Causation Assessment in Epidemiologic Research

Question one

Strength of association

According to Hill, “a strong association is more likely to have a causal component than is a modest association” (HÖfler, 2005). “Hill illustrated this point with the high-risk ratios for the association between exposure levels of smoking and incidents of lung cancer” (HÖfler, 2005). Rothman and Greenland proved Hill’s basis to be inaccurate by providing counter-examples for strong but non-causal relationships (HÖfler, 2005). They argued that the strength of association is usually very weak unless the association is at a ratio of 1:1.


The consistency criterion fails to include the different immutable properties (HÖfler, 2005) in an association to have an accurate analysis and interpretation in different study designs.


In this criterion, Hill assumes that a specific factor causes an effect on a particular outcome among a group of people. This assumption is not true since the factor might have other effects on an individual.


The criterion of temporality assumes an outcome of a process is influenced by its preceding factor. This criterion has been criticized for assuming that the preceding factor will always produce a specific outcome.

Biological gradient

The criterion of biological gradient assumes that there is a linear increase in a graph of exposure against the outcome. “Rothman and Greenland have argued that parts of J-shaped dose-response curves might be caused by the respective exposure levels while others might be due to confounding” (HÖfler, 2005).


The plausibility criterion relies on substantive explanation. “According to Rothman and Greenland, the assessment of plausibility is subject to the prior beliefs of individual researchers” (HÖfler, 2005).


The rule of coherence is criticized for using known facts as a score sheet. This is because Hill’s definitions of plausibility and coherence appear to be subtle…such contradicting knowledge might itself be wrong (HÖfler, 2005).


This criterion is criticized for using randomized experiments to identify the cause of a certain event. Susser suggests the use of Intervention and active change instead of a randomized experiment.


Hill suggested that scientists “judge by analogy” when conducting experiments. This criterion is criticized for not producing similar results compared to other settings.

Question Two

Research One: Epidemiology of Soy Exposures and Breast Cancer Risk

Strength of association

This research found that soy intake during adolescence exhibited a stronger protective effect on risk than adulthood exposure (Messina & Loprinzi, 2001). This experiment clearly shows the strong relationship between soy uptake in adolescence. As a result of this uptake, adolescences are protected against breast cancer. This experiment observes Hill’s criterion on the strength of association.

Specificity, Temporality and Biological gradient

The experiment shows an outcome in soy uptake among adolescence and adults. Each research group met Hill’s criteria by producing a specific outcome after the experiment. The research also satisfies the criteria of temporality. The soy consumed by adolescents and adults influenced the level of protection against breast cancer. The experiment showed that mammographic density is significantly reduced in association with high soy food intake (Messina & Loprinzi, 2001). This observation from the experiment satisfies the biological gradient criteria.


The various experiments were conducted to determine the effect of soy against breast cancer among adolescent and menopausal women. The research was conducted in Japan and Los Angeles. These experiments were conducted among different groups in different environments in accordance with the consistency criterion.

Research Two: Soy for Breast Cancer Survivors: A Critical Review of the Literature

Strength of Association

This research concluded that the non-U.S.-born Asians had protective effects against breast cancer compared to U.S.-born Asians. This is because U.S.-born Asians consume less Soy than non-U.S.-born Asians. The observation from the experiment satisfies the strength of association criteria.


The various experiments conducted found inconsistencies in the association between soy uptake and cancer risk. Some of the research conducted showed there was no association; others showed modestly reduced risks and some found quite pronounced effects (Wu, Yu, Tseng & Pike, 2008). Thus, the observation made from the experiment does not satisfy Hill’s criterion of consistency.


The experiment concluded that none of the Asian studies found that soy intake was associated with an increased breast cancer risk (Wu, Yu, Tseng & Pike, 2008). This research identified isoflavones as the compound which reduces the risk of breast cancer in human beings. Low levels of breast cancer in Asia are attributed to the large amounts of soy consumed as per their traditional lifestyle. Increased consumption of soy by Asians contributed significantly to the low levels of breast cancer in the region.


Studies have shown that people consuming large quantities of soy have better protection against breast cancer. This observation made from the research satisfies Hill’s temporality criterion. This experiment concludes that a person will have protection against breast cancer after consuming large quantities of soy.

Biological Gradient

Research conducted in Asia showed that people in Japan had better protection against breast cancer compared to the people in the U.S. This is attributed to the large quantities of soy that people consume in Japan than in the U.S. The experiment shows that an increase in consumption of soy results in higher protection against breast cancer. This observation satisfies Hill’s biological gradient criterion.


The research on the relationship between soy and breast cancer was inspired by an experiment conducted on rats. Early research showed that the addition of soy protein to a typical laboratory diet significantly decreased chemically induced rat mammary cancer (Wu, Yu, Tseng & Pike, 2008). This led to the test on human beings to determine if it will have the same results as on the rats. This observation supports Hill’s argument which states that causation is more likely if the evidence is based on randomized experiments (Wu, Yu, Tseng & Pike, 2008).


HÖfler, M. (2005). The Bradford Hill considerations on casualty: A counterfactual perspective. Harm Reduction Journal, 2:11. Web.

Messina, M. J., Loprinzi, C. L. (2001). Soy for Breast Cancer Survivors: A Critical Review of the Literature. The Journal of Nutrition, 131:11. Web.

Wu, A. H., Yu, M. C., Tseng, C. C., Pike, M. C. (2008). Epidemiology of soy exposures and breast cancer risk. British Journal of Cancer, 98, 9-14. Web.