ArticleIn this article, Toxicologist and Drug Safety Expert, Michael Whitekus discusses the rising use of prescription drugs in the United States and introduces two models that can be used to evaluate causality as it relates to drug use.
Association vs. Causation
When individuals develop debilitating health conditions, our experts in toxicology are often retained to determine if the negative health outcomes were caused by the use of specific prescription drugs or exposure to particular chemical agents.
There is a tendency among non-medical professionals to look at a temporal association as definitive proof of causation. However, the fact that a negative health outcome developed along with, or closely following, exposure to a particular substance does not prove causality.
Prescription Drug Usage in the U.S.
Prescription drug use has increased significantly in the United States in recent years. Comparing the intervals 1988–1994 and 2007–2010, the percentage of Americans not taking any prescription drugs in the past 30 days decreased from 61% to 53% while those taking five or more drugs increased from 4% to 10% (see figure 1a). This is significant because adverse drug reactions exponentially increase when multiple drugs are used at the same time.
Prescription drug use is highest among older age groups. In 2007-2010, ~50% of adults aged 45 years and over reported taking one to four drugs in the past 30 days, compared with just ~23% of children. The percentage of people taking five or more drugs in the past 30 days increases with age, from less than 1% of children to ~40% of adults aged 65 and over (see figure 1b).
Drugs play an increasing role in the long-term treatment of chronic conditions. Some of the most commonly used prescription drug classes include cardiovascular, hyperlipidemia, anti-acid reflux, antidiabetes, analgesics and antidepressants. Among adults aged 65 and over, ~70% took at least one cardiovascular agent and ~47% took a cholesterol-lowering drug in the past 30 days in 2007–2010. Comparing the intervals 1988-1994 and 2007-2010 for adults aged 65 and over, all drug classes showed an increase in use (as did the 18-64 age group) with cardiovascular, 1.4-fold; cholesterol lowering, 7.9-fold; anti-acid, 2.9-fold; antidiabetes, 2-fold; anticoagulants, 3.0-fold; analgesics, 1.3-fold and antidepressants, 4.6-fold (see figure 2).
Not only has prescription drug use increased significantly but adverse drug events (any undesirable experience associated with the use of a drug) submitted to the FDA have also increased from ~13,000 in 2004 to ~52,000 in 2014. (see figure 3).
Increased single and multiple drug use would be expected to increase the number of adverse drug events. On the next page we introduce two tools to help you evaluate the causality of any potential adverse drug case.
Evaluation of Causality - Hill Criteria
A set of criteria to determine if an association is causal was published in 1965 by epidemiologist Bradford Hill. Hill proposed nine criteria which outlined how to critically evaluate the causal link between a specific factor and some undesirable outcome. The principles contained in Hill’s nine criteria (originally applied to cancer causation) are applicable in evaluating causality in a number of different areas, including adverse drug reactions.
- Strength. The stronger the association, the more likely it is that the relation of “A” to “B” is causal. For example, Hill indicates that chimney sweeps died of scrotal cancer at rates 200 times of those who were not exposed to tar or mineral oils and smokers died of lung cancer at rates 10-30 times the rate of nonsmokers.
- Consistency. The association is consistent when results are replicated in different studies in different settings using different methods. If a relationship is causal, we would expect to find it consistently in different studies and among different populations. There was consistency in the many cigarette studies conducted to demonstrate that smoking causes lung cancer.
- Specificity. This is established when a single cause produces a specific effect. When specificity of an association is present, it provides additional support for a causal relationship. For example, there is a high specificity between asbestos exposure and malignant mesothelioma. However, the absence of specificity in no way negates a causal relationship.
- Temporality. The cause must proceed the effect and in the time period dictated by the mechanism of action. Flu symptoms typically develop ~1-4 days after exposure to virus.
- Dose-response. There should be a direct relationship between exposure to the risk factor and the extent of the effect. A little exposure should result in a little effect, while a large exposure should cause a large effect. For example Hill found that the death rate from lung cancer was 9-10 times greater in smokers and 20-30 times greater in heavy smokers then in non-smokers.
- Plausibility. A biologically plausible association is one in which a supportable mechanism can at least be hypothesized without biological evidence for its existence.
- Coherence. Coherence is clearly established when the mechanism of disease is defined but can also exist when there is enough evidence to support a plausible mechanism without a detailed understanding of each step in the process. The proposed mechanism should not conflict with current understanding of the system under investigation and is strongest when there are no plausible competing theories.
- Experiment. Cell based assays, animal studies, clinical trials, epidemiology studies and case reports can demonstrate a deleterious effect due to a specific treatment. Additional confidence for causality can be gained if the effect can be altered (prevented or ameliorated) by an appropriate experimental intervention (or drug therapy). N-acetyl cysteine is an effective antidote for acetaminophen overdose when administered within 8-10 hours after ingestion,
- Analogy. A clear cut analogy may add to the weight of evidence for an otherwise weak association. If smoking causes cancer in active smokers then second hand-smoke mediated by the same mechanism of action could have a similar albeit lesser effect.
Adverse Drug Reaction Probability Scale - Naranjo Scale
The Adverse Drug Reaction (ADR) Probability Scale or Naranjo scale is a set of 10 questions designed to help standardize assessment of causality for all adverse drug reactions. Once the questions are answered a score is generated which can be used to help assess the certainty that a particular outcome was caused by a particular agent.
- US Department of Health and Human Resources. Health, United States, 2013. [Online]. Available: http://www.cdc.gov/nchs/data/hus/hus13.pdf. [Accessed 20 Oct 2014].
- Federal Drug Administration. OpenFDA, 2004-2014. [Online]. Available: https://open.fda.gov/drug/event/. [Accessed 20 Oct 2014].
- Hill AB. The Environment and Disease: Association or Causation? Pro R Soc Med (1965) 58(5): 295-300.
- Naranjo CA, et al., A method for estimating the probabilty of adverse drug reactions, Clin. Pharmacol. Ther. (1981) 30(2): 239-245.
- Hursting SD, Weed DL. Biologic Plausibility in Causal Inference: Current Method and Practice. Am J Epidemiol (1998) 147(5): 415-425.
Dr. Whitekus is a toxicologist with expertise and experience in the fields of drug safety, pharmacology, inhalation toxicology, and environmental contaminants. Dr. Whitekus has held various key drug safety positons in industry, most recently working at Pfizer, Inc. as a drug safety team lead. Dr. Whitekus has authored or reviewed ~180 drug safety reports, authored or contributed to FDA drug applications (IND’s/NDA’s) and has evaluated numerous drugs (e.g., Lyrica® and Milnacipran®) in multiple therapeutic areas including allergy, vaccine development, oncology, CNS, cardiovascular disease and inflammation. Dr. Whitekus applies his expertise in drug safety and toxicology towards resolving disputes relating to adverse drug events and exposure to chemical and environmental toxins.