Aspirin reduces short-term mortality risk in critically ill patients with liver cirrhosis: a propensity-score matched retrospective analysis using the MIMIC-IV database

Abstract

Background

Liver cirrhosis not only leads to high mortality but also imposes significant economic burdens and health losses. Aspirin is a drug with potential indications for liver disease, but its benefits in cirrhosis have primarily been demonstrated in outpatient settings. This study aimed to determine whether aspirin therapy confers protective effects on the prognosis of critically ill cirrhosis patients.

Methods

Cirrhosis patients were identified from the Medical Information Mart for Critical Care Medicine IV (MIMIC-IV) database. Propensity score matching (PSM) was used to balance baseline differences. Multivariate Cox regression models assessed the association between aspirin therapy and 30-day and 90-day mortality, while multivariable logistic regression models evaluated its relationship with in-hospital mortality. Due to a lack of relevant data, this study could not assess aspirin-related bleeding events.

Results

The study included 3,105 patients, of whom 348 received aspirin and 2,757 did not. After propensity score matching, 334 matched pairs were successfully identified. The 30-day mortality rate among aspirin users was 15.27%, and the 90-day mortality rate was 16.77%, both lower than those among non-users. Multivariate Cox regression analysis demonstrated that aspirin use was associated with reduced 30-day mortality (HR 0.69, 95% CI 0.48–0.99) and 90-day mortality (HR 0.65, 95% CI 0.46–0.92). Multivariable logistic regression analysis indicated that aspirin use was associated with reduced in-hospital mortality (OR = 0.66, 95% CI 0.44–0.99). There was no significant difference in intensive care unit length of stay between the two groups. However, due to most participants (81.0%) receiving the 81 mg/d dose, the significant disparity in sample sizes between the high- and low-dose groups and the baseline imbalance precluded reliable dose-response comparisons in this study.

Conclusion

This single-Centre retrospective study found that aspirin use was associated with reduced mortality in critically ill patients with liver cirrhosis. However, the overall net therapeutic effect remains to be further validated due to the lack of data on bleeding risk. The relationship between aspirin dosage and specific outcomes requires clarification in future studies with larger sample sizes and more prospective designs.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12876-025-04499-2.

Introduction

Liver disease is one of the major public health challenges globally, causing well over 2 million deaths annually and accounting for 4% of all deaths globally [1]. Although liver disease is currently ranked as the 11th leading cause of death globally, its actual impact is likely to be underestimated [2]. Cirrhosis is the 10th leading cause of death in Africa (up 3 places from 2015), the 9th leading cause of death in Southeast Asia and Europe, and as high as the 5th leading cause of death in the Eastern Mediterranean region [3]. In addition, cirrhosis significantly contributes to the global loss of healthy life expectancy (measured in DALY), ranking 15th worldwide, and disproportionately affects individuals aged 25–49 years, where it is the 12th leading cause of DALY; potential life expectancy loss may be even higher in Europe [4, 5]. On an economic level, cirrhosis imposes a heavy healthcare burden. In the United States, for example, liver-related healthcare expenditures amounted to $32.5 billion in 2016, two-thirds of which was spent on inpatient and emergency care, and related expenditures have increased by 4% per year over the past 20 years [6]. Globally, there were 10.6 million cases of decompensated cirrhosis and 112 million cases of compensated cirrhosis in 2017 [7]. Patients with compensated cirrhosis and decompensated cirrhosis have a 5-fold and 10-fold increased risk of death, respectively, compared to the general population [8]. The 1- and 5-year survival rates for patients with compensated cirrhosis have been reported to be 87% and 67%, respectively, whereas the rates for patients with decompensated cirrhosis drop significantly to 75% and 45% [8]. Cirrhosis not only leads to high mortality rates but also carries a significant economic burden and health loss.

Aspirin (acetylsalicylic acid) is a widely used medicine that helps reduce inflammation, fight tumors, and affect certain fats in the body by blocking the action of the pro-inflammatory cyclooxygenase-2 (COX-2) and platelet-derived growth factor (PDGF) pathways [9–11]. Studies have indicated that aspirin can be used for many purposes, such as pain relief, reducing fever, managing heart and brain diseases, treating joint diseases, helping during pregnancy, and preventing cancer [12–14].

Cirrhosis is a disease characterized by chronic inflammation accompanied by multiple complications and a high mortality rate. Clinical treatment is still based on symptomatic therapy. In studies with rats that have cirrhosis, various medications such as the COX-2 inhibitor celecoxib, aspirin, curcumin, carvacrol, hexoketone cacodylate, diosmin, statins, emricasan, and silymarin have shown potential in lowering inflammation and combating oxidative stress [15]. Aspirin has demonstrated significant efficacy in inhibiting scarring and cellular proliferation in preliminary studies, presumably by decelerating liver fibrosis and reducing the risk of hepatocellular carcinoma (HCC) [16]. Observational studies back this assertion up: in patients with metabolic dysfunction-associated steatohepatopathy (MASLD), using aspirin was linked to slower progression of severe liver scarring, fewer cases of liver cancer (HCC), and lower death rates related to liver issues [17–20]. However, patients with advanced cirrhosis may experience an attenuated protective effect [21], and the results of available meta-analyses are controversial [22]. The use of aspirin in patients with cirrhosis needs to be weighed against safety. Although patients with portal hypertension are often comorbid with thrombocytopenia, which may increase the risk of bleeding [23], retrospective studies have indicated that the use of aspirin for cardiovascular indications in patients with cirrhosis does not significantly increase the number of major bleeding events [24–26]. A recent study involving 587 patients showed that taking aspirin after a trans jugular intrahepatic portosystemic shunt (TIPS) greatly helped patients with severe fluid buildup live longer without needing another procedure after 12 months, but it didn’t significantly help those with bleeding from varices, indicating that aspirin might work better for certain groups of patients.

Recent studies have explored the role of aspirin in critically ill patients with sepsis-associated liver injury (SALI) [27], but its benefits remain unknown in a broader population of critically ill patients with underlying liver cirrhosis. This study aims to address this gap.

Materials and methods

Data sources

This study examines data from the MIMIC-IV database version 3.1, which contains anonymous clinical information about patients in the Emergency Department and Intensive Care Unit (ICU) at Beth Israel Deaconess Medical Center (BIDMC) in the United States from 2008 to 2022, including details like patient demographics, vital signs, lab tests, medication records, and diagnostic codes. The construction of the database was reviewed by the BIDMC Institutional Review Board (IRB), which waived informed consent and approved data sharing, so no additional ethical approval was required for this study. The first author (Yu Yi) has completed the CITI program training (certification ID: 68122805) and passed the data use compliance test and was granted access to the MIMIC-IV data.

Study participants

This cohort study evaluated critically ill cirrhosis patients admitted to the ICU, identifying 7,830 patients through International Classification of Diseases (ICD) coding (Supplementary Table 1). This analysis was done by excluding 4,725 patients who were younger than 18 years of age, admitted to the hospital for a non-first admission or non-first ICU admission, and had an ICU stay shorter than 24 h. The final cohort consisted of 3,105 patients, 348 aspirin users, and 2,757 nonusers. After propensity score matching (PSM), 334 pairs of patients were matched (Fig. 1).

Fig. 1.

Patient selection flowchart from the MIMIC-IV database for the study on aspirin use in patients with cirrhosis. (A total of 7,830 patients with cirrhosis were initially identified. After applying the exclusion criteria (age < 18 years, ICU stay < 24 h, and non-first admission or non-first ICU admission), 3,105 patients were included in the final analysis. These patients were categorized into aspirin users (n = 348) and non-users (n = 2,757). Propensity score matching (PSM) was then performed, resulting in 334 matched pairs of patients for further comparative analysis)

Aspirin exposure

Aspirin exposure was defined as a prescription for oral aspirin within the first 72 h of ICU admission.

Data extraction

We extracted the following clinical data from the database: (1) Demographic data such as age, gender, and ethnicity were collected. (2) Vital signs measured on admission included heart rate, respiratory rate, and blood oxygen saturation (SpO2). (3) Disease severity was assessed using the model for end-stage liver disease (MELD), Sequential Organ Failure Assessment (SOFA), and Charlson comorbidity index. (4) The first laboratory tests done at admission were for white blood cells (WBC), hemoglobin, platelet count, alanine aminotransferase (ALT), aspartate aminotransferase (AST), total bilirubin, albumin (ALB), creatinine, lactate, international normalized ratio (INR), prothrombin time (PT), and partial thromboplastin time (PTT). (5) Complications or comorbidities: ascites, hepatic encephalopathy, gastrointestinal bleeding, hepatorenal syndrome, liver transplantation, hypertension, type 2 diabetes mellitus (T2DM), cerebral infarction, myocardial infarction, chronic obstructive pulmonary disease (COPD), sepsis. (6) Therapeutic interventions included antibiotics, statins, non-selective beta-blockers, diuretics, continuous renal replacement therapy (CRRT), and mechanical ventilation. Measured outcomes included 30- and 90-day survival, in-hospital mortality, and ICU length of stay.

Primary and secondary outcomes

The primary outcomes of this study were 30-day and 90-day mortality. Secondary outcomes were in-hospital mortality and length of stay in the ICU.

Statistical analysis

This study utilized R version 4.2.2 (R Statistical Computing Foundation, Vienna, Austria) as the analytical software. In our research, all variables had less than 20% missing data, which was imputed using the “mice” package in R. For normally distributed data, continuous variables were expressed as mean ± standard deviation; for non-normally distributed data, they were presented as median (IQR). Categorical variables are presented as counts and percentages. Continuous variables were compared between the aspirin and non-aspirin groups using t-tests or Wilcoxon signed-rank tests, respectively, while categorical variables were compared using Pearson chi-square (χ²) tests.

We applied a chi-square width of 0.05 logits standard deviations to the PSM to mitigate baseline imbalance. Using nearest-neighbor matching techniques, groups were matched 1:1. PSM efficacy was assessed via standardized mean difference (SMD), with SMD ≤ 0.1 indicating balanced initial feature models.

Kaplan-Meier (KM) survival analysis and log-rank tests were used to compare 30-day and 90-day all-cause mortality between aspirin and non-aspirin users. To assess aspirin’s effect on mortality, four models were constructed: Model 1 was unadjusted. Model 2 was adjusted for racing. Model 3 was further adjusted for ascites, hepatorenal syndrome, liver transplantation, hypertension, myocardial infarction, sepsis, antibiotics, statins, non-selective beta-blockers, diuretics, CRRT, and mechanical ventilation. Model 4 was further adjusted for heart rate, respiratory rate, SpO₂, WBC, hemoglobin, total bilirubin, creatinine, lactate, INR, PT, PPT, MELD score, SOFA score, and Charlson comorbidity index. Results are presented as hazard ratios (HR) or odds ratios (OR) with 95% confidence intervals (CI). Subgroup analyses were also done based on race, age, gender, and other health problems (such as cerebral infarction, hypertension, T2DM, myocardial infarction, COPD, and sepsis). These subgroup analyses aimed to validate the consistency and robustness of our findings. We also examined interactions between subgroups using variance ratio tests, with statistical significance set at P < 0.05.

Results

Patient characteristics

This retrospective cohort study may be subject to confounding factors. To reduce bias, propensity score matching (PSM) was employed to balance confounding variables across groups. The study included 3,105 patients with liver cirrhosis, of whom 348 (11.21%) received aspirin therapy and 2,757 (88.79%) did not. As shown in Table 1, patients in the aspirin group prior to PSM exhibited the following characteristics: (1) older age and a higher proportion of males. (2) More severe disease, manifested by a higher Charlson Comorbidity Index (median [IQR]: 7 [5–9] vs. 5 [4–7]), but lower MELD scores (16.79 [11.88–22.68] vs. 19.96 [14.03–27.83]), and SOFA scores (7 [4–10] vs. 8 [5–11]). (3) Laboratory parameters: Higher hemoglobin, platelet count, and alanine aminotransferase (ALT); lower aspartate aminotransferase (AST), total bilirubin, lactate, INR, and prothrombin time (PT) (P < 0.05). (4) Complication spectrum: Higher incidences of T2DM (42.53% vs. 27.09%), myocardial infarction (13.22% vs. 3.19%), and chronic obstructive pulmonary disease (16.38% vs. 12.33%) were higher, as was the liver transplantation rate (4.60% vs. 0.91%) (all p < 0.01). The incidence of liver-related complications such as ascites and hepatic encephalopathy was lower. Using a 1:1 propensity scores matching method with a p-value threshold of 0.05, 334 matched pairs were successfully identified. After propensity score matching, the standardized mean differences for all baseline characteristics were less than 0.1 (Table 1; Fig. 2).

Table 1.

Baseline characteristics of critically ill cirrhosis patients before and after PSM

Variables	Before PSM					After PSM
	Total	Non aspirin	Aspirin	P-value	SMD	Total	Non aspirin	Aspirin	P-value	SMD
	(n = 3,105)	(n = 2,757)	(n = 348)			(n = 668)	(n = 334)	(n = 334)
Age	60 (53–68)	59 (52–67)	66 (59–74)	< 0.001	0.541	66 (58–74)	66 (58–74)	65 (58–73.75.75)	0.533	0.048
Gender, n (%)				< 0.001	0.217				0.243	0.098
Female	1088 (35.04)	997 (36.16)	91 (26.15)			164 (24.55)	75 (22.46)	89 (26.65)
Male	2017 (64.96)	1760 (63.84)	257 (73.85)			504 (75.45)	259 (77.54)	245 (73.35)
Ethnicity, n (%)				0.360	0.081				0.961	0.022
White	2024 (65.19)	1793 (65.03)	231 (66.38)			442 (66.17)	222 (66.47)	220 (65.87)
Black	225 (7.25)	195 (7.07)	30 (8.62)			61 (9.13)	31 (9.28)	30 (8.98)
Other/Unknown	856 (27.57)	769 (27.89)	87 (25.00)			165 (24.70)	81 (24.25)	84 (25.15)
Vital signs
Heart rate (beats/min)	90 (78–105)	91 (78–105)	85 (74–98.25.25)	< 0.001	0.236	86 (74–101)	88 (75–102)	85 (74–99.75.75)	0.340	0.074
Respiratory rate (beats/min)	19 (15–23)	19 (15–23)	18 (16–22)	0.077	0.108	18 (16–22)	18.5 (15–22.75.75)	18 (16–22)	0.216	0.096
SpO2 (%)	98 (95–100)	98 (95–100)	99 (96–100)	0.001	0.093	98 (95–100)	98 (95–100)	99 (96–100)	0.211	0.097
Severity of illness
MELD score	19.51 (13.7–27.22.7.22)	19.96 (14.03–27.83)	16.79 (11.88–22.68)	< 0.001	0.382	16.76 (11.88–22.84)	16.85 (12.23–23.00.23.00)	16.75 (11.81–22.70)	0.908	0.009
SOFA score	8 (5–11)	8 (5–11)	7 (4–10)	< 0.001	0.272	7 (5–10)	7 (5–10)	7 (4.25–10.25)	0.532	0.048
Charlson comorbidity index	6 (4–8)	5 (4–7)	7 (5–9)	< 0.001	0.385	7 (5–9)	7 (5–9)	7 (5–9)	0.761	0.024
Laboratory measurements
WBC (K/uL)	9.6 (6.2–14.7)	9.6 (6.2–14.8)	9.95 (6.6–14.5)	0.759	0.019	10 (6.5–14.4)	10 (6.5–14.25.5.25)	9.95 (6.425–14.5)	0.848	0.015
Hemoglobin (g/dL)	9.5 (8.1–10.9)	9.4 (8–10.8.8)	10 (8.6–11.325.6.325)	< 0.001	0.299	10 (8.675–11.5)	10 (8.6–11.6)	10.05 (8.7–11.4)	0.825	0.017
Platelet (K/uL)	104 (67–158)	102 (66–157)	119 (80.75–171)	< 0.001	0.197	115 (79–170.25.25)	112 (75.25–171.25.25.25)	118.5 (80.25–169.5)	0.586	0.042
ALT (U/dL)	34 (20–74)	33 (20–70)	38.5 (19–181)	< 0.001	0.261	35 (19–130)	32.5 (17.25–107)	39 (20–177.75.75)	0.969	0.003
AST (U/dL)	69 (39–156)	70 (40–149)	68 (35–336)	0.001	0.170	63 (35–252.25.25)	58.5 (34–159)	69 (37.25–334.75.25.75)	0.380	0.068
Total bilirubin (mg/dL)	2.7 (1.2–6.9)	3 (1.3–7.5)	1.4 (1–3.75.75)	< 0.001	0.433	1.6 (1–3.5.5)	1.75 (1–3.475.475)	1.4 (1–3.85.85)	0.635	0.037
ALB (g/dL)	2.9 (2.5–3.4)	2.9 (2.5–3.4)	2.95 (2.6–3.3)	0.897	0.008	2.9 (2.6–3.4)	2.9 (2.5–3.4)	2.9 (2.6–3.3)	0.950	0.005
Creatinine (mg/dL)	1.2 (1–2.1.1)	1.2 (1–2.1.1)	1.1 (1–2)	0.148	0.084	1.1 (1–1.8.8)	1.1 (1–1.8.8)	1.1 (1–1.975.975)	0.756	0.024
Lactate (mmol/L)	2.2 (1.5–3.5)	2.2 (1.5–3.5)	2.1 (1.5–3.025.5.025)	0.005	0.180	2.1 (1.5–3.5)	2.1 (1.5–3.5)	2.1 (1.5–3.075.5.075)	0.886	0.011
INR	1.7 (1.4–2.1)	1.7 (1.4–2.1)	1.5 (1.3–1.9)	0.001	0.188	1.5 (1.3–1.9)	1.5 (1.3–1.9)	1.5 (1.3–1.9)	0.764	0.023
PT (sec)	18.1 (15.1–22.9)	18.3 (15.2–23.2)	16.85 (14.4–20.525.4.525)	0.002	0.185	16.65 (14.3–20.625.3.625)	16.55 (14–20.875.875)	16.7 (14.5–20.575.5.575)	0.706	0.029
PPT (sec)	36.7 (31.3–46.3)	36.9 (31.5–46.5)	35.5 (30.7–44.225.7.225)	0.123	0.077	35.2 (30.5–44.1)	35.15 (30.425–45.425)	35.2 (30.6–43.45.6.45)	0.979	0.002
Complications or comorbidities
Ascites, n (%)	1491 (48.02)	1375 (49.87)	116 (33.33)	< 0.001	0.340	218 (32.63)	106 (31.74)	112 (33.53)	0.680	0.038
Hepatic encephalopathy, n (%)	304 (9.79)	293 (10.63)	11 (3.16)	< 0.001	0.298	19 (2.84)	8 (2.40)	11 (3.29)	0.642	0.054
Gastrointestinal bleeding, n (%)	98 (3.16)	93 (3.37)	5 (1.44)	0.074	0.127	12 (1.80)	7 (2.10)	5 (1.50)	0.771	0.045
Hepatorenal syndrome, n (%)	447 (14.40)	429 (15.56)	18 (5.17)	< 0.001	0.346	32 (4.79)	14 (4.19)	18 (5.39)	0.587	0.056
Liver transplantation, n (%)	41 (1.32)	25 (0.91)	16 (4.60)	< 0.001	0.227	26 (3.89)	11 (3.29)	15 (4.49)	0.548	0.062
Hypertension, n (%)	991 (31.92)	877 (31.81)	114 (32.76)	0.767	0.020	230 (34.43)	119 (35.63)	111 (33.23)	0.569	0.050
T2DM, n (%)	895 (28.82)	747 (27.09)	148 (42.53)	< 0.001	0.328	269 (40.27)	130 (38.92)	139 (41.62)	0.528	0.055
Cerebral infarction, n (%)	123 (3.96)	103 (3.74)	20 (5.75)	0.096	0.095	41 (6.14)	22 (6.59)	19 (5.69)	0.747	0.037
Myocardial infarction, n (%)	134 (4.32)	88 (3.19)	46 (13.22)	< 0.001	0.372	73 (10.93)	36 (10.78)	37 (11.08)	1.000	0.010
COPD, n (%)	397 (12.79)	340 (12.33)	57 (16.38)	0.041	0.116	111 (16.62)	56 (16.77)	55 (16.47)	1.000	0.008
Sepsis, n (%)	2282 (73.49)	2045 (74.17)	237 (68.10)	0.019	0.134	450 (67.37)	222 (66.47)	228 (68.26)	0.680	0.038
Treatments
Antibiotics, n (%)	2830 (91.14)	2523 (91.51)	307 (88.22)	0.053	0.109	590 (88.32)	296 (88.62)	294 (88.02)	0.904	0.019
Statins, n (%)	191 (6.15)	138 (5.01)	53 (15.23)	< 0.001	0.344	96 (14.37)	50 (14.97)	46 (13.77)	0.741	0.034
Non-selective beta-blockers, n (%)	319 (10.27)	263 (9.54)	56 (16.09)	< 0.001	0.197	104 (15.57)	50 (14.97)	54 (16.17)	0.749	0.033
Diuretics, n (%)	1185 (38.16)	1062 (38.52)	123 (35.34)	0.276	0.066	240 (35.93)	120 (35.93)	120 (35.93)	1.000	< 0.001
CRRT, n (%)	433 (13.95)	397 (14.40)	36 (10.34)	0.048	0.123	70 (10.48)	35 (10.48)	35 (10.48)	1.000	< 0.001
Mechanical ventilation, n (%)	2479 (79.84)	2189 (79.40)	290 (83.33)	0.098	0.101	546 (81.74)	269 (80.54)	277 (82.93)	0.483	0.062

Fig. 2.

Balance of baseline characteristics between aspirin non-users and aspirin users before and after propensity score matching. (The standardized mean differences (SMDs) for each covariate are displayed. Red circles represent the SMDs before matching, and blue circles represent the SMDs after matching. The vertical dashed line indicates the SMD threshold of 0.1, commonly used to signify a meaningful imbalance in covariates between groups. After propensity score matching, the SMDs for nearly all covariates were reduced below the 0.1 threshold, indicating improved balance between the aspirin user and non-user groups. Abbreviations: SMD, Standardized Mean Difference; ALT, Alanine Aminotransferase; AST, Aspartate Aminotransferase; T2DM, Type 2 Diabetes Mellitus; PPT, Partial Prothrombin Time; CRRT, Continuous Renal Replacement Therapy; COPD, Chronic Obstructive Pulmonary Disease; ALB, Albumin)

Aspirin and primary outcomes

The 30-day mortality rate among aspirin users was 15.27%, and the 90-day mortality rate was 16.77%, both lower than among non-users (Table 2). Kaplan-Meier analysis indicated that aspirin users had a greater probability of surviving 30 days and 90 days both before and after matching compared to non-users (Fig. 3). Supplementary Table 2 compares baseline characteristics between patients who survived 90 days and those who died within 90 days. Risk factors for 90-day mortality were identified through univariate Cox regression analysis (Supplementary Table 3). We established four models using multivariate Cox regression to assess the independent effect of aspirin therapy on 30-day and 90-day mortality. In the pre-matching cohort, mortality rates were significantly reduced at 30 days (HR 0.72, 95% CI 0.54–0.97) and 90 days (HR 0.68, 95% CI 0.51–0.90); in the post-matched cohort, mortality rates were lower at 30 days (HR 0.69, 95% CI 0.48–0.99) and 90 days (HR 0.65, 95% CI 0.46–0.92) (Table 3).

Table 2.

Association between aspirin and clinical outcomes in critically ill patients with liver cirrhosis

Variables	Total	Non aspirin	Aspirin	P-value	HR/OR (95%CI)
Before PSM	n = 3,105	n = 2,757	n = 348
Primary outcomes
30-day mortality, n(%)	757(24.38)	704(25.54)	53(15.23)	< 0.001	0.57(0.43–0.75)
90-day mortality, n(%)	853(27.47)	795(28.84)	58(16.67)	< 0.001	0.54(0.41–0.71)
Secondary outcomes
Hospital mortality, n(%)	1052(33.88)	972(35.26)	80(22.99)	< 0.001	0.55(0.42–0.71)
ICU stay(days)	3.02 (1.82–5.96)	3.06 (1.83–6.15)	2.80(1.64–4.95)	0.014
After PSM	n = 668	n = 334	n = 334
Primary outcomes
30-day mortality, n(%)	129(19.31)	78(23.35)	51(15.27)	0.01	0.63(0.44–0.90)
90-day mortality, n(%)	144(21.56)	88(26.35)	56(16.77)	0.004	0.61(0.44–0.85)
Secondary outcomes
Hospital mortality, n(%)	183(27.40)	105(31.44)	78(23.35)	0.02	0.66(0.47–0.94)
ICU stay(days)	2.8(1.78–4.97)	2.83(1.87–5.01)	2.80(1.64–4.94)	0.997

Fig. 3.

Kaplan-Meier survival curves comparing 30-day and 90-day mortality rates between aspirin non-users and users among critically ill cirrhosis patients in each cohort (A; B, in the original cohort; C; D, in the matched cohort)

Table 3.

Association of aspirin and the risk of 30-day and 90-day mortality and hospital mortality

Models	Before PSM			After PSM
	HR/OR	95%CI	P-value	HR/OR	95%CI	P-value
30-day mortality
Model 1	0.57	(0.43–0.75)	< 0.001	0.63	(0.44–0.90)	0.01
Model 2	0.57	(0.43–0.75)	< 0.001	0.63	(0.44–0.89)	0.009
Model 3	0.68	(0.51–0.90)	0.008	0.62	(0.43–0.88)	0.008
Model 4	0.72	(0.54–0.97)	0.03	0.69	(0.48–0.99)	0.047
90-day mortality
Model 1	0.54	(0.41–0.71)	< 0.001	0.61	(0.44–0.85)	0.004
Model 2	0.55	(0.42–0.71)	< 0.001	0.61	(0.43–0.85)	0.003
Model 3	0.64	(0.49–0.85)	0.002	0.60	(0.43–0.84)	0.003
Model 4	0.68	(0.51–0.90)	0.006	0.65	(0.46–0.92)	0.015
Hospital mortality
Model 1	0.55	(0.42–0.71)	< 0.001	0.66	(0.47–0.94)	0.02
Model 2	0.55	(0.42–0.71)	< 0.001	0.66	(0.47–0.93)	0.019
Model 3	0.63	(0.48–0.84)	0.002	0.62	(0.43–0.89)	0.01
Model 4	0.67	(0.49–0.90)	0.009	0.66	(0.44–0.99)	0.044

Model 1: unadjusted

Model 2: adjusted for ethnicity

Model 3: adjusted for ethnicity, ascites, hepatorenal syndrome, liver transplantation, hypertension, myocardial infarction, sepsis, antibiotics, statins, non-selective beta-blockers, diuretics, CRRT and mechanical ventilation

Model 4: adjusted for ethnicity, ascites, hepatorenal syndrome, liver transplantation, hypertension, myocardial infarction, sepsis, antibiotics, statins, non-selective beta-blockers, diuretics, CRRT, mechanical ventilation, heart rate, respiratory rate, SpO2, WBC, hemoglobin, total bilirubin, creatinine, lactate, INR, PT, PPT, MELD score, SOFA score and Charlson comorbidity index

Aspirin and secondary outcomes

Before propensity score matching (PSM), the aspirin group exhibited lower in-hospital mortality and shorter intensive care unit (ICU) length of stay compared to the non-aspirin group. The difference in in-hospital mortality remained significant only after PSM (23.35% vs. 31.44%, P = 0.02). Multivariate logistic regression analysis indicated that aspirin therapy itself was an independent protective factor for in-hospital mortality both before and after PSM.

Aspirin dose and outcomes

In the cohort of critically ill cirrhosis patients included in this study (n = 3,105), a total of 348 patients (11.2%) received aspirin during their ICU hospitalization. Among these, low-dose (81 mg/day) was the primary regimen for 282 patients (81.0% of those receiving aspirin), while high-dose (325 mg/day) was administered to 52 patients (14.9% of those receiving aspirin) (Supplementary Table 4). Due to significant differences in sample size between the two groups and imbalances in baseline characteristics, we were unable to construct a comparable cohort using methods such as propensity score matching to compare outcomes between high and low doses. Therefore, all subsequent analyses in this study examining the association between aspirin exposure and outcomes treated aspirin as a single exposure factor, without conducting dose-subgroup analyses.

Subgroup analysis

In our detailed analysis, we discovered that using aspirin did not have important differences based on ethnicity, age, gender, cerebral infarction, hypertension, T2DM, myocardial infarction, COPD, and sepsis groups (Fig. 4).

Fig. 4.

Subgroup analysis of the association between aspirin use and mortality risk in critically ill patients with liver cirrhosis (A, 30-day mortality; B, 90-day mortality)

Discussion

This study, based on the MIMIC-IV database, confirms that aspirin therapy is significantly associated with improved 30-day and 90-day survival rates and reduced in-hospital mortality in critically ill patients with liver cirrhosis. This finding provides new clinical evidence for the therapeutic value of aspirin in this specific population.

Relatively well-defined mechanisms explain the protective effects of aspirin in cirrhosis. Existing research indicates it inhibits liver fibrosis progression and reduces hepatocellular carcinoma (HCC) risk through multi-target mechanisms: on one hand, it delays fibrosis by modulating the TGF-β1/Smad pathway, inhibiting COX-2 activity, and improving hepatic sinusoidal endothelial dysfunction [28, 29]; on the other hand, it exerts antitumor effects by suppressing the NF-κB signaling pathway, downregulating GLUT1 expression, and exerting antiplatelet actions [30, 31]. These mechanisms collectively form the theoretical basis for aspirin -treated cirrhosis.

Numerous clinical studies have substantiated the advantages of aspirin for patients with cirrhosis. A large cohort study based on Taiwan’s CGRD database demonstrated that long-term low-dose aspirin use significantly reduced the risk of HCC (3-year HR 0.57, 5-year HR 0.63) and all-cause mortality (3-year HR 0.43, 5-year HR 0.51) without increasing gastrointestinal bleeding risk [32]. Another single-center study further confirmed that aspirin users not only had a reduced risk of HCC but also a significantly lower risk of the composite outcome of hepatic encephalopathy and ascites [33]. Among liver transplant candidates, the combination of aspirin and statins demonstrated synergistic protective effects. The combined therapy group showed significantly better INR, total bilirubin, and MELD scores compared to monotherapy, along with markedly reduced mortality (23.5% vs. 76.5%, P = 0.021) [34]. Collectively, these findings support the potential for aspirin to confer multiple clinical benefits in patients with cirrhosis.

This study specifically focused on critically ill cirrhosis patients. Unlike the meta-analysis results by Abdel Malak et al. [35]. in the general cirrhosis population, our study found that aspirin conferred more pronounced survival benefits in the ICU setting. We believe this discrepancy reflects the critical role of clinical context: ICU patients often experience systemic inflammation and highly activated platelets, which are core drivers of multiple organ failure. Aspirin may exert more pronounced protective effects by suppressing this “inflammatory storm.” We observed that among cirrhotic patients treated in the ICU, those receiving aspirin had significantly higher rates of concomitant type 2 diabetes (42.53% vs. 27.09%), myocardial infarction (13.22% vs. 3.19%), chronic obstructive pulmonary disease (COPD) (16.38% vs. 12.33%), and liver transplantation (4.60% vs. 0.91%) (p < 0.01). This disparity likely stems from aspirin’s clinical dosing characteristics and its potential protective effects: Myocardial infarction patients typically require long-term aspirin therapy for secondary prevention due to elevated cardiovascular risk; type 2 diabetes patients may benefit from aspirin’s ability to improve insulin sensitivity and glycemic control, while COPD patients are more likely to take aspirin for its anti-inflammatory effects [36]. Furthermore, the higher liver transplantation rate may correlate with aspirin’s multifaceted protective effects—studies indicate this drug not only improves prognosis by reducing portal hypertension and bleeding risks but also extends transplant candidates’ survival by decreasing hepatocellular carcinoma (HCC) incidence, a key indication for liver transplantation [35].

These findings echo the protective effects reported by Wang et al. [27]. in sepsis-associated liver injury but extend the potential beneficiary population of aspirin beyond the specific etiology of sepsis to encompass a broader cohort of critically ill patients with underlying chronic end-stage liver cirrhosis experiencing decompensation due to various causes. This extension suggests that aspirin’s effects may have broader pathophysiological underpinnings and provides new theoretical grounds for targeted studies in highly heterogeneous critically ill cirrhotic populations.

This study has several limitations that warrant careful consideration. First, as a retrospective observational study based on single-center ICU data, its conclusions primarily apply to critically ill cirrhosis patients and may not be generalized to stable community or outpatient populations. Although propensity score matching was used to adjust for known confounders, the inherent limitations of retrospective design prevent complete exclusion of unmeasured confounders—such as the possibility that patients in the aspirin group received more comprehensive medical care due to higher prevalence of cardiovascular comorbidities. More critically, the absence of multiple core data points undermines the study’s depth and clinical applicability. First, we lacked access to patients’ pre-hospitalization aspirin history, preventing differentiation between long-term users and those who initiated treatment during hospitalization. This limitation hindered in-depth analysis of the drug’s efficacy patterns. Second, the MIMIC-IV database does not contain etiological information regarding cirrhosis (e.g., viral, alcoholic, metabolic), which is a significant issue since the cause of the disease has a major impact on the natural history of liver disease and the patient’s prognosis. For instance, patients with metabolic dysfunction-associated fatty liver disease (MASLD) typically bear a higher burden of cardiovascular comorbidities, which may independently influence mortality and potentially confound the observed association with aspirin. Similarly, differing pathophysiological and inflammatory environments across etiologies could alter aspirin’s therapeutic effects. Despite our adjustment for available disease severity measures (e.g., MELD score) and comorbidities, we cannot exclude residual confounding due to etiology. Therefore, we need to determine the universality of our findings across all cirrhosis subtypes. Future prospective studies must prioritize collecting etiological data to validate and refine our results. Third, we were unable to assess aspirin-related bleeding risks, despite high-level evidence (e.g., Abdel Malak et al., 2023)(1) demonstrating such risks in cirrhotic populations (HR = 1.11). This requires very careful interpretation of the observed mortality benefit within a risk-benefit balancing framework. Fourth, lacking precise cause-of-death information, we could only report all-cause mortality rates. This feature prevents clarification of whether aspirin exerts its protective effect by improving liver-specific complications or by influencing systemic pathological processes. Finally, due to the influence of clinical dosing patterns (81.0% receiving 81 mg/day), the high-dose group exhibited insufficient sample size and confounding indications, rendering reliable dose-response analysis unfeasible. Collectively, these limitations mean that while this study suggests potential survival benefits from aspirin, it cannot definitively establish its clinical net benefit in this critically ill population. Future studies should prospectively collect key data on the timing of medication initiation, etiology, bleeding events, and specific causes of death, and validate the efficacy and safety of aspirin across different cirrhosis subgroups. Simultaneously, additional investigation into aspirin’s distinctive efficacy in mitigating systemic inflammation and platelet activation is warranted, contingent upon the pathophysiological condition of critically ill patients.

Conclusion

This single-Centre retrospective study found that aspirin use was associated with reduced mortality in critically ill patients with liver cirrhosis. However, the overall net therapeutic effect remains to be further validated due to the lack of data on bleeding risk. The relationship between aspirin dosage and specific outcomes requires clarification in future studies with larger sample sizes and more prospective designs.

Abbreviations

ALT

Alanine Aminotransferase

AST

Aspartate Aminotransferase

ALB

Serum Albumin

COPD

Chronic Obstructive Pulmonary Disease

CRRT

Continuous Renal Replacement Therapy

HR

Hazard Ratio

INR

International Normalized Ratio

ICU

Intensive Care Unit

MELD

Model for End-stage Liver Disease

MIMIC-IV

Medical Information Mart for Intensive Care-IV

OR

Odds Ratio

PT

Prothrombin Time

PTT

Partial Thromboplastin Time

PSM

Propensity Score Matching

SpO2

Blood Oxygen Saturation

SMD

Standardized Mean Difference

SOFA

Sequential Organ Failure Assessment

T2DM

Type 2 Diabetes Mellitus

WBC

White Blood Cell Count

Authors’ contributions

Yu Yi designed the research; Yinghua Chen extracted and analyzed the data; Yu Yi wrote the manuscript; Yu Yi collated and interpreted the data; and Yinghua Chen modified the manuscript and interpreted the analysis. Yawen Luo reviewed the manuscript. All authors contributed to the article and approved the final submission.

Funding

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Data availability

Data is provided within the manuscript or supplementary information files.

Declarations

Ethics approval and consent to participate

Not Applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.