There was a refereed journal article published in 1979 (Hintz and Van Vleck) that suggested that breeding roans to roans to get roan offspring could have lethal consequences. Since the occurrence of “homozygous roans” was rare, it was thought that “roan to roan” breedings resulted in absorption of the embryo. However, the late Dr. Ann Bowling from the University of California-Davis conducted a study that refuted the earlier paper. This was reported in the Quarter Horse News ( Can a Horse be Homozygous Roan? ) with a promise for a refereed journal article in the near future. She used molecular analysis to prove the existence of homozygous roans in the American Quarter Horse breed. Homozygous roans have two alleles for the roan color, while heterozygous roans only have one allele. She documented the existence of homozygous roans and disproved the lethal gene theory since, if the gene was lethal, then homozygous roans would not survive.
There is no test available to the public for roan markers, but UC-Davis would be the best contact for any further information.