First, I'm not sure I believe that it is true to say that, "most of those mega-journals which boast broad scope focus in fact only on medicine, biology and health sciences." This might be true of PeerJ but it is not true of PLOS ONE. I have read sociological experiments, physics-based network analyses, and pure computational work published in PLOS ONE. I get the sense that bioscience makes up the core of their editorial board and bulk of their publications, but it's far from the whole thing. Also, there are other "mega-journals" that serve other communities and fields. For example, SAGE Open focuses on the humanities and social sciences. There are others and new ones being created frequently.
That said, it is true that the first and biggest mega-journals are more focused on bioscience than they are on other fields. I think that the reason is because the "mega-journal" model is closely tied to the modern open access movement which has its largest amount of support in biological and health sciences. Basically, I think that PLOS ONE attracts so many biological articles because they have a biological heavy editorial board. They have a biological heavy editorial board because they have leveraged the network that exists for PLOS. PLOS' flagship journal is, of course, PLOS Biology and the organization was started by a group of bioscientists.
Peter Suber's open access timeline is instructive. Although OA has a broader history, folks like Harald Varmus have used institutional support at places like the NIH to push for a set of norms around open access through projects like PubMedCentral and requirements from funders to publish OA. The result is an OA movement that has simply been much more successful and influential in the biosciences than it has been in other fields. There are structural reasons this might be the case. Pay-to-publish models are easier to swallow in fields that are largely grant based. Biological and health sciences makes up an enormous proportion of grant funded research through the NIH, NSF, and other private and public agencies.
But my sense is that, fundamentally, PLOS ONE has provided the inspiration for most mega-journals because it quickly became the largest journal of the world and, through that process, managed to maintain a surprisingly high impact factor (~4 in 2011!). To the extent that PLOS ONE has created the mega-journal model, there may be a bioscience bias — especially early on — that comes simply from PLOS ONE's own roots in bioscience.