This post is written by Dr Niall Morris
Prednisolone is about as tried and tested a treatment as we have in medicine. We all know how fantastic it is for settling down exacerbations of asthma. So much so that, when we see a child with a moderate exacerbation of asthma who’s fit to go home with some extra treatment, we never give it a second thought before reaching for the prescription pad. It’s a no brainer.
But have you ever considered that there might be a better way? Maybe taking a 3-day course of prednisolone isn’t ideal for children who are reluctant to take their medication. Wouldn’t a single dose be better? And prednisolone does unfortunately tend to make some children vomit. What if dexamethasone, which has a longer half life and might cause less vomiting, was just as good? We use it all the time for croup, after all.
Let’s take a look at a study that set out to answer just that question. You can find it at this link…
What was the objective of the study?
The authors wanted to know if a single dose of dexamethsaone might have the same efficacy as a course of prednisolone for children presenting to the ED with asthma exacerbations.
What was the primary outcome?
A blinded clinician assessed the Paediatric Respiratory Assessment Measure on day 4. This seems reasonable – it’s a validated tool for this purpose.
Who was included?
The study used rather broad inclusion criteria, with children as young as two included and the definition of what constituted an asthma exacerbation. A patient could be included if they demonstrated any of the following criteria: dyspnoea, wheeze, acute cough, increased work of breathing, increased requirement for ß2-agonist from baseline use, or SaO2 less than 95%.
In short, if a child felt short of breath or had a cough they could be enrolled and given steroids despite being otherwise well. Giving steroids to patients with mild exacerbations is debatable and by including this group in the study it increases the likelihood of finding non-inferiority.
How were the patients randomised?
Patients were randomised in permuted blocks of 12 eg. ABAABAAABBBB. Randomising patients in large blocks of 12 (instead of, say, 4) reduces the chances of selection bias as it makes it harder for clinicians to guess the arm to which the next patient will be allocated, although it’s not impossible. In this study, the recruiting clinician took the next available envelope from a locked storage cupboard and prescribed the particular medication. Envelope randomisation is vulnerable to all sorts of chicanery, ranging from holding it up to the light or even just opening it and having a peak, an issue the REVERT trial ingeniously overcame by having investigators sign and date across the seal before opening the envelope.
What were the treatment arms?
Subjects in one arm received a single dose of oral dexamethasone 0.3 mg/kg (maximum dose 12 mg) while those in the other arm were given a 3-day course of once-daily prednisolone 1 mg/kg per day (maximum dose 40 mg)
What patients were included?
There were 226 patients in this study. Baseline demographic characteristics were similar between the two groups apart from sex with more male patients in the prednisolone group (74.6% versus 61.8%; P=.03).
What did the results show?
The mean change in PRAM score from arrival in the ED to review at day 4 showed non-inferiority (3.48 [SD 2.66] for the dexamethasone group and 3.51 [SD 2.59] for the prednisolone group; mean difference –0.38; (95% CI –0.71 to 0.64).
More children in the dexamethsaone group received further systemic steroids with 14 days than in the prednisolone group (13.1% v 4.2%; absolute difference 8.9%; 95% CI 1.9% to 16.0%).
How were the data analysed?
The analysis was performed in an intention to treat fashion. As previously mentioned (3), in non-inferiority trials one should ideally perform per protocol analysis alongside intention to treat. When patients move between two treatment arms it increases the risk of finding no difference between the two treatments.
So no blinding?
Full blinding would have been possible in this trial and could have solved a number of problems in the paper. The authors note that more children in the dexamethasone arm received steroids as an outpatient and postulate that this may be due to the community physician’s traditional preference for prednisolone, but without full blinding we can’t truly know.
The clinician reviewing the patient on day 4 was reportedly blinded to the treatment but it cannot be guaranteed that the patient or their care provider won’t give the game away.
So what does this mean?
Despite the methodological flaws in the trial it is not the first trial that has demonstrated the non-inferiority of dexamethasone to prednisolone.
Three previous RCTs have looked at oral prednisolone and oral dexamethasone. The most robust study was a randomized, double-blinded non-inferiority trial of children 2 to 16 years, given single-dose oral Dex (0.6 mg/kg to a maximum of 18 mg) or oral Pred (1 mg/kg per dose to a maximum of 30 mg) twice daily for 5 days, and found no difference in the time taken for the Patient Self-Assessment Score to return to baseline (4). An open label randomised controlled trial also found no difference in relapse rates when children aged between 2 to 18 years were given stat oral dexamethasone or 5 days of oral prednisolone (5). The finally study was so underpowered that no meaningful conclusion can be gained from it (6).
The bottom line
Taken together with previous evidence, the unpalatable taste and issues with compliance, I would be inclined to give a single dose of dexamethasone if the patient required steroids.