Current planet formation theories are bound to comply with the observational constraint that protoplanetary disks have lifetime of ~3 Myr. This timescale is mostly based on spectroscopic studies of objects accreting matter from a circumstellar disk around pre-main sequence stars (PMS) located in low-density, nearby (d<1-2kpc) star forming regions. These objects do not reflect the conditions in place in the massive starburst clusters where most of star formation occurs in the universe. Using a new robust method to indentify PMS objects through their photometric excess in the Halpha band, we have studied with the HST and ground based facilities the PMS population several starburst clusters, namely NGC3603 in the Milky Way and several clusters in the Carina Nebula,  30 Doradus and the surrounding regions in the Large Magellanic Cloud and NGC 346 and NGC 602 in the Small Magellanic Cloud. We found a wide spread of ages (0.5 to 20 Myr) for PMS stars, clearly showing that accretion from circumstellar disks is still going on well past 10 Myr. This finding challenges our present understanding of protoplanetary disk evolution, and can imply a new scenario for the planet formation mechanism and of star clusters formation in general. Based on these results we were recently granted 175hr with OmegaCAM at the VST to carry out a deep optical wide field survey of nearby (<3kpc) star forming regions. These observations will provide physical parameters (including mass accretion rates) for over 10000 PMS stars and will establish whether the long timescales of circumstellar discs are common.