Computed Tomographic Brain Scans and Antiplatelet Therapy After Stroke
A Study of the Quality of Care in Dutch Hospitals
Background and Purpose We sought to develop a measure (“quality weight”) that indicates the severity of a deviation from optimal care with respect to secondary prevention with antiplatelet treatment after stroke. We also sought to estimate the effects that efforts to improve the quality of secondary prevention may have on health outcome and healthcare costs in the Netherlands.
Methods First, we developed quality weights with decision analysis techniques. These quality weights express the excess risk of vascular events in the first 2 years after stroke compared with the optimal strategy (CT brain scan in all patients and aspirin in case of cerebral infarction). Second, these weights were applied in a follow-up study of 738 stroke patients older than 45 years. The number of stroke patients admitted to a hospital in 1991 in the Netherlands was used to estimate nationwide effects. We used data from 23 neurological departments and from the Information Center for Health Care in the Netherlands.
Results The 2-year excess risk of fatal and nonfatal vascular events caused by omitting a CT brain scan and giving aspirin to all patients is rather small (on average, 0.6%). The 2-year excess risk caused by not giving aspirin to a patient with cerebral infarction is much higher (4.1%). The follow-up study indicated that only 6% of the admitted patients had not been evaluated with a CT brain scan and that 14% of the patients with cerebral infarction proven by CT scan did not get antiplatelet treatment at discharge. Efforts to improve the quality of secondary prevention after stroke may prevent 74 vascular events annually in the Netherlands at an expense of 6200 Dutch guilders per prevented event (1 Dutch guilder=0.53 US dollar, 1991).
Conclusions Efforts to improve the quality of secondary prevention with antiplatelet treatment might reduce the number of new vascular events within the first 2 years after stroke by approximately 3%. The total costs related to the extra diagnostic and therapeutic activities are approximately 0.2% of the total annual hospital costs for acute stroke patients in the Netherlands (250 million Dutch guilders).
Medical care for patients with stroke varies considerably in the Netherlands as well as in other countries. Variations can be observed with respect to hospital admission, diagnostic workup, and secondary prevention of new vascular events.1 2 3 A Dutch consensus conference on stroke in 1991 concluded that CT of the brain plays an important role in the assessment of stroke patients.4 It is used to confirm the diagnosis of stroke and to differentiate between cerebral infarction, intracerebral hemorrhage, and “nonstroke lesions.” The exclusion of hemorrhage is mandatory in patients who use anticoagulants. CT can also help in determining the underlying infarct mechanism, which may be important for treatment and prognosis.5 6
We investigated in this study how well these recommendations are put into practice in hospitals in the Netherlands. First, we developed a quantitative measure (“quality weight”) that expresses the severity of a deviation from optimal care on the prognosis of the stroke patient. These quality weights are based on published evidence on the prognosis of stroke and the effectiveness of antiplatelet treatment after stroke. Second, we estimated the improvement in the 2-year outcome after stroke and the costs related to extra diagnostic and therapeutic activities on a nationwide basis for the Netherlands if a CT brain scan is performed in all stroke patients who are admitted to the hospital.
Subjects and Methods
Fig 1⇓ is a schematic representation of the role of CT scanning in the treatment of patients with a stroke. It represents the choice between (A) treatment with daily aspirin, (B) a CT brain scan and aspirin only if no hemorrhage has been demonstrated, and (C) no aspirin (or any other antiplatelet drug). To assess the effects of these three strategies on the prognosis of stroke patients, we estimated the risk of fatal and nonfatal vascular events (ie, stroke, myocardial infarction, or vascular death). We limited the time horizon of this approach to 2 years because the preventive effect of aspirin on the occurrence of vascular events may be much smaller in the third year and even absent in the fourth.7
We considered stroke patients who were discharged alive from the hospital. Only two possible underlying conditions were taken into account: infarction and hemorrhage. This simplification is justified in our view because the proportion of patients with nonstroke lesions (eg, cerebral tumor, subdural hematoma, metabolic disturbances) is small (1.5%) if the clinical history indicates a sudden onset of the deficit.8 9 10 11 12 The direct therapeutic consequences of identifying nonstroke lesions are also limited: in a prospective population-based study, CT identified five nonstroke lesions, of which four were untreatable.11
The proportion of stroke patients with hemorrhage in Western Europe is approximately 15%.13 A patient’s clinical profile may help to differentiate hemorrhage from infarction.8 10 14 15 16 CT scans are highly accurate in making this distinction.5
The 2-year risk of fatal or nonfatal vascular events after infarction was estimated to be 23.0%. This risk estimate was based on the observed long-term survival in the Oxfordshire Community Stroke Project17 and a case-fatality rate for recurrent vascular events of 46% estimated from an overview of randomized trials of antiplatelet therapy after prior stroke or transient ischemic attack.7 There is little evidence on the risk of vascular events after hemorrhage.18 Therefore, we assumed that the 2-year risk of 23.0% also applies for patients with hemorrhage.19
A daily dose of aspirin after infarction reduces the odds of vascular events by approximately 22%.7 The estimated 2-year risk of vascular events in a patient with infarction who uses aspirin is then 18.9%. No studies have reported on the effects of aspirin after hemorrhage. Therefore, we arbitrarily assumed that the harmful effects of aspirin after hemorrhage are as large as its beneficial effects after infarction (in other words, an increase in the odds of suffering a vascular event of 22%). This implies that the 2-year risk of a vascular event after hemorrhage in a patient who uses aspirin is 26.7%.
The 2-year risk of fatal and nonfatal vascular events for a patient with an average risk of hemorrhagic stroke of 15% could then be estimated to be 20.1% with aspirin, 19.5% with CT scanning followed by aspirin if appropriate, and 23.0% without aspirin. Thus, the 2-year excess risk compared with CT scanning is 0.6% if aspirin is given to all patients and 3.5% if no aspirin is given at all (Fig 1⇑). These 2-year excess risks of fatal and nonfatal vascular events were used as quality weights.
A total number of 738 stroke patients older than 45 years consecutively admitted to the neurological departments of 23 Dutch hospitals (3 university and 20 general hospitals) from mid-1991 to mid-1992 were included in a study on the quality of hospital care for stroke patients. The 23 hospitals were selected from an urban region in the western part (n=17) and a more rural region in the northern part of the Netherlands (n=6). Nine hospitals have more than 400 beds, 3 of which are university hospitals. Patients were considered to have had a stroke if there was a focal neurological deficit of sudden onset with no known alternative to a vascular cause with a duration of at least 24 hours. Patients who were thought to have subarachnoid hemorrhage, transient ischemic attack, or nonstroke pathology were not included. For each stroke patient, data were collected from the medical and nursing records by trained nonmedical research assistants after the patient had been discharged from the hospital. These data relate to medical history, physical examination on admission, diagnostic procedures and their results, and medical and nonmedical therapeutic activities. Antiplatelet treatment was defined as a daily dosage of at least 38 mg of aspirin or, in case of aspirin intolerance, an equivalent dose of ticlopidine.
The actual costs of CT scanning, including capital costs, were estimated to be 277 Dutch guilders (Dfl; 1 Dfl=0.53 US dollar, 1991) in one of the participating hospitals. The annual costs of antiplatelet treatment (mostly aspirin), including the pharmacist’s fee, were estimated to be 100 Dfl.
Clinical Profile and Differential Diagnosis of Intracerebral Hemorrhage and Cerebral Infarction
The 2-year excess risk if no CT scan has been done depends on the probability that the patient has hemorrhage or infarction. We therefore developed a scoring rule to estimate the probability that a patient has hemorrhage based on the patient’s clinical profile. This was done because (1) the accuracy of the existing stroke scoring systems is poor, (2) most scoring rules cannot be scored retrospectively, and (3) the scoring rules perform poorly in populations other than those in which they were developed.20 The scoring rule was developed in all patients of the follow-up study who had a CT scan. Patients using anticoagulants were excluded because a CT brain scan is always necessary to determine whether the anticoagulant treatment should be reversed. We also excluded those patients who did not survive the first 2 days after admission (Fig 2⇓). Thus, logistic regression was performed to distinguish hemorrhage from infarction in 577 patients (Fig 2⇓). The items from a diagnostic rule developed by Panzer and coworkers16 (acute coma at stroke onset, vomiting, severe headache, marked hypertension, and new hyperglycemia) were used as potential predictors. Hypertension and hyperglycemia could be removed without a significant deterioration of the model (likelihood ratio test statistic, 0.7; 2 df). The probability of hemorrhage without any risk factor according to this model is 8%. Acute coma at onset of stroke (odds ratio [OR], 2.4; 95% confidence interval [CI], 1.3 to 4.5), vomiting (OR, 3.6; 95% CI, 2.1 to 6.3), and severe headache (OR, 3.5; 95% CI, 1.3 to 9.5) are incremental risk factors. The goodness-of-fit could not be significantly improved by adding interaction terms to the model (likelihood ratio test statistic, 3.7; 3 df). The discriminatory power of the scoring rule is moderate. The area under the receiver operating characteristic curve is 0.66 (95% CI, 0.60 to 0.72), which indicates that the probability of correctly identifying which of two patients—one with hemorrhage and one with infarction—has hemorrhage is two thirds. An area of 0.50 indicates a score of no discriminatory power. This scoring rule was developed only in the patients who had undergone a CT scan (577 of the 616 patients). However, we could not demonstrate significant differences with respect to the three risk factors for hemorrhage between patients with and without a CT scan, and therefore it is unlikely that patient selection for the CT scan will be a major source of bias.
Of the 738 patients who were included in the study, 616 did not use anticoagulants and survived the first 2 days after hospital admission (Fig 2⇑). A CT brain scan was performed in 577 (94%) of these 616 patients. The use of CT scanning was significantly higher in university and large general hospitals, in younger patients, in patients with coronary heart disease, and in patients without diabetes mellitus (Table⇓). Of the patients who had a CT scan, 59 patients with infarction and 4 patients with hemorrhage were discharged with a therapeutic regimen that was discordant with the results of the CT scan. Of the 39 patients who survived the first 2 days after admission and who were not evaluated by CT, 23 (59%) died before discharge. Nine (56%) of the 16 who were discharged alive received antiplatelet treatment, 2 (13%) received oral anticoagulant treatment, and 5 (31%) were discharged without any of these treatments.
How Severe Are These Observed Deviations From Optimal Care?
Given the observed care with respect to secondary prevention with antiplatelet treatment and the risk estimates presented in Fig 1⇑, we expect new fatal and nonfatal vascular events within the first 2 years after stroke in 98 (19.9%) of the 492 patients who were discharged alive.
If a patient with infarction does not receive antiplatelet treatment, the 2-year excess risk of fatal and nonfatal vascular events is estimated to be 4.1% (23.0%−18.9%; see “Subjects and Methods”). If a patient with hemorrhage receives antiplatelet treatment, the 2-year excess risk is assumed to be 3.7% (26.7%−23.0%). If these estimates are applied to the 59 patients with infarction who did not get antiplatelet treatment, one may expect 2.4 (4.1%×59) extra vascular events in the first 2 years after discharge. The expected number of extra vascular events in the two patients with hemorrhage who received antiplatelet treatment is 0.07 (3.7%×2). In this context we do not consider those patients with either infarction or hemorrhage who were discharged with oral anticoagulants. In the majority of these cases the reason for anticoagulant use was the occurrence of deep vein thrombosis, cardiac valvular disease and arrhythmias, or a progressive ischemic stroke.
The 2-year excess risk without CT scanning depends on the probability of hemorrhage. None of the 9 patients who were not evaluated with CT scanning and discharged alive on antiplatelet treatment had any incremental risk factor for hemorrhage (acute coma at onset, vomiting, or severe headache). The estimated probability that these patients have hemorrhage is therefore 8% (see scoring rule described in “Subjects and Methods”). The 2-year excess risk of vascular events in these patients was estimated with the decision tree (represented in Fig 1⇑) to be 0.3%, and the expected number of extra vascular events in this group is 0.03 (0.3%×9). We estimated with the scoring rule that the probability of hemorrhage of the patients who were not evaluated with CT scanning and who did not receive antiplatelet treatment is 8% for the 2 patients without any risk factor for hemorrhage and 17% for the 3 patients who became comatose immediately at stroke onset. The 2-year excess risk of vascular event in these patients was estimated with the decision tree to be 3.7% and 3.4%, respectively. The expected number of extra vascular events is 0.2 (3.7%×2+3.4%×3).
In summary, the expected number of extra vascular events that can be attributed to deviations from optimal care in all 738 patients within 2 years after stroke is 2.7 (2.4+0.07+0.03+0.2), of which approximately 90% (2.4/2.7) originates from patients with infarction proven by CT who did not receive antiplatelet treatment. In relative terms, it can be concluded that efforts to improve the quality of secondary prevention with antiplatelet treatment might reduce the number of new vascular events within the first 2 years after stroke by approximately 3% (2.7/98).
What Effects Do Efforts to Improve the Quality of Secondary Prevention Have on Health Outcome and Healthcare Costs?
In the Netherlands 20 134 patients were admitted to the hospital with intracerebral hemorrhage or cerebral infarction in 1991 (International Classification of Diseases codes 431, 432, 433, 434, 436, and 437). On the basis of this number, we estimated that the total number of extra vascular events that could be prevented annually in the Netherlands if a CT scan was performed in all admitted stroke patients and aspirin was prescribed in all patients with infarction would be 74 (20 134/738×2.7). Following the same methodology, we estimated that this requires that 1064 extra CT brain scans are performed and that antiplatelet treatment is prescribed in 1653 more patients at the time of hospital discharge. The extra cost related to these diagnostic and therapeutic activities is approximately 460 000 Dfl (1064×277 Dfl+1653×100 Dfl) annually or 6200 Dfl per prevented event.
The main findings of this study are that only 6% of the stroke patients who were admitted to the hospital had not been evaluated with a CT brain scan, whereas 14% of the stroke patients with infarction proven by CT scan had not received antiplatelet treatment at discharge. This study suggests that an extra 74 vascular events could be prevented annually in the Netherlands and that 90% of these events could be prevented by giving aspirin to patients with infarction that had been proven by CT. Efforts to improve the quality of secondary prevention with antiplatelet treatment would reduce the number of new vascular events within the first 2 years after stroke by approximately 3%. This requires an extra expense of 460 000 Dfl annually in the Netherlands or an increase of 0.2% in total annual hospital costs for stroke patients, which have been estimated to be approximately 250 million Dfl in the Netherlands (J.H.P. van der M. et al, unpublished data, 1995).
These estimates may serve as rough measures of the severity of suboptimal care. One of the main obstacles in the calculations of these estimates is that there is little empirical evidence about the risk of new vascular events after hemorrhage and about the risks of aspirin use after hemorrhage.18 We do not believe that clinical studies on the risks of aspirin use after hemorrhage will be forthcoming. It is likely that these risks are considerably lower than we have assumed in this study. The overall results will be only marginally affected, however, if we assume that the harmful effects of aspirin after hemorrhage are indeed very small. We estimated that 90% of the 74 extra vascular events that might annually be prevented by improvements in the quality of secondary prevention in the Netherlands are attributable to prescribing aspirin to patients with infarction proven by CT scan. On the other hand, if we assume for the sake of argument that the harmful effects of prescribing aspirin after hemorrhage are much larger than originally assumed in this study, then the results are again hardly affected. For instance, if prescribing aspirin after hemorrhage would double the odds of recurrent vascular events (instead of the originally estimated increase in the odds of recurrent vascular events by 22%; see “Subjects and Methods”), the total number of extra vascular events that can annually be prevented in the Netherlands would be 79, which is an increase of approximately 7% compared with our original estimate of 74.
Another source of uncertainty concerns the effect of antiplatelet treatment after the earlier years. The antiplatelet trials show a strong decrease of the effect of aspirin after the first 2 years after stroke, which can be at least partly explained by noncompliance and the occurrence in later years of vascular events that have been “postponed” by antiplatelet treatment.7 For this reason we limited the time horizon of our study to 2 years.
Finally, we supposed that only very few patients do not tolerate some form of antiplatelet treatment. This is reasonable in our view because there are no significant differences between the protective effects of high (>500 mg/d) and medium (38 to 325 mg/d) aspirin doses, whereas the gastrotoxicity of medium doses is probably lower.7 Moreover, in the case of aspirin intolerance, ticlopidine, which seems to be at least as effective as aspirin, may serve as an alternative. This study does not provide information regarding why the patients who have an infarction demonstrated on CT were not started on antiplatelet treatment. The 59 patients with infarction who did not receive antiplatelet treatment were on average older and more severely disabled at discharge than the 371 patients with infarction who did receive some sort of antiplatelet or anticoagulant treatment (mean age, 76 versus 72 years; mean Barthel Index score, 9 versus 15, respectively). This indicates that the prognosis of patients without antiplatelet treatment at discharge is worse, which suggests that we have overestimated somewhat the gain that could be achieved by putting these patients on aspirin.
At this moment it is difficult to estimate the effect that secondary prevention of vascular events has on the lifetime costs of medical care after stroke.21 On the one hand, secondary prevention will prolong poststroke life expectancy, which may increase the lifetime cost after stroke. The extent of this prolongation cannot be estimated accurately because we do not know the effect of aspirin during later years.7 On the other hand, one may also expect that the prevention of new vascular events slows down the progressive dependency of stroke patients with respect to the activities of daily life and nursing care. This implies that patients who use antiplatelet treatment in the chronic phase after stroke may require fewer resources for home help and nursing home and hospital care. The net effects of these opposite trends can only be estimated when more evidence becomes available on the long-term effects that antiplatelet treatment after stroke has on survival and the utilization of healthcare resources.
The role of CT scanning in stroke is of course not limited to the decision of whether to prescribe aspirin. Intracerebral hemorrhage in patients who need or may need anticoagulants must be excluded, and there are particular forms of stroke, such as cerebellar infarction or hemorrhage, that may benefit from early CT diagnosis and immediate neurosurgery; in these cases depressed consciousness or the typical cerebellar symptomatology itself points toward the specific site of the lesion and guides diagnostic decision making.5 Another advantage of CT scanning in the early phase after stroke may be the identification of specific pathophysiological subtypes (embolism, lacunar abnormalities) according to the site and size of infarction, but the predictive value of the CT features alone in this respect is of limited importance for establishing the diagnosis.5 6 Finally, CT scanning is also useful to exclude nonstroke pathology. Although the course may not be altered in most cases, knowledge of this diagnosis may be of importance for patient management.11 All these facets of the role of CT scanning were not included in our quality weights because their impact on the prognosis of the patients is relatively small compared with the decision to prescribe antiplatelet treatment.
Our study has shown the following: (1) Decision analysis can be used to estimate a quantitative measure (“quality weight”) of the severity of a deviation from optimal care. (2) The most important element in the process of improving the quality of secondary prevention after hospital discharge in the Netherlands is prescribing antiplatelet treatment to all patients with infarction proven by CT scan and not performing more CT scans. (3) Efforts to improve the quality of care with regard to secondary prevention with antiplatelet treatment may result in a reduction in the number of new vascular events in the first 2 years after stroke by approximately 3% at the expense of an increase of 0.2% of the total hospital costs for stroke patients in the Netherlands.
This study was supported by a grant from the National Committee on Investigative Medicine of the Health Insurance Executive Board of the Netherlands (O.G.91-037). Additional support was provided by the Ministerie van Welzijn, Volksgezondheid en Cultuur (89-700), The Netherlands’ Health Research Promotion Programme (SGO), and Nederlandse Hartstichting (40.004). Dr Limburg is a clinical investigator of The Netherlands Heart Foundation.
We thank Prof C.P. Warlow, University of Edinburgh, UK, and Prof H. van Crevel, University of Amsterdam, the Netherlands, for their critical review of the manuscript.
- Received November 7, 1995.
- Revision received January 2, 1996.
- Accepted January 5, 1996.
- Copyright © 1996 by American Heart Association
Limburg M. Stroke: Aspects of Diagnosis and Management. Amsterdam, Netherlands: University of Amsterdam; 1992. Thesis.
UK-TIA Study Group. Variation in the use of angiography and carotid endarterectomy by neurologists in the UK-TIA Aspirin Trial. Br Med J. 1983;286:514-517.
Dunbabin DW, Sandercock PAG. Investigations of acute stroke: what is the most effective strategy? Postgrad Med J. 1991;67:259-270.
Antiplatelet Trialists’ Collaboration. Collaborative overview of randomised trials of antiplatelet therapy, I: prevention of death, myocardial infarction, and stroke by prolonged antiplatelet therapy in various categories of patients. Br Med J. 1994;308:81-106.
Report of the WHO Task Force on Stroke and Other Cerbrovascular Disorders. Recommendations on stroke prevention, diagnosis, and therapy. Stroke. 1989;20:1407-1431.
Allen CMC. Clinical diagnosis of the acute stroke syndrome. QJM.. 1983;52:515-523.
Britton M, Hindmarsh T, Murray V, Tydén SA. Diagnostic errors discovered by CT in patients with suspected stroke. Neurology. 1984;34:1504-1507.
Sandercock P, Molyneux A, Warlow C. Value of computed tomography in patients with stroke: Oxfordshire Community Stroke Project. Br Med J. 1985;290:193-197.
Bamford J, Sandercock P, Dennis M, Burn J, Warlow C. A prospective study of acute cerebrovascular disease in the community: the Oxfordshire Community Stroke Project 1981-1986. J Neurol Neurosurg Psychiatry.. 1990;53:16-22.
Harrison MJG. Clinical distinction of cerebral hemorrhage and cerebral infarction. Postgrad Med J. 1980;56:629-632.
Von Arbin M, Britton M, De Faire U, Helmers C, Miah K, Murray C. Accuracy of bedside diagnosis in stroke. Stroke. 1981;12:288-293.
Dennis M, Burn JPS, Sandercock P, Bamford JM, Wade DT, Warlow C. Long-term survival after first-ever stroke: the Oxfordshire Community Stroke Project. Stroke. 1993;24:796-800.
Franke CL, van Swieten JC, Algra A, van Gijn J. Prognostic factors in patients with intracerebral haematoma. J Neurol Neurosurg Psychiatry.. 1992;55:653-657.
Bergman L, van der Meulen JHP, Limburg M, Habbema JDF. Costs of medical care after first-ever stroke in the Netherlands. Stroke. 1995;26:1830-1836.