Distinct problems in the analysis of failure times with competing causes of failure include the estimation of treatment or exposure effects on specific failure types, the study of interrelations among failure types, and the estimation of failure rates for some causes given the removal of certain other failure types. The usual formation of these problems is in terms of conceptual or latent failure times for each failure type. This approach is criticized on the basis of unwarranted assumptions, lack of physical interpretation and identifiability problems. An alternative approach utilizing cause-specific hazard functions for observable quantities, including time-dependent covariates, is proposed. Cause-specific hazard functions are shown to be the basic estimable quantities in the competing risks framework. A method, involving the estimation of parameters that relate time-dependent risk indicators for some causes to cause-specific hazard functions for other causes, is proposed for the study of interrelations among failure types. Further, it is argued that the problem of estimation of failure rates under the removal of certain causes is not well posed until a mechanism for cause removal is specified. Following such a specification, one will sometimes be in a position to make sensible extrapolations from available data to situations involving cause removal. A clinical program in bone marrow transplantation for leukemia provides a setting for discussion and illustration of each of these ideas. Failure due to censoring in a survivorship study leads to further discussion.

To determine whether allogeneic bone-marrow transplantation is associated with a graft-versus-leukemia effect, we examined the relation between relapse of leukemia and graft-versus-host disease in 46 recipients of identical-twin (syngeneic) marrow, 117 recipients of HLA-identical-sibling (allogeneic) marrow with no or minimal graft-versus-host disease, and 79 recipients of allogeneic marrow with moderate to severe or chronic disease. The relative relapse rate was 2.5 times less in allogeneic-marrow recipients with graft-versus-host disease than in recipients without it (P less than 0.01). This apparent antileukemic effect was more marked in patients with lymphoblastic than nonlymphoblastic leukemia, and in those who received transplants during relapse rather than during remission, and was most evident during the first 130 days after transplantation. Survival of all patients was comparable since the lesser probability of recurrent leukemia in patients with graft-versus-host disease was offset by a greater probability of other causes of death.

The records of 545 patients were reviewed for risk factors associated with cytomegalovirus (CMV) infection after marrow transplant. CMV infection occurred among 36% of seronegative patients and 69% of seropositive patients. Among seronegative patients, significant risk factors for CMV infection included positive serology of the marrow donor (relative rate, 2.3) and the use of granulocyte transfusions from seropositive donors (relative rate, 2.5). Among both seronegative and seropositive patients, the occurrence of acute graft-versus-host disease (GVHD) significantly increased the risk of CMV infection (average relative rate, 1.8) and of subsequent CMV pneumonia (average relative rate, 2.6). CMV excretion and viremia were each associated with subsequent pneumonia, but the positive predictive values were low. One-third of long-term survivors excreted CMV at greater than 250 days after transplantation. The only risk factor for late excretion was CMV infection that occurred in the first 150 days after transplantation. In contrast to the effect of acute GVHD on CMV infection, CMV infection did not increase the risk of either acute or chronic GVHD.

Marrow transplantation has generally been limited to patients with a sibling who is genotypically identical for HLA. In a study of the acceptable limits of HLA incompatibility, 105 consecutive patients with hematologic cancers who received marrow grafts from haploidentical donors (study group) were compared with 728 similar patients concurrently receiving grafts from HLA genotypically identical siblings (control group). The unshared haplotypes differed variably: 12 were phenotypically but not genotypically identical for HLA-A, HLA-B, and HLA-D; 63 differed at one locus (A, B, or D); 24 at two loci; and 6 at three. A higher proportion of study patients had delayed engraftment, granulocytopenia, or graft rejection. Acute graft versus host disease occurred earlier and with greater frequency in study patients. The risk of the disease did not correlate with disparity for Class I (A or B) versus Class II (D-region) loci. Thus, incompatibility for HLA has an important effect on the course after clinical marrow transplantation. In spite of these complications, there was no statistically significant difference in the survival of the study patients and control patients who received their transplants during remission.