Nature: NPAR Linked to ICU Mortality but Offers No Incremental Discrimination Over Albumin Alone
A retrospective cohort study using the MIMIC-IV database found the neutrophil percentage-to-albumin ratio (NPAR) associated with 30-day mortality in intensive care patients but offered no incremental discrimination over albumin alone, Nature reported.
The Tech TL;DR:
- NPAR showed an adjusted odds ratio of 1.15 per 5-unit increase for mortality between hour 24 and day 30.
- Discrimination AUC for NPAR was 0.627 versus 0.631 for albumin alone (p=0.639).
- Authors cite residual selection bias and lack of external validation as barriers to clinical use.
Researchers analyzed 5,130 adult ICU patients from the Medical Information Mart for Intensive Care IV (MIMIC-IV) database. The primary analysis required paired albumin and neutrophil percentage measurements taken within the first 24 hours of admission and no more than two hours apart. Patients had to survive the first 24 hours; follow-up began at that landmark and ended at day 30. The cohort recorded 1,082 deaths (21.1%) during the follow-up window.
A four-knot restricted cubic spline logistic model adjusted for age, sex, selected comorbidities, first-day SOFA score, and sepsis status. The spline model fit the data better than a linear model (AIC 4,451.4 vs. 4,475.3; likelihood-ratio chi-square = 27.85, df = 2, p < 0.001). The adjusted linear odds ratio was 1.15 (95% CI, 1.10–1.19) per 5-unit increase in NPAR.
Despite the statistical association, NPAR alone achieved only modest discrimination with an area under the receiver operating characteristic curve (AUC) of 0.627. This did not outperform albumin alone, which posted an AUC of 0.631 (p = 0.639). The evidence of non-linearity depended on distribution tails, and the incremental discrimination was characterized as small.
The authors concluded that residual selection bias, covariate-timing uncertainty, confounding, and the absence of external validation preclude clinical deployment of NPAR as a standalone prognostic tool. The requirement for temporally aligned blood draws within a two-hour window adds a logistical constraint not routinely met in standard ICU workflows.