Improving Hypertension Quality Measurement Using Electronic Health Records S Persell, AN Kho, JA Thompson, DW Baker Feinberg School of Medicine Northwestern University Chicago, Illinois Supported by award 1 K08 HS from the Agency for Healthcare Research and Quality
Problems with Current Quality Measures l Simple intermediate outcome measures (e.g., blood pressure at last visit <140/90) may not reliably indicate who is truly receiving poor care –A pt with controlled blood pressure runs out of meds and comes to clinic with BP 150/100 –A pt with coronary disease had an LDL cholesterol of 220 mg/dl, which decreased to 110 on a maximal dose of a statin but did not reach the goal of LDL < 100
Adverse Consequences l These limitations problematic as incentives based on performance measures increase l When used for internal quality improvement, measurement errors such as these may cause physicians to reject measure validity
…and the Solution? l Develop quality measures that more accurately capture what would be defined as poor care –i.e. higher specificity of failures l Electronic health records can help facilitate implementation of more complicated measures
Study Aims l To develop and apply increasingly more sophisticated measures of hypertension quality utilizing data available within an EHR l To compare the results of measured quality using simple outcome measures and more sophisticated measures
Methods l Design: retrospective observational cohort study l Setting: urban Internal Medicine practice with a commercial EHR (Epic) l Eligibility –Hypertensive adults with 3 or more clinic visits between 7/05 and 12/06
Baseline Quality Measure l Baseline: –Patients with hypertension recorded on their problem list, past medical history, or encounter diagnosis codes –Blood pressure at last visit <140/90 –Blood pressure <130/80 if comorbid diabetes
Quality Measure 2: Relax Cutoff l Include last BP ≤ goal as satisfying measure ≤ 140/90 ≤ 130/80 if diabetes
Quality Measure 3: Incorporate Average BP l If either the last or mean of last three BPs are at goal, the patient is considered to satisfy the measure
Quality Measure 4: Account for Aggressive Management l Include patients prescribed 3 or more different antihypertensive drug classes including a diuretic as satisfying the measure –Beta blocker, calcium channel blocker, ACE or ARB, peripheral alpha blocker, centrally acting anti-adrenergic drug, or direct vasodilator –AND diuretic
Quality Measure 5 Account for Low Diastolic Blood Pressure, A Safety Concern l Studies suggest that for pts with coronary artery disease and diabetes, lowering the diastolic BP below 70 mmHg may be harmful l Therefore, if patients with uncontrolled systolic blood pressure had diastolic pressure < 70 mmHg, they were consider to satisfy measure
Quality Measure 6: Include Patients with Undiagnosed Hypertension l Include in denominator patients with a mean blood pressure ≥140/90 mmHg or ≥130/80 mmHg if the patient has comorbid diabetes even if they do not have hypertension recorded as a diagnosis
Study Population NAge, mean (SD) Female, % Diagnosed hypertension No diabetes Diabetes (14) 61 (12) Undiagnosed hypertension No diabetes Diabetes (15) 51 (12) 44 46
Variation Across Measures (no DM) Last BP < 140/90 58% Last BP ≤ 140/90 67% Last or mean ≤ 140/90: 76% ≤ 140/90 or 3 drugs with diuretic: 83% ≤ 140/90 or 3 drugs w/ diuretic or low DBP: 84% Include undiagnosed hypertension: 81%
Last BP < 130/80 30% Last BP ≤ 130/80 39% Last or mean ≤ 130/80: 47% ≤ 130/80 or 3 drugs with diuretic: 73% ≤ 130/80 or 3 drugs w/ diuretic or low DBP: 76% Include undiagnosed hypertension: 73% Variation Across Measures (DM)
Results of Standard vs. Advanced Hypertension Quality Measures } Δ 23% } Δ 43%
Limitations l We used hypothetical quality measures to demonstrate concept l Single site; generalizability not known –Would be difficult, but not impossible, to apply measures at sites without an EHR l Data within EHRs may be incomplete l Still may miss important exceptions –Home blood pressure controlled
Conclusions l Small changes in measure criteria produce large changes in measured quality l Many patients who did not satisfy the simple measure were receiving aggressive care l More sophisticated measures may better align external measurement with internal quality improvement
Implications l More sophisticated measures may: –Improve detection of true quality problems that need attention by MDs and other staff –Remove incentives to stop caring for patients with resistant hypertension –Remove incentives to unsafely or unnecessarily over treat some patients