Evidence regarding the relationship between age-adjusted Charlson comorbidity index (ACCI) and in-hospital mortality is limited. Therefore, the present study investigated whether there was an independent association between ACCI and in-hospital mortality in critically ill patients with cardiogenic shock (CS) after adjusting for other covariates (age, sex, history of disease, scoring system, in-hospital management, vital signs at presentation, laboratory findings and vasopressors). ACCI, calculated retrospectively after hospitalization between 2008 and 2019, was derived from intensive care unit (ICU) admissions at the Beth Israel Deaconess Medical Center (Boston, MA, USA). Patients with CS were classified into two categories based on predefined ACCI scores (low, <8; high, ≥8). Based on baseline ACCI, the risk of in-hospital mortality in patients with CS was calculated using a multivariate Cox proportional risk model, and the threshold effect was calculated using a two-piece linear regression model. The in-hospital mortality rate was ~1.5 times greater in the ACCI high group compared with that in the ACCI low group [hazard ratio (HR)=1.45; 95% confidence interval (CI), 1.14
Cardiogenic shock (CS) is a low-cardiac-output clinical syndrome caused by severe impairment of myocardial performance and characterized by inadequate tissue perfusion and microcirculation disorders (
Mortality risk stratification should be routinely performed in patients with CS. Several predictive tools for critically ill patients are available, including the simplified acute physiology score, and acute physiology and chronic health evaluation score (
The age-adjusted Charlson Comorbidity Index (ACCI) is a weighted system based on the age of an individual and their chronic condition (
Data for this retrospective, single-center, observational cohort study were obtained from the Medical Information Mart for Intensive Care (MIMIC)-IV database (
After completing the collaborative institutional training initiative course and passing the ethics examination (certification no. 48693003), access to the database was approved by the review committee of Massachusetts Institute of Technology and Beth Israel Deaconess Medical Center, and access to the database was granted for download and use.
Ethical review and approval was not required for the study in accordance with the local legislation and institutional requirements.
Patients with CS between 2008 and 2019 were eligible for inclusion. The inclusion criteria were as follows: i) International Classification of Diseases (ICD)-9(
The exclusion criteria were as follows: i) Age <18 years; ii) intensive care unit (ICU) length of stay <24 h; and iii) incomplete or unobtainable ACCI data. Only the first admission was included in the analysis of patients admitted to the ICU multiple times.
Upon admission, the comorbidity score was calculated based on the clinical history of the patient. ACCI was generated based on comorbidity score and age of the patients. A total of 19 medical conditions were included in the comorbidity score, each rated from 1 to 6 points to calculate the index score (
During the first 24 h after admission to the ICU, data on baseline characteristics of the patients were collected, including demographics (age and sex), vital signs [heart rate and mean blood pressure (MBP)], laboratory findings [serum creatinine (Scr), white blood cell (WBC), hemoglobin and platelet counts], comorbidities [hypertension, chronic kidney disease (CKD), stroke, chronic pulmonary disease (COPD), dementia, paraplegia, peptic ulcer disease, diabetes, severe liver disease, malignant cancer and peripheral vascular disease (PVD)], and scores [ACCI and Oxford Acute Severity of Illness Score (OASIS)]. In-hospital management included mechanical ventilation (MV) and extracorporeal membrane oxygenation (ECMO). Etiologies of CS included AMI and acute heart failure (AHF). The use of vasoactive drugs, including dobutamine, norepinephrine and dopamine, was also recorded.
In-hospital mortality was the primary outcome.
The characteristics of patients were analyzed based on a predefined ACCI. Continuous variables are reported as means and standard deviations, and categorical variables are presented as percentages. Median values and interquartile ranges were calculated for parameters with a skewed distribution. Pearson's χ2, unpaired Student's t-test and Mann-Whitney U test were used to compare categorical variables, continuous variables with normal distribution and continuous variables with skewed distribution. The Kaplan-Meier and log-rank analyses were used to determine in-hospital survival curves. Multivariate Cox regression analysis was used to estimate the correlation between ACCI and in-hospital mortality. Following the Strengthening the Reporting of Observational Studies in Epidemiology statement (
Initially, patients who had an ICD-9 code of 78551 (cardiogenic shock) or 99801 (postoperative shock, cardiogenic), an ICD-10 code of R570 (cardiogenic shock) or T8111 (postprocedural cardiogenic shock) or T8111XA (postprocedural cardiogenic shock, initial encounter) or T8111XD (postprocedural cardiogenic shock, subsequent encounter) or T8111XS (postprocedural cardiogenic shock, sequela) were included. Postoperative cardiogenic shock is mostly caused by cardiac surgery and is more severe, which may affect the present conclusions. Additionally, in the current study, patients were included whenever cardiogenic shock was present, but there may be a large heterogeneity between patients with cardiogenic as the primary diagnosis or not, which may affect the conclusions. To test the robustness of the present findings, a sensitivity analysis was conducted wherein patients who presented postoperatively with CS or patients for whom CS was not indicated in the primary diagnosis were excluded.
A total of 2,547 patients with CS were identified from 71,532 MIMIC-IV admissions. The flowchart in
Baseline characteristics are presented in
In the index hospitalization, 495 patients (32.1%) died, and those with high ACCI were more likely to die during hospitalization (41.6 vs. 25.4%; P<0.001;
Univariate logistic regression analysis showed that variables including age, acute heart failure, chronic kidney disease, malignant cancer, Charlson comorbidity index, ACCI score ≥8, Oxford Acute Severity of Illness Score, hemoglobin, white blood cells, serum creatinine, mechanical ventilation, mean blood pressure, and use of norepinephrine and dopamine were associated with in-hospital mortality (
After adjustment for a series of covariates, ACCI and in-hospital mortality exhibited a non-linear dose-response relationship (
To further clarify the influence of surgery and diagnosis sequence on the results, 175 patients who received postprocedural CS and 476 patients with CS who showed diagnosis sequence number >5 were excluded from sensitivity analyses. After excluding these patients, the HR of in-hospital mortality generally increased in patients with high ACCI (
In the current retrospective cohort study, ACCI was independently associated with in-hospital mortality in critically ill patients with CS. Higher ACCI was associated with higher in-hospital mortality rate. This relationship persisted after adjusting for the appropriate variables and confounders. To the best of our knowledge, this is the first report examining the relationship between ACCI and in-hospital mortality in critically ill patients with CS. Non-equidistant trends were revealed in the effect values in different ACCI subgroups, suggesting a possible non-linear relationship between ACCI and in-hospital mortality in these patients. The effect of ACCI on in-hospital mortality rate depended on whether the ACCI score was >4.5 or <4.5; ACCI >4.5 was positively associated with in-hospital mortality, whereas ACCI <4.5 did not demonstrate a statistically significant association with in-hospital mortality, suggesting a threshold effect.
Studies have shown that patients admitted to the ICU with CS have a significantly higher mortality rate within 24 or 48 h of admission (
Evidence suggests that ACCI is a valid and widely used measure for predicting mortality risk. This tool can be used to predict clinical outcomes and to screen for sensitive conditions (
Several studies have observed the relationship between ACCI and in-hospital mortality in critically ill patients (
The present study had several limitations. First, as a single-center study and owing to the strict inclusion criteria, conclusions may be extrapolated only for patients with CS who were in the ICU for >24 h. Second, some undetected and uncontrolled confounding factors may have affected the conclusions, which is an inherent problem for all observational studies. Third, the current study could not find a plausible explanation for the lack of a correlation between ACCI and in-hospital mortality below the threshold; hence, further investigation of possible causes and mechanisms is required. Forth, the reason for patient mortality; whether this was expected; and if the patients were subjected to standard clinical practice were not provided in the database, so we cannot identify the exact cause of death. This has somewhat influenced the further search for ways to reduce mortality. However, after excluding some special populations from the sensitivity analysis, the present results were consistent with those in the whole population, indicating the robustness of the results.
In conclusion, a non-linear relationship between ACCI and mortality in critically ill patients with CS exists, with a significant increase in in-hospital mortality when the ACCI score exceeds 4.5.
The authors wish to thank Dr Liu Jie (People's Liberation Army of the China General Hospital, Beijing, China) and Dr Yang Qilin (The Second Affiliated Hospital of Guangzhou Medical University, Guangzhou, China) for helping with the revision of the manuscript.
The clinical data used to support the findings of this study were acquired from the Monitoring in Intensive Care Database IV version 2.0 (MIMIC-IV v.2.0). Although the database is publicly and freely available, users must complete the National Institutes of Health's web-based course known as Protecting Human Research Participants to apply for permission to access the database. The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.
All authors have contributed to this work and take public responsibility for appropriate portions of the content and agree 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 were appropriately investigated and resolved. DW was responsible for the design of this study. RC contributed to data acquisition and analysis. WW contributed to interpretation of the data. RC and WW confirm the authenticity of all the raw data. YS made substantial contributions to analysis of data. DW and YM made substantial contributions to interpretation of data. All authors have read and approved the final manuscript.
The use of the MIMIC-IV database was approved by the review committee of Massachusetts Institute of Technology and Beth Israel Deaconess Medical Center. Ethical review and approval was not required for the study in accordance with the local legislation and institutional requirements. And the data used in the present study were obtained from a public database, which do not contain protected health information, so individual patient consent was not needed. The protocol was performed in accordance with the Declaration of Helsinki.
Not applicable.
The authors declare that they have no competing interests.
Flowchart of patient selection. CS, cardiogenic shock; MIMIC-IV, medical information mart for intensive care-IV; ICU, intensive care unit; ACCI, age-adjusted Charlson comorbidity index.
Kaplan-Meier survival curves showing in-hospital mortality rate according to ACCI in patients with CS. ACCI, age-adjusted Charlson comorbidity index; CS, cardiogenic shock.
Non-linear dose-response relationship between ACCI and in-hospital mortality. The black line and gray area represent the estimated values and their corresponding 95% confidence intervals, respectively. HRs are adjusted for age, sex, hypertension, chronic kidney disease, stroke, chronic pulmonary disease, dementia, paraplegia, peptic ulcer disease, diabetes, severe liver disease, malignant cancer, peripheral vascular disease, acute myocardial infarction, acute heart failure, heart rate, mean blood pressure, hemoglobin, platelets, white blood cell, serum creatinine, Oxford Acute Severity of Illness Score, mechanical ventilation, extracorporeal membrane oxygenation, dobutamine, norepinephrine and dopamine. ACCI, age-adjusted Charlson comorbidity index; HR, hazard ratio.
Baseline characteristics of the study participants.
ACCI | All patients | Low ACCI (<8) | High ACCI (≥8) | P-value |
---|---|---|---|---|
Number, n | 1544 | 907 | 637 | |
Sex, n (%) | 0.340 | |||
Male | 920 (59.6) | 550 (60.6) | 370 (58.1) | |
Female | 624 (40.4) | 357 (39.4) | 267 (41.9) | |
Age, years | 70.2±14.5 | 65.8±15.5 | 76.4±9.9 | <0.001 |
Etiology | ||||
AMI, n (%) | 562 (36.4) | 307 (33.8) | 255 (40.0) | 0.015 |
AHF, n (%) | 988 (64.0) | 519 (57.2) | 469 (73.6) | <0.001 |
History of disease | ||||
Hypertension, n (%) | 326 (21.1) | 222 (24.5) | 104 (16.3) | <0.001 |
CKD, n (%) | 197 (12.8) | 37 (4.1) | 160 (25.1) | <0.001 |
Stroke, n (%) | 76 (4.9) | 37 (4.1) | 39 (6.1) | 0.088 |
PVD, n (%) | 294 (19.0) | 103 (11.4) | 191 (30.0) | <0.001 |
Dementia, n (%) | 48 (3.1) | 15 (1.7) | 33 (5.2) | <0.001 |
COPD, n (%) | 473 (30.6) | 229 (25.2) | 244 (38.3) | <0.001 |
Peptic ulcer, n (%) | 35 (2.3) | 11 (1.2) | 24 (3.8) | 0.002 |
Diabetes, n (%) | 207 (13.4) | 25 (2.8) | 182 (28.6) | <0.001 |
Paraplegia, n (%) | 44 (2.8) | 8 (0.9) | 36 (5.7) | <0.001 |
Malignant cancer, n (%) | 114 (7.4) | 19 (2.1) | 95 (14.9) | <0.001 |
Severe liver disease, n (%) | 46 (3.0) | 11 (1.2) | 35 (5.5) | <0.001 |
Scoring system | ||||
ACCI | 7.0±2.7 | 5.2±1.6 | 9.5±1.6 | <0.001 |
OASIS | 37.9±10.1 | 37.2±10.1 | 38.9±10.0 | <0.001 |
In-hospital management | ||||
MV, n (%) | 487 (31.5) | 298 (32.9) | 189 (29.7) | 0.204 |
ECMO, n (%) | 48 (3.1) | 35 (3.9) | 13 (2.0) | 0.060 |
Vital signs at presentation | ||||
HR (beats/minute) | 88.3±17.9 | 89.1±18.3 | 87.3±17.2 | 0.062 |
MBP (mmHg) | 75.1±9.4 | 76.4±9.6 | 73.2±8.9 | <0.001 |
Laboratory findings | ||||
Hemoglobin (g/dl) | 12.0±2.3 | 12.5±2.3 | 11.2±2.0 | <0.001 |
Platelet (K/µl) | 166.0 (115.0, 228.0) | 167.0 (118.0, 230.0) | 165.0 (114.0, 226.2) | 0.637 |
WBC (109/l) | 14.9 (11.0, 20.0) | 15.8 (11.7, 20.9) | 13.7 (10.5, 19.0) | <0.001 |
Scr (mg/dl) | 1.6 (1.1, 2.4) | 1.3 (1.0, 2.0) | 2.1 (1.5, 3.1) | <0.001 |
Vasopressors, n (%) | ||||
Dobutamine, n (%) | 374 (24.2) | 197 (21.7) | 177 (27.8) | 0.007 |
Norepinephrine, n (%) | 986 (63.9) | 575 (63.4) | 411 (64.5) | 0.69 |
Dopamine, n (%) | 351 (22.7) | 202 (22.3) | 149 (23.4) | 0.649 |
Outcomes | ||||
In-hospital mortality, n (%) | 495 (32.1) | 230 (25.4) | 265 (41.6) | <0.001 |
Los hospital (days) | 13.9±12.6 | 13.6±12.7 | 14.3±12.5 | 0.254 |
Los ICU (days) | 7.3±7.9 | 7.2±8.1 | 7.3±7.7 | 0.803 |
AMI, acute myocardial infarction; AHF, acute heart failure; PVD, peripheral vascular disease; CKD, chronic kidney disease; COPD, chronic pulmonary disease; ACCI, age-adjusted Charlson comorbidity index; OASIS, Oxford Acute Severity of Illness Score; HR, heart rate; WBC, white blood cell; Scr, serum creatinine; MV, mechanical ventilation; ECMO, extracorporeal membrane oxygenation; MBP, mean arterial blood pressure; Los hospital, length of hospital stay time; Los ICU, length of intensive care unit stay time.
Association between ACCI scores and in-hospital mortality in multiple regression model.
Model I | Model II | Model III | Model IV | |||||
---|---|---|---|---|---|---|---|---|
Variable | HR (95% CI) | P-value | HR (95% CI) | P-value | HR (95% CI) | P-value | HR (95% CI) | P-value |
ACCI | 1.10 (1.06-1.13) | <0.001 | 1.05 (1.01-1.10) | 0.007 | 1.12 (1.06-1.18) | <0.001 | 1.09 (1.03-1.16) | 0.004 |
ACCI <8 | 1.00 (Ref) | - | 1.00 (Ref) | - | 1.00 (Ref) | - | 1.00 (Ref) | - |
ACCI ≥8 | 1.56 (1.31-1.86) | <0.001 | 1.30 (1.08-1.57) | 0.007 | 1.64 (1.29-2.09) | <0.001 | 1.45 (1.14-1.86) | 0.003 |
Model I: Did not adjust for confounders. Model II: Adjusted for age + sex. Model III: Model II + hypertension, chronic kidney disease, stroke, chronic pulmonary disease, dementia, paraplegia, peptic ulcer disease, diabetes, severe liver disease, malignant cancer, peripheral vascular disease, acute myocardial infarction, acute heart failure. Model IV: Model III + HR, mean blood pressure, hemoglobin, platelets, white blood cell, serum creatinine, Oxford Acute Severity of Illness Score, mechanical ventilation, extracorporeal membrane oxygenation, dobutamine, norepinephrine, dopamine. HR, hazard Ratio; CI, Confidence interval; ACCI, age-adjusted Charlson comorbidity.
Threshold effect analysis of age-adjusted Charlson comorbidity on in-hospital mortality.
Threshold of ACCI | HR | 95% confidence intervals | P |
---|---|---|---|
<4.5 | 0.717 | 0.458,1.123 | 0.147 |
≥4.5 | 1.122 | 1.054,1.194 | <0.001 |
Log-likelihood ratio test | <0.001 |
HRs were adjusted for age, sex, hypertension, chronic kidney disease, stroke, chronic pulmonary disease, dementia, paraplegia, peptic ulcer disease, diabetes, severe liver disease, malignant cancer, peripheral vascular disease, acute myocardial infarction, acute heart failure, HR, mean blood pressure, hemoglobin, platelets, white blood cell, serum creatinine, Oxford Acute Severity of Illness Score, mechanical ventilation, extracorporeal membrane oxygenation, dobutamine, norepinephrine, dopamine. HR, hazard ratio.