1022 — Effect of Reporting Worrisome Results to Primary Care Providers on Processes and Outcomes of Care
Fihn SD (COE Seattle) , Bryson CL
(COE Seattle), McDonell MB
(COE Seattle), Fan VS
Although audit/feedback improves care only marginally, many healthcare systems, including VHA, routinely notify clinicians about results falling outside predetermined limits. We reexamined data from ACQUIP (Ambulatory Care Quality Improvement Project) to determine whether alerting providers to worrisome values yielded any discernible effect.
ACQUIP was a 2-year, firm-randomized trial conducted in 7 VA primary care (PC) clinics. Responses to mailed questionnaires addressing 6 chronic conditions were merged with inpatient, outpatient, pharmacy, and laboratory records. Intervention clinicians received individual visit-based reports regarding clinical findings (e.g., BP, HbA1c), symptoms (e.g., angina, depression) and health habits (e.g., excessive alcohol intake) and periodic aggregate reports including information on patients with worrisome results (criteria predefined by expert panels). Control clinicians received no reports. For patients with worrisome results, we examined number of visits, existing and new prescriptions for relevant drugs, medication adherence, and applicable clinical measures in the year following feedback (means).
Of 22,413 initial entrants, the number of patients whose provider received a report of worrisome results and total number with each condition were: coronary disease (762/8152), hypertension (438/11,945), diabetes (555/4928), COPD (812/4983), depression (1059/6598), and problem drinking (401/5858). Six of 34 comparisons favored the intervention and one favored control. Intervention patients with worrisome coronary disease (angina or nitroglycerine =1-2 times/week) had slightly more PC visits than controls (1.24 vs. 1.14, p=0.002). Although diabetics in the intervention group with HbA1c >8 had more PC visits (1.33 versus 1.19, p=0.002), more diabetic medications initiated (0.79 versus 0.63, p=0.004), and more HbA1c tests (0.59 versus 0.47, p=0.001), follow up HbA1cs were not different between groups (p=0.65). Intervention patients with uncontrolled hypertension started more new antihypertensives (1.10 versus 0.93, p=0.046), but follow-up blood pressures were not different between groups (systolic p=0.83, diastolic p=0.28). Problem drinkers (intervention) had significantly more PC visits, and depressed intervention patients filled fewer antidepressants. Mean follow-up depression and alcohol consumption scores, however, were comparable.
The ACQUIP trial demonstrated no overall effect of audit/feedback. Informing PC providers specifically about therapeutic failures minimally increased utilization but failed to improve outcomes.
Notification about worrisome results may expend resources without clinically demonstrable effect.