Association between prognostic nutritional index and mortality in older Japanese patients with severe post-stroke dysphagia requiring nutritional support: a retrospective cohort study.
PNI was inversely associated with mortality in older Japanese patients with severe post-stroke dysphagia requiring PEG or TPN and provided incremental prognostic information beyond standard clinical variables, though prospective and external validation is required.
Key Findings
Results
Higher PNI as a continuous variable was independently associated with lower all-cause mortality in older patients with severe post-stroke dysphagia.
In fully adjusted multivariable Cox regression, each 1-point increase in PNI was associated with a 7% reduction in all-cause mortality (HR = 0.93, 95% CI: 0.90-0.97).
The cohort comprised 132 older Japanese patients with severe PSD requiring PEG or TPN, treated between January 2014 and January 2017.
Mean age was 80.7 years (78 women, 54 men).
Results
Patients in the highest PNI tertile had markedly lower mortality risk than those in the lowest tertile.
Compared with the lowest PNI tertile (T1), the highest tertile (T3) showed a 75% risk reduction (HR = 0.25; 95% CI: 0.10-0.60).
Associations were estimated using multivariable Cox regression and Kaplan-Meier curves.
Robustness was assessed with subgroup and sensitivity analyses.
Results
PNI showed numerically higher discrimination for 1-year mortality than CAR, but the difference was not statistically significant.
Time-dependent AUC for 1-year mortality was 0.771 for PNI versus 0.712 for CAR (p = 0.16).
Discrimination was evaluated using time-dependent ROC curves.
Results
Adding PNI to a clinical baseline model significantly improved discrimination and reclassification, whereas adding CAR did not.
For PNI: IDI = 0.055 (95% CI 0.010-0.104, p < 0.001); NRI = 0.471 (95% CI 0.126-0.957).
For CAR: IDI = 0.015 (p = 0.31), indicating no significant improvement.
Incremental predictive value was assessed using IDI and NRI over a clinical baseline model.
Conclusions
The authors conclude PNI offers incremental prognostic information in this population but needs prospective and external validation before use in risk stratification.
The study was a single-center retrospective cohort with 132 patients.
PNI is calculated from serum albumin and lymphocyte count and had been insufficiently investigated in older PSD patients.
The authors state utility for risk stratification requires prospective and external validation.
What This Means
This study looked at 132 older Japanese patients (average age about 81) who had severe swallowing problems after a stroke and needed artificial nutrition through a feeding tube into the stomach (PEG) or intravenous nutrition (TPN). The researchers examined whether the prognostic nutritional index (PNI), a score calculated from a blood protein (albumin) and white blood cell (lymphocyte) count, was linked to the risk of death.
Patients with higher PNI scores were less likely to die. Each 1-point increase in PNI was linked to a 7% lower risk of death, and those in the top third of PNI scores had about a 75% lower risk than those in the bottom third, after accounting for other factors. PNI also improved the ability of a model based on standard clinical information to predict one-year death, while an alternative marker, the C-reactive protein/albumin ratio (CAR), did not significantly improve prediction. However, PNI's advantage over CAR on its own was not statistically significant.
This research suggests PNI, which is based on routine blood tests, may help identify patients at higher risk. The study was retrospective, took place at one center, and was small, so the findings show an association rather than proof of cause. The authors note that the results need to be confirmed in prospective studies and other settings before PNI could be used for risk assessment.
Wu W, Wang J. (2026). Association between prognostic nutritional index and mortality in older Japanese patients with severe post-stroke dysphagia requiring nutritional support: a retrospective cohort study.. Frontiers in neurology. https://doi.org/10.3389/fneur.2026.1902644