Original Articles
Development and validation of a model to estimate stroke incidence in a population*,**

https://doi.org/10.1053/jscd.2003.6Get rights and content

Abstract

Stroke is a common condition with a substantial impact on health care. Using published epidemiological data, a mathematical model was created to predict annual stroke incidence in populations over 45 years old, utilizing age, gender, ethnicity, and stroke risk factor prevalence (hypertension, atrial fibrillation, diabetes, smoking, and ischemic heart disease). The purpose of this study is to assess the models ability to reliably estimate the annual number of first strokes. The model was validated against two cohorts: the Northern Manhattan Stroke Study (NM), performed in 1995 and 1996, and the Copenhagen City Heart Study (CCHS), undertaken in 1980-84, 1984-88, and 1988-93. Both cohorts provided the actual number of first strokes for respective years and risk factor prevalence. The Mantel-Haenszel test compared actual to predicted incidence rates. The two cohorts differed in risk factor prevalences, size, and demographics. For all cohort groups/years, the predicted number of annual first strokes was not statistically different from actual first stroke incidence (P >.05). In NM, the actual number of first strokes compared to predicted was 7 versus 13 (P =.18) for 1995 and 9 versus 18 (P =.08) for 1996. Actual and predicted annual strokes in CCHS for the time frames 1980-83, 1984-88, and 1988-93 were 65 versus 69 (P =.73), 72 versus 87 (P =.23), and 75 versus 93 (P =.16), respectively. The model provides a tool for estimating annual first strokes within a population, with a tendency of bias toward overestimating the number of incident strokes. This evidence-based model may be utilized by health policy makers to predict stroke burden at a population level. Copyright © 2003 by National Stroke Association

Section snippets

The stroke model

The models assumptions were derived from published studies in the stroke medical literature and through discussions with opinion leaders in stroke epidemiology throughout the world.8, 9, 10, 11 A Medline search using the queries, stroke, epidemiology, incidence, and prevalence resulted in over 100 relevant articles, which provided information that was used to develop the models logic. Statistical information was also obtained from the following organizations: the American Heart Association, the

Model software

This proprietary model uses a Microsoft Windows operating system and requires data entry of the defined population over the age of 45 years by age and gender in 10-year increments. Risk factor prevalence is entered for all risk factors. Outputs of the model include predicted baseline annual first strokes and total annual first strokes (after adjusting for the effect of risk factors).

Results

The two cohort populations differed in their risk factor prevalence profiles, cohort size, and demographics (Tables 1 and 2).The Copenhagen cohort (Table 1) for the time periods evaluated was approximately 13,000, while the NOMASS cohort (Table 2) was comprised of 806 and 1060 participants for 1995 and 1996, respectively.

The NOMASS cohort was predominantly Caribbean-Hispanic (46% in 1995 and 40% in 1996) and African American (33% and 35%, respectively), while the Copenhagen cohort was

Discussion

Accurate epidemiological data are required to optimize planning for health care services, such as training of specialists, types of hospital and support services provided, implementation of public health programs, etc. Most countries lack accurate data regarding the true burden of stroke in their populations. Additionally, measures using hospitalization rates or administrative claims are not always an accurate reflection of stroke incidence.20, 21, 22 We present a model that utilizes commonly

Conclusion

As populations age, the impact of stroke on the population and the resultant economics of health care will grow. Although the ranking of stroke as a cause of death may not change in the near-term, acute stroke and its attendant disability is projected to increase worldwide in the next two decades. Our stroke model appears to be a useful tool for estimating annual first strokes within a population. This evidence-based model can be utilized by health care policy-makers and educators to predict

Acknowledgements

The authors would like to thank Dave Laitenen, MPH, for his analytic and technical assistance.

References (29)

  • CJL Murray et al.

    Mortality by cause for eight regions of the world: Global burden of disease study

    Lancet

    (1997)
  • WB Kannel et al.

    Prevalence, incidence, prognosis, and predisposing conditions for atrial fibrillation: Population-based estimates

    Am J Cardiol

    (1998)
  • CLM Sudlow et al.

    Comparing stroke incidence worldwide

    Stroke

    (1996)
  • O Fustinoni et al.

    Ethnicity and stroke: Beware of the fallacies

    Stroke

    (2000)
  • Blood pressure, cholesterol and stroke in eastern Asia

    Lancet

    (1998)
  • CS Anderson et al.

    Ascertaining the true incidence of stroke: Experience from the Perth Community Stroke Study, 1989-1990

    Med J Aust

    (1993)
  • Asian Acute Stroke Advisory Panel

    Stroke epidemiological data of nine Asian countries

    J Med Assoc Thai

    (2000)
  • P Thorvaldsen et al.

    Stroke incidence, case fatality, and mortality in the WHO MONICA project. World Health Organization Monitoring Trends and Determinants in Cardiovascular Disease

    Stroke

    (1995)
  • J Bamford et al.

    A prospective study of acute cerebrovascular disease in the community: The Oxfordshire Community Stroke Project 1981-86

    J Neurol Neurosurg Psychiatry

    (1988)
  • RD Brown et al.

    Stroke incidence, prevalence, and survival: secular trends in Rochester, Minnesota, through 1989

    Stroke

    (1996)
  • PA Wolf et al.

    Probability of stroke: A risk profile from the Framingham Study

    Stroke

    (1991)
  • RD Brown et al.

    Incidence of transient ischemic attack in Rochester, Minnesota, 1985-1989

    Stroke

    (1998)
  • ...
  • ...
  • Cited by (3)

    *

    Supported by Boehringer Ingelheim, GmbH, Ingelheim, Germany. The Northem Manhattan Stroke Study is supported by grants from NINDS (R01 NS 27517, 29993). The majority of the work was performed at Protocare Sciences in Santa Monica, CA.

    **

    Address reprint requests to: Carol Zaher, MD, MBA, MPH, Protocare Sciences, 2400 Broadway, Suite 100, Santa Monica, CA 90404.

    View full text