Correction of an unexpected increasing trend in glucose measurements during 7 years recruitment to a cohort study

ObjectivesTo study an unexpected, significant increase in glucose measurements during 7 year recruitment to a cohort study.Design and methodsMeasurements of quality control solutions and blood glucose in pregnant women were done by Accu-Chek Sensor glucometer. Time-trends were analysed by regression models and control charts.ResultsCohort measurements were de-trended by weighted linear regressions based on independent control values.ConclusionsBiologically implausible trends in data can be corrected by using independent control values.

► Glucose values increased unexpectedly during a 7 year study.
► Data from quality controls, BMI and insulin pointed to the point-of-care glucometer.
► Blood glucose measurements were de-trended by statistical regression techniques.
► Regular quality controls and control charts should be used.