TO THE EDITOR: We welcome the opportunity to discuss Shapiro et al.'s findings in relation to ours. Both investigations linked psychiatric morbidity (e.g., personality disorder and substance abuse) with compliance behavior and demonstrated that clinician judgment of compliance behavior and psychosocial risk and support can be related to different aspects of health behavior. We share a common view that identification of risk should lead to interventions to enhance outcome and prolong life. Our philosophy is that certain candidate behaviors (e.g., smoking or drinking during end-stage lung and liver disease) may preclude transplant but only until behavior change can be demonstrated. Patients remain in control of their own destiny provided that they demonstrate responsibility and seek help in overcoming serious personal obstacles destructive of the organs they seek to replace.

We are puzzled by Shapiro et al.'s characterization of our study as "hypothesis generating." We would reserve hypotheses for experimental manipulations or groundbreaking theoretical formulations, not verification of relationships properly credited to the earlier research that we cited in our article. Shapiro et al. also cite a briefer follow-up interval as a possible reason for the failure to relate their variables to mortality. However, they appear to have inappropriately used correlation analysis rather than proportional hazard model statistics to evaluate this relationship. They actually identified a larger group (125 cases) than we studied but obtained ratings on only 75. They did not report what portion of the 33 survivors were included in their analysis, nor did they describe needed comparisons between survivors and nonsurvivors for illness severity before transplant. It is also notable that axis I conditions were not considered or mentioned. We would encourage the authors to consider this variable and reanalyze their data by using survival analysis.

In addition to identifying treatable psychiatric conditions, our findings suggest that evaluation needs to go beyond psychiatric diagnosis and consider other psychological behavioral risk factors in order to properly forecast treatment outcome, including health care utilization and survival time. Beyond encouraging replication investigations, we hoped that our study demonstrated the feasibility of routine evaluation of all relevant (medical and behavioral health) factors with this population as part of a continuous outcomes management process integral to the clinical service. Our findings and those of our predecessors should make a persuasive case for continued coverage of consultation evaluation and treatment efforts, since psychiatric and behavioral health factors are clearly important components of the "major medical" condition (organ failure) and treatment intervention (transplant). While mental health services may have only recently gained parity with other medical conditions, legislators, health benefit administrators, and our own medical brethren still need to recognize that there is a continuum of behavioral and psychosocial aspects of health and illness that are, in fact, inseparable from physical illness.