Stanley et al. (2015) argued that the normal fall in glucose after birth is itself a mild, self-limiting form of hyperinsulinism, in which the glucose threshold for insulin release is set lower than in later life. This explains why ketones are low in healthy newborns during the first day and why a single low glucose in that window says little about underlying disease. The Pediatric Endocrine Society recommendations build on the same reasoning: evaluation for a persistent disorder is best deferred until after the first 48 hours, unless hypoglycaemia is severe, and the diagnosis rests on metabolic markers measured at the time of hypoglycaemia (Thornton et al., 2015). A prolonged but ultimately self-resolving form associated with perinatal stress, growth restriction or birth asphyxia lies between the two and may last for weeks (Hoe et al., 2006).

The commonest and most severe forms result from loss-of-function variants in ABCC8 and KCNJ11, which encode the two subunits of the ATP-sensitive potassium channel of the β-cell; the first such variants were identified in the sulfonylurea receptor gene by Thomas et al. (1995). Activating variants in GLUD1 cause the hyperinsulinism and hyperammonaemia syndrome (Stanley et al., 1998), and several rarer dominant forms are known. A large series showed that the genotype predicts the response to diazoxide and the pancreatic histology (Snider et al., 2013). Recessive channel variants on both alleles cause diffuse disease. A single paternally inherited recessive variant, combined with somatic loss of the maternal allele in a region of the pancreas, causes a focal lesion that can be cured by limited resection (Arnoux et al., 2011). The diagnostic pathway therefore has two aims: to establish hyperinsulinism quickly, and to identify the infant with focal disease.

The pathway begins when hypoglycaemia departs from the transitional pattern. The features that should prompt it are a low glucose that is severe, symptomatic, recurrent despite adequate feeding, or still present after 48 hours; a need for intravenous glucose; and hypoglycaemia in an infant with no conventional risk factor. Macrosomia without maternal diabetes, a family history of hypoglycaemia, and physical signs of Beckwith-Wiedemann syndrome or of hypopituitarism add weight. Bedside glucose meters are imprecise at low concentrations, so values that will drive decisions should be confirmed on a laboratory or blood gas analyser (British Association of Perinatal Medicine, 2017).

Treatment is never delayed for the sake of diagnosis. Glucose is restored and maintained according to the unit protocol. The pathway asks for one additional act: calculation and documentation of the glucose infusion rate needed to keep the infant normoglycaemic, expressed in milligrams per kilogram per minute and including enteral intake where relevant. A healthy newborn liver releases roughly 4 to 6 mg/kg/min. A requirement well above that range points strongly to excess insulin action, and the British framework uses a figure above 8 mg/kg/min as a trigger for investigation (British Association of Perinatal Medicine, 2017). High requirements often exceed what a peripheral vein tolerates, and the need for central access should be anticipated.

The diagnosis depends on blood taken while the plasma glucose is below the threshold set in the guideline in use, which for the Pediatric Endocrine Society is 2.8 mmol/L (50 mg/dL) (Thornton et al., 2015). This may occur spontaneously or, in a stable infant under close observation, during a supervised reduction in glucose delivery planned with specialist advice. The sample includes laboratory glucose, insulin, C-peptide, β-hydroxybutyrate, free fatty acids, cortisol, growth hormone, lactate, ammonia and an acylcarnitine profile, with urine for organic acids. Tubes and request forms should be assembled in advance and kept at the cot side, because the opportunity is brief and the glucose must then be corrected at once.

Immediately after the critical sample, glucagon is given by the route and in the amount specified by the local protocol, and glucose is measured at intervals over the following 30 to 40 minutes. In a fasted infant without hyperinsulinism the liver has little glycogen left to mobilise. A rise of more than 1.7 mmol/L (30 mg/dL) shows that glycogen has been inappropriately retained and is strong evidence of insulin excess (Thornton et al., 2015). The test doubles as treatment of the hypoglycaemic episode.

Hyperinsulinism is established when, during hypoglycaemia, ketones and free fatty acids are suppressed, the glucagon response is exaggerated and the glucose requirement is high. Insulin itself is an unreliable marker. It is often only just detectable, and assays vary; Ferrara et al. (2016) showed that suppressed β-hydroxybutyrate and free fatty acids and the glucagon response identify the condition more dependably than the insulin concentration. Raised ammonia suggests the GLUD1 form. Low cortisol or growth hormone in a single critical sample has poor specificity for true deficiency and should lead to formal testing, not to a diagnosis (Kelly et al., 2008). Ketotic hypoglycaemia, raised lactate or an abnormal acylcarnitine profile redirect the work-up towards other metabolic and endocrine causes. The interpretation is written down with the differential diagnoses considered.

Once hyperinsulinism is established the infant should be discussed with a centre experienced in the condition, and this is true whatever the apparent severity. Diazoxide, which holds the potassium channel open, is the first-line drug, and a trial lasting several days is the conventional test of responsiveness (De Leon et al., 2024). Prescribing follows specialist advice and the product information. The pathway requires that responsiveness be judged by a stated criterion, namely the ability to maintain normal glucose on ordinary feeds through an age-appropriate fast, and that adverse effects be looked for. Fluid retention is expected, and pulmonary hypertension has been reported in neonates (Herrera et al., 2018). Most infants with channel variants that abolish function do not respond.

Blood from the infant and both parents is sent for analysis of ABCC8 and KCNJ11 as soon as diazoxide unresponsiveness is suspected, and without waiting for the trial to end. Rapid sequencing of these genes changes management, because a single paternal recessive variant predicts focal disease with high probability (Banerjee et al., 2011). Such infants are referred for positron emission tomography with 18F-fluoro-L-DOPA, which localises focal lesions accurately enough to guide surgery (Hardy et al., 2007). Infants with diffuse disease continue medical treatment under specialist care.