scispace - formally typeset
Search or ask a question

Showing papers by "Ralph B. D'Agostino published in 1995"


Journal ArticleDOI
TL;DR: High plasma homocysteine concentrations and low concentrations of folate and vitamin B6, through their role in homocy steine metabolism, are associated with an increased risk of extracranial carotid-artery stenosis in the elderly.
Abstract: Background Epidemiologic studies have identified hyperhomocysteinemia as a possible risk factor for atherosclerosis. We determined the risk of carotid-artery atherosclerosis in relation to both plasma homocysteine concentrations and nutritional determinants of hyperhomocysteinemia. Methods We performed a cross-sectional study of 1041 elderly subjects (418 men and 623 women; age range, 67 to 96 years) from the Framingham Heart Study. We examined the relation between the maximal degree of stenosis of the extracranial carotid arteries (as assessed by ultrasonography) and plasma homocysteine concentrations, as well as plasma concentrations and intakes of vitamins involved in homocysteine metabolism, including folate, vitamin B12, and vitamin B6. The subjects were classified into two categories according to the findings in the more diseased of the two carotid vessels: stenosis of 0 to 24 percent and stenosis of 25 to 100 percent. Results The prevalence of carotid stenosis of >25 percent was 43 percent in the m...

1,201 citations


Journal ArticleDOI
TL;DR: After multivariable adjustment, LA enlargement remained a significant predictor of stroke in men and death in both sexes and appears to be partially mediated by LV mass.
Abstract: Background The medical literature contains conflicting reports on the association of left atrial (LA) enlargement with risk of stroke. The relation of LA size to risk of stroke and death in the general population remains largely unexplored. Methods and Results Subjects 50 years of age and older from the Framingham Heart Study were studied to assess the relations between echocardiographic LA size and risk of stroke and death. During 8 years of follow-up, 64 of 1371 (4.7%) men and 73 of 1728 (4.2%) women sustained a stroke, and 296 (21.6%) men and 271 (15.7%) women died. Sex-specific Cox proportional-hazards models were adjusted for age, hypertension, diabetes, atrial fibrillation, smoking, ECG left ventricular (LV) hypertrophy, and congestive heart failure or myocardial infarction. After multivariable adjustment, for every 10-mm increase in LA size, the relative risk of stroke was 2.4 in men (95% CI, 1.6 to 3.7) and 1.4 in women (95% CI, 0.9 to 2.1); the relative risk of death was 1.3 in men (95% CI, 1.0 t...

961 citations


Journal ArticleDOI
TL;DR: Findings support previous contentions that a "preclinical phase" of detectable cognitive deficits can precede the clinical diagnosis of probable AD by many years, and support the hypothesis that problems with secondary verbal memory are among the first signs of AD.
Abstract: Objective: To evaluate the interval between the onset of detectable cognitive impairment and clinical diagnosis in individuals with probable Alzheimer's disease (AD), and to identify the pattern of the earliest changes in cognition in probable AD. Design: Longitudinal follow-up of a community-based cohort sample. In 1976 through 1978, a screening neuropsychological examination was administered to Framingham Study participants. These subjects were then followed up prospectively for development of probable AD for up to 13 years. Setting: This study was conducted at a communitybased center for epidemiologic research. Participants: The surveillance sample consisted of 1045 participants in the Framingham Study aged 65 to 88 years who were free of dementia at the time of the neuropsychological screening examination. Main Outcome Measures: Scores on a group of neuropsychological tests were entered into a series of age-and education-adjusted multiple regression procedures, with the presence or absence of probable AD as the outcome variable. Results: Considered individually, most of the screening neuropsychological measures were significantly related to later AD diagnosis. When stepwise regression procedures were employed, only measures of verbal memory and immediate auditory attention span remained significantly related to AD diagnosis. Of note, subjects later diagnosed with probable AD performed at higher levels than normal subjects on the Digit Span test at initial screening. Regression results were essentially unchanged even when the AD sample was restricted to those individuals for whom the screening examination preceded the clinical onset of dementia by 7 years or more. Conclusions: These findings support previous contentions that a "preclinical phase" of detectable cognitive deficits can precede the clinical diagnosis of probable AD by many years, and they also support the hypothesis that problems with secondary verbal memory are among the first signs of AD.

467 citations


Journal ArticleDOI
01 Aug 1995-Stroke
TL;DR: Temporal patterns of stroke onset were observed for season, day of the week, time of day, and place in a community-based population, suggesting that there are periods of increased risk of stroke that may be amenable to preventive strategies.
Abstract: Background and Purpose Several studies have shown stroke onset to vary by season, day of the week, and time of day. These temporal patterns, which may provide insights into pathogenesis, were found mainly in clinical series, which can be subject to selection bias. To obtain a less distorted picture of stroke onset, we examined the month and season, day of the week, time, and place stroke occurred in a community-based cohort. Methods Over a 40-year period of surveillance of the Framingham Study cohort of 5070 people aged 30 to 62 years and free of stroke and cardiovascular disease at entry, 637 completed initial strokes occurred. Month, season, day of the week, time of day, and place of occurrence of stroke were ascertained systematically and related prospectively to stroke incidence, subtype, and gender. Results Winter was the peak season for cerebral embolic strokes. Significantly more stroke events occurred on Mondays than any other day, particularly for working men. For intracerebral hemorrhages, a thi...

229 citations


Journal ArticleDOI
TL;DR: Low educational attainment was associated with increased risk of non-AD dementia, perhaps because of deleterious smoking habits and other risk factors for stroke in the least-educated individuals.
Abstract: Objective: To evaluate whether low educational attainment is a risk factor for the incidence of dementia and Alzheimer9s disease (AD) in the Framingham Study and to determine whether age at onset of dementia is earlier in persons with low educational levels. Design: A community-based cohort was studied longitudinally for the development of dementia. Diagnosis was made according to strict criteria by two neurologists and a neuropsychologist. Subtype of dementia and year at onset were determined. Incidence rates were compared in three education groups: less than grade school, less than high school, and more than equals high school. Participants: A total of 3,330 men and women aged 55 to 88 years. Results: During 17 years of follow-up, 258 incident cases of dementia, including 149 AD cases, were identified. Unadjusted incidence rates were significantly elevated (p less than 0.05) for dementia and non-AD dementia among the least educated. The age-adjusted relative risk for subjects with a grade school education or less compared with those who earned a high school diploma was 1.31 (95% confidence interval [CI], 0.90 to 1.90) for dementia generally, 1.04 (95% CI, 0.62 to 1.74) for AD, and 1.75 (95% CI, 1.03 to 2.98) for non-AD dementia. Age at onset of dementia did not vary by educational attainment. Conclusions: After age adjustment, low educational attainment was not a significant risk factor for the incidence of dementia generally or of AD. Low educational attainment was associated with increased risk of non-AD dementia, perhaps because of deleterious smoking habits and other risk factors for stroke in the least-educated individuals. Adequately adjusting for age and examining subtypes of dementia are important in assessing the influence of education on dementia incidence. NEUROLOGY 1995;45: 1707-1712

229 citations


Journal ArticleDOI
01 Sep 1995-Stroke
TL;DR: Ninety-two percent of subjects presenting with newly discovered AF at the time of acute stroke continued to have this rhythm disturbance in a chronic or paroxysmal form, suggesting that in most instances AF was probably the precipitant rather than the consequence of stroke.
Abstract: Background and Purpose When atrial fibrillation (AF) is first documented at the time of onset of acute stroke, it is difficult to establish a temporal relationship between AF and stroke. Did AF precede and precipitate the stroke, or did the arrhythmia appear as a result of stroke? Following the course of the newly diagnosed AF may help to clarify this relationship. Methods The Framingham Study cohort of 5070 members, aged 30 to 62 years and free of cardiovascular disease at entry, has been under surveillance for the development of cardiovascular disease, including stroke. We followed the course of AF, which was documented for the first time on or soon after hospital admission for stroke. Results During 38 years of follow-up, 115 of 656 initial stroke events occurred in association with AF: 89 had previously documented AF, 21 had AF discovered for the first time on admission for the stroke, and 5 were admitted with sinus rhythm but developed AF after admission. Of the 21 subjects with AF diagnosed on admis...

166 citations


Journal Article
TL;DR: This measure, developed as part of Type II Diabetes Patient Outcomes Research Team project, uses patients' reports of symptoms and conditions, as well as patients' ratings of symptom intensity to characterize total disease burden, and differs from other measures of case mix in lack of dependence on diagnoses.
Abstract: Case mix has been shown to be of critical importance in studies of effectiveness and quality of care using health outcomes. How these variables are defined, combined, and used to adjust or increase precision in tests for differences in health outcomes has been a source of controversy. Because existing measures were developed to adjust mortality and have marginal relevance for the adjustment of functional status outcomes, especially in ambulatory settings, the authors developed a measure of case (or patient) mix that is specifically designed to adjust functional status outcomes measured in office practice or out-of-hospital settings. This measure, developed as part of Type II Diabetes Patient Outcomes Research Team project, uses patients' reports of symptoms and conditions, as well as patients' ratings of symptom intensity to characterize total disease burden. It differs from other measures of case mix in lack of dependence on diagnoses. Separate measures were developed for each of 15 different disease categories (e.g., chronic lung disease) grouped by body system affected. Within each measure, questionnaire items were combined to rate the severity of that disease on a 1 to 4 scale, according to definitions provided by clinicians. A single, global measure was developed by aggregating the 15 measures, weighted according to the expected impact of each disease category on functional outcomes and disability. In a sample of 1,738 patients, significant relationships were observed between the global case mix measure and functional status, disability days, and service utilization.(ABSTRACT TRUNCATED AT 250 WORDS)

120 citations


Journal Article
TL;DR: Logistic regression, classification tree, and neural network models all can provide excellent predictive performance of medical outcomes for clinical decision aids and policy models, but their ultimate limitations seem due to the availability of the information in data rather than their respective intrinsic properties.
Abstract: Background There is increasing interest in mathematical methods for the prediction of medical outcomes. Three methods have attracted particular attention: logistic regression, classification trees (such as ID3 and CART), and neural networks. To compare their relative performance, we used a large clinical database to develop and compare models using these methods. Methods Each modeling method was used to generate predictive instruments for acute cardiac ischemia (which includes acute myocardial infarction and unstable angina pectoris), using prospectivel-collected clinical data on 5773 patients, who presented over a two year period to six hospitals' emergency departments with chest pain or symptoms suggesting acute ischemia. This data set was then split into training (n = 3453) and test (n = 2320) sets. Of 200 available variables, modeling was restricted to those available within the first 10 minutes of emergency department care (history, physical exam, and electrocardiogram). Results When the number of variables was limited to eight, representing a practical number for input in the real-time clinical setting, the logistic regression's receiver-operating characteristic (ROC) curve area, as a measure of diagnostic performance, was 0.887; the classification tree model's ROC curve area was 0.858, and the neural network's ROC curve area was 0.902. When the number of variables used by a model was not limited, the logistic regression's ROC area was 0.905, the classification tree model's 0.861, and the neural network's 0.923. Among these models the neural networks had noticeably poorer calibration. When the outputs from each of these unrestricted models were presented to each of the other methods as an additional independent variable, the ROC areas of the new "hybrid" models were not significantly better than the original unlimited models (ROC areas 0.858 to 0.920). Conclusions Logistic regression, classification tree, and neural network models all can provide excellent predictive performance of medical outcomes for clinical decision aids and policy models. Their ultimate limitations seem due to the availability of the information in data (a "data barrier") rather than their respective intrinsic properties. Choices between these methods would seem to be most appropriately based on the needs of the specific application, rather than on the premise that any one of these methods is intrinsically more powerful.

107 citations


Journal ArticleDOI
TL;DR: The Age x Blood Pressure model as it pertains to older adults was not supported, but independent associations (with all covariables controlled) between the indices of blood pressure and cognitive functioning were statistically significant.
Abstract: Interactions of three indices of blood pressure (systolic blood pressure, diastolic blood pressure, and chronicity of hypertension) and age-cohort membership were examined for a sample of 1,695 stroke-free participants of the Framingham Heart Study, ages 55-88 years. Blood pressure level and chronicity of hypertension were assessed over five biennial examinations performed between 1956 and 1964, a time when few hypertensives were being treated, and were related to neuropsychological tests administered between 1976 and 1978. Multiple linear regression methods were used to examine Age x Blood Pressure (or Chronicity of Hypertension) interactions in alternative analyses involving three age groups (55-64 years, 65-74 years, and 75-88 years) and age as a continuously distributed variable (age in years). Interactions were either statistically nonsignificant or trivial with respect to magnitude of effect. This was true when interaction terms (Age x Blood Pressure Level or Age x Chronicity of Hypertension) were controlled for blood pressure, age, education, occupation, cigarette smoking, alcohol consumption, gender, and antihypertensive treatment. The Age x Blood Pressure model as it pertains to older adults was not supported, but independent associations (with all covariables controlled) between the indices of blood pressure and cognitive functioning were statistically significant.

105 citations


Journal Article
TL;DR: Various types of nonrandomized studies to measure effectiveness when a randomized control group is not available are reviewed, along with their advantages and disadvantages.
Abstract: Randomized controlled trials or studies are often considered the ideal way to evaluate the effectiveness of a treatment compared to a control. In such a study, the randomization procedure ensures that the subjects receiving the treatment and control are equal with respect to all conditions except for receiving the treatment or the control. Differences found by statistical comparisons of the results of such a study can be attributed to the effect of the treatment or how much the treatment differs from the control when all other things are held constant. Randomized controlled trials are not always possible, and even when possible they are often performed with such restrictions that they do not provide the true measure of the effectiveness of the treatment in the "real world" or under "conditions of usual practice." This article reviews the use of nonrandomized studies to measure effectiveness when a randomized control group is not available. Various types of nonrandomized studies are reviewed, along with their advantages and disadvantages. Often, these studies require statistical adjustments such as matching or covariance analysis to adjust for inequalities or to remove biases between the treatment and control groups; these are reviewed as well.

102 citations


Journal ArticleDOI
TL;DR: Blood pressures and chronicity of hypertension were inversely associated with performance on visual and verbal memory tests of the Kaplan-Albert battery and the odds of performing poorly were higher for age than for the blood pressure variables.
Abstract: The relationships of three blood pressure variables (systolic and diastolic pressures and chronicity of hypertension) and age to cognitive performance on the Kaplan—Albert Neuropsychological Test Battery were explored in a sample of 1,695 stroke-free participants of the Framingham Heart Study. Multiple blood pressure measurements were obtained over four or five exams extending over 8–10 years. Neuropsychological testing was conducted 12–14 years after the final blood pressure measurement. Data were subjected to multiple binary logistic regression analyses, and odds ratios were calculated for performance in the lower 50th and 25th percentiles for 10-mmHg increases in diastolic blood pressure, 20-mmHg increases in systolic blood pressure, proportion of exams (out of four or five) in which participants were diagnosed as hypertensive, and 10-year increases in age. Blood pressures and chronicity of hypertension were inversely associated with performance on visual and verbal memory tests of the Kaplan—...

Journal ArticleDOI
TL;DR: The extent and progression of baldness and the aforementioned outcomes were assessed using a Cox proportional hazards model, adjusting for age and other known cardiovascular disease risk factors.
Abstract: The authors assessed the relation between the extent and progression of baldness and coronary heart disease. Baldness was assessed twice, in 1956 and in 1962, in a cohort of 2,017 men from Framingham, Massachusetts. Extent of baldness was classified in terms of number of bald areas: no areas bald (n = 153), one area bald (n = 420), two areas bald (n = 587), and all areas bald (n = 857). Men who were assessed both times and who had two or fewer bald areas during the first evaluation were classified into one of three groups: "mild or no progression," "moderate progression," or "rapid progression." The cohort was followed for up to 30 years for new occurrences of coronary heart disease, coronary heart disease death, cardiovascular disease, and death due to any cause. The relations between the extent and progression of baldness and the aforementioned outcomes were assessed using a Cox proportional hazards model, adjusting for age and other known cardiovascular disease risk factors. Extent of baldness was not associated with any of the outcomes. However, the amount of progression of baldness was associated with coronary heart disease occurrence (relative risk (RR) = 2.4, 95% confidence interval (CI) 1.3-4.4), coronary heart disease mortality (RR = 3.8, 95% CI 1.9-7.7), and all-cause mortality (RR = 2.4, 95% CI 1.5-3.8). Rapid hair loss may be a marker for coronary heart disease.

Journal ArticleDOI
TL;DR: It is necessary to select patients suitable for vaginal or laparoscopic mesh placement for intranasal administration based on prior history and once they provide informed consent for surgery.
Abstract: Clinical Pharmacology & Therapeutics (1995) 58, 605–616; doi:

Journal ArticleDOI
TL;DR: Several methods for age- Adjustment of survival curves are discussed, including direct age-adjustment and proportional hazard modelling with age as a covariate.
Abstract: Kaplan-Meier curves provide descriptors of survival information for different subgroups within a data set. Investigators frequently use these figures for descriptive comparison of the effect of a particular measure upon survival. When subjects enter a study at different ages and age is associated with survival, it is often desirable to adjust survival information for age differences in the subgroups. This report discusses several methods for age-adjustment of survival curves, including direct age-adjustment and proportional hazard modelling with age as a covariate.

Journal ArticleDOI
TL;DR: The observed secular trends in diet and risk factor levels for cardiovascular disease in the Framingham population are important to guide the development and implementation of population-based strategies for promoting cardiovascular health, including nutrition interventions.
Abstract: Objective In this study we examined changes in dietary intake and risk factors for cardiovascular disease that occurred over three decades in a US-population-based sample. Design Secular trends in dietary profiles and risk factors were studied in cross-sectional samples of subjects from the Framingham Study in 1957–1960, 1966–1969, and 1984–1988. Results Dietary levels of cholesterol appeared to have declined considerably, whereas macronutrient and fatty acid intakes appeared to change only slightly. Men appeared to increase their saturated fat intakes from 16.4% in 1966–1969 to 17.0% in 1984–1988 ( P Conclusions The observed secular trends in diet and risk factor levels for cardiovascular disease in the Framingham population are important to guide the development and implementation of population-based strategies for promoting cardiovascular health, including nutrition interventions. J Am DietAssoc. 1995; 95:171–179.

Journal ArticleDOI
TL;DR: The authors used data collected on 1,959 males aged 35-69 years from the fourth Framingham Study examination to analyze the relations between total serum cholesterol levels and 409 coronary deaths, 325 cancer deaths, and 534 other deaths for a 32-year follow-up.
Abstract: The authors used data collected on 1959 [U.S.] males aged 35-69 years from the fourth Framingham Study examination to analyze the relations between total serum cholesterol levels and 409 coronary deaths 325 cancer deaths and 534 other deaths for a 32-year follow-up. The roles of smoking and alcohol drinking are also taken into account. "A significant U-shaped relation with all-cause mortality was noted as were an inverse relation to cancer mortality and a monotonic increasing relation with coronary disease mortality. In subset analyses the association of low serum cholesterol...with cancer mortality was observed in men who smoked cigarettes." (EXCERPT)

Journal ArticleDOI
TL;DR: AF has a negative impact on survival, independent of the preexisting cardiovascular conditions with which it is often associated, and the increased mortality of AF underscores the importance of its prevention.

Journal ArticleDOI
TL;DR: No relation between number of children and coronary heart disease incidence in men is found in two large prospective cohort studies, which contrasts with the previous positive association found between pregnancy and coronaryHeart disease in women.
Abstract: We examined the association between number of children and coronary heart disease risk among men in two large prospective cohort studies. Data from the Framingham Heart Study comprised a group of 1,632 men who were followed for 30 years; data from the first National Health and Nutrition Examination Survey National Epidemiologic Follow-up Study included 2,584 men who were followed for 3-5 years. We found no relation between number of children and coronary heart disease incidence in these cohorts. This lack of association in men contrasts with the previous positive association found between pregnancy and coronary heart disease in women.

Journal Article
TL;DR: The cardiovascular risk factors in the elderly are much the same as those that predispose middle-aged candidates for cardiovascular disease, and the short-term potential benefit of treatment is actually greater inThe elderly than in the middle aged.
Abstract: The cardiovascular risk factors in the elderly are much the same as those that predispose middle-aged candidates for cardiovascular disease. Some risk factors, such as blood lipids, impaired glucose tolerance, fibrinogen, and uric acid, are associated with lower risk ratios in advanced age, but this lower relative risk is offset by a high absolute risk. Thus, cardiovascular risk factors remain relevant at elderly age. Since the incidence of cardiovascular disease and the prevalence of the predisposing risk factors is so great in the elderly, the attributable risk is large, and the short-term potential benefit of treatment is actually greater in the elderly than in the middle aged. Preventive measures, as well, are useful in reducing the risk of disease in the elderly. Therefore, efforts at primary and secondary prevention should be considered in the elderly-they should not be neglected simply because of their age.

Journal ArticleDOI
TL;DR: A methodology for the development of health risk appraisal functions when the number of potential risk factors is large is discussed and sex specific functions for nursing home institutionalization are illustrated.
Abstract: A health risk appraisal function is a mathematical model designed to estimate the risk or probability of a person's mortality or morbidity for various diseases based upon risk factors such as age, medical history and smoking behaviour. The Framingham Study has contributed substantially to the development and use of these for endpoints such as mortality and incidence of coronary heart disease and other cardiovascular diseases. This paper discusses a methodology for the development of health risk appraisal functions when the number of potential risk factors is large and illustrates it with sex specific functions for nursing home institutionalization. The methodology involves grouping variables substantively into sets, applying principal component factor analysis and variable clustering to obtain substantively meaningful composite scores, ranking these in order of substantive importance, and then entering these with a hierarchical ordering into a Cox proportional hazard regression.

Journal ArticleDOI
TL;DR: The female advantage over men erodes with age, with the menopause and with acquisition of an unfavorable lipid profile and glucose intolerance, which is strongly influenced by the burden of risk factors.

Journal ArticleDOI
TL;DR: The purpose of this investigation was to assess the influence of gender in developing the TPI database and found that after adjustment for significant predictors of trial inclusion, women were 25% less likely to be included in clinical trials.
Abstract: The thrombolytic predictive instrument (TPI) was developed to identify those patients most likely to benefit from thrombolytic therapy for acute myocardial infarction as well as to facilitate the earliest possible administration of this treatment. The TPI consists of predictive models derived from clinical data obtained from both clinical trials and data registries. These models are subject to potential bias due to combinations of primary data from different sources. The purpose of this investigation was to assess the influence of gender in developing the TPI database. In this database, there were 1,096 (22%) women and 3,826 (78%) men; only 38% of the women were enrolled in clinical trials, whereas 46% of the men were (p < 0.0001). Within clinical trials, there were few significant eligibility differences between women and men, as the vast majority of patients met eligibility standards for entry in these trials. However, within clinical registries, the women were older (p < 0.0001) and more often had elevated blood pressure on admission (p = 0.002). Multivariate logistic regression indicated that after adjustment for significant predictors of trial inclusion, women were 25% less likely to be included in clinical trials (odds ratio = 0.76, 95% confidence interval = 0.60, 0.96). In order to counter bias introduced by the exclusion of women from clinical trials, the TPI database included patients from non-trial settings. Carefully including patients from clinical registries or non-trial settings may be an important strategy in constructing generally applicable predictive instruments.

Journal ArticleDOI
TL;DR: The results of fitting three longitudinal models, two autoregressive models and a compound symmetry model to data on a cohort of 1154 adult men in Boston show that the damped autore progressive model provided a significantly better fit than either of the other two models.
Abstract: We compare the results of fitting three longitudinal models, two autoregressive models (the serial correlation model and a damped autoregressive model) and a compound symmetry model, to data on a cohort of 1154 adult men in Boston. The serial correlation model assumes that the error terms are autocorrelated with correlation of the form lambda t for visits t years apart while the damped autoregressive assumes that the correlation between error terms of observations t years apart is of the form lambda t theta. The compound symmetry model assumes that the errors are correlated, with the same correlation regardless of how far apart observations are in time. These three models are all related in that the serial correlation and compound symmetry models are particular cases of the damped autoregressive models (that is, theta = 1 corresponds to the serial correlation model and theta = 0 corresponds to the compound symmetry model). For current smokers, the damped autoregressive model provided a significantly better fit than either of the other two models (p < 0.001); for never smokers the damped autoregressive and compound symmetry models were almost identical with both providing a significantly better fit than the serial correlation model (p < 0.001).