Cardiac index modifies the association between blood pressure response index and 28-day mortality in sepsis

Dear Editor,

The blood pressure response index (BPRI) relates the mean arterial pressure (MAP) achieved to the intensity of vasoactive and inotropic support. In a multicentre study, minimum BPRI showed a consistent L-shaped association with mortality, and subsequent trajectory analysis extended its value by identifying clinically distinct patterns during early vasopressor therapy [1, 2]. BPRI therefore provides a useful summary of pressure response under treatment. Cardiac output, however, may differ between patients with a similar pressure response. A recent commentary highlighted this physiologic uncertainty and the need to consider the circulatory state represented by BPRI [3]. We examined whether echocardiographic cardiac index (CI), as a measure of forward flow, provides additional context for the association between BPRI and mortality in sepsis.

We conducted a retrospective cohort study using MIMIC-IV v3.1 and MIMIC-IV-ECHO v1.0 [4, 5]. Adults meeting the derived Sepsis-3 criteria were eligible when vasopressor support began during the ICU stay and within 24 h of sepsis onset [6]. We identified the first ICU stay within each hospitalization and fixed the first structured transthoracic echocardiogram performed within 24 h after vasopressor initiation before assessing whether CI could be calculated. BPRI was obtained from the most recent concurrent MAP and Vasoactive-Inotropic Score (VIS) measurements during the 60 min preceding echocardiography; invasive MAP was preferred when available. CI was calculated from left ventricular outflow tract diameter, velocity-time integral, heart rate and body surface area recorded in that examination. We excluded examinations performed during mechanical circulatory support and retained the first fully eligible TTE-indexed admission for each patient. The outcome was death within 28 days after TTE. Logistic regression modelled the nonlinear main effects of log-transformed BPRI and CI using three-degree-of-freedom natural cubic splines and included a single interaction term between standardized log(BPRI) and standardized CI. The model adjusted for age, sex, Charlson Comorbidity Index, noncardiovascular Sequential Organ Failure Assessment score, invasive mechanical ventilation and time from vasopressor initiation to TTE. We compared the interaction model with the corresponding additive model using a likelihood-ratio test and assessed precision with sandwich-robust 95% confidence intervals and 1,000 patient-level bootstrap samples.

The cohort included 736 patients, of whom 273 (37.1%) died within 28 days. Mean age was 67.6 years, 393 patients (53.4%) were men, and 431 (58.6%) received invasive mechanical ventilation. Median BPRI was 5.3 (interquartile range, 2.8–12.2), median CI was 2.6 L/min/m² (2.0–3.5), and TTE was performed a median of 10.4 h (6.0–15.5) after vasopressor initiation. There was evidence that the association between BPRI and 28-day mortality varied with CI (interaction odds ratio, 0.79; robust 95% confidence interval, 0.65–0.96; likelihood-ratio P = 0.015). The pattern changed across the BPRI range (Fig. 1). At BPRI 1, predicted mortality was approximately 10% points higher at the 75th than at the 25th percentile of CI. The difference was minimal around BPRI 4 and reversed at BPRI 20, where predicted mortality was approximately 9% points lower at the higher CI level. Estimates were less precise toward the extremes of BPRI. Patient-level bootstrap resampling yielded a similar interaction estimate (95% confidence interval, 0.63–0.96).

Fig. 1.

Cardiac index and the continuous association between BPRI and adjusted 28-day mortality. A, Adjusted standardized 28-day mortality risk across continuous BPRI at the 25th percentile, median, and 75th percentile of cardiac index. B, Adjusted absolute risk difference comparing cardiac index at the 75th versus 25th percentile across continuous BPRI; the horizontal line denotes no risk difference. BPRI is displayed in its original units on a logarithmic axis. Estimates were standardized over the observed covariate distribution. Shaded areas in both panels indicate patient-level bootstrap 95% confidence intervals

Taken together, these findings place BPRI and CI in complementary roles. BPRI efficiently summarizes the pressure achieved relative to vasoactive and inotropic support, whereas CI adds information about the accompanying forward-flow state. Their relationship is unlikely to be uniform because arterial pressure reflects both flow and vascular tone and is actively titrated during clinical care. At low BPRI, the pattern observed with higher CI may be compatible with marked vasodilation despite preserved or increased flow, although vascular resistance was not measured and estimates were less precise at the extremes. In the middle BPRI range, CI provided little visible risk separation. At higher BPRI, a comparatively favourable pressure response could still coexist with lower forward flow and higher risk. Thus, BPRI remains a practical marker of pressure response, while CI may refine its interpretation when echocardiography is already clinically indicated and available. This interpretation is consistent with contemporary shock guidance, which places arterial pressure alongside cardiac output, cardiac function and tissue perfusion [7].

This study was retrospective and conducted at a single centre. The cohort was limited to patients who underwent TTE and had sufficient measurements to calculate CI, and a single paired assessment could not capture subsequent hemodynamic changes. Tissue-perfusion measurements were not available at the same time point. Nevertheless, the findings suggest that the association between BPRI and mortality varies with contemporaneous forward flow. CI may therefore provide additional context for interpreting BPRI, a finding that warrants confirmation in independent cohorts.

Abbreviations

BPRI

Blood pressure response index

CI

Cardiac index

ICU

Intensive care unit

MAP

Mean arterial pressure

SOFA

Sequential Organ Failure Assessment

TTE

Transthoracic echocardiography

VIS

Vasoactive-Inotropic Score

Author contributions

Rong Lei, Chongchong Ye and Chaofu Yue conceived and designed the study, contributed to data extraction and formal analysis, interpreted the results and drafted the manuscript. Qinyong Yan, Xueyan Wang, Qingsong Ma and Caimei Hu contributed to data verification, interpretation and critical revision. Feng Yue and Mei Yang supervised the study and contributed to methodology, interpretation and critical revision. All authors read and approved the final manuscript.

Funding

This work was supported by the Yunnan Province Longyun Expert Workstation (no. 202305AF150090) and the National Key Clinical Specialty Construction Project.

Data availability

The source data analyzed in this study are available through the PhysioNet repository as MIMIC-IV version 3.1 (https://physionet.org/content/mimiciv/3.1/) and MIMIC-IV-ECHO version 1.0 (https://physionet.org/content/mimic-iv-echo/1.0/). Access requires completion of the relevant training and credentialing procedures and acceptance of the PhysioNet data use agreement.

Declarations

Ethics approval and consent to participate

MIMIC-IV and MIMIC-IV-ECHO contain deidentified clinical data. Their establishment and use were approved by the institutional review boards of Beth Israel Deaconess Medical Center and the Massachusetts Institute of Technology, with the requirement for individual informed consent waived. This study was a secondary analysis of deidentified data and involved no direct patient contact or intervention.

Competing interests

The authors declare no competing interests.