Tuesday, July 7, 2009

CPR in Older Patients

The New England Journal of Medicine reports an epidemiologic study of in-hospital CPR in older patients (65 years of age and older). The study identified over 430,000 Medicare beneficiaries who underwent in-hospital CPR between 1992-2005 to answer the question of whether survival rates of in-hospital CPR have improved over that time. They also attempt to determine which patient and hospital characteristics might predict survival.

The rate of survival after CPR hasn't changed much in the time period examined with 18.3% of all CPR recipients surviving to hospital discharge (similar to other studies). Lower survival was significantly associated with male gender (OR 0.97), age (with progressive decline in survival as patients age), Deyo-Charlson score of chronic disease burden, admission from a skilled nursing facility (OR 0.60), and race (black OR of 0.70 and other non-white races OR of 0.85).

Non-metropolitan hospitals had better survival rates, perhaps because sicker patients were referred to metropolitan facilities before they required CPR. Neither the number of hospital beds nor teaching hospital status seemed to make a difference in survival.

Overall, there were 2.73 CPR events per every 1000 admissions. This increased slightly over the period of time investigated (a timeframe which also coincides with the introduction of hospital based palliative care in the United States). Over time, the number of hospital deaths that were preceded by CPR increased from 3.9% in 1992 to 5.2% in 2005.

Neither neurologic outcomes nor quality of life data were reported for survivors. I wonder if it's possible to use rough surrogates for these outcomes such as the ICD-9 code for anoxic brain injury and CPT codes for PEG tube and tracheostomy, but that wasn't done. Furthermore, it would helpful to have 6 month or one year mortality data. The study is already chock full of all sorts of data, so perhaps I'm asking for too much or maybe some of this data will be released later.

Over time, fewer patients are being discharged home after CPR with more going to another hospital (long term acute care hospitals, likely), SNFs, or inpatient hospice. As the authors note, rather than necessarily representing poorer outcomes, this likely represents the fact that patients are being discharged quicker and sicker to other places besides home, although they don't report the hospital length of stay for patients in 1992 vs. 2005.

Some thoughts on the various factors examined:


  1. Age: The percentage of patients receiving CPR increased from 14.6% in the 65-69 year old group to 22.6% in the 75-79 year old group and then decreased to 7.9% in the nonagenarian group. Survival to discharge in those three groups was 22%, 19%, and 12% respectively. The 12% survival rate in the nonagenarian group seems amazing but likely represents a selection bias, as one could envision the 8% in this group who underwent CPR being the most robust physiologically for their age group.

  2. Patients Admitted from Skilled Nursing Facility: While this only represented 2.5% of patients, this is a major risk factor for mortality after CPR. Any patient that returns from a skilled nursing facility to the hospital should have a full exploration of their goals of care, regardless of code status. I wonder how these dismal outcomes compare to being admitted from an intermediate care facility in this population, but alas Medicare doesn't pay for that, so probably doesn't track it.

  3. Chronic Disease Burden (using Deyo-Charlson score): Between a score of 0,1, and 2, survival to hospital discharge didn't vary much (around 19% in each group), but then for those with a score greater than 3 it went down to 16.1%. The greater than 3 crowd is probably quite heterogeneous in their survival rate, and it would be curious to parse this out a bit more.

  4. Race: In this study, blacks survived to hospital discharge 14.3 % of the time compared to 15.9% of other non-whites and 19.2% of whites. Of hospital deaths amongst black patients, 6.6% were preceded by episode of CPR compared to 3.9% of white patients and this incidence has increased significantly since 1992. In the multivariate analysis, they adjusted for hospital location where blacks were more likely to be admitted and this only improved survival slightly. The authors suggest several possibilities for the lower survival in blacks and other non-whites, including concern about quality of care before, during, and after CPR. They cite research which suggests resuscitation and defibrillation might be delayed in blacks leading to a more malignant rhythm at time of CPR. It's also known that blacks are less likely to elect a DNR order, and they suggest this may play a role. Another disparity factor that was not mentioned in their discussion could be the lower rates of end-of-life discussions occuring between black patients and their healthcare providers (as evidenced by a recent study from JAGS in a nursing facility setting).
The authors lament the lack of improvement in CPR outcomes over time and wonder whether it's because we haven't gotten any better at ACLS in the hospital or if we really have but there's a confounder present that makes it look like we haven't. They ponder whether the nature of acute illness has changed towards illnesses that respond less favorably to CPR. Other confounders could include an increase in age and chronic comorbidity level amongst those receiving CPR, but the change in these over time was not reported.

The authors reach one conclusion that will be no surprise to most Pallimed readers:

Of significant concern is our finding that the proportion of patients who died in the hospital after having previously undergone in-hospital CPR has increased during a time of more education and awareness about the limits of CPR in patients with advanced chronic illness and life-threatening acute disease.
As Drew has noted here and here, we still have a long way to go in educating people about CPR and outcomes.

In 1992, the 65-90 year old population was born between 1902 and 1927 whereas in 2005, this cohort was born between 1915 and 1940. There could be a cohort effect going on here, as well, as the earlier cohort was all born well before doctors could routinely save lives with wonder drugs like penicillin, whereas those born in the 1930's were mere babes when the first case report of a penicillin pulling someone back from the brink was published. (Drew pondered the possibility of this type of cohort effect in Ireland as described here.)

Perhaps the advent of hospital based palliative care in the United States has artificially suppressed an increase in non-beneficial CPR in American hospitals? Hard to say- unless you do an analysis of hospitals who had palliative care teams for most of the timeframe vs. those that don't have them to see if the same trends are present (although, once again, one could anticipate many confounders). If not, we can still hope to stem the tide in the future.

Since some elderly patients DO survive CPR with good neurologic outcomes, I'll continue to hope for more research that helps identify those that will benefit most from it so that we can be more selective in our application of resuscitation measures. And it's probably appropriate that a certain percentage of deaths are preceded by unsuccessful CPR, but is 5.2% the right number? Unless we can come up with some new-fangled bedside instant analyzer that tells us prognosis right before we start compressions, it will remain a crapshoot and we can just hope to avoid CPR in those patients where CPR is certain not to meet their goals.

Pallimed | Blogger Template adapted from Mash2 by Bloggermint