Leveraging automated decision support tools to aid performance measurement: An examination of variation in thiazide use among patients with uncomplicated hypertension

Background: Despite evidence-based guidelines which favor thiazide diuretics, use of these drugs remains low. Understanding of the extent to which apparent nonadherence to guidelines may represent clinically appropriate prescribing decisions is limited.

Objective: To characterize the patient-level factors associated with thiazide use among patients with treated uncomplicated hypertension.

Methods: We conducted a secondary analysis using baseline data from a 15-month randomized evaluation of the impact of ATHENA-HTN, an automated decision support system for hypertension management, on guideline concordance and blood pressure control. Primary care clinicians at nine sites within three VA Health Care Systems were randomized to receive either ATHENA advisories with patient-specific recommendations or pop-up hypertension reminders. We included in our analysis patients with uncomplicated hypertension who were treated with one or more antihypertensive medications during the study period, who had no strong contraindication to thiazides, and who were not on furosemide at baseline. Thiazide diuretics were preferred drugs for these patients under the JNC-6 guidelines in effect at the time; recommendations to prescribe thiazides are also included in the current JNC-7 guidelines. T-tests and χ2 statistics were used to test the relationship between a variety of patient demographic and clinical characteristics, and whether the patient had an active prescription for thiazide at baseline.

Results: Thirty-eight percent of patients with uncomplicated hypertension were on a thiazide at baseline. Patients were significantly less likely to have an active prescription for thiazide if they had a prior diagnosis of uncomplicated gout (34% vs. 39%, p=0.025), benign prostatic hypertrophy (33% vs. 41%, p<0.0001), diabetes (36% vs. 40%, p=0.005), or other relative indication for a non-thiazide antihypertensive drug (33% vs. 42%, p<0.0001). Among patients who were not on a thiazide but had a relative indication for a non-thiazide drug, 79% were being treated with the recommended non-thiazide drug; of those patients who were treated with a recommended non-thiazide drug, 63% had blood pressures above target at their first primary care visit during the study period.

Conclusions: Conditions associated with lower use of thiazide diuretics in some cases may represent appropriate clinical judgments. At the same time, a substantial proportion of patients who are treated with other recommended antihypertensive medications could benefit from the inclusion of a thiazide as part of a multi-drug regimen. Automated decision support systems, which make use of electronically available clinical information to provide patient-specific assessments, can be leveraged to conduct refined analyses of performance.