The literature search identified 18,802 citations, and 377 full-text articles were assessed for eligibility. Of 58 RCTs which included 26,351 patients in the analysis, 5 large RCTs included more than one condition — sepsis, surgery, trauma, and traumatic brain injury. Thus, we extracted the subgroup data of patients with different conditions. As a result, 23 RCTs on sepsis patients, 24 on surgical patients, 10 on trauma patients, and 4 on traumatic brain injury patients were included for further analysis (Figure 1, appendix pp 17-48). We present the risk of bias assessment for each included study in appendix 7 (appendix pp 61-70); eFigure 7.1 shows the overall risk of bias in five domains in sepsis trials, eFigure 7.2 shows the risk of bias for the individual studies, and eFigure 7.3 explained the reasons for upgrading or downgrading in every studies (appendix pp60-63, 64-66, 67-69). The reasons to downgrade are mostly inadequate randomization process, open labeled design or no detailed information. No significant differences in baseline variables between interventions were observed within our NMA (appendix pp 49-60).
Most RCTs used the 2001 International Sepsis Definitions Conference sepsis definition (24) and included sepsis patients with shock status or those who had evidence of tissue or organ hypoperfusion (eTable 5.2, appendix pp23-26). The timing for fluid resuscitation is when the patient meets the enrollment criteria: systemic hypoperfusion defined by low blood pressure, low central venous pressure or wedge pressure and elevated lactate level. We compared the mean arterial pressure among interventions at baseline (appendix pp56), ranging from 69.0 to 73.9 mmHg, and found no statistically significant differences among seven fluid types. Besides, in eTable 5.4 (appendix pp29-31), and eTable 5.6 (appendix pp35-36), we compared resuscitation targets among included trials. The resuscitation goals are generally to maintain wedge pressure around 15~18 mmHg or central venous pressure around 8~12 mmHg. The average mean study fluid volume was 2397.4 mL ± 1019.1 mL in each arm, and the total resuscitation fluid volume was 7615.6 mL ±1729.7 mL (appendix pp 22-31, 61-64). In eTable 5.3 (appendix pp26-27), we presented the baseline characteristics, including age, severity of illness, mean arterial pressure, and lactate level.
Between 1983 and 2018, 23 RCTs with 14,659 participants presented with usable results on mortality. In appendix eTable 5.1 (appendix pp 21), we provided the details of mortality outcome used in our analysis, including in-hospital mortality, 30 day-mortality, and 90-day mortality. If multiple time points were reported in a study, we chose the longest observation period for mortality analysis. Balanced crystalloids reduced mortality more than saline and L-HES with odds ratios (OR) of 0.84 (95% CI 0.74-0.95) and 0.81 (95% CI 0.69-0.95), respectively (Figure 2A). Sequential NMA further supported the difference in mortality rate between balanced crystalloids vs. saline and L-HES by demonstrating that the trend in cumulative evidence exceeded the efficacy boundary. The cumulative evidence exceeded the futility boundary in the comparison between balanced crystalloids and Albumin, but fell between efficacy and futility boundary in the comparison between balanced crystalloids and gelatin (Figure 3). According to SUCRA, balanced crystalloid appeared to be the best option; however, saline, L-HES, and H-HES were not favored (Figure 4).
Sepsis Patients- Fluid Resuscitation Volume
Thirteen trials with 10,970 participants reported usable results for fluid resuscitation volume in sepsis patients. Balanced crystalloids and saline required more fluid volume than iso-oncotic albumin with mean differences (MD) of 2122 mL (95% CI -300 to 4544 mL) and 1964 mL (95% CI 89 to 3840 mL), respectively (Figure 2B). SUCRA revealed that the colloids were associated with less resuscitation fluid volume than crystalloids (Figure 4).
Sepsis Patients- Acute kidney injury
Eleven trials with 10,569 participants reported usable results for acute kidney injury. Balanced crystalloids significantly reduced a greater risk of acute kidney injury than L-HES (OR, 0.80; 95% CI 0.65-0.99), and H-HES (OR, 0.54; 95% CI 0.37-0.84) (Figure 2C). SUCRA ranking revealed that gelatin, balanced crystalloid, saline, and iso-oncotic albumin had a lower risk of acute kidney injury than L-HES and H-HES (Figure 4).
Sepsis Patients- Red Blood Cell Transfusion Volume
Ten trials with 11,979 participants reported usable results for the packed red blood cell transfusion volume. Balanced crystalloids required less volume of red blood cell transfusion than hyperoncotic albumin (MD, 274 mL; 95% CI 5 mL to 548 mL), L-HES (MD, 232 mL; 95% CI 35 mL to 430 mL) and H-HES (MD, 497 mL; 95% CI 141 mL to 854 mL). (Figure 2D). SUCRA revealed that the crystalloids and iso-oncotic albumin were associated with less transfusion volume than other colloids (Figure 4).
The funnel plot and Egger’s test did not detect any significant publication bias (appendix pp 114-116). Loop inconsistency and design inconsistency were also not detected (appendix pp 124-129). The meta-regression did not change the ranking order (appendix pp 138-139). The evidence certainty in mortality revealed a moderate to high evidence confidence in comparison, including balanced crystalloids, saline and L-HES; low to moderate in iso-oncotic albumin and hyperoncotic albumin; very low in gelatin, and H-HES (appendix pp 139-142). Results of sensitivity analyses with the exclusion of the largest SMART trials (12) or the inclusion of the pilot SALT trial (25) in appendix 14 show no substantial differences from the main analysis.
During 1979 to 2020, 8 (34.80%), 6( 26.00%), 6 (26.00%), and 3RCTs compared different resuscitation fluids in patients receiving cardiac surgery, aortic surgery, major abdominal surgery, and hip arthroplasty and cystectomy, respectively (appendix pp 32-36). Fluid resuscitation was provided during surgical procedures to maintain hemodynamic parameters in most trials, and the mean resuscitated fluid of interest was 3327.5 mL (appendix 65-67).
Surgical Patients- Mortality
Twenty-three trials with 4,646 participants had valid results on mortality. There were no significant differences in mortality between 7 interventions (Figure 5); and SUCRA showed that hyperoncotic albumin and balanced crystalloid were associated with less mortality than gelatin, HES, and saline (Figure 4).
Surgical Patients- Fluid Resuscitation Volume
Twenty trials with 4,512 participants provided data on resuscitation fluid volume. Balanced crystalloids group required significantly more fluid resuscitation volume than iso-tonic albumin (MD, 2612 mL; 95% CI 1416-3800), hypertonic albumin (MD, 2852 mL; 95% CI 742-4962), L-HES (MD 1494 mL; 95% CI 345-2644), H-HES (MD, 1462 mL; 95% CI 418-2505), and gelatin (MD, 1154 mL; 95% CI 64-2240), (Figure 5). SUCRA ranking showed that colloids (albumin, HES, and then gelatin) were associated with less fluid resuscitation volume than crystalloids (Figure 4).
Surgical Patients- Acute kidney injury
Fourteen trials with 4,248 participants reported results for acute kidney injury. The ORs between seven treatments were not statistically significant (Figure 5). SUCRA showed iso-oncotic albumin, and balanced crystalloids were associated with less acute kidney injury than HES and gelatin.
Surgical Patients- Red Blood Cell Transfusion Volume
Sixteen trials with 2,818 participants presented usable results for red blood cell transfusion volume. Ranking probabilities showed that albumin, L-HES, and then gelatin were associated with less transfusion volume than H-HES and crystalloids (Figure 5).
Publication bias and inconsistency were not significant (appendix pp 118-121). The confidence ratings were low to very low among all comparisons in surgical trials (appendix pp 143-146).
Trauma and Traumatic Brain Injury Patients
During 1977 to 2018, 10 RCTs compared different resuscitation fluids in trauma patients who required fluid resuscitations, and 4 RCTs in traumatic brain injury patients. (appendix pp 37-39). Patients’ mean age was 48.6 years, predominantly male (69.8%), and mean resuscitation study fluid was 5,481 mL among trauma trials. (appendix pp 82-86)
Ten trials with 5,076 participants had valid results on mortality in trauma patients, and differences in mortality were not significant between interventions in trauma patients. Balanced crystalloid required less volume of red blood cell transfusion than saline (MD, 350 mL; 95% CI 160 mL to 540 mL), and L-HES (MD, 964 mL; 95% CI 400 mL to 1527 mL). Four trials with 1,970 participants had valid results on mortality in traumatic brain injury patients, and saline reduced mortality than albumin with OR of 0.55 (95% CI 0.35-0.87) (appendix pp 103-114). The confidence ratings were low to very low among all comparisons in traumatic and traumatic brain injury trials (appendix pp 123-124, 128, 148-150).