Nosocomial stroke

JStroke Cerebrovasc Dis 1993;3:112-114 © 1993 National Stroke Association

Nosocomial Stroke Stephen J. Marks, M.D., Michael D. Bennett, B.S., Ijames Zisfein, M.D., and Brij M. Singh, M.D.

Nosocomial stroke occurs during hospitalization for unrelated problems. Increased understanding of this relatively ignored entity may provide the key to improved stroke prophylaxis for the hospitalized patient at risk and provide clues to the precipitants of stroke in the general population. We compared nosocomial stroke to stroke occurring outside the hospital in a mixed prospective and retrospective analysis of 372 consecutive strokes occurring over 2 years. We excluded nosocomial stroke associated with cardiac bypass surgery, carotid endarterectomy, and cerebral angiography for cerebrovascular disease because of the known associations of these procedures with stroke. Of our 372 strokes, 47 were nosocomial. There were no significant age and sex differences between nosocomial stroke and stroke admissions. Nosocomial stroke patients were significantly more likely than stroke admission patients to be normotensive (p = 0.001), diabetic (p = O.OI),and have cardiac disease (p = 0.03). Nosocomial stroke patients were significantly less likely to have brain hemorrhages (p = 0.001), lacunar infarcts (p = 0.03), or infarcts of undetermined cause (p = 0.047). Half of the nosocomial stroke patients died versus 11% of stroke admission patients. Nosocomial stroke differs in stroke type, associated diseases, and prognosis from stroke occurring outside the hospital. Key Words: Cerebrovascular disorders-Nosocomial stroke.

Nosocomial stroke is a relatively ignored yet frequent type of stroke. In the National Survey of Stroke, 8.7% of their strokes registered occurred during hospitalization for another, unrelated problem (1). A study that focused on the "in-hospital cerebral ischemia" found that 15% of their 171 patients had a transient ischemic attack (TIA) or stroke while hospitalized (2). We felt that a larger study was indicated by the importance of nosocomial stroke. Better understanding of this entity may lead to better care of our hospital patients. Additonally, illness that requires

From the Departments of Neurology, New York Medical College, Valhalla, NY,and 1 New York Medical College, Lincoln Hospital Campus, Bronx, NY, U.S.A. Address correspondence and reprint requests to Dr. S. J. Marks at Department of Neurology, New York Medical College, Munger Pavilion, Valhalla, NY 10595, U.S.A. 112

J STROKE CEREBROVASCDIS, VOL. 3, NO.2, 1993

hospitalization seems to be a strong risk factor for stroke. Careful scrutiny of this population may provide clues as to the mechanism of stroke in general.

Subjects and Methods We analyzed in a mixed prospective and retrospective study 372 strokes at two institutions. One-hundred sixty-six strokes were prospectively and consecutively identified at one hospital from August 31, 1988, to September 19, 1990. Twenty-one of these were nosocomial strokes (NS); the remaining 145 were stroke admissions (SA). At the second hospital, every stroke occurring between July 1, 1988, and June 30, 1990, had retrospective chart analysis. Two-hundred-six strokes occurring in 204 patients were evaluated. Six stroke charts were unavailable for evaluation. Of these 206 strokes, 26 were NS and 180 were

NOSOCOMIALSTROKE

SA. TIAs were excluded. We excluded stroke associated with cardiac bypass surgery, carotid endarterectomy, and cerebral angiography for cerebrovascular disease because of the known association of these procedures with stroke. Strokes under the age of 15 were excluded. All patients were evaluated for the presence of hypertension, diabetes, and cardiac disease. Prior cerebrovascular disease was examined but excluded because of concerns regarding the accuracy of this parameter in some of the retrospectively obtained material. All patients underwent cerebral computed tomography (CT) of the brain. Strokes were classified as large-vessel disease (low flow), large-vessel disease (local embolism), cardioembolic, lacunar, other (immune-mediated thrombosis, vasculitis, fibromuscular dysplasia, amyloid, and dissection), infarct of undetermined cause, subarachnoid hemorrhage, and intracerebral hemorrhage. Criteria for classification were based on NINDS cri-

Table 2. Stroke typeoccurrences for tile hoogroups Stroke admissions (n = 325) Large vessel (low flow) Large vessel (embolic) Cardioembolic Lacun ar Infarcts of undetermined cause Other infarcts (see text) Subarachnoid hemorrhage Intracerebral hemorrhage

Nosocomial stroke (n = 47)

7 1 19 49

(2.2%) (0.3%) (5.8%) (15.1%)

8 (17.0%) 3 (6.4%)

135 3 25 86

(41.5%) (0.9%) (7.7%) (26.5%)

21 (44 .7%) 4 (8.5%) 0(0.0%) 4 (8.5%)

4 (8.5%) 3 (6.4%)

teria (3,4). Statistical analysis was performed on the combined prospective and retrospective data.

Results Table 1.

Causes for admission of our NS patients

Cardiac Congestive heart failure Myocardial infarction Angina Arrhythmia Endocarditis Infection Pneumonia Urosepsis Meningitis Cancer Leukemia

17 6 5 3 2 1 7 4 2 1 6 3

Lung

1

Rffi~

1 1 5 2 1 1 1 3 2 1 3 2 1 1 4 1 1 1 1

Gastrointestinal Surgery Neck cancer Gastrectomy S~ngrnft

Transurethral resection of prostate Dialysis Hemodialysis Peritoneal Angiography of head and neck tumors Bleeding diathesis Aplastic anemia Thrombotic thrombocytopenic purpura Miscellaneous Trauma Postpartum Acute renal failure Anaphylaxis (bee sting)

We were able to evaluate 372 strokes, of which 47 were NS. Table 1 provides a summary of the reasons for admission of our NS patients. Nosocomial stroke represented 12.6% of all strokes evaluated (12.65% at the prospectively studied institution and 12.56% at the retrospective one). Nosocomial stroke did not show a predilection for age or gender. Nosocomial stroke patients were more likely to be normotensive (p = O.OOI), have diabetes (p = O.OI), have cardiac disease (p = O.03), and have infarcts (p = O.OOI). Comparing infarcts, those of the NS group were less likely to be infarcts of undetermined cause (p =0.047) and less likely to be lacunar infarcts (p =0.03). Halfof our NS patients died, whereas only 11% of our stroke admissions died. Table 2 shows the breakdown of stroke type for the two groups.

Discussion Nosocomial stroke is an extremely prevalent form of stroke, yet it remains relatively unstudied. Nosocomial stroke made up 12.6% of both our institutions' stroke population. Large population-based studies on stroke have reported incidences of NS of 8.7% (I) and 6.5% (5). The hospital-based Stroke Data Bank reported an incidence of 7.2% (3). These studies may under-represent the true impact of this entity. The other study that evaluated "in-hospital cerebral ischemia" similarly found that 13% (21 of 156 completed strokes) were nosocomial (2). Although the pathogenetic mechanisms at work in NS may be as diverse as the reasons for hospitaliza-

J STROKE CEREBROVASCDIS, VOL. 3, NO.2,

1993

113

S. J. MARKSIT AL.

tion, our data, in conjunction with other studies, indicate that infarction is more common than hemorrhage/ and embolic mechanisms are more likely than thrombotic ones (2/6). Of our 47 NS, only 4 were hemorrhages (all parenchymal), as contrasted with SA, in which 34.1% were hemorrhages. (The large number of hemorrhagic SA may reflect the large referral base for subarachnoid hemorrhages at one of our institutions.) This finding is similar to the other study (2) in which no cases of hemorrhagic stroke were nosocomial. Infarcts that have mild hemorrhagic changes, in our study at least, were not classified as hemorrhages. We found a higher incidence of cardiac-related stroke as have other studies (2/5/6). This would seem to reflect a multifactorial effect of cardiac disease to cause stroke and cause hospitalization, in addition to the potential of other illnesses to precipitate cardioembolic stroke. Caution must be used, however, in inferring cardioembolic mechanisms in stroke patients with cardiac disease. Patients with cardiac conditions that would allow NINDS classification as cardioembolic stroke may have stroke related to other mechanisms. The strong tendency of NS not to have hypertension conforms with the finding that NS is more likely to be nonhemorrhagic and utilize mechanisms such as embolization. The lower incidence of lacunar stroke in NS would naturally follow from this observation. The differences between NS and SA suggest that NS may be atypical of the mechanisms that cause stroke in general. Iatrogenic stroke would be a clear example of this. Our impression is that iatrogenic influences were operative in only a few of our strokes. Excluding the more obvious cardiac-related strokes and those associated with angiography, illness itself may be the risk factor here at work. Clearly, illness has powerful effects on hemodynamic, metabolic, and

114

J STROKE CEREBROVASC DIS, VOL.3, NO.2, 1993

hemostatic equilibrium. The tendency for NS to have diabetes may not so much reflect diabetic vasculopathy but the tendency for hyperglycemia, stimulated by the stress of illness, to predispose the brain to infarction (7). Stroke may be viewed mechanistically in two drastically different ways. One view would hold that a linearly progressing pathologic process would ultimately reach its endpoint, i.e.,stroke. Alternatively, a predisposing state could cause stroke after certain triggering stimuli occur. The seemingly high rate of stroke in the hospitalized population provides evidence that the latter mechanism may be at work, at least in NS in which illness may act as the trigger. Such triggers may be iatrogenic, although they more likely represent conditions found in illness that create hypercoagulable, low-flow, and metabolic states that predispose brain tissue to infarction. Better elucidation of the response of the body to illness may help to reveal the triggers of stroke that may occur in certain elements of the stroke population in general.

References 1. Weinfeld FD. The National Survey of Stroke. Stroke 1981;12(Suppl1):1-27. 2. Kelley RE, Kovacs AG. Mechanisms of in-hospital cerebral ischemia. Stroke 1986;17 :430-3. 3. FoulkesMA, WolfPA/PriceTR/MohrJP/HierDB. The stroke data bank: design, methods, and baseline char-' acteristics. Stroke 1988;19:547 -54. 4. Sacco R, Ellenberg JH, Mohr JP/ et al. Infarcts of undetermined cause : the NINCDS stroke data bank. Ann Neurol 1989;25:382-90. 5. Schoenberg BS,Schoenberg DG, Whisnant}p. Stroke during hospitalization: a population study. Neurology 1983;33(SuppI2):92. 6. Hart R, Hindman B. Mechanisms of perioperative cerebral infarction. Stroke 1982;13 :766-73. 7. Helgason CM. Blood glucose and stroke. Curr Concepts Cerebrovasc Dis Stroke 1988;23:1-6.