ICU diaries are meant to prevent PTSD. Correcting for publication bias, they didn't.

A meta-analysis of 21 studies and 2,180 ICU survivors found that once results were adjusted for publication bias, ICU diaries had no statistically significant effect on PTSD incidence (RR 0.73, CI 0.47-1.12) or severity. Modest benefits for anxiety and depression survived.

Many intensive care patients come away with fragmented, sometimes delusional memories of their stay, and a widespread nursing intervention tries to help: staff and families keep a diary during the admission that the patient can later read to rebuild a coherent account. It is cheap, humane, and widely believed to reduce post-ICU PTSD. Researchers led by Wenyu Zhang tested that belief across seven databases, pooling 21 controlled studies (17 randomised, 4 not) covering 2,180 survivors, and did something most earlier reviews skipped: they formally corrected for publication bias, the tendency for studies finding nothing to go unpublished. Before adjustment, the picture looked favourable, and a trial sequential analysis crossed the boundary for benefit. After trim-and-fill correction, it did not hold. PTSD incidence gave a risk ratio of 0.73 with a confidence interval from 0.47 to 1.12, which includes no effect at all, and PTSD severity gave a standardised mean difference of -0.23 with an interval from -1.20 to 0.75, consistent with meaningful harm or meaningful benefit. Modest benefits for anxiety and depression symptoms did survive the correction, and a subgroup analysis suggested diaries written by healthcare professionals without family involvement worked better for PTSD severity than family-written ones. The authors' own conclusion is appropriately narrow: diaries belong in a broader post-ICU support package, not as a standalone treatment for any one outcome. Several caveats limit how far even the null result travels. Heterogeneity ranged from 0% to 95%, so the studies frequently disagreed. Nobody can be blinded to whether they received a diary, which builds in performance bias. The subgroup comparisons used unadjusted data, so they inherit the same publication bias found in the main result. And a substantial share of the studies came from Chinese databases, where ICU visiting policies and family involvement differ from Western practice.