Carandang R, Seshadri S, Beiser A, Kelly-Hayes M, Kase CS, Kannel WB, Wolf PA. Trends in Incidence, Lifetime Risk, Severity, and 30-Day Mortality of Stroke Over the Past 50 Years. JAMA. 2006;296(24):2939-2946. doi:10.1001/jama.296.24.2939
Author Affiliations: Department of Neurology, School of Medicine (Drs Carandang, Seshadri, Kelly-Hayes, Kase, and Wolf) and Department of Biostatistics, School of Public Health (Dr Beiser), Boston University, Boston, Mass; and the National Heart, Lung, and Blood Institute's Framingham Heart Study (Drs Seshadri, Kelly-Hayes, Kase, Kannel, and Wolf), Framingham, Mass.
Context Prior estimates of long-term trends in the incidence and severity of stroke have varied; trends in lifetime risk have not been reported.
Objective To determine long-term trends in the incidence, lifetime risk, severity, and 30-day mortality of clinical stroke.
Design, Setting, and Participants Prospective evaluation of the community-based Framingham Study original and offspring cohorts. Participants were 9152 men and women free of prevalent stroke and undergoing follow-up for up to 50 years over 3 consecutive periods (1950-1977, 1978-1989, and 1990-2004), with biennial ascertainment of stroke risk factor data and active surveillance for incident clinical stroke and cause-specific mortality.
Main Outcome Measures Incidence (age-adjusted, sex-specific), severity, 30-day mortality, and mortality-adjusted 10-year and lifetime risk of stroke in each of the specified periods.
Results There were 1030 incident clinical strokes (450 [44%] in men, 629 atherothrombotic brain infarctions [61%]) in 9152 persons 55 years or older over 174 917 person-years of follow-up. The age-adjusted incidence of first stroke per 1000 person-years in each of the 3 periods was 7.6, 6.2, and 5.3, respectively, in men (P = .02 for trend) and 6.2, 5.8, and 5.1 in women (P = .01 for trend). The lifetime risk at age 65 years decreased from 19.5% to 14.5% in men (P = .11) and from 18.0% to 16.1% in women (P = .61). Age-adjusted stroke severity did not vary across periods; however, 30-day mortality decreased significantly in men (from 23% to 14%; P = .01) but not significantly in women (from 21% to 20%; P = .32).
Conclusions In this cohort of men and women free of prevalent clinical stroke at initial examination, incidence of stroke has decreased over the past 50 years but the lifetime risk has not declined to the same degree, perhaps due to improved life expectancy. The results of this study suggest that improved control of risk factors has lowered stroke incidence but emphasize the need for continued primary prevention efforts.
Stroke continues to be a major public health concern, with more than 750 000 incident strokes occurring each year in the United States.1- 3 It is the third leading cause of death behind heart disease and cancer and the leading neurologic cause of long-term disability.4 More than half of all stroke survivors are left dependent on others for everyday activities.5 The projected increase in the elderly population with improvements in life expectancy has and is expected to further increase stroke prevalence and the overall population burden of stroke. On the other hand, there have been countervailing major developments in the identification and treatment of stroke risk factors and the management of acute stroke.6- 15
Studying the temporal trends can provide insights about the effect of these varying influences on the incidence, lifetime risk, severity, and case fatality of stroke. These data can help guide health programs, public policy, and the allocation of research funding. The Framingham Study has gathered more than 50 years of data on incident stroke, using standardized definitions and methods of data collection. In addition, it has verified survival data, permitting computation of the lifetime risk16 of stroke over the same period.
The Framingham Study17 consists of the initial cohort recruited in 1948 (5209 persons, aged 28-62 years) and the offspring and spouses of offspring (5124 persons, aged 5-70 years) who were enrolled in 1971. The cohorts are predominantly of European descent, well-educated, and middle-class. Race/ethnicity was determined both by self-report and by the investigators. There were 25 persons from the original cohort and 7 persons from the offspring cohort who were excluded because of prevalent stroke at their initial examination. Participants underwent follow-up until the last biennial examination attended or up to December 2004 (whichever was earlier) and were censored at first stroke or death.
Written informed consent was obtained from all participants at each examination. The consent form and the study design, as outlined below, were approved by the institutional review board of Boston Medical Center.
Clinical stroke was defined as rapidly developing signs of focal neurologic disturbance of presumed vascular etiology lasting more than 24 hours. This definition excludes transient ischemic attacks and silent cerebral infarctions or hemorrhages detected only by imaging. Individual stroke subtypes were categorized based on preestablished diagnostic criteria that include clinical features, imaging studies and other laboratory criteria, noninvasive vascular studies, cardiac evaluations for a source of embolus, and, when available, information from autopsy studies. Ischemic stroke was diagnosed if a focal neurologic deficit was documented and if imaging showed no hemorrhage, if imaging showed an ischemic infarction that correlated with the clinical deficit, or if an ischemic infarction was documented at autopsy. Ischemic strokes were further classified as atherothrombotic brain infarctions (ABIs) if no cardiac source of embolus was found; this category included large-artery infarctions, lacunar infarctions, and infarctions of unknown origin.
Stroke severity was categorized based on the observed neurologic deficits at the initial neurologic examination. Deficits were classified into 4 categories: none (no deficit/impairment), mild (a deficit was present in visual, communication, motor, and/or sensory realms, with impairment insufficient to affect functional independence), moderate (deficit requiring assistance in any one of the domains mentioned above), and severe (the patient was functionally dependent on others in 2 or more domains). Case fatality was defined as the proportion of patients with stroke who died within 30 days of the onset of the stroke. These definitions have remained unchanged since the inception of the cohort.
All persons who developed incident clinical stroke between 1948 and 2004 were identified as part of ongoing clinic and hospital surveillance. Methods of data collection include medical record review; laboratory testing; imaging; and collaboration with general practitioners, emergency departments, and imaging facilities in the area. If a participant saw a physician or was admitted to the hospital, visited an emergency department, or underwent any brain imaging between biennial examinations for symptoms that are suggestive of transient ischemic attack or stroke, a stroke neurologist from the Framingham Study visited the person within 48 hours. Standardized data forms, including those for a complete history and neurologic examination, Mini-Mental Status Examination, and functional assessment, were filled out at the time of the initial evaluation and at 1, 3, and 6 months. All records and imaging results were obtained for review.
A panel of 3 investigators (at least 2 neurologists) adjudicated the diagnosis of stroke in each case, based on review of all relevant medical results and on the assessment of the study neurologist who had examined the participant. An independent tracking system is in place to gather information on participants regardless of whether they still reside in the vicinity of Framingham, Mass. The study has benefited from excellent continuity among its assessing stroke neurologists, and this has contributed to the long-term stability of the ascertainment and adjudication process.
Stroke incidence was analyzed by age and sex over 3 different periods (1950-1977, 1978-1989, 1990-2004). Cutpoints of 1978 and 1990 were selected to coincide with the widespread use of computed tomography (CT) and magnetic resonance imaging (MRI), respectively, in this population, since past studies18- 20 have suggested that the introduction of a new imaging modality can affect the recorded incidence of stroke. We included all clinical strokes in participants aged 55 to 94 years, unless otherwise noted. We present, separately for men and women, annual incidence of stroke by period, standardized to the overall age distribution within sex. We used log-linear Poisson regression to estimate age-adjusted rate ratios for stroke incidence in the 3 periods, using the first period as the referent. We performed age-adjusted, sex-specific logistic regression to compare the periods with respect to severity and 30-day mortality, again using the first period as the referent.
We compared the distribution of stroke risk factors, ie, the intermediate-term and lifetime risk of stroke faced by the typical 65-year-old man or woman, in each of these periods. To compare exposure to stroke risk factors at age 65 years in each of the 3 periods, we used χ2 tests for dichotomous variables and t tests for continuously distributed variables. We used a modified Kaplan-Meier method to estimate the 10-year mortality-adjusted cumulative incidence and the lifetime risk of stroke in participants who were free of stroke at age 65 years.
In these analyses we censored data for participants 90 years and older due to unreliability of the lifetime risk estimates in the 10th decade. These analyses of 10-year mortality-adjusted cumulative incidence and lifetime risk of stroke were estimated in participants who reached the index age free of the event, using a modified double-decrement survival analysis technique (Kaplan-Meier with survival age as the time scale) that we have described previously.21 In this technique, persons reaching the index age (65 years) free of the event of interest (clinical stroke or ABI) undergo follow-up until they develop the event, die free of any stroke (or ABI), reach the upper age limit (90 years), or until their last medical examination or contact with the Framingham Study at which they are determined to be free of stroke. In standard Kaplan-Meier analyses it is assumed that persons who die would, if they had lived, develop the disease of interest at the same rate as survivors. In the lifetime risk method, death is treated as a true competing risk so that the subsequent risk of stroke for persons who die is set to zero. While the standard Kaplan-Meier analyses are appropriate for studies examining pathophysiology, lifetime risk estimates are appropriate when estimating the actual risk experience of an average individual or the public health burden in a population.
There were 1030 incident clinical strokes recorded in 9152 persons (aged ≥55 years) over 174 917 person-years of follow-up; 450 (44%) were in men, and 629 (61%) were ABIs. These data were used to estimate the age-adjusted annual incidence, stroke severity, and 30-day case fatality across the 3 periods. There were 818 incident clinical strokes recorded in 6806 persons between the ages of 65 and 89 years (number of persons at risk was small at ages ≥90 years) over 93 760 person-years of follow-up; 356 (44%) were in men, and 498 (61%) were ABIs. These data were used to estimate the lifetime risk of stroke and ABI. There were 320 incident clinical strokes (174 in men) between the ages of 65 and 74 years (56 550 person-years), with 98 ABIs; these data were used to estimate the intermediate-term risk (mortality-adjusted 10-year cumulative risk) of stroke.
The prevalence of risk factors (examined at age 65 years) improved overall, but there were some divergent trends. Mean systolic blood pressure and cholesterol levels, prevalence of hypertension, proportion of all participants receiving treatment for hypertension, and prevalence of current smoking were noted to improve significantly across the 3 periods. However, the prevalence of diabetes in women and of atrial fibrillation in men, and the mean body mass index in both sexes, worsened significantly over time. Despite this divergence, scores on the Framingham Stroke Risk Profile, a widely known and previously validated risk assessment tool,9 improved significantly for both men and women (Table 1).
The age-adjusted annual incidence of clinical stroke and ABI in participants aged 55 to 94 years decreased over the 3 periods. The incidence of clinical stroke decreased significantly, from 7.6 to 5.3 per 1000 person-years (P = .02 for trend across periods) and from 6.2 to 5.1 per 1000 person-years (P = .01 for trend) in men and women, respectively. There was a similar decrease in the incidence of ABI, from 3.7 to 2.9 per 1000 person-years (P = .01 for trend) in women, but this failed to reach statistical significance in men, with a decrease from 4.9 to 3.6 per 1000 person-years (P = .07 for trend) (Table 2).
The proportion of clinical strokes that were either moderate or severe remained unchanged at 47% to 48% across the 3 periods in men. In women, there was an apparent increase from 40% to 60% between the first and second periods, but this difference was not statistically significant (P = .26). The 30-day mortality decreased significantly from 23% to 14% in men (P = .01) but not significantly in women (from 21% to 20%; P = .32) (Table 3).
We analyzed the 10-year mortality-adjusted cumulative incidence as an index of intermediate risk at age 65 years over the 3 periods. Comparing the first and third periods for clinical strokes, this 10-year risk decreased over the 3 periods, from 7.6% to 4.8% in men (P = .01) and from 5.5% to 3.7% in women (P = .04). Data for ABI in particular similarly showed a significant decrease from the first to the third periods in both men and women (Table 4).
Across the 3 periods, the lifetime risk of clinical stroke (by age 90 years) decreased from 19.5% to 14.5% in men aged 65 years (P = .11) and from 18.0% to 16.1% in women (P = .61). Although the decrease in the lifetime risk of stroke was not statistically significant, the 10-year mortality-adjusted cumulative risks and the lifetime risk showed comparable relative decreases. Thus, for example, the 10-year mortality-adjusted cumulative risk of clinical stroke in men aged 65 years decreased from 7.6% to 4.8%, an absolute decrease of 2.8% corresponding to a relative decrease of approximately 37% from the baseline value of 7.6%. This finding was statistically significant (P = .01). In comparison, the lifetime risk of stroke in men aged 65 years, estimated over a longer 30-year period, achieved an absolute decrease of 5% as described above, corresponding to a relative decrease of approximately 30%, but this did not reach statistical significance. This difference in statistical significance may be explained by the fact that the lifetime risk estimates are based on smaller sample sizes at older ages; thus, the confidence intervals of these e stimates are wider than those for the 10-year risk estimates. The lifetime risk of ABI also did not vary significantly across the periods (Table 4).
Our study shows a decrease in the age-adjusted annual incidence of first stroke among both men and women. This is in accordance with the results of most prior studies that evaluated temporal trends in stroke risk using annual age-adjusted incidence. The earliest studies18,19 looking at trends in stroke incidence were from Rochester, Minn, and reported a decrease of 45% in the average annual incidence of stroke between 1945 to 1949 and 1975 to 1979. Several subsequent studies—including the Honolulu Heart Program22 and studies from China,23 Denmark,24 Australia,25 Japan,26- 28 Finland,29,30 England (Oxfordshire),5 New Zealand,31 and, more recently, France32—have reported decreasing or stable incidences most often attributed to better treatment of risk factors over time. Only 1 previous study has reported an increasing stroke incidence.33
Several studies18,19,26- 29,34,35 have reported an attenuation of the decrease in the stroke incidence in the 1980s-1990s, which they attributed to changes in case-finding methods18,19 and to increased diagnostic sensitivity of imaging techniques.18,19,26 We did not find a definite flattening of the decrease across periods, perhaps because our comprehensive surveillance resulted in fewer missed strokes even in the era prior to CT and MRI imaging. An additional explanation would be our consistent emphasis on clinical diagnostic criteria supplemented, rather than supplanted, by imaging.
We evaluated the mortality-adjusted 10-year cumulative incidence of clinical stroke and ABI at age 65 years as an index of intermediate-term risk and found that these risks decreased in both men and women. We then assessed the lifetime risk of clinical stroke and ABI.
The lifetime risk statistic describes the long-term individual risk for an incident event over the course of an individual's remaining lifetime, adjusting for the competing risk of mortality in the cohort. Its usefulness is that it is more easily understood by the public than measures such as age-adjusted incidence. We had previously reported16 the lifetime risk of stroke in the Framingham Study sample as 1 in 6 for men and 1 in 5 for women but had not examined temporal trends. We have now extended the prior analyses to examine the lifetime risk of stroke at age 65 years across the 3 periods and have found a decreasing trend that fails to reach statistical significance for the lifetime risk of stroke in general or of ABI in particular, for either men or women. Given that the age-adjusted annual incidence and 10-year cumulative incidence has decreased over the last 50 years, the trend in lifetime risk of clinical stroke and ABI is likely due to increasing life expectancy, which partially offsets the decreases in annual and shorter-term risk.
Several studies5,20,26,29,36 have reported a temporal trend toward decreasing severity and disability from stroke. This has been attributed to the increasing sensitivity of imaging techniques such as CT and MRI for identifying small infarctions and hemorrhages20,29 and to an increased awareness among clinicians of transient ischemic attacks.37 In our sample there was an increasing number of participants with imaging results available. In the last period, 90.6% of persons diagnosed with stroke had undergone either a CT or an MRI, whereas only 16.3% in the first period had undergone these imaging procedures. Despite this, we did not find a decrease in severity across the periods. A greater diagnostic sensitivity for small, less severe strokes may have offset the increasing prevalence of more severe strokes expected with an aging population. We had previously reported that 45% of all strokes occurring between 1953 and 1973 in the original Framingham cohort were moderate to severe,38 and our current estimates for 1990-2004 ranged from 48% in men to 60% in women.
The trend toward a greater severity in women, while not significant, is intriguing. The sex difference is not due to more hemorrhages in women, and the trends for ABI are nearly parallel to those for clinical stroke. The reasons for this divergence are unclear but may be related to the older age at initial stroke, because when we adjusted for age the difference in severity between the periods no longer reached statistical significance.
Several studies,38- 40 including a previous study from Framingham,38 assessed the influence of sex and age on disability following stroke and reported that women, compared with men, were more likely to be dependent for activities of daily living, to require assistance with ambulation, to need nursing home placement, and to have greater overall disability following stroke. After adjusting for age and stroke type it was found that the older age at initial stroke noted in women largely accounted for the sex difference in disability. Two recent studies39,40 have reported that age influenced stroke outcome because of poorer compensatory ability and fewer social supports in elderly persons.
In most prior studies the reported 30-day mortality following stroke has ranged from 15% to 30%, which is consistent with our observations.5,24- 26,29,36,41- 45 Temporal trends in stroke mortality have either remained stable24,29,44,46,47 or declined.17,26- 29,37,48- 50 The decline in mortality has been attributed to both a decreasing case fatality51- 53 and a decrease in stroke incidence.54,55
We found a divergence in temporal trends in 30-day mortality following stroke in men and women; whereas men had a significant decrease over the 3 periods, women did not. The average case-fatality rates have been higher in women in nearly all of the populations studied as part of the World Health Organization MONICA (Monitoring Trends and Determinants in Cardiovascular Disease) Project,56 but that study did not include further analysis of stroke subtype. Morikawa et al,26 in the Oyabe Study conducted in rural Japan, also reported higher case-fatality rates for women aged 65 to 74 years from 1977 to 1986, with improvement in case fatality for both men and women in later periods.
The Framingham Study benefits from the availability of 5 decades of data, a well-organized surveillance system to ascertain incident stroke cases, the direct ascertainment (ie, not based on record review) of events, the consistent use of standardized definitions and codes, and excellent continuity among assessing stroke neurologists since its inception. An obvious limitation of our study is the predominantly European ancestry of our study sample; our findings require replication in other more racially and ethnically diverse samples. The Framingham Study sample is relatively small compared with those of larger studies based on record review; hence, we had limited statistical power to detect very modest changes in stroke incidence rates and lifetime risks. Further detailed analysis of the trends in risk factor prevalence across the periods and within age subgroups and a modeling of the effects of these trends on the incidence, lifetime risk, severity, and case fatality of stroke are beyond the scope of this article.
There has been a decrease in the age-adjusted annual incidence of clinical stroke and ABI in both men and women over the last 50 years, and the 10-year risk of stroke for a man or woman aged 65 years has also decreased significantly over this period. A more modest relative decrease in lifetime risk was seen, which was not statistically significant. The severity of stroke has not decreased and 30-day mortality has decreased significantly only in men, perhaps due to an older age at onset of stroke and more severe strokes in women. These sobering trends emphasize that while improved control of risk factors has lowered incidence of stroke, there is a need for greater primary prevention efforts to reduce the lifetime risk, severity, and 30-day mortality following stroke.
Corresponding Author and Reprints: Philip A. Wolf, MD, Department of Neurology, Boston University School of Medicine, 715 Albany St, B-608, Boston, MA 02118-2526 (firstname.lastname@example.org).
Author Contributions: Drs Seshadri and Beiser had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Carandang, Seshadri, Beiser, Wolf.
Acquisition of data: Seshadri, Beiser, Kelly-Hayes.
Analysis and interpretation of data: Carandang, Seshadri, Beiser, Kase, Kannel, Wolf.
Drafting of the manuscript: Carandang, Seshadri.
Critical revision of the manuscript for important intellectual content: Seshadri, Beiser, Kelly-Hayes, Kase, Kannel, Wolf.
Statistical analysis: Beiser.
Obtained funding: Wolf.
Administrative, technical, or material support: Seshadri, Kelly-Hayes, Wolf.
Study supervision: Seshadri, Kase, Wolf.
Financial Disclosures: Dr Wolf reports that he served on the data monitoring committee of a large stroke prevention trial and as chair of the PROFESS DMC trial sponsored by Boehringer-Ingelheim Ltd. No other authors reported disclosures.
Funding/Support: This study was supported by grants from the National Institute of Neurological Disorders and Stroke (5R01-NS17950) and the National Heart, Lung, and Blood Institute's Framingham Heart Study (NIH/NHLBI Contract N01-HC-25195).
Role of the Sponsors: The National Institute of Neurological Disorders and Stroke and the National Heart, Lung, and Blood Institute had no role in the design and conduct of the study; the collection, analysis, and interpretation of the data; or the preparation and review of the manuscript. These sponsors did approve the manuscript as submitted.