Construction and preliminary validation of a nursing quality evaluation index system for enteral nutrition in preterm infants based on the Donabedian model

Background Enteral nutrition is vital for preterm infants in neonatal intensive care units (NICUs), yet a comprehensive, structured set of quality indicators for evaluating its delivery remains absent. This study therefore aimed to develop a standardized quality indicator system as a reference for enteral nutrition nursing in preterm infants. Methods Guided by the Donabedian structure–process–outcome framework, we drafted the indicator system from a systematic literature review and panel discussions, then refined it through a two-round Delphi survey with 18 experts to establish content validity and consensus. Indicator weights were derived using the Analytic Hierarchy Process (AHP). We then tested the system in a non-concurrent control design: the 14 process-quality indicators directly tied to enteral feeding were implemented as NICU interventions. Preterm infants admitted between July and December 2025 (control group) and between January and June 2026 (intervention group) were enrolled, with time to full enteral feeding and feeding intolerance as the primary outcomes. A retrospective dual-independent chart-based scoring was additionally performed to test inter-rater reliability and preliminary validity. Results Both consultation rounds achieved high response rates (90.0%, 18/20; 100.0%, 18/18). The expert authority coefficient (C r) was 0.848 and 0.880, and Kendall's W was 0.132 and 0.204 (both P < 0.001). The I- CVIs of the 39 tertiary indicators ranged from 0.67 to 1.00, with 36 reaching ≥0.78, the three items with I-CVI < 0.78 were retained after comprehensive panel review. The final system comprised 3 primary indicators, 11 secondary indicators, and 39 tertiary indicators. Retrospective dual-rater scoring was completed for 98 medical records, yielding an overall intraclass correlation coefficient ICC(A,1) of 0.982 (95% CI: 0.969–0.991). Because the raters were not blinded to group allocation, independent known-groups validity evidence could not be established. In the clinical validation, 98 preterm infants were enrolled (48 controls, 50 intervention); the time to full enteral feeding fell from (12.33 ± 4.68) days to (10.16 ± 4.04) days ( P < 0.05 ), and feeding intolerance declined from 16.7% (8/48) to 4.0% (2/50) (Fisher's exact test, P = 0.049). The overall intervention implementation rate was 70.19%. Conclusion The constructed index system demonstrated acceptable content validity ( S-CVI / Ave = 0.95) and excellent inter-rater reliability under chart-review conditions ( ICC(A,1) = 0.982, an upper-bound estimate given unblinded dual scoring. In preliminary validation, its implementation was associated with shorter time to full enteral feeding and a lower incidence of feeding intolerance. Given the single-center non-concurrent controlled design and the suboptimal implementation rate (70.19%), these changes should be regarded as concurrent with system implementation rather than attributable to the system itself.