Prophylactic fluconazole works; but we shouldn’t use it?

A new high quality multicenter RCT examined the effects of fluconazole prophylaxis in about 360 very tiny babies (VTB), that is they weighed less than 750g at birth. They were enrolled in the first 5 days of life to fluconazole prophylaxis or placebo. It was given iv or po twice a week until 42 days of age. Primary outcome was death or invasive candidiasis, with definitions given for definite and probable disease, which look very appropriate, which is as I would expect given that the PI of the study is Danny Benjamin, the authority on neonatal candidiasis, (Benjamin DK, , et al. Effect of Fluconazole Prophylaxis on Candidiasis and Mortality in Premature Infants. JAMA. 2014;311(17):1742).

The study primary outcome was less frequent in the active treatment group, but the difference could have been due to chance (p=0.24), 21% placebo and 16% fluconazole. Invasive candidiasis was reduced from 9% to 3% during the treatment period and from 11% to 4% prior to discharge, both of which reductions are unlikely to be due to chance (p < 0.02). There were no appreciable toxicities.

So what should be the conclusion? I would suggest that the conclusion as it is published is scientifically accurate, but only partially true.

Fluconazole ‘did not result in a lower incidence of the composite of death or invasive candidiasis.’

The conclusion does not state, as I think it should, fluconazole greatly reduces the incidence of invasive candidiasis, but this reduction did not lead to a reduction in mortality. (Which was 14% in each group, confidence intervals for death between -7% and +7%).

I think this wording difference could be important, and I will try and explain why.

Control group mortality was 14%, with an incidence of invasive Candida of 9%. Let’s assume that during planning of the study these numbers in these 32 NICUs were similar, and that fluconazole was expected to reduce invasive candidiasis by 2/3, without adverse effects on other serious outcomes that would increase mortality. Previous data show a mortality among VTBs who develop Candidiasis of around 30%. So we can calculate that among infants without candidiasis the mortality is 12%, to give an overall mortality of 14%. If the only effect of fluconazole is the reduction in candidiasis, then we might expect the intervention group mortality to be reduced to 12.5%.

The combined outcome of death and invasive candidiasis would then be expected to fall from 20% to 15%. (Which is in fact almost exactly what the study actually showed!) That would be: 11 deaths in the 91 of the 100 enrolled control infants who did not get candida plus 9 cases of candida, compared to 12 deaths in the 97 per 100 enrolled fluconazole infants who did not get candidiasis plus 3 cases of candida.

Which means that the study was actually quite underpowered, for that combined outcome, if the only expected result was a reduction in candidiasis and its associated mortality.

A really important feature of this study was a high quality follow up component. There was no real difference in any 18 to 22 month outcomes, either in neurological outcomes or Bayley 3 scores.

If you were to do the same exercise for these long term outcomes as for the ‘death or candidiasis’ outcome, it would also show that the study was underpowered to test the hypothesis that the only effect of fluconazole was to reduce candidiasis and the neurodevelopmental impairment that goes with it.

That certainly does not mean that these results aren’t important. We did need to know that fluconazole doesn’t have serious adverse effects, or have adverse effects on long term neurologic or developmental outcomes. We can now be reasonably confident that that is the case.

The final sentence of the abstract, states ‘These findings do not support the universal use of prophylactic fluconazole in extremely low-birth-weight infants.’ Again I am not sure that I agree entirely with that: if ‘universal’ use of prophylactic fluconazole reduces invasive candidiasis without harm, why not?

The article also suggests that fluconazole could perhaps be considered in NICUs with a high incidence. But why not, rather, in infants with risk factors that lead to a high incidence?

This is a significant issue in therapies used as prophylaxis of a complication which is relatively uncommon, if there is a small risk of harm, then patients who receive the prophylaxis who would not have developed the complication only have potential for risk and not benefit, and patients who would have developed the complication may have major benefit. But if we cannot identify those who are at risk, then we have to determine overall whether the prophylactic treatment is worthwhile.

One thing I would like to see from this study is an analysis of risk factors, in this same cohort, and also the timing of those risk factors and the onset of candidiasis. I think based on all the data that we have, that a reasonable approach would be to use prophylaxis for VTBs at increased risk, perhaps limited to those periods of high risk, such as when on broad spectrum antibiotics, when central lines and TPN are being used, and/or when on steroids.

That to me would be more rational than using prophylaxis in all infants in high incidence units, and not in low risk units, to use it during periods of high risk, and not during periods of low risk.

About Keith Barrington

I am a neonatologist and clinical researcher at Sainte Justine University Health Center in Montréal
This entry was posted in Neonatal Research and tagged , , . Bookmark the permalink.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.