What is the value of a family history of premature cardiovascular disease in predicting increased risk of cardiovascular disease?

Transcription

1 NZHTA EVIDENCE TABLES 10 January 2003 What is the value of a family history of premature cardiovascular disease in predicting increased risk of cardiovascular disease? A critical appraisal of the literature presented in Evidence Tables Carolyn Doughty New Zealand Health Technology Assessment Department of Public Health and General Practice Christchurch School of Medicine Christchurch, NZ. Division of Health Sciences, University of Otago

2 NEW ZEALAND HEALTH TECHNOLOGY ASSESSMENT (NZHTA) THE CLEARING HOUSE FOR HEALTH OUTCOMES AND HEALTH TECHNOLOGY ASSESSMENT Department of Public Health and General Practice Christchurch School of Medicine, Christchurch, New Zealand What is the value of a family history of premature cardiovascular disease in predicting increased risk of cardiovascular disease? A critical appraisal of the literature presented in Evidence Tables Carolyn Doughty NZHTA EVIDENCE TABLES 10 January 2003

3 This report should be referenced as follows: Doughty C. What is the value of a family history of premature cardiovascular disease in predicting increased risk of cardiovascular disease? NZHTA Evidence Tables New Zealand Health Technology Assessment (NZHTA) ISBN X

4

5 i ACKNOWLEDGEMENTS These evidence tables were produced by staff of NZHTA. The report was prepared by Dr Carolyn Doughty (Research Fellow), who conducted the critical appraisals, prepared the report and coordinated the project. Marita Broadstock assisted with report preparation, selection of articles and Dr Ray Kirk (Director), provided peer review. Margaret Paterson (Information Specialist) developed and undertook the search strategy and coordinated retrieval of documents. Cecilia Tolan (Administrator) provided administrative support and Mrs Ally Reid (Secretary/Word Processor) provided document formatting. Philippa Monkman assisted with retrieval of documents. The Canterbury Medical Library assisted with the retrieval of articles. NZHTA is a Research Unit of the University of Otago funded under contract to the Ministry of Health. The evidence tables were commissioned by the New Zealand Guidelines Group, funded by the New Zealand s Ministry of Health. We thank Dr Rob Cook for assisting in developing the scope of the review and providing background material for the review. DISCLAIMER New Zealand Health Technology Assessment (NZHTA) takes great care to ensure the information supplied within the project timeframe is accurate, but neither NZHTA, the University of Otago, nor the contributors involved can accept responsibility for any errors or omissions. The reader should always consult the original database from which each abstract is derived along with the original articles before making decisions based on a document or abstract. All responsibility for action based on any information in this report rests with the reader. NZHTA and the University of Otago accept no liability for any loss of whatever kind, or damage, arising from reliance in whole or part, by any person, corporate or natural, on the contents of this report. This document is not intended as personal health advice. People seeking individual medical advice are referred to their physician. The views expressed in this report are those of NZHTA and do not necessarily represent those of the University of Otago or the New Zealand Ministry of Health. These evidence tables were commissioned by Dr Rob Cook, on behalf of the New Zealand Ministry of Health. NZHTA is a Research Unit of the University of Otago and funded under contract by the New Zealand Ministry of Health. COPYRIGHT This work is copyright. Apart from any use as permitted under the Copyright Act 1994 no part may be reproduced by any process without written permission from New Zealand Health Technology Assessment. Requests and inquiries concerning reproduction and rights should be directed to the Director, New Zealand Health Technology Assessment, Christchurch School of Medicine and Health Sciences, P O Box 4345, Christchurch, New Zealand.

6 ii CONTACT DETAILS New Zealand Health Technology Assessment (NZHTA) The Clearing House for Health Outcomes and Health Technology Assessment Department of Public Health and General Practice Christchurch School of Medicine and Health Sciences PO Box 4345 Christchurch New Zealand Tel: Fax: nzhta@chmeds.ac.nz Web Site:

7 iii LIST OF ABBREVIATIONS AND ACRONYMS ANCOVA analysis of covariance ANOVA analysis of variance CAD coronary artery disease CI confidence interval CHD coronary heart disease CVD cardiovascular disease ECG electrocardiogram FCHL familial hyperlipidemia FDR first-degree relative FH family history FHTG familial hypertriglyceridemia FRS family risk score HRR hazard rate ratio ICD international classification of diseases IS ischaemic stroke NIDDM non-insulin-dependent diabetes mellitus NZGG New Zealand Guidelines Group NZHTA New Zealand Health Technology Assessment MI myocardial infarction MOH Ministry of Health (NZ) ns not significant (i.e., p value >0.05) OR odds ratio PCA primary cardiac arrest RR relative risk SD sudden death TIA transient ischaemic stroke UA unstable angina WHO World Health Organisation

8 iv GLOSSARY Anamesis - The ability to recall past events. Confounding variable - A variable that can cause or prevent the outcome of interest, is not an intermediate (mediating) variable, and is associated with the factor under investigation. Unless it is possible to adjust for confounding variables, their effects cannot be distinguished from those of the factor(s) being studied. Cox regression - This is also known as the proportional hazards model and is used in survival analyses to provide a mathematical way of relating survival to time. Grey literature - That which is produced by all levels of government, academics, business and industry, in print and electronic formats, but which is not controlled by commercial publishers. Haemorrhagic stroke - A haemorrhagic stroke results from rupture of an artery wall. Hypercholesterolemia - An elevated concentration of cholesterol in the blood. Hyperlipidemia - The presence in the blood of an abnormally high concentration of fats. Ischaemic stroke - An ischaemic stroke occurs when the flow of blood is prevented by clotting or by a detached clot that lodges in an artery. Logistic regression - This technique is based on the logit transformation of a proportion and is a useful tool for modelling categorical data. The response (dependent) variable can take one of two values 0 or 1, occurrence or non-occurrence of an event e.g., death from myocardial infarction. For such binary or dichotomous variables, the objective is to develop a method for estimating the probability of occurrence of an event, expressed as a function of a number of independent variables. Mantel-Haenszel test - A test of conditional independence that can be applied to K strata of two by two tables. Response totals are treated as fixed and data can be analysed by conditioning, within each stratum, on both group totals and the response totals. Under the null hypothesis of conditional independence, this statistic has approximately a chi-squared distribution with one degree of freedom. Mendelian stroke - Type(s) of stroke that are inherited according to the principles of heredity derived from Mendel s laws. Myocardial infarction - Death of a segment of heart muscle, which follows interruption of its blood supply. Myocardial infarction is usually confined to the left ventricle. The patient experiences a heart attack usually consisting of sudden severe chest pain, which may spread to the arms and throat. Premature family history - A first degree relative (parent, sibling or child) that has history of myocardial infarction or ischaemic stroke at ages younger than 55 years in men and younger than 65 years in women. Risk factor - An exposure or aspect of personal behaviour or lifestyle, which on the basis of epidemiologic evidence is associated with a health-related condition. Stroke - A sudden attack of weakness affecting one side of the body. It is the consequence of an interruption of blood to the brain. The primary disease is in the heart or blood vessels and the effect on the brain is secondary. Transient ischaemic attack - The result of temporary disruption of the circulation to part of the brain due to embolism, thrombosis to brain arteries or spasm of the vessel walls. The symptoms may be similar to those of a stroke but patients recover within 24 hours.

9 1 Methodology BACKGROUND The Ministry of Health is funding a collaborative venture between the New Zealand Guidelines Group, the Stroke Foundation of New Zealand and the National Heart Foundation to develop an integrated set of best practice, evidence-based guidelines for cardiovascular disease. To inform this work, the New Zealand Health Technology Assessment (NZHTA) will prepare evidence tables reporting on critical appraisals of research studies relevant to specific aspects of the guideline. There is growing evidence that most cardiovascular disease (CVD) is preventable. Results of long-term prospective studies consistently identify persons with low levels of risk factors as having lifelong low levels of heart disease and stroke. Moreover, these low levels of risk factors are related to healthy lifestyles (Stampfer et al. 2000, Stamler et al. 1999). Nevertheless, it is likely that there may be a group of high-risk patients for whom screening and intervention in first-degree relatives (including children) would be an important aspect of primary prevention (Pearson et al. 2002). It has been proposed that individuals who have a first degree relative with a history of premature CVD (younger than 55 years in men and 65 years in women) may themselves have an increased risk of CVD. This topic will inform the guideline about whether a positive family history of premature myocardial infarction (MI) or ischaemic stroke (IS) is a useful predictor of subsequent CVD. It will identify studies that have compared the CVD mortality and morbidity outcomes for people with and without a family history of premature CVD. Although family history has been proposed as one of the risk factors for CVD, its role is not unequivocal. In some studies a significant association was found between the family history and morbidity or mortality of CVD (Graffagnino et al. 1994, Jousalahti et al. 1997), while in others such an association was not found (Kiely et al. 1993). Saito et al. (2000) suggests that the disagreement may derive from complexities of the genetic role and differences in study design. The definition of a positive family history also differs and may include the presence of a family member with the past history of CVD or presence of a family member who developed the CVD under a certain age. The range of family members also differs, sometimes limited to parents (Jousalahti et al. 1997) and sometimes extending to uncles and aunts (Brass et al. 1991, Graffagnino et al. 1994). Analytical methods employed to investigate this association vary, but include the use of cross-sectional studies, community studies and family studies including twin studies. The two most common epidemiological methods are 1) retrospective comparisons of disease prevalence in first-degree relatives of heart disease patients versus healthy controls, or in relatives of patients with extensive disease versus those with less or no disease; and 2) prospective examinations of disease incidence in those with versus those without a positive family history, both approaches should seek to control for other major risk factors. Results derived from epidemiological studies, as well as details on the study design, sample, measures, limitations and study quality will be described and presented in the form of an evidence table. SELECTION CRITERIA Study inclusion criteria Publication type Studies published between 1990 and August, 2002 inclusive in the English language, including primary (original) research (published as full original reports) and secondary research (systematic reviews and meta-analyses) appearing in the published literature.

10 2 Study design Peer reviewed studies will be considered for this review if they employ one of the following study designs: systematic review or meta-analysis case-control study cohort study. Note: Any identified unpublished or grey literature will be included for New Zealand specific studies only. Comparison groups studies comparing those with and without a family history of premature CVD. 1 Population study population are adults (aged 25-75) from the general population. Sample size Studies with samples of at least 20 participants. Outcomes Outcomes considered included: mortality morbidity: - new onset angina - non-fatal myocardial infarction (MI) - ischaemic stroke (IS) - transient ischaemic attack (TIA). Study exclusion criteria Research papers were excluded if they: were not published in English were correspondence, book chapters, conference proceedings, abstracts reported studies with fewer than 20 persons included in reported relevant outcomes did not clearly describe their methods and results, or had significant discrepancies concerned primarily with: genetic analyses (e.g., testing polymorphisms) transmission of risk within families (e.g., familial aggregation, twin studies) balloon angioplasty, coronary artery bypass surgery, carotid intima-media thickness (IMT) or carotid atherosclerotic plaques 1 Premature family history will be defined as a first degree relative (parent, sibling or child) that has history of myocardial infarction or ischaemic stroke at ages younger than 55 years in men and younger than 65 years in women.

11 3 or concerned a study population (50% or more): where there was a family history of cardiovascular disease but no indication of age of onset or age at first event (MI, IS or TIA) outside the age range or without separate analyses on the year age group with haemorrhagic stroke with ruptured berry aneurysm/subarachnoid haemorrhage with peripheral vascular disease without comorbid cardiovascular disease with familial hyperlipidemia (FCHL) and/or familial hypertriglyceridemia (FHTG). SEARCH STRATEGY A systematic method of literature searching and selection was employed in the preparation of this review. Searches were limited to English language material published from 1990 onwards. The searches were completed on 10 September Principal sources of information The following databases were searched (using the search strategy outlined in Appendix 1): Bibliographic databases Cinahl Current Contents Embase Medline Science Citation Index Social Science Citation Index Index New Zealand Review databases Best Evidence Cochrane Library Database of Abstracts of Reviews of Effectiveness Health Technology Assessment database NHS Economic Evaluation database Other sources Te Puna - National Library of New Zealand National Guideline Clearing House Scottish Incollegiate Guidelines Network Center for Disease Control and Prevention Health Canada Australian Department of Health and Ageing American Heart Association Cardiovascular Research Foundation British Heart Foundation Canadian Cardiovascular Society

12 4 Search terms used index terms from Medline (MeSH terms): exp myocardial infarction, exp cerebrovascular accident, family health index terms from Embase: exp myocardial infarction, exp cerebrovascular accident, family health, sibling, exp parent, family history, heredity, cardiovascular risk the above index terms were used as keywords in databases where they were not available and in those databases without controlled vocabulary additional keywords (not standard index terms) were used in all databases: myocardial infarction, heart attack, heart infarction, ischaemic stroke, cerebral infarction, sibling*, sister, brother, mother, father, parent, parental, parents Medline floating subheading: ge (genetics). A systematic method of literature searching and selection was employed in the preparation of this review. STUDY SELECTION Studies were selected for appraisal using a two-stage process. Initially, the titles and abstracts (where available) identified from the search strategy, were scanned and excluded as appropriate. The full text articles were retrieved for the remaining studies and these were appraised if they fulfilled the study selection criteria outlined above. APPRAISAL OF STUDIES Articles were formally appraised using the NZGG s GATE FRAME checklists for case control and cohort studies (prognosis) graded using NZGG summary levels of Evidence. There were four gradings applied, as follows: 1. quality of review design (minimising bias) 2. quality of results (understandable, precise, and/or sufficient power) 3. quality of study applicability: could applicability be determined? 4. quality of study applicability: are findings applicable in usual practice/settings? Each Grade was coded as one of the following: Very well (+), Okay ( ), Poorly ( ) or as a combination of these. EVIDENCE TABLES Evidence tables for primary research studies present key information summaries described below. study source including authors, year published, and country of origin study design and setting, inclusion and exclusion criteria sample characteristics including sample size, sample characteristics including demographic variables, any comparisons between intervention groups on these variables at baseline, definitions, methods results including statistically tested comparisons and reporting relevant statistical data, authors conclusions study quality gradings (applying GATE FRAME criteria as described above) and specific study limitations.

13 5 LIMITATIONS OF THE REVIEW This study has used a structured approach to review the literature. However, there were some inherent limitations with this approach. Namely, systematic reviews are limited by the quality of the studies included in the review and the review s methodology. This review has been limited by the restriction to English language studies. Restriction by language may result in study bias, but the direction of this bias cannot be determined. In addition, the review has been limited to the published academic literature, and has not appraised unpublished work. Restriction to the published literature is likely to lead to bias since the unpublished literature tends to consist of studies not identifying a significant result. Papers published pre-1990 were not considered as these tended to concern outdated practices or poor study designs. The studies were initially selected by examining the abstracts of these articles. Therefore, it is possible that some studies were inappropriately excluded prior to the examination of the full text article. Investigators appear to have used the terms denoted by CVD, CHD or CAD more or less interchangeably; the convention followed by this review was to use the same acronym within an evidence table that was used by the study being appraised. Although family studies of high-risk groups were excluded, individuals with these conditions are included in the studies appraised where the rate of these disorders was less than 50% of the sample, and broadly typical of that usually found in the general population for the study country. In this review the definitions of early or premature onset of CVD differed from study to study (and in relation to our working definition). However, the age categories for premature onset most frequently used were <55 years, <60 years or <65 years. Where there were differences in the way age categories were reported across gender, the upper limit of <65 years was used to report results for both sexes in the evidence table. If results were presented separately for both sexes, all sub-group analyses were shown in the tables for ages <55 for men and <65 for women. Of the 15 studies appraised, only six studies clearly defined early onset as <55 years in men whereas 13 studies clearly defined early onset as <65 years in women. As a general rule, sub-grouping subjects on the basis of the age at which their relatives experienced disease morbidity or mortality reduces the statistical power available in these studies. This was more of an issue in studies where stroke was the outcome of interest as often there were small numbers of cases or events regardless of the study design. Demonstrating that the effect of family history on CVD morbidity and mortality requires careful identification of positive versus negative family history, complete assessment of known risk factors, and control for covariation of elevated risk factor profiles and positive family history is recommended (Perkins 1986). Degree of risk may vary markedly depending on the number of relatives affected (same or different generation and sex as compared with the proband), and the age at which they are affected. Both retrospective and prospective studies depend upon the reliability of the family history data obtained but often this information was incomplete due to non-response. For example, in Leander et al. (2001) a large proportion of the respondents lacked information about whether their parents or siblings had suffered from, or died of, CHD. Overall, 21% responded don t know to questions about CHD in their mothers, fathers or siblings. Hence, the true response rates were much lower if only individuals with complete family history were included (61% for cases and 58% for controls). Recall bias may be an inherent problem in the studies appraised, as information on the participants family health history and other risk factors was collected retrospectively. It is possible that reporting may be less reliable in older than younger participants, as more time has elapsed since the parental event. Furthermore, in the majority of studies appraised, family history was not validated independently so misclassification of the exposure cannot be excluded.

14 6 The quality of information on other risk factors (confounding factors) varied from study to study, with data on these being largely based on self-report (information from either self-administered questionnaires or structured interviews). Several studies specifically reported that they also conducted a physical examination (Eaton et al. 1996, Jouven et al. 1999, Li et al. 2000) while others opted for a combined approach linking data from self-reports with prescription registries and taking selected blood samples (Jousilahti et al. 1996). The study by Li et al. (2000) was the only study that looked at coronary heart disease in two different ethnic groups. They found that non-respondents compared with respondents were more likely to have low education, low income, poor health and high current smoking which suggests that results from their study may only apply to the relatively healthy, high socio-economic status members from the communities, rather than the communities as a whole. Such variation in response rates by sociodemographic group suggests caution in applying results to other communities. None of the studies in this review were conducted within New Zealand. The generalisability of results to the New Zealand population and context may be limited and need to be considered based on the demographic and clinical features of each sample reported in the tables. Data extraction, critical appraisal and report preparation was performed by a single reviewer. Within the studies conducted in this area multiple risk factors were identified and controlled for, hence only a subset of the reported (adjusted) associations have been reported in these tables. For a detailed description of methods, statistical analyses, and results used in the studies appraised, the reader is referred to the original papers cited. The review scope was developed with the assistance of NZGG s Dr Rob Cook. The review was conducted over a limited timeframe (August, 2002 December, 2002).

15 7 Results From the search strategy we identified over 2,600 potentially relevant articles/abstracts of which 85 were eligible for retrieval. Of these, 70 were excluded for the following reasons: expert opinion/narrative review/background (n=4), case reports with sample size of fewer than 20 (n=1), considered children/adolescents (n=7), family history was not early onset (n=15), fewer than 50% of sample with parental FH of premature CVD (n=2), family history not considered as a risk factor (n=4), eligible outcomes not measured (n=18), cross-sectional design/descriptive (n=4), preliminary data/protocol only (n=2), not relevant to the topic (n=5), correspondence only (n=2), abstract only (n=6). These excluded papers, annotated with the reason for exclusion, are listed in Appendix 2. Fifteen retrieved articles were appraised (listed in Appendix 3). Full details of these papers, including methods, key results, limitations and conclusions, are provided in the evidence table below. Papers are presented in decreasing order of quality (based on four quality gradings given equal weighting), and papers of equal quality are presented in reverse chronological order (i.e., more recently published first).

16 8 Table 1. Evidence table of research studies relating to the value of a family history of premature cardiovascular disease in predicting increased risk of cardiovascular disease Authors Country Study design, setting, inclusion and exclusion criteria Sample characteristics Methods Results and authors conclusions Quality gradings and limitations Sesso et al. (2001) Boston, USA Prospective cohort study Study setting This study examined men from the Physicians Health Study (PHS) and the Women s Health Study (WHS) in the United States. PHS was a completed trial of aspirin and betacarotene in the primary prevention of CVD and cancer. WHS is an ongoing trial of aspirin and vitamin E in the primary prevention of CVD and cancer. Inclusion criteria for PHS male physicians aged 40 to 84 years free from prior MI, stroke, transient ischaemic attack, and cancer (except nonmelanoma skin cancer). Inclusion criteria for WHS female health professionals aged 45 years postmenopausal or not intending to become pregnant free from prior MI, stroke, transient ischaemic attack, and cancer (except non-melanoma skin cancer). Definitions Original cohort: n=61,947 Men: n=22,071 Women: n=39,876 Among the 22,071 randomised men in the PHS, 1556 were excluded resulting in a study population of 20,515. Among the women 1891 were excluded resulting in a study population of 37,985. A further 836 men (132 CVD cases) and 453 women (14 CVD cases) having maternal or paternal history of MI but missing data on parental age had to be excluded from sub-analyses based on parental age. FH of MI for men (women) Maternal: 1159 (2995) Paternal: 5195 (8274) Both: 715 (1686) Where FH in both parents: Mean age: 54.2±8.3 (54.3±6.8) years Hypertension: 15.6% (33.8%) Diabetes: 2.5% (3.6%) Family history was considered positive: If men said yes to either Has your mother ever had a documented MI? or Has your father ever had a documented MI? Data collection Men provided self-reports of baseline coronary risk factors, including age, smoking habits, alcohol, exercise, hypertension and diabetes. At 12-month follow-up, men were asked for the age of parents with MI. Women provided baseline information on the same variables plus postmenopausal status and postmenopausal hormone use. Data analysis Separate analyses were done for men and women. Participants were first examined by categories of parental history with ANCOVA to compare mean values or chisquared tests to compare proportions of baseline coronary risk factors. Cox proportional hazards models estimated the relative risk (RR) and 95% CI of CVD, MI and stroke. Three separate sets of models were fitted for parental history of MI, age of maternal MI, and age of paternal MI. Crude models were examined with the addition of age, then adjustments made for BMI, smoking status, exercise and alcohol intake. There were 2,664 CVD events (including 653 MI cases and 613 stroke cases) in men and 563 CVD events (including 159 MI cases and 195 stroke cases) in women during follow-up. Parental history of MI and RR (95%CI) of MI and Stroke (adjusted for age and other risk factors). MI in men and FH in Mother: RR= % CI Mother <50 years: RR=1.0 95% CI Mother 50-59years: RR= % CI Father: RR= % CI Father <50 years: RR= % CI Father years: RR= % CI Both: RR= % CI Stroke in men and FH in Mother: RR= % CI Father: RR= % CI Both: RR= % CI MI in women and FH in Mother: RR= % CI Mother <50 years: RR= % CI Mother 50-59years: RR= % CI Father: RR= % CI Both: RR= % CI Stroke in women and FH in Mother: RR= % CI Father: RR= % CI Both: RR= % CI Quality gradings quality of design? + quality of results? + could applicability be determined? + are the findings applicable in usual practice? + Specific limitations history of MI in other firstdegree relatives (particularly siblings) was not assessed. Additional details on family history of MI might further improve the definition of positive family history some misclassification of history is likely because younger participants free of parental history of MI may later develop a positive history parental history of MI, parental age and mortality status was not updated through the course of the study, so it is possible that the RRs may have been underestimated, particularly for earlier parental ages of MI.

17 9 Table 1. Evidence table of research studies relating to the value of a family history of premature cardiovascular disease in predicting increased risk of cardiovascular disease (continued) Authors Country Study design, setting, inclusion and exclusion criteria Sample characteristics Methods Results and authors conclusions Quality gradings and limitations Sesso et al. (2001) Boston, USA (Continued) Exclusion criteria for PHS and WHS if there was missing data on parental history of MI if they were found to have had prerandomisation angina. Reference group Those with no parental history. If women said yes to Did any of these relatives ever have an MI? including their mother or father, plus the age category (< 40 years, 40 to 49 years, 50 to 59 years, 60 years, or unknown. In the PHS, age at MI was categorised as none, <50, 50 to 59, 60-69, 70-79, and 80 years. There were a number of other additional covariates. The parameter estimates of maternal versus paternal history of MI were compared by the inclusion of terms for maternal age of MI, maternal history, paternal age of MI, and paternal history in a multivariate model. Likelihood ratio tests compared models with maternal, paternal, and both maternal and paternal history of MI. A premature paternal history was an important, independent predictor of CVD in both men and women. In contrast, maternal history of MI followed quite a different age distribution because CVD occurred predominantly in older, postmenopausal women. Therefore maternal history of MI appears to predict CVD at least as strongly as paternal history of CVD, and also at older ages of maternal MI. the effect of age of parental stroke on association with CVD was not specifically reported, probably because the smaller sample size precluded sub-analyses this study conducted an in-depth analysis of the effect of parental age on the RR of CVD and also examined the joint associations with parental history of MI. Only selected estimates are reported here, for further detail refer to original article.

18 10 Table 1. Evidence table of research studies relating to the value of a family history of premature cardiovascular disease in predicting increased risk of cardiovascular disease (continued) Authors Country Study design, setting, inclusion and exclusion criteria Sample characteristics Methods Results and authors conclusions Quality gradings and limitations Jousilahti et al. (1996) Helsinki, Finland Prospective cohort study Study Setting Cardiovascular risk factor surveys were performed in 1972 and 1977 in the eastern Finnish provinces of North Karelia and Kuopio and in the Southwestern region of Turku- Loimaa. Independent random samples of 6.6% of the population were drawn in both areas in both years. Those who participated in both surveys were only included in the first survey cohort. Inclusion criteria aged 30 to 59 years. Exclusion criteria (of 17682) participants with a reported history of acute myocardial infarction before the risk factor survey (n=314) incomplete data on parental history of CHD (n=238) incomplete data on one or more risk factors or indicators of socio-economic status (n=1510). Original cohort: n=19,894 At the end of follow-up: n=17,682 Men: n=7,605 (791 events) Women: n=8,015 (255 events). Participation rates were 90% among men and 93% among women. The follow-up time of each participant in the present analyses was 12 years. The end point of the follow-up was either nonfatal MI or coronary death, whichever occurred first. Mean follow-up time was 8.6 years in men and 9.7 years in women. FH of CHD in men (women) Age: 42.1(42.9) years Diabetes: 5.4% (5.2%) FH in 1 parent: 22.4% (24.6%) Father only: 13.9% (14.2%) Mother only: 5.9% (7.2%) Both parents: 2.6% (3.1%) Data collection The surveys included a selfadministered questionnaire, sent in advance to participants. This included questions about medical history, health behaviour, socioeconomic status and family history. The history of premature CHD of the parents, fatal or non- fatal MI, or angina pectoris before the age of 60 years, was obtained by questionnaire. Only those participants who had complete data on both parents were included in the analysis. Analyses of variance were used to test the difference in risk factors between participants with and without parental history of CHD. Age was used as a covariate in the analyses. 1) Age and study year adjusted 2) With multifactorial adjustment FH of premature CHD for men 1) Either parent: RR=1.61, 95% CI ) The father: RR=1.72, 95% CI ) The mother: RR=1.38, 95% CI ) Both parents: RR=1.45, 95% CI ) Either parent: RR=1.54, 95% CI ) The father: RR=1.63, 95% CI ) The mother: RR=1.34, 95% CI ) Both parents: RR=1.38, 95% CI FH of premature CHD for women 1) Either parent: RR=1.85, 95% CI ) The father: RR=1.62, 95% CI ) The mother: RR=2.26, 95% CI ) Both parents: RR=1.57, 95% CI ) Either parent: RR=1.80, 95% CI ) The father: RR=1.57, 95% CI ) The mother: RR=2.20, 95% CI ) Both parents: RR=1.29, 95% CI Quality gradings quality of design? + quality of results? + could applicability be determined? + are the findings applicable in usual practice? + Specific limitations the number of CHD events among participants with a family history of both parents was small so no definate conclusions for this subset can be drawn family history was identified through selfreport. This information was not validated in any way so there is the potential for misclassification.

19 11 Table 1. Evidence table of research studies relating to the value of a family history of premature cardiovascular disease in predicting increased risk of cardiovascular disease (continued) Authors Country Study design, setting, inclusion and exclusion criteria Sample characteristics Methods Results and authors conclusions Quality gradings and limitations Jousilahti et al. (1996) Helsinki, Finland (Continued) Reference group In all analyses this consisted of those participants neither of whose parents had a reported history of premature CHD. Definitions Parental history of premature CHD was defined as either fatal or nonfatal myocardial infarction or angina pectoris before the age of 60 years. Data analysis Multivariate analyses were performed using the Cox proportional hazard model. Age and study year adjusted risk ratios (RR) of CHD incidence were calculated among the participants in different categories of positive family history of premature CHD: (1) a family history involving either parent, (2) a family history involving the father only, (3) a family history involving the mother only, or (4) a family history involving both parents. To analyse whether the effect of positive parental history on the risk of acute MI depends on the age at the onset of disease, a stratified analysis was made by age at the time of the first coronary event. Adjustment for socio-economic indicators did not alter the estimates. Risk of acute MI associated with parental history of premature CHD in either parent by age at time of first coronary event. 1) For men <55 years: RR=1.71, 95% CI ) For men 55 years: RR=1.50, 95% CI ) For men <55 years: RR=1.63, 95% CI ) For men 55 years: RR=1.45, 95% CI ) For women <55 years: RR=2.87, 95% CI ) For women 55 years: RR=1.49, 95% CI ) For women <55 years: RR=2.70, 95% CI ) For women 55 years: RR=1.46, 95% CI Positive parental history (<60 years) of CHD is associated with increased risk of non fatal MI or coronary death in both genders and that risk is independent of known major CHD risk factors and socio-economic status. The risk associated with positive paternal history of CHD was similar both in men and women but positive maternal history was associated with higher risk in women. as more time has elapsed since the parental event, it is possible that reporting may be less reliable in older than younger participants. However, positive family history was reported more often by younger than older participants the authors suggest that most of the difference between age groups can be explained by variation of frequency of CHD in the population however at least for women with maternal FH this seems unlikely participants who suffered from symptoms of angina pectoris themselves would have recalled a positive parental history better than those who did not have any symptoms of CHD.

20 12 Table 1. Evidence table of research studies relating to the value of a family history of premature cardiovascular disease in predicting increased risk of cardiovascular disease (continued) Authors Country Study design, setting, inclusion and exclusion criteria Sample characteristics Methods Results and authors conclusions Quality gradings and limitations Hassan et al. (2002) London, UK Case-control Study setting All acute stroke and transient ischaemic attack (TIA) referrals to a single cerebrovascular clinic serving an area in southeast London. Controls were recruited from the same geographical area from spouses of patients where available (n=222) and by community sampling (n=401). Inclusion criteria for cases white patients, with ischaemic stroke or TIA. Inclusion criteria for controls Selection of controls was performed to ensure that age (by decade), sex, and ethnicity were matching, though not matching of vascular risk factors. Participants Case patients: n=727 Control participants: n=623 Cases Sex: 60.2% male Mean age: years Diabetes: 13.4% Hypertension: 62.9% Controls Sex: 58.9% male Mean age: years Diabetes: 5.45% Hypertension: 9.5% Detailed family histories were obtained by interview. Participants were asked if parents and siblings were alive or dead, whether they had been affected by any stroke, and the age at first stroke in affected members. Data collection Community controls were invited randomly by letter using family practice lists covering the same geographical area as that covered by the hospital. Data analysis Differences between the groups were examined using the χ 2= test for proportions and the student t-test for mean age. Logistic regression analysis was used to determine the independent contribution of family history to ischaemic stroke risk. The relative odds ratio of sibling stroke associated with patient ischaemic stroke was also determined in a logistic regression model. For the risk of stroke (unadjusted), family history of any stroke predicted increased risk of ischaemic stroke (IS): Any age: OR= 1.27, 95% CI , p<0.05 Young IS: OR=1.29, 95% CI , p=0.07 Family history of any stroke ( 65 years) was a stronger predictor of all IS: Any age: OR= 1.63, 95% CI , <0.001 Young IS: OR=2.16, 95% CI , <0.001 For the risk of stroke (adjusted for age and other risk factors), family history remained a risk factor only for: Young IS: OR=1.59, 95% CI , p<0.01 Whereas the risk for family history of earlyonset stroke: Any age: OR= 1.47, 95% CI , p<0.05 Young IS: OR=2.25, 95% CI , <0.001 Sibling risk of stroke (unadjusted/adjusted) OR=2.12, 95% CI , p<0.01 OR=1.98, 95% CI , p<0.01 Quality gradings quality of design? +/Ø quality of results? + could applicability be determined? +/Ø are the findings applicable in usual practice? + Specific limitations it was not possible to reliably distinguish a family history of ischaemic and hemorrhagic stroke at interview, therefore a positive family history could include either possibility of recall bias as retrospective information on the participants family health history and other risk factors gathered by interview.

21 13 Table 1. Evidence table of research studies relating to the value of a family history of premature cardiovascular disease in predicting increased risk of cardiovascular disease (continued) Authors Country Study design, setting, inclusion and exclusion criteria Sample characteristics Methods Results and authors conclusions Quality gradings and limitations Hassan et al. (2002) London, UK (Continued) Exclusion criteria for cases As the study was specifically interested in polygenic multifactorial stroke, patients in whom a Mendelian stroke syndrome was suspected were excluded. Exclusion criteria for controls Any history of cerebrovascular disease. A detailed family history was unavailable in 42 patients and 12 control participants. Reasons included adopted parents (n=12), lost contact with relatives or no recall (n=40) and aphasia with no history available from other family members (n=2). Definitions Early-onset or young ischaemic stroke was defined as 65 years. In this model, a history of sibling stroke (yes/no) was entered, and the dependent variable and patient ischaemic stroke was included as a covariate. Other covariates included were age, sex, diabetes, hypertension and family size (number of siblings). A level of p<0.05 was taken to be significant for this part of the study. A preplanned analysis of families with young disease onset ( 65 years) was performed (n=367). Young sibling risk of stroke (unadjusted/adjusted) Young IS: OR=3.08, 95% CI , p<0.01 Young IS: OR=2.50, 95% CI , p<0.05 Models were adjusted for age, sex, sibship size, diabetes, hypertension and smoking. In all of the models, the likelihood of a sibling stroke increased with proband age. A family history of any stroke occurring at 65 years was an independent risk factor for stroke at all ages and for young stroke ( 65 years). Family history of premature stroke in siblings was associated with an increased risk of early onset stroke in this study. family history was not validated independently so misclassification of the exposure cannot be excluded.

22 14 Table 1. Evidence table of research studies relating to the value of a family history of premature cardiovascular disease in predicting increased risk of cardiovascular disease (continued) Authors Country Study design, setting, inclusion and exclusion criteria Sample characteristics Methods Results and authors conclusions Quality gradings and limitations Friedlander et al. (2001) Washington State, USA Case-control Study setting Participants were drawn from a study of incident cardiovascular disease among women years of age residing in three adjacent counties of Washington State. Inclusion criteria for cases women diagnosed with a first fatal or nonfatal MI between 1 July 1991 and 28 February eligible MI patients with definate or probable MI were identified, of whom 161 were living at the time they initiated recruitment. Of these 107 were recruited into the study, 4 were not approached at the request of their physicians and the remainder (n=40) refused to participate or could not be located (n=10). The response rate for cases was 66.5%, and 72.8% for controls (includes both the household screening and interview participation rates). Participants Case patients: n= 107 Control patients: n= 526 Control women were frequency matched on age, using the combined age distribution of all CVD patients recruited for the study. Cases Mean age: 39.5 ± 4.5 years Diabetes: 15.9% Hypertension: 31.8% Hypercholesterolemia: 40.2% Controls Mean age: 37.7 ± 5.3 years Diabetes: 2.7% Hypertension: 9.5% Hypercholesterolemia: 14.6% Data on each woman s family health history were collected from both case patients and controls by trained interviewers using a structured questionnaire. Criteria for MI were defined by evidence of symptoms, elevated enzymes, and electrocardiographic changes. Data collection Cases were identified through the review of hospital discharge diagnoses provided by all hospitals within the study region, incident reports from emergency medical service systems, and death certificates from CVD and related conditions. Random-digit telephone dialing was used to identify a control group of women years old living in the same area during the time period of the study. Data analysis For each participant, personyears accumulated by family members and the numbers of MI events within the family were counted. Person-years of relatives at risk were accumulated from birth until age at interview or age at death, or until age of event for relatives who experienced a MI. Risk of MI associated with family history (FH) Risk of MI (unadjusted) FH in fathers: OR= 1.49, 95% CI FH in mothers: OR= 2.21, 95% CI FH in parents: OR= 1.81, 95% CI FH in brothers: OR= 4.84, 95% CI FH in sisters: OR= 8.78, 95% CI FH in siblings: OR= 5.03, 95% CI FH in FDRs: OR= 2.15, 95% CI Risk of MI (adjusted for age) FH in fathers: OR= 1.21, 95% CI FH in mothers: OR= 1.63, 95% CI FH in parents: OR= 1.66, 95% CI FH in brothers: OR= 4.96, 95% CI FH in sisters: OR= 8.83, 95% CI FH in siblings: OR= 5.10, 95% CI FH in FDRs: OR= 2.12, 95% CI Risk of MI (adjusted for age and other risk factors) FH in fathers: OR= 0.91, 95% CI FH in mothers: OR= 1.22, 95% CI FH in parents: OR= 1.12, 95% CI FH in brothers: OR= 5.61, 95% CI FH in sisters: OR= 7.77, 95% CI FH in siblings: OR= 5.17, 95% CI FH in FDRs: OR= 1.52, 95% CI Quality gradings quality of design? +/Ø quality of results? + could applicability be determined? +/Ø are the findings applicable in usual practice? + Specific limitations possibility of recall bias as retrospective information on the participants family health history and other risk factors. However the authors did conduct a small-scale validation study of family history interview data (cases: n=59; controls: n=288) compared with family history ascertained through medical records. There was a moderately strong concordance between the two for both cases and controls (sensitivity, %; specificity, %).

23 15 Table 1. Evidence table of research studies relating to the value of a family history of premature cardiovascular disease in predicting increased risk of cardiovascular disease (continued) Authors Country Study design, setting, inclusion and exclusion criteria Sample characteristics Methods Results and authors conclusions Quality gradings and limitations Friedlander et al. (2001) Washington State, USA (Continued) Exclusion criteria for cases those who were dead by the time of recruitment. Note: This study appears to have recruited cases at the same time and from the same region as Friedlander et al (2002) however there does not appear to be a significant degree of crossover between the two samples, as this study includes women only. Other risk factors for MI were assessed and included age, race, education, diabetes, hypertension, or hypercholesterolemia, cigarette smoking, physical activity, coffee consumption and dietary fat intake. Information about FDRs (each biological parent, brother or sister) was obtained, including age at time of interview or age at time of death, occurrence of MI, and age of occurrence. Definitions Criteria for MI were defined by evidence of symptoms, elevated enzymes, and electrocardiographic changes. For each FDR (parents and siblings), specific incidence rates were calculated and the relative risks estimated by dividing the rate (history of MI among a relative per 1000 person-years) among the MI cases by the rate among the controls. Unconditional logistic regression analysis was used to assess the relationship of family history with the risk of MI, while adjusting for differences in familial personyears and for confounding and mediating factors. The authors examined whether family history of early MI was more strongly related to the early onset of MI in women. Since most events occurring among siblings are early onset events they conducted this sub analysis among parents only. The odds ratios for early parental MI (fathers before age 60 years; mothers before age 65 years) were similar to those presented above, which are for the total group. Family history of MI, particularly sibling history of MI (in this study, by definition premature), was associated with an increased risk of early onset of MI in women. the study only included MI survivors. Theoretically there is the possibility that associations seen in the study are due in part to early case fatality among the MI patients without family history, though this is unlikely data on family history was not collected for non-respondents. However, there were no significant differences found in the analyses performed on selected variables where there was data available from nonrespondents, and furthermore, adjustment for nonresponse made little difference in the associations.