We used linear mixed-effects models to assess changes in continuous outcomes over time: post-bronchodilator FEV1, post-bronchodilator FVC, SGRQ total score, CAT score, and HRCT metrics. Linear mixed-effects models, specifically proportional odds models, were used to assess changes in respiratory symptoms over time. In assessing whether marijuana use among tobacco smoking participants without COPD at baseline increased the risk of subsequent development of COPD, the primary outcome was time to development of airflow obstruction, defined by a post-bronchodilator FEV1/FVC<0.70.We used zero-inflated negative binomial models to compare the rate of exacerbations between CMSs, FMSs, and NMSs. Exacerbations were classified as moderate , severe , and total . To assess dose response relationships, the same models were used with the primary predictor of interest being categorical joint-year history at baseline.A consort diagram describing the derivation of the study cohort is shown in Figure 1. At enrollment, CMSs, when compared with NMSs, tended to be younger and more often current tobacco smokers, men, and Black .They also had fewer exacerbations during the year prior to enrollment, had a better FEV1, less frequent airflow obstruction, and less emphysema and air trapping, but had similar levels of respiratory symptoms. Similar findings were noted in comparison of FMSs with NMSs. Due to incomplete reporting, calculating the cumulative lifetime amount of marijuana use in joint years was not possible for all participants, so that the number of those classified by joint-year category is lower than that of the total analysis sample. Among those with the heaviest marijuana use , directionally similar baseline differences were noted in age, sex, the proportion of Black participants,industrial rolling racks and current tobacco-smoking status compared to those with 0 joint years, as were found in comparison between CMSs and FMSs with NMSs . The estimated rates of change in continuous outcomes by baseline marijuana-smoking status are shown in Table 3A.
While numerically higher annual rates of FEV1 and FVC decline and higher rates of worsening CAT and total SGRQ scores were found comparing CMSs with NMSs, these differences were neither clinically nor statistically significant . Similar rates of change in these parameters were found on comparison of FMSs with NMSs. Estimated rates of change in continuous outcomes between joint-year-based categories were similar across all joint-year groups and between groups . Estimated annualized FEV1 decline during follow-up by marijuana joint years stratified by former and current tobacco-smoking history were similar, irrespective of tobacco smoking status . Estimated participant-specific yearly changes in odds for worsening respiratory symptoms during follow-up compared to the baseline visit by baseline marijuana status and baseline joint years are shown in e-Table 9A and B and e-Figures 1 and 2 in the online supplement, respectively. The odds over time of more cough and sputum, but not more wheeze or dyspnea, were significantly higher in CMSs compared to FMSs or NMSs , while no significant differences were found across the different joint-year categories that included both CMSs and FMSs . Estimated yearly changes in CAT and SGRQ scores were not significantly different across marijuana-smoking status and joint-year categories as shown both in Tables 3A and B in the online supplement, respectively, and e-Tables 5 and 6 in the online supplement, respectively. Our analysis showed nominally less emphysema, air trapping, and functional small airways disease progression without statistical significance among CMSs compared to NMSs. Similarly, a comparison between NMSs, FMSs, and CMSs showed no significantly different changes in HRCT metrics, except for unadjusted increased total tissue volume loss among FMSs compared to NMSs . No difference in tissue volume loss between CMSs and NMSs was found.
Estimated rates of change in HRCT metrics were generally similar across all joint-year groups , except for a higher rate of increase in PRMf SAD on comparison of those with ≥20 joint years versus 0 joint years , with a between-group difference 0.393 when unadjusted for multiple testing . Estimated yearly rates of 1 or more total or severe exacerbations during the first 365 days or the total follow-up period by baseline marijuana-smoking status and marijuana joint years are shown in Table 4 A and B and e-Figures 3 and 4 in the online supplement. While rates of total and severe exacerbations were numerically lower among both CMSs and FMSs versus NMSs during the first follow-up year, and severe exacerbation rates were slightly higher among CMSs versus NMSs during the total follow-up period, none of these differences were statistically significant . Estimated rates of total and severe exacerbations were numerically higher among those with ≥20 versus those with 0 joint years during the first follow-up year. During the total follow-up period, rates of total exacerbations, but not severe exacerbations, were slightly higher among those with ≥20 versus those with 0 joint years. However, none of these between-group differences were statistically significant .Estimated hazard ratios for the development of COPD during follow-up by baseline marijuana-smoking status and joint years among participants without spirometric evidence of COPD at baseline are shown in Table 5 and e-Figures 5 and 6 in the online supplement. The odds of developing COPD by spirometric criteria were lower among CMSs and FMSs versus NMSs, as well as among those with ≥20 versus those with 0 joint years, although these differences were not statistically significant.The increasing prevalence of marijuana smoking among adolescents and adults,including aging adults,in the wake of a growing number of states legalizing marijuana use underscores the need to better understand the impact of marijuana use on lung health. This need is particularly evident among adult tobacco smokers in their mid- and older life who have been understudied previously.
The current analysis of the pulmonary consequences of marijuana smoking in the SPIROMICS cohort of current and former tobacco smokers with or at high risk of developing COPD is a longitudinal extension of a cross-sectional analysis of the baseline findings in the same cohort.10 While the latter cross-sectional study failed to identify deleterious effects of concomitant marijuana smoking on lung function or baseline structural radiographic abnormalities when compared with the effect of tobacco smoking alone, it could not answer the question of whether marijuana drying racks affects changes in these outcomes over 1 to several years of follow-up. In addition, the current study overlaps to some extent with a recent longitudinal analysis focused mainly on the trajectory of lung function in SPIROMICS participants limited to those with ≥3 spirometry visits.By including all those participants with ≥2, rather than only ≥3, spirometry visits at least 1 year apart, the current study has the advantage of including in the analysis larger numbers of CMSs and FMSs, most importantly of those heavy MSs with ≥20 joint years, in an effort to achieve greater statistical power in examining the influence of marijuana smoking on lung function decline. Furthermore, the current study examined changes in respiratory symptoms and HRCT metrics during follow-up that were not included in the previous report. Our study revealed trends toward higher rates of decline in post-bronchodilator FEV1 and worsening CAT and SGRQ scores among CMSs compared with NMSs and contrastingly, smaller rates of change in percentage of emphysema and functional small airways disease. However, none of these differences were statistically significant. Similarly, when we compared different categories of lifetime cumulative amounts of marijuana smoking, no significant differences were noted in rates of change in lung function, CAT or SGRQ scores, or HRCT metrics, except for an increase in PRMfSAD among the heaviest marijuana-smoking category in comparison to those with 0 joint years. It is noteworthy that significantly higher odds of worsening cough and sputum were noted among CMSs in comparison with both NMSs and FMSs, but not between FMSs and NMSs. The latter finding is consistent with previous data showing a significant reduction in symptoms of chronic bronchitis after cessation of marijuana smoking. Although some numerical differences were noted in rates of exacerbations across marijuana-use status and joint-year categories, none of the between-group differences were statistically significant. Finally, while the probability of subsequently developing COPD among tobacco smokers without COPD at baseline was lower among CMSs and FMSs compared with NMSs, as well as between the heaviest marijuana smokers versus those with no history of marijuana smoking, none of these differences reached statistical significance.
Taken together, the aforementioned data failed to demonstrate that marijuana smoking of any lifetime cumulative amount had a demonstrable effect on changes over time in clinical outcomes relevant to COPD, including respiratory symptoms, health status, HRCT metrics, or frequency of exacerbations. Our failure to find any impact of even heavy marijuana smoking on lung function decline in ever-tobacco smokers with or at risk of COPD differs substantially from the findings of Tan et al.The authors demonstrated a dose-response effect of marijuana on lung function decline in the CanCOLD study subcohort with a significantly greater rate of decline in FEV1 only among those with ≥20 joint years compared to those who never used marijuana . Surprisingly, in the same study, among those with ≥20 joint years of marijuana smoking, the rates of FEV1 decline were very similar for CMSs and FMSs, compared to NMSs. In contrast, the average rate of FEV1 decline among the heaviest former tobacco smokers was substantially lower than that of the current tobacco smokers. Since tobacco smokers with COPD have a substantial reduction in the rate of FEV1 decline after sustained smoking cessation,34 the disparate findings of Tan et al15 comparing the impact of quitting marijuana with that of quitting tobacco is surprising. The absence of a difference in the rates of decline between their current and former marijuana smoking participants, most of whom were dual smokers of marijuana and tobacco, may be a reflection of the impact of continuing tobacco smoking among those who had quit using marijuana rather than of an enduring effect of marijuana among the quitters. It is also noteworthy that the number of SPIROMICS participants who were particularly heavy marijuana smokers was almost 3 times higher than the number of CanCOLD participants with a heavy marijuana smoking history , suggesting that our analysis of the impact of heavy marijuana use on lung function decline had greater statistical power. Finally, while the reference control group in our analysis of FEV1 decline in relation to marijuana smoking consisted of NMSs with a history of at least 20 pack years of tobacco smoking, the reference group in the analysis reported by Tan et al was comprised solely of never smokers of either substance. Thus, our aim was to examine whether marijuana smoking had an impact on the progression or development of COPD in current or former smokers of tobacco who already had COPD or were at increased risk of developing COPD, while Tan et al evaluated whether marijuana smoking led to an accelerated decline in lung function in a population of whom 43% were nonsmokers of tobacco.Our findings are also at odds with the results of another recent study by Winhusen et al.Using data from electronic health records of patients treated in an integrated health care system located in Northeast Ohio, the authors reported a significantly greater risk for COPD, defined using International Classification of Diseases, 9th and 10th revisions’ codes, among persons with a diagnosis of cannabis use disorder compared to propensity-matched controls in a subgroup of patients with a diagnosis of tobacco use disorder . These findings imply an additive effect of cannabis on top of tobacco use. However, limitations of the latter study include misclassification of COPD in the absence of spirometry data, suggested by the relatively young average age of the authors’ analysis population versus ours , as well as the absence of data on the route of cannabis administration and the intensity and duration of its use. The marked disparity of these results with ours underscores the need for additional study. The possibility of a dose–response impact of marijuana exposure is suggested by our finding of a significantly larger effect of ≥20 joint years on PRMfSAD in comparison with 0 joint years , consistent with a deleterious effect of heavy marijuana use on small airways. The latter observation is consistent with the recently reported finding in a New Zealand birth cohort at age 45 years of an association of lifetime cannabis use, adjusted for tobacco pack years, with pre-bronchodilator peripheral airways resistance and reactance using impulse oscillometry.