Methods for epidemiologic analyses of multiple exposures: A review and comparative study of maximum-likelihood,preliminary-testing,and empirical-bayes regression

Many epidemiologic investigations are designed to study the effects of multiple exposures. Most of these studies are analysed either by fitting a risk-regression model with all exposures forced in the model, or by using a preliminary-testing algorithm, such as stepwise regression, to produce a smaller model. Research indicates that hierarchical modelling methods can outperform these conventional approaches. I here review these methods and compare two hierarchical methods, empirical-Bayes regression and a variant I call ‘semi-Bayes’ regression, to full-model maximum likelihood and to model reduction by preliminary testing. I then present a simulation study of logistic-regression analysis of weak exposure effects to illustrate the type of accuracy gains one may expect from hierarchical methods. Finally, I compare the performance of the methods in a problem of predicting neonatal mortality rates. Based on the literature to date, I suggest that hierarchical methods should become part of the standard approaches to multiple-exposure studies.     

Methods for trend estimation from summarized dose-response data, with applications to meta-analysis.

Meta-analysis often requires pooling of correlated estimates to compute regression slopes (trends) across different exposure or treatment levels. The authors propose two methods that account for the correlations but require only the summary estimates and marginal data from the studies. These methods provide more efficient estimates of regression slope, more accurate variance estimates, and more valid heterogeneity tests than those previously available. One method also allows estimation of nonlinear trend components, such as quadratic effects. The authors illustrate these methods in a meta-analysis of alcohol use and breast cancer.     

Identifiability and exchangeability for direct and indirect effects.

We consider the problem of separating the direct effects of an exposure from effects relayed through an intermediate variable (indirect effects). We show that adjustment for the intermediate variable, which is the most common method of estimating direct effects, can be biased. We also show that even in a randomized crossover trial of exposure, direct and indirect effects cannot be separated without special assumptions; in other words, direct and indirect effects are not separately identifiable when only exposure is randomized. If the exposure and intermediate never interact to cause disease and if intermediate effects can be controlled, that is, blocked by a suitable intervention, then a trial randomizing both exposure and the intervention can separate direct from indirect effects. Nonetheless, the estimation must be carried out using the G-computation algorithm. Conventional adjustment methods remain biased. When exposure and the intermediate interact to cause disease, direct and indirect effects will not be separable even in a trial in which both the exposure and the intervention blocking intermediate effects are randomly assigned. Nonetheless, in such a trial, one can still estimate the fraction of exposure-induced disease that could be prevented by control of the intermediate. Even in the absence of an intervention blocking the intermediate effect, the fraction of exposure-induced disease that could be prevented by control of the intermediate can be estimated with the G-computation algorithm if data are obtained on additional confounding variables.     

Evaluating anxiety in patients attending optometric practice

Although anxiety is a well-established obstacle to the delivery of effective health care, there have been no attempts to measure it in the optometric consulting room. In this paper, we introduce physiological and psychological techniques that may be used to evaluate anxiety and arousal in the consulting room and present data from a small group of patients attending for a routine eye examination. Specifically, arousal was assessed before, during, and after the examination by measuring skin conductance in five patients. Anxiety was evaluated using the State-Trait Anxiety Inventory. Our data confirm the ability of these techniques to quantify arousal and anxiety in the optometric consulting room and reveal a previously unknown but important facet of the eye examination. We conclude that these techniques are suitable for use in further experimental work and may be used to identify factors capable of reducing anxiety in the optometric consulting room.     

Identifiability, exchangeability and confounding revisited

In 1986 the International Journal of Epidemiology published "Identifiability, Exchangeability and Epidemiological Confounding". We review the article from the perspective of a quarter century after it was first drafted and relate it to subsequent developments on confounding, ignorability, and collapsibility.     

Divergent biases in ecologic and individual-level studies.

Several authors have shown that ecologic estimates can be biased by effect modification and misclassification in a different fashion from individual-level estimates. This paper reviews and discusses ecologic biases induced by model misspecification; confounding; non-additivity of exposure and covariate effects (effect modification); exposure misclassification; and non-comparable standardization. Ecologic estimates can be more sensitive to these sources of bias than individual-level estimates, primarily because ecologic estimates are based on extrapolations to an unobserved conditional (individual-level) distribution. Because of this sensitivity, one should not rely on a single regression model for an ecologic analysis. Valid ecologic estimates are most feasible when one can obtain accurate estimates of exposure and covariate means in regions with internal exposure homogeneity and mutual covariate comparability; thus, investigators should seek out such regions in the design and analysis of ecologic studies.     

Attitudes toward prevention of cardiovascular diseases among first-year students at eight American medical schools, 1983-1985

First-year medical students at eight U.S. medical schools were surveyed by written questionnaire in 1983-1985 to determine their attitudes toward cardiovascular diseases prevention at medical school entry. An overall response rate of 92% was achieved (2,654 questionnaires), and 97% of responders provided complete and analyzable survey data. Response rates at five of eight medical schools were 98-100%, and one school each had rates of 67, 84, or 90%. Differences in mean attitude responses from school to school were small, as were differences between men and women or between blacks and whites. This survey found that entering medical students have generally positive attitudes toward the effectiveness of preventive cardiology practice as well as toward the importance of research efforts in cardiovascular disease prevention. Students frequently indicated, however, that it is "extremely difficult" to change patients' unhealthful habits and that "physician encouragement" may not be sufficient to help patients achieve more healthful behaviors. These findings could be helpful in directing educational efforts for medical students. The data suggest that major emphasis should be placed on conveying facts regarding the physicians' efficacy in clinical preventive cardiology and on teaching the skills of preventive cardiology practice. Less emphasis appears to be necessary on encouraging positive attitudes about the importance of prevention since current students' attitudes appear to be already positive in this dimension.     

On the measurement of susceptibility in epidemiologic studies

Although relative effects of risk factors (relative risks) are commonly used in epidemiologic studies of disease, these measures do not provide estimates of the proportion of persons who are "susceptible" to the risk factor. Susceptibility may be defined under a simple sufficient cause model as the underlying factor (or set of factors) sufficient to make a person contract a disease following exposure. The authors derive simple estimates of the proportion of susceptibles in a population based on relative risk, and disease and exposure frequencies. The proportion of susceptibles increases with increasing disease frequency and relative risk but declines at high exposure frequency. For many chronic diseases with a lifetime risk in the range of 1-10 per cent, rare exposures suggest the presence of a large proportion of susceptibles, whereas common exposures suggest fewer susceptibles in the population. The estimation of the proportion of susceptibles is important in the search for genetic and environmental factors that interact with measured risk factors in etiologic studies of disease.     

Detection of Staphylococcal Superantigenic Toxins by a CD69-Specific Cytofluorimetric Assay Measuring T-Cell Activation

The presence of staphylococcal superantigenic toxins in the supernatants of liquid cultures was detected by an easy and rapid method assessing the activation of T lymphocytes by cytofluorimetric measurement of CD69 expression. Staphylococcus aureus cells were grown in Eagle’s minimum essential medium supplemented with 5% heat-inactivated fetal calf serum. Supernatant fluids from all S. aureus strains producing superantigen-related toxins, including enterotoxins A to E, toxic shock syndrome toxin, and exfoliative toxins A and B, induced CD69 expression in a significantly higher number of T cells than a cutoff of 2%. This CD69 assay might be used for initial detection of superantigens from S. aureus strains isolated in the context of staphylococcal toxemia or related chronic human diseases such as atopic dermatitis or Kawasaki syndrome.