Aim: We performed a meta-analysis of randomised controlled trials (RCTs) with insulin analogues in type 2 diabetes utilising a least-squared regression model in order to assess the relationship between baseline HbA1c, the magnitude of HbA1c decrease and attainment of HbA1c target of < 7%.
Methods: Randomised controlled trials involving insulin regimens (basal, prandial, biphasic and basal-bolus) were identified through electronic searches (MEDLINE, EMBASE, CINAHL and The Cochrane Library) through September 2010. We included any study arm of RCTs if they were at least 12 weeks in duration; the number of patients in any arm was more than 30 and reported the baseline HbA1c and change from baseline HbA1c.
Results: We found 87 studies, with a total of 135 arms, and 38,803 patients. The weighted R(2) values for the overall analysis assessing the association between baseline HbA1c and absolute change in HbA1c or the proportion of patients at target were 0.485 (p < 0.001) and 0.146 (p < 0.001), respectively. Subanalyses of insulin regimens for the association between basal HbA1c and absolute decrease of HbA1c produced weighted R(2), which were significant for all insulin regimens with the highest association for basal-bolus (R(2) = 0.719, p < 0.001).
Conclusions: The strong positive relationship between baseline HbA1c and the magnitude of HbA1c change we found in RCTs using insulin analogues in type 2 diabetes should be considered when assessing the clinical efficacy of insulin therapies.
© 2011 Blackwell Publishing Ltd.