Feasibility of integrating research data collection into routine clinical practice using the electronic health record

The electronic health record is becoming central to routine medical practice and has the potential to facilitate large scale clinical research. We evaluated the completeness and accuracy of data collection using designated research fields integrated into a semistructured clinical note. We hypothesized that prospective research data collection as part of routine clinical charting is feasible, with a high rate of utilization (greater than 80%) and accuracy (kappa greater than 0.80).

Infants with congenital hydronephrosis were followed prospectively at a single institution. Existing functionality in the electronic health record was used for data collection by creation of 28 different data elements captured from a hydronephrosis note or phrase template. Completeness (percent utilization) was calculated and accuracy was assessed by comparing the structured data to manual chart review. Comparisons were conducted using the chi-square test, with 2-tailed p values <0.05 considered statistically significant.

A total of 80 patients were eligible for manual chart review. Data were recorded through template use in 64 patients for an overall completeness of 80.0%. Of 28 elements 17 (60%) demonstrated "almost perfect" agreement (kappa greater than 0.80), and all variables reached at least "moderate" agreement (greater than 0.40).

Integrating research fields into routine clinical practice is feasible by using semistructured clinical templates within an electronic health record. High completion and accuracy rates were captured from a variety of fields within a hydronephrosis template.