Intensive care admission practices varied widely at Veterans Affairs hospitals, suggesting a lack of consensus about high- and low-risk patients, investigators concluded.
Out of 31,555 patients admitted directly to 118 acute-care hospitals included in the analysis, more than 50% had a 30-day predicted mortality of 2% or less, reported Lena Chen, MD, of the University of Michigan in Ann Arbor, and colleagues in Archives of Internal Medicine online.
The rate of admission for low-risk patients ranged from 1.2% to 38.9% and similar variation was observed in ICU admission practices for high-risk patients with a predicted mortality of more than 30%, the authors wrote.
In total, 66.1% of the hospitals in the study were in different quartiles of ICU use for low- versus high-risk patients.
"Hospital admitting patterns varied widely for patients at every level of severity," they said. "These findings suggest that there may be considerable lack of consensus about when to use the ICU."
Studies aimed at identifying factors associated with ICU admission could aid in the development of admission standards tailored to different levels of clinical risk, they added.
Critical care resources account for 15% of all hospital costs (Crit Care Med 2010; 38: 65-71). Variation in use of critical care resources has implications for both the quality and cost of inpatient care, the authors noted.
Several studies have examined patient characteristics and illness severity associated with ICU admissions, but no published data have documented the proportion of new inpatients admitted directly to the ICU across a broad sample of hospitals within a national healthcare system. Even less is known about how the proportion might vary from hospital to hospital, the authors added.
To examine patterns of ICU admission within the VA health system, Chen and colleagues analyzed data associated with the first nonsurgical admission from an emergency department or outpatient clinic at 118 acute-care hospitals within the system. They limited their review to a 12-month period during 2009 and 2010.
The authors sought to address three questions:
- What is the 30-day predicted mortality of patients admitted directly to the ICU?
- For patients with the same predicted mortality, to what extent does direct ICU admission vary?
- Does patient severity influence admission patterns?
The final analysis included 31,555 patients, or 10.9% of all admissions (289,310) during the study period. The most common diagnoses associated with direct admission to the ICU were sepsis (12.8%), acute myocardial infarction (8%), coronary artery disease (7.9%), and dysrhythmia (7.2%). Almost half (49.2%) of the patients had diagnoses classified as "other."
The authors found that 53.2% of the patients had a 30-day predicted mortality of 2% or less, followed by 18.7% with a predicted mortality of 2% to 5%. A little over 10% had a predicted mortality of 5% to 10%, while 10.9% had a predicted mortality of 10% to 30%. Those with a predicted mortality in excess of 30% came in at 7.1%.
ICU occupancy was ≤75% at the time of admission in 80% of cases. Overall, the ICU admission rates across the 118 hospitals ranged from 1.6% to 29.5%.
The median rate of direct admission to the ICU was 7.3%. Almost 10% of total variability was explained by patient severity and diagnosis, whereas 0.4% could be explained by ICU occupancy and the facility's complexity level.
For every one standard-deviation increase in illness severity, the odds ratio for ICU admission increased to 1.50. However, the odds varied widely across hospitals, from 0.85 to 2.22.
As a result of the hospital admitting patterns' lack of sensitivity to predicted mortality, two-thirds of the facilities (66.1%) fell into different quartiles of ICU admission for patients with the lowest versus highest 30-day predicted mortality.
After adjusting for severity at admission, the authors found that patients admitted directly to the ICU had a higher mortality risk at 30 days but not at 90 days. When treatment effect was estimated as a function of patient severity, admission to an ICU was associated with a reduced risk of death at 30 days among patients with a 30-day predicted mortality >18.4% at admission.
The study adds to evidence of the uncertain role that ICU admission has in the U.S. healthcare system, according to the authors of an invited commentary.
"The wide variation in ICU use observed in this study should prompt us to consider the drivers of this variation and which patients truly need intensive care," wrote Christopher W. Seymour, MD, and Jeremy M. Kahn, MD, of the University of Pittsburgh.
Low-risk patients in ICU beds might reflect poor triage decisions, inability to assess illness severity at the point of care, or an "institutional culture of ICU overuse." Low admission rates for high-risk patients might also reflect poor triage or, alternatively, "successful solicitation of end-of-life treatment preferences among patients likely to die."
Primary source: Archives of Internal Medicine
Additional source: Archives of Internal Medicine