 |
|
 |
|
 |
 |
|
 |
 |
 |
 |
 |
 |
 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
To the Editor.—
Doyle et al1 are to be congratulated for putting forward a study that highlights the potential of low-dose dexamethasone as a short-term intervention in ventilator-dependent extremely low birth weight (ELBW) infants after 1 week of age. I was particularly interested in their discussion of spontaneous intestinal perforations (SIPs) and the possibility that a more mature postnatal intestine might be less likely to acquire SIP with steroid exposure. It is an idea that may have merit. However, I think some perspective is in order regarding how we should "reopen the debate on corticosteroids and ... future RCTs [randomized, controlled trials]" in this patient population.
First, I want to state why I think Doyle et al might be correct in their supposition that the earlier an ELBW infant receives steroids, the greater the chance he or she will acquire SIP and, therefore, that later administration may be safer. Using the Pediatrix database, we recently obtained a national cohort of 633 patients with SIP and found that there was a significant association with both early postnatal steroid and indomethacin exposure (0–3 days) when compared with matched controls.2 However, we also looked at these same drugs in a later time period (4–7 days) and did not see an association with either drug, suggesting that there may indeed be something unique to an early window of exposure.
Unfortunately, there is a potential fallacy in the assumption that it is the timing rather than the combination of drugs that causes SIP. This same assumption is actually implicit in the design of all neonatal RCTs that test pharmaceuticals (ie, that nonrandomized drugs will not have highly specific interactions with the randomized drug). In the case of SIP, it is the potential for harmful synergy between steroids and indomethacin that is the concern. By inferring that it is the timing and not the temporal coincidence of these 2 drugs, Doyle et al could be setting the stage for round 2 of the iatrogenic SIP saga with their call for new corticosteroid RCTs.3–5
For example, a recent retrospective study published in Pediatrics showcased the late use of escalating-dose indomethacin for ductal closure in a similar patient population.6 What if centers enrolled patients in low-dose dexamethasone trials and used escalating-dose indomethacin for ductal closure concurrently? How do we know this new combination will not lead to SIP? We do not. So, what if history repeats itself and there are 3 or 4 concurrent RCTs doing the natural experiment all over again? Maybe we will get lucky and it really will be the timing, or maybe it will be a disaster and I will be publishing unpopular articles and editorials about SIP—again.
If we are to reengage in a conversation about steroid RCTs, I think we should first agree to exclude indomethacin (and all other nonsteroidal antiinflammatory drugs) during any trial periods. We should do this to minimize patient risk but also to give steroids their best chance to succeed. This is likely to narrow the eligible patient population, given that three quarters of the ELBW patients with ventilator dependence at 1 week still had patent ductus arteriosus in the DART (Dexamethasone: A Randomized Trial) study. Because SIP is only slightly more common than cystic fibrosis in the newborn population, extremely large cohorts will be required to achieve sufficient power to examine outcomes such as SIP and cerebral palsy.7 No prospective neonatal study of this size has ever been completed successfully. Thus, to seriously engage in this endeavor will require a broader scope of collaboration than we have heretofore seen. It is not my intent to say that such a project cannot succeed, only that those who accomplish it and do so in a meaningful way with respect to SIP may have to be heroic, inventive, or both.
There are some bright spots in this bleak scenario. We have made some progress in our ability to distinguish SIP from other confounding neonatal diseases. We now know with reasonable certainty that corticosteroid-related perforations are diagnosed, on average, 1 week after exposure, so this gives a reasonable window within which to expect adversity in a new paradigm.8 Likewise, we have learned that one quarter of our smallest patients have occult presentations and that Bell's criterion is not an adequate classification for SIP.9 Finally, we have learned that 20% of SIP cases are unrelated to postnatal factors and occur in infants >1000 g.8 For this reason, we should not be expanding our RCT weight ranges to 1250 g, as some investigators have done, to enhance patient recruitment. When taken together, these facts should improve our ability to correctly identify and capture steroid-related SIPs.
There can be no doubt that steroids have short-term benefits for the lungs, and the Doyle et al article nicely illustrates this for ELBW infants. But, let's be clear; when you have no SIP data in your cohort, you cannot say anything meaningful about SIP. The basal incidence of SIP in most ELBW data sets is 4% to 8%. In some RCTs it is considerably less in both the treated and control cohorts (as it was in this one), and we should probably suspect the veracity of SIP as a secondary outcome in these reports.
I sincerely hope that the DART protocol proves to be valuable, but I urge caution with regard to polypharmacy and meticulous planning of secondary outcome definitions when undertaking an investigation of this sort.
REFERENCES
Related articles in Pediatrics:
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
 |
||
|
||
Home | My Pediatrics | Journal Information | Current Issue | Past Issues | Subscriptions & Services | Contact Us Click here for faster international access © 2006 American Academy of Pediatrics. All rights reserved. |
||
|
||
