In an insightful opinion piece featured in The Daily Caller (on November 20, 2017), Dr. Carl Phillips equips readers with the tools necessary for a more informed and nuanced critique of scientific claims, with a particular focus on divisive subjects such as vaping. Through this, he advocates for a crucially critical perspective essential in deciphering the intricate world of public health research. Dr. Phillips underscores the perils of simplistic assertions and misinterpreted findings, illustrating how they can erroneously shape public discourse and policy. This call for a more discerning evaluation of scientific data emphasizes the importance of depth and precision in understanding the nuanced dynamics in health-related research, particularly in areas that spark significant debate and contention.
“Science Lesson: How Understanding ‘Confounding’ Can Combat Anti-Vaping Junk Science,” written by Dr. Carl V. Phillips, is an article that emphasizes the importance of recognizing and addressing confounding factors in epidemiological research, especially in studies related to vaping.
Confusion arises when the observed relationship between two variables is influenced by a third factor related to both the exposure and the outcome but is not a step in the causal path from exposure to outcome. This can lead to erroneous conclusions about the proper relationship between the exposure and the outcome.
Consider a study aiming to understand whether vaping increases the risk of lung diseases. Preliminary results indicate that individuals who vape have higher rates of lung diseases compared to those who have never vaped or smoked. At first glance, this could suggest that vaping is the direct cause of the increased risk of lung diseases. However, consider a vital variable not accounted for in this analysis: the smoking history among vapers. That is, many people who vape are ex-smokers or current smokers who also use electronic cigarettes. Smoking is known for its strong association with lung diseases.
In this scenario, the smoking history acts as a third variable that confuses the relationship between vaping and lung diseases. The confusion occurs because both vaping and smoking are related to the outcome (lung diseases). Still, it has not been considered that smoking alone is already a well-established risk factor for these diseases. Therefore, without adjusting for smoking history, the study’s results could erroneously attribute the risk of lung diseases to vaping when, in fact, it could be the residual effect of previous or concurrent smoking.
This is how confusion can bias a study’s results and lead to incorrect conclusions if all relevant variables are not identified and adequately controlled. Proper consideration of confusion is essential for accurate causal inferences in epidemiological research.
Beyond the Confusion: Seeking Clarity in Health Research Methodologies
Phillips introduces the topic by highlighting how many individuals who use electronic cigarettes have developed an intuitive understanding of this critical epidemiological concept known as confusion despite needing to be formally familiar with the term. In this context, confusion refers to the challenge of determining whether an observed association between two variables (such as vaping and smoking in the Gateway Process) is causal or is due to unconsidered factors.
While many intuitively understand these arguments, the author suggests that having full knowledge of the concept of confusion can be even more beneficial for critically evaluating scientific claims and refuting what he calls “junk anti-vaping science.”
In this sense, the American epidemiologist addresses a fundamental aspect in the interpretation of health research: the distinction between association and causality and the role of confusion in this context. “Confusion occurs when a difference in the outcome is observed between groups exposed and not exposed to a certain factor, but such a difference is not directly due to the exposure. This phenomenon can lead to misinterpretations of the causal relationship between variables, such as in the case of vaping and lung diseases.
According to the author, confusion manifests when the observed results are due more to unconsidered factors (confounders) than the exposure itself. Confounders are variables that researchers identify and adjust in their analyses to reduce the impact of confusion on the study’s results. These are not the direct cause of confusion but variables associated with the studied exposure and the observed outcome. Through their adjustment, researchers seek to clarify the relationship between the exposure and the outcome, trying to isolate the effect of the exposure of interest from other factors that could influence the outcome.
Suppose an epidemiological study wishes to investigate whether consuming a diet high in saturated fats is related to a higher risk of heart disease. The high saturated fat diet would be the exposure of interest, and heart disease would be the outcome.
Physical activity would be a confounder. People who consume diets high in saturated fats might be less physically active than those with healthier diets. Physical activity, known to reduce the risk of heart disease, then acts as a confounder, as it is associated with both the exposure (high-fat diet) and the outcome (heart disease). In this case, if the study does not adjust for the level of physical activity, the results could mistakenly show that a high saturated fat diet has a strong relationship with heart disease when part of that risk could be due to the lower physical activity in that group.
Clarifying the Haze: Confounders in the Debate on Vaping and Health
Phillips criticizes the flawed understanding of confusion by many readers and authors, who often confuse the terms “confounder” and “confusion,” and underscores the importance of recognizing that, although some research may suggest associations between exposures and outcomes, inferring direct causality is a much more complex process subject to interpretation. Consequently, he emphasizes the need for careful analysis to distinguish between the effects of confusion and authentic causal relationships, especially in research with significant implications for public health and health policy.
The article discusses a common challenge in health research: managing confusion, which refers to how external variables can affect the observed relationship between two variables of interest, potentially biasing the results. To address this, researchers introduce additional variables in their analyses, called “confounders,” to “control” or adjust for this confusion and isolate the actual causal relationship. However, Dr. Phillips argues that this term needs to be more accurate, as these variables are not the source of confusion but attempt to correct it. Therefore, it would be more appropriate to call them “deconfounders.”
These deconfounders are often imperfect approximations and fail to remove the effect of confusion altogether, leaving what is known as “residual confusion.” An example given is adjusting for the number of years a person has smoked to try to control the impact of previous smoking in a study on vaping. This adjustment is imperfect because it ignores factors such as the intensity of the smoking habit, leading to only partially removing the confusion.
The text criticizes the standard practice in health research of making superficial adjustments without carefully considering how to measure and correct adequately for propensities or inclinations that could introduce bias into the results. This is particularly relevant in studies seeking to identify causal effects, such as those investigating whether vaping acts as a “gateway” to the smoking habit.
Even after attempting to adjust for confounders, residual confusion suggests that study estimates may be biased. This underscores the importance of understanding and recognizing confusion in epidemiological research. Knowing the technical details of confusion and how it can affect estimates strengthens arguments when discussing these issues. It helps identify when and how study results may be misinterpreted due to inadequate adjustments.
“Confusion arises when unconsidered external factors distort the observed relationship between two variables.”
In his approach to critical interpretation of epidemiological research, Dr. Carl V. Phillips proposes a series of intellectual tools designed to unravel the complexity behind the concept of confusion and its influence on health studies. Dr. Phillips underlines the essentiality of fully grasping the nature of confusion, which arises when unconsidered external factors distort the observed relationship between two variables. This critical understanding enables readers to discern situations where associations may be misinterpreted due to external influences more accurately.
Additionally, Dr. Phillips meticulously distinguishes between the terms “confusion” and “confounders,” referring to the bias introduced by external variables and the latter to the variables that researchers attempt to adjust in their analyses to mitigate confusion. This differentiation is key to evaluating whether a study has effectively handled the factors that could skew its results.
The epidemiologist also criticizes the practice of making inadequate adjustments in studies, pointing out the tendency to use adjustment variables that do not fully capture the essence of the true confounders, leaving behind what he terms “residual confusion.” This skepticism towards the adjustments made underscores the importance of examining whether efforts to control confusion are genuinely effective.
Phillips also emphasizes the importance of recognizing how differences in the propensities of the populations studied, such as the inclination of youth towards vaping and smoking, can introduce confusion into studies. This approach is especially relevant in research on the gateway effect of vaping to smoking, suggesting that a deeper analysis of these behaviors is necessary to avoid erroneous conclusions. He also advocates for a critical analysis of the methodology used in health studies, promoting a rigorous evaluation of how confusion is managed and the selection of adjustment variables. Adjustment variables are factors that can influence both the exposure and the outcome of a study and, therefore, must be considered in the analysis to avoid erroneous conclusions.
For example, research on smoking highlights the importance of considering a wide range of variables when examining the effects of tobacco throughout history. These variables include economic, social, cultural, and political factors that can affect tobacco production, marketing, and consumption, as well as its perceptions and regulations.
In the context of tobacco research or research on any other topic, adjustment variables help researchers better understand the relationship between the exposure of interest (such as the use of tobacco) and the measured outcome (for example, health problems). By adjusting for these variables, the confusing effects of other factors that could bias the results can be minimized, allowing for a more accurate interpretation of the causal relationship.
For instance, in epidemiological studies, adjustment variables may include age, gender, socioeconomic status, and lifestyle habits such as diet and exercise. Adjusting for these variables allows researchers to isolate the specific effect of the exposure of interest on the outcome, thus providing a clearer understanding of its impact. According to Phillips, a critical attitude is essential for navigating the complex landscape of public health research and avoiding adopting conclusions and policies based on simplistic claims or poorly interpreted results.
In a few lines, Dr. Phillips equips readers with intellectual tools to address scientific claims critically, encouraging rigorous scrutiny of studies and their methodologies. This approach is vital in evaluating controversial topics like vaping, where deep understanding and critical analysis are fundamental to deriving informed conclusions and policies based on solid evidence.