Will Increased Awareness Among Physicians of the Significance of Sudden Agonizing Headache Affect the Outcome of Subarachnoid Hemorrhage?
Coventry and Warwickshire Study: Audit of Subarachnoid Hemorrhage (Establishing Historical Controls), Hypothesis, Campaign Layout, and Cost Estimation
Background and Purpose The most common symptom associated with aneurysmal minor bleed (“warning leak”) is a sudden agonizing headache. Early screening of these patients may improve the outcome of subarachnoid hemorrhage and may be highly cost-effective.
Methods We conducted an extensive retrospective audit of subarachnoid hemorrhage over the last 10 years in the region of Coventry and Warwickshire, England, and initiated a continuous campaign among all physicians in the region for early neurosurgical referral of patients with sudden agonizing headache.
Results Over the last 10 years the incidence of subarachnoid hemorrhage in the region was 8.7/100 000 per year. Surgical activity was 34% and early mortality 45.2%. Functional outcome, both overall and by grade on admission, was within internationally accepted levels. Warning leak symptoms before admission were experienced by 20% of patients. These patients sought medical advice but were not referred immediately to the neurosurgical unit.
Conclusions We have established our population as valid historical controls and outlined our campaign strategy. Lowering the clinical threshold at which patients with sudden agonizing headache are screened for aneurysms or arteriovenous malformations will undoubtedly increase diagnostic costs. However, for reasons given in the text, we estimated the cost per quality-adjusted life year gained to be £1000 ($1500).
Primary SAH is a potentially devastating neurosurgical condition. Despite advances in neurosurgical management, less than 40% of patients will have a full functional recovery.1 Furthermore, since the condition afflicts relatively young patients (mean age, 45 to 65 years) and its treatment requires the use of an enormous number of resources, the actual cost of the disorder is staggering ($1 755 600 000 annually in the United States [1991 figures]2 ).
The single most important independent predictor of outcome of SAH remains the clinical state (WFNS grade3 ) on admission.4 5 6 Patients in grades I and II have consistently better outcome in terms of both mortality (12% compared with 45% in grades III and IV) and functional recovery (78% to 85% compared with 45%).7 8 It therefore follows that any measure that could deliver a larger percentage of patients with higher clinical grades to an NSU would improve the overall outcome as well as reduce the social and economic costs of the condition.
Fifty years ago it was recognized that a significant number of intracranial aneurysms are subject to minor leaks in the weeks to days preceding the major hemorrhage.9 Many studies have since shown that approximately 20% of patients admitted with a diagnosis of SAH had experienced “warning leak” symptoms in the period before admission.3 4 10 11 12 13
By far the most common symptom associated with aneurysmal minor bleed (warning leak) is a sudden agonizing headache. It is often associated with nausea, vomiting, and neck pain, and its intensity is such that medical advice is usually sought.3 14 Unfortunately, many physicians appear to be unaware of its significance. Thus, more than 90% of these patients are not referred immediately to the appropriate neurosurgical center. They are diagnosed as suffering from a variety of conditions, such as tension headache, sinusitis, influenza, temporal arteritis, or “sprained neck.”8 12 15
Since June 1995 we have started an extensive, continuous campaign in the region of Coventry and Warwickshire, England, targeting all doctors dealing with patients suffering from SAH to improve their awareness of the significance of sudden agonizing headache.
In this report we present a retrospective overview (audit) of the overall management of SAH in the region of Coventry and Warwickshire and establish this population as a valid historical control. Furthermore, we outline our campaign strategy and discuss the cost implications of our proposals.
Subjects and Methods
The region of Coventry and Warwickshire has had a relatively steady population over the last 10 years (Table 1⇓). A mixture of urban and rural communities, it is almost equally divided between the city of Coventry and the region of Warwickshire. It is served by four district general hospitals: George Elliot in Nuneaton (north), St Cross in Rugby (east), Warwick Hospital in Warwick (south), and Walsgrave in Coventry (west), where the NSU is located. Until recently the only CT scanner in the region was also located at this hospital. During the study period, all cases of SAH were treated at the NSU, although, as is common in all parts of the United Kingdom, comatose, inoperable patients were primarily treated at the referring hospitals. Diagnosis was almost always confirmed by CT scan. After approval by the Coventry and Warwickshire Ethics Committees and over a period of 6 months, we collected the case records of all patients with SAH during a 10-year period (1985 to 1995). Details of the patients were recovered from the angiography records of Walsgrave Hospital’s radiology department, the NSU’s own archives, the audit departments, and the postmortem reports of the pathology departments of all four hospitals serving the region. Patient notes were retrieved, and all relevant information was entered into a computerized database. When necessary, the general practitioner responsible for the patient was contacted to provide further information. The WFNS system of classification was used for preoperative and postoperative clinical grading. The clinical presentation, described in the notes by the treating physicians, was used for verifying the grade on both admission and discharge. The patient’s functional outcome was graded according to the GOS16 in the same fashion.
To establish this population as valid historical controls, we compared our results with those reported in other studies on SAH. Admission grades, mortality, and overall functional outcome were compared with the population-based studies of Broderick et al6 and Fogelholm et al (Central Finland Study).17 Patient outcome by grade on admission was compared with that reported in the study of Kassell et al (Cooperative Study).7
In the Cooperative Study the level of consciousness was used for preoperative classification. We arbitrarily compared “alert” with grades I and II, “drowsy” with grade III, “stuporous” with grade IV, and “comatose” with grade V. The χ2 test as a measure of goodness of fit18 was used at every grade to verify that the observed value differences were only due to chance and therefore not statistically significant, ie, to prove that our outcome results were comparable to those in the Cooperative Study. The χ2 test was used to compare functional outcome with that of the Central Finland Study.17
An information sheet and referral guidelines were sent to all general practitioners and accident and emergency physicians of the region. The importance of sudden agonizing headache was stressed, and they were actively encouraged to refer any patients with a suspicious headache to the NSU. Regional meetings were organized, and teaching sessions for junior physicians were supplemented with relevant information. Regular updates of the campaign progress and possible results are planned to maintain the appropriate momentum.
Every referred patient is admitted immediately and undergoes CT scan. If this is negative, we perform a lumbar puncture, and the cerebrospinal fluid is examined spectrophotometrically for xanthochromia.19 If the results are still negative but clinical suspicion is high, we proceed to cerebral angiography. Unless any of these tests returns positive, the patient is discharged home in less than 48 hours. We have established a telephone hotline for direct access to the on-call neurosurgical registrar to provide better guidance to the referring physicians. The radiology department in our hospital will be equipped with a new MR scanner in the near future and will be able to provide us with MR angiography. We anticipate that the screening of patients with severe headache will then be simplified because we should be able to scan and discharge them immediately if the standard MR and MR angiography responses are negative.
Using the General Practice Research Data Base at the Office of Population Census Surveys, we estimated the current use of health services by patients with headache-related problems and attempted to calculate the potential impact of our campaign.
Six hundred forty-nine patients with a diagnosis of SAH were identified in the period from January 1985 to January 1995. The calculated incidence of SAH in our region was 8.65/100 000 per year. During this period 296 patients were admitted to the NSU with a diagnosis of SAH. Physicians followed an additional 192 patients (after a diagnostic CT scan), 99% of whom were considered inoperable at the time. Of 14 767 postmortem examinations performed in the region during the same period, 124 cases of fatal SAH occurring in the community were identified. They represent 0.8% of all deaths and 19% of all cases of SAH (Fig 1⇓). We were unable to retrieve the notes of 37 patients. The age of the patients ranged from 12 to 90 years (mean, 51 years), with 283 males and 366 females.
Of those patients admitted to the NSU, 220 (74%) were angiographically demonstrated to have intracranial aneurysms, and 208 underwent aneurysm surgery. Therefore, surgical activity for the 10-year period was 34% (208/612). The timing of surgery varied between 1 and 270 days (median, 7 days). Patients with arteriovenous malformations were routinely scheduled for elective excision ranging from a few weeks to many months after the initial hemorrhage. Thirty-two patients had a negative CT scan, with an abnormal lumbar puncture on admission. Seven of these had positive angiograms and underwent craniotomy and clipping of their aneurysms. We admitted 214 patients in grades I and II (alert), 74 in grades III and IV (drowsy, confused), and 7 in grade V (comatose) (Table 2⇓). On discharge, 67% of all patients admitted to the NSU made a full recovery (GOS score of 1), while 35 (12%) died (Table 3⇓). Our overall in-hospital mortality over the study period was 45.2% (186 of the patients treated by the physicians and 35 of the patients admitted to the NSU). The total number of patients who died from SAH in the entire population during the 10-year period was 345 (case-fatality rate, 53%).
Of those admitted to the NSU, 100 patients (33%) reported one or more episodes of headache in the weeks to days before admission. Eighty-three patients (28%) eventually visited their general practitioners or their accident and emergency departments, but only a third (25) were referred to the NSU. Interestingly, not one of the patients who was seen at an accident and emergency department was referred. Thus, 58 patients (20% of those admitted to the NSU) were not referred the first time they presented to the medical profession, when they were, by definition, grade I. When these patients were finally admitted, their clinical state was significantly worse: 32 (55%) were in grades II or worse (Table 4⇓). On discharge, 13 (22%) of them had mild to severe neurological deficits, 1 remained in a vegetative state, and 3 died (Table 5⇓).
When we compared our outcome results with those in the Cooperative Study9 (Fig 2⇓), χ2 values were not statistically significant (P>>0.05) only when we omitted deaths. At all grades our mortality was moderately but significantly less than that reported in the Cooperative Study. Furthermore, our overall functional outcome did not differ significantly from that reported in the Central Finland Study17 (Table 3⇑).
SAH management has improved considerably over the last 40 years. Both microsurgery and perioperative treatment have undergone significant changes. However, the overall outcome of the condition remains unchanged,1 with the clinical grade of the patient on admission the most important prognostic factor.
Although many investigators have highlighted the importance of early neurosurgical referral of patients with sudden agonizing headache,4 5 10 11 12 13 concerns about cost and availability of resources have hindered any progress in this direction.20 This can have grave consequences. In the prospective Danish Aneurysm Study,11 the presence of a premonitory leak was not associated with overall outcome, but these patients fared far worse than those admitted in WFNS grade I. In a retrospective study Leblanc5 showed that of patients with a minor leak before the main rupture, only 40% were categorized with more favorable clinical grades (I, II) on admission, while 89% of those with one bleeding episode were categorized with more favorable grades on admission. These findings were also confirmed in the Coventry and Warwickshire Study, with more than half of the patients who reported a premonitory headache arriving at the hospital in grade II or worse. The sad paradox of SAH is obvious: those patients who are most treatable are least likely to be diagnosed correctly and vice versa.
This report presents the results of an extensive audit of SAH in the region of Coventry and Warwickshire. We recognize the fact that in a retrospective study, the possibility of bias is significant. Since we are planning to use similar methods of data collection in the future to assess the effectiveness of our campaign, any inherent bias in the study design should cancel itself out and not affect our results. It is also possible that as more CT scanners are installed in the peripheral hospitals, diagnosis of SAH will become more accurate and consequently a smaller number of traumatic taps will be interpreted as SAH. The outcome of the future population then would be artificially worse: patients with traumatic taps but negative CT are more likely to have a better outcome. On that basis, we have deliberately omitted from our database all cases treated in the peripheral hospitals and not confirmed by CT scan.
Any study of SAH, either prospective or retrospective, is likely to suffer from bias for another very important reason: these studies are almost always conducted by neurosurgeons/neurologists and are therefore based on patients with SAH referred to NSUs. There are few true population-based studies in which it is reasonable to assume that all patients with SAH are entered. The study reported by Pakarinen21 is a good example. It is therefore likely that most studies based on data from NSUs will be biased, since by and large patients categorized in better grades are referred to such units. This is certainly true in Great Britain, where most of the NSUs do not admit comatose patients after SAH. Where this study breaks new ground in Britain is in our attempt to study the community morbidity and mortality of the condition. This is why we have gathered postmortem data and data from our physician colleagues to create as comprehensive a view as possible. It is only with such a community-based database that we can hope to detect an overall change in outcome as a result of our campaign.
To establish the Coventry and Warwickshire population as valid historical controls, we compared our results with those of population- and referral-based studies on SAH. Worldwide, the reported incidence of SAH varies from 6/100 000 to 16/100 000.21 22 If we take into consideration that not all sudden deaths were followed by a postmortem examination and that at least 30 patients during this period were not referred to Walsgrave Hospital, then the incidence of SAH in our region was approximately 9/100 000 per year. Our outcome results were comparable to those reported in the Cooperative Study only when we omitted deaths. Although there has been considerable criticism of the selection bias inherent in that study design,6 we used only the outcome results by grade on admission to compare with our in-hospital management of patients with SAH. Population-based studies6 17 report early mortality of 45%, with surgical activity of approximately 50%. The early mortality of SAH in our region was 45.2%, with surgical activity of 34%. This, along with the fact that patients in coma tended to be treated conservatively at the referring hospitals, may explain the increased overall mortality (case-fatality rate, 53%) in our region. More aggressive operating policies have produced only marginal improvement in overall outcome, although better quality of life has been reported.17 Clinical grades on admission were, however, broadly comparable (43% of patients alert on admission in the Coventry and Warwickshire study compared with 45% in the Central Finland Study), and overall functional outcomes were not significantly different (Table 3⇑). Finally, we confirmed that approximately 20% of patients admitted to the NSU had experienced warning leak headaches before the actual hemorrhage and that they had not been referred early enough. Their outcome was eventually worse than that of patients in WFNS grade I, although it did not differ significantly from the overall outcome of patients with SAH.
International experience with the management of SAH indicates that the overall outcome of the condition would improve if more patients were identified after a minor bleed or warning leak. By encouraging physicians to immediately refer all patients with sudden agonizing headache, we believe that a reduction in both morbidity and mortality of SAH will be achieved in our region. In our cost-conscious era, however, such an approach should be not only effective but economically sensible as well. We were able to determine that 1.2% of the population per year visit their general practitioners complaining of headache. Repeated visits to the physician are also included, and therefore this figure may represent an overestimation of the actual number of patients suffering from headache. Using these data, we estimated a yearly attendance in our region of approximately 10 000. We currently investigate 100 patients per year with CT, scanning for possible SAH (internal audit data of NSU). That translates into apparently one referral for every 100 patients with headache.
If we assume a threefold increase in referrals due to our campaign, we could expect 200 extra investigations for SAH per year. Until now we have been admitting approximately 30 patients per year with proven SAH. We can expect 20% of them, or 6 patients per year, to have a premonitory leak resulting in earlier referral. From our audit we know that 55% of these patients arrive in a worse clinical state when they suffer a subsequent major bleed; therefore, as a result of our campaign, 4 patients per year should be admitted in a better clinical state, resulting in a yield ratio of 2:100 extra screened patients. However, at a mean age of 45 years, with a probability for good recovery of 85%, overall probability of death (if left untreated) of 90%, and screening cost per patient of £400, the estimated quality-adjusted life year cost is £1025 ($1537). This compares very favorably with established screening programs such as breast cancer screening (cost per quality-adjusted life year, £3300 to £3400) ($4950 to $5100)23 or the cost of treating mild hypertension or single-vessel coronary artery disease.24 In calculating this cost we considered only those patients expected to be admitted in better clinical condition. If the campaign is effective, we hope to see more patients admitted with a diagnosis of SAH and therefore better overall outcome of the condition in the region (reduced overall mortality as well as morbidity). Consequently, the cost of screening will be reduced further. Furthermore, the planned introduction of MR angiography will greatly reduce the cost of screening because there will be no need for admission and lumbar puncture of those patients with negative scans.
This argument must still be proven in practice.
Selected Abbreviations and Acronyms
|GOS||=||Glasgow Outcome Scale|
|WFNS||=||World Federation of Neurological Surgeons|
The authors are grateful to the Audit and Pathology Departments of all four referring hospitals in the region of Coventry and Warwickshire for allowing them to retrieve patient records and providing discharge data.
Reprint requests to Christos M. Tolias, Department of Neurosurgery, Walsgrave Hospital, Clifford Bridge Rd, Coventry CV2 2DX, UK.
Presented in part at the meeting of the British Neurosurgical Research Group, Newcastle, UK, April 20-21, 1995.
- Received October 17, 1995.
- Revision received February 6, 1996.
- Accepted February 8, 1996.
- Copyright © 1996 by American Heart Association
Batjer HH. Intracranial aneurysm. In: Renganchari S, Wilkins R, eds. Principals of Neurosurgery. Singapore: Wolf Publishing; 1994:11.2-11.4.
Wiebers DO, Torner JC, Meissner I. Impact of unruptured intracranial aneurysms on public health in the United States. Stroke. 1992;23:1416-1419.
Broderick JP, Thomas GB, Dudler JE, Tomsick T, Leach A. Initial and recurrent bleeding are the major causes of death following subarachnoid hemorrhage. Stroke. 1994;25:1342-1347.
Rosenorn J, Eskensen V, Schmidt K, Espersen JO, Haase J, Harmsen A, Hein O, Knudsen V, Midholm S, Markunssen E, Rasmussen P, Ronde F, Voldby B, Hansen L. Clinical features and outcome in 1076 patients with ruptured intracranial aneurysms: a prospective consecutive study. Br J Neurosurg.. 1987;1:33-46.
Richardson JC, Hyland HH. Intracranial aneurysms: a clinical and pathological study of subarachnoid and intracranial hemorrhage caused by berry aneurysms. Medicine (Baltimore).. 1941;20:1-83.
Juvela S. Minor leak before rupture of an intracranial aneurysm and subarachnoid haemorrhage of unknown aetiology. Neurosurgery. 1991;30:7-11.
Ostergaard JR. Warning leak in subarachnoid haemorrhage. BMJ.. 1990;301:190-191.
Fogelholm R, Hernesniemi J, Vapalahti M. Impact of early surgery on outcome after aneurysmal subarachnoid hemorrhage. Stroke. 1993;24:1649-1654.
Armitage P, Berry G. Statistical Methods in Medical Research. 3rd ed. Oxford, England: Blackwell Scientific Publications, Inc; 1994:394-395.
Vermuelen M, Hasan D, Blijenberg BG, Hijdra A, Van Gijn A. Xanthochromia after subarachnoid haemorrhage needs no revisitation. J Neurol Neurosurg Psychiatry.. 1989;52:826-828.
Majubdar B, Partridge J. Warning leak in subarachnoid haemorrhage. BMJ.. 1990;301:390. Letter.
Pakarinen S. Incidence, aetiology and prognosis of primary subarachnoid haemorrhage: a study based on 589 cases diagnosed in a defined urban population during a defined period. Acta Neurol Scand. 1967;43(suppl 29):1-128.
Locksley HB. Natural History of Subarachnoid Haemorrhage, Intracranial Aneurysm and Arteriovenous Malformations: Based on 6368 Cases in the Co-operative Study. In: Sahs AL, Perret GE, Locksley HB, eds. Intracranial Aneurysms and Subarachnoid Hemorrhage. Philadelphia, Pa: JB Lippincott; 1969:37-108.
Cambell DI, Royle GT. Breast cancer screening programs: are they worthwhile? Recent Adv Surg.. 1992;15:1-22.