To the Editor.—Although statistically elegant, the recent article in the Archives by Reuben et al1 may be incomplete. The "terminal cancer syndrome" may indeed be a real entity; however, the predictive model of the authors that used the Karnofsky Performance Status Scale (KPS) and specific clinical symptoms lacks the additional variables we have found to be significant.
We recently conducted an historical prospective study to determine whether certain variables correlated with length of survival in patients with terminal cancer. Data were obtained from 172 newly admitted patients to a home-based community hospice service. The following information was from the history and physical examination obtained by the physician: patient acknowledgment of disease (yes or no); number of preadmission medications; systolic and diastolic blood pressures; pulse; and evaluation and defined scale assignment of mobility, anorexia, cachexia, and pain. Activities of daily living (ADLs), including transfer, bathing, dressing, and continence, were