Thursday, May 28, 2009

(Not) Discussing hospice with lung cancer patients

Archives of Internal Medicine has a paper about hospice discussions with patients with lung cancer. The data came from a large, US, multicenter/regional study which involved phone surveys of ~1500 patients diagnosed with lung cancer (who, among many other items, were asked if a health care provider had discussed hospice care with them). The surveys occurred 4-7 months after diagnosis, this analysis only looked at patients with metastatic cancer, and charts were also abstracted to see if there was any documentation of discussing hospice care. Surrogates could answer for patients who weren't able or who were already dead and in fact over half of the respondents in this analysis were actually surrogates. Response rate was ~50% for the original telephone survey. A bunch of other items were collected as well. (As a minor side note they talk about coding patients' responses to questions about their prognosis and including 'it's in God's hands' as one of the potential codes - along with different timeframes and 'I don't know.')

There's a lot of data in this paper, and I'm just going to focus on what seems most relevant. Only 53% of patients (or their surrogates) reported and/or had medical records indicating that they had discussed hospice with a health care provider. There were a variety of modest (generally 5-10%) but statistically significant differences in reporting that discussion between different groups (whites more likely than African Americans or Latinos; patients with Medicare more likely than those with Medicaid; older more likely than younger; those who received chemo vs. those that hadn't, etc.).

Time to death had a major impact on who had had a hospice discussion as you'd expect, but the results weren't heartening. Surrogates reported that for ~80% of patients who were already dead there had been a hospice discussion; this drops to 54% for those who died within 2 months of the interview; 34% for those 2-4 months, and 26% for those who died 4-6 months after the interview.

When they looked only at the patients who completed the interview themselves these rates were even lower (43% for those who died within 2 months, 27% 2-4 months, etc.). Patients' self-reported expected prognoses impacted the likelihood of having a discussion: of those who expected less than a year it was ~45%, for those who thought it was likely to be greater than 5 years (which, not-shockingly-but-still-kinda-sobering, was over a fifth of the cohort - remember these were all patients with stage 4 lung cancer which has a 5 year survival of under 5%) it was 12%. 16% of the 'In God's hands' patients reported a hospice discussion. Of the "do not know" patients, which were 40% of respondents, 18% had had a discussion.

Notably, half this patient responders said they prefer 'pain relief over life extension.' That preference did not predict having a hospice conversation.

The low response rate to the survey may is a concern with these data, however my suspicion is that a 100% response rate would likely have altered the numbers a bit, but would not have altered the trends seen here. In addition, surrogates probably aren't terribly accurate reporters of what has happened to a patient; that said the differences in what the entire cohort demonstrated and the subgroup of patient-themselves responders likely reflects the reality that the surrogate responder patients were sicker (or of course already dead) and so the reported rates of conversations were higher in the entire group.

That aside the findings here are quite clear: for this group of patients with dismal prognoses hospice discussions were happening late (if at all), patients overall admitted ignorance or inaccurate perceptions of their likely futures (time-wise), those that did have a decent understanding of their prognosis were more likely to have had a discussion, although still at low rates, and half of these patients straight-up reported that life extension was not their preeminent goal.

I'm personally mostly interested in data about patients' perceptions of prognosis. To me, that's the key breakdown here. If you have a patient who is thinking they'll likely live for more than 5 years, a discussion, even info sharing, about hospice doesn't particularly make sense. If they understand it's likely to be less than a year, that discussion makes much more sense. More than that, we're doing our patients a disservice if we are telling them/allowing them to believe (by not correcting or never even addressing) grossly over-optimistic prognoses. I think I'd make different choices in my life (and my health care decisions) if I understood my likely time left was less than a year vs. 2+. (One also wonders if that fifth who thought they had over 5 years to live thought their cancer was curable.) Not all patients want to talk about time, of course - that's fine (although not making sure they understand that their cancer is incurable and any treatment is given with the hope of life-prolongation but without cure is not fine). But half these patients were already saying life-prolongation wasn't their top priority; if we're not even discussing time and all the treatment options with them we're doing them a disservice. This is not just a problem in oncology, although it's probably the most well-defined there as it's relatively well-studied.

Sometimes I feel that the Great Big Dirty Secret of 'palliative care' that's really of course out in the open if you cared to look at the literature or, you know, talk with patients is that many and indeed most of our patients facing life-limiting illnesses want to talk about this stuff. (To some extent of course, not necessarily all the gory details). And they are in fact deeply grateful to be able to talk about it. More frightening, worrying, not to talk about it. They thank me all the time for bringing it up, actually. And it's usually pretty easy to do.

Pallimed | Blogger Template adapted from Mash2 by Bloggermint