Confounders & conditioning of analyses
Idea: Statistical associations between any two variables generally vary depending on the values taken by other "confounding" variables. We need to take this dependency (or conditionality) into account when using our analyses to make predictions or hypothesize about causes, but how do we decide which variables are relevant and real confounders?
Guidelines for annotations
Notes and annotations from 2007 course
Initial notes from PT
From PT:
Before reading this week's articles, read Gordis on confounding variables (chaps 14 old editions or 15 new edition) & chapter 3 or 4 on age adjustment (standardization).
When reading the articles, make notes on how the readings address the topic of adjusting for confounding variables (which includes age-standardization) and identify controversies or discordant views about how to do this.
Cases:
Immunization levels (Egede): Note the conclusion about racial/ethnic inequality even after adjusting for other variables thought to correlate with race/ethnicity. Do you agree with the three implications p. 326ff) drawn from the results?
SES gradients in disease (Krieger): The abstract states that "for virtually all outcomes, risk increased with CT [census tract] poverty, and when we adjusted for CT poverty, racial/ethnic disparities were substantially reduced." Where can the result of adjustment be seen in the paper? (This paper also fits in week 7 on inequalities.)
Hormone replacement therapy (Prentice vs. Petitti): Notice the adjustments used by the first paper that bring the clinical component of the WHI hormone replacement trial into line with the observational component. Do Pettiti acknowledge and rebut this in concluding that it was wrong to think that hormone therapy prevents CV disease?
Birth weight and blood pressure (Huxley vs. Davies): Along with Huxley et al's general argument that the birthweight-adult blood pressure association may well be an artifact of selective publication of studies with small sample size, they criticise the adjustment of the association for adult weight. (In other words, the association holds for people in the same stratum or slice of weight.) Try to form an opinion about whether you agree or disagree with such an adjustment. Davies et al. provide counter-evidence to Huxley et al. -- how does their study differ in methods, results, and interpretation?
Control at work and mortality (Davey-Smith 1997): This simple study shows that "control at work" is not the cause of SES gradients in health outcomes. What method(s) do they use to undermine previous claims about control at work?
Mendelian randomization to analyze environmental exposures (Davey-Smith & Ebrahim 2007): The approach introduced in this paper is cutting edge "epidemiology in the age of genomics" and has led to funding of a major new Research Center under Davey-Smith at Bristol. I suggest that you summarize for yourself the logic of this approach so you can explain it to someone who's never heard of it.
Annotations on common readings
Mendelian Randomization: Genetic variants as instruments... George Davey Smith and Shah Ebrahim
In their article, Davey-Smith & Ebrahim provide their readers with information on a genetic determinant of C-Reactive Protein (CRP) with little connection to the risk of coronary heart disease (CHD). The implications of these relationships can provide some insight as to the contributions of Mendelian randomization to causal inference in epidemiology. Mendelian randomization gives a focused evaluation of causality, but it relies on strong assumptions and may have limited usefulness in the evaluation of CRP as a marker of inflammation when the genetic variant has a modest ability to predict the phenotype of interest.
Mendelian randomization capitalizes on the random allocation of a genetic variant that influences an intermediate phenotype such as CRP to evaluate the possibly confounded relationship of that phenotype with an outcome such as CHD. A variety of approaches have been used in actual practice for causal inference about CRP and outcomes. The easiest method relies on qualitative comparisons between the effects of genetic variants on CRP and the connection between CRP and CHD.
Additionally, Mendelian randomization findings may be confounded by other genetic variants, with the variant that is being analyzed or by population stratification. But if correctly done, and carefully interpreted, Mendelian randomization studies can provide useful evidence to support or reject causal hypotheses. (CH)
Annotated additions by students
Associations and Attributions
The articles this week are closely related to my research interests, in that they attempt to establish correlation and causality among variation in SES and racial/ethnic differences to inequities or disparities in health status and outcomes. In the article by Smith et al on socioeconomic gradients, the authors explore earlier studies that attempted to establish causality between key indicators and health outcomes. These included studies pointing to ‘a lack of control’ at work and failure to own an automobile as related to early mortality. In this very brief article, the authors question these assertions pointing to the correlation of both with low-paid jobs, and suggest that while there may be an association, there is unlikely to be causation.
In the Egede and Zheng article, the authors challenge the reader to rethink attribution of outcomes to race and ethnicity, highlighting the fact that these are social constructs and therefore proxies for other, more tangible variables. The reader is asked to consider whether the lower rates of obtaining a flu vaccine for certain ethnic/racial groups [PT] are due to provider prejudice, lack of patient access, patients’ fears and biases re: vaccines and medical care in general – or something else all together.
The articles re: variance in outcomes pertaining to birth weight and blood pressure rates later in life served to demonstrate how sample size and selection can affect the research findings. In the case of the larger, screening sample, there were issues of participant recall of birth-weight, as well as rounding up or down of BP rates among this large sample that generated different findings as compared to a smaller sample where birth-weights were verified in medical records and BP was measured more precisely.
On the whole, these articles challenge ‘easy’ answers to research questions in which complex SES constructs are involved and the risk of confounding factors is particularly high. (posted by Amy H 10-14-09)
Egede and Zheng: Racial/Ethnic Differences in Adult Vaccination Among Individuals
with Diabetes
The reading provides an overview of the research that was done using the 1998 National Health Interview Survey, regarding diabetes and disparities in influenza and pneumococcal vaccinations. The aim was to determine what independent variables, if any, influence the likelihood of a diabetic person receiving vaccinations for influenza and pneumonia.
Diabetes most often affects minorities, with African Americans and Hispanics being more likely to suffer complications or death as a result of the disease. Several independent variables were used in the analysis to determine vaccination rates: race/ethnicity, age, education, household income, four census regions, origin of birth, marital status, access to quality care, health insurance, and socioeconomic status (SES). Additionally, four models were developed: “the base model, base model plus race/ethnicity and access to care, and base model plus race/ethnicity, access to care, and SES”; the tests were 2-tailed and the significance level was set at a<.05.
It was found that across all models, race/ethnicity influenced the rate of influenza and pneumococcal vaccines given to persons with diabetes. It was concluded that only 54% of diabetics who visited with a primary care physician received an influenza vaccination, which implies that not all primary care physicians are recommending that their diabetic patients receive this important vaccine. Also, the race/ethnicity variables need to be further defined so that a better understanding can be gained as to how social and cultural factors impact health outcomes across racial/ethnic groups. Lastly, cultural norms and beliefs may also point to why diabetics do not receive the influenza and pneumococcal vaccines (i.e., religion, superstition, etc.), “and differential recommendation of vaccination by physicians were responsible for disparities in vaccination coverage.”
Whites (non-Hispanics) had higher rates of vaccination than African Americans or Hispanics. And it was found that race/ethnicity predicted the administration of both vaccines. Therefore, the hypothesis put forth by the authors cannot be validated: “differences in access to care, health care coverage, and SES would explain racial disparities in influenza and pneumococcal vaccination rates in adults with diabetes.” (CH '09)
Egede and Zhange found that race/ethnicity is an important predictor of influenza and pneumococcal vaccination independent of access to care, health coverage, and socioeconomic status when looking at diabetics (p325). Even though diabetics are at higher risk of serious side effects from influenza, these researchers found that those who come from racially or ethnically disadvantaged groups have less access to important, and standard, vaccines. Further, the researchers controlled for comorbid conditions that might make a doctor more likely to recommend vaccination. This study also allowed for more distinction between races than previous studies, showing a more precise division between vaccination of whites and blacks (rather than whites versus non-whites).
This research serves as an important reminder to always consider disparities in access to care and care provision when conducting epidemiological research. (MC '09)
