Assessing Evidence for Causation

When determining whether there is sufficient evidence to imply a causal relationship between a variable and an outcome, it is important to follow a systematic process that utilizes the criteria for causation. Two examples of how this process can be applied are given below. Each has been prepared following an extensive search and analysis of the existing literature.

Example # 1 “Cigarette Smoking and Periodontitis”

Bradford Hill Criteria Cigarette Smoking and Periodontitis
Temporal Relationship (the cause must precede the outcome) Yes (Numerous observational studies)
Strength (Rank the evidence according to study Level and RR’s or OR’s reported) Strong BUT (only level 2 studies available, but OR’s are strong)
Dose-response (Biological gradient) (Disease worsens with amount of exposure) Yes (Numerous studies demonstrating this)
Consistency (Relationship is repeatedly observed) Yes (Numerous studies consistently have the same results)
Plausibility (From a biological standpoint, the relationship is possible) Yes (Numerous studies showing biological plausibility)
Experiment (Causality is more likely if demonstrated in randomized controlled trials or systematic reviews of RCT’s) No (No RCT’s or systematic reviews of RCT’s available)
Specificity (A specific factor influences a particular outcome) No (Cannot say with certainty that in every case, smoking causes periodontitis)
Coherence (Studies must not contradict each other) Yes (Studies do not contradict each other)

Conclusions from this analysis: Due to ethical reasons, it is not possible to conduct experimental studies where some participants are assigned to smoke and others not to smoke and then wait for years to see if they develop periodontal disease. Thus, the analysis falls short in the category of experiment, however the evidence is very strong in observational studies (OR’s >4). This analysis is similar to the one conducted by Bradford Hill regarding cigarette smoking and lung cancer. This analysis reveals that cigarette smoking is not only plausible but consistently associated with an increased prevalence/severity of periodontitis demonstrating a dose-response relationship. Thus, one can say with confidence, that it is suspected of playing a causal role. Thus, smoking is now listed as one of the few true Risk Factors for periodontitis.

Example # 2 “Vaccines and Autism”

Bradford Hill Criteria Vaccines and Autism
Temporal Relationship Yes (Numerous observational studies)
Strength Weak
Dose-response (Biological gradient) No
Consistency No (Study findings not consistent)
Plausibility Yes
Experiment No (No RCT’s or cohort studies)
Specificity No
Coherence Unknown

Conclusions from this analysis: Similar to Example #1, it would be unethical to conduct RCT’S, therefore there are only observational studies available. However, in comparison to the first example, although it is plausible biologically, the study results have been weak and extremely inconsistent. Additionally, dose-response and specificity have not been shown and coherence, is unknown. Thus, one can state with confidence that there is not sufficient evidence to determine or even suspect a causal role between vaccines and autism.