Introduction

Globally, life expectancy is increasing, and elderly adults comprise a growing proportion of the population1. In 2022, the world has 771 million people aged 65 or older, three times more than the population in 1980 (258 million). The number of elderly people is estimated to hit 994 million by 2030 and 1.6 billion by 2050. United Nations Department of Economic and Social Affairs, Population Division (2022)2. Sepsis is a worldwide complication of infectious processes associated with substantial morbidity and mortality. A worldwide estimate of 48.9 million cases of sepsis occurred in 2017, with 11 million deaths3. In mainland China, sepsis affects one in five patients admitted to intensive care units, with a 90-day mortality of 35%4. Based on National Health data from England, it was estimated that 77.5% of these deaths associated with sepsis occurred in individuals over the age of 755. Pre-existing heart failure significantly increases mortality in patients with sepsis by about 33%.

Dopamine, norepinephrine, and vasopressin are the three most popular vasoactive drugs used clinically for the treatment of various types of shock. Current studies of vasoactive drugs are mostly focused on cardiogenic shock or septic shock alone and based on the adult population. Little attention has been paid to the protocols for utilizing vasoactive drugs in elderly sepsis patients with pre-existing heart failure. Thus, we performed a retrospective study from a large database to investigate the optimal vasoactive drug regimen in elderly septic patients with combined heart failure.

Materials and methods

Data source

Data for this study were obtained from the public database Medical Information Mart for Intensive Care (MIMIC III) (https://mimic.mit.edu) 6. The version 1.4 MIMIC-III database, maintained by the Massachusetts Institute of Technology Laboratory for Computational Physiology, contains data on patients hospitalized in an ICU at Beth Israel Deaconess Medical Center from 2001 to 2012. One of our authors, ZBH, who was responsible for data extraction, obtained free accessibility to this database after passing the examination of the National Institutes of Health (NIH) online course and gaining the certification (certification No. 36300529). Because the MIMIC-III database is a kind of publicly, available anonymized database, ethical approval was not required.

Study population

For patients readmitted, only the first hospitalization was retained.

Inclusion criteria were as follows: (1) age of 65 years or older; (2) ICU length of stay(LOS) longer than 24 h; (3) diagnosis of sepsis and heart failure; (4) use of norepinephrine, dopamine, or norepinephrine combined with vasopressin boost as a blood pressure maintenance regimen.

Exclusion criteria included the following: (1) age less than 65 years; (2) ICU LOS less than 24 h; (3) no sepsis and no heart failure; (4) no use of norepinephrine, vasopressin, or dopamine; (5) use of a combination regimen of vasoactive drugs other than norepinephrine combined with vasopressin, such as norepinephrine combined with dopamine or vasopressin combined with dopamine.

Data extraction and management

Heart failure was identified by International Classification of Diseases, Ninth version (ICD-9) codes: 4280, 4281, 4289, 39891, 40201, 40211, 40291, 40401, 40403, 40411, 40491, 40493, 42820, 42821, 42822, 42823, 42830, 42831, 42832, 42933, 42840, 42841, 42842, and 42843. Systolic heart failure was defined using ICD-9 codes: 42820, 42821, 42822, 42823. Diastolic heart failure was defined using ICD-9 codes:42830, 424831, 42832, 42833.

Sepsis was classified according to the criteria of the 2016 Third International Consensus Definitions for Sepsis and Septic Shock (Sepsis-3)7.

All-cause mortality was the primary outcome, including 7-day mortality, 28-day mortality, and 90-day mortality). Several secondary outcome indicators were included: Hospital LOS, ICU LOS, Mechanical Ventilation (MV) incidence, Acute Kidney Injury (AKI) incidence, increased heart rate, new-onset arrhythmia, new-onset malignant arrhythmia, etc. AKI was defined by the Kidney Disease: Improving Global Outcomes (KDIGO) criteria8. Increased heart rate was defined as the gap between the peak HR and the baseline HR extracted from ECG or ECG monitoring. The term "new-onset malignant arrhythmia" is defined as ventricular tachycardia, ventricular fibrillation, and ventricular cardiac arrest. The Vasoactive-Inotropic Score (VIS), a weighted sum of various vasopressors and inotropes, is known to be an independent predictor of adverse outcomes including ventilator days, intensive care unit length of stay, and mortality9.

The VIS score is therefore calculated as dopamine dose (μg kg−1 min−1) + dobutamine dose (μg kg−1 min−1) + 100 × epinephrine dose (μg kg−1 min−1) + 100 × norepinephrine dose (μg kg−1 min−1) + 10,000 × vasopressin dose (U kg−1 min−1) + 10 × milrinone dose (μg kg−1 min−1).

Baseline data were obtained based on the first data within 24 h of ICU admission. If missing, we use the last data before admission to ICU instead.

All scripts used for demographic characterization, clinical scores, and comorbidity were obtained from the GitHub website (https://github.com/MIT-LCP/mimic-code). Data extraction was performed with PostgreSQL tools (v10.0; PostgreSQL Global Development Group) using SQL.

Statistical analysis

As a general rule, continuous variables are reported as medians and interquartile ranges, whereas categorical variables are reported as percentages. Non-parametric data were examined using the Wilcoxon rank-sum test or the Kruskal–Wallis test, whilst parametric data were studied using either analysis of variance (ANOVA). Kolmogorov–Smirnov test was used for Normality Test.

To reduce selection bias and potential confounders, propensity score matching(PSM) method10 was applied. Variables included in the matching include age, gender, weight, heart rate, CHF, Systolic HF, Diastolic HF, Acute Physiology Score III (APS III), Simplified Acute Physiology Score II (SAPS II), Sequential Organ Failure Assessment (SOFA), Glasgow (GCS) score, blood lactate, bilirubin, oxygenation index, platelets, blood creatinine, blood urea nitrogen, hemoglobin, etc. We set MT to 0.02, selected sampling without replacement, and used Maximize execution performance to perform PSM. Based on log-rank tests, Kaplan–Meier survival curves were analyzed. Cox regression analyses were used to analyze the significance of various variables on survival.

Multiple logistic regression modelling will be used for categorical outcomes and multiple linear regression modelling will be used for continuous variables. Variables with P < 0.1 at univariable analysis were included in a multivariable logistic regression model with a stepwise selection method. Statistical analyses were performed with SPSS version.25 and R version 3.5.3. Bilateral P-value < 0.05 was considered statistically significant.

Results

We found 61,532 records in MIMIC-III v1.4 and finally, 1357 individuals were enrolled. 60,175 records were excluded (33,956 for patients < 65 years of Age; 3628 for ICU LOS < 24 h; 12,295 for no sepsis; 6069 for no HF; 2 for the wrong record of LOS; 3023 due to no-use of vasoactive drugs; 365 for non-first ICU admission records; 27 for vasopressin alone and 810 for non-combinations of this study) (Fig. 1).

Figure 1
figure 1

Workflow of the Study. NE-norepinephrine, LOS-length of stay.

Full size image

Baseline demographic information and clinical outcomes

Overall, 1357 patients had an average age of 79.04 years (SD, 7.46 years), 50.7% were male and 72.7% were white. Patients were divided into norepinephrine group (NE group), norepinephrine combined with vasopressin group (NE + VAS group), and dopamine group according to the vasoactive drug used. We found differences existed among the three groups in terms of demographic characteristics, comorbid diseases, and laboratory tests at ICU admission. On clinical scores that reflected the severity of the disease, there were significant differences except for the oxygenation index (PaO2/FiO2) (Table 1). To reflect some level of volemic status, the CVP values of each group of patients were shown in Table 1. Among pediatric and adult patients, VIS is known to be an independent predictor of adverse outcomes including ventilator days, intensive care unit length of stay, and mortality11,12, so we also analyzed VIS within 6 h, 24 h, 24 to 48 h, and 48 to 72 h after ICU admission. The primary and secondary outcomes were presented in Table 2.

Table 1 Baseline clinical characteristics of the study population.
Full size table
Table 2 Outcomes of the study population.
Full size table

Clinical outcomes after PSM

To balance the baseline factors, we performed PSM. After 1:1 PSM, all of the groups were comparable concerning characteristics (Table 3), however, we found significant differences between 3 groups for outcome variables, especially mortality at 28 days and 90 days (Table 4).

Table 3 Baseline clinical characteristics after propensity score matching.
Full size table
Table 4 Clinical outcomes after propensity score matching.
Full size table

Five-year survival analysis and risk factor

5-year survival analysis of all 1367 patients suggested NE group, dopamine group, and NE + VAS group had significantly different cumulative survival rates (P < 0.001, Fig. 2). After Cox regression models, based on our study population, the combined norepinephrine and vasopressin decreased 5-year survival (HR = 1.346, P = 0.001) (Table 5).

Figure 2
figure 2

Five-year survival analysis in NE, NE + VAS and dopamine groups were 29.4%, 21.2%, 25.8%, respectively. NE-norepinephrine, VAS-vasopressin.

Full size image
Table 5 Risk factors associated with 5-year mortality in the study population.
Full size table

Association with ICU LOS/Hospital LOS

As secondary endpoints, we focused on ICU LOS and Hospital LOS. According to a multivariate linear retrospective model, dopamine alone may shorten ICU LOS and Hospital LOS (Tables 6, 7).

Table 6 Association with ICU length of stay among the three groups.
Full size table
Table 7 Association with hospital length of stay among the three groups.
Full size table

Additionally, as we all know, there were more new-onset arrhythmias observed in patients treated with dopamine than norepinephrine (high-quality evidence)13. So we used a logistic regression model to evaluate new-onset arrhythmias incidence and new-onset malignment arrhythmias among the three groups (Tables 8, 9).

Table 8 Association with new-onset arrhythmias among the three groups.
Full size table
Table 9 Association with new-onset malignment arrhythmias among the three groups.
Full size table

Dopamine alone had higher new-onset arrhythmias (OR = 1.665, P < 0.001); norepinephrine combined with vasopressin group had higher malignant arrhythmias (OR = 2.829, P < 0.001).

Discussion

In our study, we found norepinephrine combined with vasopressin worsened outcomes of elderly sepsis patients with heart failure, which suggested this combination was an independent risk factor for 5-year survival. Dopamine alone reduced the Hos Los and ICU Los but had a higher risk of new-onset arrhythmias.

Population ageing is the most important medical and social demographic challenge worldwide14. In combination with age-related changes in the human immune system, these immunologic changes may make the elderly particularly susceptible to sepsis15,16. Autopsy records for elderly people over 80 in China also confirm that infection-related diseases are the second leading cause of death accounting for 26.6% of all deaths17. The high global mortality of sepsis18 is also associated with the failure to keep the hemodynamic status. The choice of vasoactive drugs is more complex and challenging, especially in sepsis patients with heart failure.

Dopamine and epinephrine are catecholamines. An early review19 showed that norepinephrine had an advantage over dopamine in all-cause mortality and the development of arrhythmias in septic shock. SC guidelines also recommend norepinephrine in septic shock20. As the first-line treatment in cardiogenic shock, norepinephrine has replaced epinephrine21. However, the role of which vasoactive drugs in patients with septic shock with heart failure is still controversial22, especially in the elderly. Vasopressin, which is synthesized by the hypothalamic paraventricular and supraoptic nucleus23, is recommended as second-line therapy for adults suffering from septic shock with inadequate mean artery pressure levels24. However, animal experiments have shown that vasopressin may decrease coronary blood flow25. Therefore, we would like to know if norepinephrine combined with vasopressin is appropriate for elderly sepsis patients with heart failure. We found that NE combined with vasopressin may be harmful (28-d, 90-d mortality, and other outcomes) to this study population and has the higher mortality in five-year survival analysis among three groups (P < 0.001). Long-term survival is independently influenced by this combination, which is not consistent with the findings of VASST in 200826. However, the VASST study population did not include patients with NYHA III and IV, and patients were not grouped by age. More interestingly, in 2018, the same VASST Group found that 28-day mortality was significantly higher in NE + vasopressin group than in NE alone(60.8% vs. 46.2%, P = 0.009) in a retrospective study27. Although this retrospective analysis also did not group age and cardiac function, it has partially supported our opinion.

Second, dopamine alone shortened ICU-LOS and Hos-LOS compared with the other two groups, which sounds good for this population. After regression analysis, it was found that dopamine remained an independent risk factor for new-onset arrhythmias, which is consistent with SOAP II22. Meanwhile, NE + vassoprssin was the independent risk factor for new-onset malignant arrhythmias in this study population. We need to consider avoiding this combination in elderly sepsis patients with HF.

This study has the following limitations, first, we conducted a PSM analysis to minimize selection bias in a retrospective study, but the risk of residual unmeasured confounding remains possible. Therefore, the results should be considered in the target population. In addition, the limitations of this study include the lack of each patient's cardiac function and cardiorespiratory endurance before admission. Changes in blood composition may be caused by both pathogens and antibiotics. And volemic status of patients were unknown although we attempted to use CVP and CO reflect. We acknowledge that one of the limitations of our study is that data might be missing from the medical charts. Last, but not least, it was a retrospective single-center study, further multi-center prospective studies are necessary to corroborate our findings.

Conclusions

Taken together, norepinephrine in combination with vasopressin decreased survival and increased the incidence of malignant arrhythmias in elderly sepsis patients with pre-existing heart failure. Dopamine alone reduces ICU and hospital length of stay but increases the new-onset arrhythmias.