Heterogeneity within populations and subgroups
Idea: How people respond to treatment may vary from one subgroup to another--When is this a matter of chance or of undetected additional variables? How do we delineate the boundaries between subgroups?
Guidelines for annotations
Notes and annotations from 2007 course
Initial notes from PT
Lagakos provides a statistician's cautions about the significance of results derived from subgroups of the whole population, especially if the subgroups were only defined after exploring the data.
The opposite caution is that treating everyone as if they were from the same population (for good statistical reasons) distracts our attention from the clues that might lead us to seeing that the population is not one uniform whole, but is a mixture of types. This can have significant health care implications -- see case studies about different kinds of breast cancer (Regan) and aspirin resistance.
Annotations on common readings
Annotated additions by students
Eikelboom
This article speaks to the benefits of an aspirin regime in the prevention of such vascular conditions as atherothrombosis, nonfatal myocardial infarction, nonfatal stroke, or vascular death. Patients at high risk for these diseases (N=29,652) were subjected to long-term therapy and after about two years it was found that there was an odds reduction of about 23 percent, a relative risk reduction of 19 percent, and an absolute risk reduction of 3.1 percent over two years (1.5 percent per year). It was surmised that aspirin did not prevent recurrent vascular events in high risk patients (81 percent), and 12.9 percent of high risk patients, despite taking aspirin, experienced a recurrent vascular event in the next two years.
It is hypothesized that some patients are resistant to the antiplatelet effects of aspirin, which is why an aspirin regime may not be effect in all persons. And, aspirin resistant is described as a failure to produce the anticipated “response on one or more laboratory measures of platelet activation and aggregation. Therefore, many have been deemed aspirin resistant (2-8).
In 2001, the Journal reported that confidence intervals of 325 patients with a history of coronary or cerebral vascular disease were aspirin resistant (3.3 percent to 8.6 percent). 325 mg/day was the measure of aspirin intake that was deemed a failure when given to individuals for a minimum of seven days before testing. “After adjusting for 12 potential prognostic factors, multivariate analysis indicated that aspirin resistance was associated with a 4.1-fold excess adjusted hazard of serious vascular events, independent of age, gender, and conventional vascular risk factors.”
Gum et al noted that failure to render the anticipated effect of aspirin on a laboratory measure of platelet aggregation, present in about 1 in 20 high-risk patients, is an independent predictor of risk of serious vascular events in the future. However, the number of outcomes was small, which means that the estimate of the magnitude of hazard is not precise. “The 95 percent confidence intervals are consistent with an excess hazard as low as 1.4-fold and as much as 12-fold.
One can now ask if the term aspirin resistance is standardized, valid, or reliable. Can this study be generalized to other laboratories and other patients? It is hard to figure out what technique is the most sure or valid measure of aspirin resistance without direct comparison of their clinical relevance “such as capacity to discriminate patients at risk of recurrent vascular events.”
A fair amount of individuals treated with aspirin fail to receive the benefits on routinely used laboratory measures of the antiplatelet effects of aspirin. More is required to standardize and validate laboratory tests and the identification of therapeutic strategies that can modify the results of these investigations and subsequently reduce the risk of future thrombotic complications. Despite all else, an aspirin regime has proven beneficial (and successful) in preventing premature atherothrombotic vascular events in many individuals. (CH, '09)
Race in a Bottle (Kahn, 2007)
Understanding how genes work and how they influence our physical traits and our health has been an objective of modern science, medicine in particular. A great deal of genetic research aims to understand genes in order to provide more effective treatment to patients, depending on their “genetic make-up”. This brought race into play as a factor of potential differences. Although most research goals aimed at understanding genes in relation to the race were entirely altruistic, some have had a more pragmatic, sometimes commercial purpose.
The article by Kahn discusses history of an “ethnic” drug - BiDil, from the rejection of the drug by the FDA for general population to the process of targeting and patenting the drug for African-Americans, and final approval by the FDA for this sub-population.
Kahn emphasizes negative aspects of this kind of “pragmatic science” (emphasis by DBJ). While he notes that “science and commerce have always proceeded together”, he concludes that in the case of BiDil the commercial reasons exceeded the scientific ones. From the scientific standpoint, the second patent was the same as the first one, except that the use of the drug in the second patent was specified for African-Americans suffering from cardiac failure, even though the inventor of the drug himself indicated that BiDil should work for people of all races. Targeting (and marketing) the drug for African-Americans seemed to be the way to get it approved. It was also profitable for the pharmaceutical company that owned the patent.
The key issue Kahn raises is the effect this kind of ethnic “flirting” (DBJ) has on the delivery of care. On one hand, Non-African-American patients with cardiac failure will not be treated with the drug that could be beneficial for their health because it is not intended for them. On the other hand, because the pharmaceutical company owning the patent will not have competition for certain number of years, the drug can be more expensive, which will further result in limited access to the drug for those who cannot afford it or for those whose health insurer would not cover it. Either way, health benefits for many are surpassed by commercial benefits for the few. Furthermore, while trying to understand how race is related to genes and how this can improve health of certain racial or ethnic groups, some findings may cause unequal treatment due to race, especially for the commercial reason (how insurance companies will assume their risk and profit) (DBJ 09).
