BACKGROUND: Assessment of quality of care using classical threshold measures (TM) is open to debate. Measures that take into account the clinician's actions and the longitudinal nature of chronic care are more reliable, although their major limitation is that they require more sophisticated electronic health records. We created a clinical action measure (CAM) for the control of LDL and non-HDL cholesterol from low-complexity data, and investigated how quality of care in individual diabetes centres based on the CAM is related to that based on the classical TM.

METHODS: Data was used from 3421 diabetes patients treated in 95 centres, collected in two consecutive retrospective data collections. Patients met the TM when their index value was below target. Patients met the CAM when their index value was below target or above target but for whom treatment initiation or intensification, or possible contraindication, was indicated.

RESULTS: Based on the TM, 60-70 % of the patients received good care. This percentage increased significantly using the CAM (+5 %, p < 0.001). At the centre level, the CAM was associated with a higher median score, and a change in position among centres ('poor', 'good' or 'excellent' performer) for 5-10 % of the centres.

CONCLUSIONS: Judging quality of diabetes care of a centre based on a TM may be misleading. Low-complexity data available from a quality improvement initiative can be used to construct a more fair and feasible measure of quality of care.