Impact of in-hospital ACEI use on long-term prognosis in discharged type 2 myocardial infarction patients: a retrospective propensity score-matched cohort study

Abstract

Objective

This study aimed to investigate the association between angiotensin-converting enzyme inhibitor (ACEI) use during hospitalization and short- and long-term prognosis in patients with type 2 myocardial infarction (T2MI) after discharge.

Methods

We conducted a retrospective cohort study using data from the Medical Information Mart for Intensive Care IV (MIMIC-IV) database. The analysis included adult critically ill patients diagnosed with T2MI. Exposure was defined as ACEI administration during the intensive care unit (ICU) stay. The primary outcome was 3-year all-cause mortality. Propensity score matching (PSM) was performed at a 1:1 ratio, and matching quality was assessed using standardized mean differences (SMD); all absolute SMD values for matched covariates were below 0.1. Multivariate analysis was used to adjust for potential confounders.

Results

A total of 1,086 T2MI patients were included. After PSM, 590 patients (295 per group) were analyzed. ACEI use was associated with significantly lower 3-year all-cause mortality (hazard ratio [HR], 0.55; 95% confidence interval [CI], 0.41–0.74; P < 0.001). Sensitivity analysis further supported this association (HR, 0.58; 95% CI, 0.42–0.80; P < 0.001). Additionally, multivariate Cox analysis indicated that ACEI use was associated with reduced risk of 180-day all-cause mortality (HR, 0.49; 95% CI, 0.33–0.72; P < 0.001). However, no significant associations were observed between ACEI use and 3-year all-cause readmission (HR, 1.003; 95% CI, 0.776–1.295; P = 0.984) or 3-year T2MI recurrence (HR, 1.14; 95% CI, 0.64–2.05; P = 0.654).

Conclusion

In this observational study, ACEI use during hospitalization was associated with significantly lower short- and long-term all-cause mortality in patients discharged after T2MI, but was not significantly associated with readmission or T2MI recurrence rates. These findings suggest a potential clinical relevance of ACEIs in T2MI populations and may expand the understanding of their role in myocardial infarction with different pathogenesis. Further prospective studies are warranted to clarify optimal ACEI administration strategies in this patient group.

Clinical trial number

Not applicable.

Key Question

Does ACEI use during hospitalization affect the long-term prognosis of T2MI patients after discharge?

Key Finding

In discharged T2MI patients, the use of ACEI was associated with significantly lower 3-year and 180-day all-cause mortality, but was not associated with a 3-year all-cause readmission or risk of T2MI recurrence.

Take Home Message

The use of ACEI was associated with a significant lower short- and long-term all-cause mortality in patients with discharged T2MI; nevertheless, it does not affect readmission or T2MI recurrence rates. These findings broaden the clinical use of ACEI to include patients with myocardial infarction caused by different pathophysiological mechanisms.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12872-026-05697-w.

Introduction

Type 2 myocardial infarction (T2MI) is myocardial necrosis caused by an imbalance between myocardial oxygen supply and demand. This imbalance is not related to coronary plaque rupture [1]. Its main risk factors include reduced myocardial oxygen supply, such as anemia, hypoxemia, and coronary artery spasm, or increased myocardial oxygen demand, such as tachycardia, hypertensive crisis, and severe infection [1, 2]. These risk factors are more common in critically ill patients. The treatment strategy for T2MI mainly focuses on addressing the primary condition, such as correcting anemia, improving hypoxemia, or managing shock, rather than on coronary revascularization [3]. Existing studies have found that, in addition to treating the primary disease, lipid-lowering therapy (hazard ratio [HR] 0.77, 95% confidence interval [CI] 0.61–0.97) [4], sodium-dependent glucose transporters 2 (SGLT2) inhibitors (HR 0.67, 95% CI 0.41–1.10) [5, 6], and glucagon-like peptide-1 (GLP-1) agonists (HR 0.65, 95% CI 0.46–0.92) [7] can significantly reduce the incidence of T2MI. However, in patients diagnosed with T2MI, it remains unclear whether antiplatelet agents [8–10], lipid-lowering drugs [8, 11], beta-blockers [8–10], and ACEI [8, 10] provide clinical benefits. The treatment of T2MI still lacks standardized plans or recommendations to date [12–14].

It is worth noting that in patients with type 1 myocardial infarction (T1MI), ACEIs mainly exert their effects by inhibiting the renin-angiotensin system (RAS). This inhibition reduces the production of angiotensin II (Ang II) and increases bradykinin levels. These actions lead to vasodilation, reduction of cardiac load, and delay of myocardial remodeling, thereby lowering the mortality rate of patients with acute myocardial infarction (AMI) [15–17]. Additionally, relevant guidelines clearly state that for myocardial infarction patients with a left ventricular ejection fraction (LVEF) ≤ 40%, or those with complications such as hypertension, diabetes mellitus, or chronic kidney disease, ACEI is recommended to be administered as early as possible and continued over the long term in the absence of contraindications [18]. Therefore, these findings indicate that the use of ACEI is of great significance in the T1MI population. A recent study found that ACEI use may reduce 2-year mortality among T2MI patients [19], suggesting that ACEI not only benefits the T1MI population, traditionally affected by coronary plaque rupture, but also potentially offers protection to T2MI patients.

However, during the acute phase of T2MI, the severity of the primary condition and treatment strategy may have a stronger impact on short-term mortality risk in T2MI patients. This influence may reduce the prognostic value of traditional myocardial infarction treatments, such as ACEI use. These findings suggest that T2MI patients who show improvement in their primary condition following treatment require increased clinical attention. Therefore, this study aims to explore the relationship between ACEI use during hospitalization and the short- and long-term prognosis of T2MI survivors, to provide evidence supporting ACEI use in this patient population.

Methods

Data sources

All data were obtained from the Medical Information Mart for Intensive Care IV version 3.1 (MIMIC-IV 3.1) database [20]. This openly accessible critical care database contains electronic medical records of 546,028 patients. The data were collected by Beth Israel Deaconess Medical Center in Boston, Massachusetts, from 2008 to 2022. Patients' demographic information, laboratory tests, vital signs, hospital status, medication, and surgical procedures are documented in detail in the MIMIC-IV 3.1 database. All information about patients' identification has been anonymized, and all identifiable information has been removed. Therefore, informed consent was waived. The author completed the required data research training from the Collaborative Institutional Training Initiative to obtain database access permission (Record ID: 52,310,626). This study was reported in accordance with the STROBE guidelines for observational cohort studies [21].

Study population

In the MIMIC-IV 3.1 database, all patients with a first admission from 2015 to 2022 were extracted. The diagnosis of T2MI was determined using the International Classification of Diseases (ICD)−10 code I12.A1. The ICD-10 code I12.A1 in the MIMIC-IV database corresponds to “non-plaque ruptured oxygen-depletion imbalanced myocardial infarction” and its use for identifying T2MI patients has been validated in previous studies [19].

Inclusion criteria

Patients with the ICD10 diagnostic code I12.A1 were included.

Exclusion criteria

1. Patients who died during hospitalization.

2. Patients with missing data on ACEI medication.

Data collection

Data extraction was performed using Structured Query Language (SQL) scripts retrieved from the GitHub website (https://github.com/MIT-LCP/mimic-iv).

Relevant patient information was collected, including the following categories:

Demographic data

Gender and age;

Comorbidities

Charlson Comorbidity Index (CCI), old myocardial infarction, atrial fibrillation, congestive heart failure, cerebrovascular disease, chronic lung disease, diabetes mellitus, kidney disease, malignant tumor, hypertension, and hyperlipidemia;

Risk factors for T2MI

Sepsis, acute respiratory distress syndrome(ARDS), acute kidney injury, septic shock, tachyarrhythmia, bradyarrhythmia, acute respiratory failure with hypoxemia, hypotension, and acute hemorrhagic anemia;

Medication use

Vasoactive drugs, anticoagulants, antiplatelet drugs, statins, diuretics, angiotensin II receptor blockers (ARBs), digitalis drugs, antibiotics, and beta-blockers.

Special treatments received during hospitalization

Continuous renal replacement therapy (CRRT), mechanical ventilation(MV), and length of hospital stay.

Exposure and outcome

Patients were categorized into an unexposed group if they did not receive ACEI treatment during hospitalization, and into an exposed group if they did. For those in the exposed group, ACEI therapy was typically initiated after admission to the intensive care unit (ICU), provided no contraindications were present. Data on ACEI exposure were obtained from medical prescription records, and patients with missing exposure data were excluded from the analysis.

The primary endpoint was 3-year all-cause mortality. This duration was selected to capture intermediate-term cardiovascular outcomes and mortality relevant to the pathophysiology of type 2 myocardial infarction, balancing the need for sufficient event accrual with the practical constraints of follow-up in a critically ill cohort. Moreover, the 3-year follow-up period was chosen based on the availability of data in the MIMIC-IV database, where the median follow-up time is 3.2 years. This aligns with the standard follow-up duration for cardiovascular disease prognosis studies [15, 19] and adequately reflects long-term outcomes.

Secondary endpoints included 180-day all-cause mortality, 3-year readmission rate, and T2MI recurrence rate.

Statistical analysis

This was a retrospective analysis conducted without a predefined statistical analysis plan. No statistical power calculation was performed, and the sample size was determined based on the available data in the database. The study cohort was divided into two groups: patients who received ACE inhibitor (ACEI) treatment (the ACEI use group) and those who did not (the non-use group). Variance inflation factors (VIF) were used to assess multicollinearity among variables.

For continuous data, normality was evaluated using the Shapiro–Wilk test. Normally distributed data were expressed as mean ± standard deviation and compared between groups with the independent samples t-test. Non-normally distributed data were presented as median (interquartile range [IQR]) and compared using the Mann–Whitney U test. Categorical data were expressed as percentages and analyzed with the chi-square test for inter-group comparisons.

To estimate the hazard ratio (HR) and its 95% confidence interval (CI) for the primary outcome of 3-year all-cause mortality, a Cox proportional hazards model was employed to evaluate the independent effect of the exposure factor on survival time. The cumulative incidence of 3-year all-cause mortality was estimated with the Kaplan–Meier method and compared via the log-rank test. For binary outcomes, including 180-day all-cause mortality and 180-day and 3-year all-cause readmission rates, Cox regression was used to calculate the HR and 95% CI. The cumulative incidences of secondary outcomes were also estimated using the Kaplan–Meier method to aid in result interpretation.

All analyses were performed with R software (version 4.3.3), and a two-tailed P-value < 0.05 was considered statistically significant.

PSM

The primary analysis was conducted within the propensity score-matched cohort to investigate the associations between ACEI use and the primary and secondary outcomes. To mitigate potential confounding, we employed propensity score matching (PSM). The propensity score, defined as the probability of receiving ACEI treatment for each patient, was estimated using a multivariable logistic regression model.

The variables selected for the propensity score model were based on established clinical knowledge and prior literature, encompassing demographic factors, comorbidities, presenting conditions, and key treatments. These included: gender; age; Charlson Comorbidity Index (CCI) score; history of old myocardial infarction, atrial fibrillation, congestive heart failure, cerebrovascular disease, chronic lung disease, diabetes mellitus, nephropathy, malignancy, hypertension, and hyperlipidemia; as well as the presence of sepsis, acute respiratory distress syndrome (ARDS), acute kidney injury, septic shock, tachyarrhythmia, bradyarrhythmia, acute hypoxemic respiratory failure, hypotension, and acute hemorrhagic anemia. Concomitant medication/treatment use was also adjusted for, including: vasoactive agents; anticoagulants; antiplatelet agents; statins; diuretics; angiotensin II receptor blockers (ARBs); digitalis; antibiotics; beta-blockers; continuous renal replacement therapy (CRRT); invasive mechanical ventilation (MV); and length of hospital stay.

A 1:1 nearest-neighbor matching algorithm was applied without replacement, using a caliper width of 0.05 of the standard deviation of the propensity score logit. The balance of covariates between the ACEI and non-ACEI groups before and after matching was assessed using the standardized mean difference (SMD), with an SMD of less than 0.10 considered indicative of adequate balance.

In the matched cohort, univariate analyses were first performed. Variables showing an association with the outcomes at a significance level of P < 0.05 in these analyses—such as CCI score, history of old myocardial infarction, congestive heart failure, malignancy, hyperlipidemia, and use of antiplatelet drugs, statins, diuretics, ARBs, and beta-blockers—were subsequently included in the final multivariable regression models for further adjustment.

Doubly robust analysis

To evaluate the robustness of the study findings against potential misspecification of the statistical model, we performed doubly robust analyses in the propensity score-matched cohort. Four analytical approaches were compared:

1. Original propensity score matching followed by a stratified Cox model;

2. An inverse probability of treatment weighting (IPTW) Cox model with stabilized weights and robust standard errors;

3. A multivariable-adjusted Cox model incorporating covariates such as age, sex, Charlson Comorbidity Index, atrial fibrillation, sepsis, acute kidney injury, and shock;

4. An augmented IPTW estimator that combines propensity score weighting with outcome regression adjustment. Covariate balance was evaluated using standardized mean differences (SMD), with |SMD|< 0.10 indicating adequate balance.

Subgroup analysis

Subgroup analyses were conducted within the matched cohort based on the following categories: age (≤ 65 years, 65–80 years, ≥ 80 years), presence of tachycardia (yes/no), Charlson Comorbidity Index score (< 7, 7–8, > 8), and use of digitalis (yes/no) [15, 16].

Sensitivity analysis

To further assess the robustness of the matched cohort results, we performed sensitivity analyses on the full dataset. These included demographic factors (sex, age), clinical measures (CCI), comorbidities (prior myocardial infarction, atrial fibrillation, congestive heart failure, cerebrovascular disease, chronic lung disease, diabetes, nephropathy, malignancy, hypertension, hyperlipidemia, sepsis, ARDS, acute kidney injury, septic shock, tachyarrhythmia, bradyarrhythmia, acute respiratory failure with hypoxemia, hypotension, acute hemorrhagic anemia), medications (vasoactive drugs, anticoagulants, antiplatelet agents, statins, diuretics, ARBs, digitalis, antibiotics, beta-blockers), treatments (CRRT, mechanical ventilation), and length of hospital stay.

We note that the exact timing of ACEI initiation during hospitalization is not consistently documented in the MIMIC-IV database for oral medications. To address potential immortal time bias arising from this limitation, we conducted the following sensitivity analyses:

1. 7-Day Landmark Analysis: Patients with a hospital stay shorter than 7 days (n = 241) were excluded to ensure all included patients survived a minimum period during which ACEI could have been prescribed.

2. IPTW: Propensity scores were re-estimated using baseline covariates measured at admission, and IPTW was applied to balance characteristics between ACEI users and non-users.

3. Stratification by Hospital Length of Stay: Patients were stratified by duration of hospitalization (7–10 days, 11–14 days, > 14 days) to examine consistency across subgroups.

All sensitivity analyses were conducted using R version 4.3.3. The “survival” package was used for Cox proportional hazards models, “WeightIt” for IPTW estimation, and “cobalt” for balance diagnostics. Two-sided P-values < 0.05 were considered statistically significant.

Results

Patient selection

A total of 122,905 records were identified. After applying the inclusion and exclusion criteria, 1,086 patients with T2MI were finally included, of whom 302 (27.8%) received ACEI therapy during hospitalization. The matched cohort consisted of 590 patients, with 295 in each group. Figure 1 illustrates the patient selection process.

Fig. 1.

Flowchart of study population

Cohort characteristics

Table 1 presents the baseline characteristics of patients before and after matching. The VIF of each variable was less than 5, indicating no multicollinearity (Supplementary Table 1). In the entire cohort, significant differences were observed in confounding factors-including CCI, old myocardial infarction, congestive heart failure, malignant tumor, and hyperlipidemia-as well as medication use such as antiplatelet, statins, diuretics, ARBs and beta-blockers between patients who received ACEI treatment (P < 0.05) (Supplementary Table 2). Matching improved variable balance, achieving an absolute SMD of less than 0.10, as shown in Supplementary Figs. 1 and 2, which illustrate the distribution balance before and after PSM.

Table 1.

Comparison of baseline characteristics before and after PSM

Variable	Before PSM					After PSM
	Total (n = 1086)	Non-ACEI (n = 784)	ACEI (n = 302)	Statistic	SMD	Total (n = 590)	Non-ACEI (n = 295)	ACEI (n = 295)	Statistic	SMD
Demographic data
Age, Mean ± SD	73.61 ± 14.27	73.85 ± 14.71	73.00 ± 13.06	t = 0.879	−0.065	73.22 ± 13.40	73.40 ± 13.84	73.04 ± 12.96	t = 0.320	−0.027
Male, n (%)	620 (57.09)	438 (55.87)	182 (60.26)	χ2 = 1.721	0.090	337 (57.12)	161 (54.58)	176 (59.66)	χ2 = 1.557	0.104
Comorbidity
CCI, Mean ± SD	7.82 ± 2.72	7.95 ± 2.85	7.48 ± 2.35	t = 2.744	−0.198	7.50 ± 2.52	7.50 ± 2.68	7.51 ± 2.36	t = −0.016	0.001
OLD MI, n (%)	88 (8.1)	54 (6.89)	34 (11.26)	χ2 = 5.593	0.138	63 (10.68)	32 (10.85)	31 (10.51)	χ2 = 0.018	−0.011
Atrial Fibrillation, n (%)	407 (37.48)	301 (38.39)	106 (35.10)	χ2 = 1.009	−0.069	203 (34.41)	98 (33.22)	105 (35.59)	χ2 = 0.368	0.050
Congestive Heart Failure, n (%)	599 (55.16)	416 (53.06)	183 (60.60)	χ2 = 5.004	0.154	333 (56.44)	155 (52.54)	178 (60.34)	χ2 = 3.647	0.159
Cerebrovascular Disease, n (%)	178 (16.39)	127 (16.20)	51 (16.89)	χ2 = 0.075	0.018	101 (17.12)	51 (17.29)	50 (16.95)	χ2 = 0.012	−0.009
Chronic Pulmonary Disease, n (%)	265 (24.4)	195 (24.87)	70 (23.18)	χ2 = 0.339	−0.040	142 (24.07)	73 (24.75)	69 (23.39)	χ2 = 0.148	−0.032
Diabetes without cc,n(%)	258 (23.76)	174 (22.19)	84 (27.81)	χ2 = 3.803	0.125	147 (24.92)	64 (21.69)	83 (28.14)	χ2 = 3.271	0.143
Diabetes with cc, n (%)	307 (28.27)	220 (28.06)	87 (28.81)	χ2 = 0.060	0.016	155 (26.27)	68 (23.05)	87 (29.49)	χ2 = 3.159	0.141
Renal Disease, n (%)	430 (39.59)	323 (41.20)	107 (35.43)	χ2 = 3.033	−0.121	208 (35.25)	103 (34.92)	105 (35.59)	χ2 = 0.030	0.014
Malignant Cancer, n (%)	135 (12.43)	116 (14.80)	19 (6.29)	χ2 = 14.486	−0.350	49 (8.31)	30 (10.17)	19 (6.44)	χ2 = 2.693	−0.152
Hypertension, n (%)	380 (34.99)	264 (33.67)	116 (38.41)	χ2 = 2.151	0.097	206 (34.92)	94 (31.86)	112 (37.97)	χ2 = 2.417	0.126
Hyperlipidemia, n (%)	579 (53.31)	403 (51.40)	176 (58.28)	χ2 = 4.140	0.139	319 (54.07)	145 (49.15)	174 (58.98)	χ2 = 5.740	0.200
Risk factors for T2MI
Sepsis, n (%)	215 (19.8)	157 (20.03)	58 (19.21)	χ2 = 0.092	−0.021	113 (19.15)	58 (19.66)	55 (18.64)	χ2 = 0.099	−0.026
ARDS, n (%)	11 (1.01)	6 (0.77)	5 (1.66)	χ2 = 0.950	0.070	6 (1.02)	2 (0.68)	4 (1.36)	χ2 = 0.168	0.059
AKI, n (%)	545 (50.18)	403 (51.40)	142 (47.02)	χ2 = 1.675	−0.088	280 (47.46)	139 (47.12)	141 (47.80)	χ2 = 0.027	0.014
Septic Shock, n (%)	157 (14.46)	116 (14.80)	41 (13.58)	χ2 = 0.262	−0.036	77 (13.05)	38 (12.88)	39 (13.22)	χ2 = 0.015	0.010
Tachycardia, n (%)	133 (12.25)	96 (12.24)	37 (12.25)	χ2 = 0.000	0.000	75 (12.71)	39 (13.22)	36 (12.20)	χ2 = 0.137	−0.031
Bradycardia, n (%)	28 (2.58)	2 (2.55)	8 (2.65)	χ2 = 0.008	0.006	18 (3.05)	10 (3.39)	8 (2.71)	χ2 = 0.229	−0.042
ARF hypoxia, n (%)	300 (27.62)	223 (28.44)	77 (25.50)	χ2 = 0.947	−0.068	147 (24.92)	73 (24.75)	74 (25.08)	χ2 = 0.009	0.008
Hypotension, n (%)	108 (9.94)	80 (10.20)	28 (9.27)	χ2 = 0.212	−0.032	61 (10.34)	33 (11.19)	28 (9.49)	χ2 = 0.457	−0.058
Acute posthemorrhagic anemia, n (%)	189 (17.4)	140 (17.86)	49 (16.23)	χ2 = 0.404	−0.044	98 (16.61)	49 (16.61)	49 (16.61)	χ2 = 0.000	0.000
Medication status
Vasoactive agent, n (%)	606 (55.80)	445 (56.76)	161 (53.51)	χ2 = 1.052	−0.069	328 (55.59)	171 (57.97)	157 (53.22)	χ2 = 1.346	−0.095
Anticoagulant, n (%)	1025 (94.38)	735 (93.75)	290 (96.03)	χ2 = 2.131	0.117	563 (95.42)	280 (94.92)	283 (95.93)	χ2 = 0.349	0.051
Antiplatelet, n (%)	780 (71.82)	532 (67.86)	248 (82.12)	χ2 = 21.913	0.372	481 (81.53)	238 (80.68)	243 (82.37)	χ2 = 0.281	0.044
Statin drugs, n (%)	827 (76.15)	571 (72.83)	256 (84.77)	χ2 = 17.104	0.332	491 (83.22)	242 (82.03)	249 (84.41)	χ2 = 0.595	0.065
Diuretic drugs, n (%)	691 (63.63)	483 (61.61)	208 (68.87)	χ2 = 4.975	0.157	400 (67.8)	199 (67.46)	201 (68.14)	χ2 = 0.031	0.015
ARBs, n (%)	178 (16.39)	156 (19.90)	22 (7.28)	χ2 = 25.310	−0.485	43 (7.29)	21 (7.12)	22 (7.46)	χ2 = 0.025	0.013
Digitalis drugs, n (%)	36 (3.31)	23 (2.93)	13 (4.30)	χ2 = 1.279	0.068	26 (4.41)	14 (4.75)	12 (4.07)	χ2 = 0.161	−0.034
Antibiotic, n (%)	662 (60.96)	471 (60.08)	191 (63.25)	χ2 = 0.920	0.066	373 (63.22)	188 (63.73)	185 (62.71)	χ2 = 0.066	−0.021
Beta receptor blockers, n (%)	810 (74.59)	555 (70.79)	255 (84.44)	χ2 = 21.418	0.376	498 (84.41)	250 (84.75)	248 (84.07)	χ2 = 0.052	−0.019
Special treatment
CRRT, n (%)	31 (2.85)	25 (3.19)	6 (1.99)	χ2 = 1.136	−0.086	10 (1.69)	4 (1.36)	6 (2.03)	χ2 = 0.407	0.048
MV, n (%)	449 (41.34)	323 (41.20)	126 (41.72)	χ2 = 0.025	0.011	244 (41.36)	122 (41.36)	122 (41.36)	χ2 = 0.000	< 0.001
Length of stay, Mean ± SD	11.94 ± 13.41	11.87 ± 13.59	12.12 ± 12.95	t = −0.274	0.019	11.95 ± 12.40	12.20 ± 13.37	11.69 ± 11.36	t = 0.498	−0.045

Primary outcome

Figure 2 presents the Kaplan–Meier curve for 3-year all-cause mortality stratified by ACEI use. The analysis revealed that ACEI use was associated with a significant reduction in 3-year all-cause mortality (HR, 0.55; 95% CI, 0.41–0.74; P < 0.001). Univariate Cox proportional hazards analysis confirmed this favorable association (HR, 0.55; 95% CI, 0.40–0.75; P < 0.001). Other factors, including age, CCI, atrial fibrillation, congestive heart failure, chronic lung disease, kidney disease, malignant tumor, ARB use, and digitalis use, also showed statistically significant associations. Variables with a p-value < 0.1 in the univariate analysis were included in a subsequent multivariate Cox proportional hazards model to identify independent predictors. After adjusting for these relevant factors, ACEI use remained independently associated with a lower risk of 3-year all-cause death (HR, 0.59; 95% CI, 0.43–0.79; P < 0.001), as detailed in Table 2.

Fig. 2.

Kaplan–Meier curves for 3-year all-cause mortality matched by ACEI use

Table 2.

Univariate and Multivariate Cox Models for 3-Year all-cause mortality

Variables	Univariate analysis					Multivariate analysis
	Β	S.E	Z	P	HR (95%CI)	β	S.E	Z	P	HR (95%CI)
Demographic data
Age	0.05	0.01	7.03	< 0.001	1.05 (1.04 ~ 1.06)	0.04	0.01	5.61	< 0.001	1.05 (1.03 ~ 1.06)
Male	−0.20	0.15	−1.38	0.167	0.82 (0.61 ~ 1.09)
Comorbidity
CCI	0.22	0.03	7.86	< 0.001	1.25 (1.18 ~ 1.32)	0.16	0.03	5.50	< 0.001	1.18 (1.11 ~ 1.25)
OLD MI	0.05	0.23	0.23	0.819	1.05 (0.67 ~ 1.67)
Atrial Fibrillation	0.32	0.16	2.06	0.039	1.38 (1.02 ~ 1.88)
Congestive Heart Failure, n (%)	0.47	0.16	2.98	0.003	1.60 (1.18 ~ 2.19)
Cerebrovascular Disease, n (%)	0.08	0.20	0.40	0.686	1.08 (0.74 ~ 1.59)
Chronic Pulmonary Disease, n (%)	0.35	0.16	2.12	0.034	1.42 (1.03 ~ 1.96)	0.26	0.17	1.52	0.129	1.29 (0.93 ~ 1.80)
Diabetes without cc, n (%)	−0.14	0.18	−0.78	0.435	0.87 (0.61 ~ 1.23)
Diabetes with cc, n (%	0.23	0.16	1.44	0.150	1.26 (0.92 ~ 1.73)
Renal Disease, n (%)	0.53	0.15	3.51	< 0.001	1.69 (1.26 ~ 2.27)
Malignant Cancer, n (%)	0.82	0.21	3.84	< 0.001	2.26 (1.49 ~ 3.43)
Hypertension, n (%)	−0.06	0.16	−0.35	0.725	0.95 (0.69 ~ 1.29)
Hyperlipidemia, n (%)	−0.09	0.15	−0.57	0.567	0.92 (0.68 ~ 1.23)
Risk factors for T2MI
Sepsis	−0.03	0.20	−0.16	0.873	0.97 (0.65 ~ 1.44)	−0.53	0.36	−1.45	0.147	0.59 (0.29 ~ 1.20)
ARDS	−0.71	0.97	−0.73	0.465	0.49 (0.07 ~ 3.30)
AKI	0.28	0.16	1.81	0.070	1.33 (0.98 ~ 1.80)
Septic Shock	0.11	0.23	0.50	0.619	1.12 (0.71 ~ 1.76)	0.74	0.41	1.81	0.071	2.11 (0.94 ~ 4.73)
Tachycardia	0.25	0.21	1.24	0.217	1.29 (0.86 ~ 1.93)	0.46	0.22	2.11	0.035	1.58 (1.03 ~ 2.41)
Bradycardia	0.27	0.37	0.73	0.463	1.32 (0.63 ~ 2.74)
ARF hypoxia	0.18	0.17	1.06	0.290	1.19 (0.86 ~ 1.66)
Hypotension	−0.34	0.28	−1.21	0.227	0.71 (0.41 ~ 1.24)
Acute posthemorrhagic anemia	−0.14	0.21	−0.65	0.513	0.87 (0.58 ~ 1.32)
Medication status
Vasoactive agent	0.10	0.15	0.68	0.498	1.11 (0.82 ~ 1.49)
ACEI	−0.60	0.16	−3.77	< 0.001	0.55 (0.40 ~ 0.75)	−0.54	0.16	−3.43	< 0.001	0.59 (0.43 ~ 0.79)
Anticoagulant	0.22	0.37	0.59	0.558	1.25 (0.60 ~ 2.60)
Antiplatelet	0.02	0.20	0.09	0.928	1.02 (0.69 ~ 1.51)
Statin drugs	0.02	0.20	0.09	0.925	1.02 (0.69 ~ 1.49)
Diuretic drugs	0.12	0.17	0.74	0.457	1.13 (0.82 ~ 1.58)
ARBs	−0.91	0.41	−2.24	0.025	0.40 (0.18 ~ 0.89)	−0.78	0.42	−1.84	0.066	0.46 (0.20 ~ 1.05)
Digitalis drugs	0.65	0.33	1.99	0.046	1.92 (1.01 ~ 3.66)	0.88	0.30	2.90	0.004	2.42 (1.33 ~ 4.40)
Antibiotic	0.18	0.16	1.15	0.251	1.20 (0.88 ~ 1.64)
Beta receptor blockers	−0.15	0.19	−0.79	0.430	0.86 (0.60 ~ 1.25)
Special treatment
CRRT	−0.52	0.68	−0.77	0.442	0.59 (0.16 ~ 2.25)
MV	−0.13	0.16	−0.84	0.400	0.88 (0.64 ~ 1.19)
Length of stay	0.01	0.00	1.94	0.053	1.01 (1.00 ~ 1.02)	0.01	0.01	2.67	0.008	1.01 (1.01 ~ 1.03)

The multivariate model included the following variables: age, CCI, atrial fibrillation, congestive heart failure, chronic pulmonary disease, renal disease, malignant cancer, AKI, use of ACEI, ARBs, and digitalis drugs, as well as length of stay

Subgroup analysis

The use of ACEI was associated with significantly lower 3-year all-cause mortality in patients with T2MI. This mortality association was significant across several key patient subgroups, including: older patients aged ≥ 80 years (HR, 0.59; 95% CI, 0.39–0.90; P = 0.015) and those aged 65–80 years (HR, 0.43; 95% CI, 0.25–0.74; P = 0.002); patients with a high comorbidity burden, specifically those with a CCI > 8 (HR, 0.54; 95% CI, 0.35–0.83; P = 0.005) or a CCI of 7–8 (HR, 0.40; 95% CI, 0.21–0.75; P = 0.004); patients without tachycardia (HR, 0.55; 95% CI, 0.40–0.77; P < 0.001); and patients not treated with digitalis (HR, 0.58; 95% CI, 0.42–0.79; P < 0.001). This association remained consistent across these prespecified subgroups, with no significant interaction observed for age (P-interaction = 0.763), presence of tachycardia (P-interaction = 0.947), CCI score (P-interaction = 0.339), or use of digitalis (P-interaction = 0.242), as detailed in Fig. 3.

Fig. 3.

Subgroup analysis of 3-year all-cause mortality rates in the matched cohort

Collectively, these observational findings suggest that in-hospital initiation of ACEI therapy may be linked to improved long-term survival across a broad spectrum of patients with T2MI, including older individuals and those with significant comorbidities.

Sensitivity analysis

Cox regression analysis across multiple models demonstrated that in-hospital ACEI use was associated with significantly lower 3-year all-cause mortality. This association remained consistent across progressively adjusted models: crude (HR 0.55, 95% CI 0.41–0.74; P < 0.001; Model 1), adjusted for age and sex (HR 0.54, 95% CI 0.40–0.74; P < 0.001; Model 2), further adjusted for medical history (HR 0.57, 95% CI 0.42–0.78; P < 0.001; Model 3), and additionally adjusted for concomitant medications (HR 0.58, 95% CI 0.42–0.80; P < 0.001; Model 4) (Table 3).

Table 3.

Analysis of multiple Cox regression models on the association between ACEI use and 3-year all-cause mortality

Variables	Model1		Model2		Model3		Model4
	HR (95%CI)	P	HR (95%CI)	P	HR (95%CI)	P	HR (95%CI)	P
ACEI
No	1.00 (Reference)		1.00 (Reference)		1.00 (Reference)		1.00 (Reference)
Yes	0.55 (0.41 ~ 0.74)	< 0.001	0.54 (0.40 ~ 0.74)	< 0.001	0.57 (0.42 ~ 0.78)	< 0.001	0.58 (0.42 ~ 0.80)	< 0.001

Model 1: the crude (unadjusted) model

Model 2: adjusted for gender and age

Model 3: adjusted for gender, age, and comorbidities, including Old MI, atrial fibrillation, congestive heart failure, cerebrovascular disease, chronic pulmonary disease, diabetes without complications, diabetes with complications, renal disease, malignant cancer, hypertension, hyperlipidemia, and CCI

Model 4: further adjusted for medication use, such as anticoagulants, diuretics, ARBs, digitalis, antibiotics, beta-blockers, antiplatelets, statins, and vasoactive agents, in addition to all covariates included in Model 3

Sensitivity analysis for immortal time bias

Supplementary Table 3 summarizes the results from multiple analytical approaches used to evaluate the association between in-hospital ACEI use and mortality. Consistent findings were observed across the 7-day landmark analysis (both crude and adjusted), the IPTW-weighted analysis, and the analysis stratified by length of stay. All approaches indicated that ACEI use was significantly associated with lower 3-year and 180‑day mortality, with hazard ratios ranging from 0.55 to 0.68 (all P < 0.05).

After applying IPTW, covariate balance between ACEI users and non-users was excellent, as demonstrated by all standardized mean differences (SMDs) being < 0.01 and all Kolmogorov–Smirnov statistics < 0.07. The interaction P value for the stratified analysis was 0.62, suggesting no significant effect modification by length of stay.

In line with these results, Supplementary Fig. 3 displays the hazard ratios (with 95% confidence intervals) for the association between in-hospital ACEI use and 3-year mortality across the different methods. All estimates point toward a favorable association, with hazard ratios consistently ranging from 0.55 to 0.68. Supplementary Fig. 4 illustrates the standardized mean differences for key covariates before (red circles) and after (blue triangles) inverse probability of treatment weighting. The conventional balance threshold of |SMD|= 0.1 is shown as a dashed black line. The figure confirms the effectiveness of the weighting procedure, as all post-weighting SMD values fell well below this threshold, indicating adequate adjustment for measured confounders.

Doubly robust analysis results

The doubly robust analyses demonstrated remarkable consistency in their findings, as shown in Fig. 4A-D. In the matched cohort (n = 590), all analytical approaches indicated a significant association between ACEI use and lower 3-year mortality. Specifically, the HR were 0.55 (95% CI 0.41–0.74, P < 0.001) for the primary PSM analysis, 0.66 (95% CI 0.45–0.97, P = 0.037) for the IPTW-weighted analysis, 0.67 (95% CI 0.46–0.98, P = 0.042) for the multivariable-adjusted analysis, and 0.65 (95% CI 0.44–0.96, P = 0.033) for the augmented IPTW estimator. Excellent covariate balance was achieved following IPTW weighting, with a maximum |SMD| of 0.003.

Fig. 4.

Doubly robust analysis. A Effect Size Consistency: HR and 95% CI across four analytical methods in the propensity-score matched cohort (n = 590). The dashed red line indicates the null effect (HR = 1.0). All methods show protective associations with HRs < 1.0. B Estimate Precision: Widths of 95% confidence intervals for each method. Narrower intervals indicate greater precision. All robust methods show similar precision. C Covariate Balance Quality: Balance scores calculated as 1 divided by the maximum absolute standardized mean difference (1/Max |SMD|). Higher values indicate better balance between ACEI users and non-users after adjustment. D Statistical Significance: -log₁₀ transformation of P-values. The dashed red line corresponds to P = 0.05. All methods show statistical significance (transformed values > 1.3)

The narrow range of hazard ratios (0.55–0.67) across the four distinct methods indicates that the observed association is robust to different statistical adjustment strategies, thereby satisfying the doubly robust criterion. This consistency strengthens the credibility of the finding, although it is important to frame it as an observed association rather than a causal effect, given the retrospective observational design of the study. A comprehensive comparison of the analytical methods across dimensions including hazard ratio, precision, balance, statistical significance, and effective sample size is provided in Supplementary Fig. 5.

Secondary outcome

180-day mortality

Figure 5 displays the Kaplan–Meier curve for 180-day all-cause mortality stratified by ACEI use. The analysis indicates that ACEI use was significantly associated with reduced mortality (HR, 0.43; 95% CI, 0.29–0.64; P < 0.001). Univariate Cox regression yielded consistent results, with ACEI use linked to a lower risk of 180-day all-cause mortality (HR, 0.43; 95% CI, 0.29–0.65; P < 0.001). Other factors, including age, CCI, atrial fibrillation, chronic lung disease, malignant tumor, ARDS, use of digitalis, and CRRT, also showed statistical significance. After adjustment for these covariates in multivariate Cox analysis, ACEI use remained independently associated with lower 180-day all-cause mortality (HR, 0.49; 95% CI, 0.33–0.72; P < 0.001; Supplementary Table 4).

Fig. 5.

Kaplan–Meier curves for 180-day all-cause mortality matched by ACEI use

Sensitivity analyses using a multi-model Cox regression approach further supported these findings. In the unadjusted model (Model 1, Supplementary Table 5), the ACEI group had significantly lower crude 180-day mortality (HR, 0.43; 95% CI, 0.29–0.64; P < 0.001). After adjustment for age and sex (Model 2), in-hospital ACEI use remained significantly associated with reduced mortality (HR, 0.44; 95% CI, 0.30–0.65; P < 0.001). Further adjustments for prior medical history (Model 3) and concomitant medications (Model 4) continued to show significant mortality reductions (HR, 0.47 and 0.49, respectively; both P < 0.001).

3-year all-cause readmission rate and T2MI recurrence rate

The Kaplan–Meier curve for the 3-year all-cause readmission rate was plotted according to ACEI use. As demonstrated in Fig. 6(a), ACEI use was not associated with the risk of 3-year all-cause readmission (HR, 1.003; 95% CI, 0.776–1.295; P = 0.984). Univariate Cox proportional hazards analysis also indicated no significant effect of ACEI use on 3-year all-cause readmission risk (HR, 0.89; 95% CI, 0.65–1.20; P = 0.441). In contrast, factors including congestive heart failure, diabetes mellitus, kidney disease, acute kidney injury, and digitalis use were significantly associated with readmission risk. After adjustment for relevant confounders in a multivariate Cox model, ACEI use remained unrelated to 3-year all-cause readmission risk (HR, 1.00; 95% CI, 1.00–1.00; P = 0.664) (Supplementary Table 6).

Fig. 6.

Kaplan–Meier curves for readmission and recurrence rates within 3 years of T2MI. a: All-cause readmission rate; b: T2MI recurrence rate

Similarly, the Kaplan–Meier curve for 3-year T2MI recurrence (Fig. 6(b)) revealed no significant effect of ACEI use on recurrence risk (HR, 1.14; 95% CI, 0.64–2.05; P = 0.654). Univariate Cox analysis corroborated this finding (HR, 1.14; 95% CI, 0.64–2.04; P = 0.652). By comparison, congestive heart failure, diabetes mellitus, kidney disease, acute kidney injury, and digitalis use showed statistically significant associations with T2MI recurrence. Multivariate Cox analysis, adjusted for relevant factors, further confirmed that ACEI use did not affect 3-year T2MI recurrence risk (HR, 1.10; 95% CI, 0.59–2.04; P = 0.762) (Supplementary Table 7).

Discussion

This is the first study to investigate the relationship between ACEI use and prognosis in T2MI patients discharged alive. We found that in-hospital ACEI use was significantly associated with lower short-term (180-day) and long-term (3-year) all-cause mortality, but was not associated with post-discharge readmission or T2MI recurrence. These findings provide real-world evidence supporting the use of ACEI in T2MI patients and may help inform future treatment strategies.

ACEIs dilate blood vessels, reduce cardiac load, and delay myocardial remodeling, primarily by inhibiting the RAS, reducing Ang II production, and promoting bradykinin accumulation [15–17]. The cardioprotective role of ACEIs in T1MI is well established. Previous studies have reported improved 5-year survival in AMI patients treated with ACEIs compared to those not receiving such therapy [15]. ACEI treatment has also been associated with reduced all-cause mortality in AMI patients undergoing percutaneous coronary intervention [22]. Moreover, after 2 years of follow-up, patients treated with ACEIs showed lower mortality than those receiving ARBs [15]. In addition, during hospitalization for AMI, patients who did not adhere to ACEI therapy had significantly higher mortality than those who did, irrespective of statin or beta-blocker use [17]. Together, these observations suggest that ACEIs offer substantial benefit in AMI patients. However, given the distinct pathophysiology of T2MI compared with T1MI, as well as the frequent presence of comorbidities in T2MI patients, whether ACEI treatment confers similar benefits in T2MI remains uncertain.

A recent study indicated that in-hospital ACEI use may improve both in-hospital and 2-year survival in T2MI patients [19]. Our study further supports this association. Specifically, we observed that among T2MI patients discharged alive, in-hospital ACEI use was significantly associated with lower post-discharge all-cause mortality over both short- and long-term follow-up periods. This association may be explained by several potential mechanisms. First, ACEIs may reduce infarct size and attenuate the inflammatory response. Following myocardial infarction, activation of the renin–angiotensin–aldosterone system (RAAS) occurs, with Ang II as a key mediator [23]. Ang II levels rise within hours and can remain elevated for days [24], leading to coronary vasoconstriction—reducing oxygen supply while increasing demand—and promoting release of inflammatory factors that exacerbate ischemic injury [25, 26]. Although T2MI is not typically caused by acute plaque rupture, the resultant myocardial ischemia and necrosis still trigger local and systemic inflammatory responses and neurohormonal activation, including RAAS activation. Excessive RAAS activation can worsen myocardial oxygen demand by increasing afterload and heart rate, thereby promoting adverse ventricular remodeling [12]. In this setting, ACEIs may attenuate this detrimental neurohormonal cascade, mitigate maladaptive remodeling, and improve coronary perfusion, potentially disrupting the cycle of ongoing injury following the initial supply–demand mismatch [1, 2, 27, 28]. Infarct size has also been positively correlated with cardiac ACE activity [29]. Thus, by inhibiting Ang II production [30] and modulating inflammation [27], ACEIs may help limit infarct expansion and support early cardiac recovery. Second, ACEIs may help preserve cardiac structure and prevent adverse remodeling. Studies have shown that in patients with first anterior wall MI, administering captopril within 24 h of symptom onset for 14 days significantly reduced the incidence of ventricular dilatation [31] and attenuated early left ventricular remodeling [32]. Furthermore, reduced baroreflex sensitivity is linked to ventricular tachycardia, increases the risk of hemodynamic compromise during tachycardia [33], and is associated with higher cardiac mortality [34]. Treatment with captopril initiated within 4 days after MI has been shown to improve baroreflex sensitivity [35], which may lower the risk of poor outcomes.

However, this study found that ACEI use was not associated with a reduction in either all-cause readmission or T2MI recurrence. This finding diverges from prior research on ACEI in myocardial infarction. One potential explanation is that non-coronary events—such as anemia, hypoxemia, hypotension, and other major risk factors for T2MI [2, 36]—contribute substantially to readmission risk [37]. While ACEI primarily target the cardiovascular system, they may not prevent readmissions or recurrence driven by non-coronary etiologies [28]. Thus, ACEI is unlikely to significantly mitigate readmission or recurrence risks arising from non-cardiovascular causes. The prognostic benefits associated with ACEI in T2MI patients may not reflect a T2MI-specific effect, but rather an overall improvement in cardiovascular risk profile. As a first-line treatment for hypertension, heart failure, and coronary heart disease [18], ACEI can lower long-term cardiovascular event risk in T2MI patients, thereby potentially reducing all-cause mortality, without directly altering the pathophysiological process of T2MI. This interpretation aligns with the present results—that ACEI was not associated with T2MI recurrence—and suggests future research explore combined strategies of ACEI and control of T2MI triggers.

In addition, existing evidence indicates that ACEI-based regimens are associated with greater cardiocerebrovascular protection in elderly patients compared with younger individuals, including significant reductions in the risks of stroke, death, and heart failure [38]. Subgroup analyses further support that older patients may derive more benefit from ACEI. For T2MI patients aged ≥ 65 years, ACEI may offer enhanced benefit when combined with active management of underlying conditions, such as anti-infection therapy or anemia correction. Moreover, ACEI use was associated with benefit in patients without tachyarrhythmia, with a CCI score ≥ 7, and not taking digitalis. Thus, in clinical practice, ACEI may be prioritized for this patient subgroup.

The robustness of the primary findings was further assessed through doubly robust analyses. Four distinct methods—propensity score matching, inverse probability weighting, regression adjustment, and their combination—yielded highly consistent hazard ratio estimates (range: 0.55–0.67). This consistency across analytical approaches suggests that the observed association is not an artifact of any single method and remains stable under different confounding adjustment strategies. Notably, excellent covariate balance was achieved after inverse probability weighting, indicating effective control of measured confounders. Nevertheless, as with all observational studies, the possibility of residual confounding from unmeasured factors cannot be excluded.

This study is subject to several limitations. First, it is a retrospective analysis based on a single-center database (MIMIC-IV) with a relatively small sample size, necessitating further validation through multi-center prospective studies. Second, exposure was broadly defined as any ACEI use during the ICU stay, without data on dosage, treatment duration, or post-discharge continuation. This non-differential misclassification likely biases the results toward the null, suggesting that the true association might be stronger if sustained therapy is required for benefit. Thus, the observed association between ACEI use and mortality should be interpreted as a conservative estimate. Additionally, the absence of post-discharge medication follow-up data in MIMIC-IV precluded evaluation of ACEI continuation on long-term prognosis, representing an inherent database limitation and a key shortcoming. Third, some hospitalized patients did not receive ACEI due to contraindications or adverse reactions, potentially introducing selection bias. Fourth, by restricting the analysis to patients who survived to discharge, we may have selected a cohort with lower overall mortality risk, which could lead to an overestimation of the association between ACEI use and post-discharge outcomes. We emphasize that our study aimed to assess the association between in-hospital ACEI use and outcomes among survivors, and the findings should not be extrapolated to all hospitalized patients, including those who died during admission. Fifth, residual confounding related to clinical stability is a concern. Patients who received ACEI likely had more stable hemodynamics and better renal function at baseline, indicating lower baseline risk. Despite using PSM, we lacked granular data on in-hospital hemodynamic trends, lactate levels, or SOFA scores. Consequently, ACEI use may serve more as a marker of clinical stability than as a direct factor linked to improved survival. These findings should be regarded as hypothesis-generating, and causal inferences require further validation through randomized controlled trials. Specifically, ACEI use represents both an intervention and a stability indicator, as initiation typically requires stable blood pressure and renal function; thus, residual confounding from unmeasured severity metrics, such as vasopressor doses or dynamic parameters, may persist. Finally, as with any observational study, unmeasured confounding factors (e.g., socioeconomic status or genetic predispositions) may influence the results, despite adjustment for numerous variables. Future research incorporating more granular data or alternative methods, such as instrumental variable analysis, would be valuable to confirm the observed associations.

Conclusion

This study found that ACEI use during the ICU stay was associated with significantly lower short- and long-term all-cause mortality among T2MI survivors, but not with readmission or recurrence rates. This finding expands the clinical applications of ACEI in populations with myocardial infarction caused by different mechanisms. Nevertheless, due to study design limitations, clinicians must carefully evaluate the indications and contraindications for each patient. Future prospective studies should further clarify the optimal dosage, timing, and patient selection for ACEI administration.

Abbreviation

ACEI

Angiotensin converting enzyme inhibitor

T2MI

Type 2 myocardial infarction

MIMIC-IV

Medical Information Mart for Intensive Care IV

PSM

Propensity score matching

SGLT2

Sodium-dependent glucose transporters 2

GLP-1

Glucagon-like peptide-1

T1MI

Type 1 myocardial infarction

RAS

Renin-angiotensin system

LVEF

Left ventricular ejection fraction

MIMIC-IV 3.1

Medical Information Mart for Intensive Care IV version 3.1

ICD

International classification of diseases

SQL

Structured Query Language

CCI

Charlson Comorbidity Index

ARDS

Acute respiratory distress syndrome

ARBs

Angiotensin II receptor blockers

CRRT

Continuous renal replacement therapy

HR

Hazard ratio

CI

Confidence interval

SMD

Standardized mean difference

VIF

Variance inflation factor

Ang II

Angiotensin II

AMI

Acute myocardial infarction

RAAS

Renin–angiotensin–aldosterone system

ACE

Angiotensin-converting enzyme

MV

Mechanical ventilation

Authors’ contributions

ZJY, LJN and FWY: carried out the studies, participated in collecting data, and drafted the manuscript. LY, WYX and MXY: participated in collecting data and helped to draft the manuscript. LYY and LWL: performed the statistical analysis. WF and JXF: design, review and editing the manuscript. All authors read and approved the final manuscript.

Funding

The study was supported by Shanghai Municipal Health Commission Health Industry Clinical Research Project (202340100), Jiading District Central Hospital Clinical Research Support Program (2025LY14 and 2025LY18). The funding sources had no role in the design of the study; in the collection, analysis, and interpretation of data; or in the writing of the manuscript.

Data availability

The data from public databases were available on the MIMIC-IV website at https://mimic.physionet.org/.

Declarations

Ethics approval and consent to participate

The study protocol was in accordance with the Declaration of Helsinki and was reviewed and approved by the Ethics Committee of Shanghai Jiading District Central Hospital (2025–34), an affiliated teaching hospital of Shanghai University of Medicine & Health Sciences. In accordance with national laws and institutional requirements, written informed consent was not required of the participants because of the nature of the study.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.