Baseline demographic characteristics of patients are summarized in Table 1. A total
of 463 individual patients with cirrhosis, out of 529 reviewed, were identified including
whites (n=241), Hispanics (n=106), Blacks (n=50), and Asian and Pacific Islander Americans
(API, n=27). The remaining 39 patients were categorized as Other. Mean age was 57.2
± 10.8 years (range 24 to 96) and the majority of the patients were male (58.1%).
Compared to other racial groups, a larger proportion of API did not speak English
as their primary language (51.8%); p= <0.0001. A significantly higher proportion of
Blacks, Hispanics, and API had Medicaid compared to Whites (p=0.0002). Blacks and
Hispanics also had lower median income than Whites ($45,710 versus $54,844, p=0.01).
Further, Blacks and Hispanics had higher mean MELD-Sodium (MELD-Na) scores compared
to Whites (28.7±9.3 and 29.1±8.5 versus 26.4±9.5, p=0.02) (Table 2).
Regarding medical co-morbidities, API (70.4%) and Blacks (44.0%) had significantly
higher prevalence of diabetes mellitus (DM) compared to Whites (33.3%, p=0.004). Blacks
also had the highest prevalence of coronary artery disease (54.0%, p=0.0004) and chronic
hepatitis C (38.0%, p <0.0001) compared to other racial groups. API had the highest
proportion of cirrhosis from chronic hepatitis B compared to other races (25.9%, p
<0.0001) (Table 3).
Adherence to most cirrhosis quality care indicators ranged from 87% to 100% (Figure
1) suggesting high quality inpatient cirrhosis care. The exception was that only
53.2% received the proper dosing of albumin (1.5g/kg IV albumin on day 1 and 1g/kg
IV albumin on day 3 of SBP diagnosis). No racial disparities regarding receipt of
quality cirrhosis care were identified.
Regarding healthcare outcomes, Black patients had the highest mean total hospital
admissions (6.1±6.3, p=0.01), highest mean number of 30-day re-admissions (2.1±3.7,
p=0.05). Blacks also had the highest mean number of 90-day readmissions (3.7±5.5,
P=0.10) and highest mean total hospital days (46.3±54.5, p=0.07) though these did
not reach statistical significance (Table 3).
Socioeconomic factors and medical co-morbidities were found to be an independent predictor
of 30-day hospital readmission (Table 4). Multivariable proportional odds regression
showed that race was not a statistically significant predictor of 30-day readmission
(p=0.21). However when Blacks were compared to whites, OR=2.21 with 95% CI (1.14-3.95)
favoring 30-day readmission in Blacks. Coronary artery disease (CAD), chronic obstructive
pulmonary disease (COPD), hepatic encephalopathy (HE), and presence of ascites were
all independently associated with 30-day readmission on multivariable proportional
odds regression analysis. Insurance status was also found to be predictive of 30-day
readmission. The odds ratio of 30-day readmission for lack of insurance versus private
insurance was OR=0.45 with 95% CI (0.21-0.99). The odds ratio of 30-day readmission
for Medi-cal insurance versus private insurance and Medi-cal insurance versus Medicare
insurance were OR=1.15 with 95% CI (0.53, 2.47) and OR=1.35 with 95% CI (0.82-2.24),
Multivariable proportional odds regression analysis for predictors of 90-day readmission
was also performed (Table 5). Race was a statistically significant predictor of 90-day
readmission (p=0.03). The odds ratios of 90-day readmission for Black versus white
and Black versus API were OR=4.69 with 95% CI (1.57, 13.95) and OR=8.45 with 95% CI
(1.90, 37.60), respectively. Insurance status was also a statistically significant
predictor of 90-day readmission (p = 0.0001). The odds ratios of 90-day readmission
for lack of insurance versus Medi-cal insurance and Medi-cal insurance versus Medicare
insurance were OR=0.15 with 95% CI (0.06-0.39) and OR=4.72 with 95% CI (2.11, 10.6),
respectively. The odds ratio of 90-day readmission for Medi-cal insurance versus
private was OR=2.57 with 95% CI (0.78 8.52).
MELD-Na and median income were not found to be predictive of readmission on univariable
analysis so multivariable analysis was not performed on these variables regarding
their relationship to hospital readmission. However, multivariable binary logistic
regression analysis showed a statistically significant association between 30-day
mortality and MELD-Na OR=1.06, 95% CI (1.03, 1.09), p< 0.0001.