Monday, August 4, 2008
Cancer has e-pub'd a study looking at the aggressiveness of care in patients dying of non-small cell lung cancer. The data come from a prospective study looking at early specialist palliative care involvement for patients newly diagnosed with advanced (stage IIIb or IV) NSCLC - essentially a 'concurrent care' model in which patients received specialist palliative care alongside routine cancer care/therapy (see here). This study looks at data for 46 patients (mean age 66 years; 100% white) accrued at one of the sites (Massachusettes General Hospital) and examines factors associated with aggressiveness of care (use of anti-neoplastic therapy, emergency department visits, hospitalizations, hospice use, place of death, etc.). The primary framework was looking at elements of care in the last 30 days of life for the 40/46 patients who died during the study - it's kind of like a mortality follow-back design although the patients were enrolled and followed prospectively.
The findings generally support the impression of aggressive care in the final days of life: 40% received antineoplastic treatment within 30 days of dying (nearly a quarter within 14 days of dying), and ~50% visited the ED and/or had a hospital admission. 62% died in a hospice setting, although with a mean of only 16 days. One-third of patients spent less than a week in hospice before death; one-third however spent more than a month. Interestingly, in bivariate analysis of predictors of 'aggressive' care elevated baseline anxiety or depression symptoms predicted increased likelihood of anticancer treatment within 14 days of death. They don't mention whether they evaluated for a 'treatment effect' of individual oncologists (i.e. distinct variations in care based on who the oncologist was - such an effect has been observed for aggressiveness of care for dying patients in ICUs - given the intimacy of this study this may not have been an easy thing to do politically, or may just have been outside the scope of this analysis).
The discussion at the end of the paper highlights the many, many unanswered questions this study raises, as well as generally cautions against throwing up of arms in the air about this. The big question (at least from a patient-centered perspective) of course is does this seemingly high rate of aggressive care reflect 'bad care' or 'less than ideal care'? On its face it seems that it does, but I'm not convinced that that's necessarily true; if all these patients at the beginning of their last month of life had been told that it was 'highly likely' that they would die in the next 1-2 months (regardless of what care they received), were given all the available options and services (hospice, other types of home care, ongoing anticancer treatment, supportive-only treatment, etc. etc.) - would these outcomes have been any different? We really don't know, and even if they outcomes would be different it's not like the levels of hospitalization/ED visits would go to zero (sometimes the best place for people, even very sick dying people, is the hospital). In addition, chemotherapy (oral or IV) does benefit quality of life in NSCLC (for some patients) - something which the authors address in the paper's discussion - and unless one is really darn sure someone has less than 30 days to live a decision to continue that treatment is not necessarily a bad one, although it eliminates a patient's access to otherwise indicated hospice services.
Reflecting on my own experience, I think there are major barriers to good end of life care on both ends (doctors aren't having realistic conversations and advising about appropriate care plans in light of a poor prognosis until way too late; the way US hospice services are set up creates barriers to access except for patients in the final days/few weeks of life due to restrictions [real or imagined] on anticancer treatments, emergency room visits, etc.).
Temel, J.S., McCannon, J., Greer, J.A., Jackson, V.A., Ostler, P., Pirl, W.F., Lynch, T.J., Billings, J.A. (2008). Aggressiveness of care in a prospective cohort of patients with advanced NSCLC. Cancer DOI: 10.1002/cncr.23620