Objectives: To evaluate the implementation of clinical guidelines for diabetes mellitus in general practice with a specific computer-based clinical decision support system (CDSS) as part of the intervention.

Methods: Randomized study with health center as unit. General practice in Sør- and Nord-Trøndelag counties in Norway, 380,000 inhabitants. Seventeen health centers with 24 doctors and 499 patients with diabetes mellitus were in the intervention group and 12 health centers with 29 doctors and 535 patients were in the control group. Main outcome measures were group differences in fractions of patients without registrations (process evaluation) and mean group differences for the same variables (patient outcome evaluation).

Results: Statistically significant group differences were experienced for fractions of patients without registration of cigarette smoking (intervention group, 82.6%; control group 94.5%), body mass index (78.2% vs. 93.0%), and sufficient registrations for calculation of risk score for myocardial infarction (91.1% vs. 98.3%); all during 18 months. Large center variations were shown for all variables. The only statistically significant group difference was −2.3 mm Hg (95% CI, −3.8, −0.8) in diastolic blood pressure in favor of the intervention group. Statistically insignificant differences in favor of the intervention group were HbA1c, −0.1% (95% CI, −0.4, 0.1), systolic blood pressure, −1.2 mm Hg (95% CI, −4.4, 2.0). Statistically insignificant differences in favor of the control group were fractions of smokers, +3.0% (95% CI, −4.0, 10.0), body mass index, +0.3 kg/m2 (95% CI, −0.8, 1.4), risk score in female +0.1 (95% CI, −5.1, 5.2), and risk score in male +2.6 (95% CI, −14.2, 19.5).

Conclusions: Implementation of clinical guidelines for diabetes mellitus in general practice, by means of a CDSS and several procedures for implementation, did not result in a clinically significant change in doctors' behavior or in patient outcome.