Algorithms applied to electronic health record data can identify more cases of undetected diabetes in patients than claims codes, and discriminates between type 1 and type 2 diabetes, according to a study published recently in Diabetes Care.
Diabetes can be better controlled if detected and treated early. The researchers surmised that simply analyzing claims codes was not an adequate way to identify individuals who have undiagnosed diabetes.
To that end, they created an algorithm to analyze four years of data from the EHR of a large multisite, multispecialty ambulatory practice. The algorithm was applied to the EHR data in new and multiple combinations to flag patients who may have diabetes, reviewing lab test results, diagnosis codes and prescriptions.
The researchers found that the algorithm provided more complete diabetes surveillance and increased case capture. The positive predictive value of the combinations of data was 94 percent for type 1 codes alone; comparing type 1 and type 2 codes, the algorithm correctly identified 100 percent of patients with type 1 diabetes.
Other studies have shown that EHRs can improve the treatment and control of diabetes. EHR data also can improve care from a public health standpoint, as the use of diabetes registries can help improve research and create more effective prevention and treatment techniques.
Read an official abstract of the study here.