Trends in Prevalence and Outcomes of Cannabis Use Among Chronic Obstructive Pulmonary Disease Hospitalizations: A Nationwide Population-Based Study 2005–2014

Abstract

Background: Chronic obstructive pulmonary disease (COPD) is the fourth leading cause of mortality in the United States. Due to the ongoing legalization of cannabis, its acceptance, availability, and use in the in-patient population are on the rise. In this retrospective study, we investigated the association of cannabis use with important outcomes in COPD hospitalizations.

Methods: The National Inpatient Sample (NIS) data were analyzed from 2005 to 2014. The primary outcome of interest was the trends and outcomes of cannabis use among COPD hospitalizations, including in-hospital mortality, pneumonia, sepsis, and respiratory failure.

Results: We identified 6,073,862 hospitalizations, 18 years of age or older, with COPD using hospital discharge codes. Of these, 6,049,316 (99.6%) were without cannabis use, and 24,546 (0.4%) were admitted with cannabis use. The majority of COPD hospitalizations with cannabis use were aged 50–64 (60%). Cannabis use was associated with lower odds of in-hospital mortality (odds ratio [OR] 0.624 [95% confidence interval (CI) 0.407–0.958]; p=0.0309) and pneumonia (OR 0.882 [95% CI 0.806–0.964]; p=0.0059) among COPD hospitalizations. Cannabis use also had lower odds of sepsis (OR 0.749 [95% CI 0.523–1.071]; p=0.1127) and acute respiratory failure (OR 0.995 [95% CI 0.877–1.13]; p=0.9411), but it was not statistically significant.

Conclusions: Among hospitalized patients with a diagnosis of COPD, cannabis users had statistically significant lower odds of in-hospital mortality and pneumonia compared to noncannabis users. The association between cannabis use and these favorable outcomes deserves further study to understand the interaction between cannabis use and COPD.

Introduction

Cannabis is the second most widely smoked substance in our society after tobacco.¹ The acceptance, availability, and use of cannabis have been increasing with ongoing legalization throughout the United States.² As per the most recent data from 2017, 14.6% of adults in the United States reported cannabis use.³ There has also been an increase in cannabis use among hospitalized patients, with an increasing trend toward older and sicker patients with increasing rates of moderate to severe disability.⁴ Similar to traditional cigarettes, cannabis smoke contains significant amounts of volatile constituents (including ammonia, hydrocyanic acid, and nitrosamines) and tar components (including phenols, naphthalene, and the carcinogenic benzopyrene and benzanthracene). These substances promote airway edema, inflammation, and can impair bactericidal activity.^5,6 Cannabis smoke may promote the development of airflow obstruction in some individuals. A cross-sectional population-based study showed smoking both cannabis and tobacco was associated with a greater risk of chronic obstructive pulmonary disease (COPD) than smoking tobacco alone, suggesting an additive effect of the two substances.^7,8 Other studies have raised concern over the potentially harmful effects of cannabis smoke on the lung, including risk of lower respiratory tract infections and lung cancer.^5,9–12

COPD has become the fourth leading cause of mortality in the United States, leading to more than 700,000 hospitalizations per year.^13,14 Hospitalized COPD patients with higher comorbidity burden have a higher risk for mortality, hospital length of stay (LOS), readmissions, and higher health expenditures.^15–17

With regard to cannabis use, studies reporting its impact on COPD are variable and limited, and no studies have specifically examined the effects of cannabis use on important hospital outcomes in patients with COPD. We hypothesized that cannabis use among patients hospitalized for COPD would have worse outcomes in terms of mortality, pneumonia, sepsis, and acute respiratory failure compared to noncannabis users.

In this study, we aimed to measure the prevalence of and outcomes associated with cannabis use in patients hospitalized with COPD, as determined by hospital discharge coding over a 10-year (2005–2014) period.

Methods

Data source

A description of the National Inpatient Sample (NIS) database has been elaborated in prior studies.^4,18–20 The NIS is one of the largest, all-payer publicly available in-patient care database for the United States and is maintained by the Agency for Health Care Quality and Research (AHRQ).²¹ The NIS includes a 20% stratified random sample of all in-patient hospitalizations from 46 states in the United States and contains information on over 7 million hospital discharges per year. Each observation denotes a hospitalization with one primary diagnosis, up to 29 secondary diagnoses, and 15 procedure diagnoses with International Classification of Disease, 9th Revision, Clinical Modification (ICD-9-CM) codes. The validated discharge weights provided by the Healthcare Cost and Utilization Project-NIS database were used to generate national estimates for 95% of hospitalizations nationwide.²² Ethical clearance and patient consent were not sought as the NIS HCUP database contain de-identified patient data.

Study population

For our analysis, we only used NIS data from 2005 to 2014. Similar to previous studies, we used the ICD-9-CM code 491.x, 492.x, 492.0, and 496 to identify hospitalizations involving hospitalizations with principal diagnosis (dx1) of COPD.^18,19 The variables for hospitalization demographics were provided in the dataset (e.g., age, gender, LOS). The use of cannabis was identified using cannabis abuse/dependence diagnoses (“continuous,” “episodic,” “unspecified”) with the ICD-9-CM codes' 304.30, 304.31, 304.32, 305.20, 305.21, and 305.22.⁴ All hospitalizations studied were older than or equal to 18 years of age.

Outcome measured

The temporal trends in the prevalence of cannabis abuse among COPD hospitalizations in the overall cohort were studied. Next, we analyzed the demographics and outcomes of COPD hospitalizations with and without cannabis use. Our primary outcome of interest was all-cause in-patient mortality defined as “died” during the hospitalization encounter in the NIS database. The incidence of pneumonia (003.22, 481.0, 513.0, 480.xx, 482.xx, 483.xx, 485.xx, 486.xx), incidence of respiratory failure (518.81,518.8), and incidence of sepsis/bacteremia (038.xx and 790.7) were studied based on the secondary diagnosis (dx2–dx30). The ICD-9-CM codes used to define each complication/outcome are provided in Supplementary Table S1.

Statistical analysis

Survey analysis methods were used to account for the clustering and stratification of encounters for all continuous and categorical variables. SAS 9.4 (SAS Institute, Inc., Cary, NC) software was used to perform statistical analysis. We used sampling weights to estimate trends and national estimates to account for sampling design changes as recommended by the AHRQ. The demographics and comorbid diseases were compared using the chi-square test for categorical variables and Student's t-test for continuous variables. Multivariate logistic regression method was performed in SAS (proc survey logistic) to assess the association between cannabis use and in-hospital mortality after including the other variables for potential confounders. C-statistic was used for goodness of the model fit for a binary outcome. We used the weighted sample and conducted analysis with factoring the age to get a better model in predicting outcomes. A two-tailed p-value <0.05 was considered statistically significant. A checklist provided by the AHRQ was used for performing all analyses to ensure the appropriateness.²³

Results

From January 2005 to December 2014, there were an estimated 6,073,862 hospitalizations aged 18 years or older across the United States with COPD. Of these, 6,049,316 (99.6%) hospitalizations were without cannabis use, and 24,546 (0.4%) were admitted with cannabis use (Table 1). The mean age (years) was lower in hospitalizations with cannabis use (53.56 vs. 68.93, p<0.0001). Out of a total of 24,546 COPD hospitalizations with cannabis use, the majority were aged 50–64 years (n=14,677; 60%). Out of a total 6,049,316 COPD hospitalizations without cannabis use, majority were aged 65–79 years (n=2,629,594; 43%). Overall, there were a higher number of female hospitalizations than males (55.9% vs. 44.1%). The majority of COPD hospitalizations with and without cannabis use were Caucasian (55% and 69%, respectively). The mean LOS in days was lower in hospitalizations with cannabis use ([3.69 vs. 4.55]; p<0.0001). Over a period of 10 years, we noted an increasing trend in the prevalence of cannabis use and mean age among COPD hospitalizations from 0.139% and 49 years in 2005 to 0.777% and 55 years in 2014, respectively (Table 2 and Fig. 1).

Table 1.

Demographic Characteristics of Chronic Obstructive Pulmonary Disease Hospitalizations With and Without Marijuana Use

Variables	COPD without marijuana use		COPD with marijuana use		p
	N	%	N	%
Unweighted index admissions	1,247,879		5028
Weighted index admissions	6,049,316		24,546
Age in years at admission
Mean age (in years)	68.93		53.56		<0.0001
18–34	19,997	0.33	939	3.80	<0.0001
35–49	347,572	5.75	6148	25.05
50–64	1,761,732	29.12	14,677	59.79
65–79	2,629,594	43.47	2686	10.94
>80	1,290,421	21.33	96	0.39
Died during hospitalization
Did not die	5,952,014	98.39	24,394	99.38
Died	94,386	1.56	126.00	0.51
Disposition of the patient
Routine	3,963,412	65.52	19,703	80.27	<0.0001
Transfer to short-term hospital	83,694	1.38	269.891	1.10
Transfer other: includes SNF, ICF, and another type of facility	856,744	14.16	1263	5.15
HHC	965,425	1.60	1860	7.58
AMA	80,065	1.32	1288	5.25
Died in hospital	94,386	1.56	125.831	0.51
Discharged alive, destination unknown	21	0.00	0	0.00
Elective versus nonelective admission
Nonelective	5,551,791	91.78	23,623	96.24	<0.0001
Elective	481,059	7.95	870	3.54
Indicator of sex
Male	2,665,747	44.07	15,380	62.66	<0.0001
Female	3,382,605	55.92	9166	37.34
Length of hospital stay
Mean length of stay (days)	4.55		3.69		<0.0001
0–3	2,959,045	48.92	15,047	61.30	<0.0001
4–6	2,034,979	33.64	6750	27.50
7–9	645,988	10.68	1661	6.77
10–12	210,544	3.48	555	2.26
>12	197,216	3.26	534	2.17
Primary expected payer
Medicare	4,312,722	71.29	8257	33.64	<0.0001
Medicaid	608,069	10.05	8401	34.23
Private insurance	784,108	12.96	2948	12.01
Self-pay	182,809	3.02	3315	13.51
No charge	21,665	0.36	386	1.57
Other	128,702	2.13	1138	4.64
Race
White	4,197,765	69.39	13,616	55.47	<0.0001
Black	505,217	8.35	6690	27.25
Hispanic	233,247	3.86	874	3.56
Asian	48,987	0.81	69	0.28
Pacific Islander	29,114	0.48	217	0.88
Other	90,077	1.49	313	1.28
Cost of hospitalization in USD-(mean)	7412.8		6942.4		0.0068
Bed size of the hospital
Small	1,110,105	18.35	3515	14.32	<0.0001
Medium	1,625,061	26.86	6863	27.96
Large	3,287,236	54.34	14,044	57.22
Location/teaching status of the hospital
Rural	1,440,493	23.81	3010	12.26	<0.0001
Urban-nonteaching	2,685,297	44.39	8989	36.62
Urban-teaching	1,896,612	31.35	12,423	50.61

p<0.05 demonstrates that the variables are independent.

AMA, against medical advice; COPD, chronic obstructive pulmonary disease; HHC, home health care; ICF, intermediate care facility; SNF, skilled nursing facility.

Table 2.

Trends of Hospitalization for Chronic Obstructive Pulmonary Disease Admitted with Marijuana Use 2005–2014

Year	Percentage among COPD hospitalizations during each year	Weighted frequency per 100,000 hospitalizations	Weighted frequency	Mean age in years	Total hospitalizations by year
2005	0.139	138.68	790	49.29	569,611
2006	0.199	199.07	1072	50.78	538,504
2007	0.222	221.59	1186	51.17	535,223
2008	0.256	255.58	1654	51.15	647,156
2009	0.313	312.84	2067	52.67	660,728
2010	0.422	421.93	2694	53.27	638,492
2011	0.454	454.48	3009	54.03	662,074
2012	0.595	595.20	3815	54.01	640,965
2013	0.628	627.87	3870	54.98	616,374
2014	0.777	777.36	4390	55.21	564,735

FIG. 1.

Trends in the prevalence of marijuana use among COPD hospitalizations each year between 2005 and 2014. COPD, chronic obstructive pulmonary disease.

Univariate and multivariate logistic regression analyses were performed. In multivariate analysis, outcomes for death, pneumonia, sepsis, and acute respiratory failure were analyzed. All models were adjusted for age, gender, race, comorbidities, residential region, and hospital size (Table 3). Interestingly, similar to Dai and Richter, our study showed that COPD hospitalizations with more medical comorbidities are likely to use cannabis than those without such conditions.²⁴ Among hospitalizations with 1–3 comorbidities, 53% were using cannabis and 74% were not using cannabis. Higher prevalence of cannabis use were found among hospitalizations with 4–6 comorbidities than without cannabis use (41% vs. 24%) (Table 4).

Table 3.

Multivariate Logistic Regression Analysis Showing the Adjusted Odds Ratios Predicting the Specific Outcomes for Chronic Obstructive Pulmonary Disease Hospitalizations

Effects (outcome variable)	Pneumonia		Acute respiratory failure		Sepsis		In-hospital mortality
	OR (95% CI)	p	OR (95% CI)	p	OR (95% CI)	p	OR (95% CI)	p
Marijuana users	0.882 (0.806–0.964)	0.0059	0.995 (0.877–1.13)	0.9411	0.749 (0.523–1.071)	0.1127	0.624 (0.407–0958)	0.0309
Female gender	0.88 (0.869–0.891)	<0.0001	1.002 (0.982–1.022)	0.8825	0.894 (0.856–0.934)	<0.0001	0.847 (0.815–0.88)	<0.0001
Hypertension	0.9 (0.887–0.914)	<0.0001	0.892 (0.872–0.912)	<0.0001	0.651 (0.62–0.683)	<0.0001	0.519 (0.499–0.54)	<0.0001
Uncomplicated diabetes	0.967 (0.952–0.982)	<0.0001	0.898 (0.877–0.919)	<0.0001	0.952 (0.903–1.004)	0.0715	0.71 (0.676–0.746)	<0.0001
Complicated diabetes	0.91 (0.881–0.941)	<0.0001	0.889 (0.846–0.935)	<0.0001	1.093 (0.983–1.215)	0.1008	0.666 (0.596–0.745)	<0.0001
Age	1.013 (1.012–1.013)	<0.0001	0.995 (0.994–0.996)	<0.0001	1.01 (1.008–1.012)	<0.0001	1.047 (1.045–1.049)	<0.0001
Number of chronic conditions	1.001 (0.997–1.005)	0.7608	1.056 (1.05–1.062)	<0.0001	1.113 (1.103–1.123)	<0.0001	1.097 (1.087–1.108)	<0.0001
Liver disease	0.985 (0.945–1.028)	0.4927	0.984 (0.924–1.048)	0.6197	1.376 (1.215–1.558)	<0.0001	1.273 (1.121–1.446)	0.0002
Hospital location—urban non-teaching versus rural	1 (0.954–1.048)	0.0111	1.395 (1.312–1.483)	<0.0001	1.444 (1.33–1.569)	<0.0001	1.002 (0.934–1.076)	0.2425
Hospital location—urban teaching versus rural	0.91 (0.867–0.955)	<0.0001	1.545 (1.451–1.645)	<0.0001	1.594 (1.464–1.735)	<0.0001	1.076 (0.997–1.162)	0.0249
Hospital bed size medium versus small	0.996 (0.951–1.044)	0.4536	1.192 (1.129–1.26)	0.0061	1.23 (1.131–1.337)	0.0099	1.095 (1.011–1.185)	0.3835
Hospital bed size large versus small	0.963 (0.922–1.006)	0.0517	1.26 (1.195–1.329)	<0.0001	1.273 (1.18–1.374)	<0.0001	1.129 (1.057–1.207)	0.0066

CI, confidence interval; OR, odds ratio.

Table 4.

Number of Comorbidities Among the Chronic Obstructive Pulmonary Disease Hospitalizations With and Without Marijuana Use

No. of comorbidities	COPD with marijuana (%)	COPD without marijuana (%)
1–3	52.87	73.68
4–6	41.44	24.33
7–9	5.46	1.94
10–14	0.21	0.03

In-hospital mortality outcome

Cannabis use was associated with lower odds of in-hospital mortality (odds ratio [OR] 0.624 [95% confidence interval, CI: 0.407–0.958]; p 0.0309) among COPD hospitalizations. Apart from cannabis use, female gender, presence of hypertension, and complicated and uncomplicated diabetes mellitus were associated with lower odds of in-hospital mortality. Presence of liver disease (OR 1.273 [95% CI: 1.121–1.446]; p=0.0002) and the number of chronic conditions (OR 1.097 [95% CI: 1.087–1.108]; p<0.0001) were associated with a higher odds of in-hospital mortality. Urban teaching and large-sized hospital were associated with higher odds of mortality when compared to rural (OR 1.076 [95% CI: 0.997–1.162]; p=0.0249) and small-sized hospitals (OR 1.129 [95% CI: 1.057–1.207]; p=0.0066), respectively (Fig. 2).

FIG. 2.

Adjusted odds ratios of clinical variables predicting the mortality for COPD hospitalizations.

Pneumonia outcomes

Cannabis use was associated with lower odds of pneumonia (OR 0.882 [95% CI: 0.806–0.964]; p=0.0059) among COPD hospitalizations. In addition to cannabis use, the female gender, presence of hypertension, uncomplicated, and complicated diabetes mellitus were associated with lower odds of pneumonia.

Sepsis outcomes

Cannabis use was associated with a statistically insignificant lower odds of sepsis (OR 0.749 [95% CI: 0.523–1.071]; p=0.1127). Apart from cannabis use, the female gender and the presence of hypertension were associated with statistically significant lower odds of sepsis. Higher odds of sepsis were seen in hospitalizations with liver disease (OR 1.376 [95% CI: 1.215–1.558]; p<0.0001) and the number of chronic conditions (OR 1.113 [95% CI: 1.103–1.123]; p<0.0001).

Acute respiratory failure outcomes

Cannabis use was associated with a statistically insignificant lower odds of acute respiratory failure (OR 0.995 [95% CI: 0.877–1.13]; p=0.9411). Lower odds of acute respiratory failure were also seen in patients with hypertension and complicated and uncomplicated diabetes mellitus. Again, the number of chronic conditions was associated with higher odds of acute respiratory failure.

Discussion

The results of our study showed that over a 10-year period, cannabis use was increasingly prevalent in patients hospitalized for COPD. Since legalization of cannabis started in November 2012, we believe that legalization would not have affected the prevalence of cannabis use in our sample for the most part of the study period.²⁵ Cannabis use among COPD hospitalizations was associated with lower in-hospital mortality, a decreased risk for pneumonia, and reduced LOS compared to patients without cannabis use. Our findings do not support the hypothesis that cannabis is associated with worse in-hospital outcomes in COPD.

The association between cannabis use and more favorable clinical outcomes was unexpected. After all, there are biologically plausible reasons to expect that cannabis use would have a negative impact on patients hospitalized for COPD. Cannabis smoke leads to the generation of reactive oxygen species, which may induce edema, inflammation, and cell injury in the airways and lungs.^9,26 Tetrahydrocannabinol (THC) also has an immunosuppressant effect on alveolar macrophages with significant impairment in their phagocytic and bactericidal activity.^27,28 Endobronchial biopsies and video-bronchoscopy have shown vascular proliferation, goblet cell hyperplasia, basal cell hyperplasia, loss of cilia, chronic airway inflammation, and submucosal edema.^29,30 There is a concern that cannabis may be contaminated with pathogenic bacteria or fungi-like aspergillus.³¹ This can potentially predispose to an increased risk of pulmonary infections in cannabis users.^32–35

Interestingly, however, our analysis is not alone in showing favorable or neutral associations between clinical outcomes and cannabis use. Similar results have been reported in patient groups other than COPD. For example, among 387,608 hospitalized patients, Vin-Raviv et al. found that cannabis use, defined by ICD codes, was associated with significantly lower odds of experiencing in-hospital mortality compared to nonusers, and significantly reduced the odds of heart failure and cardiac disease. In a systematic review of postmyocardial infarction (MI) outcomes, Pradhan et al. found that in-hospital mortality in patients with MI was significantly reduced among cannabis users compared with nonusers.^36,37 Ajibawo et al. found lower odds of mortality among congestive heart failure hospitalizations with cannabis.³⁸ Taghavi et al. examined the effect of preinjury use of cannabis in trauma patients and reported lower mortality in patients with severe injury.³⁹ Jolley and Welsh reported that cannabis use in patients with human immunodeficiency virus (HIV) was not associated with increased pneumonia severity.⁴⁰

Although the reasons for these observations are uncertain, there are some data that suggest that cannabis could have beneficial or neutral effects in patients with COPD. Cannabidiol (CBD) is a nonpsychoactive constituent of cannabis. Some animal models suggest that THC has anti-inflammatory and immunomodulatory effects that may confer downstream clinical benefits in chronic inflammatory disorders such as COPD and asthma.^41,42 CBD also have therapeutic application in epilepsy, multiple sclerosis, and neuropathic pain due to its anti-inflammatory and immune suppressive properties.^43,44 Animal models have also demonstrated anti-inflammatory effects on colon epithelial cells by the nonpsychoactive component of cannabis, D9-tetrahydrocannabinolic acid.⁴⁵ Also, the toxic effects of cannabis smoke on lung function may be less than one may otherwise predict. In a population study by Hancox et al., cannabis use did not impair airflow or gas transfer, although it was associated with lung hyperinflation.⁴⁶ The SPIROMICS study found that cannabis use was not associated with increased symptoms of wheezing, cough, or bronchitis.⁴⁷ Other meta-analyses have found no association between cannabis use and impairment in spirometric indices.^9,48 In addition, THC may have a bronchodilator effect, and when studied in healthy subjects, it actually increased airway conductance.^49,50

There are other potential explanations for our findings. It is possible that cannabis use does not directly impact hospital outcomes in COPD, but rather coclusters with some other, not yet identified clinical or behavioral factor that accounts for our study's observations.

Regardless of the reasons, the association of cannabis use with lower in-hospital mortality, pneumonia, and hospital LOS in hospitalized patients with COPD is compelling and consistent with previously reported data in non-COPD patient populations. Additional studies are warranted to clarify the factors that underlie these observations.

Strengths

The major strength of this study is the large sample size, which is representative of almost 95% of U.S. hospitalizations. Our study is the first direct evidence regarding the association between cannabis use and health outcomes among COPD hospitalizations in the United States.

Study limitations

The use of administrative databases has certain limitations. Because this is a cross-sectional observational study, the possibility of sampling error, selection bias, and residual measured and unmeasured confounding cannot be completely eliminated.⁵¹ However, the potential limitation may be partially compensated for by the large size of the database and the ability to obtain nationwide estimates using the discharge weights provided by the Healthcare Cost and Utilization Project (HCUP). It is possible that we may have underestimated the prevalence of cannabis due to the fact that coding for these diagnoses had occurred only when there was documentation of the clinician advising the patient for the cessation of cannabis. During the initial study time period, there might be social desirability bias in disclosing cannabis use.^52,53

Another significant limitation is that this study is constrained by data elements provided by HCUP-NIS. It lacks data on readmissions and patient-level clinical information such as the body mass index, the cause of death, quantification, and route of cannabis or cigarette smoking. In addition, we cannot determine the results of pulmonary function tests and the severity of COPD.

Conclusion

Our study shows that cannabis use among patients hospitalized with COPD may be at lower risk for pneumonia and in-hospital mortality than nonusers. The adverse effects of cannabis on respiratory health have only scant evidence, which has many limitations; the full impact of cannabis use on the respiratory system is yet to be known. Additional studies are needed further to clarify the effects of cannabis use on COPD hospitalizations.

Abbreviations Used

AHRQ

Agency for Health Care Quality and Research

AMA

against medical advice

CBD

cannabidiol

CI

confidence interval

COPD

chronic obstructive pulmonary disease

HCUP

Healthcare Cost and Utilization Project

HHC

home health care

HIV

human immunodeficiency virus

ICD-9 CM

International Classification of Diseases-9 Clinical Modification

ICF

intermediate care facility

LOS

length of stay

MI

myocardial infarction

NIS

National/Nationwide Inpatient Sample

OR

odds ratio

SNF

skilled nursing facility

THC

tetrahydrocannabinol

Authors' Contributions

K.G., D.C.V., M.S.R., K.E., S.R., A.G., and J.K. each made substantial contributions to the conception or design of the work, the acquisition, analysis, or interpretation of data for the work, drafted and assisted in critical revisions to work for important intellectual content, provided final approval of the version to be published, and are in agreement to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.

Author Disclosure Statement

No competing financial interests exist.

Funding Information

No funding was received for this article.

Supplementary Material

Supplementary Table S1

Cite this article as: Gunasekaran K, Voruganti DC, Singh Rahi M, Elango K, Ramalingam S, Geeti A, Kwon J (2021) Trends in prevalence and outcomes of cannabis use among chronic obstructive pulmonary disease hospitalizations: a nationwide population-based study 2005–2014, Cannabis and Cannabinoid Research 6:4, 340–348, DOI: 10.1089/can.2020.0133.