Management of type 2 diabetes mellitus and prediction of cardiovascular complications

Cardiovascular disease is the leading complication of type 2 diabetes and approximately half of the patients with type 2 diabetes will die of a cardiovascular cause. To prevent the occurrence of cardiovascular disease, effective diabetes treatment is necessary. Up-titrating diabetes medication by addition of an oral drug or commencement of insulin improves glycemic control. However it has been suggested that glucose lowering agents are associated with weight gain, which could increase the incidence of cardiovascular disease. In our study medication combinations including insulin or thiazolidinedione were associated with increase in weight of 3 kg, while sulphonylureas and metformin were weight neutral. Moreover we assessed the effect of blood pressure lowering agents among moderate-high and very high cardiovascular risk groups. Absolute treatment effects tended to be different between the two risk groups with greater absolute risk reduction for the very high risk group, but this did not reach significance. To accurately examine the cardiovascular risk of a patient general practitioners often use risk scores. Performing a systematic review of the literature 12 cardiovascular risk scores were identified specifically designed for diabetes patients, and 33 prediction models applicable to diabetes patients because they included diabetes as a factor. Even though many risk scores have not been validated or after validation performed only moderately to poor, several risk scores have been included in clinical practice guidelines. A complete assessment of all cardiovascular risk scores designed specifically for diabetes patients was conducted. Discrimination of all risk scores was poor to moderate. Most risk scores severely overestimated the risk except for the ADVANCE risk score which underestimated the risk. After recalibration, the calibration of all models was good, (all p-values>0.05) with a slight overestimation of the actual risk. The model with the best calibration in our cohort was the Swedish NDR risk score which was then used to assess the added prognostic value of Heart-type Fatty Acid-Binding Protein (h-FABP) for the prediction of cardiovascular disease among diabetes patients. A non-significant positive association between H-FABP and CVD was observed. The small improvement in the discrimination has only little clinical value for cardiovascular risk assessment. Patients with type 2 diabetes routinely visit their physician regularly, so in clinical practice repeated risk factor information is collected over time. We assessed how to incorporate repeated measurements in clinical prediction models and if this affected the predictive performance using cardiovascular risk assessment among diabetes patients as an example. This study suggests that use of a baseline model is as predictive as recalculating the risk at new time points or fitting a new model using repeated measurements. Only small changes in discrimination were observed for models using updated information