Validation of Family History in Subarachnoid Hemorrhage
Background and Purpose In 6% to 9% of patients with subarachnoid hemorrhage (SAH), familial aggregation occurs; truly familial cases carry a worse prognosis than sporadic cases and raise the question of screening. If relatives have died from SAH, the family history is often the only available clue to the diagnosis, but the sensitivity and predictive value of such a history for SAH are unknown.
Methods We contacted a next of kin for a consecutive series of patients who had died in the hospital of subarachnoid hemorrhage (n=20), intracerebral hemorrhage (n=22), or ischemic stroke (n=23) between 3 and 5 years previously, and we compared the diagnosis based on the history from this next of kin with the medical diagnosis confirmed by a CT scan.
Results The positive predictive value of the diagnosis of “probable SAH” from the history in our study sample was 0.7; when adjusted for incidence rates in the general population it was 0.6 (95% confidence interval, 0.3 to 0.8). The sensitivity of the diagnosis based on the history was 0.5 (95% confidence interval, 0.3 to 0.7); 10 of the 20 cases of SAH were not identified.
Conclusions The family history of SAH, without confirmation from medical documents, is an insufficiently accurate tool to prove or disprove the diagnosis of familial SAH.
Familial aggregation of subarachnoid hemorrhage (SAH) occurs in 6% to 9% of patients admitted with SAH.1 Because outcome in patients with familial SAH is worse than that in patients with sporadic SAH,2 screening of unaffected relatives should be considered in such cases. For many relatives who have reportedly had SAH and died, medical records are no longer available. Since in these cases the family history is the only available source of information for the diagnosis of familial SAH, it is important to know the reliability of recalling an SAH in a relative who is no longer alive. Therefore, in a consecutive series of patients who were known to have died of either SAH, intracerebral hemorrhage (ICH), or ischemic stroke, we used the history provided by a next of kin 3 or more years after the fatal event to assess the positive predictive value and sensitivity of a family history of SAH.
Subjects and Methods
Medical documents of all patients admitted to Utrecht University Hospital between January 1, 1988, and December 31, 1990, with diagnoses of SAH, ICH, or ischemic stroke were reviewed. Only those patients in whom the stroke had occurred at home and who had ultimately died were selected as index patients. A trained nurse, who was ignorant of the diagnosis, interviewed a next of kin (the partner or, if unavailable, a parent or a child) by telephone. This interview was preceded by a letter providing information about the study. In a semistructured manner, the relative was questioned about the illness and cause of death of the patient (including details about the onset and course of the disease) and the diagnosis as explained to them by the attending physician. All histories thus retrieved were assessed independently by two observers (J.E.C.B., G.J.E.R.) who were also ignorant of the diagnosis. In case of disagreement, the history was assessed by a third observer (J. van G.). Descriptions such as “bleeding between the membranes around the brain” or “a burst balloon in the head” were classified as SAH, a “hemorrhage in the brain” was classified as ICH, and “a clot in the vessels of the brain” was classified as ischemic stroke. Whenever the next of kin was unable to give such a description, the circumstantial criteria shown in Table 1⇓ were used to classify the episode as probable, possible, or probably not SAH. We did not classify any history as “definite SAH,” since we prefer to reserve this term for cases confirmed by CT scan or angiography. The classification based on history was then compared with the diagnosis in the medical records, which was in all cases confirmed by a CT scan. The sensitivity and the positive predictive value of the diagnosis based on family history were calculated. In addition, we compared the positive predictive value of the history in young patients (aged 50 years or less) with that in older patients (aged 51 years or more) to analyze whether the accuracy of reporting was influenced by the age of the patient.
We found 32 patients with ischemic stroke, 31 with ICH, and 30 with SAH who met our inclusion criteria. For 18 of these 93 patients, relatives could not be traced because they had no known relatives (7 patients), the partner had died in the meantime (4 patients), the address of the relative was not traceable (4 patients), or the family practitioner considered the family too upset to participate (3 patients). For 9 other patients, the relatives declined to take part in the study. Thus, 65 patients were included in the study: 20 with SAH, 22 with ICH, and 23 with cerebral infarction.
In the general population, the incidence of ICH is at least twice that of SAH,3 4 whereas in our study sample we had almost equal numbers of patients with SAH and ICH. Because the predictive value is dependent on the incidence, we doubled the number of patients with intracranial hemorrhage in each category to recalculate the positive predictive value that would be expected in the general population.
On the basis of family history, 14 patients were classified as probable SAH, 8 as possible SAH, and 43 as probably not SAH (Table 2⇓). Ten of the 14 patients with the diagnosis of probable SAH had indeed died from SAH (positive predictive value of the diagnosis of probable SAH, 0.7). In 10 of the 20 cases of SAH, the diagnosis was not retrieved from the family history (sensitivity, 0.5; 95% confidence interval [CI], 0.3 to 0.7). None of the patients with ischemic stroke were misclassified as probable or possible SAH. Thus, with our criteria for assessing the family history, only ICH and SAH could be confused. In our study sample, the number of patients with SAH equaled that of patients with ICH. In the general population, however, the incidence of ICH is at least twice that of SAH.3 4 Therefore, if our criteria are applied to the general population, 8 instead of 4 patients with ICH will be classified as probable SAH, and the predictive value of the diagnosis of probable SAH will drop to 0.6 (95% CI, 0.3 to 0.8).
There was a trend toward more accurate reporting on younger patients (aged 50 years or less) compared with older patients (51 years or more). The positive predictive value for younger patients was 0.88 (95% CI, 0.47 to 0.99), and for older patients it was 0.50 (95% CI, 0.12 to 0.88).
According to our criteria the positive predictive value of the diagnosis of probable SAH was 0.7 in our study sample, and it would be 0.6 when extrapolated to the general population. Thus, in almost one half of the patients in whom the family history, without further data, suggests that a relative has died from SAH, the actual diagnosis is ICH. In addition, in almost one half of fatal episodes of SAH, the diagnosis is missed on the basis of the relatives’ accounts.
As can be expected in a retrospective study, we had to exclude 27 of 93 patients, which could represent a bias for the study. However, in most cases the relatives could not be traced for reasons that appear to be unrelated to the disease of the patient and are therefore unlikely to have influenced the results.
Our study population consisted of patients with fatal stroke only, rather than a complete spectrum of neurological disease. Therefore, the predictive value of the negative diagnosis of probably not SAH cannot be calculated from our data. The positive predictive value we found is based on the assumption that relatives of patients dying from diseases other than stroke do not confuse these disorders with stroke in general or SAH in particular. For nonfatal episodes of stroke, the predictive value of the history may well be better because the patient’s own account is likely to be more accurate than that of a bereaved relative.
In our study, the history was always taken by the same trained nurse in a standardized manner, which may have caused overestimation of the positive predictive value of a family history of SAH because in clinical practice the information from relatives will be obtained in a less uniform and systematic fashion.
Although not statistically significant, we found a trend toward more accurate reporting on younger than on older patients. Possible explanations are that SAH occurring in younger patients has a greater emotional impact and is remembered more clearly by relatives and that SAH is a relatively more common cause of death in younger than older patients. Moreover, our criteria were in part based on the age of the patient.
The positive predictive value and sensitivity of the family history for SAH that we found are slightly worse than those found for myocardial infarction5 and epilepsy6 but similar to observations for migraine and severe headache.7 The higher values obtained for myocardial infarction and epilepsy may be explained by the greater familiarity of the lay public with epilepsy and myocardial infarction than with SAH and by the confusion caused by the similarity in clinical presentation and nature of SAH and ICH.
Thus, despite the impact one would expect a serious disease such as SAH to have on the close relatives of patients, it proves to be very difficult to recognize episodes of SAH from family histories alone.
- Received January 9, 1996.
- Revision received January 11, 1996.
- Accepted January 11, 1996.
- Copyright © 1996 by American Heart Association
Bromberg JEC, Rinkel GJE, Algra A, Greebe P, van Duyn CM, Limburg M, Hasan D, ter Berg HWM, Wijdicks EFM, van Gijn J. Subarachnoid haemorrhage in first and second degree relatives of patients with subarachnoid haemorrhage. BMJ. 1995;311:288-289.
Bromberg JEC, Rinkel GJE, Algra A, Limburg M, van Gijn J. Outcome in familial subarachnoid hemorrhage. Stroke. 1995;26:961-963.
Bamford J, Sandercock P, Dennis M, Burn J, Warlow. A prospective study of acute cerebrovascular disease in the community: the Oxfordshire Community Stroke Project 1981-86, II: incidence, case fatality rates and overall outcome at one year of cerebral infarction, primary intracerebral and subarachnoid haemorrhage. J Neurol Neurosurg Psychiatry. 1990;53:16-22.
Kee F, Tiret L, Robo JY, Nicaud V, McCrum E, Evans A, Cambien F. Reliability of reported family history of myocardial infarction. BMJ. 1993;307:1528-1530.
Ottman R, Hong S, Lipton RB. Validity of family history data on severe headache and migraine. Neurology. 1993;43:1954-1960.