Association of Early Palliative CareWith Survival in PatientsWith Advanced Lung Cancer

In Reply We thank Skelin and colleagues for their comments regarding our study¹ evaluating the association of early palliative care with survival among patients with advanced lung cancer. Performance status is an important parameter in cancer treatment decision-making, and the Eastern Cooperative Oncology Group scale is one set of criteria used to encourage standardized reporting of treatment toxic effects and response, especially in the conduct of randomized clinical trials. Unfortunately, interrater reliability is inconsistent,² inherent clinician bias may contribute to inconsistencies,³ and missing values are common in administrative data. Instead, we included the Charlson Comorbidity Index score, which measures comorbidities, and the Functional Comorbidity Index score, which measures physical function to predict survival in modeling, as surrogates of performance status because both are well validated in administrative data. Ultimately, treatment decision-making in oncology is complex, multifactorial, and influenced by clinician-related and patient-related factors; therefore, no single measure of patient suitability for treatment is likely to be sufficient.

The goals of treatment among patients with incurable cancer are to promote quality of life and prolong survival. However, both may be worsened by treatment in some patients with advanced lung cancer.⁴ Additionally, the initiation of early palliative care may reduce treatment receipt among these patients,^5,6 which is one suggested mechanism to explain the improved survival.⁴ Therefore, we believe it is possible that patients who received early palliative care had reduced exposure to cancer treatment owing to enhanced goals-of-care conversations rather than intolerance to therapy, as suggested by Skelin et al. Regardless, it is difficult to predict whether treatment received would lead to unequal prognosis among groups; we adjusted for treatment received in modeling given its hypothesized potential to affect survival (positively or negatively), although findings from unadjusted models (ie, those that did not include treatment) were similar.

Finally, the results of our study¹ are consistent with the landmark palliative care randomized clinical trial by Temel et al.⁶ This trial excluded patients who were already receiving palliative care, had lower performance status, and did not have pathologic confirmation of lung cancer (ie, were unable to tolerate invasive procedures). Patients were also recruited from outpatient oncology clinics and were enrolled within 8 weeks of diagnosis and received the intervention within 3 weeks of enrollment. As a result, many of the patients in our study¹ who received palliative care 0 to 30 days after diagnosis would have been ineligible to participate because many of these patients received cancer care as inpatients only and a significant proportion died within a few weeks of diagnosis. To note, outpatient palliative care, regardless of timing, was associated with a survival benefit in our study (eTables 6 and 7 in the Supplement).¹ The best comparator patient group, based on patient characteristics, from our study with the aforementioned trial⁶ is likely the group receiving palliative care 31 to 365 days after diagnosis. We do not believe that the lack of an associated survival benefit (or decreased survival) diminishes the importance of palliative care in the group that received palliative care 0 to 30 days after diagnosis because this supportive approach was likely intended to ease the dying process in these patients by offering bereavement support to loved ones and coordinating hospice services rather than focusing on prolongation of life.