Nationwide outcomes of cardiac surgery in patients with cannabis use disorder

Abstract

Objective

To characterize in-hospital morbidity and mortality in patients undergoing cardiovascular surgery with and without cannabis use disorder (CUD) using a national database.

Methods

Between 2016 and 2018, 846,837 patients who underwent cardiovascular surgery were identified from the Nationwide Readmissions Database. Of these, 11,724 (1.4%) with CUD were identified. A 1:1 balancing-score matching was performed to compare outcomes while controlling baseline characteristics and comorbidities.

Results

Concomitant substance abuse, including smoking/nicotine, opioid abuse, cocaine/stimulants abuse, and alcohol abuse, was also significantly higher (P < .001 for all). Patients with CUD had a higher prevalence of deficiency anemia, congestive heart failure, chronic lung disease, depression, drug abuse, liver disease, neurological disorder, peripheral vascular disease, psychoses, pulmonary circulatory disorders and weight loss. Prior to matching, stroke (3.4% vs. 2.8%, P = .008), pneumonia (13% vs. 9.4%, P < .001), sepsis (5.5% vs. 3.2%, P < .001), and pulmonary embolism (3.6% vs. 1.8%, P < .001) were more common among patients with CUD. However, after matching, there were no differences in overall or individual complications. Mortality among patients with CUD was lower in both unmatched (1.8% vs. 3.0%, P < .001) and matched comparisons (1.8% vs. 2.7%, P < .001).

Conclusions

Cannabis abuse was not associated with additional in-hospital morbidity or mortality and may not preclude cardiovascular surgery. However, CUD is a marker for polysubstance abuse, which deserves management.

Relevance to clinical practice

Patients with CUD are a vulnerable population with concomitant smoking, opioid abuse, cocaine abuse, and alcohol abuse that deserves treatment while undergoing cardiovascular surgery.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13019-025-03755-6.

Introduction

Despite increasing legalization in the United States and increasing reported usage, little is known regarding the true effects of cannabis use on the cardiovascular system and by extension the postoperative outcomes following cardiovascular surgery. Many temporal associations between marijuana and cardiovascular disease have been described, but long-term implications are unknown. Cardiovascular-related mortality increased by 2.3% in men and 1.3% in women in states where cannabis use was legal compared to where it was not [1]. Cannabis use has been linked to the development arrhythmias in individuals without any other notable predisposing factors [2]. In addition, marijuana use is also associated with acute cardiac ischemia ST-elevation myocardial infarction (STEMI), and non-STEMI [3]. One study noted an 8% increase in the odds of developing a myocardial infarction with recreational cannabis use [4].

Studies regarding the effect of cannabis use on surgical outcomes are scant. The few that exist do not report any meaningful difference in outcomes [5, 6]. However, patients with a history of marijuana use presenting for percutaneous coronary intervention were at higher risks of post-procedural bleeding and stroke [7]. Given the associations between cardiovascular and neurological complications after cannabis use, potential risks of major cardiovascular surgery in this population remain an important unanswered question. We sought to characterize in-hospital morbidity and mortality in patients undergoing cardiovascular surgery with and without cannabis use disorder (CUD).

Patients and methods

Data source and patients

The Nationwide Readmissions Database (NRD) (HCUP, Agency for Healthcare Research and Quality, Rockville, MD) is an administrative designed to support various types of analyses of national readmissions for all patients regardless of the expected payer for the hospital stay [8]. Discharge-level data are available in the form of International Classification of Diseases, tenth revision, clinical modifications (ICD-10-CM) diagnosis and procedure codes. Patient linkage numbers allow for HIPAA-compliant tracking of patients across hospitalizations within states.

The NRD was queried from January 2016 to December 2018 for all adult (≥ 18 years) patients undergoing isolated coronary bypass, valve, aortic surgery, or a combination of these using Clinical Classifications Software (CSS) procedure codes 44, 43, and 52, respectively. Patients with CUD were identified using ICD-10-CM diagnostic codes for cannabis abuse and dependence (F12.1, F12.2, F12.9).

Readmissions and follow-up

Readmissions were examined 90 days after the index operation. To ensure a 90-day follow-up, patients undergoing cardiac surgery were identified from January to September for each year. Patients who died during index admission or for whom length-of-stay data were missing were excluded. Readmission was defined as an admission following the index admission when surgery was performed.

Variables and outcomes

Demographics included age, sex, race, insurance status, income level, and admission status. Comorbidities were derived from the Elixhauser Comorbidity Index, a validated tool for measuring risk among patients in administrative databases [9]. Primary outcomes included mortality and complications on index admission while discharge disposition and 90-day readmissions were secondary outcomes. Complications included stroke, bleeding, blood transfusion, renal failure, pneumonia, surgical site infection, sepsis, post-procedural or postoperative respiratory complications, pulmonary embolism, deep venous thrombosis, cardiac complications, and gastrointestinal complications. ICD codes used for these can be found in Supplementary Table 1.

Statistical analysis

All analyses were conducted using the SAS System for Unix, version 9.4 (SAS Institute, Inc, Cary, NC). Frequency counts and percentages were calculated for categorical variables. For all comparisons between patients with and without CUD, chi-square and Fisher’s exact test were used. Discharge weights were used to create national estimates.

Balancing-score matching was conducted to adjust for covariables between patients with and without CUD and to mitigate the effects of confounding factors on outcomes. We used a 1:1 greedy matching algorithm with a caliper of 0.2 times the standard deviation of the linear balancing score. Because patient samples were separate for the 90-day analysis to ensure follow-up, this was matched individually. Covariables used for balancing matching included baseline characteristics such as demographics and hospital characteristics including elective surgery as well as comorbidities and concomitant substances listed in Fig. 1. Kaplan-Meier curves for readmissions were created and compared using the log-rank test.

Fig. 1.

Standardized Difference Plot before and after propensity-score matching

Results

Patient characteristics

Of 846,837 hospitalizations for cardiac surgery between 2016 and 2018, 11,724 (1.4%) were identified with documented cannabis abuse or dependence. Baseline demographics and characteristics can be found in Table 1. Patients with CUD undergoing cardiovascular surgery had more aortic and combination surgeries, urgent procedures in metropolitan teaching hospitals and were younger males in the highest income quartile less likely to have private insurance. Concomitant substance abuse, including smoking/nicotine, opioid abuse, cocaine/stimulants abuse, and alcohol abuse, was all significantly higher as well. Additionally, these patients had a higher prevalence of deficiency anemia, congestive heart failure, chronic lung disease, depression, drug abuse, and liver disease, neurological disorder, peripheral vascular disease, psychoses, pulmonary circulatory disorders and weight loss.

Table 1.

Preoperative patient characteristics (Unmatched)

Variable	No-CUD	CUD	P-value
	(N = 835,113)	(N = 11,724)
	N (%)	N (%)
Type of procedure
CABG	534,785 (64)	7,489 (64)	0.83
Valve	170,086 (20)	2,587 (22)	0.006
Aortic	19,737 (2)	476 (4)	< 0.001
Combination	110,505 (13)	1,173 (10)	< 0.001
Age (years)			< 0.001
18–44	38,245 (5)	2,350 (21)
45–54	90,148 (11)	2,796 (24)
55–64	223,692 (27)	4,304 (37)
65–74	300,869 (36)	2,058 (18)
75+	182,158 (22)	215 (2)
Sex			< 0.001
Male	591,914 (71)	9,541 (81)
Female	243,199 (29)	2,183 (19)
Admission status			< 0.001
Urgent	384,854 (46)	8,062 (69)
Elective	447,767 (54)	3,643 (31)
Missing	2,492 (0.3)	19 (0.2)
Median household income			< 0.001
1 st quartile	218,158 (26)	4,183 (36)
2nd quartile	234,084 (28)	3,300 (28)
3rd quartile	212,087 (25)	2,510 (21)
4th quartile	158,286 (19)	1,522 (13)
Missing	12,498 (2)	209 (2)
Insurance			< 0.001
Medicaid	61,648 (7)	2,157 (18)
Medicare	476,134 (57)	2,194 (19)
Other	27,053 (3)	443 (4)
Private	251,923 (30)	2,126 (18)
Uninsurance	18,355 (2)	653 (6)
Hospital teaching status			0.004
Metropolitan non-teaching	135,649 (16)	1,640 (14)
Metropolitan teaching	679,826 (81)	9,910 (85)
Non-metropolitan hospital	19,639 (2)	174 (2)
Comorbidities
Alcohol abuse	27,805 (3)	2,074 (18)	< 0.001
Smoking/Nicotine	139,944 (17)	5,876 (50)	< 0.001
Opioid abuse	8,726 (1)	877 (8)	< 0.001
Cocaine/Stimulants abuse	6,532 (1)	2,014 (17)	< 0.001
Deficiency anemia	133,870 (16)	2,375 (20)	< 0.001
Congestive heart failure	19,819 (2)	380 (3)	0.007
Chronic lung disease	185,030 (22)	3,696 (32)	< 0.001
Coagulopathy	77,271 (9)	1,128 (10)	0.45
Depression	79,402 (10)	1,766 (15)	< 0.001
Diabetes mellitus - uncomplicated	128,310 (15)	1,174 (10)	< 0.001
Diabetes mellitus - complicated	212,888 (26)	2,490 (21)	< 0.001
Drug abuse	14,287 (2)	5,547 (47)	< 0.001
Hypertension	523,674 (63)	6,759 (58)	< 0.001
Liver disease	23,462 (3)	707 (6)	< 0.001
Neurological disorders	52,398 (6)	974 (8)	< 0.001
Obesity	229,354 (28)	2,727 (23)	< 0.001
Peripheral vascular disease	143,999 (17)	2,320 (20)	< 0.001
Psychoses	13,744 (2)	714 (6)	< 0.001
Pulmonary circulation disorders	2,460 (0.3)	181 (2)	< 0.001
Renal failure	154,393 (19)	1,648 (14)	< 0.001
Weight loss	35,706 (4)	718 (6)	< 0.001

Unmatched outcomes on index admission

Unmatched outcomes can be found in Supplemental Table 2. Mortality was lower among CUD patients, whereas stroke, pneumonia, sepsis, and pulmonary embolism were higher. Cardiac complications were lower among those with CUD (42% vs. 49%; P <.001). Most of these were due to arrhythmia (41% vs. 48%; P <.001). Patients with CUD were more often discharged home (45% vs. 37%; P <.001) and less often to a skilled nursing facility (12% vs. 19%; P <.001).

Matched outcomes on index admission

After matching, 6,220 hospitalizations were identified both with and without documented CUD and were well-matched (Fig. 1). There were no differences in overall complications (65% vs. 65%, P =.66) (Table 2) or in stroke, pneumonia, sepsis or pulmonary embolism. Mortality remained lower (1.8% vs. 2.7%; P <.001). At 90 days postoperatively, there was no difference in readmissions or reoperations among matched pairs (21% vs. 21%, P =.73) (Table 3; Fig. 2).

Table 2.

Propensity-score matched outcomes after cardiac surgery in patients with or without CUD

Variable (index admission)	Non-CUD (N=6,220)	CUD (N=6,220)	P-value
	No. (%)	No. (%)
Hospital Mortality	170 (2.7)	109 (1.8)	< 0.001
Any complication	4,066 (65)	4,043 (65)	0.66
Stroke	240 (4)	207 (3)	0.1
Bleeding	341 (6)	369 (6)	0.28
Blood transfusion	185 (3)	209 (3)	0.22
Pneumonia	832 (13)	789 (13)	0.25
Sepsis	390 (6)	325 (5)	0.01
Respiratory complications	707 (11)	692 (11)	0.67
Pulmonary embolism	230 (4)	241 (4)	0.36
DVT
Gastrointestinal complications	74 (1)	76 (1)	0.87
Renal failure	1,472 (24)	1,442 (23)	0.53
Cardiac complications	2,707 (44)	2,683 (43)	0.66
Myocardial infarction	48 (1)	34 (0.5)	0.12
Cardiac tamponade	80 (1)	64 (1)	0.18
Cardiac insufficiency	*	*	0.32
Cardiac arrest	14 (0.2)	14 (0.2)	1
Heart failure	21 (0.3)	20 (0.3)	0.88
Shock	*	14 (0.2)	0.3
Cardiogenic shock	135 (2)	102 (2)	0.03
Arrhythmia	2,573 (41)	2,596 (42)	0.68
Right heart failure	*	*	0.04
Discharge status			0.006
Home health care	2,392 (38)	2,489 (40)
Home	2,746 (44)	2,776 (45)
Skilled nursing facility	797 (13)	741 (11)
Transfer to short-term skilled facility	58 (1)	37 (1)
Other	223 (4)	175 (3)
Missing	*	*

*Cells with frequency <11 are suppressed because of requirements in the Healthcare Costand Utilization Project (HCUP) data-use agreement

Table 3.

Frequency of readmission within 90 days of index operation among CUD patients after cardiac surgery

Variable	No-CUD (N = 4,540)	CUD (N = 4,512)	P-value
	No. (%)	No. (%)
Total readmitted in 90 days	953 (21)	934 (21)	0.73
Readmitted day 0–30	476 (11)	476 (11)	0.97
Readmitted day 31–90	477 (11)	458 (10)	0.58

Fig. 2.

Readmissions among patients with and without cannabis use disorder after cardiac surgery

Discussion

Multiple recent studies have raised the concern that cannabis use contributes significantly to the development of cardiovascular disease, including arrhythmias acute coronary syndromes and cardiac arrest in those with otherwise no notable risk factors [2–4, 7]. Despite that some complications were more frequent before adjustment in patients with CUD — probably due to an associated higher comorbidity burden — these did not differ when accounting for baseline characteristics and some comorbidities. In the context of polysubstance abuse, CUD is not just a naive covariable. The slight improvement in mortality and home discharge can also be explained due to the lack of access to more granular data. Moreover, the significance of these findings does not represent causality, and these associations must be validated with other databases.

Before matching, the unadjusted frequency of some complications among patients with CUD exceeded that of the general population. We found a significantly higher prevalence of stroke, pulmonary embolism, sepsis, and pneumonia. At face value, this seems to be supported by literature. Higher occurrence of stroke has been also shown among patients with CUD after percutaneous coronary intervention in a large series of 3,970 patients with CUD in Michigan [7]. Similarly, a prior analysis using the National Inpatient Sample database found an increased incidence of acute myocardial infarction, arrhythmia and stroke among young patients cannabis users compared to non-users [10].

In this analysis, the prevalence of chronic lung disease was higher among patients with CUD. Cannabis smoke contains many of the same carcinogens, mutagens, and compounds seen in tobacco smoke, though the physiological impact of these compounds has yet to be characterized [11–13]. Cannabis is also associated with dose-dependent immunosuppressive effects, including reduction in T cell proliferation and IFN-γ production, which has translated into bronchitis and poor lung function in chronic users [14, 15]. Taken together, these would seem to support the role of CUD in the development of pneumonia after cardiovascular surgery seen in the unadjusted analysis.

However, after accounting for preoperative characteristics and comorbidities, there were no individual differences in any of the above complications. Nearly 50% of CUD patients also presented with tobacco use, 18% with alcohol abuse, 17% with known cocaine/stimulants abuse, and 7.5% with opioid use disorder. Although CUD is closely associated with other concomitant substance abuse, these data suggest that it does not independently contribute to poor in-hospital outcomes after cardiovascular surgery.

There is scarce evidence of cannabis use on outcomes after cardiovascular procedures. Lack of data in this population, despite increasing prevalence, has led to ambiguity and a lack of standardization in both patient selection and management. What, in turn, does this mean for clinical practice? Preclusion of cardiovascular surgery due to cannabis abuse may not be warranted given the lack of any ​complication. However, CUD is a marker for polysubstance abuse, which deserves prompt identification and management in the perioperative setting to optimize outcomes.

Our institution in collaboration with the university has shown a parallel rise in opioid abuse among patients undergoing cardiovascular surgery, associated with higher complications and readmissions [16, 17]. Similarly, an association between cocaine abuse and aortic dissections has been reported [18]. Furthermore, other than smoking cessation, there are no standardized screening protocols to guide preoperative or postoperative management of patients with marijuana or polysubstance abuse undergoing cardiac surgery or other elective procedures [19, 20]. Prospective studies are needed to study both the preoperative risk and downstream effects of cannabis after surgery in order to counterbalance misinformation and inform clinical practice.

This study is subject to the limitations of a retrospective study using administrative data, including the potential for miscoding, overcoding, and the lack of clinical granularity afforded by single-center databases. These data also likely underrepresent the true number of patients who present with CUD due to a combination of infrequent screening and inconsistent patient disclosure. Despite these limitations, the NIS affords a well-powered analysis of a real-world cohort of patients undergoing cardiovascular surgery that has not been studied to date in this surgical discipline. It will be important for future clinical studies to validate the results of this work, to further evaluate subsets of this population that may benefit from targeted screening and intervention, and to define what the ideal screening and intervention protocol.

Conclusions

Cannabis use may not preclude cardiovascular surgery. This analysis suggests that CUD was not associated with additional in-hospital morbidity and mortality in patients undergoing cardiovascular surgery. However, CUD is a marker for polysubstance abuse, which deserves prompt identification and management in the perioperative setting.

Author contributions

KCD: conceptualization, investigation, writing - original draft, writing - review& editing. RM: Investigation, project administration, writing - review& editing. SX: investigation, writing - review& editing. GZ: data curation, formal analysis, visualization, writing - review& editing. JAM: investigation, writing - review& editing. CA: investigation, writing - review& editing. JDF: project administration, writing - review& editing. SMK: methodology, writing - review& editing. AJW: supervision, writing - review& editing. EGS: supervision, writing - review& editing.

Funding

This work was supported internally by the Cleveland Clinic Department of Thoracic and Cardiovascular Surgery.

Data availability

Data is publicly available and may be shared upon reasonable request.

Declarations

Ethics approval and consent to participate

As the NRD is a public, de-identified database, the Institutional Review Board at Case Western Reserve University deemed this study exempt from review. Thus, this study was conducted in accordance with the ethical standards of the Office of Human Research Protection and with the Helsinki declaration and its later amendments or comparable ethical standards.

Disclosure

None.

Competing interests

The authors declare no competing interests.