Livingston et al.1 studied the state-level correlation between opioid overdose deaths and recreational marijuana legislation. In the article’s introduction, the authors claim that the study will contribute to the literature on “whether cannabis is substituted for opioids in pain management,” and the reader is given the premise that “with this substitution, an immediate reduction in opioid-related poisonings would be expected.” Therefore, those who read the study’s findings—that recreational marijuana legislation was followed by 0.7 fewer opioid overdose deaths statewide per month—are led to believe that the study constitutes evidence that pain-afflicted individuals in Colorado are substituting marijuana for opioids.
This conclusion is an example of the well-known “ecological fallacy”—the often-incorrect assumption that population-level trends will be replicated on the individual level. In fact, despite the negative correlation that Livingston et al.1 found at the population level, the correlation between marijuana use and opioid use among the underlying individuals could be negative, negligible, or even positive. That is, the findings by Livingston et al. do not provide compelling evidence of an individual-level substitution effect between marijuana and opioid use.
The ecological fallacy is explained thoroughly elsewhere,2–4 but the lesson of the ecological fallacy is conveyed nicely in the title of an AJPH letter5 regarding a similar study: “State-Level Relationships Cannot Tell Us Anything About Individuals.” As applied to the article by Livingston et al., we can infer that state-level data surrounding marijuana legislation and opioid death rates cannot tell us anything about individuals’ substitution behaviors.
Studies about marijuana legalization are likely to attract substantial attention from the media, the general public, and activists, many of whom will not appreciate methodological nuances. Therefore, it is important that authors clearly lead their readers away from drawing overly expansive conclusions from research findings. Unfortunately, Livingston et al.1 do the opposite.
ACKNOWLEDGMENTS
T. L. Caputi acknowledges funding from the George J. Mitchell Scholarship.
REFERENCES
- 1.Livingston MD, Barnett TE, Delcher C, Wagenaar AC. Recreational cannabis legalization and opioid-related deaths in Colorado, 2000–2015. Am J Public Health. 2017;107(11):1827–1829. doi: 10.2105/AJPH.2017.304059. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Finney JW, Humphreys K, Kivlahan DR, Harris AH. Why health care process performance measures can have different relationships to outcomes for patients and hospitals: understanding the ecological fallacy. Am J Public Health. 2011;101(9):1635–1642. doi: 10.2105/AJPH.2011.300153. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Caputi TL, Humphreys K. Medicare recipients’ use of medical marijuana. Health Aff (Millwood) 2016;35(10):1936. doi: 10.1377/hlthaff.2016.1108. [DOI] [PubMed] [Google Scholar]
- 4.Caputi TL, Humphreys K. Medical marijuana users are more likely to use prescription drugs medically and non-medically. J Addict Med. doi: 10.1097/ADM.0000000000000405. In Press. [DOI] [PubMed] [Google Scholar]
- 5.Harris AH, Humphreys K, Finney JW. State-level relationships cannot tell us anything about individuals. Am J Public Health. 2015;105(4):e8. doi: 10.2105/AJPH.2015.302604. [DOI] [PMC free article] [PubMed] [Google Scholar]