Inpatients with 18 years or older undergoing hemodialysis three times a week for at least three months, fully performing the hemodialysis session, and spontaneously breathing were invited to take part of the study. Were excluded inpatients with a clinical suspicion or a confirmed diagnosis of severe heart disease, systemic sclerosis, interstitial lung disease of any nature, acute pulmonary infection, ongoing neoplasm at any site, deformities of the rib cage preventing ultrasound evaluation, a pulmonary, hepatic, or heart transplant. Thirty patients were recruited for the study.

After signing a written informed consent and immediately before the beginning of the hemodialysis session, we collected sociodemographic data, history of comorbidities, and recent laboratory tests. Subsequently, an ultrasound examination was performed by an experienced bedside ultrasound operator (MR), trained and certified by the World Interactive Network Focused On Critical UltraSound, who operates the equipment on a daily basis, following the SAFE protocol (Killu et al., 2020) [11], and using a portable ultrasound device model Butterfly iQ+ (Butterfly Network, Inc., Guilford, CT, United States). The images were stored and reviewed by two physicians, a pulmonologist and a nephrologist, also experienced in bedside ultrasound.

For each organ, the modes and frequencies in the handheld ultrasound device were altered as follows: for the cardiac examination, the cardiac mode was used, with a frequency ranging from 5.0 to 7.5 MHz, depending on the patient’s adipose tissue; for the pulmonary examination, the linear transducer mode was employed; for IVC, the curvilinear transducer mode was used; for the examination and measurements of IJV, the headboard was raised at a 30-degree angle and the linear transducer mode was employed, placed laterally at the level of the cricoid cartilage. We used the following formula to calculate IVCCI and IJVCI:

$$\:\left[\frac\text\text\text\text\text\text\:\text\text\text\text\text\text\text\text-\:\text\text\text\text\text\text\text\:\text\text\text\text\text\text\text\text\right)}\text\text\text\text\text\text\:\text\text\text\text\text\text\text\text}\right]\times\:\:100$$

The same sonographic protocol was applied immediately after the hemodialysis session. In addition, we collected some data after the hemodialysis session such as ultrafiltrate (UF), UFNET, medium blood pressure, and adverse events.

On this study, the following scores were assigned: cardiac function – (a) hyperkinetic = 1; (b) normal = 2; (c) hypokinetic = 3; pulmonary evaluation – (a) < 1 B-lines mean per field = 1; (b) 1–2 B-lines mean per field = 2; (c) ≥ 3 B-lines mean per field = 3; IVC diameter and spontaneous respiratory variation – (a) < 2.5 cm in widest diameter and > 50% respiratory variation in diameter = 1; (b) 1.5–2.5 cm in widest diameter and < 50% respiratory variation in diameter = 2; (c) > 2.5 cm in widest diameter and < 50% respiratory variation in diameter = 3; IJV – (a) > 40% respiratory variation = 1; (b) 20–40% respiratory variation = 2; (c) < 20% respiratory variation = 3. The scores of all four exams were added up to have a final score (SAFE score) for compound of cardiac contractility, extravascular pulmonary edema, and relative intravascular volume, and finally interpreted as: (a) 4 to 6 = hypovolemia; (b) 7 to 9 = normovolemia; (c) 10 to 12 = hypervolemia.

Regarding the cardiac evaluation methodology of the SAFE protocol, it was strictly followed by the researcher, utilizing the same steps that included obtaining a long-axis view of the heart, observing cardiac function, and estimating ejection fraction using either the eyeballing method or M-mode with maximum systole and diastole measurements; additionally, and also a short-axis view of the heart was obtained to assess cardiac function and estimate ejection fraction.

It is important to mention that in the SAFE protocol, the number of B-lines counted from all examined segments was added together, and then divided by the total number of segments examined to calculate the average. In our study, the same approach was employed, with scanning performed in 4 lung regions for each hemithorax.

In the original study, the hypovolemic, normovolemic, and hypervolemic profiles were scored as -1, 0, and + 1, respectively. In the present study, we adapted them to + 1, +2, and + 3. Therefore, it was termed the Adapted SAFE Protocol (SAFE-a).

We characterized the demographic profile, hemodialysis session data, and sonographic findings using absolute frequency and relative frequency for categorical variables and mean and standard deviation for continuous variables. We verified data parametricity using the normalized Q-Q plot and standardized residue histogram [16].

We evaluated the distribution of ultrasound findings before and after hemodialysis by applying the McNemar test followed by post-hoc analysis [17]. We compared the SAFE-A score before and after hemodialysis using the paired t-test and performed multiple regression analysis between the UFNET with the variation of the score of the SAFE-A protocol (ΔSAFE-A), variation of the number of B-lines (Δnumber of B-lines), variation of the echocardiography (Δechocardiography), variation of the inferior vena cava (ΔIVC), and variation of the internal jugular vein (ΔIJV). We adopted the Backward method to select the model with greater accuracy and predictive power and the Pearson’s correlation matrix to evaluate the relationship between the variations. We analyzed data applying the Statistical Package for Social Science version 26.0 (IBM SPSS Statistics for Windows, IBM Corporation, Armonk, NY, United States) and the significance level of 5% (p < 0.05)

Informed written consent was obtained from all participants and the study was apprioved by the Institutional Review Board (IRB) of the Hospital Estadual Alberto Rassi (Goiânia, GO, Brazil) on August 18, 2022 (CAAE: 59768822.3.0000.0035).