Thursday, June 20, 2013
This month, the Cochrane Collaboration has released a review of home palliative services for adults with advanced illness and their caregivers. The authors, led by Barbara Gomes, MSc, PhD, set the primary outcome as occurrence of death at home. Secondary outcomes included the time the patient spent at home, satisfaction with care, management and degree of pain and other symptoms, symptom burden, physical function, quality of life and caregiver pre and post bereavement outcomes. Economic data was also examined and included hospital costs, other institutional care costs, community care costs, informal care costs, and equipment and medications prescribed.
A total of 23 studies were included in the review, consisting of 16 randomized control trials (RCTs), four controlled clinical trials (CCTs), two controlled before and after studies (CBAs) and one interrupted time series (ITS). There were 37,561 patients covered in these 23 studies, with diagnoses of cancer, congestive heart failure, COPD, HIV/AIDS, multiple sclerosis and some small numbers of other conditions. About half of the studies (eleven) were from the US, five in the United Kingdom, five in continental Europe, one in Canada and one in Australia. Two of the studies dated back to the mid to late 1970s (both were in the US), with ten of the studies being of this century.
In terms of how “home palliative care” was defined, there also seemed to be great variance in the models for each of the studies. They varied in how long the service was in place, what team members/disciplines composed the care team, the intensity of services (one was a “telephone model” and another was an “intensive home care” model), in what system the services were based (for example, a community versus a hospital based model).
To meet the primary outcome of identifying whether death occurred in the home or elsewhere, only nine of the 23 studies actually included this data. Six looked at death in a hospital, five looked at death in a nursing home, another five at death in an inpatient hospice or palliative care unit, and five looked at the time the patient spent at home. Secondary outcomes, as listed above, were also gathered from the various studies. Sixteen of the studies provided information relating to resource use (ED visits, ICU stays, nursing home and hospital admissions, clinic visits, among others) and eight of those calculated costs associated with the resource use, while six calculated total cost.
In short, Gomes and her group had a monumental undertaking and sought to pull a great deal of data from these studies in formulating their review, and the headline finding is very impressive: Patients on home palliative care are more than twice as likely to die at home as patients not on home palliative care.
Now, I’m an inherently trusting kind of gal. If you tell me an intervention worked, I’m likely to believe you. If you tell me you read the 80 something articles that discussed the 23 studies that have been done, I’m definitely going to believe you. Since I’m a big fan of palliative care and somewhat biased to that end, if a 280 page report stamped with “Cochrane Review” that also does all of those things I just talked about ends with a verdict of “people have a better chance of dying at home, they have less symptoms, and their caregivers seem to do okay,” I’m going to believe you, and be really excited about it.
So if a Cochrane review comes out with a headline that patients getting home based palliative care have DOUBLE that chances of dying at home than patient not getting home-based palliative care, you can imagine the glee and enthusiasm I might feel.
And then someone suggests that this review be, well, reviewed. So I start digging. And bit by bit, my glee and enthusiasm wane.
Not too long ago, back in 2011, there was a Cochrane review on “Hospital at home: home based end of life care,” which reviewed four studies. This review calculated a relative risk, and found that those on home-based end of life care had an RR of 1.33. My first thought was, "Wow, a lot of great research must have come out since way back in 2011 to get us from 1.33 to 2.21." So I forged ahead.
In terms of looking forward to the exciting new data, four studies from the 2011 piece were included in the 2013 review, with another three added, none of which were new since the 2011 review. So, in all, for purposes of assessing likelihood of death at home there were seven studies reviewed (of the 23 studies included for all combined purposes in 2013 review). For this review, Gomes and her group chose to assess via odds ratio for dying at home from those even studies, which, again was initially (and with fanfare) found to be 2.21. But…
By dropping just one of the studies (dropping the number included to six) as a possible outlier, this dropped to 1.7.
Further, when the 2013 study dropped out the two CCTs, and included only “high quality RCTs,” (so now looking at just four studies) the calculated OR dropped further, to 1.28 – essentially on par with what Shepperd’s group found in 2011 (recall that their relative risk was 1.33).
Now, before all you statistics experts jump all over this by pointing out that one study looked at the relative risk while the other looked at the odds ratio, I will accept your point. It remains, though, that by modifying their inclusion criteria to eliminate one potential outlier and two studies not deemed “high-quality” that we dropped from the very exciting number of 2.21 to the still interesting, but less thrilling, 1.28. And, that’s not a change one can pick up just by looking at the abstract or the “Plain Language Summary,” which states that “when someone with an advanced illness gets home palliative care, their chances of dying at home more than double.”
Now we ask, after all that, IS the proof in the pudding**? Do we have enough information to start quoting this line about doubling the chances of dying at home? Is this on par with MGH’s early intervention for lung cancer study in terms of breaking ground and shedding new light on the utility and benefits of home palliative care services? My conclusion is that this is a bit of an overstatement. And this is coming from someone who is regularly accused of hyperbole by her spouse (although I’m pretty sure I have asked him 500 times to move his shoes off the stairs).
Finally, I also wonder, if the final product of this research diminishes the full impact. Why were important topics such as location of death, benefit to symptom management, cost efficacy and effects on the caregivers lumped together in this mammoth report? After all, 280 pages is, well, it’s a lot. Looking at other Cochrane Reviews on other “palliative care” topics, none came in with nearly this much muscle. For one, most of the other reviews are much more focused on what seems to be a particular area of treatment/modality of treatment (i.e. use of methadone in relieving cancer pain, drug therapy for symptoms of anxiety), and hence can understandably, perhaps, seem more focused. Most of the outcomes of these areas ended up being inconclusive due to a fair amount of conflicting evidence among the studies. If the outcomes had ended up being more conclusive, I wonder if they would have been peeled off into their own review, rather than tucked into this larger work.
I remain immensely impressed with the work that Gomes and her group did when looking at these studies and their product articles, examining all of these outcomes, and, with their results, leading a public call that additional attention, funding, and support be put toward palliative care programs. It adds to the literature an provides an excellent resource for go-to articles and studies and assesses and analyzes them thoroughly. What it also tells me, and which the review authors also point out, is that research into these outcomes is tough, and further, high quality research remains a need. If this review helps achieve any of these, it has been a huge success.
On the horizon, the Marie Curie Nursing Service and the Nuffield Trust released a research summary last fall, now available in the BMJ Supportive and Palliative Care, which looks at nearly 30,000 patients receiving home based end of life care matched to nearly 30,000 patients without such services. Their research summary is already available online, and the BMJ piece is available for a fee at this time. If you have a chance, and you want to reboost the enthusiasm and glee you initially felt about the Cochrane review, check out the research summary (before I have a chance to deflate your good vibes).
**linguistic tidbit: the original phrase, and the one till used in the UK, is “the proof of the pudding is in the eating.”
2. Shepperd S, Wee B, Straus SE (2011). Hospital at home: home-based end of life care. Cochrane Database of Systemic Reviews (7) DOI: 10.1002/14651858.CD009231
3. Chitnis, X., Georghiou, T., Steventon, A., & Bardsley, M. (2013). Effect of a home-based end-of-life nursing service on hospital use at the end of life and place of death: a study using administrative data and matched controls BMJ Supportive & Palliative Care DOI: 10.1136/bmjspcare-2012-000424
"Pudding Proof" by Christian Sinclair
"Long Road" by Chris Robertshaw/flickr via Creative Commons