Search

Background: Postmenopausal estrogen replacement therapy has been associated with favorable levels of cardiovascular disease risk factors, but these associations and the relations between estrogen use and subclinical disease have not been examined in large samples of older women., Methods and Results: Present and past estrogen use was ascertained in 2955 women > or = 65 years old in the Cardiovascular Health Study, a study of risk factors for coronary heart disease and stroke in the elderly. Present estrogen use was reported by 12% of these women and past use by an additional 26.5%. Estrogen use (past or present) was strongly associated with lower low-density lipoprotein cholesterol, fibrinogen, glucose, insulin, obesity, and age and higher high-density lipoprotein cholesterol and socioeconomic status (all P < .0001). Estrogen users also had lower levels of subclinical disease as measured by carotid intimal-medial thickness, carotid stenosis grade, ECG left ventricular mass, and Doppler mitral peak flow velocities (each P < .02). Relations were similar in younger and older women (65 to 74 versus > or = 75 years) and smokers and nonsmokers and were unchanged after women with poor medication compliance were excluded. After adjustment for other factors, estrogen use was associated with decreased carotid wall thickness, although this association was of borderline significance after further adjustment for lipids., Conclusions: Postmenopausal estrogen use in this sample of older women was associated with favorable cardiovascular disease risk factor profiles and with lower measures of subclinical disease. These findings suggest that postmenopausal estrogen use may be associated with lower risk of cardiovascular disease in women well into the eighth decade of life.

Objectives: To estimate the incidence of newly treated hypertension and to describe the patterns of antihypertensive medication use among those aged 65 years and older., Design: Medicare eligibility lists from four US communities (Forsyth County, North Carolina; Washington County, Maryland; Sacramento County, California; and Pittsburgh, Pa) were used to obtain a representative sample of 5201 community-dwelling elderly for the Cardiovascular Health Study, a prospective cohort study of risk factors for coronary heart disease and stroke. Participants were examined at baseline and again 1 year later. The two examinations included standardized questionnaires, blood pressure measurements, and the assessment of medication use by medication inventory. In this cohort analysis, we excluded 231 subjects (4.4%) who did not return for follow-up, 69 (1.3%) who had missing data for medications, and another 495 (9.5%) who were taking "antihypertensive" medications for an indication other than high blood pressure., Interventions: None., Results: Among the 4406 participants, 1613 used antihypertensive medications at both visits. Between the two visits, 144 started and 115 stopped antihypertensive therapy. Among nonusers at baseline, the annual incidence of newly treated hypertension was 5.2% in women and 5.6% in men. Due to the number of participants who stopped therapy, the overall prevalence of antihypertensive treatment increased only slightly, from 40.7% to 41.1% in women and from 37.1% to 38.2% in men, during 1 year of follow-up. After adjustment for age, systolic blood pressure, number of antihypertensive drugs, diabetes, and cardiovascular disease, the newly treated hypertensives were about half as likely as the previously treated hypertensives to receive diuretics (odds ratio [OR], 0.59; P = .008) or beta-blockers (OR, 0.52; P = .01); and they were about twice as likely to receive calcium channel blockers (OR, 1.88; P < .004) or angiotensin converting enzyme inhibitors (OR, 2.40; P < .001). A similar pattern of within-person changes over time was apparent among the continuous users., Conclusions: Between June 1990 and June 1991, physicians were increasingly prescribing angiotensin converting enzyme inhibitors and calcium channel blockers in place of diuretics and beta-blockers for the treatment of hypertension in elderly patients, especially for those just starting therapy.

Communities differ in the prevalence of various health behaviors, but it is not known to what extent these differences are due to "different types" of people living in them. We used data from the evaluation of the Henry J Kaiser Family Foundation Community Health Promotion Grant Program to study individual-level and community-level variation in health behaviors for 15 communities. Our results show (1) there was significant variation among these communities in prevalences of smoking, consumption of alcohol and dietary fat, and use of seatbelts; (2) these differences persisted after control for demographic, health status, and other health behavioral characteristics of the people in the communities; (3) the community effect on a particular person's behavior, as represented by R2, was very small (less than 1%); and (4) the adjusted differences in prevalences among communities were potentially large (for example, a 7 percentage point difference in the probability of smoking). Unique features of communities may influence health behaviors. These findings affirm the potential importance of contextual effects on individual health behavior and thus support the theory that changing the community environment may offer effective ways to change individual health behavior.

The Cardiovascular Health Study (CHS) is a longitudinal population-based study of coronary heart disease and stroke in men and women 65 years and older. The initial CHS cohort consisted of 5201 men and women recruited from a random sample of the Health Care Financing Administration (HCFA) Medicare eligibility lists in four communities in the United States. Extensive historical, physical, and laboratory evaluations were performed at the baseline examination in 1989 to 1990 to identify risk factors and subclinical disease. Periodic contacts are carried out to ascertain and verify incident cardiac and stroke events and their sequelae. Since only a short time has passed since entry of all the patients into the study, data are not available on time trends in the mortality rate of stroke, but we expect to contribute in this area in the years ahead. This article then is a description of the CHS, of methods of assessing stroke in the CHS cohort, and of prevalence of stroke and transient ischemic attacks at the initial examination.

Objective: Reducing serum cholesterol lowers the risk for ischemic heart disease, but its effects on other vascular diseases are unknown. Published trials were reviewed to determine the effect of cholesterol-lowering interventions on fatal and nonfatal stroke., Design: Meta-analysis of randomized, controlled trials., Data Identification: A literature search of English-language studies examining the effect of modified diets or medications on cardiovascular end points from 1965 to 1992 using MEDLINE and a review of references of five quantitative overviews of cholesterol reduction and coronary disease., Data Analysis: Thirteen studies met three eligibility criteria: patients randomized to intervention or control; fatal or nonfatal stroke reported separately; and end points assessed without knowledge of treatment status. Heterogeneity among studies and overall effects of treatment on fatal and nonfatal stroke were estimated using the Mantel-Haenszel-Peto method to combine independent study results. The influence of various study designs and interventions was explored using subgroup comparisons., Results: For fatal stroke, the overall odds ratio associated with cholesterol-lowering interventions in 13 trials was 1.32 (95% Cl, 0.94 to 1.86), and the odds ratio for the 10 single-intervention trials was 1.34 (Cl, 0.91 to 1.96). Among eight trials reporting nonfatal events, the summary odds ratio for nonfatal stroke for treated participants compared with controls was 0.88 (Cl, 0.70 to 1.11), and the odds ratio for total strokes was 0.98 (Cl, 0.80 to 1.19). Among three trials using clofibrate, treatment significantly increased the risk for fatal stroke (odds ratio, 2.64; Cl, 1.42 to 4.92) but not for nonfatal stroke (odds ratio, 0.87; Cl, 0.61 to 1.26). Regression analysis showed no statistical association between the magnitude of cholesterol reduction and the risk for fatal stroke., Conclusions: Lowering serum cholesterol through modified diets or medications does not reduce stroke mortality or morbidity in middle-aged men. Clofibrate appears to increase the risk for fatal strokes, but the mechanism for this effect is unknown.

The purpose of this article was to review, with special attention to the hypothesized mechanisms of atherosclerosis and thrombosis, the literature on the association of estrogens and progestins with cardiovascular disease. The data sources included recent reviews and their citations as well as literature searches of Medline. For coronary heart disease, we relied on a recent meta-analysis; for the lipid effects of estrogens and progestins, we refer to recent reviews and studies; for stroke, we identified all cohort and case-control studies; and for the effects of hormones on coagulation factors, we identified all relevant studies. The lipid effects of estrogens in postmenopausal women probably prevent atherosclerosis, and we would expect long duration of use rather than current use to provide the greatest benefit. Few epidemiologic studies have, however, assessed duration of estrogen use. High doses of estrogens are likely to be thrombogenic during current use, and it is possible that even moderate doses may increase the risk of clotting among women who smoke or who have existing coronary atherosclerosis. Compared with the lipid effects of estrogens alone, the lipid effects of combined therapy with progestins may increase atherosclerosis. The effect of progestins on coagulation factors is largely unknown, and no epidemiologic study has assessed the risk of cardiovascular disease associated with the use of combined hormone therapy in postmenopausal women. Cardiovascular risk or benefit associated with the use of postmenopausal hormones may involve several competing mechanisms, including effects on prostaglandins and vascular tone as well as atherosclerosis and thrombosis.

Objective: To determine whether making a low-fat entree available as one of two choices in an elementary school lunch program would reduce the fat content of meals selected by children., Design: In this before-after trial, students had a daily choice between two entrees, one of which was sometimes low-fat. For 93% of school days during 14 consecutive school months, daily entree choices were recorded for 619,976 student meals. The fat content of entrees was assessed with a computerized nutrient database supplemented by food manufacturer's data., Setting: Sixteen elementary schools in the Bellevue (Washington) School District., Participants: The number of students eating school lunch averaged 2440 per day, of whom 25% were less than 185% of poverty., Intervention: After a baseline period of 6 months, the intervention increased the number of days per month when one of the two entrees had 30% or fewer calories from fat., Main Outcome Measures: Data for the entire district were combined to estimate for each month the percent of students who selected low-fat entrees and the percent of calories from fat in the average meal selected by students., Results: During the 6 months before the intervention, a low-fat entree was available on 23% of days; it was selected by 39% of students; and the average meal selected by students had 36% of calories from fat. By the end of the 8-month intervention, a low-fat option was available on 71% of days; it was selected by 29% of students; and the fat content of the average meal dropped from 36% to 30% of calories from fat (P = .02)., Conclusions: In this school district, many students, given the choice, selected low-fat entrees. Recommendations for dietary fat were met simply by the environmental intervention of increasing the availability of low-fat foods.

Background: This article examines whether an in-store unobtrusive survey of grocery store product displays can be used to track community-level dietary behavior., Methods: The survey was conducted in 12 western communities two different times to measure two aspects of the grocery store environment: (a) the relative availability of low-fat and high-fiber products and (b) the amount of store-provided health-education information. Self-reported dietary intake of residents was obtained in the same 12 communities using a telephone survey. We compared the individual and store-level measures both cross-sectionally and over time., Results: We found positive and statistically significant correlations between the availability of healthful products in stores and the reported healthfulness of individual diets in cross-sectional analyses, but correlations between changes over time in the two measures were weaker and not statistically significant. The variance of the grocery store measures was nonetheless sufficiently small that a grocery store survey of 15 stores in each of 8 communities (n = 120 surveys) had power comparable to that of a telephone survey of 200 individuals/community (n = 1,600) surveys, at a fraction of the cost., Conclusion: Although the results provide further validation of cross-sectional measures of the grocery store environment, additional efforts are required to establish the validity of the grocery store survey as a method of measuring dietary change.

Objectives: To date, evaluations of community-based prevention programs have focused on assessing outcomes, not the process of organizing communities for health promotion. An approach was developed to analyze community organization efforts aimed at advancing community health objectives. These organizational processes are referred to as community activation., Methods: Information was gathered from 762 informants through a key informant survey conducted in 28 western communities. The data collected included informant ratings of community activation and information about interorganizational activities analyzed through network analytic techniques., Results: Activation levels, as measured by informant ratings, varied across communities. Program coordination, as measured by network analysis, occurred, on average, approximately 30% of the time. Higher income communities tended to be more activated than lower income communities., Conclusions: There is a widely recognized need for improved information about health-related community organization activities. It appears possible to gather such information through key informant surveys and to develop measures of community organization status that can be used in the evaluation of community health promotion programs.

A basic premise of community-based health promotion is that enduring changes in health behaviors are facilitated by changing community norms or the standards of acceptable behavior in the community. We examine whether community-level influences on individuals' normative attitudes can be related to cigarette smoking, alcohol consumption, and dietary fat intake. We conducted a random-digit dialing survey of 8,849 adults in 15 communities in the western United States as part of the evaluation of the Henry J. Kaiser Family Foundation Community Health Promotion Grant Program. We found independent associations among attitudes for tobacco, alcohol, and diet and both individual-level and community-level characteristics. A significant community effect on attitudes remained even when we controlled for individual demographics and health behaviors, as well as for the overall prevalence of the target health behavior in the community. The specific community characteristics that account for this effect were not clear in our analyses, suggesting that influential characteristics vary from community to community.

The Cardiovascular Health Study is a population-based longitudinal study of 5,201 adults aged 65 years and older. Prevalences of myocardial infarction, angina pectoris, congestive heart failure, peripheral artery disease, stroke, and transient ischemic attack were ascertained between June 1989 and May 1990 in participants recruited from Forsyth County, North Carolina; Washington County, Maryland; Sacramento County, California; and Pittsburgh, Pennsylvania. A medical history was taken to obtain self-reports of prevalent disease. For all participants, use of nitrates was ascertained to document angina, electrocardiograms were used to document prevalent myocardial infarction, and ankle-arm blood pressure studies were used to document peripheral artery disease. Self-reports of disease that were not confirmed by examination findings were further investigated by examination of medical records. Reported disease that was confirmed by examination findings or by medical records was classified as "definite." Disease that was documented by examination, but not reported by the participant, was classified as "unreported." The prevalence rates of definite myocardial infarction and angina were 11% and 15%, respectively, among men aged 65-69 years, 18% and 17% among men aged 80-84 years, 4% and 8% among women aged 65-69 years, and 3% and 13% among women aged 80-84 years. Twenty-three percent of men and 38% of women with electrocardiographic evidence of myocardial infarction did not report it. These results suggest that prevalent disease estimates based only on self-report may underestimate the prevalence of cardiovascular diseases in older Americans.

Objective: To assess the association between isolated systolic hypertension (ISH) and subclinical disease in adults aged 65 years and above., Design: Medicare eligibility lists were used to obtain a representative sample of 5201 community-dwelling elderly persons for the Cardiovascular Health Study, a National Heart, Lung, and Blood Institute--sponsored cohort study of risk factors for coronary heart disease and stroke. In this cross-sectional analysis of baseline data, we excluded 3012 participants who were receiving antihypertensive medications, had clinical cardiovascular disease, or had a diastolic blood pressure of at least 90 mm Hg., Main Outcome Measures: For electrocardiogram: myocardial infarction, left ventricular hypertrophy, and left ventricular mass as measures of myocardial damage and strain; for echocardiography: left ventricular mass, fractional shortening, and Doppler flow velocities as measures of cardiac systolic and diastolic function; and for carotid sonography: carotid arterial intima-media thickness as a measure of atherosclerosis., Results: Among the 2189 men and women in this analysis, 195 (9%) had ISH (systolic blood pressure, greater than or equal to 160 mm Hg) and 596 (23%) had borderline ISH (systolic blood pressure, 140 to 159 mm Hg). Systolic blood pressure was associated with myocardial infarction by electrocardiogram (P = .02). Borderline and definite ISH were strongly associated with left ventricular mass (P less than .001). While there was little association with cardiac systolic function, borderline and definite ISH were associated with cardiac diastolic function (P less than .001). Isolated systolic hypertension was also strongly associated with increased intima-media thickness of the carotid artery (P less than .001)., Conclusions: While cohort analyses of future repeated measures will provide a better assessment of risk, both borderline and definite ISH were strongly related to a variety of measures of subclinical disease in elderly men and women.

The Cardiovascular Health Study (CHS) was designed to examine cardiovascular disease and its risk factors in older adults. We report here the distributions of the coagulation factors fibrinogen, factor VII, and factor VIII in a population-based cohort of men and women 65 years or older. In other studies of middle-aged individuals, these factors were shown to be associated with cardiovascular risk. In the CHS cohort, all three factors were elevated, compared to levels reported in middle-aged individuals, and fibrinogen and factor VIII values were higher in each successive age group; factor VII values, in contrast, declined slightly with age in the CHS cohort. Compared to white subjects, blacks had higher values for fibrinogen and factor VIII and lower values for factor VII. While women had markedly higher values for factor VII and factor VIII than men at all ages in the CHS, mean fibrinogen values were not different between men and women.

Knowledge of the characteristics of survey non-respondents is important to determine generalizability to the population of interest. In a recent random-digit dialling survey of health behaviours only 73 per cent of the households contacted provided any information about household composition, and only 74 per cent of those actually completed the extended interview, for an overall response rate of 54 per cent. To identify possible biases we grouped all attempted phone numbers by their prefix, and looked for the association between the response rate for that prefix and other summary variables known about the prefix. A simulation study showed that the method can identify non-response biases if certain assumptions are correct. The analysis suggested that our survey data under-represent older people and those with a college education. We found no significant biases in health behaviours, possibly because the basic assumptions did not hold. This method may assist in identification of non-response bias in other studies.

We assessed the medical outcomes and costs associated with the pharmacologic treatment of patients with peripheral arterial disease (PAD) in a population-based historical cohort study of patients enrolled in a health maintenance organization. For up to 2 years, we compared 58 patients who used therapeutic amounts of pentoxifylline with a comparison group of 112 patients who received a minimal subefficacious trial of pentoxifylline. Medical records data were used to assess and control for the severity of PAD and other potentially confounding factors. Continuous use of a therapeutic amount of pentoxifylline during an initial 120-day period significantly reduced the incidence of PAD-related invasive therapeutic and diagnostic procedures in the first year of follow-up (adjusted relative risk, 0.35; 95% confidence interval, 0.12 to 0.99). However, there were no significant differences in the risk of a PAD-related hospitalization or cost of PAD-related care between continuous pentoxifylline users and the comparison group. Pentoxifylline therapy may reduce the risk of vascular surgery while not increasing the total cost of PAD care.

Most intrapopulation epidemiologic studies have been unable to find a significant association between sodium intake and blood pressure. The researchers have offered 2 opposing explanations: the hypothesis of the genetic susceptibility to sodium and the hypothesis of a weak association in general population. The evidence for and against each hypothesis is reviewed. The direction of future research is suggested.

The Cardiovascular Health Study (CHS), a cohort study of risk factors for coronary heart disease and stroke, recruited 5201 community-dwelling adults aged 65 years or older. To assess the prevalence of medication use at baseline, we used the method of medication inventory and transcribed information about drug names and doses from prescription bottles. Using a specially-written computer program, persons without a knowledge of drug nomenclature coded 10,511 (89%) of the 11,846 medicines entered. We compared the results of the medication inventory and answers to questions on specific medications for reliability and validity. The use of beta-blockers and beta-agonists assessed by the method of medication inventory, but not by the method of directed recall, was associated with a significant effect on mean heart rate. Among 5197 participants with medication data, 76.1% were taking at least one medicine, and the mean number of drugs per person was 2.28. Among those with a reported history of high blood pressure, participants with cardiovascular disease (CVD) were more likely to be treated, and they were more likely to be taking beta-blockers and calcium-channel blockers than those without CVD. Daily aspirin use was also more common among those with CVD (30.5% of women and 43.2% of men) than among those without CVD (14.0% of women and 14.0% of men). The prevalence of post-menopausal estrogen use differed significantly among the four clinical centers (range = 5.5%-22.5% of women). We conclude that this method of assessing medications was easy to use and provided estimates of exposure to drugs that may affect risk of cardiovascular disease.

Electrocardiographic abnormalities are often found in older patients, but their prevalence in free-living elderly populations is not well-defined. In addition, the clinical significance of many of these abnormalities is uncertain. The prevalence of major electrocardiographic abnormalities was determined in 5,150 adults aged greater than or equal to 65 years from the Cardiovascular Health Study--a study of risk factors for stroke and coronary heart disease in the elderly. Ventricular conduction defects, major Q/QS waves, left ventricular hypertrophy, isolated major ST-T-wave abnormalities, atrial fibrillation and first-degree atrioventricular block were collectively categorized as major electrocardiographic abnormalities. Prevalence of any major electrocardiographic abnormality was 29% in the entire cohort, 19% among 2,413 participants who reported no history of coronary artery disease or systemic hypertension, and 37% among 2,737 participants with a history of coronary artery disease or hypertension. Prevalence of major electrocardiographic abnormalities was higher in men than in women regardless of history, and tended to increase with age. Major Q/QS waves were found in 5.2%, and more than half were in those who did not report a previous myocardial infarction. Major electrocardiographic abnormalities are common in elderly men and women irrespective of the history of heart disease.

To examine the association between the level of treated blood pressure and the incidence of myocardial infarction, we conducted a population-based case-control study of 912 members of a health maintenance organization who were receiving standard clinical treatment for hypertension. We found a J-shaped relationship between the most recently measured diastolic blood pressure and the risk of myocardial infarction, the lowest risk occurring at 84 mm Hg. The relative risk of myocardial infarction at 60 mm Hg was 2.07 (95% confidence interval, 0.86 to 5.01), and at 100 mm Hg, 1.45 (95% confidence interval, 1.02 to 2.06). Treated systolic pressure bore a linear relationship to the risk of infarction. We conclude that the optimum target range for diastolic blood pressure in hypertensive patients may be 84 to 90 mm Hg. Levels outside this range may be associated with increased risk of myocardial infarction.

Community-based approaches to health promotion and disease prevention are often based on underlying theoretical models of behavior change, but evaluations of these programs have not attempted to assess empirically the validity of these models. This article uses data from the Henry J. Kaiser Family Foundation Community Health Promotion Grants Program (CHPGP) evaluation to examine a conceptual model which guided the program evaluation. The model posits relationships among six theoretical constructs: community activation, program activity, norms, environment, role models, and health behaviors. Correlational analyses using two waves of survey data were used to examine whether the hypothesized relationships were actually present. Positive cross-sectional associations between the constructs at each point in time were found but the associations between changes in the constructs over time were weak. The stability of these cross-sectional patterns suggests that there may be forces leading communities to have an underlying "consistency" with respect to community norms, the environment, health behaviors and perhaps other community characteristics. If so, effective community-based programs may need to target a wide range of community characteristics rather than to focus narrowly on a single dimension.

The current paradigm in medicine generally distinguishes between genetic and environmental causes of disease. Although the word "paradigm" has become a commonplace, the theories of Thomas Kuhn have not received much attention in the journals of medicine. Kuhn's structuralist method differs radically from the daily activities of the scientific method itself. Using linguistic theory, this essay offers a structuralist reading of Thomas Kuhn's The Structure of Scientific Revolutions. Our purpose is to highlight the similarities between these structuralist models of science and language. In part, we focus on the logic that enables Kuhn to assert the priority of perception over interpretation in the history of science. To illustrate some of these issues, we refer to the distinction between environmental and genetic causes of disease. While the activity of scientific research results in the revision of concepts in science, the production of significant differences that shape our knowledge is in part a social and linguistic process.

Carotid sonography is being performed on more than 5,000 participants in the Cardiovascular Health Study, a prospective, multicenter study of cardiovascular disease in men and women aged 65 years and older. The sonographic methods used to examine and measure the extracranial carotid arteries are described. Initial validation studies were performed on 61 subjects with a mean age of 68.6 years. Analysis of within- and between-sonographer differences and between-reader differences were performed for selected variables. In general, the mean absolute differences for within- and between-sonographer comparisons were small, with even less variability between readers. Variability was less for the common carotid artery than for the internal carotid artery. These data suggest that carotid sonography is a reliable and reproducible method for use in the study of carotid atherosclerosis in population studies.

We conducted a controlled trial of two interventions to increase the number of patients receiving alcohol counseling in an academic, general medicine clinic. Ongoing randomization yielded similar groups of patients and providers. By applying a two-item alcoholism-screening instrument incorporated into the routine intake process, nursing staff were successful in screening 90.4% of the 1,328 eligible patients seen during a 4-month study period. Of the patients screened, 428 (35.6%) were found to be positive. In those groups where nurses were allowed to refer directly on the basis of a positive screening result, over four times as many patients (10.8% versus. 2.3%) accepted appointments for counseling, as did patients seen by physicians providing standard care. Patients who were briefly introduced to the alcohol counselor at the time referral were no more likely to keep their first, formal appointments than were patients scheduled in the usual manner. Patients who did receive counseling had their diagnoses confirmed by the 25-item Michigan Alcoholism Screening Test. We conclude that the adoption of an alcohol screening program in our clinic increased the number of patients appropriately referred for counseling.

A study of health behaviors in four communities in the western United States in 1988 provided the opportunity to compare two methods of sampling elderly respondents for a telephone interview. The Polk telemarketing lists were used to identify 1,407 respondents aged 65 years and older in four communities, where 253 respondents in the same age group were also identified by the method of random digit dialing. Individuals identified from the Polk lists received a letter prior to the initial telephone contact. The overall response rate was 49.3% for random digit dialing and 57.3% for the Polk lists. On the average, the identification of one elderly respondent using the Polk lists required about 20-25% as much interviewer time per subject identified as was required by the random digit dialing method. The elderly identified by the Polk lists were significantly older than those identified by random digit dialing, and the proportions of the Polk sample who were married, white, or had an income of greater than $10,000 were slightly higher than those of the random digit dialing sample. Among 40 variables measuring various health behaviors, indicators of health status, and participation in health-related programs and classes, only three differed significantly between the two samples. The authors conclude that sampling from commercial telemarketing lists was an efficient method of identifying elderly respondents and that in these four communities, the estimates of health behaviors and health status were comparable with those obtained by random digit dialing techniques.

In this paper we discuss approaches to two distinct problems in using large computerized databases to conduct population-based case-control studies. The first topic, concerning case ascertainment, is a methodologic problem, and the second, concerning confounding by indication, is an analytic problem. The first involves attempting to ascertain all incident cases of coronary disease among enrollees of Group Health Cooperative of Puget Sound (GHC), a large health-maintenance organization. Methodologic studies reported here have helped us improve the efficiency of using the large computerized databases for case ascertainment at GHC. The second problem involves the issue of confounding by drug indication. Drugs such as beta-blockers have multiple indications, including the treatment of both high blood pressure and angina. These two indications may make it difficult in an observational study to determine whether beta-blockers may prevent coronary disease in patients with high blood pressure. We discuss here our current thinking about the best analytic approach to this problem.

Background: This article examines the relationship at the community level between individual dietary practice and the grocery store environment., Methods: Individual dietary practice was measured in 12 communities using a telephone survey to obtain self-reported diet. A protocol was developed to measure two aspects of the grocery store environment in these same 12 communities: the relative availability of healthful (low-fat and high-fiber) products, and the amount of health-education information provided. Comparisons were made between individual and store-level measures at two levels of geographic aggregation: community (typically a county) and zip code within community (n = 34)., Results: We found positive and statistically significant correlations at both the community and the zip code level between the availability of healthful products in stores and the reported healthfulness of individual diets. Positive correlations were also found between measures of the amount of health-education material provided by stores and the healthfulness of individual diets, but these correlations did not reach statistical significance., Conclusions: The results provide support for including measures of the grocery store environment as part of a community-level assessment of dietary behavior.

The Kaiser Family Foundation's Community Health Promotion Grant Program (CHPGP) provides funding and technical assistance in support of community-based efforts to prevent major health problems. The first phase of the program was implemented in 11 communities in the western United States. This paper describes the evaluation design of the CHPGP in the West, the methods of data collection, and the baseline comparability of intervention and control communities. Major features of the evaluation design include: (1) the randomization of qualified communities making application into funded and unfunded comparison groups; (2) a second set of matched control communities for some intervention sites; (3) data gathering through repeated surveys of community residents (probability samples of adults and adolescents) and institutions (health-related organizations and randomly sampled grocery stores and restaurants); and (4) the use of secondary data to monitor health events. Selected baseline data show that intervention and control communities differ in racial/ethnic composition, but relevant health behaviors and ratings of community activation for health promotion appear comparable.

The growing interest in community-based approaches to health promotion and disease prevention (HP/DP) has been accompanied by a growing need to evaluate the effectiveness of such programs. Special issues that arise in these evaluation studies include (1) entire communities are assigned to intervention and control groups, (2) only a small number of communities can usually be studied, (3) the time course of changes in behavior and other outcomes is often of interest, and (4) surveys to measure such changes over time can be conducted with either repeated cross-sectional samples or with longitudinal samples. This paper shows how these issues can be addressed under a mixed-model analysis of variance approach. This approach serves to unify several ideas in the literature on evaluation of community studies, including use of time-series regression and the question of whether the individual or the community should be the unit of analysis. We also describe how the method can be used to estimate sample size requirements, statistical power, or minimum detectable program effect.

A population-based, case-control study was conducted to determine whether beta blockers, used for the treatment of high blood pressure, prevent first events of coronary heart disease. All study subjects were health-maintenance organization enrollees with pharmacologically treated hypertension. Patients presented in 1982 to 1984 with new coronary heart disease, and control subjects were a probability sample of eligible hypertensive enrollees free of coronary heart disease. With the investigators blind to case-control status, the subjects' medical records were reviewed for other coronary risk factors, and the health-maintenance organization's computerized pharmacy database was used to ascertain the use of beta blockers. A larger proportion of controls than cases were using beta blockers. This difference was confined to the subgroup with nonfatal myocardial infarctions. For current use, the estimated relative risk for nonfatal myocardial infarction was 0.62 (95% confidence interval, 0.39 to 0.99). Among current users of beta blockers, higher doses conferred greater protection. Past use and total lifetime intake of beta blockers were only weakly associated with case-control status. The current use of beta blockers may prevent first events of nonfatal myocardial infarction in patients with high blood pressure.

A pilot test of a survey of grocery store product displays was conducted to measure the amount of health-education information provided and the proportion of the display devoted to "healthier" products. Inter-rater reliability ranged between 0.73 and 0.78 for the healthiness indices and between 0.30 and 0.67 for the health education measures. Test-retest reliability ranged from 0.44 to 1.0.

We conducted a population-based, case-control study of risk factors for first events of coronary heart disease in patients with high blood pressure. All subjects had hypertension treated with medication. The 248 cases presented with new coronary heart disease from 1982 through 1984, and the 737 controls were a probability sample of health maintenance organization patients free of coronary heart disease. The health maintenance organization's computerized pharmacy database identified recent stoppers--patients who did not fill their prescriptions regularly enough to be at least 80% compliant. After adjustment for potential confounding factors, subjects who had recently stopped using beta-blockers had a transient fourfold increase in the relative risk of coronary heart disease (relative risk, 4.5; 95% confidence interval, 1.1 to 18.5). The association was specific to beta-blockers but not diuretics. A withdrawal syndrome immediately following the cessation of beta-blocker use may be an acute precipitant of angina and myocardial infarction in hypertensive patients who have no prior history of coronary heart disease.

This article reviews the nature of the effects of hypertension, smoking and cholesterol on the incidence of cardiovascular disease and emphasizes how these effects vary by age. In the Methods section, we discuss briefly the concepts of additive and multiplicative statistical models as tools for summarizing data. In the results section, we summarize available data on the association between incident stroke and coronary heart disease in the elderly and each of these major risk factors. The traditional multiplicative model parsimoniously characterizes the individual and joint effects of age and high blood pressure in terms of risk ratios; but, for smoking and cholesterol, an additive model appears to be the most parsimonious. We discuss the consequences of these observations for the study and prevention of cardiovascular disease in the elderly.

To evaluate the performance of serum iron studies as a diagnostic test for iron-deficiency anemia in a county hospital, the authors identified retrospectively all general medicine patients who had had bone-marrow aspirates for the work-up of non-macrocytic anemias from 1978 through 1983. Re-reading a sample of aspirates from the 254 study patients (42 with iron deficiency) verified the presence or absence of iron. Analysis with logistic regression, likelihood ratios, and receiver operating characteristic curves demonstrated that the total iron-binding capacity (TIBC) performed markedly better as a diagnostic test than did the transferrin saturation test. While no single TIBC level was diagnostic, the TIBC provided a good estimate of the probability of iron-deficiency anemia.

To develop a decision rule to aid in the diagnosis of myocardial infarction, we evaluated clinical and ECG data on 540 adults treated in an urban hospital emergency room for acute chest pain. Of 62 (11.5%) patients who had acute infarctions, 54 were admitted to intensive care (sensitivity 87%); 103 of 478 patients without infarctions were also admitted to intensive care (specificity 78%). Thirty-four percent of all patients admitted had infarctions. Multivariate analysis identified only four clinical variables which carried independent information predicting infarction: two from the ECG and two from the clinical history. A predictive model based on these four variables had significantly greater specificity (86% vs. 78%, p = .003) and accuracy of overall patient classification (88% vs. 79%, p = .013) but somewhat lower sensitivity (81% vs. 87%, p = .46) than physician judgments. However, a decision rule which would have admitted to intensive care those patients with a high probability of infarction who were not admitted by the emergency room physicians, would have increased the sensitivity for detecting infarction to 95% with no appreciable decrease in specificity or yield of infarctions among patients admitted to intensive care.

The authors conducted a population-based case--control study to determine the risk of myocardial infarction in patients who reported angina-like symptoms. The cases studied were those of patients who had high blood pressure and had sought treatment in 1984 with myocardial infarction as the first manifestation of coronary artery disease. Controls, a random sample of patients who had hypertension, were frequency-matched to cases by age and gender. Blind to case--control status, the authors reviewed the medical records of the 32 cases and 64 controls for reports of angina-like symptoms. While controls reported such symptoms at a constant rate, the events for the cases clustered near their infarctions. When a patient with hypertension sought medical advice for angina-like symptoms, the risk of infarction within 30 days was 14.2 (95% confidence interval, 2.8 to 71), and after 30 days it fell to 1.03. Among patients who have high blood pressure but no history of angina, presentations with prodromal symptoms in the primary care setting are so common that only about one in 100 such visits actually heralds myocardial infarction.

We conducted a population-based, case-control study to determine whether beta-blockers, used for the treatment of hypertension, prevent first events of coronary heart disease. Cases were patients who had high blood pressure treated with medicines and who presented in 1982 to 1984 with angina or fatal or nonfatal myocardial infarction. Controls were a probability sample of health maintenance organization patients with pharmacologically treated hypertension and free of coronary heart disease. Blinded to case-control status, we reviewed the medical records of the 248 cases and 737 controls. The health maintenance organization's computerized pharmacy database was used to ascertain the use of beta-blockers. Fewer cases than controls were taking beta-blockers. This difference was confined to those with nonfatal infarctions. After adjustment for confounding, the estimated relative risk was 0.62 (95% confidence interval, 0.39 to 0.99). Higher doses of beta-blockers conferred greater protection. We conclude that beta-blockers may prevent first events of nonfatal myocardial infarction in patients with high blood pressure.

The validity and reliability of auditory screening tests were evaluated in 34 demented and 31 non-demented elderly outpatients. In reference to an audiometric gold standard (40-dB HL hearing loss in speech frequencies), 512-Hz and 1024-Hz tuning forks, finger rub, and whispered voice tests performed well (ROC curve areas = 0.82 to 0.94). Simultaneously high (greater than 0.80) sensitivities and specificities were achievable for all these tests in demented patients. In non-demented patients, however, only the whispered voice test achieved simultaneously high specificity and sensitivity. The most accurate rule for air conduction screening audiometry was the inability to hear greater than or equal to two of four 40-dB HL speech frequencies (sensitivity = 1.0, specificity = 0.75 in non-demented patients; sensitivity = 0.97, specificity = 0.74 in demented patients). Interobserver/test-retest reliability was generally high for tuning forks, finger rub, and whispered voice tests (range of intraclass correlation coefficients = 0.38 to 0.90), and was somewhat higher in demented than in non-demented patients. These results suggest that some of the simple, traditional methods of auditory screening may have considerable validity and reliability in demented and non-demented older adults.

To evaluate the ability of emergency room physicians to estimate the probability of myocardial infarction in patients with acute chest pain, the authors gathered historical, physical, and electrocardiographic information from 492 patients at the time of their presentation. The physicians admitted 30% of them to intensive care: 53 of the 61 patients with infarctions (sensitivity = 87%) and 96 of the 431 without infarctions (specificity = 78%). Overall, 36% of those admitted had infarctions. The physicians' numeric estimate of the probability of infarction was a good univariate discriminator of infarction, as demonstrated by Receiver Operator Characteristics analysis, and, as indicated by their actual operating point, they seemed to maximize the accuracy of patient classification rather than sensitivity or specificity. Logistic regression analysis identified the physicians' probability estimate as the strongest multivariate predictor of infarction, considering all other clinical information available.