INFLUENCE OF PROJECTED COMPLICATION RATES ON ESTIMATED APPROPRIATE USE RATES FOR CAROTID ENDARTERECTOMY

Objective. To examine specifically the influence of estimated perioper ative mortality and stroke rate on the assessment of appropriateness o f carotid endarterectomy. Data Sources/Study Setting. An expert panel convened to rate the appropriateness of a variety of potential indicat ions for carotid endarterectomy based on Various rates of perioperativ e complications. We then applied these ratings to the charts of 1,160 randomly selected patients who had carotid endarterectomy in one of th e 12 participating academic medical centers. Study Design. An expert p anel evaluated indications for carotid endarterectomy using the modifi ed Delphi approach. Charts of patients who received surgery were abstr acted, and clinical indications for the procedure as well as periopera tive complications were recorded. To examine the impact of surgical ri sk assessment on the rates of appropriateness, three different definit ions of risk strata for combined perioperative death or stroke were us ed: Definition A, low risk <3 percent; Definition B, low risk <5 perce nt; and Definition C, low risk <7 percent. Principal Findings. Overall hospital-specific mortality ranged from 0 percent to 4.0 percent and major complications, defined as death, stroke, intracranial hemorrhage , or myocardial infarction, varied from 2.0 percent to 11.1 percent Mo st patients (72 percent) had surgery for transient ischemic attack or stroke; 24 percent of patients were asymptomatic. Most patients (82 pe rcent) had surgery on the side of a high-grade stenosis (70-99 percent ). When the thresholds for operative risk were placed at the values de fined by the expert panel (Definition A), only 33.5 percent of 1,160 p rocedures were classified as ''appropriate.'' When the definition of l ow risk was shifted upward, the proportion of cases categorized as app ropriate increased to 58 percent and 81.5 percent for Definitions B an d C, respectively. Conclusions. Despite the high proportion of procedu res performed for symptomatic patients with a high degree of ipsilater al extracranial carotid artery stenosis and generally low rates of sur gical complications at the participating institutions, the overall rat e of ''appropriateness'' using a perioperative complication rate of <3 percent was low. However, the rate of ''appropriateness'' was extreme ly sensitive to judgments about a single clinical feature, surgical ri sk. These data show that before applying such ''appropriateness'' rati ngs, it is crucial to perform sensitivity analyses in order to assess the stability of the results. Results that are robust to moderate in v ariation in surgical risk provide a much sounder basis for policymakin g than those that are not.