Original Article
Self-reported stroke and myocardial infarction had adequate sensitivity in a population-based prospective study JPHC (Japan Public Health Center)–based Prospective Study

https://doi.org/10.1016/j.jclinepi.2008.07.016Get rights and content

Abstract

Objective

We sought to clarify the validity of self-reported stroke and myocardial infarction (MI) among Japanese population, because information on the validity, particularly on the sensitivity, of self-reported cardiovascular disease is limited and may differ among countries.

Study Design and Setting

Using the 10-year follow-up questionnaire and a stroke and MI registry in the Japan Public Health Center–based prospective Study (JPHC Study) cohort (n = 91,186), we calculated sensitivity and positive predictive values of self-reported stroke and MI incidence over 10 years.

Results

Sensitivity of self-reported incident stroke was 73%, and that for MI was 82%. Positive predictive values were 57% for stroke and 43% for MI. The supplemental inclusion of self-reported angina pectoris increased the sensitivity of MI to 89%, but attenuated the positive predictive value to 18%. Sensitivity of self-reported stroke was highest for subarachnoid hemorrhage (88%), but did not differ greatly among other stroke subtypes, affected sites or size.

Conclusion

Self-reported stroke and MI seem sensitive enough to use for exclusion of stroke and MI at baseline in Japanese cohort studies. However, self-report has too many false positives to be used as the only criterion for incident stroke and MI.

Introduction

Self-reported information on cardiovascular disease (CVD), such as stroke or myocardial infarction (MI), has been often used in cohort studies to exclude individuals with prevalent CVD or to capture incident CVD events. Although several Western studies have reported the accuracy of self-reports, reports of the sensitivity of self-reported CVD has been limited in Japan, where the case mix of incident CVD is quite different from Western countries, with high incidence of stroke, especially hemorrhagic and lacunar stroke, but low incidence of MI or large-artery occlusive stroke [1], [2]. Furthermore, the saturation level of computed tomography (CT) and/or magnetic resonance imaging (MRI) in Japanese hospitals is the highest in the world [3], which may permit Japanese physicians to easily detect small cerebral infarctions. The severity and fatality of MI in Japan is considered to be lower than that in Western countries [4], [5], [6]. These ecological differences, that is, higher proportion and detectability of mild stroke and lower proportion of severe MI, may lower the sensitivity of self-reported stroke and MI compared with Western countries. However, data on this issue among non-Western countries have been scant.

We thus examined the accuracy of self-reported stroke and MI in a cohort of middle-aged Japanese. The Japan Public Health Center–based prospective Study on cancer and CVDs (JPHC Study) [7] is a nationwide community-based prospective study of approximately 100,000 Japanese participants, with systematic surveillance of cancer, stroke, and MI. Previously, the sensitivity of self-reported cancer in this cohort was reported [8]. The aim of this study was to clarify the accuracy of self-reported incident stroke and MI over 10 years in the general Japanese population. We focused particularly on the sensitivity of self-reported stroke and MI as the primary aim of this study, because not many studies have reported sensitivity of these diseases [9], [10], [11], [12], [13], [14], [15].

Section snippets

Study population

Details of the JPHC Study protocol were described elsewhere [7]. Briefly, the JPHC Study included two local community cohorts based on Public Health Center (PHC) areas in Japan; Cohort I (started in 1990, with four PHC areas: Ninohe PHC area, northeastern Japan; Yokote PHC area, northeastern Japan; Saku PHC area, central Japan; and Ishikawa [currently Chubu] PHC area, southwestern isolated island), and Cohort II (started in 1993, with five PHC areas: Kasama [currently Mito] PHC area,

Results

During the 10-year follow-up, 2,760 first-ever strokes and 591 first-ever MIs were registered. Of these, 767 strokes (28%) and 151 MIs (26%) were fatal during follow-up, and 546 nonfatal strokes (20%) and 98 nonfatal MIs (17%) did not respond to the 10-year follow-up questionnaire. These cases were excluded from the analyses. Thus, 1,447 strokes and 342 MIs with 10-year follow-up questionnaires were included in the analysis.

Table 1 presents the sensitivity and positive predictive values for

Discussion

Sensitivities of self-reported stroke and MI were 73% and 82%, respectively, in the present study. This estimate for MI improved after including self-reported angina pectoris (89%), but with a large increase in false positives. Sensitivity was higher for subarachnoid hemorrhage than other subtypes of stroke.

The sensitivities in the present study were somewhat lower than those reported from most of the previous Western studies, in particular, for stroke. For example, a study from Olmsted County

Acknowledgments

The authors thank Mizue Fujii, Tomoko Seki, Keiko Nemoto, and Dr. ChoyLye Chei, University of Tsukuba, for technical assistances, and all staff members in each study area and those in the central office for their painstaking efforts to conduct the baseline survey and follow-up. The authors also appreciate Prof. Aaron R. Folsom, University of Minnesota, for valuable comments.

Members of the Japan Public Health Center–based Prospective Study Group (JPHC Study Group, the principal investigator: S

References (18)

There are more references available in the full text version of this article.

Cited by (46)

  • Association of the length of oestrogen exposure with risk of incident stroke in postmenopausal women: Insights from a 20-year prospective study

    2021, International Journal of Cardiology
    Citation Excerpt :

    The prevalence and bias adjusted kappa was 0.98 [23,39]. Further, stroke as an outcome has been consistently shown to have good sensitivity (~81%), specificity (~90%), positive predictive value (~64%) and kappa-agreement (~0.70) between physicians diagnosed stroke events (collected through questionnaire surveys) and hospital diagnosed events collected through hospital admission data [40–44]. Though, most cases of stroke in this age group would likely to be ischemic in origin, this study did not have sufficient information to define and exclude haemorrhagic stroke them from the analysis.

  • Is Corneal Arcus Independently Associated With Incident Cardiovascular Disease in Asians?

    2017, American Journal of Ophthalmology
    Citation Excerpt :

    Nevertheless, we have no reason to believe that the self-reported CVD rates would be differentially higher in people with and without CA. We also note that several studies have shown self-reported outcomes for CVD to be valid in the estimation of actual outcomes35–37 and have also stated that classification accuracy further improves when single event definitions such as myocardial infarction, stroke, and angina are combined into composite outcomes.38,39 Combined with our reliability results of ∼75%, the self-reported CVD outcomes captured in our study may remain an appropriate indicator of actual CVD events.

  • Age, sex, and race/ethnic temporal trends in metabolic syndrome prevalence among individuals with myocardial infarction or stroke in the United States

    2017, Journal of the Neurological Sciences
    Citation Excerpt :

    Although NHANES has not validated self-reporting of stroke and MI, the use of this approach to estimate the prevalence of stroke is frequent in epidemiological studies, with sensitivity and specificity of 90% and 96% respectively [18]. Similarly, the level of sensitivity for self-reported history of MI ranges from 80 to 82% with a level of specificity of 99.4% [19]. Our study's major strengths include the rigorous, validated, standardized, and consistent methods for collecting data in NHANES each year and nationwide scope.

  • Long-term survival of older australian women with a history of stroke

    2015, Journal of Stroke and Cerebrovascular Diseases
    Citation Excerpt :

    The primary limitation of this study is that it is based on self-reported stroke, without verification from clinical information, which may result in both over- and under-reporting. However reports of the validity of self-reported stroke vary considerably, from low/moderate55,56 to good/very good.57,58 A second limitation is the problem of survivor bias, as participants were recruited when they were already aged 70-75 years.

View all citing articles on Scopus
View full text