Relationship Between Nursing Diagnoses, Mortality and Length of Stay in the ICU: A Retrospective Cohort Study.

The study was retrospective and single-centre, so it can show association but not cause. Nursing diagnoses were recorded during routine care, the diagnostic questionnaire was optional, and diagnostic accuracy was not formally assessed, so counts may reflect documentation habits as well as patient condition. Nurse-level variables were not assessed, and the primary nursing model with fortnightly bed rotation may have introduced variability in documentation practices. The PROCEnf-USP system may not have reflected current NANDA-I taxonomy updates, potentially affecting which diagnoses were recorded. Only diagnoses recorded at ICU admission were analysed, not the daily re-evaluation described in routine care. Nursing workload was measured only for the first 24 hours, so it does not show workload across the ICU stay. Patients with ICU length of stay under 24 hours or incomplete nursing diagnosis records were excluded, which may limit generalisability. The decision-tree analysis had low sensitivity (19.6%) and a cross-validation error of 19.03%, so it is not a ready-made prediction model.

The authors declared no conflicts of interest. Funding came from CAPES, CNPq and Brazil's Ministry of Health, including support to a nursing residency programme.

Do not conclude that recording more nursing diagnoses causes death or longer stay. The length-of-stay association was not significant after adjustment (p = 0.072), and the study is retrospective, so it shows association only. Do not use the decision tree as a validated risk score because its sensitivity was only 19.6%.