Hidden nursing complexity within diagnosis-related groups (DRGs): a one-year retrospective study of standardized nursing diagnoses and actions among adult hospitalizations in Italy

Background: Diagnosis-related groups (DRGs) are used within prospective payment systems to classify hospitalizations and standardize reimbursement but may insufficiently capture nursing complexity. This study aimed to describe the variability in nursing complexity within the most prevalent DRGs among adult inpatients and to examine its relationship with DRG-specific length of stay thresholds and DRG weight. Methods: Adult hospitalizations discharged in 2022 from a large acute-care hospital in Rome, Italy, were analyzed. Nursing complexity was measured using counts of nursing diagnoses documented within 24 hours of admission and nursing actions recorded throughout hospitalization. Variability within and across DRGs was explored descriptively. Comparisons were conducted between hospitalizations within and exceeding DRG-specific length of stay thresholds. Linear regression models examined the proportion of variability in nursing complexity explained by DRG weight, both before and after adjustment for demographic and clinical variables. Results: The study included 14,169 hospitalizations across the 20 most frequent DRGs. Marked variability in nursing diagnoses and nursing actions was observed not only between DRGs but also within the same DRG, with counts ranging from 1 to 17 for nursing diagnoses and from 1 to 770 for nursing actions. Patients classified as medically similar under the same DRG exhibited substantial heterogeneity in nursing complexity. Hospitalizations exceeding DRG-specific length of stay thresholds showed consistently higher nursing complexity compared to those remaining within expected limits, both in terms of nursing diagnoses (mean 7.5, SD 4.3 vs. 4.0, SD 2.5; Welch's t(265.2) = -13.0, p < 0.001; Cohen's d = 1.0, 95% CI 0.9-1.1) and nursing actions (median 50, IQR 45 vs. 17, IQR 10; Mann-Whitney U = 311249, Z = -23.1, p < 0.001; r = 0.19). In univariable models, DRG weight showed modest associations with nursing diagnoses (β = 0.140, p < 0.001; R2 = 0.020) and nursing actions (β = 0.100, p < 0.001; R2 = 0.010). In multivariable models, the magnitude of these associations was markedly reduced, and DRG weight remained weakly associated with both nursing diagnoses (β = 0.030, p = 0.001; model R2 = 0.130) and nursing actions (β = -0.018, p = 0.040; model R2 = 0.097). Conclusions: DRGs do not adequately capture nursing complexity. Integrating standardized nursing data into hospital resource utilization and risk-adjustment models may improve the visibility of nursing care and support more accurate resource allocation in hospital financing systems.