Earthquakes preceding large events are commonly referred as foreshocks. They are often considered as precursory signals reflecting the nucleation process of the main rupture. Such foreshock sequences may also be explained by cascades of triggered events. Recent advances in earthquake detection is a motivation to re-evaluate seismicity variations prior to mainshocks. Based on a highly complete earthquake catalog, Trugman and Ross (2019) and van den Ende and Ampuero (2020) suggested that mainshocks in southern California are often preceded by anomalously elevated seismic activity. These studies assume a time-independent seismicity and thus neglect earthquake interactions. In this study, we test the same catalog against the Epidemic Type Aftershock Sequence model that accounts for earthquake clustering. We find that less than 5 out of 53 selected mainshocks (10%) are preceded by significantly elevated seismicity rates. This suggest that foreshock observations can generally be explained by background seismicity and by cascades of earthquakes even in highly complete earthquakes catalogs.