**Background:** Mother's own milk (MOM) is the optimal nutrition for preterm infants, reducing risks of necrotising enterocolitis (NEC), late-onset sepsis (LOS), and bronchopulmonary dysplasia (BPD). However, MOM shortfall is common, and units vary widely in whether they use donor human milk (DHM) or preterm formula (PF) as the supplement. Observational data suggest DHM availability may influence maternal breastfeeding behaviour, but no randomised trial has tested this. The HUMMINGBIRD study addresses whether longer DHM exposure increases breastfeeding rates at discharge.
**Methods:** This is a non-blinded, pilot RCT conducted in two tertiary neonatal units in North East England (second site added May 2022). Infants born <33 weeks' gestation or with birth weight <1500 g, whose mothers intend to express breast milk and accept DHM, are randomised within 7 days of birth. Exclusion criteria: major congenital abnormalities, prior formula exposure. Randomisation uses a web-based tool (sealedenvelope.com) with stratification by gestation (<28 weeks yes/no) and multiple birth status. Twins/triplets are co-randomised to the same arm.
**Intervention arm:** DHM is used for any MOM shortfall until 36 weeks' corrected age (ca) or discharge, but only if the mother continues expressing. If no MOM is received for 1 week or the mother stops expressing, DHM is discontinued and PF is used.
**Control arm:** DHM is used for MOM shortfall only until full feeds are achieved (150 mL/kg/day for 48 hours); thereafter PF is used.
Both arms receive standard care; human milk feeds are fortified once full feeds are reached.
**Primary outcome:** Any breast feeding or active maternal milk expression at 36 weeks' ca or discharge.
**Secondary outcomes:** Weekly growth (weight, length, head/mid-arm/thigh circumference), neonatal morbidities (NEC ≥Bell stage 2, confirmed/suspected LOS, severe BPD, ROP, IVH), length of stay, total volumes of MOM/DHM/PF received, age at fortifier initiation, feeding type at discharge, and post-discharge feeding and growth at 6 and 12 weeks. Feasibility outcomes include rates of declination, dropout, loss to follow-up, and protocol deviations. Maternal questionnaires (BSES-SF at ~day 10 and 36 weeks ca; EPDS at ~day 10) are administered. Qualitative semi-structured interviews with 15–20 mothers explore perceptions of DHM and barriers/facilitators to expressing MOM; thematic analysis will be used.
**Sample size:** To detect an increase in breastfeeding at discharge from 35% (local rate in 2019) to 60% (UK national average), 58 infants per group are needed. Assuming 10% mortality and up to 20% transfer before 36 weeks ca, 130–156 infants need to be recruited (80% power, 18–24 months recruitment).
**Analysis:** Intention-to-treat primary analysis; additional per-protocol and singleton-only analyses. Categorical data: χ² or Fisher's exact test. Continuous data: Student's t-test or Mann-Whitney U test. Logistic regression for morbidities. Two-tailed tests, p<0.05 significant. Qualitative data: reflexive thematic analysis.
**Clinical Implications:** This pilot trial will determine whether a full-scale RCT is feasible and whether longer DHM exposure can increase breastfeeding rates at discharge. If confirmed, it would support a shift in NICU feeding policies toward extended DHM use, potentially improving maternal and infant outcomes. The qualitative component will clarify mechanisms—whether DHM affects maternal beliefs, staff behaviours, or both—and whether an RCT design is acceptable to mothers. The study is registered (ISRCTN57339063) and approved by Nottingham 2 Research Ethics Committee (IRAS 281071); recruitment began 7 June 2021.