW, respectively}. Compared with intermediate/long-acting human insulin, using long-acting insulin analogues had a lower hypoglycaemia risk: aHRs (95% CI) 0.681 (0.498, 0.930), 0.662 (0.466, 0.943) and 0.639 (0.471, 0.867) from the PS-matching approach, SMRW and IPTW, respectively. No statistical differences were found between two types of insulin on individual chronic complications. Conclusion: A trend of lower diabetes-related complications associated with long-acting insulin analogues vs. intermediate/long-acting human insulin was observed. A reduced hypoglycaemia risk with long-acting insulin analogues was confirmed in this ‘real-world’ study.