Back to BaSICS: Early Treatments Matter in Critical Illness

Intravenous fluid is the most common therapy received by critically ill adults. Recent randomized trials have examined the optimal volume and composition of intravenous fluid in critical illness. Because the effects of fluid on physiology and outcomes may be greatest during the earliest phases of critical illness, randomized trials of fluid volume have predominantly focused on initial management in the emergency department or operating room before ICU admission (1, 2). In contrast, trials examining fluid composition have frequently controlled fluid therapy only after ICU admission (3–5), an approach that could predispose to finding no difference between trial groups either by missing the phase of illness in which patients receive the most fluid (decreasing “separation between groups”) or by exposing patients to the nonassigned fluid before enrollment (“contamination”). This trial design consideration may apply to the four large trials that recently compared balanced crystalloids versus saline in acutely ill adults (6–9). The two trials in which fluid composition was controlled in the emergency department before admission reported a benefit to the use of balanced crystalloids (6, 7), whereas the two trials in which fluid composition was controlled only after ICU admission reported no statistically significant difference (8, 9). Could fluid therapy early in critical illness be a key determinant of the effect of fluid composition on outcomes?

In this issue of the Journal, Zampieri and colleagues (pp. 1419–1428) address this question through a secondary analysis of one of the two large trials of balanced crystalloids versus saline that enrolled patients in the ICU (10). BaSICS (Balanced Solutions in Intensive Care Study) was a multicenter, randomized trial comparing balanced crystalloids versus saline among 10,520 patients in the ICU (8). Most patients were enrolled within 1 day after ICU admission, and 68% had received balanced crystalloids, saline, or both before enrollment. The primary outcome, 90-day mortality, did not significantly differ between the balanced crystalloid and saline groups (hazard ratio, 0.97; 95% confidence interval, 0.9–1.05).

This secondary analysis of the BaSICS trial examined whether the type of fluid patients received in the 24 hours before enrollment modified the effect of trial group assignment on mortality. Patients were categorized as having received only balanced crystalloid, only saline, a mix of both, or no recorded fluid before enrollment. The authors hypothesized that among patients who had received only balanced crystalloids before enrollment, those randomized to balanced crystalloids would experience lower 90-day mortality than those randomized to saline.

The results confirmed the authors’ hypothesis. Among the 3,202 patients who had received only balanced crystalloids before enrollment, mortality was 16% in the balanced crystalloid group and 20% in the saline group, and the probability that balanced crystalloids decreased mortality compared with saline was 92%. This high probability of benefit from balanced crystalloids was consistent across planned admissions (97%), unplanned admissions with sepsis (96%), and unplanned admissions without sepsis or traumatic brain injury (94%). As in the original trial (8), balanced crystalloids increased mortality compared with saline for patients with traumatic brain injury, potentially because of differences in tonicity.

In summary, this secondary analysis suggests that the effects of balanced crystalloids versus saline on mortality in the BaSICS trial may have been influenced by the type of fluid patients received in the emergency department, operating room, and ICU before enrollment. For patients who had only received balanced crystalloids before enrollment, the use of balanced crystalloids in the ICU appeared to decrease mortality. For patients who had received saline before enrollment, the choice of fluid in the ICU did not appear to affect outcomes.

This analysis has several strengths. First, BaSICS was a multicenter, randomized trial of balanced crystalloids versus saline, making it well-suited to this effect modification analysis. Second, the sample size allowed thousands of patients in each subgroup. Third, while this analysis was post hoc, the hypothesis it examined originated in a prior study (11); the concordance in findings decreases the likelihood that they are because of chance.

This analysis also has limitations. First, it was post hoc and reported multiple associations. Second, fluid receipt before enrollment was determined by treating clinicians and not trial protocol. Thus, the apparent effect modification by fluid type before enrollment may actually represent effect modification by other patient or clinician factors that are correlated with the type of fluid received before enrollment. Third, the mechanism is uncertain by which receipt of saline before enrollment could mitigate differences in outcomes between patients randomized to balanced crystalloids and saline. The data on chloride and hyperchloremia from this analysis do not appear to support these factors as mediating the observed effect modification.

Overall, this secondary analysis of the BaSICS trial modestly adds to the evidence supporting the use of balanced crystalloids for most critically ill adults in the emergency department and ICU. The findings are consistent with a recent meta-analysis of randomized trials (including BaSICS) that found a 90% probability that balanced crystalloids decrease mortality compared with saline (12). They are consistent with prior analyses suggesting that the beneficial effects of balanced crystalloids may be greater for patients with sepsis (13) and that balanced crystalloids may increase mortality in patients with traumatic brain injury (6, 8). The findings also reinforce that fluid management early in critical illness may meaningfully affect longer-term outcomes.

Finally, the analysis by Zampieri and colleagues informs the design of future randomized trials. The effects of fluid therapy, antimicrobial therapy (14), ventilator management (15), and other fundamental treatments may be greatest early in critical illness. Randomized trials that begin enrollment in the days after ICU admission risk missing the key phase of the critical illness and biasing toward falsely “null” results. Going forward, ICU, emergency department, and prehospital clinicians and researchers must continue to develop the collaborations and research methods needed to conduct randomized trials that begin as soon as possible after the onset of critical illness. Critical illness research must transcend its historical focus on the ICU and expand to include the full continuum of critical illness, from prehospital care to posthospital recovery.