DKA at Diagnosis: Why Infants Slip Through the Net

The under-1s carry the peak, and it has not improved

Around a third of children worldwide present in DKA at the moment they are first diagnosed with T1D. The proportion sits highest in the youngest band: roughly sixty percent of children diagnosed under their first birthday in the largest international study we have (Cherubini 2020, Diabetologia, 13 countries, 11 years of data). The figure falls through preschool age and rises again at ten to twelve years. The international band shows wide variation between countries, from around twenty percent at the lowest to over forty percent at the highest, but the under-1s peak holds across every health system that has been measured.

The harder line in the Cherubini paper is the trajectory: the rate did not improve over the eleven years of follow-up, and rose slightly in some countries. Public-awareness campaigns alone have not solved the problem. The pattern crosses country borders, sits more heavily in children from ethnic-minority backgrounds, and tracks inversely with the country’s overall Human Development Index. None of those facts implicate any individual family; they describe a health-system signal that is everyone’s problem and no individual’s fault.

Why infants slip through the net

The textbook signs of T1D, the polyuria-polydipsia-weight-loss triad and the fruity breath, were written for children who can tell you they are thirsty and old enough to wet the bed at night when they were not the night before. Infants present differently. Vomiting, sleepiness, faster breathing, fewer wet nappies that all look about right because the toddler is a toddler. The signal that something has changed is often a feeling more than a fact: she is not herself, the milk is not going down, the cot is too quiet at the wrong time of day. Parents are usually right when they have that feeling; the system around them is not always set up to act on it before the bloods are drawn.

The contributors compound. Out-of-hours general practice triages on textbook signs. The first encounter often books a follow-up rather than a same-day blood test. Ethnic-minority families and families in lower-Human-Development-Index countries are diagnosed later more often, a finding that crosses every health system measured (Cherubini 2020 again). None of this is a moral failure; it is a structural one. What helps individual families is to know the signs that matter for an infant (vomiting plus sleepiness plus changed breathing in a child who has been off their feeds is enough), and to know that a finger-prick glucose at the GP surgery is a five-minute test that costs nothing and rules T1D in or out on the spot.

What screening has actually shifted

The cleanest evidence that the under-2s peak can move comes from the TEDDY study (Elding-Larsson 2011, Diabetes Care). TEDDY enrolled children with high-risk HLA genetics from three months of age, gave families three-monthly follow-up with autoantibody testing and OGTT, and compared DKA-at-diagnosis rates against four national registries (DPV in Germany and Austria, SEARCH in the United States, Swediabkids in Sweden, and the Finnish Pediatric Diabetes Register). In the under-2s, the registries showed DKA at diagnosis in around fifty-four percent of new cases; in TEDDY it was around fifteen percent. In the under-5s, the registries sat near thirty-two percent; TEDDY at around eleven. The difference is not pharmacological. It is structural: the family already knows what to watch for, the child is already being seen every three months, and the OGTT and antibody tracks identify Stage 3 onset before the crisis.

That is a different conversation about screening than the public-awareness one. Awareness asks the wider population to recognise rare signs in time; structured longitudinal follow-up asks a smaller, defined population to be watched closely with tests that detect the disease before symptoms. The two are not in competition. Both are slow to scale. NICE, ISPAD, and the European screening collaboratives are all currently working on the case for general-population paediatric autoantibody screening, with cost-effectiveness in NHS settings as the open question. The TEDDY data are the proof of concept; the implementation is national policy and not a single family’s job.

Once the child is admitted: what is actually evidence-based

The two pieces of paediatric DKA care that families ask about most often are the fluid plan and the cerebral-oedema risk. Glaser 2001 (NEJM, multicentre case-control, 61 cases of cerebral oedema and 174 matched controls) showed that the strongest predictors at presentation were lower partial pressure of carbon dioxide and higher blood urea nitrogen, both signs of severity at the front door rather than treatment choices. Bicarbonate treatment was associated with around a four-fold higher risk of cerebral oedema after adjustment, which is why current paediatric DKA pathways reserve bicarbonate for life-threatening hyperkalaemia or pH below 6.9. Fluid rate, sodium rate, and insulin rate were not associated with cerebral oedema after adjustment in that case-control work.

The PECARN DKA FLUID trial (Kuppermann 2018, NEJM, 1,389 episodes across 13 paediatric emergency departments) was the four-arm randomised test of the fluid question that the Glaser data had reframed. Children with paediatric DKA were randomised to four combinations of fluid rate (rapid versus slow) and saline composition (0.9 percent versus 0.45 percent). Glasgow Coma Scale decline below 14 occurred in around three and a half percent of episodes; clinically apparent brain injury in just under one percent; and there was no statistically significant difference between any of the four arms. Memory and IQ at two to six months post-recovery did not differ either. The practical message is that current paediatric DKA pathways are robust to the fluid choices the team makes within standard ranges; the variable that matters most is severity at presentation.

What the family can take to the team

If a parent or carer is worried that an infant or young child has new vomiting plus sleepiness plus changed breathing in a child who has been off their feeds, the right next step is a same-day clinical review. A finger-prick glucose at the GP surgery (or in an emergency department triage room) takes less than five minutes and is enough to rule T1D in or out. The figure does not need to be requested politely; it can be asked for plainly: “please can we have a finger-prick glucose before we go home today?” The team is set up to do this; the system rewards parents who ask early.

If a sibling has been diagnosed and the family is wondering about screening, the conversation now exists. UK families can ask their paediatric diabetes service about the ELSA general-population screening study, the Type 1 Diabetes Risk Test, or the local pathway for sibling autoantibody screening. The medical case for screening is still being made at policy level; the family case for asking the question early is already settled. The next part of this guide goes deeper on what happens if a screen comes back at Stage 2.

References for this part