Comment on ‘Effect of marijuana use on cardiovascular and cerebrovascular mortality’

Yankey and co-authors reported that marijuana use was associated with more than triple the risk of death from hypertension compared to the non-use of marijuana.¹ A major concern about this paper is the rather puzzling description of the study cohort and the mortality follow-up, raising questions about the validity of the analysis. The authors describe a cohort of 1213 eligible participants who responded to questions on marijuana use during the 2005 US National Health and Nutrition Examination Survey (NHANES), apparently the first year that marijuana use was queried. The participant database was linked to the 2011 public use linked mortality data of the National Center for Health Statistics. The follow-up period for the cohort of eligible participants was 1991–2011. There were 332 deaths (27.4% of the cohort).

Since the marijuana questions were first included in NHANES in 2005, it is not possible that the eligible participants died prior to 2005. It is clear that cohort follow-up occurred prior to 2005 because there was an average of 16.1 years of follow-up (1213/19,569 person-years). It is also clear that deaths occurring prior to 2005 were included in the analysis because the authors noted that the International Classification of Disease version 9 (ICD-9) codes were used to code deaths occurring prior to 1999, and the cumulative hazards graphs shown in Figure 1 included the full 20 years of follow-up from 1991 to 2011.

Interpretation of the study is further complicated by the observation that the number of deaths in the study cohort was far higher than would have been expected based on United States mortality data. In particular, we examined US mortality data under the assumption that mortality follow-up began in 1991 for the entire group that was aged 25 years or less and that they received follow-up for mortality for up to 20 years, until 2011. Each person in that group would have been no older than 45 years in 2011. The annual age-specific total mortality of the 35–44 years age group during the 2000–2015 period ranged from 170 to as high as almost 204 per 100,000 person-years.² Age-specific mortality was, as expected, lower in younger age groups. At a rate of 204 deaths per 100,000 person-years, we would expect no more than 4080 deaths per 100,000 persons in this age group over 20 years, or about a 4% death rate. Applying 4% to the total of 226 individuals noted in the 25 years or less age group, we would have expected no more than nine deaths. The actual number of deaths reported in Table 1 for this age group was 63, seven times as high. Therefore, the number of deaths reported in the paper for this cohort is implausible because it is so much higher than that which would be expected based on published US data.

The most reasonable conclusion from the assessment provided in the prior paragraphs is that the cohort includes another group of people, not described in this paper, for whom there are no exposure data for marijuana, explaining the occurrence of deaths prior to 2005 and the high number of deaths. This is a potential fatal flaw in the analysis and needs to be explained by the authors.

In conclusion, there was an apparent large gap in the description of the study cohort and the follow-up for mortality, raising a major question about the validity of the analysis and the study conclusions.

Author contribution

SS contributed to the conception of the work and drafted the letter to the editor. JSR and RA critically revised the letter. All authors gave final approval and agree to be accountable for all aspects of the work ensuring integrity and accuracy.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

References