Je fonde ma réponse sur une explication des méthodes statistiques utilisées par la FDA que vous trouverez ici : http://www.fda.gov/ohrms/dockets/ac/02/briefing/3908B1_01_A-...
"In accordance with the WLW approach, the time to the first occurrence of a Type 1 event and the time to the first occurrence of a Type 2 event were modeled, using the same proportional hazards model, with treatment group as the main effect *** and pooled country as stratum. Data from countries in which less than 50 patients were randomized were pooled based on geography and/or similar medical practice***."

À la lumière de cette réf., je pense que 'pooled' ne qualifie que 'country' et non 'age', mais bien sûr cela demande à être vérifié