Bioelectric Impedance in Non-invasive Ventilated Patients: The NIVAR Study.

Overview

Investigators performed a retrospective observational single-centre study in a cohort of COVID-19 patients with AHRF who required NIV support and intensive care unit (ICU) admission from May 2021 to January 2022. A BIA assessment was performed within the first 24 hours after NIV onset. Patient characteristics, severity scores, HACOR score, NIV settings, and mortality were recorded. Cox-hazard multivariate analysis was used to assess the association between BIA parameters and NIV failure, adjusting for potential covariates.

This was a retrospective observational cohort study performed at Hospital Universitari de Bellvitge, a tertiary-care public hospital with a 48-bed Intensive Care Unit (ICU). Between May 2021 and January 2022, investigators included all consecutive patients admitted to our ICU with moderate-to-severe AHRF (PaO 2 /FiO 2 \&lt;200mmHg) due to SARS-Cov2 infection who underwent NIV support for at least 24h. All patients admitted to the ICU with AHRF due to confirmed COVID-19 infection by using real-time reverse-transcription chain reaction in respiratory fluids with the Light- Cycler 480 System instrument II (Roche Life Science, Indianapolis, IN, USA). To improve homogeneity, only subjects with onfirmed COVID-19 infection were included. The main exclusion criteria were the presence of electrical implants, wounds or skin damage at the designated electrode sites or the inability to perform BIA. Patients who met criteria for immediate intubation upon ICU admission were also excluded. Other circumstances or clinical conditions at the discretion of the investigator that precluded the participation of the patient in the study were also considered (i.e., limitation of therapeutic effort and poor prognosis). Because of the observational design, informed consent was waived. Data collection Several clinical variables were recorded, including age, gender, comorbidities (e.g.,Body mass index (BMI)), laboratory values and severity scores on ICU admission, characteristics of NIV support, HACOR score on ICU admission and 2h after ICU admission, and data related with outcome, such as time from NIV initiation to ETI. Gas exchange variables were obtained after initiation of NIV therapy. Disease severity scores (i.e. Acute Physiology and Chronic Health disease Classification System (APACHE) II and Sequential Organ Failure Assessment (SOFA) scores) were both calculated based on available parameters from the first 24h after ICU admission. BIA parameters were measured within the first 24h of ICU admission, once the patient was receiving NIV support. The presence of acute kidney injury (AKI) on ICU admission was evaluated based on the Kidney Disease Improving Global Outcomes (KDIGO) criteria. The primary outcome was NIV failure, defined as the need for ETI after NIV initiation, and the secondary outcome was in-hospital mortality. BIA parameters BIA parameters were measured by trained researchers with the InBody S10® (InBody Co., Ltd., Seoul, Korea) within the first 24h ICU admission. This multifrequency, segmental impedance analyser requires height, weight, and sex as input parameters. Height and weight measured upon admission were used. BIA measurements were performed in supine position with two electrodes attached to each of the four extremities. The InBody S10® measured impedance at multiple frequencies and calculated a whole-body phase angle (PhA) of 50 kHz, and segmental measurements were used to calculate thorax PhA. PhA values derived from BIA (calculated as \[arctangent (reactance/resistance) × 180°/π\]) represent the impedance of the human body. Parameters related with fluid status (i.e., total body water (TBW), intracellular water (ICW), extracellular water (ECW) and ECW ratio (ECW/TBW)), muscle status (i.e., soft lean mass (SLM) average, muscle skeletal mass, muscle skeletal mass index, TBW and Fat Free Mass ratio), and other body composition related parameters (i.e., Fat Body Mass and percentage of Body Fat) were also recorded. Normal ranges for BIA parameters depend on anthropometric values of each patient, but they are compared with population studies to assess normality. Statistical analysis Data was expressed as means and standard deviations, median (interquartile ranges), or numbers and percentages, as appropriate. In all cases, the Kolmogorov-Smirnov test and Agostino-Pearson omnibus normality test were used to check for normality of variables distribution. The Mann-Whitney U test or the two-sample t-test were used to compare between groups (NIV failure vs NIV success), as appropriate, and the Chi2 test or Fisher exact test were used to evaluate categorical prognostic factors in the univariate analysis. Analysis of variance was used to compare differences in characteristics between smaller subgroups, whereas the Bonferroni post-hoc test was performed to determine significant differences in the pairwise comparisons. Subsequently, multivariate analysis was performed using an adjusted multiple stepwise Cox regression. Variables were included in the initial model if they had a P-value \<0.2 in the univariate analysis. Moreover, those variables previously associated with NIV failure were also included. The change-in-estimates criterion and backward deletion with a 10% cutoff were used to eliminate variables from the final model. To avoid destabilising the multivariate analyses, interactions between all introduced variables were tested to minimise confounders and the influence of illness severity when analysing outcomes. Results are presented as adjusted hazard ratios with 95% confidence intervals. Based on population's variable distribution (median value), PhA was categorized into subgroups for further survival analysis, which was conducted using the Kaplan-Meier estimator for each subgroup (i.e., Log-rank test). The statistical analyses were conducted using IBM SPSS version 20.0 (IBM Corp., Armonk, NY, USA), and two-tailed P-values \<0.05 were considered statistically significant.