Since the publication of the study “SMART: Is saline on the tightrope?”1, several authors have worked with the hypothesis that the infusion of balanced solutions in critical patients improves the outcomes in terms of renal function and mortality (Table 1). No studies have evidenced benefit from the use of saline solution; in the best of cases, performance has been reported to be equivalent2.
Principal characteristics of randomized controlled studies comparing 0.9% saline solution versus balanced solutions in critical patients after 2018.
Author, year | Name of study | Country | Patients randomized to SS 0.9% | Patients randomized to BS | Type of balanced solution | Follow-up |
---|---|---|---|---|---|---|
Semler, 2018 | SMART | USA | 1169 | 1167 | RL or ABS | 30 days |
Pagano, 2020 | Italy | 49 | 35 | RL | – | |
Golla, 2020 | India | 80 | 80 | RL | 30 days | |
Zampieri, 2021 | BaSICS | Brazil | 1017 | 1987 | ABS | 90 days |
Finfer, 2022 | PLUS | Australia and New Zealand | 1026 | 1068 | ABS | 90 days |
RL: Ringer lactate; BS: balanced solution; ABS: acetate balanced solution; SS 0.9%: 0.9% saline solution
A recent systematic review has been made of randomized clinical trials comparing balanced crystalloids with saline solution in adult critical patients, with 90-day mortality as the primary endpoint. The frequency statistics analysis revealed no significant differences. However, Bayesian analysis showed the posterior probability that balanced crystalloids reduce mortality to be 89.5%3.
Physicians naturally use the Bayes theorem in decision-making referring to patient diagnosis, prognosis and treatment. Based on an effect, we decide on the cause of the effect and intuitively assign a probability to the cause we are considering. Different information sources added in sequence – development of hyperchloremia, acidosis, renal failure – define in an increasingly precise way the probability of a certain cause, and always on an intuitive basis4. However, when considering research methodology, priority is given to the simplicity of being able to decide whether a treatment, a diagnostic method or a procedure is effective or not, based on a pre-established p < 0.05. The problem is that the p-value, which is no more than a tool used to quantify such uncertainty, is too often overvalued and misinterpreted.