Multivariable analysis of association of beta-blocker use and survival in advanced ovarian cancer
Rafael González ManzanaresA.A. GockleyAlexander MelamedR.M. ClarkJohn O. SchorgeM.G. del CarmenRoss S. BerkowitzN.S. HorowitzMichael G. MutoMichael J. Worley
0
Citation
0
Reference
10
Related Paper
Keywords:
BETA (programming language)
Beta blocker
BETA (programming language)
Beta blocker
Beta adrenoceptor
Adrenergic beta-Antagonists
Cite
Citations (0)
Abstract This chapter begins with a discussion of definition and theoretical background of confounding. It then focuses on the quantification of potential confounding, evaluation of confounding, and integrated assessment of potential confounding.
Cite
Citations (3)
Effect modification
Cite
Citations (0)
Research Design
Cite
Citations (0)
Confounding may be present in nonrandomized etiological research involving human populations. It can result in erroneous conclusions about the effect of exposure on a disease outcome or about any form of causality between predictors and outcomes. Confounding can wholly or partially account for the apparent effect of the risk factor under consideration or mask the underlying, true association. Not controlling for the effects of confounding can lead to biased results, thus compromising the validity of study conclusions. The three goals of this article are: (1) to define a confounder or a confounding variable, (2) to discuss strategies for controlling the effects of confounding, and (3) to illustrate the perverse effects of confounding with the help of an example.
Causality
Etiology
Cite
Citations (1)
Abstract The first part of this chapter discusses the conditions under which a factor can confound the association between exposure and disease, and the conditions under which this cannot occur. It also differentiates confounders from antecedents or mediators. The next part discusses methods devised to neutralize the effects of confounders. Two standard methods are presented: matching to prevent confounding in the data by equalizing the exposed and the unexposed on a potential confounder, and statistical adjustment to compensate for confounding in the data by separating the effects of the exposure from the effects of the confounder.
Cite
Citations (2)
Abstract Confounding biases study results when the effect of the exposure on the outcome mixes with the effects of other risk and protective factors for the outcome that are present differentially by exposure status. However, not all differences between the exposed and unexposed group cause confounding. Thus, sources of confounding must be identified before they can be addressed. Confounding is absent in an ideal study where all of the population of interest is exposed in one universe and is unexposed in a parallel universe. In an actual study, an observed unexposed population represents the unobserved parallel universe. Thinking about differences between this substitute population and the unexposed parallel universe helps identify sources of confounding. These differences can then be represented in a diagram that shows how risk and protective factors for the outcome are related to the exposure. Sources of confounding identified in the diagram should be addressed analytically and through study design. However, treating all factors that differ by exposure status as confounders without considering the structure of their relation to the exposure can introduce bias. For example, conditions affected by the exposure are not confounders. There are also special types of confounding, such as time‐varying confounding and unfixable confounding. It is important to evaluate carefully whether factors of interest contribute to confounding because bias can be introduced both by ignoring potential confounders and by adjusting for factors that are not confounders. The resulting bias can result in misleading conclusions about the effect of the exposure of interest on the outcome.
Cite
Citations (39)
Abstract Misclassification of exposure in epidemiologic investigations has been extensively studied and is now well understood. In contrast, misclassification of confounding factors has been much less investigated. First, we consider a situation with confounding by age, in which misclassification is introduced through stratification of this inherently continuous variable. This misclassification turns out to be benign: 75% of the original confounding is removed by stratification into two age classes and more than 90% by using three age classes. Second, we consider a situation with serious confounding and serious misclassification of the confounding factor but no misclassification of the exposure. In this situation, the misclassification turns out to be of importance. After stratification for the misclassified confounding factor, it appears as though the exposure has a stronger effect on the incidence than the confounder, which is the reverse of the true situation.
Stratification (seeds)
Cite
Citations (37)
Cite
Citations (0)
An association that is not causal (Alternate explanations) is seen between outcome and exposure wheneverthe study has confounding factors, bias and random error/chance error. Bias in any study leads to systematicvariation of inferences or results or interpretations from the true picture. One of the most important biasencountered during research is confounding. Confounding is seen when a variable is observed to be associatedwith both the exposure and the outcome but is not a part of the causal pathway. Confounding factors, whenpresent, in any study, are the “Nuisance” and can be the cause in part or in full of the observed associationbetween the disease and exposure. Effect modification, on the other hand, is seen when various effects arebrought about among different subgroups by an exposure and this can be handled by doing stratification. It isthe outcome that is linked to the effect modification and not the exposure. This article will define and discussin detail confounding, and the concept of effect modification so that drawing conclusions of a study can bedone after considering these errors.
Effect modification
Cite
Citations (0)