My EPPPIC prediction
Progestagens won’t work. If they appear to in a subgroup, test in a new trial
Many people are sceptical about progestagens to prevent pre-term birth. Although necessary to maintain pregnancy, no deficiency syndrome has ever been described, early trials were of poor quality, and prospectively registered trials were all, bar one, negative for their predefined primary outcome (click here).
Despite this, enthusiasts have sliced and diced – singles/multiples, progesterone/medroxyprogesterone acetate, short cervix/positive fibronectin, membranes intact/ruptured – and claimed to find subgroups where it works. PubMed has more “systematic” reviews than registered trials. With half a dozen subgroups, selected secondary outcomes, and reviews redone after each new trial, it’s hardly surprising that we keep reading “progestagens work”.
The solution is to collect the original individual patient data (IPD) from the largest trials, and combine them together in an IPD meta-analysis (IPD-MA). So long as the poor quality trials are excluded, and all patients and all outcomes from the high quality ones included, the scope for selective reporting is limited.
The Evaluating Progestogens for Prevention of Preterm birth International Collaborative (EPPPIC) individual participant data (IPD) meta-analysis started last year. It is funded by the Patient-Centered Outcomes Research Institute (PCORI), an independent non-profit, set up by the US government, which receives a $2 tax from every US health insurance plan. The project is registered here. Website here.
Everything is being done properly. In particular, although they have a steering group which includes many involved triallists, they have excluded people with strong vested interests or who have expressed strong ideas – even me! – from the main study team. That’s as it should be. The result should be the last word on this topic. I’ll certainly take it as such. I’ve committed to do so here.
When all the data are in and analysed, if progestagens really reduce something important like perinatal death or brain damage, I’ll stop arguing and start prescribing them.
And if there’s no overall benefit, but progestagens seem to work for one arbitrarily defined subgroup, let’s say women who entered the trials via the finding of a short cervix, I’ll take such a finding seriously and support a further trial to test that new hypothesis.
Jim Thornton
Read about AsPredicted here.
Ripe-tomato.org 
I read this with interest and then read the leader in the on line BJOG that hit my in-box. The new Editor in chief Micheal ZMarsh states if I read him correctly that Progestogens reduce preterm delivery by 45%.
You both can’t be right.
Indeed not. Marsh cites Romero who 1 includes unregistered trials. 2. Says progesterone works but not MPA, 3. who says it works for short cervix but not fibronectin pos, 4 gets most of his beneficial effect from Hassan trial which FDA declined to use for licensing cos all benefit came from two tiny centres. He also ignores the fact that 21/22 registered trials were plum negative. And finally the best trial of all OPPTIMUM, was completely negative. In a nutshell I think Romero slices and dices to get positive result. But I might be wrong. EPPPIC is not slicing and dicing so we’ll soon find out who’s right. 🙂