Medicare Beneficiaries and Three Year Mortality After ICU Stay

If you have an older patient who is mechanically ventilated in the ICU, the chances that the patient will die within the next three years seem pretty high when they are in the middle of the acute illness, don't they? But what if that patient survives the hospitalization? What would you say about the patient's long term outlook? The answer will largely depend on individual factors, of course, but I find cognitive dissonance in many scenarios when a patient survives the ICU: The hope that comes with surviving such an illness is pitted against the fear that the illness was a harbinger of worse things to come (especially in the context of chronic illness) . How does this patient population fare in general?

JAMA published a matched retrospective cohort study which examines the outcomes of Medicare beneficiaries after an ICU stay. The study also reports some notable data related to mortality in patients sent to a skilled nursing facility.

The investigators started with a 5% random sample of all Medicare beneficiaries over the age of 65 and divided that sample into two. The first sample comprised the study cohort of ~35,000 patients who survived a hospitalization that included an ICU stay in 2003. The second sample comprised two control cohorts: an "I survived the hospitalization and didn't have to go to the ICU" cohort and a general population cohort. The latter two cohorts were matched with the study cohort based on age, race, sex, and whether the hospitalization had a surgical or medical DRG.

The cohorts on the whole were similar with respect to gender, race, and age. Not surprisingly, those patients admitted to the ICU had a greater burden of comorbid chronic illnesses, and the mechanically ventilated ICU patients were the sickest, with 57% having three or more comorbid conditions.

Some of the results:

• 6 month mortality: ICU survivor mortality was 14.1% (30.1% for those receiving mechanical ventilation) compared with 10.9% for hospital controls and 2.7% for the general controls
• 3 year mortality: ICU survivor mortality was 39.5% (57.6% for those receiving mechanical ventilation) compared with 34.5% for hospital controls and 14.9% for general controls
  
• Discharge to a skilled care facility was an independent predictor of mortality at both 6 months and 3 years, regardless of ICU status during hospitalization (6 month and 3 year mortalities, respectively: 24.1%/54.6% for ICU/Hospitalized controls who went to SNF vs. 7.5%/29.4% discharged home in both cohorts.)
• Readmission to the hospital was a common event in all hospitalization cohorts, but most common in those requiring ICU care (45% in first year for MV patients and 43% for non-MV ICU patients).
  

As a palliative care consultant, have you ever had the experience of being consulted on a really sick patient in the ICU (or anywhere for that matter) only to be called back a short time later with a "consult revocation" because a) it was decided that the patient's prognosis was "better" than initially thought or b) "we're still treating aggressively"? That happens to me occasionally, and my response usually is "we'll plan on seeing the patient anyway, if that's ok." (ie No take backs unless there's a really compelling reason.) This study proves that mechanically ventilated older patients who have a "good" acute prognosis are still faced with a future of uncertainty with more than half having a high burden of comorbidities, close to half requiring recurrent hospitalization in the first year, and close to 6/10 dying in the subsequent three years.

The results are difficult to apply to an individual patient. A previously healthy 70 year old patient who survives mechanical ventilation for pneumonia obviously has a different prognosis than a 70 year old who has end stage renal disease, CHF, and diabetes who survives mechanical ventilation for line sepsis.

But the results should be heeded by hospitals, ICUs, and palliative care teams. Wide variation exists in the volume and type of utilization of palliative care in ICUs. Palliative care is typically consulted when it's clear that the patient is not getting better. As I've said recently, that's not the only type of patient we should be seeing. A poor prognosis need not be certain. The higher mortality/comorbid burdened patients described in this study (especially the mechanically ventilated cohort) could be perfect for palliative care, even though they all survived hospitalization. The challenge to our field would be providing the capacity to meet the demands of this population. Questions that we need to ask of ourselves include:

1. Do we offer a time-limited consultation for advance care planning/goals of care delineation/complicated discharge planning with a plan to "remain on the sidelines" after we've helped answer those initial questions?
2. Or do we provide chronic care for patients like this (really would need increased provider capacity)?
3. Related to 2, what about palliative care at SNFs?
   
4. How many of the palliative care interventions should be "organic" within the ICU (without PC consultation)? How do we best ensure PC issues remain a focus after the patient is discharged?

The answers may depend on the setting but I'm hopeful that some uniform answers will evolve.

Here's the last paragraph in the discussion:

The magnitude of the postdischarge use of skilled care facilities for both ICU survivors and hospital controls and the high long-term mortality for all of these patients call into question whether discharge to skilled care facilities is merely a marker for higher severity of illness with appropriate delivery of care. These patients could have been discharged prematurely from acute care hospitals, and needed a higher level of care than they received. It also is possible that these patients could have had better outcomes if discharged home, but were not able to be sent there due to lack of sufficient support from family or friends to act as caregivers. These findings highlight the need for a much more detailed understanding of the long-term care needs of these patients.

In the age of throughput, there's no doubt in my mind that some of these patients were discharged from the hospital too early. However, I'm bemused in trying to sort out how these very sick patients might have better outcomes at home if they only had better caregiver support, assuming that by "outcome," the authors refer to mortality. The only outcome that might have been better for some of these patients at home would be a palliative care outcome: receiving care in the setting of choice. To that end, I do agree that we need many more details regarding the long-term care needs of this population.

See other related posts on prognosis in the ICU here and here.

(Illustration courtesy of Niels Olson on Flickr. See here for the original, which has scroll over notes.)