"Optimistic Bias:" Clinicians' Overestimation of Cancer Patient Survival Affects End-of-Life Outcomes, Including Lower Hospice Use
A cohort study that followed advanced cancer patients who engaged in a palliative care (PC) consultation between 2013 and 2016 found that patients' likely survival time was often overestimated by palliative care clinicians, and that this overestimation occurred more frequently when patients were African American or Latino. Further, the researchers determined that overestimation of prognosis corresponded with less hospice use, which suggests that this “optimistic bias” has an effect on the course of cancer patients' end-of-life (EOL) care.
Do Clinicians Know When EOL Is Happening?
“Estimating survival time in serious illness is a challenging and important task for clinicians who care for the seriously ill,” write the study authors, who note that the majority of people report a preference for comfort care over longevity near the EOL, yet more than 80% of Medicare beneficiaries die without any hospice enrollment or are enrolled within 72 hours of death.
One hypothesis explaining this “mismatch” between many patients' EOL preferences and their actual care is that “clinicians' optimistic bias in survival estimation prevents patients and clinicians from knowing when EOL is happening,” the authors write.
In order to assess the prevalence of survival overestimation and its effect on patients, researchers analyzed data and survey responses of 230 English-speaking advanced cancer patients and their PC clinicians at two US academic medical centers. Patients were asked about their EOL preferences—whether they would prefer comfort and quality of life over survival time—to gauge opinions affecting hospice use. After six months, patients or their informants were contacted to confirm survival time, hospice enrollment, and whether aggressive treatments “with little or no value to comfort in advanced cancer” were used within 14 days of death. The treatments included cardiopulmonary resuscitation, mechanical endotracheal ventilation or feeding tube.
PC clinicians were asked to make a “best guess” about the patients' survival time following the PC consultation, “assuming that their illnesses are allowed to take their natural course.” Clinicians marked each patient's likely survival time as: < 24 hours; 24 hours to < 2 weeks; 2 weeks to < 3 months; 3 months to < 6 months or > 6 months.
- Patients were a median age of 63 years (interquartile range [IR], 54 to 71 years); female gender, 50%; self-identifying as non-white, 24%; self-identifying as either African American or Latino, 20%.
- 53% of patients said they “strongly agree” they would prefer comfort and quality of life to survival time, as opposed to 6% who chose “strongly disagree.”
- Patients lived for a median of 37 days (IR, 12 days to 97 days).
- PC clinicians effectively determined patients' survival time relative to one another, and were accurate with respect to absolute survival time in 41% of cases, consistent with previous research.
- PC clinicians were nearly six times more likely to overestimate than to underestimate survival time (85% vs 15%; P < 0.001), and 50% of patients had an overestimated survival time.
- Among patients who died within six months, overestimation of survival was associated with no or late hospice use (odds ratio [OR], 0.45; 95% confidence interval [CI], 0.24 to 0.83) and more aggressive treatment at EOL (P < 0.05). Controlling for potential confounding variables did not weaken this association, and when the 15 patients who did not prefer comfort EOL care were excluded, the association was stronger (OR, 0.39; 95% CI, 0.21 to 0.76).
Higher Survival Overestimation for African American and Latino Patients
- PC clinicians were significantly more likely to overestimate survival time of patients who identified as either African American or Latino (66% vs 46%), even when adjusting for potential confounding (adjusted OR, 3.89; 95% CI, 1.64 to 9.22). This association between racial and ethnic identity and prognosis overestimation persisted when patients who did not prefer comfort EOL care were excluded (adjusted OR, 4.03; 95% CI, 1.69 to 9.63).
- African American and Latino patients survived as long as other patients (hazard ratio, 1.07; 95% CI, 0.76 to 1.51) and were not significantly more likely to be opposed to comfort-focused EOL care (11.1% vs 5.1%; P = 0.13).
Implications for Improving EOL Care
“To our knowledge, this study is the first to identify an empirical link between clinician survival overestimation and actual patient EOL outcomes,” the authors state. “We hypothesize that overestimation might act via direct pathways (i.e., communication of prognostic estimate or confirmation of hospice eligibility) or indirect ones (i.e., focus of clinician's diagnostic attention to treatment-related suffering) to influence hospice use,” they write.
Whereas previous research found that a source of racial/ethnic disparity in EOL care may be poor communication of estimated prognosis, the authors write that these findings “suggest that the accuracy of what clinicians believe to be true regarding survival prognosis differs by race/ethnicity.”
While they note that these outcomes do not identify a cause for the racial and ethnic disparities observed, the study authors stress that greater awareness of the common survival overestimation African American and Latino patients face seems like an important focus when it comes to reducing the effects of overestimation. They describe these racial and ethnic disparities in preference-concordant EOL care as “substantial” but “plausibly mutable.”
Overall, the authors write that the research findings “identify prognosis overestimation in advanced cancer to be an important, timely and a promising target for interventions that promote preference-concordant and equitable EOL care,” and they endorse the “systematic use” of prognostic tools to help give seriously ill patients the EOL care they prefer.
Source: “Palliative Care Clinician Overestimation of Survival in Advanced Cancer: Disparities and Association with End-of-Life Care,” Journal of Pain and Symptom Management; October 2018; DOI: 10.1016/j.jpainsymman.2018.10.510. Gramling R, Gajary-Coots E, Norton SA, et al; University of Vermont, Burlington, Vermont; and University of Rochester, Rochester, NY.