Rethinking the Use of Intensive Care Beds in California Hospitals Introduction This issue brief will address the following Intensive care units (ICUs) have become an questions: increasingly important part of the American ■ How do the costs of critical care compare to health care delivery system. In 2000, critical ISSUE BRIEF the prevailing reimbursement amounts? care cost $55.5 billion—13.3 percent of ■ How does utilization of critical care in total hospital costs and half a percent of the California compare to the national rate? United States’ gross domestic product.1 As the population continues to age and the severity of ■ Is there an opportunity to reduce utilization hospital cases continues to increase, the role of of intensive care in California’s hospitals? the intensive care in hospital economics and in ■ What benefits could California’s hospitals patients’ lives is expected to become even more realize from such a reduction, either in terms important. of lower costs, eased demand for nurses, or lessened pressure for additional ICU beds? Now is an appropriate time to review the utiliza- tion of intensive care units (ICUs) in California, The data suggest that California hospitals may as many of the state’s hospitals have yet to begin have the opportunity to make more efficient construction to meet new seismic standards.2 use of an expensive resource. Improving the With experts projecting greater demand for utilization of ICU beds has the potential to save critical care beds, many hospitals may be consid- both money and labor, enhancing the financial ering adding more ICU capacity as part of their viability of critical care and freeing hundreds of construction plans. However, before launch- nurses for duties in other departments. ing such projects, it is important to consider whether there is an opportunity to improve the Hospital Reliance on Intensive Care utilization of existing critical care beds. Is Growing Over the past two decades, hospitals across the Compounding the problem, California is facing United States have successfully changed their a severe shortage of nurses over the next 25 role in many ways, shortening the average length years. Experts estimate that the state already of stay for most conditions and moving numer- lacks 14,000 nurses. Given that ICUs are heavy ous surgical procedures out of the hospital and users of nursing staff—commonly maintaining into outpatient settings. One consequence of a ratio of one nurse to two patients compared these cost-cutting strategies is that the popula- to one nurse to five patients on regular medical/ tion of patients at the average hospital is now surgical floors—reducing ICU utilization could made up of sicker people presenting more have a beneficial impact on the demand for complicated cases. nurses. M ARCH 2007 The composition of hospital beds reflects this trend Barring significant changes in practice, experts antici- as hospitals across the country and in California now pate that both the United States and California will devote a higher portion of their total beds to critical see further increases in critical care patient days as the care than they did 15 years ago.3 During this period, population ages. Patients aged 65 and older already the supply of non-critical care beds in the United consume over half of all critical care days, and the States shrunk by 31 percent, while the supply of percentage of the population over 65 is projected to critical care beds increased 26 percent. As a result, grow 10 percent from 2000 to 2020.5 The Advisory the nationwide portion of total beds devoted to Board Company estimates that ICU and coronary critical care grew from 7.8 percent to 13.4 percent.4 care unit (CCU) patient days will grow 10 percent In California, the portion of total beds devoted to from 2005 through 2015.6 critical care has grown from 12 percent in 1992 to 15.2 percent in 2005. ICUs as a Source of Financial Losses for Hospitals This expansion of critical care capacity has been An analysis of costs and reimbursements for Medicare closely matched by an increase in utilization. patients admitted to United States hospitals in 2000 Critical care patient days in California climbed found that, on average, hospitals lose money on nearly 9 percent from 1992 to 2005, resulting in the patients who spend at least one day in an ICU or growth of the portion of total patient days devoted a CCU, while they make money on patients who to critical care from 15.5 percent in 1992 to 17 do not spend any time in such units.7 Medicare percent in 2005 (Figure 1). pays, on average, 83 percent of costs for cases with Figure 1. Percent of Total General Acute Care an ICU stay compared to 105 percent of costs for Devoted to Critical Care, California Hospitals cases without an ICU or CCU stay. For the average 18% 1992 2005 17.02% Medicare discharge with an ICU stay, hospital costs 16% are $14,135, while reimbursement is only $11,704, 15.24% 15.50% resulting in a $2,431 loss on the average case. In 14% 2000, hospitals across the nation lost an aggregate 12.04% $5.8 billion on Medicare patients who spent at least 12% a day in the ICU, while they made $2 billion on patients who spent no time in the ICU or CCU.8 10% A more recent study confirmed the finding that 8% hospitals lose money on Medicare ICU cases and suggested that the economics may be getting worse as 6% the average Medicare reimbursement for admissions requiring an ICU stay has decreased, as it did from 4% 2002 to 2004. In contrast, Medicare reimbursement for admissions not requiring an ICU stay increased 2% over the same time period. (See Figure 2 for the average margin and reimbursement details.9) 0% Critical Care Beds Critical Care Days Source: American Hospital Association 2 | CALIFORNIA HEALTHCARE FOUNDATION That hospitals lose money on Medicare ICU cases is ICU Utilization in California of particular concern to California providers, given While California is often one of the states with the that 60 percent of California’s ICU discharges in lowest overall hospital utilization, recent research 2004 involved Medicare patients. An additional suggests that in terms of critical care at the end of 16 percent of ICU cases involved payers with the life, California physicians and hospitals are provid- lowest reimbursement rates: 11 percent of ICU cases ing more care than their counterparts in other states. are Medi-Cal patients and 5 percent are uninsured The Dartmouth Atlas of Health Care has collected patients.10 Even if the remaining cases, which are paid data on hospital utilization for Medicare patients who by private health plans, cover some of these losses died (for each of the years 1999 through 2003). The through cost shifting, it still appears there would be Atlas data set includes a number of metrics for inten- significant financial benefit to hospitals if they could sive care utilization during the last six months of life reduce ICU utilization. and during hospitalizations in which death occurred. For each of these measures, California exceeded the national average in 2003. In addition, for each of these measures, California ranks among the states with the highest utilization rates for the year 2003 (Table 1).11 Figure 2. Economics of Medicare Cases, 2002 vs. 2004 $14,000 $12,859 $12,833 2002 2004 $12,000 $10,000 $7,594 $8,000 $6,940 $6,000 $4,000 $2,000 $0 -$357 -$287 -$2,000 -$2,426 -$2,668 -$4,000 Average Reimbursement, Average Margin, Average Reimbursement, Average Margin, ICU Cases ICU Cases Non-ICU Cases Non-ICU Cases Source: Cooper, L. and Linde-Zwirble. “Medicare intensive care unit use: Analysis of incidence, cost, and payment.” Critical Care Medicine. 32(11): 2247-2253, November 2004. Rethinking the Use of Intensive Care Beds in California Hospitals | 3 Within California, there is considerable varia- CCU days per Medicare decedent within the last tion at both the regional and the hospital level in six months of life. Fifty-nine, or about 16 percent, terms of how much critical care is used at the end perform near the national average of 3.6 days for the of life. Looking at the number of ICU/CCU days years 1999 through 2003, using between three and per decedent during the last six months of life, 11 four days. And 90 (25 percent) of California hospitals of California’s 24 hospital referral regions (HRRs) provide less than three days of care per decedent. performed at or below the national average of 3.25 days in 2003, while the majority of regions used No Correlation between ICU Case Mix more days than the national average (Table 2). A and Utilization few regions stand out due to their high utilization: The California HealthCare Foundation commis- Los Angeles at 7.05 ICU/CCU days per decedent sioned Kurt Salmon Associates (KSA) to analyze and Orange County at 5.41 ICU/CCU days per the variance in ICU length-of-stay across California decedent.12 hospitals. Using 2004 ICU utilization data from 302 California hospitals, KSA compared ICU average As a group, California’s 358 hospitals exceed the length of stay (ALOS) against case mix index (CMI) national average, providing more than four ICU/ scores, which are used as a measure of severity of Table 1. Use of Intensive Care for Medicare Beneficiaries at the End of Life, 2003 U.S. Average California Average Percent Difference Percent of Medicare decedents admitted to ICU/CCU during the 17.8% 20.9% +17.4% hospitalization in which death occurred ICU/CCU days per decedent during the hospitalization in which 1.21 1.62 +33.8% death occurred ICU/CCU allowed charges per decedent during the hospitaliza- $2,759.15 $6,542.21 +137.1% tion in which death occurred Percent of Medicare decedents admitted to ICU/CCU at least 37.4% 43.8% +17.1% once during the last six months of life ICU/CCU days per decedent 3.25 4.52 +39.0% during the last six months of life Percent of decedents spending 7 or more days in ICU/CCU during 13.4% 18.7% +39.5% the last six months of life ICU/CCU charges per decedent $6,581.44 $15,869.94 +141.1% during the last six months of life Source: The Dartmouth Atlas of Health Care (www.dartmouthatlas.org) 4 | CALIFORNIA HEALTHCARE FOUNDATION Table 2. California Regional Variation, 2003 Hospital Referral Region ICU/CCU allowed charges ICU/CCU days per per decedent during the decedent during the last hospitalization in which six months of life (2003) death occurred (2003) Los Angeles 10,395.83 7.05 Orange County 5,208.58 5.41 Palm Springs/Rancho 6,248.33 5.20 Ventura 5,756.59 4.81 San Bernardino 5,068.15 4.61 San Diego 4,621.76 4.44 Bakersfield 4,647.54 4.29 Salinas 3,810.98 4.05 San Jose 8,629.81 3.91 San Mateo County 6,668.78 3.61 Modesto 6,741.29 3.50 Alameda County 6,927.90 3.46 Redding 4,308.62 3.26 Santa Cruz 4,473.75 3.25 San Luis Obispo 5,416.62 3.24 Fresno 3,644.36 3.20 Contra Costa County 7,461.74 3.14 San Francisco 7,588.68 3.07 Chico 6,284.33 3.05 Sacramento 4,568.02 2.91 Stockton 6,302.95 2.85 Santa Barbara 3,308.17 2.85 Napa 4,438.05 2.37 Santa Rosa 3,148.19 1.74 Source: The Dartmouth Atlas of Health Care (www.dartmouthatlas.org) Rethinking the Use of Intensive Care Beds in California Hospitals | 5 patient illness. Their analysis segmented hospitals illness and length of stay in the ICU, but the graphic into three groups based on the number of ICU beds: demonstrates the lack of a meaningful correlation in ■ 2 to 8 beds, typically critical access and other California.14 A significant correlation did not exist for small community hospitals; any of the three hospital groups based on size.15 It is particularly notable that 89 California hospitals— ■ 9 to 23 beds, typically mid- to large-sized nearly 30 percent of all the hospitals analyzed—have community hospitals; and excessive ICU length of stays relative to their ICU ■ 24 or more beds, typically large-sized community CMI scores. hospitals, regional referral centers, and academic medical centers. Reducing Variation in Utilization Would In addition, KSA excluded outlier hospitals, those Produce Significant ICU Savings facilities with CMI or ALOS values beyond two The variation in utilization of critical care units standard deviations from the mean in each bed size revealed by the Dartmouth Atlas data on end-of-life segment.13 care for Medicare beneficiaries, along with the KSA data on ICU utilization for all California patients, KSA’s analysis reveals a wide degree of variation in suggests a significant opportunity to reduce patient ICU ALOS among California’s hospitals that cannot days in critical care units. Looking first at the oppor- be explained by variation in severity of patient illness tunity presented by reducing ICU and CCU use (as measured by CMI). California’s average ICU during the last six months of Medicare patients’ length of stay for 2004 was four days, but individual lives, California hospitals could save 130,557 ICU hospital performance ranged from one day to over and CCU days if they could reduce their days per 13 days. decedent to the national average. Figure 3 plots the ICU ALOS and CMI for each of Likewise, if all of the hospitals with ICU ALOS in the 302 hospitals in the sample. One would expect excess of their CMI-predicted figure reduced their to see a fairly linear relationship between severity of LOS to conform with that value, California hospitals Figure 3: Linear Regression of CMI and ICU ALOS would reduce ICU utilization by 130,364 patient for CA Hospitals days. This would represent a reduction in average 14.0 ICU length of stay from 3.71 days to 3 days. < – 8 Beds ■ 12.0 Reduced ICU patient days should enable California ■ 9-23 Beds 10.0 > ▲ – 24 Beds hospitals to achieve significant financial benefits through: ALOS in ICU 8.0 ■ Improved utilization of existing ICU beds, allow- 6.0 ing a hospital to accommodate additional ICU volume without adding capacity; 4.0 ■ Better use of scarce nursing resources; and 2.0 R2 = 0.0 ■ Reduced costs. 0 Because the potential reduction in ICU days achieved 0.5 1.0 1.5 2.0 Hospital CMI using Dartmouth Atlas data and the KSA data is Source: Kurt Salmon Associates similar, this brief models the benefits of California 6 | CALIFORNIA HEALTHCARE FOUNDATION hospitals reducing their ICU patient days by free up nearly two million critical care nursing 130,000. This would represent a 9 percent reduc- hours. The net savings in annual nursing hours, tion in total ICU patient days, and nearly a half a after accounting for additional floor staffing, day reduction in ICU ALOS ( 3.5 days vs. 3.9 days). would be 1.4 million nursing hours.18 Viewed in Payoffs from implementing such a strategy could terms of nurses, the reduction in ICU patient days include: would save a modest but still significant number of critical care nurses: 828 in total. The net ■ Better use of existing ICU beds. Reducing ICU savings, after accounting for additional medical/ utilization by 9 percent yields a savings of 356 surgical unit nurses, would be 601 nurses.19 ICU beds, just over 6 percent of the total licensed ICU beds in California. If these beds were used as Quality-Driven Initiatives Produce ICU an alternative to building new ICU capacity, this Utilization Savings would save California hospitals $356 million in While many may worry that efforts to reduce ICU capital expenses.16 utilization will harm patients, evidence suggests that ■ ICU cost savings. Substituting medical/surgical reducing utilization and improving care quality are floor days for 130,000 ICU patient days would compatible. Often, patient stays in critical care units yield $159 million in operating cost savings for are extended due to adverse events such as a hospi- California hospitals.17 For Medicare patients and tal-acquired infection or medication error, many of other patients under case rate payments, the which are preventable. A recent observational study benefit of this reduction would accrue to the of patients in two critical care units in an academic hospitals, at least in the short run. medical center found that 20 percent of critical care ■ Better utilization of nursing resources. One of patients experienced an adverse event, 45 percent the potential benefits of reduced ICU days is the of which were considered preventable.20 Hospitals reduction in nurses needed in the ICU. Given directly addressing quality problems in critical care the nursing shortage in California, any efforts units often achieve significant reductions in length of to “create” additional nurse supply are valuable. stay and costs. Examples of successful quality initia- Based on the 2000 California average of 15.3 tives are presented below. registered nurse hours per patient day in the ICU, ■ Remote intensivist monitoring. In 2000, Sentara a reduction of 130,000 ICU patient days would Health System implemented eICU, a telemedicine Understanding the Paradox mix adjustment cannot completely account for In the face of significant financial benefit of reducing these differences. ICU utilization, why does so much variation persist ■ Hospitals with high occupancy rates may have among hospitals? There are many factors: more difficulty transferring patients to step- ■ Individual physician judgment differs regard- down units. ing patient acuity and need for higher levels of ■ Patient and family preferences plays a role. care. The existence of hospitalist and intensivist The availability of hospice and palliative care programs may help reduce this difference. can reduce ICU utilization if patients and family ■ Service mix contributes to differences, as members desire this type of care. hospitals with high surgical volumes and Additional research in these areas may help programs such as trauma, transplant, and burn to identify opportunities to use expensive ICU units will have higher ICU utilization rates. Case resources more efficiently. Rethinking the Use of Intensive Care Beds in California Hospitals | 7 program that provides remote intensivist monitor- Conclusion ing of hospital critical care units, to reduce ICU Given the greater role critical care is playing and will mortality and improve quality. The eICU helped continue to play in the health care system, hospi- Sentara reduce its ICU and hospital mortal- tal leaders and physicians should work together to ity rates by 26.7 and 26.4 percent, respectively. examine whether their utilization of critical care beds At the same time, the average length of stay is appropriate. Data suggest that a sizeable number decreased by 16 percent and variable costs per of California hospitals could reduce ICU patient ICU case declined by 24.6 percent.21 days. Doing so would create many benefits, including ■ Addressing ventilator-associated pneumonia. reducing hospital losses on ICU cases and “creating” VHA, Inc.’s “transformation of the ICU” initia- additional supply of nurses. Most importantly, it tive focused on reducing ventilator-associated would encourage more judicious use of the current pneumonia (VAP), which affects many ICU supply of ICU beds. Finally, reducing ICU utilization patients and leads to extended stays and high costs does not come at the expense of patient quality or in the ICU. Nineteen ICUs participating in the satisfaction, as preventable adverse events in the ICU first round of the initiative reduced their rates of lead to prolonged stays and overly intensive care at VAP 29 percent, from 7.5 to 5.3 cases per 1,000 the end of life. ventilator days. Average length of stay for these 19 ICUs dropped 15 percent, from 4.0 to 3.4 days. One participating hospital estimated that it saved ENDNOTES $700,000 annually from reduced LOS in its two 1. Halpern, N. et al. “Changes in Critical Care Beds and ICUs.22 Occupancy in the United States 1985-2000: Differences Attributable to Hospital Size. Critical Care Medicine. 34(8): In addition, efforts addressing end-of-life care can 2105-2111. produce reductions in ICU utilization. While this 2. “Summary of Requests for Extensions to Seismic Safety subject is sometimes controversial, the reality is that Deadlines.” California Office of Statewide Health Planning and Development. Updated 11/22/06. most people do not want to die in the hospital, 3. Within this issue brief. For this data, “critical care” refers to the let alone in an ICU. Approximately 90 percent of following units: intensive care units (ICUs), coronary care units Americans would prefer to die at home, but research- (CCUs), acute respiratory care units, neonatal intensive care units (NICUs), and burn units. ers at the University of Pittsburgh found that over 4 . Halpern, N. et al. (See note 1.) 20 percent of deaths occur after admission to an 5. The Critical Care Workforce: A Study of the Supply and ICU and 38 percent of deaths occur in the hospital.23 Demand for Critical Care Physicians. U.S. Department of Health and Human Services, Health Resources and Services Well-conceived efforts focused on reducing intensive Administration. (http://bhpr.hrsa.gov/healthworkforce/reports/ care at the end of life, such as palliative care programs criticalcare/default.htm) and do-not-resuscitate orders, should be able to both 6. Innovations Center Futures Database. The Advisory Board Company. 2006. reduce critical care utilization and improve patient 7. Cooper, L. and Linde-Zwirble. “Medicare intensive care unit and family satisfaction. use: Analysis of incidence, cost, and payment.” Critical Care Medicine. 32(11): 2247-2253, November 2004. 8. Ibid. 9. “ICU stays having a significant impact on hospital margins.” Healthcare Financial Management. July 2006. 10. California Office of Statewide Health Planning and Development. 2004 Hospital Utilization Data. 8 | CALIFORNIA HEALTHCARE FOUNDATION 11. The Dartmouth Atlas of Health Care. (www.dartmouthatlas. org.) AUTHORS 12. Ibid. Cyrus Yang and J. Charles Cosovich, strategists, 13. Outliers constituted 31 out of 333 hospitals, or 9.3 percent of Kurt Salmon Associates the original hospital population. 14. The regression yielded an R2 value of .01, which indicates that Jennifer Joynt, consultant CMI lacks predictive ability for ALOS in the ICU. In other words, a high CMI value does not mean the ICU will have a high ALOS. An R2 of 1.00 or -1.00 would indicate perfect predictive ability. 15. Regressions yielded an R2 of 0.01 for small hospitals, 0.02 for medium hospitals, and 0.02 for large hospitals. 16. This assumes a cost of one million dollars in capital expenses F O R M O R E I N F O R M AT I O N , C O N TA C T : for each new ICU bed built. Current estimates of the cost to build new beds in California range from one million to two California HealthCare Foundation million dollars. 476 Ninth Street 17. The cost savings is based on a daily cost of $2,462 per ICU Oakland, CA 94607 patient day compared to $1,236 per medical/surgical day (See tel: 510.238.1040 note 7). If hospitals were able to reduce ICU LOS without increasing medical/surgical LOS, the total savings would fax: 510.238.1388 increase to $320 million. www.chcf.org 18. Donaldson, N. et al. “Nurse Staffing in California Hospitals 1998-2000: Findings from the California Nursing Outcome Coalition Database Project.” Policy, Politics, & Nursing Practice. 2(1): 20-29. This article reports mean nursing hours per patient day of 16.7 for critical care units and 7.6 for medical/surgical units. 19. California Office of Statewide Health Planning and Development, 2005 Hospital Utilization Data. 20. Rothschild, J. et al. “The Critical Care Safety Study: The incidence and nature of adverse events and serious medical errors in intensive care.” Critical Care Medicine. 33(8): 1694- 1700. 21. Breslow, M. et al. “Effect of a multiple-site intensive care unit telemedicine program on clinical and economic outcomes: An alternative paradigm for intensivist staffing.” Critical Care Medicine. 32(1): 31-38. 22. McCarthy, D. and Blumenthal D. “Committed to Safety: Ten Case Studies on Reducing Harm to Patients.” The Commonwealth Fund. April 2006. 23. “End of Life ICU Utilization May Require Re-evaluation According to University of Pittsburgh-Led Study.” UPMC News Bureau. April 6, 2004. CALIFORNIA HEALTHCARE FOUNDATION | 476 Ninth Street, Oakland, CA 94607 | tel: 510.238.1040 | fax: 510.238.1388 | www.chcf.org