Implementation of a screening program to detect previously undiagnosed dysglycemia in hospitalized patients

Many people with dysglycemia are unaware that they have the condition. We conducted a study to determine whether a screening program for hospitalized patients could identify new cases of unrecognized dysglycemia and affect the actions of attending care providers during hospitalization.

We measured A1C in 466 participants with no history of diabetes who had been admitted to hospital for coronary heart disease or elective joint replacement surgery. Participants with A1C <6.0% were considered normoglycemic and those with A1C ≥6.0% were considered dysglycemic. Notifications to care providers were placed on the charts of participants who had dysglycemia, along with recommendations for in-hospital monitoring and care. Oral glucose tolerance tests were completed 6 weeks post-hospitalization for participants with dysglycemia and a subsample of participants who were normoglycemic. Sensitivity and specificity of in-hospital dysglycemia criteria were calculated. Provider practices were determined by chart review.

In-hospital dysglycemia was present in 10.4% of patients with coronary heart disease and 11.4% of participants with elective joint replacement surgery. Attending care providers took few of the recommended actions, despite the chart notification of dysglycemia; glucose monitoring occurred <30% of the time. The in-hospital dysglycemia criterion of ≥6% demonstrated moderate sensitivity (47.5%) and high specificity (96.2%) in detecting dysglycemia based on oral glucose tolerance tests.

Dysglycemia was a relatively common finding in patients with no history of diabetes who had been admitted for coronary heart disease or elective joint replacement surgery. The in-hospital A1C screening criteria generated a high level of false-negative tests, and a chart notification had limited effects on the practices of attending care providers. Future studies examining lower A1C thresholds and the barriers to and facilitators of attending care providers' behaviours are warranted.