Objectives

The use of continuous subcutaneous insulin infusion (CSII) is a widely used therapeutic strategy for achieving near-normal blood glucose for people with type 1 diabetes (T1D). Clinical trials have shown that CSII can improve glycated hemoglobin (HbA1c) compared with multiple daily injections (MDI). However, whether this is the case in the real world remains largely unresolved. Thus, the objective of this study was to estimate the effect of CSII on HbA1c compared with MDI, using real-world population data.

Methods

A quasi-experimental, longitudinal study from 2010-2020 of all people with T1D in Denmark was designed based on prospectively collected, register-based data. All living adult individuals with T1D were identified using a validated algorithm and were available for inclusion in the final cohort. The cohort was open, i.e., new individuals entered the cohort upon diagnosis with T1D. CSII users were distinguished from MDI users algorithmically based on procedure codes for CSII and CSII-specific prescribed insulin.

The identification strategy to estimate the average treatment effect of the treated (ATT) of CSII exploited the longitudinal and treatment-staggered data structure by using a doubly robust difference-in-differences estimator. Event study analyses were conducted, and heterogeneity effects were examined for sex, age, education, HbA1c at treatment initiation and continuous glucose monitor (CGM) use.

Results

An ATT of -3.54mmol/mol (95%CI: -4.15 – -2.94) was identified for HbA1c; event study analysis revealed the largest decrease within the first year of treatment, which then stabilized. Decreases were largest among women, the relatively older, those with high HbA1c at treatment initiation, and CGM users.

Conclusions

On average, CSII therapy results in decreased HbA1c compared with MDI, particularly within the first year of therapy. Compared to clinical trials, however, the decease is smaller in magnitude. Importantly, not everyone benefits the same as evident in subgroup analyses.