To evaluate, under routine conditions, the relation between different diabetes care policies and glycemic control through a by-center analysis procedure aimed at reducing some drawbacks of cross-sectional data.


A survey on insulin-treated diabetes care management (IDDM and NIDDM) involved 16 Italian randomly selected diabetes outpatient clinics. A total of 2,142 representative patients were investigated. The standardized HbA1c average value of each center was related, by regression models, to some indicators of center care policy (average number of injections, average BMI, proportion of cases with recent fundus oculi examinations, or frequent visits) as well as to patients' average social levels (employment type). Homogeneity in patient admission criteria is assumed among the investigated centers as a basic condition for the procedure validity. Some known imbalance were controlled for both design and analysis.


HbA1c showed a univariate inverse relation with daily number of injections in IDDM (P = 0.0009, r2 = 0.56) but not in NIDDM (P = 0.33). It was inversely related to both fundus examination (IDDM P = 0.04; NIDDM P = 0.099) and qualified employment (IDDM P = 0.06; NIDDM P = 0.026). A stepwise regression analysis left in the model insulin injections (P = 0.0002) in IDDM (total r2 = 0.68) and qualified employment (P = 0.016) and fundus examination (P = 0.14) in NIDDM (total r2 = 0.53), after controlling for age, sex, disease duration, insulin therapy starting delay, and insulin dose per kilogram.


These results suggest that the confirmed benefits of a multiple-injection regimen in IDDM cannot be simply extrapolated to NIDDM, where patients' awareness and medical attention to complications proved to be the most important factors in current practice.

This content is only available via PDF.