Nine key informant interviews were conducted with one pediatrician, five nurses, two clinical officers and one newborn-unit manager, with 2–10 years of neonatal-care experience (Table 5). Thematic analysis generated eight interrelated care-related themes—timeliness of admission and triage, staffing and workload, warm-chain maintenance, infection-prevention practices, drug and equipment availability, referral coordination, monitoring and documentation, and caregiver delays and transport barriers (Table 6)—together with cross-cutting subthemes around the fragile “first hour” of care, pre-referral stabilisation and the role of mothers as co-providers of care through Kangaroo Mother Care and feeding support.

Providers consistently described the first hour after birth or arrival as decisive for LBW neonates, when delayed warming, oxygen support, glucose assessment or antibiotics could worsen prognosis. Staffing and workload were among the most prominent concerns; as one nurse explained, care could be slow because “there is low staffing which makes the care of the babies difficult in the facility from time of triage to admission of the babies is slow.” Warm-chain maintenance was repeatedly linked to hypothermia and to system constraints, including unreliable power: “Due to unreliable power supply within the facilities it makes the care for the babies difficult as incubators stop working hence babies start to lose heat.” Infection prevention was described as difficult in a crowded unit with intermittent supplies, with sepsis attributed partly to shared incubators and inconsistent decontamination. Commodity and equipment gaps—particularly shortages of continuous positive airway pressure (CPAP) machines and essential drugs—were said to delay management, and referral was viewed as poorly coordinated: “There are always delays for admission of babies… from other lower facilities since the referral system is not clear, [and the] mode of transport to the hospital also is not efficient as there are fewer ambulances in the county.” Caregiver-level delays were linked to limited danger-sign recognition, family decision-making and cultural beliefs, with some families interpreting LBW as “a curse or bad omen.” An overarching theme was that adverse outcomes reflected the interaction between clinical skill and system readiness.

Providers framed the earliest period of care as decisive. A pediatrician argued that newborn care should be treated as an emergency service, “because for these small babies, even one hour of delay can change the outcome” (KII 01), a view a clinical officer echoed in insisting that “the first hour after admission should be treated as critical” (KII 07). Competing admissions repeatedly eroded this window; as one nurse put it, “you may be stabilizing one baby on oxygen, then another baby arrives from theatre or maternity” (KII 05).

Staffing pressure was tied directly to monitoring and documentation. A nurse observed that during emergencies “documentation suffers because the priority becomes keeping the baby alive first” (KII 02), while a more experienced colleague cautioned that “you cannot manage these babies like ordinary patients because small changes matter” (KII 04). The unit manager conceded that although staff were skilled, “burnout is real” (KII 09).

Thermal care, equipment and crowding were closely intertwined. When incubators were full, staff improvised—“a stable baby may have to be moved to KMC earlier so that a more critical baby can use an incubator” (KII 03)—and congestion undermined infection control, because “when the unit is crowded, it becomes difficult to separate infected and non-infected babies properly” (KII 04).

Referral gaps produced cold, undocumented arrivals. A pediatrician described babies “referred on motorbikes or public vehicles, wrapped in ordinary clothes, with no oxygen or warmth,” so that “by the time they arrive, the situation is already worse” (KII 01), and a clinical officer noted that such neonates often “arrive with several problems at once: cold, hypoglycemic, and breathing poorly” (KII 08). Referral notes were frequently too sparse to guide management—one nurse received a note that “only [said] ‘premature baby’ without temperature, blood sugar, treatment given, or Apgar score,” which “makes us repeat assessment and lose time” (KII 02)—leading the unit manager to insist that “a small baby should not be transported cold” (KII 09).

Delays often began before admission. Providers attributed these to weak danger-sign recognition—“poor feeding, fast breathing, and cold skin may not be taken seriously early” (KII 08)—and to family decision-making and cost, with families delaying “because they do not know danger signs or they are looking for money” (KII 04). The mother’s own state then shaped care on the ward: warm-chain maintenance was easier “when the mother understands KMC,” but “if the mother is tired, post-operative, or afraid to hold a small baby, it becomes difficult” (KII 03).

The joint display (Table 7) shows convergence between the two strands. The near-universal burden of respiratory and thermal-metabolic complications aligned with provider accounts of a fragile first-hour stabilisation window; the independent effect of birth weight matched providers’ emphasis on the intrinsic vulnerability of the smallest neonates; the independent maternal-PIH effect was consistent with provider observations linking maternal antenatal conditions to fragile newborns; and the warm-chain and system themes explained why thermal and care-readiness variables were statistically entangled with illness severity. Together, the strands indicate that neonatal vulnerability and maternal hypertensive disease set the level of risk, while the readiness and timeliness of newborn-unit care shaped whether that risk translated into a severe outcome. This reading was crystallised by the unit manager, who observed that “staff can do their best, but without enough personnel, equipment, and referral coordination, some preventable complications will continue” (KII 09).