Axis Imaging The Script
The Script · Published by Axis Imaging, Melbourne

Friday 13 November 2026

3 stories today

1. Over a third of infant peanut allergy diagnoses are wrong

Of 778 Australian infants with a peanut reaction in their history and a positive allergy test, 271 ate their way through a supervised peanut challenge without reacting. That's 34.8% of babies carrying an allergy label they didn't earn.

The data comes from ADAPT, the national infant peanut oral immunotherapy program. It's run by Murdoch Children's Research Institute, the National Allergy Centre of Excellence and 10 paediatric public hospitals across Victoria, NSW, Queensland, SA and WA. Since 2024, eligible infants get a threshold oral food challenge before starting OIT. To qualify, they need a CoFAR grade 1 or 2 reaction before 12 months, plus an SPT of at least 3mm or sIgE of at least 0.35kUA/L. The challenge is seven escalating doses, from 15mg to 1300mg, given 15 minutes apart.

The tests helped only at the extremes. Every infant with Ara h 2 sIgE of 5.0kUA/L or more, or an SPT of 12mm or more, reacted. In between, history plus a positive test wasn't enough to call it. Neither test predicted how severe a reaction would be. Of 29 infants with a grade 3 history who had this challenge, 48.3% tolerated peanut.

Challenges carry real risk. Anaphylaxis occurred in 7.8% of all challenges. All were managed safely, which is why the authors want challenges done in centres equipped for severe reactions. For infants who did react, the challenge threshold set their OIT starting dose, and 97.5% started successfully.

Peanut allergy affects about one in 30 Australian kids. That adds up to a lot of families avoiding peanut, and possibly facing years of treatment, without needing to.

The takeaway: When an infant comes in with a reaction and a positive SPT or sIgE, tell parents the diagnosis isn't confirmed yet. Refer to a paediatric allergy service for a supervised challenge rather than locking in lifelong avoidance.

Read the source

2. Cochrane finds cannabis-for-insomnia evidence too weak to draw conclusions

Pool the trials of medical cannabis for insomnia and you get an extra 7.68 minutes of sleep a night, give or take half an hour either way. That's roughly one snooze-button press, and five of the nine trials behind it were run in Australia.

The Cochrane review, published 13 October, combined nine studies from 2010 to 2025 covering 537 adults with diagnosed insomnia. Most enrolled fewer than 50 people and only two passed 100. Eight compared cannabis oil, capsules or soft gels against placebo. One used amitriptyline.

Sleep quality and satisfaction (ISI, PSQI) showed a slight, non-significant benefit across 374 participants. Sleep onset latency moved by about four minutes and daytime functioning barely shifted. Certainty for the main outcomes was rated very low, and five of the placebo-controlled trials carried a high risk of bias.

Adverse events were the one place the numbers got loud. Cannabis had an odds ratio of 11.25 versus placebo, though that came from just three studies and 98 people, at low certainty.

Lead author Dr Lucia Varela, a neurologist at the University of Manitoba, was clear: "This does not mean that medical cannabis is ineffective." It means nobody can yet say it works. Few trials included older people with comorbidities, so the evidence says even less about them.

Five placebo-controlled trials were still running at the April 2025 search cut-off, so a future update could change the picture.

The takeaway: When a patient asks about cannabis for sleep, tell them the best available review found no reliable evidence it helps and a low-certainty signal of more side effects. Watch for those five ongoing trials before changing your answer.

Read the source

3. Some antidepressants may maintain depression improvements better than others

Fluoxetine just lost a head-to-head on staying power. A new Lancet Psychiatry review found escitalopram, duloxetine and agomelatine were all more likely than fluoxetine to keep people's depression improvements going.

The researchers reviewed 115 randomised controlled trials involving more than 30,000 people, counting six months or more as long term. Their main analysis pooled 35 trials that compared one antidepressant directly against another. People on escitalopram or duloxetine were also less likely to drop out than those on fluoxetine, which the authors used as their measure of acceptability.

Now the fine print. More than half the trials in the main analysis carried a high risk of bias, and some drugs had very few studies behind them. Only six trials followed people beyond six months. Australians take antidepressants for an average of four years, so "long term" here really means the first half-year.

The local context makes this land. Escitalopram is already Australia's most prescribed antidepressant, ranked 13th of all PBS-dispensed prescriptions in 2024-25, so plenty of your patients are on the one that scored well. Duloxetine came in at 45th. Agomelatine isn't on the PBS, so its strong showing comes with a non-PBS script attached.

With 3.8 million Australians dispensed an antidepressant in 2024-25, expect "am I on the good one?" in clinic. Guidelines say continue for at least six months, preferably 12, after an acceptable response. One Australian study found nearly a quarter of people were still getting scripts two years later.

The takeaway: Don't switch anyone who is doing well, and the authors say the same. When the question comes up, check whether the patient is past the six to 12 month mark after response and whether a gradual, supervised taper is worth discussing.

Read the source

Also noted

Get tomorrow's issue by email at 6:30am.