Causation is the relationship between cause and effect, where one event (cause) brings about another (effect). Concepts of Causation explores the connection and influence between variables, explaining the occurrence of phenomena in a sequential and dependent manner.
Establishing causation often involves interpreting statistical measures such as relative risk (RR) in cohort studies or odds ratios (OR) in case-control studies. These statistical indicators offer researchers vital insights into the strength of associations observed between a particular disease and exposure to suspected risk factors. While relative risk quantifies the likelihood of disease occurrence among exposed versus unexposed individuals in cohort studies, odds ratios assess the odds of exposure among individuals with the disease against those without it in case-control studies. These metrics serve as crucial tools, shedding light on the potential relationships between diseases and their suspected risk factors, allowing researchers to discern and evaluate these associations.
We will explore the principles associated with establishing causation. By the end of this page you should be able to:
- define the term causal factor.
- discuss characteristics associated with causal factors.
- describe the process of establishing causation.
- discuss the Bradford Hill’s criteria for establishing causation.
Causal Factors
A causal factor, in the context of epidemiology and health, refers to any element, circumstance, or attribute that contributes to or elevates the risk or likelihood of a particular disease or health condition.
A factor is typically deemed causal when its increase or presence is associated with a higher occurrence or frequency of the disease or health issue in question. The strength of this association further solidifies the theory of causation. Moreover, if reducing or eliminating the factor results in a subsequent decrease in disease incidence or severity, it provides substantial evidence supporting its causal relationship with the health condition.
In essence, identifying and understanding causal factors are pivotal in preventive medicine and public health as they offer valuable insights into developing interventions, policies, or strategies aimed at reducing the burden of diseases or health problems by targeting these influential factors
Characteristics of Causal Factors
Direct Factors
Direct factors directly influence changes in the body’s cellular structure or functioning. For instance, exposure to a toxic substance affecting cell structure can be considered a direct factor or coronary artery disease (CAD) directly resulting in myocardial infarction.
Indirect Factors
These factors heighten the likelihood of pathogenic changes without directly causing them. They might create an environment conducive to disease development. For example, an unhealthy diet, sedentary lifestyle, and dyslipidemia indirectly influence the risk for CAD significantly influencing incidences of cardiovascular diseases such as myocardial infarction.
Single Event vs. Complex Event
A single event involves a sole contact leading to an outcome, such as one exposure to the HIV virus. Conversely, a complex event entails multiple interacting factors causing an outcome, like the causal web in chronic non-communicable diseases such as diabetes or hypertension, involving various contributing factors like obesity, age, lifestyle, and genetics.
Necessary Factors
These factors must be present for a disease to occur. For instance, in the case of tuberculosis or AIDS, the specific pathogen must be present, along with a compromised immune system. Using the Web of Causation, hypertension is a necessary factor for the occurrence of hypertensive disease.
Necessary and Sufficient Factors
In rare instances, a factor alone is both necessary and sufficient for disease development. For example, Vibrio Cholerae causing severe watery diarrhea without any additional contributing factors. From figure 3.3, hypertension alone may result in hemorrhagic cerebrovascular disease.
Necessary but Not Sufficient
Some factors are necessary but not solely responsible for causing a disease. Multiple factors are required, typically in a specific sequence, for the disease to manifest. Carcinogenesis is an example, involving multiple steps like oxidative stress, impaired repair mechanisms, and abnormal cellular proliferation. In order for thrombosis to cause cerebrovascular disease such as a transient ischemic attack (TIA), atherosclerosis and clotlysis (thrombolysis) must be present. Hence, these are necessary factors for TIA but are not sufficient to cause TIA on their own.
Sufficient but Not Necessary
Certain factors, like exposure to benzene or radiation, can independently cause a disease like leukemia, but other factors can also cause the disease without their presence. Again, using Figure 3.3, Atherosclerosis is a sufficient factor for myocardial infarction however, it is not necessary since hypertension or thrombolysis can cause myocardial infarction in the absence of atherosclerosis.
Neither Sufficient nor Necessary
Some factors cannot cause a disease on their own, nor are they the only factors contributing to the disease. This model is typical for chronic disease relationships where multiple factors interact to influence the disease’s development.
Principles in Establishing Causation
Principles guiding the establishment of causation have evolved through various seminal works, such as those developed by the Surgeon General Advisory Committee on Smoking and Health in the United States. This committee, notably active in 1964, significantly contributed to understanding the correlation between smoking and numerous health ailments, particularly lung cancer and heart disease. The report titled “Smoking and Health: Report of the Advisory Committee of the Surgeon General of the Public Health Service” presented several criteria instrumental in solidifying the causal relationship between smoking and adverse health outcomes. These criteria encompassed Consistency, Strength of Association, Dose-Response Relationship, Temporal Relationship, Biological Plausibility, and Coherence.
Subsequently, the British scientist Sir Bradford Hill further refined and broadened this set of criteria. The assessment of whether an association is causal relies on the nature and breadth of the available evidence. Below, we will explore this expanded list to scrutinize and understand these critical components influencing the determination of causation
Establishing Causation
Establishing causation involves several stages.
Stage 1
- Establish Statistical Association (SA)
- Employ descriptive statistics such as the median, mean, and mode, which summarize data distribution
- Use inferential statistics to draw conclusions from observed data.
In this stage, researchers aim to identify a statistical association between exposure to a potential causal factor and the occurrence of a disease.
Stage 2
- Once a statistical association is identified, the next step is to establish Causal Inference.
- This phase involves exploring the possibility of a cause-and-effect relationship between the identified exposure and disease occurrence.
It entails rigorous analysis to determine if the observed statistical association is indicative of a genuine causal link or if other factors might explain the relationship observed statistically
The Bradford Hill Criteria for Causation
Strength of Association
The strength of association represents the likelihood of a causal relationship between a risk factor and a disease outcome. Measured through metrics like relative risk or odds ratio, a stronger association implies a higher probability of causation.
For instance, in observations of chimney sweeps who succumbed to scrotal cancer at rates 400 times higher than the general population, the robust correlation between scrotal cancer and chimney sweeps highlighted a compelling connection due to environmental exposure. Similarly, John Snow’s evaluation during the 1855 cholera outbreak, while not significantly impacting mortality rates, still showcased a compelling association with contaminated water.
Consistency with Other Studies
The consistency of findings across various studies strengthens the evidence for causation. Consistent observations in diverse populations and different settings bolster the likelihood of a genuine effect, provided these studies are free from bias. In essence, it is the coherence and harmony of findings across various research investigations or studies regarding a specific relationship between a risk factor and a disease outcome. Here’s an example:
Suppose several epidemiological studies conducted across different regions and diverse populations consistently show a strong positive association between excessive sugar consumption and the risk of developing type 2 diabetes mellitus. Each study, regardless of its location or the demographic characteristics of the subjects, consistently demonstrates a higher incidence of diabetes among individuals with high sugar intake compared to those with lower intake levels. This consistency in findings across multiple studies, despite different methodologies or varied populations, provides compelling evidence supporting the association between sugar consumption and the risk of developing type 2 diabetes mellitus.
Specificity of Association
This criterion proposes that a causal factor should predominantly lead to a single disease, indicating that the disease arises from a singular cause. It suggests that when a very particular population at a distinct location experiences a specific disease without any other apparent explanations, it reinforces the likelihood of a causal relationship. The more precise and exclusive the link between a factor and its resulting effect, the stronger the likelihood of a causal connection.
However, this criterion’s strength is often weakened due to the intricate interplay between disease occurrence and the interaction of various exposures. While specificity adds weight to establishing causal links, it is essential to acknowledge that diseases can result from multiple factors and may not always align with a single causative agent. Thus, while specificity is supportive, it is not always definitive in establishing causation due to the complex nature of disease development.
For example, in the context of HIV causing AIDS, the specific relationship between HIV viral load and its depletion of CD4 cells illustrates the high specificity of HIV for causing AIDS.
Temporal Relationship
The temporal relationship criterion stipulates that the exposure to the causal factor should precede the onset of the disease. This concept poses challenges, especially in case-control studies compared to cohort studies. This criterion essentially requires that the cause occurs before the effect. It is crucial for researchers to establish that the cause was present prior to the onset of the disease. If there is an anticipated delay between the cause and the expected effect, the effect should occur after that duration.
Example- In an investigation regarding the association between smoking and lung cancer, researchers conducted a cohort study where individuals who smoked were followed up for several years to observe the development of lung cancer. The temporal relationship criterion was met as smoking (the cause) preceded the diagnosis of lung cancer (the effect) in the majority of cases, establishing a plausible temporal sequence..
Biological Plausibility
Biological plausibility signifies that a causal explanation aligns with our understanding of biological mechanisms and disease processes. It can sometimes be challenging to establish, especially when epidemiological observations precede scientific knowledge of the biology involved.
The discovery of the link between high oxygen levels and retinopathy of prematurity (ROP) in premature infants exemplifies biological plausibility. In the 1940s, it was noted that increased exposure to oxygen-rich environments led to higher rates of ROP among premature infants. The excessive oxygen caused abnormal blood vessel growth in the retina, resulting in this condition. This revelation transformed neonatal care, necessitating meticulous regulation of oxygen levels to prevent ROP while ensuring proper infant development. Today, early detection and intervention are vital in managing ROP and preventing severe vision impairment in premature infants.
During the 1854 cholera outbreak in London, John Snow’s investigation challenged the prevailing belief that diseases like cholera were spread through “miasma” or foul air. He suspected contaminated water as the source and mapped cases around a public water pump on Broad Street. His observations, although predating a complete understanding of microbial contamination, aligned with the emerging germ theory. By removing the pump handle, new cases declined, supporting the theory of waterborne cholera transmission. Snow’s actions and findings provided early evidence that later shaped the biological understanding of disease transmission, emphasizing the plausibility of waterborne cholera, despite limited biological knowledge at the time
Dose-Response Relationship
Dose-response relationship is a principle indicating that as the level, intensity, duration, or overall exposure to an agent increases, so does the associated risk. For instance, in smoking-related lung cancer cases, the risk escalates significantly with higher numbers of cigarettes smoked. The principle implies that with a causal relationship, a larger exposure (dose) would lead to a more pronounced effect on the outcome (response).
For instance, if individuals who smoke lightly or for a shorter duration exhibit a higher risk of lung cancer compared to heavy, long-term smokers, it would raise questions about the causal link between smoking and lung cancer.
Consistency with Other Available Evidence
Consistency with other available evidence in establishing causation implies that the evidence regarding the natural progression, biology, and epidemiology of the disease should align and create a unified picture. The proposed cause-and-effect relationship should not contradict information gathered from diverse sources such as experiments, laboratory studies, clinical trials, pathology reports, and epidemiological data. The harmony among these different sources of knowledge strengthens the likelihood of a genuine causal association between the identified factor and the observed effect.
Experimentation
Experimentation in establishing causation typically involves conducting experimental epidemiological studies, which can sometimes resemble natural experiments where human groups are observed, akin to the work of John Snow. Clinical trials and community trials, when implemented with controlled measures to mitigate confounding factors, serve as robust evidence for causation. However, epidemiological experimentation for establishing causation can be challenging due to practical and ethical concerns. As an alternative, in vitro experiments utilizing animal models were frequently employed to provide additional evidence supporting causation.
Given the increasing concerns regarding animal welfare and ethical considerations related to in vitro experiments involving animals, some researchers are adopting alternative methodologies. One such replacement is the utilization of advanced computational models and simulations. These models employ computational techniques, artificial intelligence, and mathematical algorithms to simulate biological processes, drug interactions, disease mechanisms, and toxicological effects. These computational approaches provide valuable insights, often eliminating the need for animal experimentation while delivering substantial scientific data for research and drug development purposes
Analogy
Analogy in epidemiology signifies similarities found in various studies or situations. For instance, if a pharmaceutical drug like thalidomide causes severe birth defects, similar effects might be expected from other medications. However, it’s important to note that analogy does not offer definitive proof of cause and effect. Rather, establishing causation requires a combination of factors including empirical evidence, judgment, and experimental support. No single criterion alone can be regarded as a decisive condition; rather, a collective evaluation is needed for robust conclusions.
Summary
Establishing causation in research is pivotal for drawing meaningful conclusions from study evidence. An essential aspect of determining causality involves ensuring associations are unaffected by confounding factors. While clinical trials play a crucial role in epidemiological studies to explore causation, they primarily focus on comparing treatment efficacy rather than establishing causality outright. The stringent criteria of randomization and reduced confounding factors in clinical trials contribute to this distinction.
Epidemiological studies are instrumental in identifying links between exposures and diseases within populations, aiding in understanding disease patterns, risk factors, and potential causes. However, epidemiological studies, despite their significance, have limitations in definitively recommending treatments based solely on observed associations.
Published: 2023- 11- 22, Last updated: 2024- 3- 4