Insulin poisoning — a screening assay used as forensic proof
The prosecution’s claim: Blood samples from two infants — Children F and L — returned results suggesting raised insulin with low C-peptide. Normally insulin and C-peptide are released together by the pancreas. A high-insulin-low-C-peptide pattern, the prosecution argued, is only explicable by insulin administered from outside the body. The jury was told this was proof of deliberate poisoning.
Prosecution claim
Blood samples from two infants — Children F and L — returned results suggesting raised insulin with low C-peptide. Normally insulin and C-peptide are released together by the pancreas. A high-insulin-low-C-peptide pattern, the prosecution argued, is only explicable by insulin administered from outside the body. The jury was told this was proof of deliberate poisoning.
Counter-evidence
The Roche Cobas immunoassay used is a screening test. Its own manufacturer's guidance requires confirmation by mass spectrometry before a result can be treated as diagnostic of exogenous insulin. That confirmation was never done. Independent endocrinologists (including Adel Ismail and contributors to science4justice.nl) have shown the assay is prone to false positives in neonates because of interfering antibodies and cross-reactivity. The Shoo Lee Panel went further than the assay critique. It found that the insulin/C-peptide ratios in Children F and L were within the range normal for preterm infants — a pattern that does not prove exogenous insulin at all — and it attributed the babies' hypoglycaemia to natural and iatrogenic causes: for Child F, sepsis, prematurity, intrauterine growth restriction, a tissued long line and poor management; for Child L, preterm birth and severe intrauterine growth restriction, inadequately managed.
Key point: A screening immunoassay was never designed for forensic use. In every other British criminal case involving insulin, confirmatory mass spectrometry is performed. That did not happen here.
What the jury heard
Biochemists from the Royal Liverpool laboratory described the Roche immunoassay results as showing an insulin level inconsistent with endogenous pancreatic release. The Crown's witnesses told the jury the pattern left no explanation other than exogenous insulin in the TPN bag.
What the Panel says
The Panel concludes that the insulin/C-peptide ratios in Children F and L were within the range normal for preterm infants and do not prove exogenous insulin. It attributes Child F's prolonged hypoglycaemia to sepsis, prematurity, intrauterine growth restriction, a tissued long line and poor management, and Child L's to preterm birth and severe intrauterine growth restriction, inadequately managed. Separately, the screening result was never confirmed by the mass spectrometry the assay manufacturer recommends for forensic use — without which a result cannot reliably discriminate exogenous insulin from interference.
What independent experts add
- Dr Adel Ismail, a clinical biochemist, has published detailed commentary on interference and false-positive rates in the Cobas assay in neonates.
- Prof. Geoff Chase (University of Canterbury NZ) has modelled the physiological plausibility of the reported values and concludes several alternative explanations are consistent with the observed data.
- science4justice.nl catalogues every known false-positive case in the assay's published literature, and identifies the specific Royal Liverpool laboratory protocol change: the 2010 protocol (used at the time of testing) did not state the lab could not diagnose exogenous insulin, while the 2012 protocol did. For forensic purposes this matters.
- The Roche Cobas immunoassay also reports positive where insulin auto-antibodies are present (transiently from a diabetic mother in utero, or chronically). Adrenal suppression, infection, certain genetic variants, liver disease, kidney disease, certain drugs, and alcohol can all generate false 'factitious hyperinsulinism' readings.
- The sample processing protocol requires centrifugation within 30 minutes and freezing at -20°C to prevent degradation. Samples delayed beyond 6 hours prior to centrifuge are not usable forensically, regardless of whether they were 'frozen' on arrival at Liverpool.
- Treatment for hypoglycaemia itself suppresses C-peptide. If the test was taken after repeated dextrose infusions (as it was), the C-peptide reading does not reflect the baby's baseline.
- The numerical insulin value reported (4,657 pmol/L) is on the order of values seen in adult attempted-suicide patients who injected 200+ units of insulin. It is physiologically implausible on the prosecution's own theory of a 0.6 ml spike of insulin into a slow-running TPN bag.
- The correct lab for forensic exogenous-insulin diagnosis in the UK is Guildford, not Liverpool. Neonatal samples should go in a red-top, heparinised, separated collection tube; Liverpool's protocol used gel tubes.
- No TPN bags were retained or tested; the prosecution theory required LL to have 'spiked' TPN stock in the ward fridge, accessible to all nursing staff on the unit.
- The Panel's 2025 Joint Expert Witness Insulin Report on Babies F and L sets out each of these problems in technical detail and concludes the insulin evidence cannot support a criminal finding of exogenous insulin administration.