Key words: disposition; outcorne; prolonged m echanical ventilation; regional weaning center; survival; weaning

Post-ICU Mechanical Ventilation* Treatment of 1,123 Patients at a Regional Weaning Center David f. Scheinhorn, MD, FCCP; David C. Chao, MD, FCCP; Meg Stearn-Hassenpjlug, MS, RD; Laurie D. LaBree, MS; and Diane]. Heltsley, LCSW Study objectives: To update our database, reporting changes in the results of weaning attempts and profile of patients transferred to us after prolonged mechanical ventilation (PMV) in the ICU. Design: Retrospective record review, with prospective recording of physiologic measurements on admission from mid-1994. Setting: Regional weaning center (RWC). Patients: We studied 1,123 consecutive ventilator-dependent patients transferred for attempted weaning over an 8-year period. Measurements and results: Median (range) time of mechanical ventilation prior to transfer to the RWC declined from 37 (I to 249) days in 1988 to 29 (I to 120) days in 1996 (p<0.05). Acute physiology score of acute physiology and chronic health evaluation (APACHE) III was 32 (6 to I23) on RWC admission, equaling reported scores soon after ICU admission. Comparing other data on admission from 1988 to 1996, mean (±SO) serum albumin level declined from 2.92±0.58 to 2.43±0.50 gldl, and alveolar-arterial oxygen pressure difference widened from I06±50 to I39±99 mm Hg. Prevalence of stage II or worse pressure ulceration on admission increased from 34% in 1988 to 46% in I995. Despite these trends, there has been no significant change in patient outcome (55.9% weaned, 15.6% failed to wean, 28.8% died) or in median time to wean (29 [I to 226] days). Overall survival at 1 year after discharge for the 8-year period is 37.9%, improving from 29% in 1988-I99I to 45% since I992; survival in weaned patients discharged to home has improved from 45 to 59% during the respective time periods. Conclusions: Patients are being transferred from the ICU to our RWC for attempted weaning sooner in their course ofpmv. Although more severely ill on arrival than in past years, mortality is unchanged, more than half of the patients continue to be successfully weaned, and survival after RWC discharge is improved. (CHEST 1997; 111:1654-59) Key words: disposition; outcorne; prolonged m echanical ventilation; regional weaning center; survival; weaning Abbreviations: APACHE = acute physiology and chronic health evaluation; APS=acute physiology score; BRH = Barlow Respiratory Hospital; ECF=extended care facility; P(A-a)0 2 =alveolar-arterial oxygen pressure difference; PMV=prolonged mechanical ventilation; R\VC = regional weaning center; TVD=total ventilation days J n the 1990s, economic pressures to maximize resource utilization have resulted in the transfer of hemodynamically stable mechanically ventilated patients out of the I CU setting. Post-I CU care for these *From the Barlow Respiratory Hospital and Research Center (Drs. Scheinhorn and Chao, and Mss. Stearn-Hassenpflug and Heltsley), and the University of Southern California School of Medicine (Ms. LaBree), Los Angeles. Presented in Part at Weaning '96: Weaning from Prolonged Ventilation, April1996, Palm Springs, Calif; and at the Amelican Lung Association/ American Thoracic Society International Conference, May 12-16, 1996, New Orleans. Supported by grants from the Barlow Foundation, Harold R. And Winifred R. Stanton Foundation, and the Lluella Morey Murphey Foundation. Manuscript received S eptember 17, 1996; revision accepted November 14. Reprint requests: Dr. Scheinhorn, Barlou; Respiratory Hospital, 2000 Stadium Way, Los Angeles, CA 90026-2696 patients is now recognized as part of a continuum of care not only for the recipient patients, but also for critical care practitioners. In the Los Angeles area, Barlow Respiratory Hospital (BRH) functions as a regional weaning center (RWC), accepting such ventilator-dependent patients for attempted weaning. It is a private 49-bed, free-standing, not-forprofit, diagnosis-related group exempt, long-term acute care facility in a city park. In 1988, we began collecting data on this population, reporting on our first series of 421 patients in 1994.1 We have since accumulated information on more than 1,100 consecutive ventilator-dependent patients transferred to BRH over a period of 8 years. This report presents the data and identifies and discusses changes, trends, and implications for the future care of these patients. 1654 Clinical Investigations in Critical Care

MATERIALS AND METHODS Medical records of 917 ventilator-dependent patients admitted to BRH between May 1988 and June 1994 were retrospectively reviewed. As pa1i of ongoing protocols, data were gathered prospectively on 206 ventilator-dependent patients admitted from July 1994 through May 1996. Outcome data were scored in 1,037 patients admitted through December 1995. One-year survival data were gathered on 618 patients discharged alive through May 1995. Etiology and duration of ventilator dependency prior to transfer to the RWC were obtained from outside hospital records and from the transferring physician. In the last 206 patients admitted, we recorded the diagnosis that led to initial ICU en try, and from one to Hve diagnoses believed to contribute to ventilator dependence at the time of BRH admission. Demographic data, physiologic data on admission to BRH, time to wean, outcome of weaning attempts, and disposition at discharge were obtained from the patients' BRH medical records. Measurements were made to calculate patients' acute physiology scores (APS) of the acute physiology and chronic health evaluation (APACHE) III.2 Outcomes of weaning efforts were scored at BRH discharge. Successful weaning was defined as at least 1 week of complete ventilator independence. Patients who required partial support, such as nocturnal ventilation, were scored with those who failed to wean. Twenty-seven patients transferred to other acute-care facilities for treatment of intercurren t surgical illnesses prior to determination of outcome were not included in outcome analysis. Dedicated personnel in the BRH Social Services Department followed up patients postdischarge. Telephone contacts seven times during the first year were used to track patients' status. Discharge to a board and care facility was counted as discharged to home; discharge to subacute care was counted as discharged to an extended-care facility (ECF). Survival at 1 year after discharge was scored for each patient. If patients could not be contacted 1 year after discharge, they were included in the "Died" group for data analysis. Mean values are expressed ::'::SD; median values are reported with the range of values (range) when the distribution of data was not normal. Statistical tests used in year-to-year comparisons to trends included the following: Wilcoxon Rank Sum tests for comparisons of median times spent ventilator dependent prior to and at the RWC; Kruskall-Wallis nonparametric tests for differences in time to wean; multiple comparison t tests for changes in P(A-a)0 2 (alveolar-arterial oxygen pressure difference) and albumin; Fisher's Exact Test for comparing percentages in the pressure ulceration, weaning outcome, and disposition an a l y. ~ : e results; and x 2 test for comparing percentages of overall survival, and to seek correlation b etween stage of pressure ulceration and mortality. Correlation between serum albumin and P(A-a)0 2 values and time to wean were calculated using the Spearman correlation coefficient. RESULTS For the 8-year period, there were no significant changes in the mean patient age of 69 ± 13 years, the 57% female gender distribution, or the history of cigarette smoking in 67% of patients. Diagnoses that led to ICU admission were essentially unchanged from those reported in 1994.1 Most of the 206 patients most recently admitted fall into three diagnostic categories (vs the percent in the prior report): chronic lung disease, 27.7% (vs 24.5% ); acute lung disease, 29.1% (vs 31.8%); and postoperative, 22.5% (vs 23.5%). In the same subgroup of patients, the most frequent combinations of diagnoses at BRH admission follow (with percent weaning success): acute pneumonia vvith underlying COPD (n=28, 57% weaned); aspiration pneumonia with underlying neuromuscular disease (n = 18, 50% weaned); cardiac/thoracic surgery with underlying COPD (n = 15, 33% weaned); coronary artery bypass surgery with severe congestive heart failure (n = 14, 43% weaned); sepsis complicated by ARDS (n = 12, 50% weaned). Selected measurements on RWC admission are shown in Table 1, with data grouped to compare our earlier report on 421 patients from May 1988 to June 19911 with later and aggregate data. All variables show significant changes at the p< 0.05 level. APACHE III APS was calculated on admission for patients admitted since late 1994 (n =206). The median score was 32 (6 to 123); mean score was 34.0 ±19.2. Median age and mortality in this subgroup of patients were 72 (17 to 94) years and 30.6%, respectively. Median time of mechanical ventilation prior to transfer to the RWC was 37 (1 to 249) days in 1988 and 29 (1 to 120) days in 1996 (p< 0.05 ), with the decline during intervening years shown in Figure 1. Table 1 shows the comparison between the periods of time selected for discussion. The mean serum albumin and mean P(A-a)0 2 values on admission to BRH are shown in Figure 2. Serum albumin level on admission has declined from 2.92±0.58 g/dl in 1988 to 2.43±0.50 gldl in 1996. Significant differences between years are noted (p< 0.001 ), with lower albumin level for each year Table!-Comparison of Selected Measurements in 1,123 Ventilator-Dependent Patients Admitted to the RWC During Two Time Periods Spanning 8 Years Time Period 5188-6191 (n=421) 7191-5196 (n=702) 8 Years (n=l,l23) p Value Prior ventilator ti me, d 37 (0-325) 31 (0-395) 33 (0-395) < 0.001 Albumin, gldl 2.73:!::0.60 2.55:!::0.59 2.61:!::0.60 <0.001 P(A-a)0 2, mm J-Ig lj 1.5 :!::62.4 124.5:!::78.7 119.7:!::73.4 0.003 Pressure ulcers, % 32.9 41.8 36.1 0.03 CHEST I 111 I 6 I JUNE, 1997 1655

"' 45 1-e- Prior MV Duration I 5 0, - - - - - - - - - ~ = = = = = = = = = = ~ - - - - - - - - - -, 40 iii' 35 0 30 25-2 0~ - - J -- - - L- - - ~ - - ~ - - ~ L - - - ~ ~ 1B 1 ~ 1 ~ 1 ~ 1 ~ 1 ~ 1 ~ 1 ~ 1M Year FIGURE l. Trend in duration of mechanical ventilation (MV) prior to transfer to the RWC over an 8-year petiod. Median values (days) are shown. 1993 to 1996 than each year 1988 to 1992 at p<0.05 level, except for 1990. The P(A-a)0 2 has widened from 105.5±50.5 mm Hg in 1988 to 139.1±99.1 mm Hg in 1996 (p=0.02). The increase from 34% of patients admitted from 1988-1991 with stage II or worse pressure ulcers to 42% in 1991-1996 is statistically significant (p<0.001). For the 8-year period, mortality of patients admitted with stage II or worse pressure ulcers was 36%, compared to 20% for those with intact skin (p<o.ool). Median time to wean from mechanical ventilation at the RWC was 27 (3 to 207) days in 1988, which did not differ significantly from 32 (6 to 182) days in 1995. Year-to-year comparisons among all 8 years also failed to show significant differences in time to wean (p=0.29). In Table 2, outcomes of weaning attempts are grouped to compare more recent data with our earlier report. Weaning outcome is shown for the 1,037 patients admitted through December 1995, 3.2 3!+Albumin P(A-a)021-140,- - 130 ------ol --- - 'U ::J' - ~ :!2 Ol 9 2.8 120 0 c E 3 2.4 100 2.2 90 1988 1989 1990 1991 1992 1993 1994 1995 1996 Year FIGURE 2. Trend in levels of serum albumin and of P(A-a)0 9 measured on admission to the H.\VC over an 8-year/.eriod. Units are displayed on the ordinate scales as indicate. Values are means for the years shown. "' s and 1-year survival data are collected for the 618 patients discharged alive through May 1995. Outcome data for 1996 were not included since not all in the cohort have reached outcome. There are no significant changes in weaning outcome (p=0.13) or disposition (p=0.94) between the two time periods. Overall 1-year survival postdischarge for the 8-year period is 37.9%, increased significantly from 28.8 to 44.9% between the two time periods (p<0.001). Days ventilated at BRH for patients who weaned was 23 (2 to 207) days in 1-year survivors vs 36 (2 to 226) days in those who died during the year after discharge (p<0.001). Prior ventilated time was almost identical in the two groups, and adding it to BRH time, producing total ventilation days (TVD) did not change the statistical significance of the result (p=0.002). In patients discharged ventilator dependent, TVD did not correlate with 1-year survival. Transfer Out of the ICU DISCUSSION The choice of post-icu level of care is part of a plan of care that addresses the patient's clinical needs and cost of care. The cost of prolonged mechanical ventilation (PMV) is considerable. The Health Care Financing Administration has stated that mechanical ventilation is prolonged after 3 weeks of at least 6 h!d. 3 It has been estimated that patients who require 3 weeks of PMV use approximately 40% of the JCU's budget. 4 5 If this assumption is correct, the national cost of PMV is approximately $25 billion per year. 6 The transfer out of the ICU of the patient who requires PMV is driven by this high cost of traditional ventilator care in the ICU, and also by the need for ICU beds to treat incoming critically ill patients. When transferring the ventilator-dependent patient out of the ICU, the main clinical determinants in the choice of care level are the ability to safely treat ongoing medical and surgical problems, and whether weaning efforts will continue. More severely ill patients would be expected to be directed to the acute-care level. With this and cost operant, we expected to find trends toward earlier transfer of patients to our R\VC. These expectations were borne out in our data review. Factors independent of acuity of illness, such as patients' age, gender, smoking history, and diagnosis necessitating ICU admission are unchanged from our earlier report. Our more recent method of assigning diagnoses responsible for continued ventilator dependence has confirmed our impression that patients suffering PMV have multiple contributory 1656 Clinical Investigations in Critical Care

Table 2-0utcome, Discharge Disposition, and 1-Year Survival in 1,037 Ventilator-Dependent Patients Admitted to BRR Over 8 Years, With Comparison of Two Time Periods 8 Years Time Period n % 5/88-6/91, % 7/91-12/95, % Weaned 580* 559 53.5 57.5 Discharged home 268 47.1 47.2 47.0 Alive at 1 yr 1 130 52.8 44.6 58.6 Discharged ECF 301 52.9 52.8 53.0 Alive at 1 yr 88 31.8 22.1 38.4 Ventilator-dependent 159 15.3 18.1 13.5 Discharged horne 43 27.0 3l.l 23.5 Alive at 1 yr 18 46.2 39.1 56.3 Discharged ECF 116 73.0 68.9 76.5 Alive at l )T 17 16.2 7.8 24.1 Died 298 28.8 28.4 29.0 *Eleven patients transferred to other acute-care hospitals after weaning are excluded from analysis of disposition and 1-year survival. 1 0 ne-year smvival data shown are for the 618 patients discharged alive through May 1995. medical and postsurgical problems. Larger numbers of patients in each combined-diagnosis subgroup would allow us to translate the subgroup weaning outcomes into reasonable expectations of weaning success. This may prove useful in identifying which patients are best transferred to the RWC level for continued weaning attempts in the future. A clear trend to earlier transfer out of the ICU is demonstrated (Fig 1) as the median number of days of prior ventilated time dropped from 37 days in 1988 to 29 days in 1996, approaching the Health Care Financing Administration "definition" of PMV. The leveling off of earlier transfer time in recent years may represent a limitation imposed by the severity of the patient's illness or an equilib1ium of economic pressures-that of the hospital to move the patient out of the ICU and that of the physician wanting to continue to attend the patient. Earlier transfers in general could result from faster recovery from the catastrophic illness that necessitated ICU treatment or transfer while the patient was more ill; the latter would have clinical importance to the facility receiving the patient in transfer. Measures of Severity of Illness Since mid-1994, we scored severity of illness on admission to the RWC, ie, on discharge from the transferring ICU, using the APACHE III APS, a measure of "physiological instability and risk of death." 2 We found a mean score of 34 and median score of 32 in our patients, with two benchmarks for comparison: (1) the recently reported mean APS of 56.7 in almost 6,000 patients requiring mechanical ventilation on the day of admission to 42 different ICU's,7 and (2) the median score of 32 in >4,800 patients 48 hours after entry into the ICUs of five teaching hospitals, also recently reported. 8 The APS of patients entering the ICU ventilated on day 1 would be expected to be higher than that of our patients, whose APSis measured on the day of ICU discharge. The second benchmark indicates that patients transferred to us are still as ill as patients commonly granted ICU entry. Similar in-hospital mortality figures for RWC patients (30.6%) and the latter benchmark group (25.5% ), with an RWC cohort of greater age, further validates this observation. Clearly, our admission APS strongly suggests that at the time of transfer to us, these patients cannot be safely treated at the subacute, nursing home, or home care levels. Lower serum albumin and higher P(A-a)0 2 values in more recent admissions (Table 1) are consistent with transfer out of the ICU closer to the acute events resulting in respiratory failure. If the higher P(A-a)0 2 means that patients come to us earlier in the natural history of their acute pulmonary illness, and the lower serum albumin level reflects physiologic response to stress, 9 one would expect prolongation of the time it takes to wean patients at the RWC. P(A-a)0 2 did not correlate with time to wean (p=0.32), but serum albumin level did (r= -0.21, p<0.001). The yearly mean values (Fig 2) show little evidence of leveling off. Our observed trend toward longer time to wean may become significant if the transfer of patients with increasing physiologic derangement continues. Pressure ulceration develops in patients at high risk in the ICU. A recent Canadian study reported a 51% incidence of pressure ulceration in patients treated on standard ICU bed surfaces.l 0 In that study, significant predictors of pressure ulcer development included length of stay, APACHE II score, and serum albumin level-variables similar to those we found to correlate with prevalence of pressure ulceration on admission to CHEST I 111 I 6 I JUNE, 1997 1657

BRH. The high prevalence of patients with pressure ulcers and the severity of ulceration on admission to BRH (Table 1) is therefore expected. As the correlation between low serum albumin level, pressure ulceration, and acute hospital mortality is well established, 9 it is also not surprising that 36% of patients admitted to BRH with stage II or worse pressure ulcers died, in contrast to 20% mortality for those with intact skin (p<0.001). Outcome of Weaning Attempts The reduction in prior ventilated time (36.5 days in 1988 vs 29 days in 1995, p=0.005) is largely balanced by the trend, though not statistically significant, toward longer median time to wean (27 days in 1988 vs 32 days in 1995). A prolongation of weaning time at the RWC proportional to the decrease in ICU ventilated days would be expected if, as has been reported from the ICU setting, the natural history of the catastrophic illness is the primary determinant of the time that mechanical ventilation will be required. 7 Therefore, we also looked at TVD, ICU plus RWC days, finding a trend downward (80 days in 1988 vs 66 days in 1995), that while not statistically significant (p=0.39), might be most interesting from a cost standpoint. If use of specialized weaning teams hastens weaning in the ICU setting, 11. 12 transfer to the specialized RWC may account for this downward trend in TVD. Furthermore, we found a decrease in the number of patients with eventually successful, but very long weaning attempts. In 1988 to 1991, 75% of patients weaned by 55 days, while for 1991 to 1995, the 75% quartile dropped to 50.5 days. We attribute this to heightened utilization review activity, driven by pressure to be competitive in the marketplace. There is no significant change in outcome of weaning attempts (Table 2) at the RWC. In the face of several indicators that patients are "sicker" on admission to the RWC in recent years, as shown above, we believe that we are serving this population well. Reports of weaning success from units to which patients are transferred after PMV in the ICU range from 32%13 to 87%; 14 survival to discharge of 97% 15 has been reported. Comparisons are difficult due to differences in reporting methods. For example, our 56% weaning success rate for the 8-year period increases to 80% using the method of Gracey et al, 14 in which only patients surviving to discharge are scored. We think our results will prove similar to those at other free-standing long-term acute care units, with comparable intake criteria and reporting methods, when more of those data are published. Disposition and Survival Scoring of disposition at discharge (Table 2) and 1-year survival postdischarge deserve comment. Very few patients go to board and care, at which high levels of independence and activity are required, so counting them as discharged to "home" seems reasonable. However, in recent years, most patients who failed weaning, and many weaned patients with tracheostomy, go to the subacute level of care, a change since our last report. Our current records track discharges to subacute care, but because we do not have that information for the whole database, we score subacute discharges here as ECF discharges. The percent of patients who were weaned and returned home vs that of patients discharged to institutional care has remained the same since our 1994 report. The expanded menu of lower levels of care in recent years may explain the almost 8% decrease in patients being discharged home ventilator-dependent, with a reciprocal increase in patients discharged to ECF. Overall, 1-year survival has increased significantly since 1991. The increase in 1-year survival at the ECF levels of care probably reflects the growth of subacute care and the general upgrading of ability to care for these patients in our community since our first report. As expected, patients with adequate functional status and caregiver support to return to a home environment more often survived to 1 year than those requiring institutional extended care. The increase in survival at home from 44.6% to 58.6% after weaning is probably multifactorial. Attention to postweaning issues, including emphasis on patient, family, and caregiver education during rehabilitation, may play an important role. That almost half of our patients who weaned were discharged to their homes after PMV is one of the things that keeps our teams "going." When a patient who was very ill and debilitated on admission is able to return home after treatment, it is satisfying to the team as well as the patient and family. In future follow-up efforts, it will be important to note any change in a patient's disposition between discharge and 1-year postdischarge. Patients initially discharged to ECF may have gone home eventually, and some patients discharged home may have ultimately required institutional care. Comparative survival data for similar patients is scarce. In 250 patients mechanically ventilated for at least 10 days, a 28% 1-year survival has been repmted;16 a 76% 1-year survival was reported in 119 patients discharged from a ventilator-dependent unit. 14 Although others have reported no correlation between duration of PMV and long-term survival in a small number of patients, 13 we found a highly 1658 Clinical Investigations in Critical Care

significant one that could not be predicted from days ventilated in the ICU. For patients who weaned, days ventilated at the RWC were more than 50% greater in those who did not survive 1 year than in patients alive 1 year postdischarge. An area of future study will be to identify those variables responsible for increased time to wean in these patients and to test any appropriate clinical interventions that might influence long-term survival by decreasing time to wean at the RWC. CONCLUSIONS PMV is a final common pathway on which patients find themselves after a variety of insults that lead to respiratory failure. A dedicated subspecialty hospital, functioning as an RWC, can wean a substantial number of patients from PMV and return them to their homes, where almost 60% survive at least 1 year. The trend is toward admitting sicker patients who are transferred out of the ICU sooner, with more physiologic impairment and debility on arrival at the RWC. We therefore expect that achieving or maintaining our current weaning and disposition outcomes \:vill become more difficult. REFERENCES 1 Scheinhorn DJ, Artinian B, Catlin J. Weaning fi om prolonged mechanical ventilation: the experience at a regional weaning center. Chest 1994; 105:534-39 2 Knaus WA, Wagner DP, Draper EA, et al. The APACHE III prognostic system. Chest 1991; 100:1619-36 3 Cohen IL, Booth FVM. Cost containment and mechanical ventilation in the United States. New Horizons 1994; 2:283-90 4 Daly RJ, Rudy EB, Thompson KS, et al. Development of a special care unit for chronically critically ill patients. Heart Lung 1991; 20:45-52 5 Halpern NA, Bettes L, Greenstein R. Federal and nationwide intensive care units and healthcare costs: 1986-1992. Crit Care Med 1994; 22:2001-07 6 Levit K, Sensenig AL, Cowan CA, et al. National health expenditures, 1993. Health Care Financing Review, Fall 1994; 16:280-81 7 Seneff MG, Zimmerman JE, Knaus WA, et al. Predicting the duration of mechanical ventilation: the importance of disease and patient characteristics. Chest 1996; 110:469-79 8 SUPPORT Principal Investigators. A controlled trial to improve care for seriously ill hospitalized patients. JAMA 1995; 274:1591-98 9 Doweiko JP, Nompleggi DJ. The role of albumin in human physiology and pathophysiology: III. Albumin and disease states. JPEN 1991; 15:476-83 10 Inman KJ, Sibbald WJ, Rutledge FS, et al. Clinical utility and cost effectiveness of an air suspension bed in the prevention of pressure ulcers. JAMA 1993; 269:1139-43 11 Cohen IL, Bari N, Strosberg MA, e t al. Reduction of duration and cost of mechanical ventilation in an intensive care unit by use of a ventilatory management team. Crit Care Med 1991; 19:1278-83 12 Kupfer YY, Seneviratne C, Bauch J, et a!. The rapid wean team: a cost effective method for management of mechanically ventilated patients. Chest 1995; 108:101S 13 Elpern EH, Larson R, Douglass P, et al. Long-term outcomes for elderly survivors of prolonged ventilator assistance. Chest 1989; 96:1120-24 14 Gracey DR, Naessens JM, Viggiano RW, et al. Outcomes of patients cared for in a ventilator-dependent unit in a general hospital. Chest 1995; 107:494-99 15 Criner GJ. Reining in RICU costs. Adv Respir 1995; 4:57-60 16 Indihar FJ. A 10-year r eport of patients in a prolonged respiratory care unit. Minn Med 1991; 74:23-7 CHEST /111 /6/ JUNE, 1997 1659