Cost-savings of PC teams

Archives of Internal Medicine published a large study looking at cost savings associated with hospital palliative care services in the US. There have been press-releases about this circulating the palliative care email circuit, which many of you have likely seen. The study uses hospital administrative data (from 8 major CAPC/PCLC affiliate hospitals - a mix of academic and community hospitals throughout the US), and compares costs (costs were measured the same way across hospitals) between patients who received PC consultation and those who didn't. Propensity scoring was used - PC-consultation patients were matched with nearly-identical patients (based on a propensity score model) who didn't receive PC consultation, and then the two groups were compared. In-hospital-death patients were analyzed separately from those who were discharged alive. (FYI: my institution is one of the sites, but the data are from before my time here which is probably a good thing since I have a habit of giving Xigris, TPN, and Epo to my dying patients. I have no involvement with this paper or research project.)

The analysis included ~ 2600 PC-discharged patients (matched to 18000 usual care patients) and ~ 2300 PC-deaths patients (matched to ~2100 usual care patients): both categories included ~90% of all PC patients eligible for the study. Length of stay was the same between PC and usual care patients (which is clearly important given that hospital costs are closely related to LOS). Major outcomes: for discharged-alive patients PC consultation was associated with ~$2600/admission (~$280/day) in cost-savings. For patients who died, savings were ~$6900 per admission ($549/day). When adjusted for costs of PC services the numbers were still in the many thousands per admission. For those of you who are interested they break down where the cost savings lie (e.g. lab, pharmacy, imaging, ICU costs).

All well and good, and not surprising (perhaps it's not surprising to me since my boss tracks such data locally and shares it with us all the time). The obvious question, though, is to ask if the involvement of the PC service was incidental to the cost savings (ie - the decisions to limit certain treatments were already made and PC was called in afterwards to assist with enacting the 'new plan'). To rephrase - while PC consultation is associated with lower hospital costs it is not necessarily the cause of those cost savings. The authors looked at that question by looking at when the cost savings occurred in relation to consultation, and by including mean cost per day before consultation (or a similar time reference for nonPC patients derived from overall length of stay and usual LOS before PC consultation) in the propensity score model.  Basically they showed that:

1. Costs were similar prior to PC consultation between groups (they weren't dropping off in the PC group compared to the non-PC group prior to PC consultation - that is, the cost savings did, in fact, accrue after PC consultation)
   
2. Cost savings didn't begin to occur until ~48 hours after PC consultation, again suggesting that the PC consultation itself impacted costs and was not 'after the fact.'  

One could still argue that those patients, due to their goals, would have 'gone palliative' even without specialist PC services and those costs would have dropped off anyway, but, jeez, the only way to put that one to rest is an actual randomized trial of PC services which is not going to happen.  And those of us who do this I think can attest that we frequently are called not because goals of care are clear and all the decisions to be made are made and the treatment limitations are enacted but we are called because the goals aren't clear, the treatment options/decisions haven't been decided, and patients, families, and their doctors are looking for help as to what to do.  I personally am convinced by the data and doubt there will be better data coming about cost-savings for a while (unless it's a similarly designed but larger trial - although hell - when was the last time you saw a palliative care study involving over 20,000 patients?).  

However there is always room for hand-wringing and nay-saying!

1. How will this be perceived by patients and families?  These findings are great for the profession, and bolster the currently dominant model supporting hospital-based PC services (many programs are supported by hospitals with the idea that they save the hospital money, as it's difficult to support a program by income generated by billing alone). More and more of my patients have some idea of what the word palliative means (it's usually something along the lines of the morphine-hospice-death&dying-pull-the-plug doctors): if it 'gets out' that we're also the save-The-Man-some-money via stopping medical treatment - people may mistrust us some more.  I'm not too worried about this: there's tons of evidence that PC consultation improves patient/family-centric outcomes too, and I think most of my patients/families are happy we're involved in the long run, but this is not necessarily a message we need to broadcast from the mountain top.
2. What if the dreams of some PC leaders come true and we get involved earlier in disease courses, expand our patient base to include more and more of the not-close-to-dying patient population and the highly-morbid-and-life-threatening-but-not-necessarily-incurable-illness patient population (e.g. BMT patients)?  Will these cost savings diminish, or even become immeasurable unless you only include the 'close-to-death' patients?  What then will happen to the economic rationale underpinning hospital-based PC services? 

(The same issue also has a look at what end-of-life 'services' are provided for patients dying in Belgium.)