Continuous Subcutaneous Insulin Infusion Versus Multiple Daily Injections: Modeling predicted benefits in relationship to baseline A1c

With either continuous subcutaneous insulin infusion (CSII) or multiple daily insulin injection (MDII) therapy, the optimal meal insulin is a rapid-acting analog (lispro or aspart). To date, the efficacy of CSII versus MDII therapy has been evaluated in a limited number of randomized controlled trials in which rapid-acting analogs were used for both regimens. In this context, we recently conducted a pooled analysis (1) using raw trial data from three such studies undertaken in adults with type 1 diabetes (2–4). This analysis suggested that CSII is associated with better glycemic control, particularly in those patients with poor initial control. Indeed, the relative benefit of CSII over MDII was found to increase with higher baseline A1c. To provide direct clinical context, we have now re-analyzed this data to evaluate the impact of CSII and MDII in relation to specific baseline A1c categories using the pooled dataset (139 patients representing 529 patient-months on MDII and 596 patient-months on CSII).

Treatment effect on A1c was studied using a mixed linear modeling approach (MIXED procedure in SAS 9.1.3), with an isotropic exponential spatial covariance structure used to model intrasubject correlation of the repeated measurements and random effects used to model patient, patient*treatment, study, and study*month effects. All fixed and random effects were initially allowed to differ between studies, with the Akaike Information Criterion (AIC) approach used to reduce model complexity. Fixed effects in the final model included the following: 1) baseline A1c, 2) treatment modality, and 3) the interaction between baseline A1c and treatment effect.

These models predicted that, with both MDII and CSII, the reduction in A1c will progressively increase as baseline A1c rises (Fig. 1). Importantly, however, CSII is predicted to achieve a greater reduction in A1c at each level of initial glycemia. Moreover, the relative benefit of CSII over MDII in lowering A1c (i.e., difference in treatment effect between the two modalities) increases as baseline A1c rises. Indeed, for a patient with an A1c of 12%, CSII would be expected to reduce A1c by a full 1% more than MDII.

In summary, this analysis extends our earlier work by providing a quantitative estimate of the anticipated reduction in A1c associated with CSII and MDII therapy, respectively, in relation to baseline A1c. Importantly, the relative benefit of CSII over MDII in lowering A1c increases as baseline A1c rises. This analysis has significant clinical implications in that 1) it provides an anticipated response to intensive insulin regimens in adults with type 1 diabetes, and 2) it highlights the importance of baseline glycemic control as a factor to consider when choosing between CSII and MDII therapy.

It should be noted that NPH insulin provided basal replacement in MDII therapy in the current analysis. Given the superior basal pharmacokinetics of long-acting analogs, future clinical trial comparison between CSII and MDII using both rapid- and long-acting analogs will be of great interest. Based on the current findings, comparison of these treatment modalities across a broad range of baseline glycemic control would be particularly relevant.