Case-Fatality Rates and Functional Outcome After Subarachnoid Hemorrhage

Methods

To find population-based studies on SAH, a MEDLINE search was performed (key words: subarachnoid hemorrhage and stroke, outcome, mortality, case fatality, population, or epidemiology in different combinations). For studies thus found, reference lists were checked to find additional studies. This method of cross-checking was continued until no further studies were found. For population-based studies that reported on incidence but not on outcome, a new search with all authors of the incidence study was performed to detect any further outcome studies.

Inclusion criteria for this review were as follows: (1) publication between 1965 and 1995 and period of survey after 1960; (2) study period not longer than 10 years, unless separate results per decade are given; (3) population-based design, defined by the following criteria: the study population is representative of the population in general, all hospitals serving the study population have been involved, all death certificates have been examined or data on deaths of nonhospitalized patients have been obtained from the coroner; (4) upper limit of age for the study not below 75 years and lower limit not above 25 years; (5) for population-based studies about stroke in general, SAH should be considered as a separate entity with defined clinical criteria and a defined number of cases that include or at least allow calculation of case-fatality rates; and (6) review of at least two thirds of the records by the study investigators.

All eligible studies were independently reviewed by two authors (J.W.H. and G.J.E.R.) to extract data on case-finding methods, diagnostic criteria for SAH, number of patients, proportion of patients who underwent CT scanning, time of follow-up, number of deaths among patients with SAH, and, if available, functional outcome among patients who had survived the bleed. In case of disagreement, the article was reread and discussed until agreement was achieved. Studies that described two or more periods were considered separate studies.^{7 8 9} In one study describing two periods, the first period concerned only hospital-based data⁹ ; since the hospitals in question are designated emergency centers and since all patients, even if moribund, are immediately referred, both periods were included.

Data on functional outcome after SAH were extracted if available, and if necessary they were adjusted to fit into the modified Rankin Scale, a handicap scale that grades patients on their overall level of independence with reference to their previous lifestyle.^{10 11} If functional outcome was assessed by the Glasgow Outcome Scale (GOS),¹² this adjustment was as follows: GOS 5=Rankin 0+1; GOS 4=Rankin 2+3; GOS 3=Rankin 4; and GOS 2=Rankin 5. If no scale was used, patients considered to have “returned to their previous lifestyle” were graded Rankin 0 to 2.

Results

We selected 21 studies, describing 25 study periods between 1960 and 1992. Table 1⇓ summarizes the characteristics of these studies.

A total number of 2155 patients with SAH were included in these studies: 1068 women and 998 men (for 89 patients sex was not reported). Mean age at the time of the hemorrhage was 55 years (data derived from 15 studies). Mean age for women was 58 years; this increased from 57 years in the period 1977 to 1984 to 59 years between 1985 and 1992. Mean age for men was 53 years, with no increase over time (mean of 53 years for 1977 to 1984 and 52 years for 1985 to 1992).

Weighted linear regression analysis of all studies showed that the case-fatality rate decreased 0.8% per year (95% confidence interval [CI], 0.1 to 1.5; Table 2⇓, Figure⇓). The most recent study had an exceptionally low case-fatality rate (8%; range in other studies, 32% to 67%).³² Exclusion of this study altered the result to a decrease of 0.5% per year (95% CI, −0.1% to 1.2%). Because of this substantial difference, this study was excluded in all other analyses.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

Case-fatality rate after subarachnoid hemorrhage. Point estimates and 95% confidence intervals are shown.

Exclusion of studies with outcome assessment at an interval other than 1 month did not essentially alter the results. After adjustment for age and sex in the 12 studies in which these data were available, the crude decline in case-fatality rate (0.6% per year) increased to 0.9% per year (95% CI, −0.7% to 2.6%). The decline in case-fatality rate was not influenced by adjustment for proportion of patients that underwent a CT scan or by the examination of only coroners’ data and not all death certificates. After recalculation of the percentage of patients who died before admission to 15%, the results were essentially the same.

Table 3⇓ shows data from eight studies that provided information on functional outcome. Ten percent to 20% of all patients became disabled, with loss of independence. Although only eight of 21 studies provided data on functional outcome and the time of outcome assessment varied from 1 to 48 months, weighted linear regression of these data showed an increase in the proportion of patients who remained independent after their SAH of 1.5% per year (95% CI, 0.1% to 2.8%).

Discussion

The results of our review show a 15% decrease in case-fatality rate of SAH during the last three decades. Although the confidence intervals include zero, indicating that the results might also be consistent with no trend in case-fatality rate, this decrease is consistent through all analyses we performed and becomes stronger if data are adjusted for age and sex.

The mean age of patients included in the studies with SAH has increased over time, which negatively affects the unadjusted case-fatality rates in the later studies because outcome is less favorable in older patients.^{38 39}

Our results should be interpreted with some caution since a number of other factors may have influenced our results. First, although we used predefined criteria for methods of case finding, considerable differences in completeness of case finding remain between studies. Because patients who die at home or in a nursing home are most difficult to detect, studies with more complete case finding in this subset of patients may report higher case-fatality rates, but analysis in which we used recalculated data with a minimum of 15% dead on admission did not alter the results. Also, whether data on out-of-hospital deaths were obtained from the coroners’ office or by the examination of all death certificates had no influence. Second, unpublished data may not have been included. Since the studies in this review do not concern clinical trials or treatment evaluations with possible “negative” results, the potential issue of publication bias is not likely in this review. Third, not all studies used the same clinical criteria for the diagnosis of SAH. We assume that the majority of cases studied have been true SAH cases because the definitions show a great overlap and because most patients have been investigated by lumbar puncture or CT. However, before CT scanners became widely available, some patients with intracerebral hemorrhage were probably misdiagnosed as having SAH.¹

Since case-fatality rates for SAH and intracerebral hemorrhage are similar,⁴⁰ inclusion of some patients with intracerebral hemorrhage will not have biased the results. Fourth, we compared studies from different countries, with possible differences in management strategies and hence different case-fatality rates. In this review, however, there is a random distribution of countries studied over time. It is very unlikely that the most recent studies were all undertaken in places with “better” management strategies and therefore lower case-fatality rates. Two studies that were undertaken in the same place but at different time intervals^{8 9} both showed a decrease in case-fatality rate over time and thus supported our findings in all other analyses. Moreover, we excluded the most recent study, which had a very low case-fatality rate. Exclusion of this study resulted in an underestimation of the actual decrease in case-fatality rate, and its findings support the presence of a true decline.

A plausible explanation for the decrease in case-fatality rate over time is the improved management of patients with SAH. Management strategies that may have contributed to the improved outcome include (1) better case-finding methods that identify less severe cases; (2) more accurate diagnostic tools, such as CT and MR techniques; and (3) improved medical and surgical treatment regimens.

The substantial effects on outcome by treatments such as nimodipine in randomized trials⁴¹ are much stronger than the modest effects on the case-fatality rate in population-based samples. This discrepancy might be explained by two factors. First, new treatment strategies have predominantly been introduced in the past 5 to 10 years. The effect on case-fatality rate might become more substantial in the future. Second, the effect is diluted by extremes of the clinical spectrum: patients who will recover in any case or those who die early after the initial bleed. A considerable and stable number of patients (10% to 20%) die at home or arrive dead on admission,^{8 20 42 43} and another 12% to 30%^{7 23} never recover from the initial bleed. Thus, only a small proportion of all patients with SAH will profit from new medical or surgical treatment regimens.

Our analysis showed that the proportion of patients who remain independent has increased. This is consistent with the decrease in case-fatality rate and indicates that the smaller proportion of deaths is associated with a larger proportion of independent patients. However, many patients who survived their bleed are left severely disabled, and even patients who remain independent may experience great difficulties in resuming their previous lifestyle.⁴⁴ Evaluation of new modes of treatment should therefore focus not only on case-fatality rate but also on improvement of functional outcome and quality of life.