Introduction

Estimates suggest that over 300 million patients globally undergo surgical procedures annually, and this number continues to rise1,2. Despite considerable progress in surgical outcomes in recent decades, post-major surgery survival remains a significant global public health concern. This concern is particularly pronounced in patients subjected to significant abdominal surgery, placing them at a heightened risk of inadequate organ perfusion. The extended duration and intricate nature of major abdominal surgery often culminate in conditions such as hypovolemia, vasodilation, and diminished cardiac output. To counteract these challenges, the infusion of fluids and the administration of vasoactive drugs, such as dopamine, dobutamine, and norepinephrine, prove instrumental in sustaining hemodynamic stability and augmenting organ perfusion in the presence of compromised vascular tone or reduced cardiac output3.

The use and dosage of vasoactive drugs to sustain hemodynamics are often considered indicators of disease severity, with a prevailing belief that mortality rates increase when higher doses of these drugs are necessitated4,5,6. Vasoactive–inotropic score (VIS), first proposed in 20107, was initially used to quantify hemodynamics vasoactive and inotropic support after cardiac surgery in pediatric patients, and has since been extensively used in adult cardiac surgery and critically ill patients in intensive care unit (ICU)8,9,10,11,12.VIS is a quantifying system that quantifies the amount of inotropic and vasopressor agents, including dopamine, dobutamine, epinephrine, milrinone, vasopressin, and norepinephrine7. Typically, it is computed as the average inotropic support administered over a predefined period. Research findings consistently assert that an elevated VIS stands as an independent risk factor for adverse clinical outcomes in various scenarios, including pediatric and adult cardiac surgery8,9,10,13, as well as in septic patients11,12. Koponen et al.8 reported that VISmax provided good prediction accuracy for the adverse outcomes [area under the curve (AUC), 0.72; 95% confidence interval (CI) 0.69–0.75]. The VISmax may be a useful indicator for distinguishing adult patients at high mortality risk after cardiac surgery. In patients with sepsis, VISmax had a better prognostic value for 30-day mortality than (Sequential Organ Failure Assessment) SOFA score (AUC = 0.724; 95% CI 0.694–0.753)12, a common ICU scoring system. However, the connection between an elevated VIS and mortality in patients undergoing major abdominal surgery remains unelucidated. This study utilizes the medical information mart in intensive care-IV (MIMIC-IV) database to investigate the correlation between the maximum vasoactive-inotropic score (VISmax) and the prognosis of adult patients subjected to major abdominal surgery. We hypothesize that an elevated VISmax within the initial postoperative 24 h is associated with both short-term and long-term all-cause mortality in adult patients undergoing major abdominal surgery.

Methods

Data source

We conducted a retrospective cohort study utilizing data extracted from the MIMIC-IV database14. MIMIC-IV version 2.2 constitutes an electronic health record dataset, which is a an extensive and publicly accessible critical care medical database. It comprised information from over 50,000 patients admitted various ICUs at the Beth Israel Deaconess Medical Center (Boston, MA, USA) between 2008 and 2019. The MIMIC project was approved by both Beth Israel Deaconess Medical Center (2001-P-001699/15) and Massachusetts institute of technology (Approval ID: 10734458). Individual patient consent was waived because unidentified health information of patients was used. A database access (certification number 1564657) was secured by one of the authors (JH). The writing of the manuscript adheres strictly to the guidelines outlined in the strengthening the reporting of observational studies in epidemiology (STROBE) statement15.

Study population

We included all adult patients admitted to the ICU following major abdominal surgeries, such as colorectal, pancreatic, gastric, liver, and splenic resection. Among them, colon surgeries included left colon, right colon, transverse colon, ascending colon, descending colon, and sigmoid colon surgeries. The presence of open or endoscopic surgery, radical or palliative surgery, and oncological or non-oncological surgery were included in this study. Exclusion criteria comprised: (1) Patients lacking VIS data within 24 h post-ICU admission; (2) ICU stay duration less than 24 h. Additionally, cases with a VISmax exceeding 100 were excluded, indicating an abnormal increase. In instances of multiple admissions during the study period, only records from the initial hospitalization were considered. Eligible patients were categorized into five groups, utilizing the quintiles of VISmax as delineated by Koponen et al.8: 0–5, 6–15, 16–30, 31–45, and > 45.

Data extraction

Navicat premium 16 was employed in conjunction with structured query language (SQL) for data extraction. The SQL script codes used herein were sourced from the GitHub repository (https://github.com/MIT-LCP/mimic-iv). Patient characteristics, encompassing age, sex, race, body mass index (BMI), and the Charlson comorbidity index (CCI), were systematically collected. Information relating to comorbidities, including hypertension, congestive heart failure, myocardial infarction, chronic pulmonary disease, cerebrovascular disease, diabetes, severe renal disease, and malignant cancer, was extracted in accordance with the international classification of diseases coding systems. Records from the initial admission to the intensive care unit (ICU) were scrutinized, encompassing the severity of illness assessed through the simplified acute physiology score II (SAPS II) and SOFA score. Relevant laboratory tests, including lactate, base excess, white blood cell count, platelet count, PaCO2, PO2, glucose level, hemoglobin level, C-reactive protein, neutrophils, bicarbonate, potassium level, calcium level, chloride level, activated partial thromboplastin time (APTT), prothrombin time (PT), and international normalized ratio (INR), were also systematically extracted. Furthermore, treatments administered within the initial 24 h post-ICU admission, such as vasopressor usage and Continuous Renal Replacement Therapy (CRRT), were documented. The VISmax was computed according to the following formula, using the maximum dosing rates of vasopressors and inotropes within the initial 24 h post-ICU admission: VISmax = dopamine dose (µg/kg/min) + dobutamine dose (µg/kg/min) + 100 × epinephrine dose (µg/kg/min) + 10 × milrinone dose (µg/kg/min) + 10,000 × vasopressin dose (units/kg/min) + 100 × norepinephrine dose (µg/kg/min)7.

Outcomes measures

The primary outcomes of this study comprised short-term (30-day) and long-term (1-year) all-cause mortality. Secondary outcomes encompassed in-hospital mortality, as well as the lengths of both ICU stay and hospital stay. Adverse outcomes such as surgical site infections, sepsis, acute kidney injury (AKI), pneumonia, intestinal obstruction and postoperative hemorrhage within 7 days postoperatively were also extracted. We further performed subgroup analyses based on age (< 65 vs ≥ 65 yr), CCI (≤ 6 vs > 6), mechanical ventilation (yes vs no), malignant cancer patients (yes vs no), and surgery approach (open surgery vs laparoscopic surgery).

Statistical analysis

Given the retrospective nature of this analysis, no a priori statistical analysis plan was delineated, and the sample size was derived from the available data within the database. Descriptive statistics are presented as median [inter-quartile range (IQR)] or mean (standard deviation) for continuous variables and as n (%) for categorical variables. Group comparisons of patient and clinical characteristics were conducted using the Kruskal−Wallis test for continuous variables and the chi-square test or Fisher’s exact test for categorical variables. The missing rate for each variable is outlined in Supplementary Table S1. C-reactive protein and neutrophils were eventually eliminated due to data missing exceeding 30%. Cox proportional hazard models were used to estimate the association between VISmax and survival. Schoenfeld residual test and the significance of time-varying effects for each variable were used to evaluate the Cox proportional hazards model. Significance was defined by a P value of < 0.05. We performed two models to adjust for confounding factors. The confounding factors in Cox regression model were selected according to the results of univariate regression and clinical experience. Model 1 was adjusted only for age, gender, and BMI. Model 2 was further adjusted for SAPSII, SOFA, CRRT, mechanical ventilation, myocardial infarct, diabetes, malignant cancer, surgical type, surgical approach, lactate, creatinine, and INR. The cumulative incidence of 30-day and 1-year all-cause mortality in each group was analyzed using the Kaplan−Meier method, and differences were evaluated through log-rank tests. The relationships between VISmax and mortality were also explored using smooth curve fitting. Logistic regression models were executed to compute the odds ratio (OR) with a 95% CI for dichotomous secondary outcomes, while linear regression was employed to assess the association between VISmax and continuous secondary outcomes. Hazard ratios (HR) were calculated using the formula HR = eβi. The predictability of increased VIS regarding 30-day and 1-year mortality was evaluated through the area under the curve (AUC) of the receiver operating characteristics (ROC) curve. Analyses were conducted using R software (version 4.2.3; R Foundation for Statistical Computing, Vienna, Austria) and EmpowerStats (https://www.empowerstats.com) (version 4.1; X&Y Solutions, Inc., Boston, MA). A P-value < 0.05 (two-sided) was considered significant for all tests.

Ethics approval and consent to participate

The study was approved by the institutional review boards of the Massachusetts Institute of Technology and Beth Israel Deaconess Medical Center and was granted a waiver of informed consent.

Results

Patient characteristics

Figure 1 shows the process of patient selection. Initially, 10,748 records were identified. After excluding the unqualified records, 512 patients were included in the analysis. Among them, the VISmax groups were categorized as follows: ≤ 5 (n = 51), > 5−15 (n = 118), > 15−30 (n = 114), > 30−45 (n = 105), and > 45 (n = 124) (Fig. 1). Of the included 512 patients undergoing major abdominal surgery, 280 (54.7%) were male, with a median age of 67.5 years. Colorectal surgery accounted for the largest proportion among all the major abdominal surgeries included. Of 512 patients, 412 (80.5%) underwent open major abdominal surgeries. Patients who underwent open major abdominal surgery were more likely to receive vasoactive and inotropic drugs. Among the six drugs in VIS components, norepinephrine and epinephrine are the two drugs with the largest proportion. The baseline characteristics are shown in Table 1. Patients with higher VISmax exhibited more severe illness (as evidenced by higher SAPS II and SOFA scores), received more treatments (such as Continuous Renal Replacement Therapy [CRRT] and mechanical ventilation), and manifested a greater prevalence of underlying comorbidities, such as diabetes. Furthermore, compared to patients with VISmax ≤ 5, those with elevated VISmax demonstrated higher blood lactate levels, base excess, creatinine, and poorer coagulation function. Patients with elevated VISmax were also more prone to experiencing higher mortality and a prolonged ICU stay.

Figure 1
figure 1

Flow chart of patient selection. MIMIC-IV, medical information mart in intensive care-IV.

Table 1 Characteristics for patients following major abdominal surgery with different VISmax score.

Primary outcome

Short-term all-cause mortality (30-day mortality)

Figure 2 illustrates the correlation between VISmax and the risk of 30-day mortality. When considering VISmax as a continuous variable, a notably positive association with 30-day mortality was observed in multivariable-adjusted models. This association was further evaluated through multivariable Cox regression analysis, as delineated in Table 2. In the fully adjusted model, after adjusting for SAPSII, SOFA, CRRT, mechanical ventilation, myocardial infarct, diabetes, malignant cancer, surgical type, surgical approach, lactate, creatinine, and INR. An increase of one standard deviation (SD) in VISmax was linked to a 2% higher risk of 30-day mortality (HR 1.02, 95% CI 1.01, 1.03, P = 0.0037). In the evaluation of VISmax on a dichotomous scale, patients in the VISmax > 30−45 group and VISmax > 45 group exhibited a significantly elevated risk of 30-day mortality in comparison to those in the low VISmax group (≤ 5), with hazard ratios (HRs) of 3.96 (95% CI 1.26, 12.51, P = 0.02) and 3.73(95% CI 1.16, 12.02, P = 0.03), respectively (Table 2). The results of Schoenfeld residuals satisfied the proportional hazards assumption, and the DFBETA diagnostic plot indicated robust Cox analysis results (Supplementary Figure S1, S2).

Figure 2
figure 2

Smooth curve of the trend of the relationship between VISmax and 30-day mortality (A,B) and 1-year mortality (C,D). B and D model were adjusted for SAPSII, SOFA, CRRT, mechanical ventilation, myocardial infarct, diabetes, malignant cancer, surgical type, surgical approach, lactate, creatinine, and INR.

Table 2 Association of VISmax and mortality in patients after major abdominal surgery.

Long-term all-cause mortality (1-year mortality)

In a multivariate-adjusted model, a smoothed spline demonstrates a significant positive association between increases in VISmax and 1-year mortality (Fig. 2). In a multivariate Cox regression model, an SD increase in VISmax was associated with a 2% higher risk of 1-year mortality in the fully adjusted model (HR 1.02, 95% CI 1.01, 1.043, P = 0.0002). When VISmax was evaluated on a dichotomous scale, patients in the VISmax > 30–45 group, and VISmax > 45 group exhibited an increased risk of 1-year mortality, with HRs of 3.48 (95% CI 1.44, 8.42, P = 0.0055) and 2.76 (95% CI 1.09, 6.95, P = 0.0285), respectively (Table 2).

Figure 3 displays Kaplan−Meier survival curves for each VIS group. The three highest VISmax groups showed an increased mortality risk of up to 1 year. Compared with VISmax ≤ 5 group, significant difference was observed in VISmax > 30–45 group and VISmax > 45 group (log-rank test P < 0.003 and P = 0.0002, respectively), but there is no difference between the two groups (log-rank test: P = 0.12) (Supplementary Table S2).

Figure 3
figure 3

Survival curves for 5 VISmax group.

Subgroup analyses.

Table 3 shows the results of subgroup analyses for 30-day mortality and 1-year mortality. Regarding 30-day mortality, VISmax was associated with an increased risk in all subgroups. For 1-year mortality, the lower limit of the 95% CI was > 1.00 in all subgroups, indicating higher 1-year mortality with elevated VISmax, irrespective of baseline patient characteristics.

Table 3 Association of VISmax and 30-day and 1-year mortality in patients after major abdominal surgery.

ROC curve analysis

VISmax demonstrated a 30-day mortality prediction with an AUC of 0.69 (95% CI 0.64–0.75), exhibiting performance comparable to that of both SOFA (AUC, 0.65; 95% CI 0.60–0.69, Delong method with P = 0.13) and SAPS II (AUC, 0.65; 95% CI 0.60–0.69, Delong method with P = 0.84) (Fig. 4). This congruence in predictive capability was similarly observed for 1-year mortality. The ROC analysis for VISmax predicting 1-year mortality produced an AUC of 0.67 (95% CI 0.62–0.72), comparable to SOFA (AUC, 0.64; 95% CI 0.59–0.68, DeLong method with P = 0.18) and SAPS II (AUC, 0.71; 95% CI 0.66–0.75, DeLong method with P = 0.18) (Fig. 4). The optimal cutoff value for VISmax in predicting 30-day mortality and 1-year mortality was determined as 30.3 (sensitivity: 67.7%, specificity: 63.4%) and 15.9 (sensitivity: 81.2%, specificity: 41.7%), respectively.

Figure 4
figure 4

Receiver operating curves (ROC) of 30-day mortality and 1-year mortality on VISmax, SOFA, SAPS II. (A) ROC for 30-day mortality. (B) ROC for 1-year mortality.

Secondary outcomes

In-hospital mortality

In contrast to the low VISmax group (≤ 5), a heightened VISmax was linked with a progressively elevated risk of in-hospital mortality, as revealed by the multivariable logistic regression analysis. The odds ratio (OR) for the VISmax group > 30−45 was 3.59 (95% CI 1.14, 11.27, P = 0.03), while for the VISmax group > 45, the OR was 3.82 (95% CI 1.19, 12.32, P = 0.02) (refer to Supplementary Table S3).

Length of ICU and hospital stay

The linear regression analysis indicated that an elevated VISmax was correlated with a prolonged length of ICU stay, with a hazard ratio (HR) of 3.57 (95% CI 1.70, 7.37, P = 0.001). However, this association was not observed for the length of hospital stay, where the HR was 1.63 (95% CI 0.32, 8.19, P = 0.56).

Postoperative complications

Compared with VISmax < 5 group, the elevated VISmax group had a higher incidence of postoperative AKI (P = 0.023), sepsis (P < 0.001) and postoperative hemorrhage (P = 0.025).

Discussion

To our knowledge, this is the first to study to explore the correlation between the maximum VIS and mortality in patients following major abdominal surgery. In this retrospective investigation, patients were classified into five groups according to the quintile of VISmax. Our findings underscore that an elevated VISmax within the initial 24 h following ICU admission correlates with both short-term and long-term all-cause mortality in patients undergoing major abdominal surgery. This correlation persists across subgroup analyses. Furthermore, we noted that an elevated VISmax is linked with increased in-hospital mortality and an extended length of ICU stay.

Short-term mortality rates among patients undergoing major abdominal surgery vary widely, ranging from 1 to 20%16,17,18,19, with higher incidence rates observed among the elderly19. The mean age in our study cohort was 66.2 years, and the overall 30-day mortality was recorded at 24.2%. Notably, a significant proportion of patients necessitating ICU admission following major abdominal surgery and subsequent vasoactive drug administration fall within the elderly demographic. However, our findings indicate that age does not emerge as an independent risk factor for vasoactive drug utilization. This is exemplified by our observations across the five VISmax groups, where no discernible pattern has been observed associating higher scores with advanced age. The VIS, formulated through a simple formula to standardize dosages of diverse drugs and objectively quantify the degree of cardiovascular support, has witnessed growing adoption. Initially applied in pediatric and adult cardiac surgery contexts8,9,10,13, it subsequently became apparent that an elevated VIS was linked with heightened mortality in septic patients11,12. Our study broadens the clinical utility of VIS, illustrating that a high VISmax is associated with postoperative mortality in patients undergoing major abdominal surgery. This finding aligns with earlier investigations focused on adult cardiac surgery8,9,10. A previous study indicated that patients with a high VISmax level experienced prolonged hospital and ICU stays8,10. However, in our study, patients with higher VISmax values did not exhibit a significantly extended duration of hospital stay. This discrepancy may be attributed to the increased mortality associated with elevated VISmax values within the initial 24 h of ICU admission. In addition, variations in discharge criteria across different hospitals could contribute to the observed differences in length of stay.

This study has several implications for clinical practices. VISmax was an independent predictor of 30-day mortality in adult patients after major abdominal surgery. It is helpful for clinicians to stratify the high risk of postoperative death in these surgical patients. In addition, VISmax may be superior to traditional ICU scoring systems, such as SOFA scoring. Traditional SOFA scoring systems measure circulation dysfunction in critically ill patients and only roughly quantify the vasoactive drug support in the cardiovascular system. Given that cardiovascular dysfunction is one of the most common organ failures in the ICU20,21, a more elaborate assessment of circulatory function may be beneficial to improve predictive power. Several common vasopressors or inotropes used in clinical practice have also been quantified in VIS (e.g., vasopressin or dobutamine). Adding VIS to the cardiovascular component of future SOFA score may improve its circulatory dysfunction and mortality accuracy. Furthermore, considering the strong correlation between VISmax and mortality, future research may consider combining VIS and other clinically relevant factors to establish a predictive model for evaluating postoperative mortality in patients undergoing major abdominal surgery, helping clinicians distinguish high-risk patients for postoperative death early and take appropriate treatment.

While the VIS is increasingly integrated into clinical practice, certain unresolved issues merit attention. Our study, akin to previous investigations by Tohmea et al.10, KARA et al.22, and Koponen et al.8, adopted VISmax at postoperative 24 h admission to the ICU, affirming a robust association between elevated VISmax and postoperative mortality. However, a study by Davidson and co-workers23 prospectively assessed VIS 48–72 h after infants' cardiac surgery, revealing that VIS at 48 h exhibited a more sensitive prediction of prolonged mechanical ventilation than VISmax calculated within 48 h after surgery. They highlighted that the duration of vasopressor support served as a superior predictor of an unfavorable prognosis compared to the intensity of VIS. In a separate study, Crow and colleagues24 explored the relationship between postoperative 72 h VISmax and prognosis in cardiac surgery, establishing that VISmax within 48 h after surgery was correlated with prolonged ventilation time and length of hospital stay. Furthermore, the absence of a defined threshold for categorizing VIS as high or low is a current limitation. Different studies delineate patients with “high” VIS from those with "low" VIS, employing cutoff values ranging from 5.5 to 42.57,8,9,10,11,12,13,20,23,24. Given the pivotal role of intervention timing in achieving a clinically favorable outcome, future research should prioritize determining the optimal cutoff time for designating VIS as high.

This study is beset by several limitations. First, its retrospective observational design exposes the results to potential residual bias and unmeasured confounders, notwithstanding the application of multivariable analyses. Secondly, the study did not establish a cause-and-effect relationship. Next, the patients with secondary or multiple abdominal surgeries may have a higher 30-day mortality rate, and we did not exclude this part of the patients, which may have some impact on the outcomes. Finally, the reliance of the study on data from the MIMIC-IV database, limited to United States adults, may curtail the generalizability of our findings.

Conclusion

Elevated VISmax within the initial postoperative 24 h following ICU admission was associated with both short-term and long-term mortality in patients undergoing major abdominal surgery.