When I got to bed last night, I suddenly realised that I may have missed out a fairly obvious cause, as I had not taken into account the fact that the patients were all pregnant.
I just wonder if these patients have all made anti-Lea and anti-Leb, as it is not unusual for the Lewis antigens to "disappear" in pregnancy, and quite often they transiently make Lewis antibodies. If none of the cells you are using are themselves Le(a-b-), then this antibody mixture (actually, it isn't a mixture, but anti-Lea+b) can look like an antibody directed against a high prevalence antigen.
If you cannot obtain sufficient Le(a-b-) red cells to test this theory (and it is only a theory), you could try to inhibit the antibody with saliva from an Le(a-b+) individual (but don't forget to control this by diluting another aliquot of the plasma with saline).
The attached may be of interest/use.
The Lewis Blood Group System and Secretor Status.docx