The not-so-crystal ball: The ongoing challenge of physician prognostication

Abstract

Grulke et al. examined whether the prognostic estimations Survival estimates are further complicated by the followprovided by physicians for patients undergoing allogeneic haematopoietic stem cell transplantation were associated with their actual survival [1]. They also investigated whether patient-associated objective factors might influence the predictions given by the physicians [1]. The authors can be complemented for investigating and composing this interesting report. The authors concluded physicians on transplant units were able to give a concise prognosis estimate that corresponded to real outcomes. However, it is well known in the literature that physicians' prognostic estimates of survival are far from accurate. Physicians tend to overestimate survival in their patients as shown in seven of eight studies in a recent review of prognostic accuracy [2]. Only 20% of physician predictions of survival were accurate, with 63% of predictions overly optimistic and 17% overly pessimistic [3]. Based on their gross overestimation, oncologists would probably make better predictions if they divided their estimates by two [4]. Though physician predictions have been strongly associated with actual survival [2,3,5], they are by no means exact [2]. Doctors may be able to discriminate patients in immediate danger of death (within 3 to 6 months) from those who are not [6–8], but are unable to accurately calibrate the survival of patients who will live longer.