Commentary: Critical care bed management: Can we do better?

Is it possible to give the best critical care while spending less money and resources doing it? Can we reduce waste while improving quality in a so-called lean approach to critical care? I believe that we have too many critical care beds, and we fill some of those beds with patients who can be taken care of at less intense levels of care—which are also less expensive.

Most work that is done to improve critical care looks at the quality of care. This is an area where a lot of data are accumulating. Take septic shock, for example. In the recently published ProCESS trial (The ProCESS Investigators. N Engl J Med. 2014. 370[18]:1683), the 60-day in-hospital mortality for septic shock was 18.2% to 21.0%. A lot of institutions (including mine) are struggling to get their septic shock mortality rate under 30%. Although some people critique the ProCESS trial mortality rate on patient selection, most of us try to figure out how to duplicate that lower rate. We do this in areas other than septic shock. If we are comparable in whatever quality statistic, we applaud our success. If we aren’t comparable, we look at ways to improve, often based on what was done in that particular study.

How big of a financial burden is our critical care spending? According to an analysis of critical care beds by Halpern and colleagues (Crit Care Med. 2004;32[6]:1254), the number of hospital beds decreased 26.4% between 1985 and 2000, and the absolute number of critical care beds increased 26.2% (quantitated at 67,357 adult beds in 2007 per SCCM.org, www.sccm.org/Communications/Pages/CriticalCareStats.aspx). Critical care beds cost $2,674 per day in 2000, up from $1,185 (our CFOs tell us it is more like $3,500 to $4,000 per day now). They represented 13.3% of hospital costs, 4.2% of national health expenditures (NHE), and 0.56% of gross domestic product (GDP). There are 55,000 critically ill patients cared for each day in the United States, representing 5 million ICU patients per year. This is an enormous expenditure of money and it is growing.

Another interesting observation by Halpern and colleagues (Crit Care Med. 2004;32:1254) was that critical care beds were only at 65% occupancy. This reflects my own experience where we operate at a 70% average ICU bed occupancy. We have created a larger financial burden with the fixed costs of one-third more ICU beds than we actually use. Some bed availability is desirable, but how much is too much? Are we doing the best job to give quality care and spend money wisely? Can we be more efficient in the throughput of patients and in their care? Admission criteria should be part of any unit, designed to place all patients who need ICU care appropriately in the ICU and exclude those whose care can be managed at a lesser level of intensity and cost. Discharge criteria, care protocols (e.g., wake up and wean), checklists, and daily attention to the usual parameters (e.g., DVT prophylaxis) are essential for high-quality but efficient care. Done 24/7, we can maximize efficiency and quality with a minimum of ICU readmissions. Throughput is part of every physician’s job description. The physician who wants one more day for his or her patient in the ICU simply because the nurse has fewer patients misses a number of points. Why would anyone want more exposure to resistant organisms, more noise, more awakenings, and less sleep, just to name a few? Keeping that non-ICU patient in the ICU bed might even delay the transfer of another patient coming from the ED, where we know they often don’t get good ICU care.

Are the beds filled only with what we intensivists would consider legitimate ICU patients, defined by both generally accepted (endotracheal tube in place) and individually specified criteria (unit specific related to other unit capabilities)? That would impact cost. An interesting article by Gooch and Kahn (JAMA. 2014; 311[6]:567) discussed the demand elasticity of the ICU. They considered the changes in case mix of patients between days of high and low bed availability. They contended that when ICU beds were available, there was an increase in patients who were unlikely to benefit from ICU admission. This group included a population of patients likely to survive and whose illness severity was low and a population of patients who were unlikely to survive and had a high illness severity. In other words, admissions expand to fill the staff-able beds. If this is true, it is another area where better management could lower costs without reducing the quality of care.

What if bed availability truly is reduced, often by a lack of critical care nursing staff if not physical beds? Here the answer is unclear. Town (Crit Care Med. 2014;42[9]:2037) looked at ICU readmission rates and the odds of having a cardiac arrest on the ward related to bed availability. Five ICUs with 63 beds total were examined. As ICU bed availability decreased, the odds of patients who were discharged from the ICU being readmitted to the ICU went up. Also, the odds of patients having a cardiac arrest on the ward increased when medical (not total) ICU beds were less available. In 2013, Wagner and colleagues (Ann Intern Med. 2013;159[7]:447) looked at 155 ICUs with 200,730 patients discharged from ICUs to hospital floors from 2001 to 2008. They examined what they call the strain metrics. These included the standardized ICU census, the proportion of new admissions, and the average predicted probability of death of the other patients in the ICU on the days of ICU discharge. When the strain metrics increased, ICU patients had shorter ICU length of stay and ICU readmission odds went up. They didn’t, however, see an increased odds of death, a reduced odds of being discharged home, or a longer total hospital LOS. In a third study reported in 2008 in the Annals of Internal Medicine by Howell and colleagues (Ann Intern Med. 2008;149[11]:804), an innovative method of bed management was described. Because of an overcrowded ED and a high ambulance diversion rate, hospitalists implemented a system of bed control that was based on knowledge of ICU beds and ED congestion and flow. Bed assignments were better controlled by twice-daily ICU rounds and regular visits to the ED: throughput for admitted patients decreased by 98 minutes and time on diversion decreased significantly.

Mery and Kahn reported in 2013 (Crit Care. 2013;17[3]:315) that when ICU bed availability was reduced, there was a reduction in the likelihood of ICU admission within 2 hours of a medical emergency team (MET) activation. What is interesting about this study done in three hospitals in Calgary, Alberta, Canada, is that there was an increased likelihood that the patient goals of care changed to comfort care when there was no bed availability, compared with two ICU beds being available. Even more interesting is that hospital mortality did not vary significantly by ICU bed availability: More patients were moved to palliative care yet no more people died. Perhaps a lack of ICU beds expedited appropriateness of care.

To summarize, we have more patients in critical care beds where we spend ever-increasing amounts of our health-care dollars, but we seem to have more critical care beds than we need. We still have patients in our ICUs who would be better cared for elsewhere in our institutions. We can perform more cost-effective throughput when we are pressed to do so and usually we can do it safely.

I contend that the next improvement in lean ICU medicine will be better management tools. Comprehensive checklists have helped me where computer solutions have yet to be developed. I am working to create hardware/software management solutions that will make my job more cost-effective and provide a sustainable process for what comes after me.

Dr. Waxman is associate professor of medicine, KU School of Medicine, Kansas City, Kan.; medical director, Medical Surgical ICU/PCU, Research Medical Center; and adjunct professor, Rockhurst University, Helzberg School of Management, Kansas City, Mo.

Is it possible to give the best critical care while spending less money and resources doing it? Can we reduce waste while improving quality in a so-called lean approach to critical care? I believe that we have too many critical care beds, and we fill some of those beds with patients who can be taken care of at less intense levels of care—which are also less expensive.

Most work that is done to improve critical care looks at the quality of care. This is an area where a lot of data are accumulating. Take septic shock, for example. In the recently published ProCESS trial (The ProCESS Investigators. N Engl J Med. 2014. 370[18]:1683), the 60-day in-hospital mortality for septic shock was 18.2% to 21.0%. A lot of institutions (including mine) are struggling to get their septic shock mortality rate under 30%. Although some people critique the ProCESS trial mortality rate on patient selection, most of us try to figure out how to duplicate that lower rate. We do this in areas other than septic shock. If we are comparable in whatever quality statistic, we applaud our success. If we aren’t comparable, we look at ways to improve, often based on what was done in that particular study.

How big of a financial burden is our critical care spending? According to an analysis of critical care beds by Halpern and colleagues (Crit Care Med. 2004;32[6]:1254), the number of hospital beds decreased 26.4% between 1985 and 2000, and the absolute number of critical care beds increased 26.2% (quantitated at 67,357 adult beds in 2007 per SCCM.org, www.sccm.org/Communications/Pages/CriticalCareStats.aspx). Critical care beds cost $2,674 per day in 2000, up from $1,185 (our CFOs tell us it is more like $3,500 to $4,000 per day now). They represented 13.3% of hospital costs, 4.2% of national health expenditures (NHE), and 0.56% of gross domestic product (GDP). There are 55,000 critically ill patients cared for each day in the United States, representing 5 million ICU patients per year. This is an enormous expenditure of money and it is growing.

Another interesting observation by Halpern and colleagues (Crit Care Med. 2004;32:1254) was that critical care beds were only at 65% occupancy. This reflects my own experience where we operate at a 70% average ICU bed occupancy. We have created a larger financial burden with the fixed costs of one-third more ICU beds than we actually use. Some bed availability is desirable, but how much is too much? Are we doing the best job to give quality care and spend money wisely? Can we be more efficient in the throughput of patients and in their care? Admission criteria should be part of any unit, designed to place all patients who need ICU care appropriately in the ICU and exclude those whose care can be managed at a lesser level of intensity and cost. Discharge criteria, care protocols (e.g., wake up and wean), checklists, and daily attention to the usual parameters (e.g., DVT prophylaxis) are essential for high-quality but efficient care. Done 24/7, we can maximize efficiency and quality with a minimum of ICU readmissions. Throughput is part of every physician’s job description. The physician who wants one more day for his or her patient in the ICU simply because the nurse has fewer patients misses a number of points. Why would anyone want more exposure to resistant organisms, more noise, more awakenings, and less sleep, just to name a few? Keeping that non-ICU patient in the ICU bed might even delay the transfer of another patient coming from the ED, where we know they often don’t get good ICU care.

Are the beds filled only with what we intensivists would consider legitimate ICU patients, defined by both generally accepted (endotracheal tube in place) and individually specified criteria (unit specific related to other unit capabilities)? That would impact cost. An interesting article by Gooch and Kahn (JAMA. 2014; 311[6]:567) discussed the demand elasticity of the ICU. They considered the changes in case mix of patients between days of high and low bed availability. They contended that when ICU beds were available, there was an increase in patients who were unlikely to benefit from ICU admission. This group included a population of patients likely to survive and whose illness severity was low and a population of patients who were unlikely to survive and had a high illness severity. In other words, admissions expand to fill the staff-able beds. If this is true, it is another area where better management could lower costs without reducing the quality of care.

What if bed availability truly is reduced, often by a lack of critical care nursing staff if not physical beds? Here the answer is unclear. Town (Crit Care Med. 2014;42[9]:2037) looked at ICU readmission rates and the odds of having a cardiac arrest on the ward related to bed availability. Five ICUs with 63 beds total were examined. As ICU bed availability decreased, the odds of patients who were discharged from the ICU being readmitted to the ICU went up. Also, the odds of patients having a cardiac arrest on the ward increased when medical (not total) ICU beds were less available. In 2013, Wagner and colleagues (Ann Intern Med. 2013;159[7]:447) looked at 155 ICUs with 200,730 patients discharged from ICUs to hospital floors from 2001 to 2008. They examined what they call the strain metrics. These included the standardized ICU census, the proportion of new admissions, and the average predicted probability of death of the other patients in the ICU on the days of ICU discharge. When the strain metrics increased, ICU patients had shorter ICU length of stay and ICU readmission odds went up. They didn’t, however, see an increased odds of death, a reduced odds of being discharged home, or a longer total hospital LOS. In a third study reported in 2008 in the Annals of Internal Medicine by Howell and colleagues (Ann Intern Med. 2008;149[11]:804), an innovative method of bed management was described. Because of an overcrowded ED and a high ambulance diversion rate, hospitalists implemented a system of bed control that was based on knowledge of ICU beds and ED congestion and flow. Bed assignments were better controlled by twice-daily ICU rounds and regular visits to the ED: throughput for admitted patients decreased by 98 minutes and time on diversion decreased significantly.

Mery and Kahn reported in 2013 (Crit Care. 2013;17[3]:315) that when ICU bed availability was reduced, there was a reduction in the likelihood of ICU admission within 2 hours of a medical emergency team (MET) activation. What is interesting about this study done in three hospitals in Calgary, Alberta, Canada, is that there was an increased likelihood that the patient goals of care changed to comfort care when there was no bed availability, compared with two ICU beds being available. Even more interesting is that hospital mortality did not vary significantly by ICU bed availability: More patients were moved to palliative care yet no more people died. Perhaps a lack of ICU beds expedited appropriateness of care.

To summarize, we have more patients in critical care beds where we spend ever-increasing amounts of our health-care dollars, but we seem to have more critical care beds than we need. We still have patients in our ICUs who would be better cared for elsewhere in our institutions. We can perform more cost-effective throughput when we are pressed to do so and usually we can do it safely.

I contend that the next improvement in lean ICU medicine will be better management tools. Comprehensive checklists have helped me where computer solutions have yet to be developed. I am working to create hardware/software management solutions that will make my job more cost-effective and provide a sustainable process for what comes after me.

Dr. Waxman is associate professor of medicine, KU School of Medicine, Kansas City, Kan.; medical director, Medical Surgical ICU/PCU, Research Medical Center; and adjunct professor, Rockhurst University, Helzberg School of Management, Kansas City, Mo.

Is it possible to give the best critical care while spending less money and resources doing it? Can we reduce waste while improving quality in a so-called lean approach to critical care? I believe that we have too many critical care beds, and we fill some of those beds with patients who can be taken care of at less intense levels of care—which are also less expensive.

Most work that is done to improve critical care looks at the quality of care. This is an area where a lot of data are accumulating. Take septic shock, for example. In the recently published ProCESS trial (The ProCESS Investigators. N Engl J Med. 2014. 370[18]:1683), the 60-day in-hospital mortality for septic shock was 18.2% to 21.0%. A lot of institutions (including mine) are struggling to get their septic shock mortality rate under 30%. Although some people critique the ProCESS trial mortality rate on patient selection, most of us try to figure out how to duplicate that lower rate. We do this in areas other than septic shock. If we are comparable in whatever quality statistic, we applaud our success. If we aren’t comparable, we look at ways to improve, often based on what was done in that particular study.

How big of a financial burden is our critical care spending? According to an analysis of critical care beds by Halpern and colleagues (Crit Care Med. 2004;32[6]:1254), the number of hospital beds decreased 26.4% between 1985 and 2000, and the absolute number of critical care beds increased 26.2% (quantitated at 67,357 adult beds in 2007 per SCCM.org, www.sccm.org/Communications/Pages/CriticalCareStats.aspx). Critical care beds cost $2,674 per day in 2000, up from $1,185 (our CFOs tell us it is more like $3,500 to $4,000 per day now). They represented 13.3% of hospital costs, 4.2% of national health expenditures (NHE), and 0.56% of gross domestic product (GDP). There are 55,000 critically ill patients cared for each day in the United States, representing 5 million ICU patients per year. This is an enormous expenditure of money and it is growing.

Another interesting observation by Halpern and colleagues (Crit Care Med. 2004;32:1254) was that critical care beds were only at 65% occupancy. This reflects my own experience where we operate at a 70% average ICU bed occupancy. We have created a larger financial burden with the fixed costs of one-third more ICU beds than we actually use. Some bed availability is desirable, but how much is too much? Are we doing the best job to give quality care and spend money wisely? Can we be more efficient in the throughput of patients and in their care? Admission criteria should be part of any unit, designed to place all patients who need ICU care appropriately in the ICU and exclude those whose care can be managed at a lesser level of intensity and cost. Discharge criteria, care protocols (e.g., wake up and wean), checklists, and daily attention to the usual parameters (e.g., DVT prophylaxis) are essential for high-quality but efficient care. Done 24/7, we can maximize efficiency and quality with a minimum of ICU readmissions. Throughput is part of every physician’s job description. The physician who wants one more day for his or her patient in the ICU simply because the nurse has fewer patients misses a number of points. Why would anyone want more exposure to resistant organisms, more noise, more awakenings, and less sleep, just to name a few? Keeping that non-ICU patient in the ICU bed might even delay the transfer of another patient coming from the ED, where we know they often don’t get good ICU care.

Are the beds filled only with what we intensivists would consider legitimate ICU patients, defined by both generally accepted (endotracheal tube in place) and individually specified criteria (unit specific related to other unit capabilities)? That would impact cost. An interesting article by Gooch and Kahn (JAMA. 2014; 311[6]:567) discussed the demand elasticity of the ICU. They considered the changes in case mix of patients between days of high and low bed availability. They contended that when ICU beds were available, there was an increase in patients who were unlikely to benefit from ICU admission. This group included a population of patients likely to survive and whose illness severity was low and a population of patients who were unlikely to survive and had a high illness severity. In other words, admissions expand to fill the staff-able beds. If this is true, it is another area where better management could lower costs without reducing the quality of care.

What if bed availability truly is reduced, often by a lack of critical care nursing staff if not physical beds? Here the answer is unclear. Town (Crit Care Med. 2014;42[9]:2037) looked at ICU readmission rates and the odds of having a cardiac arrest on the ward related to bed availability. Five ICUs with 63 beds total were examined. As ICU bed availability decreased, the odds of patients who were discharged from the ICU being readmitted to the ICU went up. Also, the odds of patients having a cardiac arrest on the ward increased when medical (not total) ICU beds were less available. In 2013, Wagner and colleagues (Ann Intern Med. 2013;159[7]:447) looked at 155 ICUs with 200,730 patients discharged from ICUs to hospital floors from 2001 to 2008. They examined what they call the strain metrics. These included the standardized ICU census, the proportion of new admissions, and the average predicted probability of death of the other patients in the ICU on the days of ICU discharge. When the strain metrics increased, ICU patients had shorter ICU length of stay and ICU readmission odds went up. They didn’t, however, see an increased odds of death, a reduced odds of being discharged home, or a longer total hospital LOS. In a third study reported in 2008 in the Annals of Internal Medicine by Howell and colleagues (Ann Intern Med. 2008;149[11]:804), an innovative method of bed management was described. Because of an overcrowded ED and a high ambulance diversion rate, hospitalists implemented a system of bed control that was based on knowledge of ICU beds and ED congestion and flow. Bed assignments were better controlled by twice-daily ICU rounds and regular visits to the ED: throughput for admitted patients decreased by 98 minutes and time on diversion decreased significantly.

Mery and Kahn reported in 2013 (Crit Care. 2013;17[3]:315) that when ICU bed availability was reduced, there was a reduction in the likelihood of ICU admission within 2 hours of a medical emergency team (MET) activation. What is interesting about this study done in three hospitals in Calgary, Alberta, Canada, is that there was an increased likelihood that the patient goals of care changed to comfort care when there was no bed availability, compared with two ICU beds being available. Even more interesting is that hospital mortality did not vary significantly by ICU bed availability: More patients were moved to palliative care yet no more people died. Perhaps a lack of ICU beds expedited appropriateness of care.

To summarize, we have more patients in critical care beds where we spend ever-increasing amounts of our health-care dollars, but we seem to have more critical care beds than we need. We still have patients in our ICUs who would be better cared for elsewhere in our institutions. We can perform more cost-effective throughput when we are pressed to do so and usually we can do it safely.

I contend that the next improvement in lean ICU medicine will be better management tools. Comprehensive checklists have helped me where computer solutions have yet to be developed. I am working to create hardware/software management solutions that will make my job more cost-effective and provide a sustainable process for what comes after me.

Dr. Waxman is associate professor of medicine, KU School of Medicine, Kansas City, Kan.; medical director, Medical Surgical ICU/PCU, Research Medical Center; and adjunct professor, Rockhurst University, Helzberg School of Management, Kansas City, Mo.