Monday, January 16, 2012

The Prognostic Frame of Healthcare for Older Adults (and Everyone Else)

I have a 1 in 542 chance of dying in the next year. In other words, for every 542 men who are the same age as me, one will die in 2012. For those of us who survive until 2017, the probability of death during that year will be 1 in 364. This probability will increase fairly predictably over the years. Gompertz Law of human mortality suggests that the rate doubles every eight years. Supposedly, this law stands the test of time and across populations.

I'm certain actuarialists routinely figure out probabilities that a man my age will become seriously ill or disabled as well. Statistics put mathematical meat onto a mild, chronic anxiety about dying or becoming ill. It's something that most of us feel and I don't trust people who say that they don't have any anxiety about it.

Source: Meandering Through Mathematics
What does one do with that anxiety? Prepare. Prepare for the "worst case" scenarios. Name a surrogate. Talk to that surrogate about theoretical scenarios and make sure they know what you value most. Put it in writing. Get life insurance. Learn to respond to the anxiety you have. If you make it through this next year, the chances are lower that you'll make it through the subsequent year, so you might as well get ready.

Don't be superstitious. Acknowledging death does not make it happen.

Live your life. There is a 99.8% chance that you'll live through the next year if you are my age and your life expectancy going forward is an additional 40 years. Save for retirement. Anticipate being around to pay for your kids' college education. Exercise. Eat well. Seek appropriate preventative healthcare. Find what you love to do and do it. Breathe.

And if you live long enough, you'll likely develop a chronic disease. People are remarkably resilient and can adjust, so don't worry about this too much. With illness, you'll likely develop a greater awareness of your own mortality. Hopefully, you've prepared. If not, it's never too late.

But you'll go to your doctor with the expectation that he'll help improve your quality of life in the face of the chronic disease and also that he'll know what to do to help you avoid death. You've seen the ads on TV for such and such a pill which will improve your vigor. You'll look ten years younger than the average person with your condition, be able to win the stuffed animal for your grandkid, and the lights will go off in your bedroom a full 30-45 minutes before you plan on going to sleep. You imagine it to be so perfect that if a life-threatening condition arises and your doctor doesn't have a pill to fix the problem, you'll think, "This is the year 20XX, why don't they have a pill for this?"

This is the generic frame for the typical clinic visit for patients in the early 21st century. You know that doctors can't fix everything. Your doctor knows she can't fix everything. Both facts frequently go unspoken between the four walls of the clinic room as both sides labor to prevent a "premature" death. It's a noble and worthy goal but Gompertz Law remains. It occasionally crosses your mind: What would it be like to get sick, what type of interventions will you have to go through, what will happen if you become ill and don't have enough support to remain at home? It's crossed your doctor's mind, too. She's seen others go through it, but it's easier to talk about A1C's, LDL's, and holiday plans so concerns go unspoken. Advance care planning in the office remains challenging for multiple reasons, but the frame of the visit does not easily permit acknowledgement of death or other difficult outcomes.

Enter ePrognosis. Researchers from UCSF (including colleagues from Geripal) reviewed prognostic indices for older adults in JAMA last week. The research and the slick new ePrognosis website with easy to use prognostic indices has been widely publicized (all the way around Le Monde!). Their review identified several indices for predicting prognosis in community dwelling, hospitalized, or nursing home bound older adults. The indices aren't perfect. For instance, most need more testing for validity in heterogeneous populations. (See Ken Covinksy's wise analysis at Geripal.) If you're looking for an "expiration date" (who would do that?), then don't hold your breath. We can look at actuarial data or Gompertz Law to get a general sense of prognosis. The tools on ePrognosis sharpen the lens based on variables related to disease state, functional status, etc.

I've written about the importance of framing for discussions about code status in hospitalized patients. In that setting, it's helpful for the patient and doctor to have a common understanding of the patient's prognosis before discussing the code status decision. The same principle applies to the outpatient setting in a non-terminally ill population. My hope is that ePrognosis will help providers prepare the frame for the advance care planning discussion. As Carol Levine points out in her comment below Ken's Geripal post, ePrognosis may unintentionally turn into a "parlor game" for the general population. Even though ePrognosis is intended for provider use, I'm not convinced that use amongst the general population is a bad thing. It may be quite the opposite, as long as you can go "talk to your doctor" about it!

It's always appropriate to prepare for multiple scenarios. Explicit discussion of prognosis, while not an absolute prerequisite, can serve as an entree to discuss the possible scenarios. The healthy young man looks at his probability of death and buys life insurance but prepares at the same time for life as a 50, 60, and 70 year old. In old age, it's still appropriate to prepare for multiple scenarios. If you are an 80 year old with a Mazzaglia Index score is 3, then you have an 8% chance of dying in the next 15 months and a 30% risk of being hospitalized, it makes sense to do some preparation for those events. However, you also have a 92% chance of NOT dying in the next 15 months. What then? What will it take for you to age in place? What happens if frailty or other circumstances make this impossible? Prepare for the most likely and worst case scenarios. It should be your expectation that your healthcare team can help you prepare.

In the JAMA article, the authors suggest using the indices to guide discussions about preventative care (such as whether to pursue a screening colonoscopy in a woman with advanced COPD and comorbidities). This use of the indices makes sense. Colonoscopy, while generally safe, is not without risk of complications, especially in older adults with comorbidities. It's appropriate for physicians to use the indices along with their clinical judgment to guide their recommendations. This has brought about accusations of bedside rationing, which is really a separate discussion. Medicare covers screening colonoscopy once every ten years. I wouldn't anticipate physicians refusing to order a colonoscopy for a non-terminally ill elderly patient based on a prognostic score.

It remains to be seen whether ePrognosis will increase awareness of prognosis in non-terminal, older adults or whether this will impact clinical decisions. I think it will. Time will tell.

(See also this recent NEJM perspective piece and related Geripal commentary.)

Monday, January 16, 2012 by Lyle Fettig, MD ·

Monday, January 2, 2012

The Alignment of HPM and Oncology: Palliative Care Influence on Chemotherapy Use in Metastatic NSCLC

We continue to learn more from the landmark randomized controlled trial conducted by Temel et al from NEJM which demonstrated that outpatient palliative care consultation for patients with newly diagnosed Stage IV lung cancer improves quality of life and may also improve survival by a few months.

Although the survival advantage was a secondary outcome in the original NEJM study, it remains the headliner outcome.  For many who were stuck in an old paradigm of considering palliative medicine as the option of last resort after all other therapies have failed, there was a strong association (mentally, not scientifically) with an idea that hospice and palliative care for patients with cancer meant “letting go of the rope” (i.e. stopping chemotherapy) and that when the transition was made, the patient would go into a free fall towards their death as a result of “letting go.”  Although some lesser quality studies had already suggested that hospice and related interventions might not shorten life and may even prolong it, it was still pleasantly surprising (and for some, inexplicable) when this randomized controlled trial demonstrated a survival advantage for patients seen by a palliative care team. 

"Aikido" from
Wikimedia Commons

As with any result in a clinical trial, it helps to have a plausible mechanism of action to explain the result.  Since this was a “bundled” intervention, it may be difficult to tease out the exact reason why palliative care patients lived longer. Many explanations have been proposed.  Chemotherapy definitely has a role for life-prolongation but has side effects.  Could it be that palliative care helps patients with those side effects, allowing them to receive more of it?  Conversely, could palliative care “protect” patients from toxic chemotherapy near the end of life when chemo may do more harm than good?  Alternatively, is there something intrinsic to palliative care which prolongs life irrespective of chemotherapy?

Last week, the Journal of Clinical Oncology released a secondary analysis of the data from the initial NEJM study data which further elucidates the differences in chemotherapy utilization between the palliative care and standard groups to try to help answer some of these questions.  The authors also reexamined healthcare utilization at the end of life.  Here are some bottom lines.  The palliative care wing did not appear to receive more chemotherapy.  The two groups did not differ with respect to initial anti-cancer therapy (and only one patient…in the standard care wing, incidentally.... didn’t receive anti-cancer therapy at the onset of the trial).  They examined the use of multiple lines of chemotherapy (first, second, third, fourth, and beyond).  Once again, there was no difference between the groups nor was there any difference in how much time it took for the groups to arrive at the next line of chemo. 

How did the two groups differ, then?
  • Days between last IV chemotherapy dose until death (Palliative care group median 64 days vs. Standard Care median 40 days, p=.02).  For oral chemo, the results were 27 days and 17 days respectively but this was not statistically significant.
  • Chemotherapy use in last 60 days of life (Palliative care group 52.5% vs. 70% in the Standard Care group, p=0.05).  A trend towards less chemo use in last 30 and last 14 days of life as well but not statistically significant.
  • Hospice utilization:  While there was no difference between groups with respect to the percentage of patients utilizing hospice, the palliative care group received hospice services longer (median of 24 days vs. 9.5 days in the standard care group).  In the palliative care group, 60% of patients received hospice >7 days before death compared to 33% in the standard care group.  (More on this below.)
When examining location of death, a slightly greater proportion of the palliative care group died at home compared to an inpatient hospice, hospital, or nursing facility, but this was not statistically significant.

It should be noted that the study was not powered to look at many of these outcomes such as chemo use in the last month and 14 days of life as well as location of death.  For instance, with a larger sample size, it’s possible that the increased incidence of home death in the palliative care group might have risen to statistical significance.  However, the lack of statistically significant difference here is not surprising.  While hospital based palliative care can prepare patients/families for the challenges which may come near the end of life, the provision of adequate care within the home to ensure a peaceful death involves many factors which may be outside the purview of the intervention described in the study (e.g. the development of unpredictable symptoms which are difficult to manage at home, capability of caregivers, capacity of the home hospice, etc). 

The study still doesn’t definitively answer the survival mechanism of action question, it only suggests that aggressive chemo at the end of life is not the reason.  The median survival advantage in the original study was about 2.7 months.  In this study, patients receiving the palliative care intervention were free from IV chemo 24 days longer at the end of life and received 14 days more of hospice which doesn’t add up to 2.7 months.  Avoidance of toxic chemo near the end of life and provision of supportive hospice likely play a role, but in the shadows, there appears to be a survival benefit which is unrelated to chemo avoidance or hospice. Improved symptom control and psychological well-being, perhaps?  In Temel et al, the primary outcome was improved QOL 12 weeks after initiation of the intervention.

While we are talking about the study being underpowered, I do have to point out that there was a statistically insignificant trend towards the palliative care group having a longer time from first to second line and from second to third line chemo (0.8 months longer in each case).  This result came about only when participants were censored at time of analysis--when the numbers were crunched censoring participants at either date of death or date of analysis, there was no difference between groups in time to next line of chemo.  I'm not sure which method is better for detecting a difference between groups, but this does suggest that with greater power, the study might have detected a statistically significant difference.  In other words, we still can't rule out the possibility that palliative care helped patients receive more chemo early on in the course of their cancer therapy.

It was never my hypothesis that palliative care helped patients receive more chemo, thus resulting in the survival advantage.  I've always thought that the benefits of excellent symptom/psychosocial support and avoidance of chemo near the end of life were more plausible explanations.  Don’t confuse this with a bias against chemotherapy for patients with advanced lung cancer.  I fully embrace the role for chemotherapy in advanced lung cancer and the palliative care practitioners in this study were obviously not chemotherapy obstructionists.  I am familiar enough with oncology clinical trials (without being an oncologist) to know they lend more guidance to oncologists regarding when to start the agents and less guidance regarding when to stop. While sometimes the decision to stop is clear cut (patient has catastrophic complication or documented progression of disease on imaging with performance status slipping to ECOG 3), there are other times when the decision is more arbitrary.

For example, consider a patient in the middle of third line chemo who is experiencing more fatigue whose ECOG slips from 1 to 2 where everyone is suspecting progression but it may be just as easy to continue with the status quo until the next scan or until a complication comes up.  At this point, many patients may be thinking that pushing forward with chemo isn't worth it to them, but what's next?  That question is often wrought with fear.  Palliative care might help “tip the scales” towards discontinuation in this circumstance a couple of ways.  Through advance care planning discussions, "what next" becomes clearer: The role of hospice may be better understood and end of life fears already addressed.  Even if hospice isn’t initiated immediately, palliative care is already part of the status quo, perhaps enabling either the oncologist or the patient to be more proactive with saying “hey, this (chemo) doesn’t seem like it’s working” with less of a feeling that they are “giving up.”

One of the most frequently noted caveats to this study is the fact that it's a single institution. Other institutions vary with respect to patient population, oncology practice, and palliative care team practice.   The palliative care intervention brought the median hospice length of stay just above the national median (21.1 days in 2009) and 60% of the palliative care patients received hospice >7 days which is below the national rate of 65%. (Source: NHPCO Facts and Figures).   It's worth remembering that the palliative care intervention followed guidelines from the National Consensus Project for Quality Palliative Care. Adherence to those guidelines at other institutions should also mean that patients will see many of the same benefits even if the results of Temel et al and the present study couldn't be exactly replicated.
It’s notable that the American Society of Clinical Oncology (ASCO) is looking carefully at the quality of end of life care within oncology programs (quote from present study):
“With accumulating evidence for the benefits of earlier referral to hospice for patients with terminal cancer, leaders in oncology have established guidelines for quality end-of-life care. The American Society of Clinical Oncology Quality Oncology Practice Initiative has delineated several key metrics for determining high-quality treatment near death, such as no chemotherapy within the last 2 weeks of life, referral to hospice, and enrollment in hospice care more than 1 week before death.  Palliative care clinicians, who are accustomed to practicing within a multidisciplinary team, are ideally suited to support oncology providers, patients, and families in discussing and planning for such treatment at the end of life. To date, to our knowledge, no studies have shown the utility of these services for altering the quality of end-of-life care.”
The QOPI measures are worth reviewing.  As palliative care programs track metrics and achieve TJC certification, it's only natural that we should work with our oncology colleagues to help with the end of life QOPI metrics as well.  The results of this study indicate that our work helps oncologists with those measures. 

Lastly, a friendly editorial which accompanies the study is also worth a read. 

Monday, January 2, 2012 by Lyle Fettig, MD ·