Multicausality: Compounding

Multicausality: Compounding

Multicausality: Confounding Assignment
These estimates include the influence of other extraneous variables, such as confounders.
Confounding is often considered a type of bias, but it is a real relationship that requires an
adjustment in the study design or analysis. Understanding how to identify confounding is
important as most associations have multiple causal factors. Recognizing if a study
adjusted for the appropriate confounding variables is important to determine the validity
of the association. To assist your proficiency with the concept of confounding, and how it
ultimately affects public health, this practice assignment has been provided.
Complete Problems 1 to 4 from the “Multicausality: Confounding – Assignment” by
Schoenbach, located in your Topic Materials. Check your answers against the solutions
presented in the “Multicausality: Confounding – Assignment Solutions” Topic Material.

Multicausality: Confounding

a) Studies have accentuated the fact that the use of anti-hypersensitive (AH)
medication such as reserpine for more than five years was associated with
increased risk of invasive breast cancer (RR= 1.18, 95% CI 1.02- 1.36) as
compared with lack of use. These findings are similar to those in other studies
indicating a statistically significant interaction based on the use of AH drugs at
baseline and the diastolic blood pressure (Schoenbach, 2001). Similarly, for those
women not using reserpine, the level of diastolic blood pressure was positively
associated with the subsequent risk of breast cancer. Conversely, women with AH
medication at the baseline portrayed an opposite effect in the blood pressure of
borderline significance. As such, these studies concluded that breast cancer
incidence among hypersensitive patients does not differ from the general

population (Largent et al. 2010). Therefore, in this case, reserpine is a risk factor
for breast cancer as the incidence of the latter is 10.47 per 100,000 women-years
who use the medication as compared to 6.14 in non-users. The consideration of
the non-obese women eliminates potential compounding in obesity as a risk factor
based on the rate ratio, 6.40/4.10 = 1.6.
b) Similarly, elevated levels of diastolic blood pressure have been associated with
an increased risk of breast cancer among non-pharmacologically treated women.
Therefore, obesity, in this case, is a risk factor based on the overall incidence rate
of 4.22 per 100,000 women-years as compared to 8.72/ 100,000 for obese women
(Schoenbach, 2001). These are coupled with the rate of 8.30/4.10 = 2.2 among the
women who did not use reserpine.
c) Due to the high prevalence rate of hypertension, anti-hypersensitive medications
are among the most common drugs prescribed for the treatment or mitigation of
the condition. Most fundamentally, the elucidation of the relationship between
hypertension, AH medication use, and the risk of developing breast cancer has
facilitated the examination of potential side effects. Studies have thus indicated
that long-term use (more than five years) of AH medication leads to an increased
risk of invasive breast cancer (Largent et al. 2010). As such, obesity, especially
among women, has been associated with the risk of hypertension, thereby
associating AH medications such as reserpine with obesity. As highlighted in the
solution provided, the prevalence of reserpine is relatively higher in obese women
based on the increased risk to develop hypertension.

d) Despite the positive correlation between reserpine and breast cancer, their
association is not attributable to obesity. As indicated in the above sections,
reserpine and obesity are risk factor for breast cancer, but there is no definitive
association between the medication and the condition (Largent et al. 2010).
Nonetheless, the evaluation of the crude rate ratio (10.47/6.14= 1.7) leads to a
slightly greater value as compared to the stratum-specific rate ratios (1.5 in obese
and 1.6 in non-obese). This may presumably indicate a weak association in the
attribution of reserpine to obesity.

a) According to the cohort study, there is a significant difference in the relative risk
for COPD among the smelters (10.5) and the truck drivers due to smoking.
However, since the same proportion (55%) of the source of population are
smokers, it cannot be deduced that smoking is the confounding factor that leads to
the difference in their risk of developing the lung disease due to the variation in
their exposure rate.
b) The development of chronic obstructive pulmonary disease (COPD) entails a
significant health problem in that the estimation if its prevalence is characterized
by considerable variations. Most fundamentally, most of these variations depict
the differences in the population under study, data quality control; rules used in
the definition of the condition, and spirometry methods. Studies have indicated
that the use of the lower limit of normal (LLN) forced expiratory volume in 1 s
(FEV1/FVC<LLN) criterion as opposed to the FEV1/FVC<0.7 leads to the
minimization of age biases (Lamprecht et al. 2013). It further reflects clinically

significant airflow limitation that is irreversible. Therefore, based on the study,
the supposition that low FEV1 is not an independent risk factor for COPD would
not serve as a suitable reason for failing to control it as a potential confounder.

  1. The implications associated with the models showing that homocysteine level (HCS)
    would need to be considered as a potential confounder of the association between oral
    contraceptive use (OC) and myocardial infarction (MI) entail:
    i. Homocysteine level (HCS) is a risk factor for the development of myocardial
    infarction (MI) due to an independent pathway of the oral use of contraceptives.
    Consequently, the control of the incidence rate of HSC, as a potential
    confounder, would lead to the avoidance of the association between oral
    contraceptive use and the development of myocardial infarction.
    ii. Since HCS is associated with risk of developing MI, it cannot serve as a
    confounder of the relationship between oral contraceptive use and the
    occurrence of MI. Moreover, the control of the effects associated with HCS
    would eliminate the causal pathway of OC (Lamprecht et al. 2013).
    Alternatively; the most appropriate approach would involve determining the
    causal relationship between HCS and MI as well as that between OC and HCS
    while controlling other influences.
  2. The three potential sources of the bias apparent from the characteristics of the different
    contraceptive groups include cigarette smoking where 17 subjects aged 25 to 29 years
    were reported to smoke fifteen or more cigarettes per day as compared to 7 and 12
    participants in the diaphragm and IUD users respectively (Schoenbach, 2001). This
    difference would largely influence the appearance of a causal association between oral

contraceptive use and circulatory deaths since cigarette smoking is a directly associated
with severe circulatory illnesses. These would be coupled with bias in the age factor,
which reported a significant difference in the age of the oral contraceptive, diaphragm,
and IUD users (Lamprecht et al. 2013). Similarly, the development of circulatory
illnesses would be directly associated with oral contraceptive use at higher rates as
compared to diaphragm and IUD users among young women in older ones.
Moreover, the incidence rate of hypertension in oral contraceptive users is higher (0.91%)
as compared to diaphragm and IUD users. This implies that the difference in the
incidence rates would influence the appearance of the causal association through the
depiction of the risk factors, incidence, and relative rates that lead to further development
of the condition and eventually circulatory death (Schoenbach, 2001). Similarly, the
history of rheumatic heart disease was relatively low in OC users as compared to
diaphragm and IUC users, thereby indicating a causal association between the two



Lamprecht, B., Mahringer, A., Soriano, J. B., Kaiser, B., Buist, A. S., & Studnicka, M. (2013).
Is spirometry properly used to diagnose COPD? Results from the BOLD study in
Salzburg, Austria: a population-based analytical study. Primary Care Respiratory
Journal, 22(2), 195-200.
Largent, J. A., Bernstein, L., Horn-Ross, P. L., Marshall, S. F., Neuhausen, S., Reynolds, P., …
Anton-Culver, H. (2010). Hypertension, antihypertensive medication use, and breast
cancer risk in the California Teachers Study cohort. Cancer Causes & Control, 21(10),
Schoenbach, V. J. (2001). Multicausality: Confounding – Assignment solutions.