Only a few studies have reported a long-term follow-up in a significant number of patients using continuous subcutaneous insulin infusion (CSII) (1). Who stands to benefit more from this costly insulin therapy is still unclear.
The aim of our observational, retrospective study was to evaluate the possible predictors of the degree of improvement of metabolic control with CSII in 82 consecutive type 1 diabetic patients (age 37.9 ± 13.4 years, 42 men and 40 women, duration of diabetes 19.7 ± 9.9 years) who started CSII in the Diabetes Unit of Bergamo Hospital between June 1999 and March 2004.
The patients had been treated with multiple daily injection (MDI) therapy (regular [n = 22] or rapid-acting analog insulin [n = 60] before meals plus NPH [n = 72] or glargine [n = 10] as basal insulin) for at least 1 year. During CSII treatment, lispro or aspart analogs were used.
All patients were evaluated every 3 months both before and during CSII. The mean duration of CSII treatment was 31.9 ± 14.5 months (range 4–55).
Only three patients discontinued CSII. Every patient performed self-monitoring of blood glucose (SMBG) (four to seven daily determinations).
Data are expressed as means ± SD. The means of parametric data of the period of MDI treatment were compared with those of the CSII period using the Student’s t test for paired data. The differences between groups were compared using the Student’s t test for unpaired data.
Compared with MDI therapy, HbA1c (A1C) significantly decreased with 3 months of CSII therapy (CSII vs. MDI: 8.35 ± 1.06 vs. 9.39 ± 1.35%, P < 0.001). The significant decrease of A1C was maintained over the whole CSII treatment with a mean change in A1C of 1.15 ± 0.84% (P < 0.001).
During CSII treatment, as compared with MDI treatment, there was a significant decrease of severe hypoglycemic episodes (0.35 ± 0.07 per patient/year during MDIs vs. 0.10 ± 0.02 during CSII, P < 0.001) and insulin requirement (52.1 ± 17.5 units/day vs. 38.8 ± 12.3, P < 0.001). CSII was not associated with any significant increase in BMI. Incidence of ketoacidosis was negligible during both MDI and CSII treatment.
To evaluate the possible predictors of CSII effect on A1C changes, multiple linear regression analysis performed on all patients revealed that age (β = 0.16, P = 0.05) and baseline A1C (β = 0.21, P = 0.008) were independently associated with A1C improvement after 3 months of CSII (F = 5.41, adjusted R2 = 0.28). BMI, diabetes duration, insulin requirement, and frequency of SMBG were unrelated to A1C changes. Age and baseline A1C were even better predictors of the mean A1C changes during the whole follow-up period (F = 11.87, adjusted R2 = 0.48).
This observational, clinic-based study of a significant number of type 1 diabetic patients who represent all pump-treated patients in the province of Bergamo, Italy, confirmed that CSII significantly decreased A1C levels with respect to MDIs, as reported in a recent meta-analysis of 52 CSII studies (2). The reduction of severe hypoglycemic episodes and the concomitant negligible frequency of ketaoacidosis confirmed the safety of CSII (3,4).
Although most of our patients during MDI treatment used rapid-acting analogs, only a few used glargine as basal insulin. It is possible that the extensive use of this long-term analog during MDI treatment would reduce the difference in metabolic control, as suggested (5).
There are discordant data about the persistence of initial lowering of A1C achieved with CSII (6). Our data showed that improved glucose control persisted during the whole long-term follow-up period, which was longer (mean duration 2.6 years) than most CSII studies.
Most important, our study demonstrated that CSII was particularly advantageous in patients with the poorest metabolic control, as suggested by a randomized controlled trial that compared CSII and MDI treatments using a rapid-acting analog (7). In our population, those with a baseline A1C >10% (n = 25) had an average decline in A1C of 1.5 ± 0.6%, significantly greater (P < 0.001) than that observed in those with a baseline value <8% (n = 16; average decline 0.6 ± 0.5%).
CSII was also more advantageous in patients older than 50 years (n = 14; average A1C decline 1.45 ± 0.7%) than in those younger than 20 years (n = 11; average A1C decline 0.5 ± 0.8%; P < 0.01). A similar observation suggested that CSII is useful and safe in older adults with type 1 diabetes (8).
In conclusion, we suggest that in type 1 diabetic patients who have sufficient ability to master CSII therapy, a poor metabolic control, despite MDI therapy, and older age are better predictors of CSII efficacy.
