Key words: cost; mechanical ventilation; mortality; respiratory failure; survival; tracheostomy

Transcription

1 Hospital and Long-term Outcome After Tracheostomy for Respiratory Failure* Milo Engoren, MD, FCCP; Cynthia Arslanian-Engoren, PhD, RN; and Nancy Fenn-Buderer, MS Objective: To determine the patient characteristics, hospital course, hospital cost, posthospital survival, and functional outcome in a group of patients with tracheostomy for respiratory failure. Design: Retrospective chart review combined with prospective evaluation of functional status. Setting: An urban, tertiary-care medical center. Patients: Adult patients with tracheostomy for respiratory failure between January 1, 1998, and December 31, Methods: Retrospective chart review and prospective administration of the Short Form-36 (SF-36) for health status outcome. Results: Four hundred twenty-nine patients were studied. Hospital mortality was 19%. Only 57% of survivors were liberated from mechanical ventilation. At 100 days, 6 months, 1 year, and 2 years after discharge, 24%, 30%, 36%, and 42% of hospital survivors had died, respectively. Patients liberated from mechanical ventilation and having their tracheostomy tubes decannulated had the lowest mortality (8% at 1 year); the mortality of ventilator-dependent patients was highest (57%). Sixty-six patients completed the SF-36 for functional status. While emotional health was generally good, physical function was quite limited. Median hospital direct variable cost was $29,340. Conclusion: Overall survival and functional status are poor in patients with tracheostomy for respiratory failure. Patients who are liberated from mechanical ventilation and have their tracheostomy tubes removed have the best survival; however, it comes at a higher hospital cost and longer length of stay. (CHEST 2004; 125: ) Key words: cost; mechanical ventilation; mortality; respiratory failure; survival; tracheostomy Abbreviations: CI confidence intervals; DRG diagnosis-related group; SF-36 Short Form-36 Tracheostomies may be performed to ensure a safe and patent airway in patients; however, most are performed to facilitate mechanical ventilation for respiratory failure. 1 While there are complications, including death, associated with tracheostomy, its performance has become common. 2,3 The development of bedside percutaneous tracheostomy has *From the Departments of Anesthesiology and Internal Medicine (Dr. Engoren) and Research Oversight and Education (Ms. Fenn-Buderer), St. Vincent Mercy Medical Center, Toledo, OH; and School of Nursing (Dr. Arslanian-Engoren), University of Michigan Ann Arbor, MI. This study was presented in part at the 32nd Critical Care Conference, January 28 to February 3, 2003, San Antonio, TX. Funding was supplied solely by institutional and departmental resources. Manuscript received February 11, 2003; revision accepted June 24, Reproduction of this article is prohibited without written permission from the American College of Chest Physicians ( permissions@chestnet.org). Correspondence to: Milo Engoren, MD, FCCP, Department of Anesthesiology, St. Vincent Mercy Medical Center, 2213 Cherry St, Toledo, OH 43608; engoren@pol.net obviated the need for a trip to the operating room (with the attendant risks of transporting a critically ill patient), has lowered costs, and may be one of the For editorial comment see page 7 factors driving the increased incidence of tracheostomy. 4 One tenth of patients receiving mechanical ventilation undergo tracheostomy. 4,5 The early use of tracheostomy has been advocated to promote more rapid liberation from mechanical ventilation and lower hospital costs. 6 Nevertheless, patients who require tracheostomy for respiratory failure, diagnosis-related group (DRG) 483, have a high consumption of resources. By DRG categories, they have the highest patient costs and the highest hospital reimbursement. Studies that evaluated survival in ICU patients admitted with respiratory failure have found a high hospital mortality rate, approximately 40%. 7,8 However, DRG 483 patients are a subset of these patients 220 Clinical Investigations in Critical Care

2 Table 1 Patient Characteristics: Nonsurvivors vs Survivors* Subject Characteristics Nonsurvivors, Dead or Hospice (n 95) Survivors (n 334) p Value Male gender 48 (51) 198 (59) 0.13 Marital status Married 53 (56) 144 (43) Single 22 (23) 125 (37) 0.04 Divorced 4 (4) 22 (7) Widowed 16 (17) 43 (13) Race White 73 (78) 254 (77) Black 19 (20) 58 (17) 0.3 Other 2 (2) 20 (6) Insurance Medicare 71 (76) 155 (47) Medicaid 9 (10) 23 (7) Medicaid pending 1 (1) 13 (4) Commercial 7 (7) 71 (22) HMO/PPO 3 (3) 32 (10) Other 0 (0) 7 (2) None 2 (2) 30 (9) Admitting service Medicine 75 (79) 157 (47) Surgery 19 (20) 159 (48) Family practice 1 (1) 18 (5) Admitting category MI or angina 17 (18) 25 (8) CVA 2 (2) 34 (10) Arrest 1 (1) 5 (2) Other medical 55 (58) 133 (40) Elective surgery 10 (11) 19 (6) Emergency surgery 7 (7) 7 (2) Trauma 3 (3) 110 (33) Head injury 1 (1) 68 (20) Comorbidities Hypertension 65 (68) 161 (48) Coronary artery disease 46 (48) 110 (33) 0.01 Peripheral vascular disease 11 (12) 23 (7) 0.14 MI (not on admission) 28 (29) 58 (17) 0.01 Stroke 14 (15) 40 (12) 0.5 COPD 24 (25) 84 (25) 1.0 Diabetes mellitus 36 (38) 90 (27) 0.04 Cancer 14 (15) 33 (10) 0.18 Renal dysfunction 43 (45) 82 (25) Transfusion 89 (94) 246 (74) New renal dysfunction 63 (66) 65 (19) Dialysis New 38 (40) 26 (8) No 54 (57) 298 (89) Chronic 3 (3) 10 (3) Percutaneous gastrostomy tube 36 (38) 143 (43) 0.7 Open gastrostomy or jejunostomy tubes Gastrostomy only 13 (14) 57 (17) Jejunostomy only 2 (2) 20 (6) 0.05 Both gastrostomy and jejunostomy 8 (9) 53 (16) Neither gastrostomy nor jejunostomy 71 (76) 203 (61) Any permanent feeding tube 58 (62) 266 (80) Surgeries Craniotomy 1 (1) 32 (10) 0.01 Thoracic 24 (25) 50 (15) 0.02 Abdominal 24 (25) 51 (15) 0.02 Major vascular 9 (9) 16 (5) 0.08 Major orthopedic 2 (2) 31 (9) CHEST / 125 / 1/ JANUARY,

3 Table 1 Continued Subject Characteristics Nonsurvivors, Dead or Hospice (n 95) Survivors (n 334) p Value Surgeries (continued) Other 13 (14) 51 (15) 0.70 Any of above 59 (62) 191 (57) 0.4 Inferior vena caval filter 7% 59 (18) 0.01 Age, yr 95 (71; 20 88) 333 (64; 18 98) Height, cm 90 (168; ) 307 (170; ) 0.01 Weight, kg 94 (74; ) 330 (77; ) 0.4 Admission hemoglobin, g/dl 95 (11.4; ) 334 (12.6; ) Nadir hemoglobin, g/dl 95 (7.6; ) 332 (8.1; ) Admission creatinine, mg/dl 95 (1.3; ) 334 (1.0; ) Peak creatinine, mg/dl 95 (3.4; ) 334 (1.3; ) *Data are presented as No. (%) or No. (median; range) unless otherwise indicated. HMO/PPO health maintenance organization/preferred provider organization; MI myocardial infarction; CVA cerebrovascular accident. who may have a better hospital outcome. By physician and family choice, they may exclude the sickest patients: those who die within several days of respiratory failure, and those with multisystem organ dysfunction who are not expected to survive. Yet, little is known about this group of patients. The few studies 4,9 11 evaluating outcome have been limited by small patient populations, specialized populations, limited follow-up, and no cost information. The purposes of this study were as follows: (1) to describe patient characteristics, hospital course, posthospital discharge survival, functional outcome, and hospital cost in a group of DRG 483 patients; and (2) to determine those patient characteristics that are associated with survival and functional outcome. Materials and Methods This study was approved by the institutional review board. Oral consent was obtained from all participants for the telephone interview portion of the study. The computerized database of the hospital was queried for all patients who underwent tracheostomy or were discharged in DRG 483 (tracheostomy for respiratory failure) between January 1, 1998, and December 31, Most tracheostomies are performed as surgical tracheostomies under general anesthesia in the operating room or the cardiovascular ICU. The remainder were performed percutaneously in the ICU under local anesthesia. Four hundred twenty-nine patients were included in the study. Another 80 patients who had undergone tracheostomies were excluded: charts were unavailable (n 8); tracheostomy was performed as part of a planned surgical procedure (n 12), for emergency airway control (n 8), in pediatric patients 18 years old (n 23), for secretion management (n 15), or for upper airway obstruction in an intubated patient (n 11); or if the patient already had a tracheostomy on admission to the hospital (n 3). Charts were reviewed for age; sex; marital status (married, single, widowed, divorced); race (white, black, other); insurance status (none, private commercial, Medicare, Medicaid, Medicaid pending on admission, health maintenance organization/preferred provider organization); height; weight; hospital admitting service (internal medicine, family practice, surgery); admission diagnosis; presence of closed head injury on admission; medical history of hypertension, coronary artery disease, peripheral vascular disease, myocardial infarction, stroke, COPD, diabetes mellitus, and cancer; admission and nadir (lowest) hemoglobin; admission and peak creatinine; acute renal failure (peak creatinine 2 mg/dl if normal on admission, else increased by 1 mg/dl if elevated on admission) 9 ; transfusion; dialysis (none, acute, chronic); insertion of an inferior vena cava filter; insertion of a surgical feeding tube (open gastrostomy, open jejunostomy, both open gastrostomy and jejunostomy, percutaneous endoscopic gastrostomy, or none); type of operation (craniotomy, major vascular [operations on the abdominal aorta or carotid, iliac, or femoral arteries]; thoracic; abdominal; major orthopedic [operations on the pelvis, femur, or tibia, or hip or knee replacement]; other [all remaining operations]); dates of admission, tracheostomy, last day of mechanical ventilation, and discharge; discharge location (home, extended-care facility, rehabilitation hospital, another acute care hospital, hospice facility, or death); respiratory status on discharge (partial mechanical ventilation [on tracheostomy collar for at least 6 h/d], mechanical ventilation dependent [receiving at least 18 h/d], liberated from mechanical ventilation but tracheostomy tube still present, liberated from mechanical ventilation and tracheostomy tube decannulated); the ability to walk in the hall; and nutrition (total parental nutrition, tube feedings, neither). Death after discharge was determined from hospital records, telephone calls, and Social Security Death Index. Survivors were contacted by telephone, and if they consented to participate, were administered the Short Form-36 (SF-36). 12 Direct variable cost for each patient s care was obtained from the internal accounting system of the hospital. Data were collected at an urban, tertiary-care, universityaffiliated, level I trauma and referral medical center. Fifty of the licensed 588 beds are adult critical care beds. The adult ICUs are open units, but ventilator privileges are restricted to anesthesiologists, pulmonologists, and trauma surgeons with critical care certification. Statistics Univariate associations between categorical patient characteristics and binary outcomes were analyzed with either 2 statistics or Fisher exact tests depending on applicability (SAS Institute; Cary, NC). Data are presented as odds ratios and 95% confidence intervals (CI). Continuous patient characteristics were analyzed univariately with either the unpaired Student t test or the 222 Clinical Investigations in Critical Care

4 Table 2 Multivariate Predictors of Hospital Mortality Effect Odds Ratio 95% CI p Value Medical admit* No gastrostomy or jejunostomy feeding tube New renal dysfunction Age Nadir hemoglobin *Admission by any medical service. nonparametric Mann-Whitney rank-sum test depending on normality. Two-tailed p values 0.01 were used to indicate statistical significance. Multivariate associations between patient characteristics and binary outcomes were examined by logistic regression; odds ratios and 95% CI are presented for significant effects in the multivariate models. 13 Cox proportional hazards methodology was used to model the probability of survival as a function of time, looking for differences in survival associated with various patient characteristics. Risk ratios (also referred to as hazards ratios) and 95% CI are presented for significant effects in the multivariate models. 14 Multivariate modeling was done using forward selection and confirmed by backwards selection. No variable was forced to remain in the models; p values 0.05 were used to enter and keep variables in the models. Results Four hundred twenty-nine of the 5,142 adult patients (8.3%) who required mechanical ventilation received tracheostomy for respiratory failure (DRG 483). Eighty-two of these 429 patients (19%) died in the hospital. Another 13 patients were transferred to hospice. All hospice patients died within 9 days and were included for analysis purposes with in-hospital deaths making total hospital mortality 22%. Patient characteristics of all patients are shown in Table 1. By multivariate analysis, higher age, lower nadir hemoglobin, admission by a medical service, no surgically inserted feeding tube, and most strongly new renal dysfunction were significantly associated with hospital death (Table 2). When analyzed without processes of care, medical service admission (odds ratio, 2.3; 95% CI, 1.3 to 4.3; p 0.006), lower nadir hemoglobin (odds ratio, 5.1; 95% CI, 3.0 to 9.0; p 0.001), and new renal dysfunction (odds ratio, 1.5; 95% CI, 1.2 to 1.8; p 0.001) remained significantly associated with hospital death. Despite similar lengths of stay, nonsurvivors had higher direct variable costs ($35,078 vs $28,202, p 0.001) than survivors (Table 3). Of the 334 hospital survivors, 118 patients (35%) were discharged completely ventilator dependent, and 25 patients (7%) were partially ventilator dependent. One hundred ninety-one patients (57%) were completely liberated from mechanical ventilation, although only 57 patients (30%) had their tracheostomy tubes removed before discharge. By multivariate analysis, younger age, ability to eat without tube feeds or total parenteral nutrition, ability to walk in Table 3 Direct Variable Cost and Length of Stay* Variables Nonsurvivors (n 95) Ventilator Dependent (n 143) Liberated With Tracheostomy Tube (n 134) Liberated and Decannulated (n 57) Total cost (direct variable cost), $1, (50 322), 24 (2 67),#, ** 30 (5 184),#, 39 (36 288), ** No. of days from Admission to tracheostomy 17 (3 51),, ** 14 (2 54) 13 (0 53) ** 13 (1 47) Admission to death or liberation 30 (12 134), ** 23 (7 102) 25 (3 64) ** Admission to death or discharge 30 (12 134), 28 (6 162),, ** 32 (14 130), 37 (14 155),, ** Tracheostomy to death or liberation 12 (0 114), 16 (1 93), 24 (8 138), Liberation to discharge 8 (1 44) 12 (5 118) *Data are presented as median (range). Nonsurvivors patients who died in hospital; Ventilator Dependent patients who were discharged still requiring mechanical ventilation; Liberated With Tracheostomy Tube patients liberated from mechanical ventilation, but tracheostomy tube was still present on hospital discharge; Liberated and Decannulated patients liberated from mechanical ventilation and whose tracheostomy tube had been removed before discharge. All pairwise comparisons p 0.05 except where indicated. No. of days from admission to death in the nonsurvivor group and from admission to liberation in the liberated groups. No. of days from admission to death in the nonsurvivor group and from admission to discharge in the other three groups. No. of days from tracheostomy to death in the nonsurvivor group and from tracheostomy to liberation in the liberated groups. p 0.001, pairwise comparison. p 0.01, pairwise comparison. #p 0.01, pairwise comparison. **p 0.001, pairwise comparison. p 0.05, pairwise comparison. p 0.05, pairwise comparison. p 0.05, pairwise comparison. p 0.05, pairwise comparison. CHEST / 125 / 1/ JANUARY,

5 the hall, having any operation, and being admitted for other than medical reasons (excluding angina and myocardial infarction) were associated with liberation from mechanical ventilation (Table 4). Patients who were liberated from mechanical ventilation had longer hospitalizations than those patients discharged receiving mechanical ventilation (34 days vs 28 days, p 0.001) and 35% higher hospital costs ($32,221 vs $23,802, p 0.04) [Table 3]. One hundred thirty-four of the 191 patients liberated from mechanical ventilation still had a tracheostomy tube present at discharge. The multivariate analysis showed that ability to walk and eat and absence of hypertension were significantly associated with tracheal decannulation. Both groups of patients had similar length of time from hospital admission to liberation from mechanical ventilation (25 days vs 23 days, p 0.05). However, the patients who were decannulated had a longer total length of stay (37 days vs 32 days, p 0.02) and a longer length of stay after liberation from mechanical ventilation (12 days vs 8 days, p 0.001). They also incurred higher costs ($39,137 vs $30,432, p 0.04) [Table 3]. Overall survival was poor: 24% of hospital survivors were dead by 100 days, 30% by 6 months, 36% by 1 year, and 42% by 2 years after discharge. However, Kaplan-Meier survival varied dramatically by ventilator and tracheostomy tube status on discharge. Only 40 5% ( SE) of ventilator-dependent patients were alive 1 year after discharge, compared to 86 5% of decannulated patients (p 0.001; Fig 1). Being discharged ventilator dependent or even with a tracheostomy tube if liberated from mechanical ventilation was associated with a twofold to fourfold higher chance of death. Coronary artery disease and higher admission creatinine also increased the risk of death, while closed head injury and being married lowered the risk of dying (Table 5). Sixty-six patients completed the SF-36. Another 38 patients refused or were unable per the caregivers. The remaining 67 survivors could not be reached. Table 4 Multivariate Predictors of Liberation From Mechanical Ventilation in the 334 Survivors on Discharge* Effects Odds Ratio 95% CI p Value Younger age, yr Nonmedical admission Any operation Walking in hall Eating without tube feeds or TPN *TPN total parenteral nutrition. Figure 1. Kaplan-Meier survival plot (in days) of the four groups of hospital survivors. Kaplan-Meier survival of ventilator-dependent patients (dashed and dotted line), partially dependent patients (dotted line), patients liberated from mechanical ventilation but tracheostomy tube is still present (dashed line), and patients liberated and tracheostomy tube removed (straight line). Survival was best in the liberated and tracheostomy tuberemoved group (p vs patients liberated from mechanical ventilation but tracheostomy tube is still present; p vs ventilator-dependent patients and vs partially dependent patients). Survival was intermediate in the patients liberated from mechanical ventilation but tracheostomy tube is still present (p compared to ventilator-dependent patients.) Survival did not differ between ventilator-dependent patients and partially dependent patients, or between partially dependent patients and patients liberated from mechanical ventilation but tracheostomy tube still present. Decannulated patients had better social functioning than patients discharged partially (n 3) or totally (n 16) ventilator dependent. Otherwise, the three groups were similar. Most responders had good emotional health, but remained with major physical limitations (Table 6). Discussion Hospital survival in patients who underwent tracheostomy for respiratory failure (DRG 483) was 78%. Survival was associated with both factors present on hospital admission and factors developed in the hospital, particularly renal dysfunction. The survival rate was similar to a small study 4 of 51 patients that found 86% hospital survival, but better than that found in a Medicaid population (49%). 11 In contrast, survival in our Medicaid population was 72%, which may be related to improved medical care over the past decade or to differences in practice patterns that decide who receives a tracheostomy. Our risk factors agree with previous studies 15,16 that 224 Clinical Investigations in Critical Care

6 Table 5 Predictors of Postdischarge Death in the 334 Survivors by Cox Hazard Models Effects Relative Risk 95% CI p Value Ventilator status (compared to liberated from ventilator and tracheostomy tube decannulated) Ventilator dependent Partial ventilator dependence Liberated from ventilator, tracheostomy tube present Coronary artery disease Admission creatinine Closed head injury Married found new renal dysfunction associated with increased hospital mortality. The higher risk of death associated with patients admitted to the medical service probably represents the fact that these patients received a tracheostomy for worsening of chronic medical problems, which may be irreversible. However, surgically admitted patients may have had tracheostomies necessitated by complications of surgery, which may be more self-limited and reversible, and hence they were more likely to recover. We also found that the lack of a permanent feeding tube was associated with an increased the risk of dying. This probably reflects local practice of which patients receive these feeding tubes: terminally ill patients are not given a permanent feeding tube. Patients expected to live, but unable to eat, would frequently receive a permanent feeding tube as part of discharge preparation. While lower nadir hemoglobin levels were also associated with increased mortality, patients with lower hemoglobin levels were also more likely to be transfused and it is not possible to separate the effects of anemia from transfusion. Lower hemoglobin levels mean lower oxygen concentrations and may result in lower oxygen delivery, which may lead to end-organ damage and death. 17 In other studies, however, Spiess et al 18 showed fewer complications in patients with lower hemoglobin concentrations after cardiac surgery, and Hebert et al 19 found that transfusion to a higher hemoglobin threshold may have increased mortality in critically ill patients. While the physiologic derangements of older age that lead to a greater risk of death are not understood, other studies 11,20,21 have found that age predicts mortality in critically ill patients. Of the 334 hospital survivors, only 191 patients (57%) were completely liberated from mechanical ventilation. Survivors were younger and more rehabilitated on discharge. Whether patients were more rehabilitated because they had longer hospital stays or whether they had longer hospital stays because their physicians saw continued progress in walking, eating, and being liberated cannot be determined from this study and is an area for future research. Of the 191 liberated patients, only 57 patients (30%) had their tracheostomy tubes decannulated before discharge. Again, the better outcome was associated with longer stays, higher costs, and a higher functional recovery on discharge; that is, they could walk and eat. Patients discharged without tracheostomy tubes also had the best 1-year survival (92%). Table 6 SF-36 Results at Follow-up* Variables Ventilator (n 19) With Tracheostomy Tube (n 25) Without Tracheostomy Tube (n 22) Physical functioning 22 (10 49) 25 (9 76) 50 (15 79) Role-physical 12 (0 25) 0 (0 38) 50 (0 75) Bodily pain 50 (41 77) 82 (42 100) 62 (41 89) General health 52 (32 57) 57 (38 87) 70 (47 82) Vitality 40 (22 61) 45 (30 60) 48 (30 60) Social functioning 50 (38 75) 75 (47 91) 88 (75 100) Role-emotional 83 (33 100) 100 (33 100) 100 (33 100) Mental health 76 (62 76) 70 (63 80) 72 (58 76) *Data are shown as median (interquartile range). Both completely (n 16) and partially (n 3) dependent patients at hospital discharge. Patients discharged from hospital liberated from mechanical ventilation but tracheostomy tube is still present. Patients discharged from hospital liberated from mechanical ventilation and the tracheostomy tube decannulated. p CHEST / 125 / 1/ JANUARY,

7 We found that functional status the ability to eat and walk was an important association with better outcome. This is similar to other studies 9,22,23 that found that the inability to walk predicts 30-day mortality in pneumonia and heart failure patients and 6-month mortality in survivors of prolonged mechanical ventilation. Posthospital survival was predominantly related to ventilator status (Table 5). Admission creatinine, coronary artery disease, closed head injury, and marital status were also significantly associated with death. Patients who were completely ventilator dependent on hospital discharge had a fourfold increased risk of death. Being partially ventilator dependent or being liberated from mechanical ventilation but still retaining the tracheostomy tube placed patients at a lesser but still elevated risk of death. We found that an elevated admission creatinine is predictive of late mortality. While a similar effect has been found in patients undergoing percutaneous coronary interventions and in postmenopausal women with coronary heart disease, a large population-based study found that while renal insufficiency on hospital admission was associated with an increased mortality after discharge, it could be attributed to the factors that cause renal dysfunction, such as diabetes mellitus and hypertension However, our result was independent of the presence of diabetes mellitus and hypertension. Coronary artery disease is a well-known risk factor for death. 27 This increased risk occurred despite the fact that 28% of hospital survivors with coronary artery disease had surgical (25%) or angioplastic (4%) revascularization (70% of nonrevascularized patients compared to 45% of revascularized patients were dead at follow-up). We found that closed head injury patients had a better posthospital survival. This is similar to our other finding that these patients were less likely to die in hospital. We speculate that their good outcome is related to these patients mostly being younger and healthier before hospital admission. Being married was also beneficial to survival. A previous study 28 has found similar benefits in married patients after hospitalization for respiratory failure from COPD. Our 30% 6-month mortality was similar to the 33% found by Seneff et al 29 in chronically ventilated patients being considered for transfer to a specialized weaning center. Similar 6-month mortality rates of 25% and 22% were found in patients receiving at least 5 days or at least 7 days, respectively, of mechanical ventilation. 9,30 After this initial high 6-month mortality, mortality curves tended to flatten out (Fig 1). At follow-up, we found that most responders had good emotional health but remained with major physical limitations. This is similar to Chatila et al, 31 who evaluated 25 survivors of a ventilator rehabilitation unit. They found mild-to-moderate impairment in quality of life as measured by the sickness impact profile in most patients. Subjects with chronic diseases did worse. There are several limitations to this study. The study was conducted at only one hospital and may not be representative of other patient populations. In particular, we had a high proportion of trauma and head injury patients, reflecting level I trauma status. Although timing of tracheostomy and choice of patients receiving tracheostomy may differ from center to center and physician to physician, we chose to study tracheostomy patients rather than a defined number of days of mechanical ventilation because the performance of a tracheostomy is a natural decision point for patients and their families to decide on the amount of care to continue to provide. In the United States, tracheostomy also moves the patient into a particular DRG with an increased reimbursement. However, this limits its comparison with other studies that used a defined number of days, such as 4, 7, or 10, to define respiratory failure. Another limitation is the low level of patient participation (39%) in completing the SF-36. This may have limited the ability to detect clinically important differences between groups. Many patients died before the study was conducted. They tended to be sicker and more likely to require ventilator or airway support at hospital discharge. If studied before they died, they may have had even lower scores on the SF-36. However, among survivors, there were only minor differences in demographics, comorbidities, and hospital course between participants, those who refused, and those unable to be contacted. In summary, we found that overall survival and functional status are poor in patients with tracheostomy for respiratory failure. Patients who are liberated from mechanical ventilation and have their tracheostomy tubes removed have a much better survival; however, it comes at a higher hospital cost and longer length of stay. References 1 Heffner JE. Medical indications for tracheostomy. Chest 1989; 96: Stauffer JL, Olson DE, Petty TL. Complications and consequences of endotracheal intubation and tracheotomy: a prospective study of 150 critically ill adult patients. Am J Med 1986; 14: Kapadia FN, Bajan KB, Raje KV. Airway accidents in intubated intensive care unit patients: an epidemiological study. Crit Care Med 2000; 28: Kollef MH, Ahrens TS, Shannon W. Clinical predictors and 226 Clinical Investigations in Critical Care

8 outcomes for patients requiring tracheostomy in the intensive care unit. Crit Care Med 1999; 27: Fischler L, Erhart S, Kleger GR, et al. Prevalence of tracheostomy in ICU patients: a nation-wide survey in Switzerland. Intensive Care Med 2000; 26: Brook AD, Sherman G, Malen J, et al. Early versus late tracheostomy in patients who require prolonged mechanical ventilation. Am J Crit Care 2000; 9: Esteban A, Anzueto A, Frutos F, et al. Characteristics and outcomes in adult patients receiving mechanical ventilation: a 28-day international study. JAMA 2002; 287: Vasilyev SS, Chaap RN, Mortensen JD. Hospital survival rates of patients with acute respiratory failure in modern respiratory intensive care units. Chest 1995; 107: Engoren M, Fenn Buderer N, Zacharias A. Long-term survival and health status after prolonged mechanical ventilation after cardiac surgery. Crit Care Med 2000; 28: Muir JF, Girault C, Cardinaud JP, et al. Survival and long-term follow-up of tracheostomized patients with COPD treated by home mechanical ventilation: a multicenter French study in 259 patients; French Cooperative Study Group. Chest 1994; 106: Kurek CJ, Cohen IL, Lambrinos J, et al. Clinical and economic outcome of patients undergoing tracheostomy for prolonged mechanical ventilation in New York state during 1993: analysis of 6,353 cases under diagnosis-related group 483. Crit Care Med 1997; 25: Ware JE Jr, Snow KK, Kosinski M, et al. SF-36 health survey: manual and interpretation guide. Lincoln, RI: Quality Metric, Stokes ME, Davis CS, Koch GG. Categorical data analysis using the SAS system. Cary, NC: SAS Institute, 1995; Lawless JF. Statistical models and methods for lifetime data. New York, NY: John Wiley & Sons, 1982; Mangano CM, Diamondstone LS, Ramsay JG, et al. Renal dysfunction after myocardial revascularization: risk factors, adverse outcomes, and hospital resource utilization: the Multicenter Study of Perioperative Ischemia Research Group. Ann Intern Med 1998; 128: Brivet FG, Kleinknecht DJ, Loirat P, et al. Acute renal failure in intensive care units-causes, outcome, and prognostic factors of hospital mortality; a prospective, multicenter study; French Study Group on Acute Renal Failure. Crit Care Med 1996; 24: Bishop MH, Shoemaker WC, Appel PL, et al. Prospective, randomized trial of survivor values of cardiac index, oxygen delivery, and oxygen consumption as resuscitation endpoints in severe trauma. J Trauma 1995; 38: Spiess BD, Ley C, Body SC, et al. Hematocrit value on intensive care unit entry influences the frequency of Q-wave myocardial infarction after coronary artery bypass grafting: The Institutions of the Multicenter Study of Perioperative Ischemia (McSPI) Research Group. J Thorac Cardiovasc Surg 1998; 116: Hebert PC, Wells G, Blajchman MA, et al. A multicenter, randomized, controlled clinical trial of transfusion requirements in critical care: Transfusion Requirements in Critical Care Investigators, Canadian Critical Care Trials Group. N Engl J Med 1999; 340: Knaus WA, Wagner DP, Draper EA, et al. The APACHE III prognostic system: risk prediction of hospital mortality for critically ill hospitalized adults. Chest 1991; 100: Cohen IL, Lambrinos J. Investigating the impact of age on outcome of mechanical ventilation using a population of 41,848 patients from a statewide database. Chest 1995; 107; Keeler EB, Kahn KL, Draper D, et al. Changes in sickness at admission following the introduction of the prospective payment system. JAMA 1990; 264: Daley JD, Jencks S, Draper D, et al. Predicting hospitalbased mortality for Medicare patients: a method for predicting mortality for patients with stroke, pneumonia, acute myocardial infarction, and congestive heart failure. JAMA 1988; 260: Best PJ, Lennon R, Ting HH, et al. The impact of renal insufficiency on clinical outcomes in patients undergoing percutaneous coronary interventions. J Am Coll Cardiol 2002; 39: Shlipak MG, Simon JA, Grady D, et al. Renal insufficiency and cardiovascular events in postmenopausal women with coronary heart disease. J Am Coll Cardiol 2001; 38: Garg AX, Clark WF, Haynes RB, et al. Moderate renal insufficiency and the risk of cardiovascular mortality: results from the NHANES I. Kidney Int 2002; 61: Jones WB, Riley CP, Reeves TJ, et al. Natural history of coronary artery disease. Bull N Y Acad Med 1972; 48: Almagro P, Calbo E, Ochoa de Echaguen A, et al. Mortality after hospitalization for COPD. Chest 2002; 121: Seneff MG, Wagner D, Thompson D, et al. The impact of long-term acute-care facilities on the outcome and cost of care for patients undergoing prolonged mechanical ventilation. Crit Care Med 2000; 28: Douglas SL, Daly BJ, Brennan PF, et al. Outcomes of long-term ventilation patients: a descriptive study. Am J Crit Care 1997; 6: Chatila W, Keimer DT, Criner GJ. Quality of life in survivors of prolonged mechanical ventilatory support. Crit Care Med 2001; 29: CHEST / 125 / 1/ JANUARY,