Cannabis use history is associated with increased prevalence of ascites among patients with nonalcoholic fatty liver disease: A nationwide analysis

Table 1 Patient demographics, hospital characteristics, and outcomes among patients with nonalcoholic fatty liver disease, by history of cannabis use

	Cannabis users, n = 3820	Non-cannabis users, n = 7625	P value
Sex			NS1
Female	36.0%	36.1%
Male	64.0%	63.9%
Patient age, mean (SD)	41.4 (12.9)	42.0 (12.9)	0.032
Patient age, in 10 years age groups3			NS4
18-27	16.8%	16.6%
28-37	25.3%	25.3%
38-47	20.4%	20.5%
48-57	25.8%	25.8%
58-67	11.0%	11.0%
68-77	0.8%	0.8%
Race/ethnicity			NS4
White	60.6%	60.7%
Black	21.6%	21.6%
Hispanic	13.9%	13.9%
Asian or pacific islander	0.4%	0.4%
Native American	0.9%	0.9%
Others	2.6%	2.4%
ECI, mean (SD)	2.1 (9.4)	3.6 (8.1)	< 0.012
ECI, by category			NS4
≤ 0	47.6%	47.7%
1-5	15.8%	15.9%
6-10	19.6%	19.5%
11 or higher	16.9%	16.9%
Insurance			< 0.054
Medicare	16.6%	17.2%
Medicaid	42.8%	30.5%
Private	21.9%	42.6%
Self-Pay	13.4%	6.0%
Others	5.4%5	3.7%
Cannabis abuse
Non-dependent use	94.1%	0 (by definition)
Dependent use	5.9%	0 (by definition)
Length of stay (days)	5.1	4.9	0.182
Total hospitalization charges	$42503	$43183	NS2
Comorbidities
Diabetes	29.2%	34.8%	< 0.055
Obesity	29.5%	49.4%	< 0.055
Hyperlipidemia	24.2%	32.4%	< 0.055
Clinical outcomes
Portal hypertension	80 (2.1%)	165 (2.2%)	NS5
Ascites	170 (4.5%)	275 (3.6%)	0.035
Varices and variceal bleeding	50 (1.3%)	130 (1.7%)	0.115
Cirrhosis	140 (3.7%)	275 (3.6%)	NS5

¹Chi-square, 2-tailed.

²Student t-test, 2-tailed.

³Patient ages ranged from 18 to 73, there was no one over 78.

⁴Chi-square, 2-tailed, for 2 by n table: Statistical significance demonstrates that the two groups differ.

⁵Condition absent vs present in the 2 groups, chi-square, 2-tailed. NS: Not statistically significant; ECI: Elixhauser comorbidity index.

Table 2 Independent predictors of ascites among patients with nonalcoholic fatty liver disease, stratified by cannabis use history

Cannabis users (n = 3820)					Non-cannabis users (n = 7625)
Ascites	OR (95%CI)	P value	aOR (95%CI)	P value	Ascites	OR (95%CI)	P value	aOR (95%CI)	P value
	Univariate logistic regression		Multivariate logistic regression1			Univariate logistic regression		Multivariate logistic regression1
Diabetes	1.17 (0.84-1.62)	NS			Diabetes	1.16 (0.91-1.49)	0.23
Obesity	1.00 (0.71-1.40)	NS			Obesity	0.41 (0.31-0.53)	< 0.01	0.61 (0.46-0.81)	< 0.01
Hyperlipidemia	1.32 (0.94-1.85)	0.11	1.53 (1.08-2.17)	0.02	Hyperlipidemia	0.57 (0.43-0.77)	< 0.01	0.59 (0.44-0.80)	< 0.01
Age (continuous)	0.99 (0.98-1.01)	0.22			Age (continuous)	1.02 (0.01-1.03)	< 0.01	1.02 (1.01-1.03)	< 0.01
Race/ethnicity (categoric)					Race/ethnicity (categoric)
White	Reference				White	Reference
Black	1.13 (0.78-1.64)	NS			Black	0.80 (0.58-1.10)	0.16
Hispanic	0.87 (0.53-1.42)	NS			Hispanic	1.00 (0.70-1.42)	NS
Asian or pacific islander	1				Asian or pacific islander	1
Native American	2				Native American	1.96 (0.78-4.92)	0.15	2.60 (1.03-6.61)	0.04
Others	2.46 (1.24-4.87)	0.01	2.57 (1.24-5.31)	0.01	Others	0.71 (0.29-1.74)	NS
Comorbidity (continuous)	1.08 (1.07-1.10)	< 0.01	1.08 (1.07-1.10)	< 0.01	Comorbidity (continuous)	1.06 (1.05-1.07)	< 0.01	1.05 (1.04-1.06)	< 0.01
Sex (female)	0.80 (0.58-1.09)	0.15			Sex (female)	0.99 (0.77-1.27)	NS

¹Univariate analysis with screening P < 0.02 was used to determine variables to include in the final multivariate analysis. Hyperlipidemia, Race/ethnicity, comorbidity index, and sex included in the regression.

²Univariate analysis with screening P < 0.02 was used again, and obesity, hyperlipidemia, age, race/ethnicity, and comorbidity index were included in the multivariate regression. NS: Not statistically significant.