Are you aware of any large-scale RCTs comparing NSAID vs paracetamol?

I don't think there are any large trials published, there are 5 or 6 trials registered, but most are fewer than 200 patients total. 

Arne Ohlsson and Prakesh Shah just published an updated Cochrane review. They were able to include 2 randomized trials, both of oral therapy, and with a total of 250 patients. They showed no real difference in efficacy or in long term outcomes. 

We certainly do need a good trial now, I think there are enough data to support such a trial, if paracetamol works equally well, but maybe has fewer effects of fluid balance (for example) that might lead to better outcomes, if we do the studies right. 

I agree completely, a well-powered comparative trial is def needed.

BTW, I heard from the US-based neonatologist/researcher Matthew Laughton about a planned trial that will be a placebo-controlled evaluation of NSAIDs (i.e. randomization to NSAID or placebo). As the natural course of duct closure (regardless of what's done) is a critical issue that trial will be very interesting (once completed).

However, my personal opinion is that findings from clinical PDA research in general, most commonly including infants around 28-30 weeks, might be difficult to generalize for infants born at 23-25 weekers (i.e. the association between therapy-outcome might be effect-modified by gestational age).

I think you are likely correct, certainly ibuprofen is less effective at earlier gestational ages, and paracetamol might have a different relationship with lower GA. Certainly need to get good studies done. A brand new article since I wrote the above blog post found no effect of paracetamol among VLBW infants who had failed ibuprofen.

(Roofthooft DW, et al. Limited effects of intravenous paracetamol on patent ductus arteriosus in very low birth weight infants with contraindications for ibuprofen or after ibuprofen failure. Eur J Pediatr. 2015:1-8.)