Stroke associated with cardiac surgery. Determinants, timing, and stroke subtypes.

Objective: To determine the nature of strokes complicating cardiac surgery.
Design: A medical record review was undertaken of all patients who underwent cardiac surgery (either coronary bypass, valvular surgery, or both) between January 1990 and July 1995. Univariate and multivariate analyses were done using odds ratios (ORs) and logistic regression.
Results: A total of 2211 patients underwent cardiac surgery. A total of 44 patients (2%) had postoperative strokes. They were compared with 104 surgical controls. Univariate analysis revealed that the patients with stroke were significantly older and had significantly higher rates of transient ischemic attack, congestive heart failure, and peripheral vascular disease by history. A multivariate logistic regression revealed the significant variables associated with stroke were congestive heart failure (OR, 6.8) and transient ischemic attack (OR, 1.2). Analyses of surgical variables revealed that bypass pump time of more than 120 minutes (OR, 1.40) was the only significant predictor. The majority of strokes (61%) had occurred by postoperative day 2, but 39% occurred between postoperative days 3 to 9. Hemispheric syndromes occurred in 70%, brain stem and cerebellar syndromes in 14%, and lacunar syndromes in 16%. Computed tomographic scans revealed that 29 patients had new infarcts, of which 20 (69%) were territorial, 5 (17%) were small deep, and 4 (14%) were border zone. Hemorrhagic infarction was found in 2 cases (5%). One patient (2.5%) had a cardiorespiratory arrest while undergoing computed tomography. Discharge disposition was good in 88% of patients with stroke (12% with poor outcomes) vs 97% of controls with good outcomes (3% with poor outcomes) (P = .04).
Conclusions: Strokes may be delayed following cardiac surgery perhaps because of ongoing risk of embolism or a hypercoagulable state. A surprisingly high proportion of patients have lacunar syndromes or small-deep infarcts shown on computed tomography. Early computed tomographic imaging may be useful to exclude hemorrhage if anticoagulation is considered, but is not without risk in these potentially unstable patients. Greater understanding of risk factors for stroke, timing, and subtypes may ultimately allow identification of patients at particularly high risk for perioperative stroke.