Abstract
Selective serotonin reuptake inhibitors (SSRIs) are the most commonly prescribed drugs for mental disorders in critically ill patients. We performed a retrospective cohort study to investigate the association between pre-ICU use of SSRIs and mortality in critically ill adults with mental disorders. We identified critically ill adults with mental disorders based on the Medical Information Mart in Intensive Care-IV database. The exposure was the use of SSRIs during the period after hospital admission and before ICU admission. The outcome was in-hospital mortality. Time-dependent Cox proportional hazards regression models were used to estimate the adjusted hazard ratio (aHR) with 95% confidence interval (CI). To further test the robustness of the results, we performed propensity score matching and marginal structural Cox model estimated by inverse probability of treatment weighting. The original cohort identified 16601 patients. Of those, 2232 (13.4%) received pre-ICU SSRIs, and 14369 (86.6%) did not. Matched cohort obtained 4406 patients, with 2203 patients in each group (SSRIs users vs. non-users). In the original cohort, pre-ICU use of SSRIs was associated with a 24% increase in the hazard for in-hospital mortality (aHR, 1.24; 95% CI, 1.05–1.46; P = 0.010). The results were robust in the matched cohort (aHR, 1.26; 95% CI, 1.02–1.57; P = 0.032) and the weighted cohort (aHR, 1.43; 95% CI, 1.32–1.54; P < 0.001). Pre-ICU use of SSRIs is associated with an increase in the hazard for in-hospital mortality in critically ill adults with mental disorders.
Subject terms: Bipolar disorder, Depression, Schizophrenia
Introduction
Mental disorders remain among the top ten leading causes of burden worldwide [1]. In critically ill patients, mental disorders are common before intensive care unit (ICU) admission, during ICU stay, and after discharge [2–4]. Both pre-existing and newly developed mental disorders are associated with unfavorable outcomes in critically ill patients [5–7]. Common mental disorders are typically treated with antidepressants and/or antipsychotics. Of these, selective serotonin reuptake inhibitors (SSRIs) are the first-line pharmacological treatments [8], with approximately one in five critically ill patients receiving them before ICU admission [9]. Previous studies suggested that the use of SSRIs was associated with increased mortality in diabetic patients due to ischemic stroke [10] or among older adults with chronic obstructive pulmonary disease [11]. However, data on the association between the use of SSRIs and outcomes in critically ill patients with mental disorders is scarce.
In 2014, a retrospective study on the topic was conducted using the Medical Information Mart in Intensive Care-II (MIMIC-II) database. The study found that the use of SSRIs prior to ICU admission was associated with increased in-hospital mortality in critically ill patients [9]. However, the results need to be interpreted with caution due to not adjusting for potential confounding factors, such as delirium and coma, which are known to increase mortality in critically ill patients [12, 13]. In the present study, we took into account the two factors and performed the mediation analysis to determine whether the association was mediated by delirium or coma. Furthermore, we used the marginal structural Cox model (MSCM) to adjust the daily Sequential Organ Failure Assessment (SOFA) score to explore the association between pre-ICU use of SSRIs and mortality. Moreover, we utilized the most updated version of MIMIC-IV (version 2.0), which offers more detailed medical records than MIMIC-II [14]. Here, we performed a retrospective cohort study based on data from MIMIC-IV version 2.0 to evaluate the association between pre-ICU use of SSRIs and in-hospital mortality in critically ill adults with mental disorders.
Methods
Study design
We employed a large-sample retrospective cohort design to determine the association between pre-ICU use of SSRIs and in-hospital mortality in critically ill adults with mental disorders. The study is based on the MIMIC-IV version 2.0 database, in which data were deidentified and publicly available. It was approved by the institutional review boards of the Beth Israel Deaconess Medical Center and Massachusetts Institute of Technology without a requirement for individual patient informed consent. This study is reported following the Strengthening the Reporting of Observational Studies in Epidemiology statement [15].
Data source
Data were derived from the latest MIMIC-IV version 2.0 database. MIMIC-IV contains electronic health records from over 50000 patients admitted to the ICUs at the Beth Israel Deaconess Medical Center (Boston, Massachusetts, USA) from 2008 to 2019. Data items included baseline characteristics, illness severity, coexisting conditions, vital signs, laboratory tests, therapeutic interventions, prescribed medications, diagnostic codes, imaging reports, and survival data. The data can be downloaded after completing relevant courses and obtaining a certificate. Data were extracted by structured query language using the patients’ stay_id as the index.
Patients, exposures, and outcomes
All critically ill adults (age ≥ 18 years) with mental disorders were selected from the MIMIC-IV version 2.0 database. For patients with multiple admissions, only data from the first ICU admission were used. Mental disorders, including depression, anxiety, schizophrenia, bipolar disorders, and other unspecified mental disorders, were identified by the International Classification of Diseases, Ninth Revision (ICD-9) and Tenth Revision (ICD-10) codes with the assistance of a psychiatrist colleague (Supplementary Table 1). The exposure was the use of SSRIs during the period after hospital admission and before ICU admission and extracted from the prescriptions in the MIMIC database. This study included six common SSRIs prescriptions: citalopram, escitalopram, fluoxetine, sertraline, paroxetine, and fluvoxamine. The outcome was in-hospital mortality. Follow-up began with hospital admission and ended at either in-hospital death or hospital discharge. All variables included in this study were missing <2% (Supplementary Fig. 1). Given the low level of missingness, patients with incomplete or missing data were excluded.
Covariates
We collected patients’ records after ICU admission, including age, sex, weight, admission type, first care unit, Charlson comorbidity index, Simplified Acute Physiology Score (SAPS) II, daily SOFA score, delirium during ICU stay (assessed by Confusion Assessment Method for the ICU) [16], coma during ICU stay (assessed by Richmond Agitation Sedation Scale –4 or −5) [17], therapy at baseline (antibiotic, vasopressors, continuous renal replacement therapy, and invasive mechanical ventilation), coexisting conditions (depression, cerebrovascular disease, dementia, hypertension, myocardial infarct, congestive heart failure, chronic pulmonary disease, diabetes mellitus, renal disease, liver disease, cancer, and sepsis), and medications (other antidepressants, antipsychotics, and Z drugs; Supplementary Table 2). We extracted data on the first vital signs (heart rate, mean arterial pressure, respiratory rate, body temperature, and SpO2) and first laboratory tests (white blood cell, hemoglobin, platelet, potassium, sodium, chloride, bicarbonate, glucose, creatinine, and urea nitrogen).
Statistical analysis
Continuous variables were expressed as median (interquartile range [IQR]) and were compared by the Mann-Whitney U test. Categorical variables were expressed as numbers (percentages) and were compared by the Chi-square test or Fisher’s exact test. Time-dependent Cox proportional hazards regression models were used to estimate the adjusted hazard ratio (aHR) with 95% confidence interval (CI) for the association between pre-ICU use of SSRIs and mortality. The Schoenfeld residual test was used to check the proportional hazard assumption, and variables with a P-value > 0.05 met the assumption. When it is not met, time-varying effects occur. The coxph() function has a tt argument to specify the specific time transformation. In this study, the tt function was applied to the time-varying variables and defined as “tt = function(x, t,…) x * log(t + 20)”, where x is a time-varying variable and t is the analysis time [18]. Based on clinical expertise and previous literature in combination with a statistical perspective, in the multivariate time-dependent Cox proportional hazards regression models, we included the following variables: age, sex, admission type, first care unit, Charlson comorbidity index, depression, sepsis, delirium, coma, SAPS II, SOFA, antibiotic, vasopressors, invasive mechanical ventilation, other antidepressants, antipsychotics, and Z drugs [12, 19–22]. The variance inflation factors (VIFs) were used to examine the presence of multicollinearity among variables.
We performed propensity score matching (PSM) to balance the confounding factors before treatment between the groups. PSM was constructed with a matched cohort of patients with SSRIs and those without SSRIs with similar propensity score values. Specifically, we used a multiple logistic regression model to estimate patients’ propensity scores while applying one-to-one nearest neighbor matching with a caliper width of 0.02 without replacement [23].
In the MSCM, potential variables included age, sex, admission type, first care unit, Charlson comorbidity index, depression, sepsis, delirium, coma, SAPS II, antibiotic, vasopressors, invasive mechanical ventilation, other antidepressants, antipsychotics, and Z drugs. Besides, the daily SOFA score was included in the MSCM as a time-varying confounding factor. Inverse probability of treatment weighting (IPTW) was used to estimate the parameters of MSCM [24, 25]. IPTW applied the inverse of the propensity score as a weight to construct two pseudo-populations using the original population, also known as the weighted cohort [23]. XGBoost (Extreme Gradient Boosting) is an efficient gradient boosting decision tree algorithm that can be used to estimate propensity scores, achieved with the R package “twang” [26]. To reflect the degree of influence of different covariates on the groups or the degree of imbalance between the groups, we applied the XGBoost model to rank the 45 covariates included in this study. The main parameters of the XGBoost model can be found in Supplementary Table 3. The validity of PSM and IPTW was assessed by calculating the standardized mean differences (SMDs), with SMD < 0.1 considered to establish a balanced distribution of confounding factors between the groups [27]. Additionally, we utilized the absolute standardized differences and the distribution balance for the variables before and after weighting to provide the diagnostics for PSM and IPTW [28, 29].
Subgroup analyses
We performed subgroup analyses to investigate the association between pre-ICU use of SSRIs and mortality in different subgroups. The original population was classified into subgroups based on the following characteristics: age (<65 and ≥65 years), sex (male and female), first care unit (medical/surgical ICU and other ICUs), and population (depression and other mental disorders).
Further analyses
In the original cohort, we performed several further analyses. First, the mediation analysis to determine whether delirium or coma was the mediator of the association between pre-ICU use of SSRIs and mortality [30]. Second, the class effect analysis (i.e., single drug of SSRIs). Third, the analysis to compare the effect of continuing or stopping SSRIs during ICU stay on mortality.
All statistical analyses were performed using R version 4.1.0 (R Foundation). A two-sided P-value < 0.05 was considered to indicate statistical significance.
Results
Patient selection
Figure 1 shows the process of patient selection. We identified 16601 patients in the original cohort. Of those, 2232 (13.4%) received pre-ICU SSRIs, and 14369 (86.6%) did not. In the matched cohort, we obtained 4406 patients, with 2203 patients in each group. In the weighted cohort, we also obtained two hypothetical populations with a ratio close to 1:1 (Supplementary Table 4).
Fig. 1. Flow diagram of patient selection.
ICU intensive care unit, SSRIs selective serotonin reuptake inhibitors.
Baseline characteristics
Table 1 displays the detailed baseline characteristics. In the original cohort, compared with patients in the non-users group, those in the SSRIs users group were older (median [IQR], 65.0 [55.0, 75.0] years vs. 62.0 [51.0, 73.0] years; P < 0.001), had higher Charlson comorbidity index (median [IQR], 6.0 [4.0, 8.0] vs. 5.0 [3.0, 7.0]; P < 0.001), were more likely to be receiving antibiotic (1405/2232 [62.9%] vs. 8273/14369 [57.6%]; P < 0.001) and invasive mechanical ventilation (587/2232 [26.3%] vs. 3368/14369 [23.4%]; P = 0.003), and had higher prevalence of coexisting conditions, including hypertension, myocardial infarct, congestive heart failure, chronic pulmonary disease, diabetes mellitus, and renal disease (all P < 0.05). In the matched and weighted cohorts, the imbalances in variables before receiving SSRIs were significantly improved between the groups, such as age, sex, and Charlson comorbidity index, with SMDs <0.1 (Supplementary Fig. 2).
Table 1.
Baseline patient characteristics in the original and matched cohorts.
| Variables | Original cohort | Matched cohort | ||||
|---|---|---|---|---|---|---|
| Patients, No. (%) | P-value | Patients, No. (%) | P-value | |||
| SSRIs users (n = 2232) | Non-users (n = 14369) | SSRIs users (n = 2203) | Non-users (n = 2203) | |||
| Age, median (IQR), year | 65.0 (55.0, 75.0) | 62.0 (51.0, 73.0) | <0.001 | 65.0 (55.0, 75.0) | 66.0 (56.0, 76.0) | 0.429 |
| Sex | ||||||
| Male | 972 (43.5) | 6687 (46.5) | 0.009 | 964 (43.8) | 959 (43.5) | 0.903 |
| Female | 1260 (56.5) | 7682 (53.5) | 1239 (56.2) | 1244 (56.5) | ||
| Weight, kg | 79.6 (66.7, 95.6) | 78.0 (65.2, 94.1) | 0.001 | 79.6 (66.7, 95.6) | 80.1 (67.2, 96.8) | 0.273 |
| Charlson comorbidity index, median (IQR) | 6.0 (4.0, 8.0) | 5.0 (3.0, 7.0) | <0.001 | 6.0 (4.0, 8.0) | 6.0 (4.0, 8.0) | 0.595 |
| SAPS II, median (IQR) | 33.0 (26.0, 41.0) | 31.0 (23.0, 40.0) | <0.001 | 33.0 (26.0, 41.0) | 33.0 (25.0, 42.0) | 0.621 |
| SOFA, median (IQR) | 2.0 (0.0, 3.0) | 1.0 (0.0, 3.0) | <0.001 | 2.0 (0.0, 3.0) | 1.0 (0.0, 3.0) | <.001 |
| Admission type | ||||||
| Emergency | 1048 (47.0) | 8522 (59.3) | <0.001 | 1040 (47.2) | 993 (45.1) | 0.364 |
| Urgent | 418 (18.7) | 2337 (16.3) | 410 (18.6) | 428 (19.4) | ||
| Othera | 766 (34.3) | 3510 (24.4) | 753 (34.2) | 782 (35.5) | ||
| First care unit | ||||||
| Medical/surgical ICU | 1360 (60.9) | 10580 (73.6) | <0.001 | 1341 (60.9) | 1466 (66.5) | <0.001 |
| Otherb | 872 (39.1) | 3789 (26.4) | 862 (39.1) | 737 (33.5) | ||
| Delirium during ICU stay | 529 (23.7) | 4157 (28.9) | <0.001 | 519 (23.6) | 643 (29.2) | <0.001 |
| Coma during ICU stay | 614 (27.5) | 3374 (23.5) | <0.001 | 603 (27.4) | 579 (26.3) | 0.434 |
| Therapy at baseline | ||||||
| Antibiotic | 1405 (62.9) | 8273 (57.6) | <0.001 | 1382 (62.7) | 1319 (59.9) | 0.055 |
| Vasopressors | 335 (15.0) | 2501 (17.4) | 0.006 | 327 (14.8) | 415 (18.8) | <0.001 |
| CRRT | 19 (0.9) | 154 (1.1) | 0.400 | 19 (0.9) | 22 (1.0) | 0.754 |
| Invasive mechanical ventilation | 587 (26.3) | 3368 (23.4) | 0.003 | 580 (26.3) | 506 (23.0) | 0.011 |
| Coexisting conditions | ||||||
| Depression | 1766 (79.1) | 8769 (61.0) | <0.001 | 1740 (79.0) | 1726 (78.3) | 0.633 |
| Cerebrovascular disease | 291 (13.0) | 1859 (12.9) | 0.923 | 289 (13.1) | 316 (14.3) | 0.255 |
| Dementia | 78 (3.5) | 689 (4.8) | 0.008 | 78 (3.5) | 67 (3.0) | 0.398 |
| Hypertension | 949 (42.5) | 5711 (39.7) | 0.014 | 941 (42.7) | 952 (43.2) | 0.761 |
| Myocardial infarct | 383 (17.2) | 2164 (15.1) | 0.011 | 380 (17.2) | 384 (17.4) | 0.905 |
| Congestive heart failure | 767 (34.4) | 3844 (26.8) | <0.001 | 756 (34.3) | 780 (35.4) | 0.467 |
| Chronic pulmonary disease | 831 (37.2) | 4585 (31.9) | <.001 | 817 (37.1) | 853 (38.7) | 0.277 |
| Diabetes mellitus | 714 (32.0) | 4278 (29.8) | 0.036 | 705 (32.0) | 739 (33.5) | 0.290 |
| Renal disease | 529 (23.7) | 2836 (19.7) | <0.001 | 522 (23.7) | 538 (24.4) | 0.597 |
| Liver disease | 299 (13.4) | 2072 (14.4) | 0.210 | 294 (13.3) | 285 (12.9) | 0.721 |
| Cancer | 264 (11.8) | 1627 (11.3) | 0.507 | 260 (11.8) | 275 (12.5) | 0.518 |
| Sepsis | 922 (41.3) | 6703 (46.6) | <0.001 | 905 (41.1) | 1028 (46.7) | <0.001 |
| Vital signs, median (IQR) | ||||||
| Heart rate (beats/min) | 83.0 (74.0, 98.0) | 88.0 (76.0, 103.0) | <0.001 | 83.0 (74.0, 98.0) | 87.0 (75.0, 101.0) | <0.001 |
| Mean arterial pressure (mm Hg) | 81.0 (71.0, 93.0) | 84.0 (73.0, 97.0) | <0.001 | 81.0 (71.0, 93.0) | 83.0 (71.0, 95.0) | 0.003 |
| Respiratory rate (breaths/min) | 18.0 (15.0, 22.0) | 18.0 (15.0, 23.0) | <0.001 | 18.0 (15.0, 22.0) | 18.0 (15.0, 23.0) | 0.008 |
| Body temperature (°C) | 36.7 (36.4, 37.1) | 36.8 (36.5, 37.1) | 0.001 | 36.7 (36.4, 37.1) | 36.8 (36.5, 37.1) | 0.095 |
| SpO2 (%) | 98.0 (95.0, 100.0) | 98.0 (95.0, 100.0) | 0.970 | 98.0 (95.0, 100.0) | 98.0 (95.0, 100.0) | 0.297 |
| Laboratory tests, median (IQR) | ||||||
| White blood cell (k/uL) | 10.5 (7.5, 14.3) | 10.1 (7.1, 14.1) | 0.012 | 10.5 (7.5, 14.3) | 10.3 (7.2, 14.4) | 0.742 |
| Hemoglobin (g/dL) | 10.0 (8.6, 11.5) | 10.6 (9.0, 12.2) | <0.001 | 10.0 (8.6, 11.5) | 10.2 (8.9, 11.8) | 0.002 |
| Platelet (k/uL) | 197.5 (143.0, 266.0) | 201.0 (146.0, 265.0) | 0.266 | 197.0 (143.0, 266.0) | 198.0 (144.0, 261.0) | 0.662 |
| Potassium (mEq/L) | 4.1 (3.7, 4.5) | 4.1 (3.7, 4.5) | 0.015 | 4.1 (3.7, 4.5) | 4.1 (3.7, 4.5) | 0.757 |
| Sodium (mEq/L) | 138.0 (136.0, 140.0) | 139.0 (136.0, 141.0) | <0.001 | 138.0 (136.0, 140.0) | 138.0 (136.0, 141.0) | 0.019 |
| Chloride (mEq/L) | 103.0 (99.0, 107.0) | 104.0 (100.0, 108.0) | 0.083 | 104.0 (99.0, 107.0) | 104.0 (99.0, 107.0) | 0.948 |
| Bicarbonate (mEq/L) | 24.0 (22.0, 27.0) | 23.0 (20.0, 26.0) | <0.001 | 24.0 (22.0, 27.0) | 23.0 (21.0, 26.0) | <0.001 |
| Glucose (mg/dL) | 122.0 (102.0, 152.0) | 122.0 (100.0, 158.0) | 0.418 | 122.0 (102.0, 152.0) | 124.0 (102.0, 161.0) | 0.028 |
| Creatinine (mg/dL) | 0.9 (0.7, 1.4) | 0.9 (0.7, 1.4) | 0.562 | 0.9 (0.7, 1.4) | 0.9 (0.7, 1.4) | 0.153 |
| Urea nitrogen (mg/dL) | 18.0 (12.0, 29.0) | 18.0 (12.0, 29.0) | 0.093 | 18.0 (12.0, 29.0) | 19.0 (12.0, 31.0) | 0.021 |
| Medications | ||||||
| Other antidepressantsc | 596 (26.7) | 1161 (8.1) | <0.001 | 572 (26.0) | 583 (26.5) | 0.732 |
| Antipsychotics | 277 (12.4) | 524 (3.6) | <0.001 | 257 (11.7) | 221 (10.0) | 0.090 |
| Z drugs | 85 (3.8) | 136 (0.9) | <0.001 | 74 (3.4) | 65 (3.0) | 0.490 |
| Outcomes | ||||||
| In-hospital mortality | 189 (8.5) | 1089 (7.6) | 0.155 | 185 (8.4) | 179 (8.1) | 0.784 |
| Length of hospital stay (days) | 6.1 (4.1, 9.8) | 6.1 (3.7, 10.6) | 0.319 | 6.1 (4.1, 9.8) | 6.3 (3.9, 10.9) | 0.279 |
CRRT continuous renal replacement therapy, ICU intensive care unit, IQR interquartile range, SAPS Simplified Acute Physiology Score, SOFA Sequential Organ Failure Assessment, SSRI selective serotonin reuptake inhibitor.
aOther admission type included elective, observation, and surgical same day admission.
bOther first care unit included cardiovascular, trauma, and neurology ICU.
cOther antidepressants included monoamine oxidase inhibitors, tricyclics and tetracyclics, selective serotonin reuptake inhibitors, serotonin‐noradrenaline reuptake inhibitors, serotonin partial agonist and reuptake inhibitors, noradrenergic and specific serotoninergic antidepressants, and noradrenaline and dopamine reuptake inhibitors.
Multivariate analyses
The VIF of each variable was <5, demonstrating the absence of multicollinearity (Supplementary Table 5). Schoenfeld residual plots identify time-varying variables in the original cohort (Supplementary Fig. 3). In the original cohort, multivariable adjustment analysis indicated that pre-ICU use of SSRIs was associated with a 24% increase in the hazard for in-hospital mortality (aHR, 1.24; 95% CI, 1.05–1.46; P = 0.010; Table 2).
Table 2.
Risk of in-hospital mortality associated with pre-ICU use of SSRIs in critically ill adults with mental disorders.
| Multivariable analyses | HR | 95% CI | P-value |
|---|---|---|---|
| Original cohorta | 1.24 | 1.05–1.46 | 0.010 |
| Propensity score-matched cohorta | 1.26 | 1.02–1.57 | 0.032 |
| MSCM in the weighted cohortb | 1.43 | 1.32–1.54 | <0.001 |
CI confidence interval, HR hazard ratio, MSCM marginal structural Cox model, SSRI selective serotonin reuptake inhibitor.
aAdjusted variables included age, sex, admission type, first care unit, Charlson comorbidity index, depression, sepsis, delirium, coma, SAPS II, SOFA, antibiotic, vasopressors, invasive mechanical ventilation, other antidepressants, antipsychotics, and Z drugs.
bAdjusted variables included age, sex, admission type, first care unit, Charlson comorbidity index, depression, sepsis, delirium, coma, SAPS II, daily SOFA, antibiotic, vasopressors, invasive mechanical ventilation, other antidepressants, antipsychotics, and Z drugs.
Propensity score analysis and marginal structural Cox model
Schoenfeld residual plots identify time-varying variables in the matched cohort (Supplementary Fig. 4). Figure 2 illustrates the extent to which the individual variables in the XGBoost-based IPTW model contribute to the final propensity score. The balance diagnostics for PSM and IPTW were provided in Supplementary Figs. 5–7. In the matched cohort, pre-ICU SSRIs users had a higher risk of in-hospital mortality than non-users (aHR, 1.26; 95% CI, 1.02–1.57; P = 0.032; Table 2). In the weighted cohort, MSCM showed that pre-ICU SSRIs use was associated with increased risk of in-hospital mortality (aHR, 1.43; 95% CI, 1.32–1.54; P < 0.001; Table 2).
Fig. 2. Relative influence factor of covariates.
CRRT continuous renal replacement therapy, SAPS Simplified Acute Physiology Score, SOFA Sequential Organ Failure Assessment.
Subgroup analyses
The effect of pre-ICU use of SSRIs on in-hospital mortality varies between different subgroups (Fig. 3). The association was observed only in patients aged ≥65 years (aHR, 1.37; 95% CI, 1.12–1.69; P = 0.002), female (aHR, 1.33; 95% CI, 1.07–1.64; P = 0.009), admitted into medical/surgical ICU (aHR, 1.45; 95% CI, 1.20–1.74; P < 0.001), and with the diagnosis of depression (aHR, 1.28; 95% CI, 1.05–1.54; P = 0.012).
Fig. 3. Forest plot for subgroup analyses in the original cohort.
CI confidence interval, HR hazard ratio, ICU intensive care unit.
Further analyses
The results of the mediation analysis are shown in Supplementary Table 6. No significant mediation effect was found for either delirium (indirect effect: 0.002, P = 0.154) or coma (indirect effect: 0.002, P = 0.062). The class effect analysis showed that the increased in-hospital mortality was observed only in fluoxetine (aHR, 1.56; 95% CI, 1.12–2.17; P = 0.008; Supplementary Fig. 8). Compared to patients with stopping SSRIs, those with continuing SSRIs had higher in-hospital mortality (aHR, 2.18; 95% CI, 1.59–3.02; P < 0.001; Supplementary Table 7).
Discussion
Main findings
In this retrospective cohort study, we found that pre-ICU use of SSRIs was associated with an increase in the hazard for in-hospital mortality in critically ill adults with mental disorders. The results were stable in the matched and weighted cohorts.
Relation with previous evidence
Our study expands on current knowledge regarding the association between the pre-ICU use of SSRIs and mortality in critically ill patients. In a previous retrospective cohort study published in 2014, Ghassemi et al. used data from the MIMIC-II database and found that pre-ICU use of SSRIs was associated with increased in-hospital mortality in critically ill patients (adjusted odds ratio, 1.19; 95% CI, 1.02–1.40) [9]. In line with the previous study, our study found an increased risk of in-hospital mortality associated with pre-ICU use of SSRIs in critically ill patients. However, differences between this study and the previous study should be noted. First, we used the latest MIMIC-IV version 2.0, with more detailed medical records than MIMIC-II. Second, our study is the first to use PSM and IPTW analyses to ensure the robustness of the results. Third, we performed the mediation analysis to determine whether delirium or coma was mediator of the association between pre-ICU use of SSRIs and mortality. Fourth, we performed subgroup analyses and the class effect analysis. Fifth, we further analyzed the effects of continuing versus stopping SSRIs on mortality.
Possible explanations for findings
Why are critically ill patients with mental disorders receiving pre-ICU use of SSRIs at increased risk of mortality? The underlying mechanisms are far from being fully elucidated and require further exploration. Several potential explanations have been posited. First, SSRIs exert immunosuppressive effects by inhibiting proliferation and enhancing apoptosis of T cells. Gobin et al. observed that SSRIs induce an anti-proliferative effect by reducing T cell proliferation [31]. Besides, SSRIs have been shown to induce T cell apoptosis and decrease activated T cell viability [32]. These immune suppressive changes may increase the susceptibility of critically ill patients to secondary infections and high mortality. In a multicenter study on ICU patients receiving mechanical ventilation, the incidence of multidrug-resistant bacteria and mortality was significantly higher among immunocompromised patients than among non-immunocompromised patients [33]. Another study also found that immunosuppression status was independently associated with mortality in ICU patients with de novo acute hypoxemic respiratory failure treated with noninvasive ventilation [34]. Second, SSRIs may regulate mitochondrial machinery via one important mitochondria-related immune regulator (i.e., mitochondrial DNA) [35]. Circulating mitochondrial DNA has been associated with mortality in critically ill patients [36]. Third, the clinical investigation found that long-term use of SSRIs was associated with increased risks of cerebrovascular disease, cardiovascular disease mortality, and all-cause mortality [37]. A review also summarized the evidence on the potential adverse cardiovascular effects of treatment with SSRIs in patients with geriatric depression [38].
Implications for clinical practice
Our findings have potential implications for clinical practice, particularly for critically ill patients with pre-existing mental health disorders. A previous systematic review demonstrated that ~19% of adults in ICU suffer from such conditions, making them a significant subgroup [39]. Among the medications prescribed for mental health disorders, SSRIs are commonly used, including for depression, anxiety, and posttraumatic stress disorders [8]. However, there is limited information on the association between the use of SSRIs and outcomes in critically ill patients. A review by Kelly et al. indicated that there might be excess morbidity associated with SSRIs in critically ill patients, but the evidence was primarily derived from low-quality case reports or series [40]. Our study found that pre-ICU use of SSRIs was associated with increased mortality in critically ill adults with mental disorders, highlighting the need to address these gaps in evidence. It is important to note that our findings do not imply that SSRIs should not be used in critically ill patients. Rather, our results may encourage prescribers to consider the potential risk of increased mortality in critically ill patients receiving SSRIs.
Our study provides novel insights into the association between pre-ICU use of SSRIs and mortality. However, we cannot conclude that there is a causal relationship between pre-ICU use of SSRIs and mortality. Even if such a link exists, it is unclear whether any harm is related to the chronic effects of pre-ICU use of SSRIs, SSRIs use during critical illness, or acute SSRIs withdrawal. It is no doubt of interest to critical care physicians that pre-ICU admission use of SSRIs is associated with increased mortality in critical illness; meanwhile, whether critically ill patients should continue or stop using SSRIs during ICU stay is an important question. A retrospective cohort study found that discontinuation of chronic use of SSRIs was associated with a higher risk of antidepressant discontinuation syndrome symptoms in critically ill adults [41]. A recent review summarized the literature on the reinitiation of preadmission neuropsychiatric medications in ICU patients, including SSRIs drugs [42]. La et al. found that early reinitiation of neuropsychiatric medications within 5 days after ICU admission was associated with lighter sedation and less delirium [43]. Conversely, Cucci et al. found no association between early reinitiation within 72 h and delirium after adjustment for baseline differences [44]. Unfortunately, these studies had no data on mortality. Our study showed that continuing SSRIs during ICU stay was associated with increased in-hospital mortality compared to stopping SSRIs, highlighting the importance of this information to help inform clinical decision-making. Overall, evidence on stopping or reinitiating SSRIs and outcomes in critically ill patients is still lacking. More high-quality studies are warranted to help physicians weigh the benefits and harms of SSRIs in critically ill patients.
Study limitations
Our study has several limitations. First, given that our study is observational, there may be inherent risks, including the possibility of residual and unmeasured confounding that cannot be entirely ruled out. While we attempted to address visible bias by utilizing the PSM and IPTW approach, unmeasured covariates could still be a source of bias. Second, since our retrospective cohort study only established a correlation between exposure and outcomes, further studies using a prospective design are required to establish causality and validate our findings [45]. Third, the use of ICD-9/10 codes to identify mental disorders raises significant reliability concerns, particularly since primary care physicians primarily diagnose mental disorders. As a result, many ICU patients in the MIMIC database may not have received their primary care, resulting in the possibility that the mental disorder codes and medications may not have been documented for these patients in the database. Furthermore, the severity of mental disorders and medication compliance at the time of ICU admission remains unknown. These factors could potentially affect mortality and bias our results. Fourth, delirium/coma may change over time with the severity of the illness. We analyzed the occurrence of delirium/coma during the entire ICU stay rather than the daily status or ICU days with delirium/coma, potentially impacting the results. Fifth, we did not consider pre-ICU mental disorder severity, medication compliance, and the dose/duration of individual SSRIs used in the ICU, and further research is necessary to confirm any association. Lastly, our study only included patients from a single center, which limits its generalizability to other ICUs with varying practices or resources.
Conclusions
In summary, this retrospective cohort study suggested that pre-ICU use of SSRIs is associated with an increase in the hazard for in-hospital mortality in critically ill adults with mental disorders.
Supplementary information
Author contributions
W-JG, L-MZ, H-YY, and JL contributed to the conception and design of the work. All authors contributed to the acquisition, analysis, and interpretation of data for the work. W-JG and L-MZ contributed to drafting the work. All authors contributed to revising it critically for important intellectual content. All authors contributed to the final approval of the version to be published. All authors agreed to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.
Funding
This study received financial support from the National Natural Science Foundation of China (grant number 82072232), the Science and Technology Program of Guangzhou, China (grant number 202201020028), the Special Projects in Key Areas of General Colleges and Universities in Guangdong Province (grant number 2022ZDZX2003), the 2021 Annual Medical Teaching and Education Management Reform Research Project of Jinan University (grant number 2021YXJG029), the Clinical Frontier Technology Program of the First Affiliated Hospital of Jinan University, China (grant number JNU1AF-CFTP-2022-a01235), and the Science and Technology Projects in Guangzhou, China (grant number 202201020054 and 2023A03J1032).
Data availability
The datasets used and/or analyzed during the current study are available from the corresponding author upon reasonable request.
Code availability
All codes are available from the corresponding author upon reasonable request.
Competing interests
The authors declare no competing interests.
Ethics approval and consent to participate
The study is based on the latest MIMIC-IV version 2.0 database. It was approved by the institutional review boards of the Massachusetts Institute of Technology and Beth Israel Deaconess Medical Center without a requirement for individual patient informed consent because data were deidentified and publicly available.
Footnotes
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
These authors contributed equally: Wan-Jie Gu, Lu-Ming Zhang.
Contributor Information
Hai-Yan Yin, Email: yinhaiyan1867@126.com.
Jun Lyu, Email: lyujun2020@jnu.edu.cn.
Supplementary information
The online version contains supplementary material available at 10.1038/s41398-023-02487-2.
References
- 1.GBD 2019 Mental Disorders Collaborators. Global, regional, and national burden of 12 mental disorders in 204 countries and territories, 1990-2019: a systematic analysis for the Global Burden of Disease Study 2019. Lancet Psychiatry. 2022;9:137–50. [DOI] [PMC free article] [PubMed]
- 2.Wunsch H, Christiansen CF, Johansen MB, Olsen M, Ali N, Angus DC, et al. Psychiatric diagnoses and psychoactive medication use among nonsurgical critically ill patients receiving mechanical ventilation. JAMA. 2014;311:1133–42. doi: 10.1001/jama.2014.2137. [DOI] [PubMed] [Google Scholar]
- 3.Olafson K, Marrie RA, Bolton JM, Bernstein CN, Bienvenu OJ, Kredentser MS, et al. The 5-year pre- and post-hospitalization treated prevalence of mental disorders and psychotropic medication use in critically ill patients: a Canadian population-based study. Intensive Care Med. 2021;47:1450–61. doi: 10.1007/s00134-021-06513-z. [DOI] [PubMed] [Google Scholar]
- 4.Geense WW, Zegers M, Peters MAA, Ewalds E, Simons KS, Vermeulen H, et al. New physical, mental, and cognitive problems 1 year after ICU admission: a prospective multicenter study. Am J Respir Crit Care Med. 2021;203:1512–21. doi: 10.1164/rccm.202009-3381OC. [DOI] [PubMed] [Google Scholar]
- 5.Oh TK, Park HY, Song IA. Depression and long-term survival among South Korean sepsis survivors: a nationwide cohort study from 2011 to 2014. Crit Care Med. 2021;49:1470–80. doi: 10.1097/CCM.0000000000005030. [DOI] [PubMed] [Google Scholar]
- 6.Mazeraud A, Polito A, Sivanandamoorthy S, Porcher R, Heming N, Stoclin A, et al. Association between anxiety and new organ failure, independently of critical illness severity and respiratory status: a prospective multicentric cohort study. Crit Care Med. 2020;48:1471–9. doi: 10.1097/CCM.0000000000004495. [DOI] [PubMed] [Google Scholar]
- 7.Teixeira C, Rosa RG, Sganzerla D, Sanchez EC, Robinson CC, Dietrich C, et al. The burden of mental illness among survivors of critical care-risk factors and impact on quality of life: a multicenter pospective cohort study. Chest. 2021;160:157–64. doi: 10.1016/j.chest.2021.02.034. [DOI] [PubMed] [Google Scholar]
- 8.Pilling S, Whittington C, Taylor C, Kendrick T, Guideline Development Group. Identification and care pathways for common mental health disorders: summary of NICE guidance. BMJ. 2011;342:d2868. [DOI] [PubMed]
- 9.Ghassemi M, Marshall J, Singh N, Stone DJ, Celi LA. Leveraging a critical care database: selective serotonin reuptake inhibitor use prior to ICU admission is associated with increased hospital mortality. Chest. 2014;145:745–52. doi: 10.1378/chest.13-1722. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Würtz M, Schmidt M, Grove EL, Horváth-Puhó E, Christiansen CF, Sørensen HT. Preadmission use of selective serotonin reuptake inhibitors and short-term mortality in diabetic patients hospitalized due to stroke. J Intern Med. 2016;280:407–18. doi: 10.1111/joim.12512. [DOI] [PubMed] [Google Scholar]
- 11.Vozoris NT, Wang X, Austin PC, Stephenson AL, O’Donnell DE, Gershon AS, et al. Serotonergic antidepressant use and morbidity and mortality among older adults with COPD. Eur Respir J. 2018;52:1800475. doi: 10.1183/13993003.00475-2018. [DOI] [PubMed] [Google Scholar]
- 12.Fiest KM, Soo A, Hee Lee C, Niven DJ, Ely EW, Doig CJ, et al. Long-term outcomes in ICU patients with delirium: a population-based cohort study. Am J Respir Crit Care Med. 2021;204:412–20. doi: 10.1164/rccm.202002-0320OC. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Klein Klouwenberg PM, Zaal IJ, Spitoni C, Ong DS, van der Kooi AW, Bonten MJ, et al. The attributable mortality of delirium in critically ill patients: prospective cohort study. BMJ. 2014;349:g6652. doi: 10.1136/bmj.g6652. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Johnson A, Bulgarelli L, Pollard T, Horng S, Celi LA, Mark R MIMIC-IV (version 2.0). PhysioNet. 10.13026/7vcr-e114 (2022).
- 15.von Elm E, Altman DG, Egger M, Pocock SJ, Gøtzsche PC, Vandenbroucke JP, et al. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. PLoS Med. 2007;4:e296. doi: 10.1371/journal.pmed.0040296. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Ely EW, Margolin R, Francis J, May L, Truman B, Dittus R, et al. Evaluation of delirium in critically ill patients: validation of the Confusion Assessment Method for the Intensive Care Unit (CAM-ICU) Crit Care Med. 2001;29:1370–9. doi: 10.1097/00003246-200107000-00012. [DOI] [PubMed] [Google Scholar]
- 17.Sessler CN, Gosnell MS, Grap MJ, Brophy GM, O’Neal PV, Keane KA, et al. The Richmond Agitation-Sedation Scale: validity and reliability in adult intensive care unit patients. Am J Respir Crit Care Med. 2002;166:1338–44. doi: 10.1164/rccm.2107138. [DOI] [PubMed] [Google Scholar]
- 18.Zhang Z, Reinikainen J, Adeleke KA, Pieterse ME, Groothuis-Oudshoorn CGM. Time-varying covariates and coefficients in Cox regression models. Ann Transl Med. 2018;6:121. doi: 10.21037/atm.2018.02.12. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Stavem K, Hoel H, Skjaker SA, Haagensen R. Charlson comorbidity index derived from chart review or administrative data: agreement and prediction of mortality in intensive care patients. Clin Epidemiol. 2017;9:311–20. doi: 10.2147/CLEP.S133624. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Wewalka M, Warszawska J, Strunz V, Kitzberger R, Holzinger U, Fuhrmann V, et al. Depression as an independent risk factor for mortality in critically ill patients. Psychosom Med. 2015;77:106–13. doi: 10.1097/PSY.0000000000000137. [DOI] [PubMed] [Google Scholar]
- 21.Vincent JL, Marshall JC, Namendys-Silva SA, François B, Martin-Loeches I, Lipman J, et al. Assessment of the worldwide burden of critical illness: the intensive care over nations (ICON) audit. Lancet Respir Med. 2014;2:380–6. doi: 10.1016/S2213-2600(14)70061-X. [DOI] [PubMed] [Google Scholar]
- 22.Vai B, Mazza MG, Delli Colli C, Foiselle M, Allen B, Benedetti F, et al. Mental disorders and risk of COVID-19-related mortality, hospitalisation, and intensive care unit admission: a systematic review and meta-analysis. Lancet Psychiatry. 2021;8:797–812. doi: 10.1016/S2215-0366(21)00232-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Deb S, Austin PC, Tu JV, Ko DT, Mazer CD, Kiss A, et al. A review of propensity-score methods and their use in cardiovascular research. Can J Cardiol. 2016;32:259–65. doi: 10.1016/j.cjca.2015.05.015. [DOI] [PubMed] [Google Scholar]
- 24.Robins JM, Hernán MA, Brumback B. Marginal structural models and causal inference in epidemiology. Epidemiology. 2000;11:550–60. doi: 10.1097/00001648-200009000-00011. [DOI] [PubMed] [Google Scholar]
- 25.Zhang Z, Zhu C, Mo L, Hong Y. Effectiveness of sodium bicarbonate infusion on mortality in septic patients with metabolic acidosis. Intensive Care Med. 2018;44:1888–95. doi: 10.1007/s00134-018-5379-2. [DOI] [PubMed] [Google Scholar]
- 26.McCaffrey DF, Griffin BA, Almirall D, Slaughter ME, Ramchand R, Burgette LF. A tutorial on propensity score estimation for multiple treatments using generalized boosted models. Stat Med. 2013;32:3388–414. doi: 10.1002/sim.5753. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Stuart EA, Lee BK, Leacy FP. Prognostic score-based balance measures can be a useful diagnostic for propensity score methods in comparative effectiveness research. J Clin Epidemiol. 2013;66:S84–90.e1. doi: 10.1016/j.jclinepi.2013.01.013. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Zhang Z, Kim HJ, Lonjon G, Zhu Y, written on behalf of AME Big-Data Clinical Trial Collaborative Group. Balance diagnostics after propensity score matching. Ann Transl Med. 2019;7:16. doi: 10.21037/atm.2018.12.10. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Austin PC. Balance diagnostics for comparing the distribution of baseline covariates between treatment groups in propensity-score matched samples. Stat Med. 2009;28:3083–107. doi: 10.1002/sim.3697. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Valeri L, Vanderweele TJ. Mediation analysis allowing for exposure-mediator interactions and causal interpretation: theoretical assumptions and implementation with SAS and SPSS macros. Psychol Methods. 2013;18:137–50. doi: 10.1037/a0031034. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Gobin V, Van Steendam K, Denys D, Deforce D. Selective serotonin reuptake inhibitors as a novel class of immunosuppressants. Int Immunopharmacol. 2014;20:148–56. doi: 10.1016/j.intimp.2014.02.030. [DOI] [PubMed] [Google Scholar]
- 32.Gobin V, Van Steendam K, Fevery S, Tilleman K, Billiau AD, Denys D, et al. Fluoxetine reduces murine graft-versus-host disease by induction of T cell immunosuppression. J Neuroimmune Pharm. 2013;8:934–43. doi: 10.1007/s11481-013-9463-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Moreau AS, Martin-Loeches I, Povoa P, Salluh J, Rodriguez A, Thille AW, et al. Impact of immunosuppression on incidence, aetiology and outcome of ventilator-associated lower respiratory tract infections. Eur Respir J. 2018;51:1701656. doi: 10.1183/13993003.01656-2017. [DOI] [PubMed] [Google Scholar]
- 34.Coudroy R, Pham T, Boissier F, Robert R, Frat JP, Thille AW. Is immunosuppression status a risk factor for noninvasive ventilation failure in patients with acute hypoxemic respiratory failure? a post hoc matched analysis. Ann Intensive Care. 2019;9:90. doi: 10.1186/s13613-019-0566-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Wang Y, McLean AS. The role of mitochondria in the immune rresponse in critical illness. Crit Care. 2022;26:80. doi: 10.1186/s13054-022-03908-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Harrington JS, Huh JW, Schenck EJ, Nakahira K, Siempos II, Choi AMK. Circulating mitochondrial DNA as predictor of mortality in critically ill patients: a systematic review of clinical studies. Chest. 2019;156:1120–36. doi: 10.1016/j.chest.2019.07.014. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Bansal N, Hudda M, Payne RA, Smith DJ, Kessler D, Wiles N. Antidepressant use and risk of adverse outcomes: population-based cohort study. BJPsych Open. 2022;8:e164. doi: 10.1192/bjo.2022.563. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Ungvari Z, Tarantini S, Yabluchanskiy A, Csiszar A. Potential adverse cardiovascular effects of treatment with fluoxetine and other selective serotonin reuptake inhibitors (SSRIs) in patients with geriatric depression: implications for atherogenesis and cerebromicrovascular dysregulation. Front Genet. 2019;10:898. doi: 10.3389/fgene.2019.00898. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Pilowsky JK, Elliott R, Roche MA. Pre-existing mental health disorders in patients admitted to the intensive care unit: a systematic review and meta-analysis of prevalence. J Adv Nurs. 2021;77:2214–27. doi: 10.1111/jan.14753. [DOI] [PubMed] [Google Scholar]
- 40.Kelly JM, Rubenfeld GD, Masson N, Min A, Adhikari NKJ. Using selective serotonin reuptake inhibitors and serotonin-norepinephrine reuptake inhibitors in critical care: a systematic review of the evidence for bbenefit or harm. Crit Care Med. 2017;45:e607–16. doi: 10.1097/CCM.0000000000002308. [DOI] [PubMed] [Google Scholar]
- 41.Bainum TB, Fike DS, Mechelay D, Haase KK. Effect of abrupt discontinuation of antidepressants in critically ill hospitalized adults. Pharmacotherapy. 2017;37:1231–40. doi: 10.1002/phar.1992. [DOI] [PubMed] [Google Scholar]
- 42.Cucci MD, Chester KW, Hamilton LA. Concise definitive review for reinitiation of antidepressants, antipsychotics, and gabapentinoids in ICU patients. Crit Care Med. 2022;50:665–73. doi: 10.1097/CCM.0000000000005415. [DOI] [PubMed] [Google Scholar]
- 43.La MK, Thompson Bastin ML, Gisewhite JT, Johnson CA, Flannery AH. Impact of restarting home neuropsychiatric medications on sedation outcomes in medical intensive care unit patients. J Crit Care. 2018;43:102–7. doi: 10.1016/j.jcrc.2017.07.046. [DOI] [PubMed] [Google Scholar]
- 44.Cucci MD, Cunningham BS, Patel JS, Shimer AT, Mofleh DI, Mullen CL. Impact of early reinitiation of neuropsychiatric medications on agitation and delirium in the intensive care unit: a retrospective study. Ann Pharmacother. 2021;55:15–24. doi: 10.1177/1060028020935589. [DOI] [PubMed] [Google Scholar]
- 45.Zhang Z, Jin P, Feng M, Yang J, Huang J, Chen L, et al. Causal inference with marginal structural modeling for longitudinal data in laparoscopic surgery: a technical note. Laparosc Endosc Robot Surg. 2022;5:146–52. [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The datasets used and/or analyzed during the current study are available from the corresponding author upon reasonable request.
All codes are available from the corresponding author upon reasonable request.