James H. Ware

Random-Effects Models for Longitudinal Data

Models for the analysis of longitudinal data must recognize the relationship between serial observations on the same unit. Multivariate models with general covariance structure are often difficult to apply to highly unbalanced data, whereas two-stage random-effects models can be used easily. In two-stage models, the probability distributions for the response vectors of different individuals belong to a single family, but some random-effects parameters vary across individuals, with a distribution specified at the second stage. A general family of models is discussed, which includes both growth models and repeated-measures models as special cases. A unified approach to fitting these models, based on a combination of empirical Bayes and maximum likelihood estimation of model parameters and using the EM algorithm, is discussed. Two examples are taken from a current epidemiological study of the health effects of air pollution.

An association between air pollution and mortality in six U.S. cities

BACKGROUND Recent studies have reported associations between particulate air pollution and daily mortality rates. Population-based, cross-sectional studies of metropolitan areas in the United States have also found associations between particulate air pollution and annual mortality rates, but these studies have been criticized, in part because they did not directly control for cigarette smoking and other health risks. METHODS In this prospective cohort study, we estimated the effects of air pollution on mortality, while controlling for individual risk factors. Survival analysis, including Cox proportional-hazards regression modeling, was conducted with data from a 14-to-16-year mortality follow-up of 8111 adults in six U.S. cities. RESULTS Mortality rates were most strongly associated with cigarette smoking. After adjusting for smoking and other risk factors, we observed statistically significant and robust associations between air pollution and mortality. The adjusted mortality-rate ratio for the most polluted of the cities as compared with the least polluted was 1.26 (95 percent confidence interval, 1.08 to 1.47). Air pollution was positively associated with death from lung cancer and cardiopulmonary disease but not with death from other causes considered together. Mortality was most strongly associated with air pollution with fine particulates, including sulfates. CONCLUSIONS Although the effects of other, unmeasured risk factors cannot be excluded with certainty, these results suggest that fine-particulate air pollution, or a more complex pollution mixture associated with fine particulate matter, contributes to excess mortality in certain U.S. cities.

Developmental and Neurologic Status of Children after Heart Surgery with Hypothermic Circulatory Arrest or Low-Flow Cardiopulmonary Bypass

Background Deep hypothermia with either total circulatory arrest or low-flow cardiopulmonary bypass is used to support vital organs during heart surgery in infants. We compared the developmental and neurologic sequelae of these two strategies one year after surgery. Methods Infants with D-transposition of the great arteries who underwent an arterial-switch operation were randomly assigned to a method of support consisting predominantly of circulatory arrest or a method consisting predominantly of low-flow bypass. Developmental and neurologic evaluations and magnetic resonance imaging (MRI) were performed at one year of age. Results Of the 171 patients enrolled in the study, 155 were evaluated. After adjustment for the presence or absence of a ventricular septal defect, the infants assigned to circulatory arrest, as compared with those assigned to low-flow bypass, had a lower mean score on the Psychomotor Development Index of the Bayley Scales of Infant Development (a 6.5-point deficit, P = 0.01) and a hig...

Random-effects models for serial observations with binary response.

This paper presents a general mixed model for the analysis of serial dichotomous responses provided by a panel of study participants. Each subjects serial responses are assumed to arise from a logistic model, but with regression coefficients that vary between subjects. The logistic regression parameters are assumed to be normally distributed in the population. Inference is based upon maximum likelihood estimation of fixed effects and variance components, and empirical Bayes estimation of random effects. Exact solutions are analytically and computationally infeasible, but an approximation based on the mode of the posterior distribution of the random parameters is proposed, and is implemented by means of the EM algorithm. This approximate method is compared with a simpler two-step method proposed by Korn and Whittemore (1979, Biometrics 35, 795-804), using data from a panel study of asthmatics originally described in that paper. One advantage of the estimation strategy described here is the ability to use all of the data, including that from subjects with insufficient data to permit fitting of a separate logistic regression model, as required by the Korn and Whittemore method. However, the new method is computationally intensive.

A comparison of the perioperative neurologic effects of hypothermic circulatory arrest versus low-flow cardiopulmonary bypass in infant heart surgery

Background Hypothermic circulatory arrest is a widely used support technique during heart surgery in infants, but its effects on neurologic outcome have been controversial. An alternative method, low-flow cardiopulmonary bypass, maintains continuous cerebral circulation but may increase exposure to known pump-related sources of brain injury, such as embolism or inadequate cerebral perfusion. Methods We compared the incidence of perioperative brain injury after deep hypothermia and support consisting predominantly of total circulatory arrest with the incidence after deep hypothermia and support consisting predominantly of low-flow cardiopulmonary bypass in a randomized, single-center trial. The criteria for eligibility included a diagnosis of transposition of the great arteries with an intact ventricular septum or a ventricular septal defect and a planned arterial-switch operation before the age of three months. Results Of 171 patients with D-transposition of the great arteries, 129 (66 of whom were assign...

Effects of Cigarette Smoking on Lung Function in Adolescent Boys and Girls

BACKGROUND Little is known about the sex-specific effects of cigarette smoking on the level and growth of lung function in adolescence, when 71 percent of people in the United States who smoke tried their first cigarette. METHODS We studied the effects of cigarette smoking on the level and rate of growth of pulmonary function in a cohort of 5158 boys and 4902 girls 10 to 18 years of age, examined annually between 1974 and 1989 in six cities in the United States. RESULTS We found a dose-response relation between smoking and lower levels of both the ratio of forced expiratory volume in one second to forced vital capacity (FEB1/FVC) and the forced expiratory flow between 25 and 75 percent of FVC (FEF25-75). Each pack per day of smoking was associated with a 3.2 percent reduction in FEF25-75 for girls (P=0.01) and a 3.5 percent reduction in FEF25-75 for boys (P=0.007). Whereas the FVC level was elevated in smokers, the rate of growth of FVC and FEV1 was reduced. Among adolescents of the same sex, smoking five or more cigarettes a day, as compared with never smoking, was associated with 1.09 percent slower growth of FEV1 per year in girls (95 percent confidence interval 0.70 to 1.47) and 0.20 percent slower growth in boys (95 percent confidence interval, -0.16 to 0.56), and with 1.25 percent slower growth of FEF25-75 per year in girls (95 percent confidence interval 0.38 to 2.13) and 0.93 percent slower growth in boys (95 percent confidence interval, 0.21 to 1.65). Whereas girls who did not smoke reached a plateau of lung function at 17 to 18 years of age, girls of the same age who smoked had a decline of FEV1 and FEF25-75. CONCLUSION Cigarette smoking is associated with evidence of mild airway obstruction and slowed growth of lung function in adolescents. Adolescent girls may be more vulnerable than boys to the effects of smoking on the growth of lung function.

Linear Models for the Analysis of Longitudinal Studies

Abstract Longitudinal investigations play an increasingly prominent role in biomedical research. Much of the literature on specifying and fitting linear models for serial measurements uses methods based on the standard multivariate linear model. This article proposes a more flexible approach that permits specification of the expected response as an arbitrary linear function of fixed and time-varying covariates so that mean-value functions can be derived from subject matter considerations rather than methodological constraints. Three families of models for the covariance function are discussed: multivariate, autoregressive, and random effects. Illustrations demonstrate the flexibility and utility of the proposed approach to longitudinal analysis.

Bivalirudin in patients with acute coronary syndromes undergoing percutaneous coronary intervention: a subgroup analysis from the Acute Catheterization and Urgent Intervention Triage strategy (ACUITY) trial

BACKGROUND The aim of this study was to assess anticoagulation with the direct thrombin inhibitor bivalirudin during percutaneous coronary intervention in individuals with moderate and high-risk acute coronary syndromes. METHODS 13,819 individuals in the Acute Catheterization and Urgent Intervention Triage strategy (ACUITY) trial were prospectively randomly assigned to receive heparin (unfractionated or enoxaparin) plus glycoprotein IIb/IIIa inhibitors, bivalirudin plus glycoprotein IIb/IIIa inhibitors, or bivalirudin alone. Of these individuals, 7789 underwent percutaneous coronary intervention after angiography. The effect of the three regimens on the primary 30-day endpoints of composite ischaemia (death, myocardial infarction, or unplanned revascularisation for ischaemia), major bleeding, and net clinical outcomes (composite ischaemia or major bleeding) was assessed in this subgroup. Analyses were done by intention to treat. This trial is registered with ClinicalTrials.gov, with the number NCT00093158. FINDINGS Of the individuals who underwent percutaneous coronary intervention, 2561 received heparin plus glycoprotein IIb/IIIa inhibitors, 2609 received bivalirudin plus glycoprotein IIb/IIIa inhibitors, and 2619 received bivalirudin alone. 26 (0.3%) individuals dropped out or were lost to follow-up. There was no significant difference in the proportion of individuals with composite ischaemia, major bleeding, or net clinical outcomes at 30 days between those who received bivalirudin plus glycoprotein IIb/IIIa inhibitors and those who received heparin plus glycoprotein IIb/IIIa inhibitors (composite ischaemia: 243 [9%] patients vs 210 [8%] patients, p=0.16; major bleeding: 196 [8%] patients vs 174 [7%] patients, p=0.32; net clinical outcomes: 389 [15%] patients vs 341 [13%] patients, p=0.1). Rates of composite ischaemia were much the same in those who received bivalirudin alone and those who received heparin plus glycoprotein IIb/IIIa inhibitors (230 [9%] patients vs 210 [8%] patients, p=0.45); however, there were significantly fewer individuals who experienced major bleeding among those who received bivalirudin alone than among those who received heparin plus glycoprotein IIb/IIIa inhibitors (92 [4%] patients vs 174 [7%] patients, p<0.0001, relative risk 0.52, 95% CI 0.40-0.66), resulting in a trend towards better 30-day net clinical outcomes (303 [12%] patients vs 341 [13%] patients, p=0.057; 0.87, 0.75-1.00). INTERPRETATION Substitution of unfractionated heparin or enoxaparin with bivalirudin results in comparable clinical outcomes in patients with moderate and high-risk acute coronary syndromes treated with glycoprotein IIb/IIIa inhibitors in whom percutaneous coronary intervention is done. Anticoagulation with bivalirudin alone suppresses adverse ischaemic events to a similar extent as does heparin plus glycoprotein IIb/IIIa inhibitors, while significantly lowering the risk of major haemorrhagic complications.

Randomized clinical trials. Perspectives on some recent ideas.

In spite of the controversy over the role of randomized clinical trials in medical research, the rationale underlying such trials remains persuasive as compared to recent suggestions for alternative non-randomized studies such as those relying on the use of historical controls and adjustment technics. Others have suggested that recent statistical innovations for improving clinical trials, including adaptive allocation of treatment to patients and sequential stopping procedures, are underutilized. These innovations, though theoretically interesting, are not easily adapted to large-scale, complex medical trials in which there may be multiple end points and delayed response times. Ethical considerations suggest that randomized trials are more suitable than uncontrolled experimentation in protecting the interests of patients. Randomized clinical trials remain the most reliable method for evaluating the efficacy of therapies.

Auranofin therapy and quality of life in patients with rheumatoid arthritis. Results of a multicenter trial

In a six-month, randomized, double-blind study at 14 centers, auranofin (3 mg twice daily) was compared with placebo in the treatment of patients with classic or definite rheumatoid arthritis. All patients had unremitting disease for at least the previous six months and at least three months of therapy with nonsteroidal anti-inflammatory drugs (NSAIDs). NSAIDs, oral steroids, and analgesics were allowed throughout the trial. Efficacy was analyzed in 154 patients who received auranofin and 149 who received placebo. To reflect an expanded view of outcome assessment, the measures used included some 20 nontraditional measures of functional performance, pain, global impression, and utility (worth or value) in addition to five standard clinical measures of rheumatoid synovitis (e.g., number of tender joints). The nontraditional measures were mainly in the form of structured questionnaires administered by trained interviewers. To minimize the statistical problem of multiple comparisons, most of the measures were grouped into four composites--clinical (standard measures), functional, global, and pain--and the treatment effect for each composite was tested at the 0.0125 level of significance. Auranofin was superior to placebo in the clinical (p = 0.003), functional (p = 0.001), and global (p = 0.007) composites and trended similarly in the pain composite (p = 0.021). Individual measures within the composites consistently favored auranofin. Other measures, not part of the composites, also favored auranofin, including a patient utility measure designed for this study, the PUMS (p = 0.002). Results confirm the hypothesis that the favorable effect of auranofin on clinical synovitis is accompanied by improvements across a range of outcomes relevant to the patients quality of life.