Balanced Crystalloids versus Saline in Critically Ill Adults
Journal Club AINS, 2018 · paper · Clinical trials
A cluster-randomized trial in five intensive care units comparing balanced crystalloid fluids with saline in 15,802 critically ill adults. Major adverse kidney events within 30 days were less frequent with balanced crystalloids (14.3% vs 15.4%).
0 findings
None that passed a second check.
4 minor (typos and the like; most checked by the judge only)
Wording The Figure 3 "Overall" row prints "0.91 (0.83–0.99)" for the primary outcome. It has the same counts (1139/7942 vs 1211/7860) as the primary analysis, for which the text, abstract and Table 2 footnote give "marginal odds ratio, 0.91; 95% CI, 0.84 to 0.99; conditional odds ratio, 0.90; 95% CI, 0.82 to 0.99".
- p. 13 Figure 3, Overall row
0.91 (0.83–0.99)
- p. 6 Results, Primary outcome
marginal odds ratio, 0.91; 95% confidence interval [CI], 0.84 to 0.99; conditional odds ratio, 0.90; 95% CI, 0.82 to 0.99
- p. 16 Table 2, primary outcome
0.90 (0.82 to 0.99)
- p. 1 Abstract, Results
marginal odds ratio, 0.91; 95% confidence interval [CI], 0.84 to 0.99
Only the authors can say what is right here.
Judge's note: Viewed page 13 (Figure 3), the p.6 crop (Results) and the p.17 crop (Table 2 footnote ‡). The Overall row reads 0.91 (0.83–0.99), P 0.04, with the same event counts as the primary analysis. Its point estimate matches the marginal OR (0.91, CI 0.84–0.99), but its lower bound does not. Its CI is closer to the conditional estimate (0.90, CI 0.82–0.99), but its point estimate does not match that one. The figure legend and the Statistical Analysis section (p.5, which says subgroup analyses used 'a similar approach' and refers to the Supplementary Appendix, not in the files) do not say which model the Overall row comes from. A different model specification could account for it, so only the authors can say which estimate the row shows. The difference is small.
Second check: another model agreed
Report a problem with this finding
Wording Sample-size paragraph: the saline-group incidence is given as "approximately 15.0%", but the target is "a relative difference of 12% (an absolute difference of 1.9 percentage points)". 12% of 15.0% is 1.8 points.
- p. 4 Statistical analysis, sample size
difference of 12% (an absolute difference of 1.9 percentage points)
- p. 4 Statistical Analysis, sample-size paragraph
a relative difference of 12% (an absolute difference of 1.9 percentage points)
Only the authors can say what is right here.
Judge's note: Viewed the crop of p.4. The text gives the expected saline incidence as approximately 15.0% and then states that a 12% relative difference equals 1.9 percentage points. 0.12*15.0 = 1.8, and 1.9 would need a baseline of about 15.8%. Nothing in the article explains the gap (the earlier 22.0% baseline gives 2.64). The protocol (ref 13) and the supplement are not in the files, so the authors should say which figure is intended. This is a small inconsistency in a design parameter and does not affect any reported result.
Second check: another model agreed
Report a problem with this finding
Wording The creatinine threshold for stage 2 or higher AKI is worded "greater than 4.0 mg per deciliter" in the Table 1 footnote and the Figure 3 legend but "at least 4.0 mg per deciliter" in the Table 2 footnote.
- p. 15 Table 1 footnote **
both a value greater than 4.0 mg per deciliter (350 μmol per liter)
- p. 13 Figure 3 legend
was both greater than 4.0 mg per deciliter (350 μmol per liter)
- p. 17 Table 2 footnote ||
or at least 4.0 mg per deciliter (350 μmol per liter)
Judge's note: Checked all three crops. The baseline definition (first post-enrollment value) and the post-enrollment definition differ in other respects, but each cites KDIGO (ref. 22), and the strict and inclusive forms of the 4.0 mg/dl cutoff are inconsistent. Rests on outside knowledge: KDIGO uses ≥4.0 mg/dl.
Second check: another model agreed
Report a problem with this finding
Typo In Table 2, for ICU-free days the P value "0.94" sits on the header row while the odds ratio "1.00 (0.89 to 1.13)" sits on the Median row. For the other free-day outcomes both values sit on the header row.
- p. 16 Table 2, ICU-free days rows
1.00 (0.89 to 1.13)
Judge's note: The crop and page 16 confirm the misalignment. It is a layout issue only, and the meaning is recoverable.
Second check: none (judge only)