Investigating the impact of ramelteon on all-cause mortality in ischemic stroke ICU patients: a retrospective propensity-matched study from the MIMIC-IV database

Abstract

The study’s objective was to assess the connection between ramelteon usage and mortality from any cause in ischemic stroke patients in the intensive care unit (ICU). Utilizing the MIMIC-IV database, we analyzed a cohort of adult patients who had been diagnosed with ischemic stroke. During their time in the hospital, patients were sorted into ramelteon and non-ramelteon groups according to drug exposure. To achieve balance in baseline covariates, propensity score matching (PSM) was utilized. The primary focus was on mortality from all causes over 28 days, with secondary focuses on all-cause mortality over 90 and 365 days. Using Cox proportional hazards models, hazard ratios (HR) and 95% confidence intervals (CI) were estimated, taking into account potential confounding factors. Subgroup analyses were done to assess effect modification across clinical strata. The study encompassed 3413 patients, with 535 pairs matched after PSM. Ramelteon use was significantly associated with decreased 28-day all-cause mortality rates both before and after PSM, with adjusted HR of 0.34 and 0.23, both with P < 0.001. The fully adjusted post-PSM models showed similar protective associations for 90-day (HR = 0.43) and 365-day (HR = 0.55) all-cause mortality, both with P < 0.001. Prolonged use of ramelteon for more than 14 days and cumulative doses exceeding 300 mg were consistently linked to lower all-cause mortality at all time intervals. Through subgroup and sensitivity analyses, the associations were confirmed to be consistent across different clinical strata. In critically ill patients with ischemic stroke, ramelteon usage was independently associated with lower short- and long-term all-cause mortality.

Graphical Abstract

Supplementary Information

The online version contains supplementary material available at 10.1007/s00210-025-04653-3.

Introduction

Across the globe, ischemic stroke is a primary cause of death and ongoing disability, where intensive care unit (ICU) care is vital for influencing patient outcomes (Saini et al. 2021; Valburg et al. 2020). In spite of progress in acute interventions like thrombolysis and thrombectomy, secondary complications, including delirium, sleep issues, and systemic inflammation, significantly affect prognosis negatively (Powers et al. 2019). Sleep disruption is notably common in ICU settings and has been linked to greater mortality, extended periods of hospitalization, and cognitive decline (Wilcox et al. 2021; Chang et al. 2020; Marchasson et al. 2024). Ramelteon, acting selectively on melatonin receptors, has surfaced as a potential therapy because it can manage circadian rhythms and ease neuroinflammation (Miyamoto 2009; Sateia et al. 2017). Yet, the influence on significant clinical endpoints, such as overall mortality among ischemic stroke patients in the ICU, has not been extensively studied.

The role of ramelteon in critical care has been the subject of recent studies, which have produced mixed results. Previous studies have observed that septic ICU patients receiving ramelteon had reduced delirium and better survival, which suggests that the drug may have broader neuroprotective effects (Yu et al. 2023; Han et al. 2024). In contrast, the retrospective analysis by Kinouchi M et al. revealed no considerable difference in the occurrence of postoperative delirium between ramelteon and placebo in elderly patients who experienced general anesthesia (Kinouchi et al. 2023). These inconsistencies point to the requirement for disease-specific assessments, particularly concerning stroke, where disruption of the sleep–wake cycle intensifies secondary brain damage (Hurtado-Alvarado et al. 2016; Zamore and Veasey 2022; Logan and McClung 2019). Preclinical findings suggest that ramelteon has the potential to reduce ischemic damage by curbing oxidative stress and preserving the blood–brain barrier, yet its clinical application remains limited (Sun et al. 2022; Aslankoc et al. 2022). Considering that up to 47% of stroke ICU patients experience sleep disturbances (Duss et al. 2023; Baylan et al. 2020), it is important to specifically examine the impact of ramelteon on mortality.

The study investigates the association between ramelteon administration and all-cause mortality in ischemic stroke ICU patients by leveraging the Medical Information Mart for Intensive Care IV (MIMIC-IV) database and applying propensity score matching. By taking into account confounding elements like age, comorbidities, and the severity of illness, we aim to determine if ramelteon offers survival benefits to this high-risk group. Our study’s findings could guide clinical practice by either promoting the integration of ramelteon into stroke ICU protocols or underscoring the need for alternative sleep-modulating treatments. In light of the heightened attention on post-stroke neuroprotection, this research offers insights into the broader dialogue on supplementary treatments in critical care.

Methods and materials

Origin of data and group of study participants

This study, designed as a retrospective cohort, utilized patient data extracted from the MIMIC-IV (version 3.0) database for analysis. Curated by the Beth Israel Deaconess Medical Center and MIT, the MIMIC-IV database is publicly available. Anonymized clinical data from over 70,000 critically ill patients treated at BIDMC between 2008 and 2019 is included in the dataset. The documentation covered all test outcomes, medication plans, vital signs, hospital stay durations, and other particular information for each patient. Upon completing the authorization protocol, a credentialed researcher from our team (certification: 13,230,663) was given access to the MIMIC-IV database. The requirement for ethical approval and informed consent was waived because the study was retrospective and involved de-identified data. Ischemic stroke identification was based on the diagnostic codes from the 9th and 10th Revisions of the International Classification of Diseases. The criteria for exclusion were being under 18, not in their first intensive care unit admission (ICU, n = 577), or having an ICU stay of less than 24 h (n = 374). In addition, those who received melatonin analogues different from ramelteon were left out (n = 1). After the criteria were applied, 3413 patients were deemed eligible for the final analysis. In the ICU, patients were sorted into two groups according to their exposure to ramelteon: 543 were exposed, and 2870 were not. 1:1 propensity score matching (PSM) was implemented to equalize covariates and lessen confounding between the groups. The process produced 535 matched pairs for analysis and comparison (Fig. 1). The requirement for ethical approval and informed consent was waived because the study was retrospective and involved de-identified data.

Fig. 1.

Flowchart of sample selection. Abbreviation: MIMIC-IV, Medical Information Mart for Intensive Care IV; ICU, intensive care unit; PSM, propensity score matching

Data collection

Structured Query Language (SQL) was employed to extract data using PostgresSQL and Navicate Premium. Demographic data, vital signs, clinical interventions, comorbidities, lab data, and treatment results were the six core elements of the extraction process. The study began with the selection of variables that were clinically relevant and had the potential to influence the relationship between ramelteon use and mortality outcomes in patients with ischemic stroke. Table 1 displays an extensive list of the variables that have been extracted. The follow-up period initiated on the admission date and terminated on the death date. The first 24 h after ICU admission were used to extract all laboratory variables and disease severity scores.

Table 1.

Characteristics of patients at baseline before PSM

Characteristics	Total (n = 3413)	Non-ramelteon (n = 2870)	Ramelteon (n = 543)	P-value	SMD
Age, (years)	70.99 (60.03, 81.25)	71.04 (60.14, 81.46)	70.96 (59.41, 80.43)	0.541	− 0.022
Gender, n (%)				0.003
Female	1670 (48.93)	1436 (50.03)	234 (43.09)		− 0.140
Male	1743 (51.07)	1434 (49.97)	309 (56.91)		0.140
Race group, n (%)				0.392
Asian/Hispanic or Latino/other	1055 (30.91)	884 (30.80)	171 (31.49)		0.015
Black	385 (11.28)	333 (11.60)	52 (9.58)		− 0.069
White	1973 (57.81)	1653 (57.60)	320 (58.93)		0.027
Weight, (kg)	77.00 (64.85, 91.70)	76.70 (64.50, 91.50)	79.10 (66.05, 92.10)	0.076	0.075
HR, (beats/min)	99.00 (87.00, 114.00)	98.00 (86.00, 114.00)	100.00 (88.00, 116.00)	0.045	0.079
SBP, (mmHg)	160.00 (145.00, 177.00)	161.00 (146.00, 178.00)	156.00 (140.00, 171.00)	< 0.001	− 0.248
DBP, (mmHg)	97.00 (84.00, 111.00)	97.00 (85.00, 111.00)	97.00 (81.00, 112.00)	0.286	− 0.044
MBP, (mmHg)	113.00 (100.00, 127.00)	113.00 (101.00, 127.00)	112.00 (98.00, 126.50)	0.061	− 0.102
RR, (breaths/min)	27.00 (24.00, 31.00)	27.00 (24.00, 31.00)	28.00 (24.00, 32.00)	0.007	0.104
T, (℃)	37.28 (37.00, 37.67)	37.28 (37.00, 37.67)	37.22 (37.00, 37.67)	0.393	− 0.006
SPO2 (%)	100.00 (99.00, 100.00)	100.00 (99.00, 100.00)	100.00 (99.00, 100.00)	0.891	− 0.007
SOFA	3.00 (2.00, 5.00)	3.00 (2.00, 5.00)	3.00 (2.00, 6.00)	0.010	0.124
GCS	14.00 (11.00, 15.00)	14.00 (11.00, 15.00)	14.00 (13.00, 15.00)	0.008	0.164
CCI	7.00 (5.00, 8.00)	7.00 (5.00, 8.00)	7.00 (5.00, 9.00)	0.002	0.148
SAPSII	33.00 (26.00, 42.00)	33.00 (26.00, 42.00)	35.00 (26.00, 44.00)	0.055	0.088
OASIS	31.00 (26.00, 37.00)	31.00 (26.00, 37.00)	31.00 (26.00, 38.00)	0.580	0.036
PT (s)	13.40 (12.10, 15.70)	13.31 (12.10, 15.60)	13.40 (12.30, 15.90)	0.186	− 0.042
PTT (s)	30.70 (27.30, 41.20)	30.70 (27.30, 41.27)	30.50 (27.54, 40.68)	0.946	− 0.027
CKD, n (%)				< 0.001
No	2748 (80.52)	2339 (81.50)	409 (75.32)		− 0.143
Yes	665 (19.48)	531 (18.50)	134 (24.68)		0.143
Diabetes mellitus, n (%)				0.050
No	2274 (66.63)	1932 (67.32)	342 (62.98)		− 0.090
Yes	1139 (33.37)	938 (32.68)	201 (37.02)		0.090
Hypertension, n (%)				< 0.001
No	1711 (50.13)	1399 (48.75)	312 (57.46)		0.176
Yes	1702 (49.87)	1471 (51.25)	231 (42.54)		− 0.176
Atrial fibrillation, n (%)				0.096
No	2058 (60.3)	1748 (60.91)	310 (57.09)		− 0.077
Yes	1355 (39.7)	1122 (39.09)	233 (42.91)		0.077
Heart failure, n (%)				< 0.001
No	2540 (74.42)	2175 (75.78)	365 (67.22)		− 0.182
Yes	873 (25.58)	695 (24.22)	178 (32.78)		0.182
Malignancy, n (%)				0.095
No	3135 (91.85)	2646 (92.20)	489 (90.06)		− 0.072
Yes	278 (8.15)	224 (7.80)	54 (9.94)		0.072
COPD, n (%)				0.530
No	3054 (89.48)	2564 (89.34)	490 (90.24)		0.030
Yes	359 (10.52)	306 (10.66)	53 (9.76)		− 0.030
Moderate to severe liver disease, n (%)				0.348
No	3354 (98.27)	2823 (98.36)	531 (97.79)		− 0.039
Yes	59 (1.73)	47 (1.64)	12 (2.21)		0.039
Mannitol, n (%)				< 0.001
No	3192 (93.52)	2662 (92.75)	530 (97.61)		0.317
Yes	221 (6.48)	208 (7.25)	13 (2.39)		− 0.317
Thrombolysis, n (%)				< 0.001
No	2937 (86.05)	2513 (87.56)	424 (78.08)		− 0.229
Yes	476 (13.95)	357 (12.44)	119 (21.92)		0.229
Antiplatelet agents, n (%)				0.167
No	795 (23.29)	681 (23.73)	114 (20.99)		− 0.067
Yes	2618 (76.71)	2189 (76.27)	429 (79.01)		0.067
Anticoagulant drugs, n (%)				< 0.001
No	224 (6.56)	209 (7.28)	15 (2.76)		− 0.276
Yes	3189 (93.44)	2661 (92.72)	528 (97.24)		0.276
CRRT, n (%)				< 0.001
No	3275 (95.96)	2770 (96.52)	505 (93.00)		− 0.138
Yes	138 (4.04)	100 (3.48)	38 (7.00)		0.138
Loop diuretic, n (%)				< 0.001
No	1950 (57.13)	1705 (59.41)	245 (45.12)		− 0.287
Yes	1463 (42.87)	1165 (40.59)	298 (54.88)		0.287
Vasoactive agent, n (%)				0.017
No	2752 (80.63)	2294 (79.93)	458 (84.35)		0.122
Yes	661 (19.37)	576 (20.07)	85 (15.65)		− 0.122
Vasopressin input, n (%)				0.854
No	3321 (97.3)	2792 (97.28)	529 (97.42)		0.009
Yes	92 (2.7)	78 (2.72)	14 (2.58)		− 0.009
Mechanical ventilation, n (%)				0.141
No	2095 (61.38)	1777 (61.92)	318 (58.56)		− 0.068
Yes	1318 (38.62)	1093 (38.08)	225 (41.44)		0.068

SOFA, sequential organ failure assessment; GCS, Glasgow coma scale; CCI, Charlson comorbidity index; SAPSII, simplified acute physiology score II; OASIS, Oxford acute severity of illness score; PT, prothrombin time; PTT, partial thromboplastin time; CKD, chronic kidney disease; HR, heart rate; SBP, systolic blood pressure; DBP, diastolic blood pressure; MBP, mean blood pressure; RR, respiratory rate; T, temperature; SPO2, oxygen saturation; CRRT, continuous renal replacement therapy; COPD, chronic obstructive pulmonary disease; PSM, propensity score matching

Clinical prescribing scenarios and definition of thiamine exposure

Ramelteon exposure was identified as its prescription and administration during the ICU stay, relying on documented clinical indications instead of routine admission prophylaxis. In our cohort, the ICU team prescribed ramelteon when patients exhibited clinically recognized sleep disturbances, including insomnia, circadian rhythm disruption, or significant sleep fragmentation. These symptoms were detected during routine clinical assessments and entered into the notes by doctors or nurses. Ramelteon was prescribed primarily to boost sleep quality and correct circadian rhythm alignment. The timing for starting treatment and the total dose varied based on what each patient required. The standard adult dose for Ramelteon was 8 mg, administered once daily at bedtime. For patients in the ICU who cannot take medications orally, the most common method of administration is through enteral routes such as nasogastric or orogastric tubes. To facilitate analysis, we collected detailed prescription records and determined the cumulative doses given during the ICU stay. The cumulative exposure was divided into three predefined groups: less than 100 mg, 100 to 300 mg, and greater than 300 mg.

Prior studies and clinical guidelines have indicated that ramelteon is generally well tolerated, with the most common side effects being mild somnolence, dizziness, and fatigue, and serious adverse events being rare (Miyamoto 2009; Sateia et al. 2017; Erman et al. 2006). Moreover, limited clinical trials in perioperative and hospitalized settings have also demonstrated good tolerability (Mayanagi et al. 2023; Hokuto et al. 2020; Tanifuji et al. 2022). No evidence suggests that the rate or seriousness of adverse events related to ramelteon is enough to affect survival outcomes meaningfully (Tanifuji et al. 2022). In the MIMIC-IV database, only these frequent mild adverse effects are mentioned, and they are unlikely to influence our mortality analyses or modify the study’s overall conclusions.

Exposure and endpoint events

In this study, the administration of ramelteon during the hospital stay was the only exposure factor examined. The study’s primary focus on all-cause mortality within 28 days of the first ICU admission, which pertained to deaths occurring during the hospital stay. The study’s secondary outcomes were all-cause mortality at 90 and 365 days, tracking deaths that happened after discharge during the follow-up period.

Statistical analysis

We performed a detailed assessment to analyze the extent and pattern of missing data for every variable. Supplementary Table 1 illustrates that less than 5% of the data for most baseline physiological variables, including weight, vital signs such as heart rate, respiratory rate, blood pressure, temperature, and oxygen saturation, and GCS, were missing. The missing data for PT and PTT was moderately high, at about 11–12%. We excluded variables with more than 20% missing data from our analyses based on our predefined criteria. In instances where 20% or fewer observations were missing, the MICE method was applied, assuming the data were missing at random (MAR). In accordance with current methodological recommendations, we developed imputed datasets to ensure that the number of imputations was at least equal to the proportion of incomplete cases. To ensure congeniality, the imputation model incorporated all covariates and outcome variables. The mean (SD) is used to represent normally distributed continuous data, while the median (IQR) is used for data that is not normally distributed. The representation of categorical variables was in the form of numbers and percentages (%). The t-test or Mann–Whitney U test was employed to compare continuous variables, and the Pearson chi-square test was used for categorical variables when assessing baseline characteristics between the ramelteon and non-ramelteon groups. By employing Schoenfeld residuals and deviance residual plots, the proportional hazards (PH) assumption of the Cox regression models was examined, and no notable violations were identified. Variance inflation factors (VIFs) were used to check for potential multicollinearity among covariates, with a VIF higher than 5 indicating significant multicollinearity. In the models, all variables had VIFs that did not exceed this threshold (Supplementary Table 2). To compare survival between ramelteon users and non-users, Kaplan–Meier survival curves for 28-, 90-, and 365-day outcomes were produced, and differences were evaluated using the log-rank test. To minimize potential baseline imbalances, we utilized 1:1 nearest-neighbor propensity score matching without replacement. The logit-transformed propensity scores were used for matching, with a caliper width defined as 0.2 times the standard deviation of the logit, consistent with Austin’s methodological recommendations showing that this threshold yields an optimal compromise between bias reduction and precision (Austin 2011a). We opted for a 1:1 matching ratio, as prior evidence suggests that this strategy, despite resulting in a smaller matched sample compared to higher ratios, typically offers superior covariate balance and diminishes residual confounding (Austin 2008, 2011b; Stuart 2010). Following the matching process, covariate balance was assessed using standardized mean differences (SMD), with an SMD of 0.1 or less deemed satisfactory (Austin 2011a, 2011b; Lunt 2014). To evaluate the association between the use of ramelteon and clinical outcomes, hazard ratios (HR) and 95% confidence intervals (CI) were estimated using Cox proportional hazards models. To investigate the association between the use of ramelteon and clinical outcomes, researchers formulated three Cox proportional hazards models. No adjustments were made to Model 1. Model 2 was adjusted for demographic, physiological, and comorbidity-related variables, including age, weight, heart rate, Sequential Organ Failure Assessment, respiratory rate, Glasgow Coma Scale, Charlson Comorbidity Index, Oxford Acute Severity of Illness Score, Simplified Acute Physiology Score II, prothrombin time, partial thromboplastin time, and comorbidities such as chronic kidney disease, hypertension, diabetes mellitus, atrial fibrillation, malignancy, heart failure, chronic obstructive pulmonary disease and moderate to severe liver disease. Model 3 included all covariates in Model 2 and was further adjusted for mechanical ventilation, continuous renal replacement therapy, mannitol use, antiplatelet agents, thrombolysis, anticoagulant drugs, vasoactive agents, vasopressin administration, and loop diuretics. Subgroup analyses were executed to determine the consistency of the primary outcome among various pre-specified groups of patients. To formally examine the heterogeneity of the effect across these subgroups, we included an interaction term between the subgroup variable and the main exposure in the Cox proportional hazards model. An interaction P-value of under 0.05 was regarded as evidence of a statistically significant difference in the effect between subgroups. To further investigate potential duration- and dose–response relationships, patients were organized based on how long they took ramelteon and the total dose they accumulated. Integer scores were given to each level and considered as a continuous variable in the Cox model to test for a linear trend across these ordered categories and control for multiple comparisons, resulting in a P-value for trend (P for trend). The robustness of the observed associations was assessed through a sensitivity analysis that excluded patients with documented sleep disorders while hospitalized in the ICU. Following the exclusion of these individuals, PSM was reapplied to ensure that baseline covariates remained balanced. The Cox proportional hazards models (Models 1–3) were re-analyzed to check if the association between ramelteon use and mortality was still present in this restricted cohort. The analysis employed the same model specifications and covariate adjustments as the main analysis to ensure consistency in methodology. The data processing and statistical analyses were performed using R (version 4.3.3, 2024–02–29) along with Zstats (version 1.0; www.zstats.net), a web-based statistical platform. Using R (version 4.3.3, 2024–02–29) and Zstats (version 1.0; www.zstats.net), a web-based statistical service, data processing and statistical analyses were executed, with P < 0.05 deemed significant for two-tailed tests.

Results

Baseline characteristics

Before applying PSM, the study included 3413 patients, divided into 543 in the ramelteon group and 2870 in the non-ramelteon group (Table 1). The two groups showed significant differences at the baseline. The group of patients treated with ramelteon had a higher percentage of males (56.91% compared to 49.97%, P = 0.003), slightly lower systolic blood pressure (156 mmHg versus 161 mmHg, P < 0.001), and increased respiratory rates (28 breaths per minute compared to 27, P = 0.007). They also exhibited higher SOFA (Sequential Organ Failure Assessment) and GCS (Glasgow Coma Scale) scores compared to the non-ramelteon group (P = 0.010 and P = 0.008, respectively). The prevalence of chronic kidney disease (24.68%vs.18.50%, P < 0.001), diabetes mellitus (37.02%vs.32.68%, P = 0.050), and heart failure (32.78%vs.24.22%, P < 0.001) was also higher among ramelteon users. Conversely, the occurrence of hypertension was lower in the ramelteon group (42.54% compared to 51.25%, P < 0.001). In terms of treatment characteristics, patients in the ramelteon group had a lower frequency of mannitol administration (2.39%vs.7.25%, P < 0.001), but were more likely to receive thrombolysis (21.92%vs.12.44%, P < 0.001), anticoagulant drugs (97.24%vs.92.72%, P < 0.001), and loop diuretics (54.88%vs.40.59%, P < 0.001). After performing PSM (Table 2), 1070 patients remained, with 535 in each group, and the baseline characteristics were well balanced, with all standardized mean differences (SMDs) being less than 0.1. In Supplementary Fig. 1, the overlap of propensity scores for the ramelteon and non-ramelteon groups is shown through kernel density plots, both before and after matching. The changes in the overlap of propensity score distributions before and after matching are shown in this figure, reflecting the quality of the matching process. The standardized mean differences (SMDs) for each baseline variable are depicted in Supplementary Fig. 2, assessing covariate balance before and after matching, with values under 0.1 denoting a satisfactory balance. The overall balance in distribution is analyzed in Supplementary Fig. 1, and Supplementary Fig. 2 highlights the balance for individual variables. Together, these figures demonstrate the effectiveness of the PSM procedure.

Table 2.

Characteristics of patients at baseline after PSM

Characteristics	Total (n = 1070)	Non-ramelteon (n = 535)	Ramelteon (n = 535)	P	SMD
Age, (years)	69.75 (59.05, 80.40)	69.00 (58.57, 80.21)	70.99 (59.48, 80.54)	0.183	0.086
Gender, n (%)				0.951
Female	469 (43.83)	235 (43.93)	234 (43.74)		− 0.004
Male	601 (56.17)	300 (56.07)	301 (56.26)		0.004
Race group, n (%)				0.727
Asian/Hispanic or Latino/other	331 (30.93)	164 (30.65)	167 (31.21)		0.012
Black	112 (10.47)	60 (11.21)	52 (9.72)		− 0.050
White	627 (58.6)	311 (58.13)	316 (59.07)		0.019
Weight, (kg)	78.00 (65.73, 92.45)	77.90 (65.12, 93.83)	78.80 (65.95, 91.80)	0.747	− 0.004
HR, (beats/min)	100.00 (88.00, 117.00)	100.00 (88.00, 117.50)	101.00 (88.00, 116.00)	0.608	− 0.063
SBP, (mmHg)	156.00 (141.25, 171.00)	155.00 (143.00, 172.00)	157.00 (140.00, 171.00)	0.976	− 0.004
DBP, (mmHg)	97.00 (82.00, 112.00)	96.00 (84.00, 110.50)	97.00 (81.00, 112.00)	0.853	− 0.009
MBP, (mmHg)	111.00 (98.00, 126.75)	110.00 (98.00, 126.00)	112.00 (98.00, 127.00)	0.873	− 0.028
RR, (breaths/min)	28.00 (24.00, 32.00)	28.00 (24.00, 32.00)	28.00 (24.00, 32.00)	0.820	− 0.032
T, (℃)	37.28 (37.00, 37.67)	37.28 (37.06, 37.72)	37.22 (37.00, 37.67)	0.108	− 0.057
SPO2 (%)	100.00 (99.00, 100.00)	100.00 (99.00, 100.00)	100.00 (99.00, 100.00)	0.696	− 0.002
SOFA	3.00 (2.00, 6.00)	3.00 (2.00, 6.00)	3.00 (2.00, 6.00)	0.560	0.024
GCS	14.00 (13.00, 15.00)	14.00 (13.00, 15.00)	14.00 (13.00, 15.00)	0.278	− 0.053
CCI	7.00 (5.00, 9.00)	7.00 (5.00, 9.00)	7.00 (5.00, 9.00)	0.150	0.093
SAPSII	35.00 (26.00, 43.00)	35.00 (26.00, 43.00)	34.00 (26.00, 43.50)	0.900	0.028
OASIS	31.00 (26.00, 38.00)	31.00 (25.00, 37.00)	31.00 (26.00, 38.00)	0.548	0.054
PT (s)	13.40 (12.17, 15.60)	13.30 (12.10, 15.40)	13.40 (12.30, 15.80)	0.197	0.072
PTT (s)	30.50 (27.40, 40.56)	30.60 (27.35, 40.42)	30.40 (27.50, 40.68)	0.875	− 0.041
CKD, n (%)				0.887
No	810 (75.7)	406 (75.89)	404 (75.51)		− 0.009
Yes	260 (24.3)	129 (24.11)	131 (24.49)		0.009
Diabetes mellitus, n (%)				0.751
No	681 (63.64)	343 (64.11)	338 (63.18)		− 0.019
Yes	389 (36.36)	192 (35.89)	197 (36.82)		0.019
Hypertension, n (%)				0.951
No	615 (57.48)	308 (57.57)	307 (57.38)		− 0.004
Yes	455 (42.52)	227 (42.43)	228 (42.62)		0.004
Atrial fibrillation, n (%)				0.710
No	620 (57.94)	313 (58.50)	307 (57.38)		− 0.023
Yes	450 (42.06)	222 (41.50)	228 (42.62)		0.023
Heart failure, n (%)				0.555
No	729 (68.13)	369 (68.97)	360 (67.29)		− 0.036
Yes	341 (31.87)	166 (31.03)	175 (32.71)		0.036
Malignancy, n (%)				0.674
No	970 (90.65)	487 (91.03)	483 (90.28)		− 0.025
Yes	100 (9.35)	48 (8.97)	52 (9.72)		0.025
COPD, n (%)				0.158
No	977 (91.31)	495 (92.52)	482 (90.09)		− 0.081
Yes	93 (8.69)	40 (7.48)	53 (9.91)		0.081
Moderate to severe liver disease, n (%)				0.527
No	1047 (97.85)	522 (97.57)	525 (98.13)		0.041
Yes	23 (2.15)	13 (2.43)	10 (1.87)		− 0.041
Mannitol, n (%)				0.527
No	1047 (97.85)	525 (98.13)	522 (97.57)		− 0.036
Yes	23 (2.15)	10 (1.87)	13 (2.43)		0.036
Thrombolysis, n (%)				0.881
No	844 (78.88)	421 (78.69)	423 (79.07)		0.009
Yes	226 (21.12)	114 (21.31)	112 (20.93)		− 0.009
Antiplatelet agents, n (%)				0.709
No	229 (21.4)	117 (21.87)	112 (20.93)		− 0.023
Yes	841 (78.6)	418 (78.13)	423 (79.07)		0.023
Anticoagulant drugs, n (%)				0.309
No	36 (3.36)	21 (3.93)	15 (2.80)		− 0.068
Yes	1034 (96.64)	514 (96.07)	520 (97.20)		0.068
CRRT, n (%)				0.475
No	994 (92.9)	494 (92.34)	500 (93.46)		0.045
Yes	76 (7.1)	41 (7.66)	35 (6.54)		− 0.045
Loop diuretic, n (%)		0.540
No	500 (46.73)	255 (47.66)	245 (45.79)		− 0.038
Yes	570 (53.27)	280 (52.34)	290 (54.21)		0.038
Vasoactive agent, n (%)				0.934
No	899 (84.02)	449 (83.93)	450 (84.11)		0.005
Yes	171 (15.98)	86 (16.07)	85 (15.89)		− 0.005
Vasopressin input, n (%)				0.377
No	1037 (96.92)	516 (96.45)	521 (97.38)		0.059
Yes	33 (3.08)	19 (3.55)	14 (2.62)		− 0.059
Mechanical ventilation, n (%)				0.803
No	634 (59.25)	319 (59.63)	315 (58.88)		− 0.015
Yes	436 (40.75)	216 (40.37)	220 (41.12)		0.015

SOFA, sequential organ failure assessment; GCS, Glasgow coma scale; CCI, Charlson comorbidity index; SAPSII, simplified acute physiology score II; OASIS, Oxford acute severity of illness score; PT, prothrombin time; PTT, partial thromboplastin time; CKD, chronic kidney disease; HR, heart rate; SBP, systolic blood pressure; DBP, diastolic blood pressure; MBP, mean blood pressure; RR, respiratory rate; T, temperature; SPO2, oxygen saturation; CRRT, continuous renal replacement therapy; COPD, chronic obstructive pulmonary disease; PSM, propensity score matching

Correlation between ramelteon administration and primary outcomes

In patients suffering from ischemic stroke, the 28-day all-cause mortality rate was significantly reduced in the group treated with ramelteon compared to those who did not receive it, both prior to (6.81%vs.15.47%) and following (6.54%vs.16.64%) PSM (Table 3). The analysis using univariate Cox regression indicated a significant link between ramelteon use and a reduction in 28-day all-cause mortality (pre-PSM: HR = 0.29, 95% CI: 0.21–0.41; post-PSM: HR = 0.28, 95% CI: 0.19–0.41; P < 0.001 for both). In multivariable-adjusted analyses (Model 3), this association remained robust after controlling for a comprehensive set of clinical covariates, including disease severity scores, comorbidities, and treatment variables (pre-PSM: HR = 0.34, 95% CI: 0.24–0.48; post-PSM: HR = 0.23, 95% CI: 0.15–0.35; P < 0.001; Table 4). The Kaplan–Meier survival curves provided additional evidence of this finding, revealing that the ramelteon group had notably higher 28-day survival probabilities than the non-ramelteon group, both prior to and following PSM (log-rank P < 0.0001; Fig. 2A–B).

Table 3.

The association between ramelteon administration and clinical outcomes among patients with ischemic stroke

Characteristics	Total	Non-ramelteon	ramelteon	HR (95% CI)	P-value
Before PSM	n = 3413	n = 2870	n = 543
Primary outcome, n (%)
28-day mortality	481 (14.09)	444 (15.47)	37 (6.81)	0.29 (0.21–0.41)	< 0.001
Secondary outcomes, n (%)
90-day mortality	984 (28.83)	868 (30.24)	116 (21.36)	0.51 (0.42–0.62)	< 0.001
365-day mortality	1217 (35.66)	1049 (36.55)	168 (30.94)	0.54 (0.44–0.66)	< 0.001
After PSM	n = 1070	n = 535	n = 535
Primary outcome, n (%)
28-day mortality	124 (11.59)	89 (16.64)	35 (6.54)	0.28 (0.19–0.41)	< 0.001
Secondary outcomes, n (%)
90-day mortality	283 (26.45)	169 (31.59)	114 (21.31)	0.50 (0.39–0.63)	< 0.001
365-day mortality	366 (34.21)	201 (37.57)	165 (30.84)	0.65 (0.53–0.79)	< 0.001

PSM, propensity score matching; HR, hazard ratio; 95% CI, 95% confidence interval

Table 4.

Multivariate-adjusted HR (95%Cl) of ramelteon administration and clinical outcomes among patients with ischemic stroke

Endpoint event	Before PSM		After PSM
	HR (95% CI)	P-value	HR (95% CI)	P-value
Primary outcome
28-day mortality
Model 1	0.29 (0.21–0.41)	< 0.001	0.28 (0.19–0.41)	< 0.001
Model 2	0.28 (0.20–0.39)	< 0.001	0.26 (0.17–0.38)	< 0.001
Model 3	0.34 (0.24–0.48)	< 0.001	0.23 (0.15–0.35)	< 0.001
Secondary outcome
90-day mortality
Model 1	0.51 (0.42–0.62)	< 0.001	0.50 (0.39–0.63)	< 0.001
Model 2	0.47 (0.38–0.57)	< 0.001	0.44 (0.35–0.57)	< 0.001
Model 3	0.54 (0.44–0.66)	< 0.001	0.43 (0.34–0.55)	< 0.001
365-day mortality
Model 1	0.65 (0.56–0.77)	< 0.001	0.65 (0.53–0.79)	< 0.001
Model 2	0.60 (0.51–0.71)	< 0.001	0.57 (0.46–0.71)	< 0.001
Model 3	0.69 (0.58–0.81)	< 0.001	0.55 (0.45–0.68)	< 0.001

Model 1: Crude;

Model 2: Age, weight, HR, RR, SOFA, GCS, CCI, SAPSII, OASIS, PT, PTT, CKD, diabetes mellitus, hypertension, atrial fibrillation, heart failure, malignancy, COPD, moderate to severe liver disease

Model 3: Age, weight, HR, RR, SOFA, GCS, CCI, SAPSII, OASIS, PT, PTT, CKD, hypertension, moderate to severe liver disease, mechanical ventilation, CRRT, mannitol, antiplatelet agents, anticoagulant drugs, thrombolysis, vasoactive agent, vasopressin input, loop diuretic

SOFA, sequential organ failure assessment; GCS, Glasgow coma scale; CCI, Charlson comorbidity index; SAPSII, simplified acute physiology score II; OASIS, oxford acute severity of illness score; PT, prothrombin time; PTT, partial thromboplastin time; CKD, chronic kidney disease; HR, heart rate; SBP, systolic blood pressure; DBP, diastolic blood pressure; MBP, mean blood pressure; RR, respiratory rate; T, temperature; SPO2, oxygen saturation; CRRT, continuous renal replacement therapy; COPD, chronic obstructive pulmonary disease; PSM, propensity score matching

Fig. 2.

28-day all-cause mortality Kaplan–Meier survival curves for ramelteon users in unmatched and PSM cohorts. (A) In the unmatched cohort, ramelteon use was associated with improved 28-day survival, with a HR of 0.34 (0.24–0.48). (B) In the PSM cohort, ramelteon use was associated with further improved 28-day survival, with a HR of 0.23 (0.15–0.35) Abbreviation: PSM, propensity score matching; HR, hazard ratio; 95% CI, 95% confidence interval

Correlation between ramelteon administration and secondary outcomes

In patients treated with ramelteon, the 90-day all-cause mortality rate was 21.36%, whereas it was 30.24% in those not treated with ramelteon after PSM (Table 3). Table 3 indicates that the one-year (365-day) all-cause mortality rate was 30.94% for those on ramelteon and 36.55% for those not on the medication after PSM. Using unadjusted Cox regression analysis, a significant correlation was found between ramelteon use and decreased 90-day and 365-day all-cause mortality risk. Before the implementation of PSM, the HR were 0.51 (95% CI: 0.42–0.62) for 90-day all-cause mortality and 0.54 (95% CI: 0.44–0.66) for 365-day all-cause mortality. Post-PSM, the associations were still significant, with HR of 0.50 (95% CI: 0.39–0.63) and 0.65 (95% CI: 0.53–0.79), respectively (P < 0.001 for all; Table 3). Further evidence for these findings was provided by Cox models with multivariable adjustments. In the fully adjusted Model 3, ramelteon use remained independently associated with a reduced risk of both 90-day (pre-PSM: HR = 0.54, 95% CI: 0.44–0.66; post-PSM: HR = 0.43, 95% CI: 0.34–0.55, P < 0.001 for all) and 365-day all-cause mortality (pre-PSM: HR = 0.69, 95% CI: 0.58–0.81; post-PSM: HR = 0.55, 95% CI: 0.45–0.68, P < 0.001 for all), as shown in Table 4. Further differences were illustrated by Kaplan–Meier survival analyses, where patients in the ramelteon group consistently showed better survival probabilities at both time points and in both unmatched and matched cohorts (log-rank P < 0.0001; Fig. 3A–D).

Fig. 3.

90 and 365-day all-cause mortality Kaplan–Meier survival curves for ramelteon users in unmatched and PSM cohorts. (A) In the unmatched cohort, thiamine use was associated with improved 90-day survival, with a HR of 0.54 (0.44–0.66). (B) In the PSM cohort, ramelteon use was associated with further improved 90-day survival, with a HR of 0.43 (0.34–0.55) (C) In the unmatched cohort, thiamine use was associated with improved 365-day survival, with a HR of 0.69 (0.58–0.81). (D) In the PSM cohort, aspirin use was associated with further improved 365-day survival, with a HR of 0.55 (0.45–0.68). Abbreviation: PSM, propensity score matching; HR, hazard ratio; 95% CI, 95% confidence interval

Multivariable-adjusted analyses of duration-response and dose–response relationships between ramelteon use and outcomes

The analyses, adjusted for various variables, indicated a clear link between the length of ramelteon administration and the risk of death among ischemic stroke patients. Patients who received ramelteon for under 7 days showed a significant reduction in the risk of all-cause mortality at 28-day (HR = 0.36, 95% CI: 0.21–0.61), 90-day (HR = 0.48, 95% CI: 0.35–0.66) and 365-day (HR = 0.59, 95% CI: 0.45–0.77; all P < 0.001) compared to those who did not receive the treatment, all with P < 0.001 (Table 5). Intermediate treatment duration (7–14 days) was also associated with reduced 28-day and 90-day all-cause mortality (HR = 0.29 and 0.58, respectively; both P < 0.01), although the association with 365-day all-cause mortality was not statistically significant (HR = 0.74, 95% CI: 0.53–1.03, P = 0.072). Those who received treatment for more than 14 days experienced the greatest benefit, with the lowest mortality risks observed at 28 days (HR = 0.09, 95% CI: 0.04–0.21, P < 0.001), 90 days (HR = 0.28, 95% CI: 0.18–0.42, P < 0.001) and 365 days (HR = 0.41, 95% CI: 0.29–0.57, P < 0.001).

Table 5.

Duration-response and dose–response relationships between ramelteon administration and clinical outcomes among patients with ischemic stroke after PSM

Categories		28-day mortality		n (%)	90-day mortality		n (%)	365-day mortality
Duration (days)	n (%)	HR (95% CI)	P-value	n (%)	HR (95% CI)	P-value	n (%)	HR (95% CI)	P-value
Without ramelteon	89 (71.77)	Reference (1)		169 (59.72)	Reference (1)		201 (54.92)	Reference (1)
< 7	17 (13.71)	0.36 (0.21–0.61)	< 0.001	50 (17.67)	0.48 (0.35–0.66)	< 0.001	74 (20.22)	0.59 (0.45–0.77)	< 0.001
7–14	12 (9.68)	0.29 (0.15–0.56)	< 0.001	35 (12.37)	0.58 (0.40–0.85)	0.005	45 (12.30)	0.74 (0.53–1.03)	0.072
> 14	6 (4.84)	0.09 (0.04–0.21)	< 0.001	29 (10.25)	0.28 (0.18–0.42)	< 0.001	46 (12.57)	0.41 (0.29–0.57)	< 0.001
P for trend			< 0.001			< 0.001			< 0.001
Cumulative dose (mg)	n (%)	HR (95% CI)	P-value	n (%)	HR (95% CI)	P-value	n (%)	HR (95% CI)	P-value
Without ramelteon	89 (71.77)	Reference (1)		169 (59.72)	Reference (1)		201 (54.92)	Reference (1)
< 100	17 (13.71)	0.36 (0.21–0.61)	< 0.001	51 (18.02)	0.48 (0.35–0.67)	< 0.001	75 (20.49)	0.59 (0.45–0.78)	< 0.001
100–300	12 (9.68)	0.29 (0.15–0.56)	< 0.001	34 (12.01)	0.57 (0.39–0.84)	0.004	44 (12.02)	0.73 (0.52–1.02)	0.065
> 300	6 (4.84)	0.09 (0.04–0.21)	< 0.001	29 (10.25)	0.28 (0.18–0.42)	< 0.001	46 (12.57)	0.41 (0.29–0.57)	< 0.001
P for trend			< 0.001			< 0.001			< 0.001

Model 1: Crude;

Model 2: Age, weight, HR, RR, SOFA, GCS, CCI, SAPSII, OASIS, PT, PTT, CKD, diabetes mellitus, hypertension, atrial fibrillation, heart failure, malignancy, COPD, moderate to severe liver disease

Model 3: Age, weight, HR, RR, SOFA, GCS, CCI, SAPSII, OASIS, PT, PTT, CKD, hypertension, moderate to severe liver disease, mechanical ventilation, CRRT, mannitol, antiplatelet agents, anticoagulant drugs, thrombolysis, vasoactive agent, vasopressin input, loop diuretic

SOFA, sequential organ failure assessment; GCS, Glasgow coma scale; CCI, Charlson comorbidity index; SAPSII, simplified acute physiology score II; OASIS, oxford acute severity of illness score; PT, prothrombin time; PTT, partial thromboplastin time; CKD, chronic kidney disease; HR, heart rate; SBP, systolic blood pressure; DBP, diastolic blood pressure; MBP, mean blood pressure; RR, respiratory rate; T, temperature; SPO2, oxygen saturation; CRRT, continuous renal replacement therapy; COPD, chronic obstructive pulmonary disease; PSM, propensity score matching

Considering cumulative dose, individuals administered more than 300 mg of ramelteon showed a marked decrease in mortality risks across all timepoints, with HR of 0.09 (95% CI: 0.04–0.21) for 28-day, 0.28 (95% CI: 0.18–0.42) for 90-day, and 0.41 (95% CI: 0.29–0.57) for 365-day all-cause mortality (P < 0.001 for all comparisons). Those receiving cumulative doses below 100 mg experienced significant declines in all-cause mortality rates at 28 days (HR = 0.36, 95% CI: 0.21–0.61), 90 days (HR = 0.48, 95% CI: 0.35–0.67), and 365 days (HR = 0.59, 95% CI: 0.45–0.78; all P < 0.001). While patients in the 100–300 mg group also showed significant benefits for 28-day (HR = 0.29, 95% CI: 0.15–0.56, P < 0.001) and 90-day all-cause mortality (HR = 0.57, 95% CI: 0.39–0.84, P = 0.004), the association with 365-day all-cause mortality did not reach statistical significance (HR = 0.73, 95% CI: 0.52–1.02, P = 0.065).

Subgroup analysis

To evaluate the uniformity of the relationship between ramelteon usage and all-cause mortality, subgroup analyses were performed across clinically relevant categories (Supplementary Table 3). The survival benefit provided by ramelteon was consistently observed across all examined subgroups, with hazard ratios remaining below 1.0 and P values under 0.05 in the majority of cases. Both younger patients, those under 65, and older patients, those 65 and older, had similar decreases in all-cause mortality (HR = 0.24 for both; P for interaction = 0.950). The protective effect was also consistent across sex (female: HR = 0.28; male: HR = 0.20; P for interaction = 0.464) and observed regardless of the presence or absence of comorbidities such as hypertension, diabetes, chronic kidney disease, atrial fibrillation, heart failure, malignancy, and COPD (all P for interaction > 0.05). Importantly, ramelteon use was associated with reduced all-cause mortality among patients undergoing critical interventions such as mechanical ventilation (HR = 0.23), CRRT (HR = 0.35), and those receiving vasoactive agents (HR = 0.18), antiplatelet agents (HR = 0.22), or thrombolysis (HR = 0.30), without significant effect modification across these subgroups (P for interaction all > 0.1). Overall, the absence of statistically significant interaction effects indicates that the beneficial effect of ramelteon on all-cause mortality was broadly consistent across various clinical scenarios and patient populations.

Sensitivity analysis

To verify the reliability of the main results, a sensitivity analysis excluded 151 ischemic stroke patients with known sleep disorders. The 919 patients left were re-evaluated after being matched 1:1 based on propensity scores. Consistent with the main analysis, ramelteon use was significantly associated with reduced all-cause mortality across all time points (Supplementary Table 4). For the primary endpoint of 28-day mortality, the multivariate-adjusted hazard ratios were 0.30 (95% CI: 0.20–0.46) in Model 1, 0.29 (95% CI: 0.19–0.44) in Model 2, and 0.28 (95% CI: 0.19–0.43) in Model 3 (all P < 0.001). The secondary endpoints showed similar protective associations. The adjusted hazard ratios for 90-day mortality were 0.51 (95% CI: 0.39–0.65) in Model 1, 0.45 (95% CI: 0.35–0.58) in Model 2, and 0.47 (95% CI: 0.36–0.60) in Model 3. For a 365-day mortality period, the hazard ratios were 0.68 (95% CI: 0.54–0.84), 0.60 (95% CI: 0.48–0.74), and 0.63 (95% CI: 0.50–0.78), respectively (all P < 0.001). These combined results affirm the persistent association between ramelteon use and a reduced risk of mortality, even after excluding patients diagnosed with sleep disorders.

Discussion

The retrospective cohort study, which was large and involved critically ill ischemic stroke patients, demonstrated that providing ramelteon during ICU hospitalization was independently correlated with a substantial reduction in all-cause mortality in both the short and long term. The robustness of this protective association was maintained after propensity score matching and adjusting for a diverse range of demographic, clinical, and therapeutic variables. The analyses of duration-response and dose–response demonstrated that longer administration periods, exceeding 14 days, and higher cumulative doses, over 300 mg, of ramelteon were consistently linked to lower all-cause mortality risks at 28, 90, and 365 days. The consistency of this association across various clinical strata, including age, sex, comorbidities, and treatment interventions, was demonstrated by subgroup analyses, suggesting a widely applicable survival benefit of ramelteon in this high-risk population. Patients with documented sleep disorder diagnoses during their ICU stay were additionally identified, and a sensitivity analysis was performed excluding these cases. The association of ramelteon with all-cause mortality was stable, suggesting that the findings are not exclusively driven by patients suffering from sleep disorders. The results imply that ramelteon could provide benefits related to ischemic stroke injury in addition to its potential sleep-improving effects.

The exploration of ramelteon for managing strokes is a novel research field. Ramelteon was first approved as an oral medication for insomnia treatment (Stroethoff et al. 2018). Nonetheless, recent studies have initiated an exploration of its potential role in ischemic stroke. The identification of ramelteon as a neuroprotective agent in ischemic stroke models has unlocked new research possibilities (Zhang et al. 2024). Ramelteon was recognized as a potential neuroprotective drug in a mouse model experiencing middle cerebral artery occlusion (MCAO). The neuroprotection provided by ramelteon was attributed to the MT1 receptor, with its effects being lessened in MT1-deficient mice and neurons (Zhang et al. 2024). The discovery brings new understanding to stroke therapy and has led to more investigations into the potential use of ramelteon in stroke treatment. In a former study, the protective effects of ramelteon on LPS-induced blood–brain barrier (BBB) disruption were investigated by assessing changes in vascular inflammation markers (Liu et al. 2021). They specifically quantified the levels of the pro-inflammatory cytokines IL-1β and MCP-1, and also examined the expression of the tight junction proteins ZO-1 and occludin. The use of ramelteon led to lower IL-1β and MCP-1 production and higher ZO-1 and Occludin expression, implying these could serve as indicators of ramelteon's influence on BBB integrity in ischemic stroke. By activating mitochondrial potassium channels and preventing pathways that cause mitochondrial death, ramelteon likely preserves mitochondrial integrity and offers protective effects (Stroethoff et al. 2018; Zhang et al. 2024). Therefore, fluctuations in mitochondrial membrane potential, the creation of mitochondrial reactive oxygen species, and the expression of proteins tied to mitochondrial function could potentially serve as biomarkers for determining ramelteon's efficacy. Stroke survivors often experience sleep disorders, which can slow down their recovery process. Ramelteon may assist in improving recovery as a sleep aid. Research on rats with transient ischemic stroke demonstrated that repeated zolpidem treatment in the subacute phase could improve circadian rhythm issues, decrease sleep fragmentation, and increase the depth of sleep (Zhong et al. 2024). Despite being about zolpidem, this study points out the significance of sleep in the recovery process after a stroke. Ramelteon, acting as a melatonin receptor agonist, could have comparable effects. In a mouse model of ischemic stroke, administering ramelteon was connected to enhanced sleep parameters and lessened ischemic injury (Wu et al. 2020). When administered orally at 3.0 mg/kg, ramelteon significantly mitigated ischemic injury, even when given 4 h after ischemia onset, and it also inhibited autophagy in the peri-infarct cortex by regulating the AMPK/mTOR signaling pathway. Moreover, ramelteon might have a direct effect on the neuronal activity that controls sleep and wake regulation. In an experiment conducted on rats, ramelteon treatment improved non-rapid eye movement (NREM) sleep and elevated fast gamma power in the primary motor cortex during NREM sleep. Modifying neural oscillations during sleep could play a role in offline information processing and might aid in recovering from a stroke (Yoshimoto et al. 2021). The combination of ramelteon with standard treatments for stroke is a current research focus. In managing ischemic stroke, standard approaches often involve the use of thrombolytic therapies and antiplatelet agents (Phipps and Cronin 2020; Powers et al. 2018). Research on a mouse pre-clinical stroke model revealed that administering ceruletide and alpha-1 antitrypsin together reduced infarct volume by 39.42%, a reduction not observed with individual administration (Simats et al. 2022). Ramelteon, when used alongside standard treatments, could also result in synergistic effects. Since ramelteon has been shown to have neuroprotective effects through mechanisms such as activating mitochondrial potassium channels and counteracting oxidative stress, combining it with thrombolytic agents that aim to restore blood flow may enhance the overall treatment effect (Stroethoff et al. 2018; Zhang et al. 2024). Existing studies offer some insights, even though clinical trials on ramelteon in stroke patients are still scarce. A study on ramelteon’s preventive effect against postoperative delirium after elective liver resection revealed a significantly lower delirium incidence in the ramelteon group (Hokuto et al. 2020). Although this was not a stroke-specific trial, it suggests that ramelteon may have a beneficial effect on reducing delirium, which is also a common complication in stroke patients. Understanding the safety and efficacy of ramelteon requires a comparative study of mortality rates in its users. According to a retrospective cohort study, septic patients exposed to ramelteon experienced a significant reduction in mortality at 30 and 90 days compared to those who were not exposed (Han et al. 2024). Following PSM, ramelteon-exposed patients had a 30-day mortality HR of 0.53 (95% CI: 0.47–0.61, P < 0.001), and the HR for 90-day mortality was also significantly decreased. In other patient groups, the impact of different drugs on mortality has been inconsistent. For example, in a study on patients with pre-dialysis advanced diabetic kidney disease, the use of ketoanalogue supplements was associated with a lower 5-year all-cause mortality rate (34.7% vs 42.7%) and an adjusted hazard ratio of 0.73 (Chen et al. 2021). Juxtaposing these results with data on mortality associated with ramelteon can give a broader context for evaluating its impact on mortality. Research indicates that Ramelteon can improve the blood–brain barrier (BBB) in rats with focal cerebral ischemia, possibly reducing the risk of post-stroke depression (Qi et al. 2023). Ramelteon pretreatment in this study enhanced depressive-like behaviors and decreased the infarct area in MCAO rats. It also controlled the expression of occludin, a tight-junction protein, by inhibiting Egr-1, thereby protecting the BBB. Future research could focus on optimizing this mechanism by figuring out the ideal timing and dosage of ramelteon to enhance its protective effects on the BBB and its neuroprotective properties.

Ramelteon, known for its selective action on melatonin receptors, may offer neuroprotective effects during ischemic stroke. The pathophysiological process of ischemic stroke is complex, with oxidative stress being a significant contributor (Orellana-Urzúa et al. 2020; Pawluk et al. 2024). Reactive oxygen species are produced during reperfusion after ischemia, resulting in oxidative stress and damage to brain tissue. Brain cells can undergo apoptosis, autophagy, and necrosis due to oxidative stress (Aslankoc et al. 2022). Ramelteon could play a part in reducing this oxidative stress. The treatment with ramelteon in mice was shown to lessen LPS-induced blood–brain barrier hyperpermeability by triggering the Nrf2 signaling pathway (Liu et al. 2021). This activation led to a reduction in oxidative stress within cerebral vessels, as indicated by decreased levels of interleukin-1β (IL-1β) and monocyte chemoattractant protein-1 (MCP-1), along with increased expression of the tight junction proteins zonula occludens-1 (ZO-1) and occludin. Moreover, the role of non-coding RNAs (ncRNAs) in the pathophysiology of ischemic stroke is significant, particularly in terms of oxidative stress (Su et al. 2022). The oxidative stress process post-ischemic stroke is mediated by glial cells, essential components of the brain, with ncRNAs participating in this mechanism (Su et al. 2022; Zhu et al. 2022). Ramelteon’s direct link to ncRNAs in ischemic stroke is not fully explored, but its antioxidant properties could potentially affect ncRNA-mediated oxidative stress pathways.

Even with its advantages, this study has several limitations that should be pointed out. Initially, Using the MIMIC-IV database, which originates from routine clinical care and not controlled experiments, the study inherently employs a retrospective cohort design for its analysis. Second, residual confounding could still be present despite rigorous adjustments and propensity score matching, as unmeasured or unknown variables might influence the observed associations. Third, the database lacked information on stroke severity, such as NIHSS scores, lesion locations, or imaging results, which restricts our capacity to categorize patients based on stroke burden. Fourth, missing information on sleep patterns or circadian rhythm disruptions, which are key mechanistic pathways that ramelteon might influence, obstructed the mechanistic interpretation. Fifth, Another important limitation is the absence of stroke-specific severity measures in the MIMIC-IV database, such as NIHSS scores, infarct volume, or lesion location.These factors are recognized as predictors of mortality and functional outcomes following a stroke, and their lack might have led to residual confounding concerning the baseline neurological injury burden. Despite adjusting for several global illness severity scores such as SOFA, OASIS, SAPS II, and GCS, along with critical interventions to partially assess patient acuity, these surrogate measures cannot entirely replace dedicated stroke severity metrics.

Conclusion

In this large retrospective cohort study, the use of ramelteon during ICU hospitalization among critically ill ischemic stroke patients was independently associated with a significant decrease in all-cause mortality, both in the short and long term. These associations were resilient after propensity score matching and multivariable adjustments, and were further corroborated by consistent findings across dose- and duration-response analyses, as well as various clinical subgroups. Even though the study's observational design restricts causal inference, our findings propose that ramelteon might have therapeutic advantages for this high-risk population. To verify these results and explore the underlying mechanisms, future prospective studies and randomized controlled trials should be conducted.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contribution

The data was collected by LQ, LHW, and HMM, who also prepared the manuscript draft. The idea was conceived and the data were analyzed by LQ and ZMH. Funding for the acquisition was provided by LHW and HMM, who also proposed the idea. FHX, LHW, along with HMM, made revisions and edits to the manuscript. Data collection was supported by LQ, ZMH, FHX, LHW, and HMM, who also provided important feedback on the manuscript draft. The final manuscript version was contributed to, reviewed, and approved by all authors before submission. The authors declare that all data were generated in-house and that no paper mill was used.

Funding

Support for this study came from the Youth Program of the Shanxi Province Science and Technology Innovation Talent Team (Grant No. 202304051001039).

Data availability

Upon a reasonable request, data will be made available.

Declarations

Ethics approval and consent to participate

This study was exempt from ethical review as it involved secondary analysis of a publicly available, de-identified third-party database (MIMIC-IV). In line with the Declaration of Helsinki’s ethical standards, this study was approved by the Ethics Committee of Massachusetts Institute of Technology and Beth Israel Deaconess Medical Center's review boards (approval number: 2001P-001699/14). All individuals provided their written informed consent prior to joining the study.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.