AAST 2013 PLENARY PAPER
Are all deaths recorded equally? The impact of hospice care on risk-adjusted mortality Rosemary A. Kozar, MD, PhD, John B. Holcomb, MD, Wei Xiong, MSc, and Avery B. Nathens, MD, PhD, Houston, Texas
BACKGROUND: Hospice care provides dignity and comfort at the end of life. While patients transferred to hospice die, they are often not recorded as inhospital deaths in trauma registries or in some administrative discharge data. Mortality rates for the purpose of database research, performance improvement, or public reporting may therefore be artificially low. The current study sought to determine the impact of discharges to hospice on risk-adjusted mortality for trauma deaths reported to the Trauma Quality Improvement Program. METHODS: Performance from Trauma Quality Improvement Program centers in 2011 was evaluated using risk-adjusted mortality with observed-toexpected mortality ratios derived from a logistic regression model. The impact of discharge to hospice on performance was measured by determining changes in performance if hospice cases were treated as survivors rather than deaths. Differences between groups were compared by nonparametric Wilcoxon rank-sum test. RESULTS: From the 167 centers with 126,259 injured patients, there were 8,862 deaths: 746 (8.4%) were discharged to a hospice, and the remainder was counted as in-hospital deaths. Overall, 106 centers (63.5%) reported at least one discharge to hospice, with the proportion of deaths ranging from 1.6% to 57%. Logistic regression demonstrated that age greater than 70 years (odds ratio [OR], 4.3; 95% confidence interval [CI], 3.5Y5.1), male sex (OR, 0.7; 95% CI, 0.6Y0.8), nonblack race (OR, 1.9; 95% CI, 1.3Y2.7), noncommercial insurance (OR, 1.4; 95% CI, 1.1Y1.7), and comorbidity counts greater than 2 (OR, 1.3; 95% CI, 1.1Y1.6) were associated with hospice care. If patients transferred to a hospice were treated as survivors in the estimation of risk-adjusted mortality, 34 centers (20%) would have a change in status. Changes would be in both directions for average-performing centers, while high-performing centers would seem worse and poor-performing centers would seem better. For centers that reported hospice deaths, the relative risk-adjusted mortality decreased by 8.8% for every 10% increase in the proportion of deaths recorded as discharged to a hospice. CONCLUSION: Given the large variation in the proportion of deaths recorded as discharged to a hospice rather than as in-hospital deaths, there is the potential for significant distortion of actual performance. Failure to consider this potential may misguide efforts directing performance improvement, research, and national reporting. Discharges to a hospice should be included with in-hospital deaths when reporting risk-adjusted mortality. (J Trauma Acute Care Surg. 2014;76: 634Y641. Copyright * 2014 by Lippincott Williams & Wilkins) KEY WORDS: Risk-adjusted mortality; trauma benchmarking; hospice care; public reporting; Trauma Quality Improvement Performance (TQIP).
P
alliative care is a philosophy of care that focuses on enhancing patient comfort and quality of life. It is becoming increasingly sought in the care of critically ill patients and is recognized as an important aspect of care for the surgical patient by the American College of Surgeons.1,2 Hospice care is one type of palliative care that is appropriate for patients during their last 6 months of life.3 Hospice care can provide dignity and comfort at the end of life. However, we noted that trauma patients at our institution (Memorial Hermann Hospital; R.A.K. and J.B.H.) transferred to a hospice care were not being counted in our trauma registry
Submitted: September 13, 2013, Revised: November 11, 2013, Accepted: November 12, 2013. From the Department of Surgery (R.A.K., J.B.H.), and Center for Translational Injury Research (J.B.H.), The University of Texas Health Sciences Center at Houston, Houston, Texas; and Department of Surgery (W.X., A.B.N.), Sunnybrook Health Sciences Centre, Toronto, Canada. This study was presented at the 72nd annual meeting of the American Association of Surgery for Trauma, September 18Y21, 2013, in San Francisco, California. Address for reprints: Rosemary Kozar, MD, PhD, 6431 Fannin, MSB 4.284; email: [email protected]. DOI: 10.1097/TA.0000000000000130
634
as inpatient mortalities and therefore not undergoing our multidisciplinary peer review process for mortality. We had 16 hospice transfers in 2010, 23 in 2011, and 51 in 2012, representing 8.5%, 11.8%, and 18.5% of all deaths, respectively. Although we use the American College of Surgeons’ Trauma Quality Improvement Performance (TQIP) for trauma-specific benchmarking,4 our hospital, Memorial Hermann Hospital, also relies on University Health Care Consortium (UHC), an alliance of academic medical centers that provides comparative data from their large administration discharge database and now includes trauma, for quality comparisons.5 Both UHC and TQIP use risk adjustment calculations in the determination of center performance. TQIP considers patients transferred to a hospice care as deaths, while UHC considers these patients as discharged alive. We became concerned that quality improvement measures may be jeopardized and that benchmarking may be compromised with variable treatment of hospice transfers. We therefore queried other large databases to determine their means of handling hospice patients. Both TQIP and National Surgical Quality Improvement Program consider patients discharged to a hospice as deaths. Medicare and Medicaid consider hospice admissions as deaths if death occurs within 30 days from transfer.6 At the time we initiated this study, UHC and the National Trauma Data Bank both J Trauma Acute Care Surg Volume 76, Number 3
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
J Trauma Acute Care Surg Volume 76, Number 3
considered transfers to a hospice care as a survivor to discharge. We hypothesized that if trauma patients transferred to a hospice care were reported as discharged alive, mortality rates for the purpose of performance improvement, benchmarking, registrybased research, or public reporting would be artificially low. The purpose of the current study was therefore to determine the impact of discharges to a hospice care on risk-adjusted mortality for trauma deaths reported to TQIP.
PATIENTS AND METHODS Centers and Subjects Data were derived from TQIP participating centers in 2011 and included all TQIP eligible patients (age 9 18 years with at least one body region with an Abbreviated Injury Scale [AIS] score Q 3). TQIP centers are limited to Level 1 or 2 centers that were verified by either the American College of Surgeons or regional authorities. TQIP excludes patients with no signs of life on arrival, those with a preexisting advanced directive to withhold life-sustaining care, and elderly patients (age Q 65 years) with an isolated hip fracture as a result of a fall. We evaluated the entire TQIP population and then separately analyzed elderly patients (age Q 65 years) and patients with isolated severe traumatic brain injury (TBI) defined as those with a head AIS score of 4 or greater or head AIS score of 3 or greater with initial motor Glasgow Coma Scale (GCS) score of 4 or lower and other body regions AIS score of 2 or lower.
Statistical Analyses The primary objective was to determine if hospital’s riskadjusted mortality relative to peer centers would be affected if patients discharged to a hospice were considered to be survivors rather than deaths. Mortality risk adjustment within TQIP has been well described7 but, briefly, is based on a logistic regression model with age, sex, injury mechanism, transfer status, systolic blood pressure, GCS motor, AIS score by body region, and comorbidities. Injury severity is included in the model using an DRG International Classification of DiseasesV9th Rev.Ybased approach (ICISS), which captures every injury in the DRG International Classification of DiseasesV9th Rev. injury code range.8 The probability of mortality for each patient was derived from the logistic regression model. Probabilities were summed for each hospital to estimate the hospital’s risk-adjusted mortality. For each hospital, the observed number of events (O) was divided by the risk-adjusted expected number of events (E) to produce an O/E ratio. If the lower limit of the 95% confidence interval (CI) of the O/E ratio was greater than unity, then the center was considered to be a poor performer (higher than expected mortality). By contrast, if the upper limit of the 95% CI of the O/E ratio was lower than unity, the center was considered to be a good performer. All centers with a 95% CI for their O/E ratio that included unity were considered to be average performers. Missing data for blood pressure (1.2%), GCS (3.1%), and heart rate (1.9%) were imputed using multiple imputation.9 Impact of discharge to a hospice care on center performance was measured by determining the changes in the performance if hospice cases were treated as survivors rather than deaths. Differences between groups (hospice vs. nonhospice deaths) were compared by nonparametric Wilcoxon rank-sum test for skewed
Kozar et al.
continuous data. Odds ratios are presented with their 95% CIs to demonstrate the strength of association between specific patient characteristics and hospice discharge.
RESULTS Data were submitted to TQIP by 167 centers on 126,259 injured patients. Baseline patient characteristics that were used in models of risk-adjusted mortality are shown in Table 1. Patients were most often white and male, sustained a fall, were not transferred, had noncommercial insurance, and were severely injured. There were a total of 8,862 deaths, of which 746 (8.4%) were discharged to a hospice and 8,116 (91.6%) were captured as in-hospital deaths. Overall, 106 centers (63.5%) reported at least one discharge to a hospice care. Across centers, the proportion of deaths recorded as discharged to a hospice ranged from 0% to 57% (Fig. 1), with a mean (SD) of 8.7% (11%) deaths as hospice. There were 61 centers that reported zero patients transferred to a hospice. Of the 746 hospice patients, time from hospital admission to transfer to a hospice care was 9.7 (11.4) days. When comparing hospice patients with in-hospital deaths, univariate analysis revealed that patients transferred to hospice care were older (77.1 [16.2] years vs. 55.9 [23.6] years), had lower Injury Severity Score (ISS) (19.2 [8.4] vs. 28.3 [13.2]), and a lower head AIS score (3.1 [1.9] vs. 3.3 [2.1]) compared with patients recorded as in-hospital deaths. After controlling for age, sex, race, payment status, and comorbidities, logistic regression demonstrated that older age, female sex, nonblack race, noncommercial insurance, and a higher burden of comorbidities were all significantly associated with hospice care (Table 2). If patients transferred to a hospice care were treated as survivors, crude mortality at centers that reported hospice discharges would decrease from 7.0% to 6.4%, a relative reduction of more than 8%. For every 10% increase in the proportion of deaths a center recorded as discharged to a hospice, there was an 8.8% relative reduction in their risk-adjusted mortality (Fig. 2). In addition, if transfers to hospice were treated as survivors for the estimation of risk-adjusted mortality, 34 centers (20.4%) demonstrated a significant change in their performance status (Table 3). Changes were in both directions for average-performing centers, while high-performing centers seemed worse and poorperforming centers seemed better. This same pattern of change held true when patients with TBI or elderly patients were analyzed separately (Table 3).
DISCUSSION In the current study, we sought to determine the impact of discharges to a hospice care on risk-adjusted mortality for trauma deaths reported to TQIP. Hospice transfers occurred in the majority of participating centers, and older age, female sex, nonblack race, noncommercial insurance, and comorbidities were all significantly associated with hospice care. Importantly, if patients transferred to a hospice care were treated as survivors, both crude and risk-adjusted mortality rates would decrease, with associated changes in performance status for some centers.
* 2014 Lippincott Williams & Wilkins
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
635
J Trauma Acute Care Surg Volume 76, Number 3
Kozar et al.
TABLE 1. Demographics for all Patients Cohort (n = 126,259) Male, % Race, % Asian Black White Other Missing Transfer, yes, % Mechanism, % Cut/pierce Firearm Fall Motorcyclist Motor vehicle collision Pedestrian Struck Other Insurance status, % Commercial Noncommercial Other Age, mean (SD), y Emergency department systolic blood pressure, mean (SD) Pulse, mean (SD) ISS, mean (SD) GCS motor, mean (SD) Head AIS score, mean (SD) Face AIS score, mean (SD) Neck AIS score, mean (SD) Chest AIS score, mean (SD) Abdomen AIS score, mean (SD) Spine AIS score, mean (SD) Comorbidity, % Heart disease Cancer Liver disease Alcoholism Smoking Stroke Diabetes Hypertension Renal disease Sensorium Obesity Respiratory disease Functional dependence Bleeding disorder Peripheral vascular disease Steroid use
All Patients
Hospice Deaths (n = 746)
In-hospital Deaths (n = 8,116)
p (Hospice vs. In-hospital)
66.4
53.2
70.7
G0.0001 G0.0001
1.6 14.5 71.4 9.7 2.8 31.6
1.1 4.8 87.9 4.3 1.9 37.3
2.0 15.0 70.5 8.9 3.7 27.2
3.2 6.6 38.3 7.7 24.8 6.9 6.2 6.5
0.4 2.7 80.4 1.9 9.0 2.5 2.0 1.1
1.4 17.9 37.2 6.6 22.2 8.8 2.6 3.4
37.2 53.7 9.1 50.1 (22.2) 138 (28.6)
16.8 78.6 4.7 77.1 (16.2) 147 (32.2)
27.5 61.5 11.0 55.6 (23.6) 128 (44.0)
89 (20.9) 17 (9.0) 5.4 (1.5) 1.7 (1.9) 0.4 (0.7) 0.05 (0.3) 1.2 (1.6) 0.5 (1.1) 0.6 (1.2)
88.3 (21.0) 19.2 (8.4) 4.8 (1.9) 3.1 (1.9) 0.4 (0.7) 0.03 (0.26) 0.6 (1.3) 0.1 (0.6) 0.5 (1.1)
94.2 (30.5) 28.3 (13.7) 2.9 (2.2) 3.3 (2.1) 0.5 (0.8) 0.1 (0.5) 1.5 (1.8) 0.7 (1.4) 0.7 (1.4)
4.8 0.9 0.6 9.6 15.7 2.3 11.3 28.5 0.7 7.5 3.1 6.9 1.4 5.4 0.2 0.4
17.0 3.8 2.0 5.1 5.5 8.4 20.0 53.6 1.7 20.8 2.4 11.5 7.0 16.9 0.4 1.5
8.0 1.9 1.4 6.7 5.2 3.6 12.8 27.7 1.4 6.6 3.0 6.4 1.7 9.2 0.3 0.3
G0.0001
With the heightened emphasis on public reporting, there is the potential for significant distortion of actual performance, particularly for high-performing centers. As we showed, 34.8% (8 of 23) of high-performing trauma centers would drop to 636
G0.0001 G0.0001
G0.0001 G0.0001 G0.0001 G0.0001 G0.0001 G0.0001 G0.0001 G0.0001 G0.0001 G0.0001 0.0001 G0.0001 0.0006 0.1859 0.0940 0.7389 G0.0001 G0.0001 G0.0001 0.4573 G0.0001 0.3782 G0.0001 G0.0001 G0.0001 0.5183 G0.0001
average performers if hospice transfers were not considered deaths. Centers for Medicare and Medicaid Services currently reports mortality measures for acute myocardial infarction, heart failure, and pneumonia patients,10 but this is expected to * 2014 Lippincott Williams & Wilkins
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
J Trauma Acute Care Surg Volume 76, Number 3
Kozar et al.
Figure 1. Variation of hospice care among 2011 TQIP centers.
increase in scope. There are commercial agencies such as US News & World Report, HealthGrade, and Thomson-Reuters as well as health care consortia, such as UHC and Leapfrog, that now report mortality statistics available to the public.11 How these agencies take into account hospice care varies widely. For example, Centers for Medicare and Medicaid Services reports on 30-day mortality. Thus, patients transferred to a hospice care who died during the index admission or within 30 days would be considered a death unless the patient was already enrolled in Medicare’s Hospice Benefit plan or if death occurred within the first day of hospital admission. In contrast, Thomson-Reuters considers patients discharged to a hospice care during the index admission as survivors11 as does UHC. Before beginning this project, the National Trauma Data Bank was excluding hospice patients in its death analysis but, as a result of this work, have since modified its practice and now report hospice transfers as deaths beginning with the 2012 report, similar to TQIP and National Surgical Quality Improvement Program. We have been in contact with UHC, and they are investigating the possibility of changing their practice at least for trauma patients, although this has not yet taken effect. Performance improvement remains an essential component of trauma center verification.12 Whether intentional, deaths may not undergo peer review if hospice patients are considered survivors. We (R.A.K. and J.B.H.) learned that this was happening at our own institution for patients who were not on the trauma service. We are now reviewing all in-hospital hospice transfers as deaths. Multidisciplinary peer review is paramount to effective quality improvement, and missed opportunities for improvement could occur if these patients were not subject to the same rigorous reviews as all inpatient deaths. The study by Franklin et al.,12 however, suggested that there were few missed opportunities for improvement when they looked at trauma patients who had withdrawal of care during the index admission. One potential disadvantage of considering patients transferred to a hospice care as deaths is that the inclusion of these patients leads to a reported increase in crude and risk-adjusted mortality. This same concern has been expressed for patients who are given do-not-resuscitate status, have advanced
directives, or undergo withdrawal of care while in the intensive care unit (ICU).13 As shown by others, most of these patients are elderly with minimal traumatic injuries.12Y15 Ideally, risk adjustment models should account for preexisting conditions and comorbidities. Thompson et al.16 have reported a comorbidity index for use in risk adjustment for injured patients; however, this may still not capture all needed fields in the minimally injured elderly patient.12 More importantly, no risk adjustment model can take into consideration the wants and needs of individuals with respect to end-of-life decisions, which are highly dependent on local culture and the sociocultural context of providers and their patients.17Y19 As such, it will be extraordinarily challenging to ‘‘adjust’’ away the effects for decisions about end-of-life care, which often happen in unique and personal discussions between providers, their patients, and/or surrogate decision makers. A limitation of the current study is that TQIP considers patients transferred to a hospice care as deaths, although death is not verified by participating centers. When we reviewed the 51 patients transferred to a hospice care at the University of Texas-Houston, 20 of our patients remained in the same hospital unit but had a virtual transfer of status, and all died in the hospital within 24 hours. We were surprised to learn that the other 31 patients were transferred to outside hospices. Follow-up data were available in 22 of these patients, and all but one patient died in the hospice facility during the index TABLE 2. Independent Risk Factors for Discharge to Hospice Among All Deaths Risk Factor Age 9 70 y Sex, male Race, nonblack Payment, noncommercial insurance Comorbidity counts 9 2
Adjusted Odds Ratio (95% CI)
p
4.3 (3.5Y5.1) 0.7 (0.6Y0.8) 1.9 (1.3Y2.7) 1.4 (1.1Y1.7)
G0.0001 G0.0001 0.0006 G0.0001
1.3 (1.1Y1.6)
0.0060
Logistic model receiver operating characteristic curve = 0.740 and Hosmer-Lemeshow calibration p = 0.2102
* 2014 Lippincott Williams & Wilkins
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
637
J Trauma Acute Care Surg Volume 76, Number 3
Kozar et al.
Figure 2. Increased reporting of discharges to a hospice decreased O/E mortality ratios for 2011 TQIP centers. Every 10% increase of hospice percentage decreases relative risk of O/E ratio by 8.8%.
admission. The average time from transfer to death was 17 days but ranged between 1 day and 180 days. Whether our data are representative of participating centers is unknown and raises an important question, when should we consider hospice patients transferred out of our trauma centers as deaths? Should all deaths be recorded equally? The time and effort to track these patients after discharge were substantial, and some patients did not die for several months. As an initial step in addressing this problem, we suggest that all hospice transfers that occur within the same facility as the index admission be considered as deaths and undergo mortality review. Similar to the results identified with transfers to a hospice care, Kelly et al.20 reported that a liberal policy of transfers to long-term acute care facilities based on a discharge protocol might lead to reductions in reported hospital mortality. A recent study by Brinkman et al.21 examined the influence of using mortality 1, 3, or 6 months after ICU admission instead of inhospital mortality on their standardized mortality ratio. They used a large Dutch database from almost 81,000 ICU patients in mixed medicine-surgical units and found that mortality ratios and rank differed depending on whether in-hospital mortality or mortality at 1, 3, or 6 months after ICU admission was used. As differences in discharge policies and mortality at time points after ICU discharge affected reported mortality, they suggested that later time points should be used as a quality indicator for benchmarking purposes. This issue has also been examined in trauma patients. Bulger et al.22 examined inhospital and 28-day mortality in severely injured patients who received out-of-hospital hypertonic saline and found less than a 1% difference in mortality. Others have evaluated later time points and have shown that trauma-related deaths continue for at least the first year after injury.22Y26 At the current time, there are no data in TQIP on the time to death of trauma patients once they leave the trauma center. We also do not have information on the reason for transfer to a hospice or the do-notresuscitate status of included patients. However, TQIP is 638
now capturing withdrawal of care, which might provide some information on different approaches to end-of-life care across participating centers. Lastly, we do not know the extent to which palliative care teams are involved with the care of injured patients that do not involve hospice transfer. Since death was not imminent on discharge to a hospice, perhaps, we should encourage the more frequent use of palliative care. It is being increasingly recognized as beneficial, and models to integrate palliative care into surgical and ICUs have been described.27 These all represent opportunities to enhance end-of-life care and should be without the fear of adversely effecting performance status. There are in fact a number of reports showing that early involvement of the palliative care team may shorten hospital or ICU length of stay without impacting mortality.28Y30 In conclusion, public reporting of risk-adjusted mortality is increasing. Given the large variation in proportion of deaths TABLE 3. Change in Center Outlier Status When Patients Transferred to a Hospice Were Coded as Survivors Center Performance All patients Average Good Poor TBI patients Average Good Poor Elderly patients Average Good Poor
Total No. No. Changed Centers to Good 120 23 24
7
143 11 13
10
135 12 20
7
No. Changed to Average
No. Changed to Poor 10
8 9 8 8 4 9 6 11
* 2014 Lippincott Williams & Wilkins
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
J Trauma Acute Care Surg Volume 76, Number 3
Kozar et al.
recorded as discharged to a hospice rather than as an inpatient death, there is the potential for distortion of actual performance. Failure to consider this potential may also misguide efforts directing performance improvement activities, registry-based research, and national reporting. Discharges to hospice should be included with in-hospital deaths when reporting risk-adjusted mortality, although better methods to capture preexisting conditions and preferences at the end of life are needed to accurately risk-adjust for mortality in these patients.
14.
15.
16.
17. AUTHORSHIP R.A.K. conducted the literature search for this study, which R.A.K., J.B.H., and A.B.N. designed. W.X. collected the data and performed statistical analysis. R.A.K., W.X., and A.B.N. contributed to data analysis. R.A.K., W.X., and A.B.N. wrote the manuscript. R.A.K., J.B.H., and A.B.N. contributed to critical revision.
18.
19.
20. ACKNOWLEDGMENT We thank Sheila Lopez, RN, BSA, MA, and Edmundo Dipasupil, CSTR, for their assistance with our (R.A.K., J.B.H.) trauma center data.
DISCLOSURE The authors declare no conflicts of interest.
REFERENCES 1. Surgical Palliative Care: A resident’s guide. Dunn G, Martensen R, Weissman D, Bradley CT, Brasel KJ, Castillo EH, Cole TR, Deville K, Dixon P, Hinshaw DB et al. Available at: http://www.facs.org/palliativecare/ surgicalpalliativecareresidents.pdf. Accessed August 7, 2013. 2. The surgeon and palliative care. Available at: http://www.facs.org/ palliativecare/index.html. Accessed August 7, 2013. 3. Hui D, De La Cruz M, Mori M, Parsons HA, Kwon JH, Torres-Vigil I, Kim SH, Dey R, Hutchins R, Liem C, et al. Concepts and definitions ‘‘supportive care,’’ best supportive care, palliative care, and hospice care in the published literature, dictionaries, and textbooks. Support Care Cancer. 2013;21:659Y685. 4. Trauma quality improvement program (TQIP). Available at: http:// www.facs.org/trauma/ntdb/tqip.html. Accessed August 4, 2013. 5. About University Healthcare Consortium. Available at: https://www.uhc. edu/12443.htm. Accessed August 7, 2013. 6. Centers for Medicare and Medicaid. Outcome measures. Available at: http:// www.cms.gov/Medicare/Quality-Initiatives-Patient-Assessment-Instruments/ HospitalQualityInits/OutcomeMeasures.html. Accessed August 7, 2013. 7. Newgard CD, Fildes JJ, Wu L, Hemmila MR, Burd RS, Neal M, Mann NC, Shafi S, Clark DE, Goble S, et al. Methodology and analytic rationale for the American College of Surgeons Trauma Quality Improvement Program. J Am Coll Surg. 2013;216(1):147Y157. 8. Osler T, Rutledge R, Deis J, Bedrick E. ICISS: an International Classification Of Disease-9 based Injury Severity Score. J Trauma. 1996;41: 380Y386. 9. Rubin DB. Inference and missing Data. Biometrika. 1976;63:581Y592. 10. Mortality measures overview. Publicly reporting risk-standardized, 30-day mortality measures for AMI, HF, and PN. Available at: http://www.qualitynet. org. Accessed August 21, 2013. 11. Cassel JB, Jones AB, Meier DE, Smith TJ, Spragens LH, Weissman D. Hospital mortality rates: how is palliative care taken into account? J Pain Symptom Manage. 2010;40:914Y925. 12. Franklin GA, Cannon RW, Smith JW, Harbrecht BG, Miller FB, Richardson JD. Impact of withdrawal of care and futile care on trauma mortality. Surgery. 2011;150:854Y860. 13. Adams SD, Cotton BA, Wade CE, Kozar RA, Dipasupil E, Podbielski JM, Gill BS, Duke JR, Adams PR, Holcomb JB. Do not resuscitate status, not
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
age, affects outcomes after injury: an evaluation of 15,227 consecutive trauma patients. J Trauma Acute Care Surg. 2013;74(5):1327Y1330. Sise MJ, Sise CB, Thorndike JF, Kahl JE, Calvo RY, Shackford SR. Withdrawal of care: a 10-year experience at a Level I trauma center. J Trauma Acute Care Surg. 2012;72:1186Y1193. Kahl JE, Calvo RY, Sise MJ, Sise CB, Thorndike JF, Shackford SR. The changing nature of death on the trauma service. J Trauma Acute Care Surg. 2013;75(2):195Y201. Thompson HJ, Rivara FP, Nathens A, et al. Development and validation of the mortality risk for trauma comorbidity index. Ann Surg. 2010;252: 370Y375. Hanchate A, Kronman AC, Young-Xu Y, Ash AS, Emanuel E. Racial and ethnic differences in end-of-life costs: why do minorities cost more than whites? Arch Intern Med. 2009;169(5):493Y501. Adames HY, Chavez-Duen˜as NY, Fuentes MA, Salas SP, Perez-Chavez JG. Integration of Latino/a cultural values into palliative health care: a culture centered model. Palliat Support Care. 2013;17:1Y9. Periyakoil VS, Stevens M, Kraemer H. Multicultural long-term care nurses’ perceptions of factors influencing patient dignity at the end of life. J Am Geriatr Soc. 2013;61(3):440Y446. Kelly KB, Koeppel ML, Como JJ, Carter JW, McCoy AM, Claridge JA. Continued rationale of why hospital mortality is not an appropriate measure of trauma outcomes. Am J Surg. 2012;203(3):366Y369; discussion 369. Brinkman S, Abu-Hanna A, de Jonge E, de Keizer NF. Prediction of longterm mortality in ICU patients: model validation and assessing the effect of using in-hospital versus long-term mortality on benchmarking. Intensive Care Med. 2013;39:1925Y1931. Bulger EM, May S, Kerby JD, Emerson S, Stiell IG, Schreiber MA, Brasel KJ, Tisherman SA, Coimbra R, Rizoli S, et al.; ROC investigators. Out-ofhospital hypertonic resuscitation after traumatic hypovolemic shock: a randomized, placebo controlled trial. Ann Surg. 2011;253(3):431Y441. Probst C, Zelle BA, Sittaro NA, Lohse R, Krettek C, Pape HC. Late death after multiple severe trauma: when does it occur and what are the causes? J Trauma. 2009;66(4):1212Y1217. Claridge JA, Leukhardt WH, Golob JF, McCoy AM, Malangoni MA. Moving beyond traditional measurement of mortality after injury: evaluation of risks for late death. J Am Coll Surg. 2010;210(5):788Y794, 794Y796. Davidson GH, Hamlat CA, Rivara FP, Koepsell TD, Jurkovich GJ, Arbabi S. Long-term survival of adult trauma patients. JAMA. 2011;305(10): 1001Y1007. Mosenthal AC, Weissmann DE, Curtis JR, Hays RM, Lustbader DR, Mulkerin C, Puntillo KA, Ray DE, Bassett R, Boss RD, et al. Integrating palliative care in the surgical and trauma intensive care unit: a report from the Improving Palliative Care in the Intensive Care Unit (IPAL-ICU) Project Advisory Board and the Center to Advance Palliative Care. Crit Care Med. 2012;40(4):11999Y11206. MacKenzie EJ, Rivara FP, Jurkovich GJ, Nathens AB, Frey KP, Egleston BL, Salkever DS, Scharfstein DO. A national evaluation of the effect of trauma-center care on mortality. N Engl J Med. 2006;354:366Y378. Mosenthal AC, Murphy PA, Barker LK, Lavery R, Retano A, Livingston DH. Changing the culture around end-of-life care in the trauma intensive care unit. J Trauma. 2008;64(6):1587Y1593. Elsayem A, Smith ML, Parmley L, Palmer JL, Jenkins R, Reddy S, Bruera E. Impact of a palliative care service on in-hospital mortality in a comprehensive cancer center. J Palliat Med. 2006;9(4):894Y902. Norton SA, Hogan LA, Holloway RG, Temkin-Greener H, Buckley MJ, Quill TE. Proactive palliative care in the medical intensive care unit: effects on length of stay for selected high-risk patients. Crit Care Med. 2007; 35(3):1530Y1535.
DISCUSSION Dr. Michael J. Sise (San Diego, California): Dr. Kozar’s well-presented study is accompanied by a well-written manuscript, which I have reviewed and have recommended for publication in the Journal of Trauma. It was helpful for the authors to share their thinking that led to this study in their manuscript: ‘‘We noted that trauma patients at our institution transferred to hospice care were not
* 2014 Lippincott Williams & Wilkins
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
639
J Trauma Acute Care Surg Volume 76, Number 3
Kozar et al.
being counted in our trauma registry as inpatient mortalities and, therefore, not undergoing our multidisciplinary peer review process.’’ This was a great introduction to their thinking, which, again, I think produced a very readable manuscript and obviously what we saw was a great presentation. This was not a small number of patients transferred to hospice at their center: 16 in 2010; 23 in 2011; and 51 in 2012, representing 8.5% to 18.5% of their trauma deaths. They further examined a variety of important administrative databases described in their presentation and they discovered the alarming discrepancies. When they examined the impact of these discrepancies across trauma centers that are compared to each other, comparative outcomes changed significantly. So this study addresses the important issue of often vexing limitations of administrative databases. The authors’ focus on post-discharge mortality identifies an important issue. If we are to truly have an inclusive trauma system, post-discharge death and disability outcomes are extremely important to include in any accounting of outcomes after injury. They are essential to the evaluation of the effectiveness of our practice guidelines and the organization of our clinical services. This is particularly true in tracking outcomes of our care of the injured elderly. Dr. Kozar and colleagues have appropriately identified disparities in calculating injury related to mortality between large and often quoted administrative databases. I congratulate the authors for identifying this problem and measuring the impact. We are left with some important questions. First, now that you have been swimming around in a variety of different data collection pools, tell us what you recommend for the practical elements of an administrative database to track outcomes. Specifically, what duration of survival is reasonable in order to count a patient as a survivor? And, secondly, what functional outcome variables in survivors should we also included in this tally? Although you discuss a variety of studies that have addressed these questions in your manuscript, what do you recommend, having become somewhat of a subject matter expert? And, finally, with what some have called the ‘‘silver tsunami’’ of rapidly increasing numbers of elderly, minimally injured trauma patients with severe comorbidities who we admit as trauma patients shortly before they die from their medical illnesses, how should we measure their outcomes? And how do we appropriately determine the quality of their care through our databases? I thank my colleagues of the AAST for the privilege of discussing the paper. Dr. Jay N. Collins (Norfolk, Virginia): Thank you for a good presentation. I think it is a very interesting topic as we all look at quality outcomes and how our mortality compares to other trauma centers. My question to you is how do you define hospice? Is this a failure to rescue? Is this a nonsurvivable injury? We have several patientsVolder patients with bad head injuriesVthat, after talking with the family and neurosurgeons, we all agree will have a terrible outcome. We often will withdraw care or make them comfort care only and they die 640
very soon after admission in our facility. Should this be counted as a death against us or as a failure to rescue? Along the same lines, in your study from Houston why does it matter where these patients die? If somebody has moved to hospice, that doesn’t count as a mortality, but if they actually die in your facility waiting to go to hospice, why do you count it against yourself? That seems contradictory as that could be related to many other factors beyond your control rather than your inability to rescue somebody. Dr. Steven Ross (Camden, New Jersey): In light of the ‘‘silver tsunami,’’ we see an increasing numberVand I’m sure everyone is seeing itVof patients who are in hospice and fall and suffer injuries and then are brought to the trauma center for care, contrary to their ‘‘do not hospitalize’’ orders. We return them fairly rapidly to their hospice, assuming they survive long enough for that. How would you count those in your mortality statistics? Dr. Karen J. Brasel (Milwaukee, Wisconsin): Rosemary, I wonder if you could comment on what your PI activity would trigger because this still is counting mortality as the important endpoint and I would argue that potentially those that survive but didn’t have a discussion about goals of care might more appropriately have been transferred to hospice and then captured in a mortality statistic. And I just wonder how you have reconciled that because you have a patient who survives but might have a devastating brain injury, and in another center, where goals-of-care discussions were had, the patient would have been transferred to hospice; in your scheme, this would then trigger PI, but if they did not have that goals-of-care discussion, it would not trigger anything in your current approach. I just wonder how you reconcile that and if you could comment on that. Dr. Rosemary A. Kozar (Houston, Texas): Thank you everyone for the questions and Dr. Sise, in particular. In terms of the questions by Dr. Sise, you asked me a difficult question, ‘‘What is the optimal time?’’ As you all know, it is not easy to get discharge data on trauma patients after they leave our trauma centers. We can say that we should get discharge data on hospice patients but that raises the question of follow up on other patient populations that we know have a high post-discharge mortality. Many of these patients, such as patients discharged to long-term acute care facilities, we know die although we don’t have good data on mortality. So it’s actually a really complicated question and not an easy one to answer. It is also very resource-intensive to suggest that trauma centers obtain post-discharge mortality data. When we tried to find out what happened to our patients discharged to outside hospices, it was not easy. We knew what institutions they were transferred to and still could not get complete data. Having said all that, however, I think if given the optimal choice, I like the definition of Medicare and Medicaid, deaths within 30 days. I think there is some data in the literature, at least from Eileen Bulger’s ROC study, looking at 30-day mortality, and it is fairly close to discharge figures. In terms of what functional outcomes, I think we first need a little more data. I don’t know if our center is unique in the fact that we send patients to outside hospices. * 2014 Lippincott Williams & Wilkins
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
J Trauma Acute Care Surg Volume 76, Number 3
I think at our center we have a lot of head injury patients going to hospice but there may be other centers that have more cancer patients, more elderly specific patients, etc, so I think in terms of the functional outcomes we probably need to get some more information on who these patients actually are. What would my specific recommendation be? I think as a first step for all of us, if we have patients under the hospice care at our institutions and they die at our institutions, we have got to say they are dead, look at them like deaths, and their care should undergo multidisciplinary peer review. I think that’s at least a first step. Dr. Collins, your question was how do you define hospice? The typical definition is that a patient should have less than six months to live. Does it matter where you die? It shouldn’t matter where you die. But if hospice patients are considered discharged alive, even though they are dying in our institutions, to me that just sounds like a huge discrepancy that we really shouldn’t tolerate at our institutions. Dr. Ross, I’m not sure how to answer your question. We sort of have a similar situation in that we’re across the street from M.D. Anderson and it’s not unusual to get a patient going to their cancer follow-up visit with Stage IV cancer and fall on the way then show up at our institution. If you’re sending them back to a hospice that they were at previously, I think that patient could be considered as discharged alive.
Kozar et al.
All I could say is right now I think as a first step we need to think about our own institution and then as an organization we have to really take a step back and think about this. It’s a pretty complicated issue. It’s got implications for verification visits as well. And, finally, Dr. Brasel’s question, you’re correct. I think it really depends on your institution. Some places will have a patient with a devastating head injury, withdraw care, and within a couple of days they are dead. Other places will transfer these patients to LTACs, SNFs and so forth, and they are never considered a death. And you are right. There isn’t an easy answer to that. The only thing I can say is that I think we need to think very carefully about our risk-adjusted calculations. Specific to the hospice patients, Dr. Nathens and I actually talked quite a bit about the different comorbidities we included in our analysis. TQIP uses a long list and a number of them would have implications for the elderly. Maybe, however, we need to come up with some elderly-specific criteria that we want to look at. Should we be recording frailty indexes, sarcopenia, dementia? Factors more specific to the elderly? Until we have those factors sorted out I don’t have a great answer for your question. I think for now we have to all think about what we want to look at and how we want to look at it and we should at least adjust our outcome measures and comorbidities in our calculations. I thank everyone.
* 2014 Lippincott Williams & Wilkins
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
641
Are all deaths recorded equally? The impact of hospice care on risk-adjusted mortality Rosemary A. Kozar, MD, PhD, John B. Holcomb, MD, Wei Xiong, MSc, and Avery B. Nathens, MD, PhD, Houston, Texas
BACKGROUND: Hospice care provides dignity and comfort at the end of life. While patients transferred to hospice die, they are often not recorded as inhospital deaths in trauma registries or in some administrative discharge data. Mortality rates for the purpose of database research, performance improvement, or public reporting may therefore be artificially low. The current study sought to determine the impact of discharges to hospice on risk-adjusted mortality for trauma deaths reported to the Trauma Quality Improvement Program. METHODS: Performance from Trauma Quality Improvement Program centers in 2011 was evaluated using risk-adjusted mortality with observed-toexpected mortality ratios derived from a logistic regression model. The impact of discharge to hospice on performance was measured by determining changes in performance if hospice cases were treated as survivors rather than deaths. Differences between groups were compared by nonparametric Wilcoxon rank-sum test. RESULTS: From the 167 centers with 126,259 injured patients, there were 8,862 deaths: 746 (8.4%) were discharged to a hospice, and the remainder was counted as in-hospital deaths. Overall, 106 centers (63.5%) reported at least one discharge to hospice, with the proportion of deaths ranging from 1.6% to 57%. Logistic regression demonstrated that age greater than 70 years (odds ratio [OR], 4.3; 95% confidence interval [CI], 3.5Y5.1), male sex (OR, 0.7; 95% CI, 0.6Y0.8), nonblack race (OR, 1.9; 95% CI, 1.3Y2.7), noncommercial insurance (OR, 1.4; 95% CI, 1.1Y1.7), and comorbidity counts greater than 2 (OR, 1.3; 95% CI, 1.1Y1.6) were associated with hospice care. If patients transferred to a hospice were treated as survivors in the estimation of risk-adjusted mortality, 34 centers (20%) would have a change in status. Changes would be in both directions for average-performing centers, while high-performing centers would seem worse and poor-performing centers would seem better. For centers that reported hospice deaths, the relative risk-adjusted mortality decreased by 8.8% for every 10% increase in the proportion of deaths recorded as discharged to a hospice. CONCLUSION: Given the large variation in the proportion of deaths recorded as discharged to a hospice rather than as in-hospital deaths, there is the potential for significant distortion of actual performance. Failure to consider this potential may misguide efforts directing performance improvement, research, and national reporting. Discharges to a hospice should be included with in-hospital deaths when reporting risk-adjusted mortality. (J Trauma Acute Care Surg. 2014;76: 634Y641. Copyright * 2014 by Lippincott Williams & Wilkins) KEY WORDS: Risk-adjusted mortality; trauma benchmarking; hospice care; public reporting; Trauma Quality Improvement Performance (TQIP).
P
alliative care is a philosophy of care that focuses on enhancing patient comfort and quality of life. It is becoming increasingly sought in the care of critically ill patients and is recognized as an important aspect of care for the surgical patient by the American College of Surgeons.1,2 Hospice care is one type of palliative care that is appropriate for patients during their last 6 months of life.3 Hospice care can provide dignity and comfort at the end of life. However, we noted that trauma patients at our institution (Memorial Hermann Hospital; R.A.K. and J.B.H.) transferred to a hospice care were not being counted in our trauma registry
Submitted: September 13, 2013, Revised: November 11, 2013, Accepted: November 12, 2013. From the Department of Surgery (R.A.K., J.B.H.), and Center for Translational Injury Research (J.B.H.), The University of Texas Health Sciences Center at Houston, Houston, Texas; and Department of Surgery (W.X., A.B.N.), Sunnybrook Health Sciences Centre, Toronto, Canada. This study was presented at the 72nd annual meeting of the American Association of Surgery for Trauma, September 18Y21, 2013, in San Francisco, California. Address for reprints: Rosemary Kozar, MD, PhD, 6431 Fannin, MSB 4.284; email: [email protected]. DOI: 10.1097/TA.0000000000000130
634
as inpatient mortalities and therefore not undergoing our multidisciplinary peer review process for mortality. We had 16 hospice transfers in 2010, 23 in 2011, and 51 in 2012, representing 8.5%, 11.8%, and 18.5% of all deaths, respectively. Although we use the American College of Surgeons’ Trauma Quality Improvement Performance (TQIP) for trauma-specific benchmarking,4 our hospital, Memorial Hermann Hospital, also relies on University Health Care Consortium (UHC), an alliance of academic medical centers that provides comparative data from their large administration discharge database and now includes trauma, for quality comparisons.5 Both UHC and TQIP use risk adjustment calculations in the determination of center performance. TQIP considers patients transferred to a hospice care as deaths, while UHC considers these patients as discharged alive. We became concerned that quality improvement measures may be jeopardized and that benchmarking may be compromised with variable treatment of hospice transfers. We therefore queried other large databases to determine their means of handling hospice patients. Both TQIP and National Surgical Quality Improvement Program consider patients discharged to a hospice as deaths. Medicare and Medicaid consider hospice admissions as deaths if death occurs within 30 days from transfer.6 At the time we initiated this study, UHC and the National Trauma Data Bank both J Trauma Acute Care Surg Volume 76, Number 3
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
J Trauma Acute Care Surg Volume 76, Number 3
considered transfers to a hospice care as a survivor to discharge. We hypothesized that if trauma patients transferred to a hospice care were reported as discharged alive, mortality rates for the purpose of performance improvement, benchmarking, registrybased research, or public reporting would be artificially low. The purpose of the current study was therefore to determine the impact of discharges to a hospice care on risk-adjusted mortality for trauma deaths reported to TQIP.
PATIENTS AND METHODS Centers and Subjects Data were derived from TQIP participating centers in 2011 and included all TQIP eligible patients (age 9 18 years with at least one body region with an Abbreviated Injury Scale [AIS] score Q 3). TQIP centers are limited to Level 1 or 2 centers that were verified by either the American College of Surgeons or regional authorities. TQIP excludes patients with no signs of life on arrival, those with a preexisting advanced directive to withhold life-sustaining care, and elderly patients (age Q 65 years) with an isolated hip fracture as a result of a fall. We evaluated the entire TQIP population and then separately analyzed elderly patients (age Q 65 years) and patients with isolated severe traumatic brain injury (TBI) defined as those with a head AIS score of 4 or greater or head AIS score of 3 or greater with initial motor Glasgow Coma Scale (GCS) score of 4 or lower and other body regions AIS score of 2 or lower.
Statistical Analyses The primary objective was to determine if hospital’s riskadjusted mortality relative to peer centers would be affected if patients discharged to a hospice were considered to be survivors rather than deaths. Mortality risk adjustment within TQIP has been well described7 but, briefly, is based on a logistic regression model with age, sex, injury mechanism, transfer status, systolic blood pressure, GCS motor, AIS score by body region, and comorbidities. Injury severity is included in the model using an DRG International Classification of DiseasesV9th Rev.Ybased approach (ICISS), which captures every injury in the DRG International Classification of DiseasesV9th Rev. injury code range.8 The probability of mortality for each patient was derived from the logistic regression model. Probabilities were summed for each hospital to estimate the hospital’s risk-adjusted mortality. For each hospital, the observed number of events (O) was divided by the risk-adjusted expected number of events (E) to produce an O/E ratio. If the lower limit of the 95% confidence interval (CI) of the O/E ratio was greater than unity, then the center was considered to be a poor performer (higher than expected mortality). By contrast, if the upper limit of the 95% CI of the O/E ratio was lower than unity, the center was considered to be a good performer. All centers with a 95% CI for their O/E ratio that included unity were considered to be average performers. Missing data for blood pressure (1.2%), GCS (3.1%), and heart rate (1.9%) were imputed using multiple imputation.9 Impact of discharge to a hospice care on center performance was measured by determining the changes in the performance if hospice cases were treated as survivors rather than deaths. Differences between groups (hospice vs. nonhospice deaths) were compared by nonparametric Wilcoxon rank-sum test for skewed
Kozar et al.
continuous data. Odds ratios are presented with their 95% CIs to demonstrate the strength of association between specific patient characteristics and hospice discharge.
RESULTS Data were submitted to TQIP by 167 centers on 126,259 injured patients. Baseline patient characteristics that were used in models of risk-adjusted mortality are shown in Table 1. Patients were most often white and male, sustained a fall, were not transferred, had noncommercial insurance, and were severely injured. There were a total of 8,862 deaths, of which 746 (8.4%) were discharged to a hospice and 8,116 (91.6%) were captured as in-hospital deaths. Overall, 106 centers (63.5%) reported at least one discharge to a hospice care. Across centers, the proportion of deaths recorded as discharged to a hospice ranged from 0% to 57% (Fig. 1), with a mean (SD) of 8.7% (11%) deaths as hospice. There were 61 centers that reported zero patients transferred to a hospice. Of the 746 hospice patients, time from hospital admission to transfer to a hospice care was 9.7 (11.4) days. When comparing hospice patients with in-hospital deaths, univariate analysis revealed that patients transferred to hospice care were older (77.1 [16.2] years vs. 55.9 [23.6] years), had lower Injury Severity Score (ISS) (19.2 [8.4] vs. 28.3 [13.2]), and a lower head AIS score (3.1 [1.9] vs. 3.3 [2.1]) compared with patients recorded as in-hospital deaths. After controlling for age, sex, race, payment status, and comorbidities, logistic regression demonstrated that older age, female sex, nonblack race, noncommercial insurance, and a higher burden of comorbidities were all significantly associated with hospice care (Table 2). If patients transferred to a hospice care were treated as survivors, crude mortality at centers that reported hospice discharges would decrease from 7.0% to 6.4%, a relative reduction of more than 8%. For every 10% increase in the proportion of deaths a center recorded as discharged to a hospice, there was an 8.8% relative reduction in their risk-adjusted mortality (Fig. 2). In addition, if transfers to hospice were treated as survivors for the estimation of risk-adjusted mortality, 34 centers (20.4%) demonstrated a significant change in their performance status (Table 3). Changes were in both directions for average-performing centers, while high-performing centers seemed worse and poorperforming centers seemed better. This same pattern of change held true when patients with TBI or elderly patients were analyzed separately (Table 3).
DISCUSSION In the current study, we sought to determine the impact of discharges to a hospice care on risk-adjusted mortality for trauma deaths reported to TQIP. Hospice transfers occurred in the majority of participating centers, and older age, female sex, nonblack race, noncommercial insurance, and comorbidities were all significantly associated with hospice care. Importantly, if patients transferred to a hospice care were treated as survivors, both crude and risk-adjusted mortality rates would decrease, with associated changes in performance status for some centers.
* 2014 Lippincott Williams & Wilkins
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
635
J Trauma Acute Care Surg Volume 76, Number 3
Kozar et al.
TABLE 1. Demographics for all Patients Cohort (n = 126,259) Male, % Race, % Asian Black White Other Missing Transfer, yes, % Mechanism, % Cut/pierce Firearm Fall Motorcyclist Motor vehicle collision Pedestrian Struck Other Insurance status, % Commercial Noncommercial Other Age, mean (SD), y Emergency department systolic blood pressure, mean (SD) Pulse, mean (SD) ISS, mean (SD) GCS motor, mean (SD) Head AIS score, mean (SD) Face AIS score, mean (SD) Neck AIS score, mean (SD) Chest AIS score, mean (SD) Abdomen AIS score, mean (SD) Spine AIS score, mean (SD) Comorbidity, % Heart disease Cancer Liver disease Alcoholism Smoking Stroke Diabetes Hypertension Renal disease Sensorium Obesity Respiratory disease Functional dependence Bleeding disorder Peripheral vascular disease Steroid use
All Patients
Hospice Deaths (n = 746)
In-hospital Deaths (n = 8,116)
p (Hospice vs. In-hospital)
66.4
53.2
70.7
G0.0001 G0.0001
1.6 14.5 71.4 9.7 2.8 31.6
1.1 4.8 87.9 4.3 1.9 37.3
2.0 15.0 70.5 8.9 3.7 27.2
3.2 6.6 38.3 7.7 24.8 6.9 6.2 6.5
0.4 2.7 80.4 1.9 9.0 2.5 2.0 1.1
1.4 17.9 37.2 6.6 22.2 8.8 2.6 3.4
37.2 53.7 9.1 50.1 (22.2) 138 (28.6)
16.8 78.6 4.7 77.1 (16.2) 147 (32.2)
27.5 61.5 11.0 55.6 (23.6) 128 (44.0)
89 (20.9) 17 (9.0) 5.4 (1.5) 1.7 (1.9) 0.4 (0.7) 0.05 (0.3) 1.2 (1.6) 0.5 (1.1) 0.6 (1.2)
88.3 (21.0) 19.2 (8.4) 4.8 (1.9) 3.1 (1.9) 0.4 (0.7) 0.03 (0.26) 0.6 (1.3) 0.1 (0.6) 0.5 (1.1)
94.2 (30.5) 28.3 (13.7) 2.9 (2.2) 3.3 (2.1) 0.5 (0.8) 0.1 (0.5) 1.5 (1.8) 0.7 (1.4) 0.7 (1.4)
4.8 0.9 0.6 9.6 15.7 2.3 11.3 28.5 0.7 7.5 3.1 6.9 1.4 5.4 0.2 0.4
17.0 3.8 2.0 5.1 5.5 8.4 20.0 53.6 1.7 20.8 2.4 11.5 7.0 16.9 0.4 1.5
8.0 1.9 1.4 6.7 5.2 3.6 12.8 27.7 1.4 6.6 3.0 6.4 1.7 9.2 0.3 0.3
G0.0001
With the heightened emphasis on public reporting, there is the potential for significant distortion of actual performance, particularly for high-performing centers. As we showed, 34.8% (8 of 23) of high-performing trauma centers would drop to 636
G0.0001 G0.0001
G0.0001 G0.0001 G0.0001 G0.0001 G0.0001 G0.0001 G0.0001 G0.0001 G0.0001 G0.0001 0.0001 G0.0001 0.0006 0.1859 0.0940 0.7389 G0.0001 G0.0001 G0.0001 0.4573 G0.0001 0.3782 G0.0001 G0.0001 G0.0001 0.5183 G0.0001
average performers if hospice transfers were not considered deaths. Centers for Medicare and Medicaid Services currently reports mortality measures for acute myocardial infarction, heart failure, and pneumonia patients,10 but this is expected to * 2014 Lippincott Williams & Wilkins
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
J Trauma Acute Care Surg Volume 76, Number 3
Kozar et al.
Figure 1. Variation of hospice care among 2011 TQIP centers.
increase in scope. There are commercial agencies such as US News & World Report, HealthGrade, and Thomson-Reuters as well as health care consortia, such as UHC and Leapfrog, that now report mortality statistics available to the public.11 How these agencies take into account hospice care varies widely. For example, Centers for Medicare and Medicaid Services reports on 30-day mortality. Thus, patients transferred to a hospice care who died during the index admission or within 30 days would be considered a death unless the patient was already enrolled in Medicare’s Hospice Benefit plan or if death occurred within the first day of hospital admission. In contrast, Thomson-Reuters considers patients discharged to a hospice care during the index admission as survivors11 as does UHC. Before beginning this project, the National Trauma Data Bank was excluding hospice patients in its death analysis but, as a result of this work, have since modified its practice and now report hospice transfers as deaths beginning with the 2012 report, similar to TQIP and National Surgical Quality Improvement Program. We have been in contact with UHC, and they are investigating the possibility of changing their practice at least for trauma patients, although this has not yet taken effect. Performance improvement remains an essential component of trauma center verification.12 Whether intentional, deaths may not undergo peer review if hospice patients are considered survivors. We (R.A.K. and J.B.H.) learned that this was happening at our own institution for patients who were not on the trauma service. We are now reviewing all in-hospital hospice transfers as deaths. Multidisciplinary peer review is paramount to effective quality improvement, and missed opportunities for improvement could occur if these patients were not subject to the same rigorous reviews as all inpatient deaths. The study by Franklin et al.,12 however, suggested that there were few missed opportunities for improvement when they looked at trauma patients who had withdrawal of care during the index admission. One potential disadvantage of considering patients transferred to a hospice care as deaths is that the inclusion of these patients leads to a reported increase in crude and risk-adjusted mortality. This same concern has been expressed for patients who are given do-not-resuscitate status, have advanced
directives, or undergo withdrawal of care while in the intensive care unit (ICU).13 As shown by others, most of these patients are elderly with minimal traumatic injuries.12Y15 Ideally, risk adjustment models should account for preexisting conditions and comorbidities. Thompson et al.16 have reported a comorbidity index for use in risk adjustment for injured patients; however, this may still not capture all needed fields in the minimally injured elderly patient.12 More importantly, no risk adjustment model can take into consideration the wants and needs of individuals with respect to end-of-life decisions, which are highly dependent on local culture and the sociocultural context of providers and their patients.17Y19 As such, it will be extraordinarily challenging to ‘‘adjust’’ away the effects for decisions about end-of-life care, which often happen in unique and personal discussions between providers, their patients, and/or surrogate decision makers. A limitation of the current study is that TQIP considers patients transferred to a hospice care as deaths, although death is not verified by participating centers. When we reviewed the 51 patients transferred to a hospice care at the University of Texas-Houston, 20 of our patients remained in the same hospital unit but had a virtual transfer of status, and all died in the hospital within 24 hours. We were surprised to learn that the other 31 patients were transferred to outside hospices. Follow-up data were available in 22 of these patients, and all but one patient died in the hospice facility during the index TABLE 2. Independent Risk Factors for Discharge to Hospice Among All Deaths Risk Factor Age 9 70 y Sex, male Race, nonblack Payment, noncommercial insurance Comorbidity counts 9 2
Adjusted Odds Ratio (95% CI)
p
4.3 (3.5Y5.1) 0.7 (0.6Y0.8) 1.9 (1.3Y2.7) 1.4 (1.1Y1.7)
G0.0001 G0.0001 0.0006 G0.0001
1.3 (1.1Y1.6)
0.0060
Logistic model receiver operating characteristic curve = 0.740 and Hosmer-Lemeshow calibration p = 0.2102
* 2014 Lippincott Williams & Wilkins
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
637
J Trauma Acute Care Surg Volume 76, Number 3
Kozar et al.
Figure 2. Increased reporting of discharges to a hospice decreased O/E mortality ratios for 2011 TQIP centers. Every 10% increase of hospice percentage decreases relative risk of O/E ratio by 8.8%.
admission. The average time from transfer to death was 17 days but ranged between 1 day and 180 days. Whether our data are representative of participating centers is unknown and raises an important question, when should we consider hospice patients transferred out of our trauma centers as deaths? Should all deaths be recorded equally? The time and effort to track these patients after discharge were substantial, and some patients did not die for several months. As an initial step in addressing this problem, we suggest that all hospice transfers that occur within the same facility as the index admission be considered as deaths and undergo mortality review. Similar to the results identified with transfers to a hospice care, Kelly et al.20 reported that a liberal policy of transfers to long-term acute care facilities based on a discharge protocol might lead to reductions in reported hospital mortality. A recent study by Brinkman et al.21 examined the influence of using mortality 1, 3, or 6 months after ICU admission instead of inhospital mortality on their standardized mortality ratio. They used a large Dutch database from almost 81,000 ICU patients in mixed medicine-surgical units and found that mortality ratios and rank differed depending on whether in-hospital mortality or mortality at 1, 3, or 6 months after ICU admission was used. As differences in discharge policies and mortality at time points after ICU discharge affected reported mortality, they suggested that later time points should be used as a quality indicator for benchmarking purposes. This issue has also been examined in trauma patients. Bulger et al.22 examined inhospital and 28-day mortality in severely injured patients who received out-of-hospital hypertonic saline and found less than a 1% difference in mortality. Others have evaluated later time points and have shown that trauma-related deaths continue for at least the first year after injury.22Y26 At the current time, there are no data in TQIP on the time to death of trauma patients once they leave the trauma center. We also do not have information on the reason for transfer to a hospice or the do-notresuscitate status of included patients. However, TQIP is 638
now capturing withdrawal of care, which might provide some information on different approaches to end-of-life care across participating centers. Lastly, we do not know the extent to which palliative care teams are involved with the care of injured patients that do not involve hospice transfer. Since death was not imminent on discharge to a hospice, perhaps, we should encourage the more frequent use of palliative care. It is being increasingly recognized as beneficial, and models to integrate palliative care into surgical and ICUs have been described.27 These all represent opportunities to enhance end-of-life care and should be without the fear of adversely effecting performance status. There are in fact a number of reports showing that early involvement of the palliative care team may shorten hospital or ICU length of stay without impacting mortality.28Y30 In conclusion, public reporting of risk-adjusted mortality is increasing. Given the large variation in proportion of deaths TABLE 3. Change in Center Outlier Status When Patients Transferred to a Hospice Were Coded as Survivors Center Performance All patients Average Good Poor TBI patients Average Good Poor Elderly patients Average Good Poor
Total No. No. Changed Centers to Good 120 23 24
7
143 11 13
10
135 12 20
7
No. Changed to Average
No. Changed to Poor 10
8 9 8 8 4 9 6 11
* 2014 Lippincott Williams & Wilkins
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
J Trauma Acute Care Surg Volume 76, Number 3
Kozar et al.
recorded as discharged to a hospice rather than as an inpatient death, there is the potential for distortion of actual performance. Failure to consider this potential may also misguide efforts directing performance improvement activities, registry-based research, and national reporting. Discharges to hospice should be included with in-hospital deaths when reporting risk-adjusted mortality, although better methods to capture preexisting conditions and preferences at the end of life are needed to accurately risk-adjust for mortality in these patients.
14.
15.
16.
17. AUTHORSHIP R.A.K. conducted the literature search for this study, which R.A.K., J.B.H., and A.B.N. designed. W.X. collected the data and performed statistical analysis. R.A.K., W.X., and A.B.N. contributed to data analysis. R.A.K., W.X., and A.B.N. wrote the manuscript. R.A.K., J.B.H., and A.B.N. contributed to critical revision.
18.
19.
20. ACKNOWLEDGMENT We thank Sheila Lopez, RN, BSA, MA, and Edmundo Dipasupil, CSTR, for their assistance with our (R.A.K., J.B.H.) trauma center data.
DISCLOSURE The authors declare no conflicts of interest.
REFERENCES 1. Surgical Palliative Care: A resident’s guide. Dunn G, Martensen R, Weissman D, Bradley CT, Brasel KJ, Castillo EH, Cole TR, Deville K, Dixon P, Hinshaw DB et al. Available at: http://www.facs.org/palliativecare/ surgicalpalliativecareresidents.pdf. Accessed August 7, 2013. 2. The surgeon and palliative care. Available at: http://www.facs.org/ palliativecare/index.html. Accessed August 7, 2013. 3. Hui D, De La Cruz M, Mori M, Parsons HA, Kwon JH, Torres-Vigil I, Kim SH, Dey R, Hutchins R, Liem C, et al. Concepts and definitions ‘‘supportive care,’’ best supportive care, palliative care, and hospice care in the published literature, dictionaries, and textbooks. Support Care Cancer. 2013;21:659Y685. 4. Trauma quality improvement program (TQIP). Available at: http:// www.facs.org/trauma/ntdb/tqip.html. Accessed August 4, 2013. 5. About University Healthcare Consortium. Available at: https://www.uhc. edu/12443.htm. Accessed August 7, 2013. 6. Centers for Medicare and Medicaid. Outcome measures. Available at: http:// www.cms.gov/Medicare/Quality-Initiatives-Patient-Assessment-Instruments/ HospitalQualityInits/OutcomeMeasures.html. Accessed August 7, 2013. 7. Newgard CD, Fildes JJ, Wu L, Hemmila MR, Burd RS, Neal M, Mann NC, Shafi S, Clark DE, Goble S, et al. Methodology and analytic rationale for the American College of Surgeons Trauma Quality Improvement Program. J Am Coll Surg. 2013;216(1):147Y157. 8. Osler T, Rutledge R, Deis J, Bedrick E. ICISS: an International Classification Of Disease-9 based Injury Severity Score. J Trauma. 1996;41: 380Y386. 9. Rubin DB. Inference and missing Data. Biometrika. 1976;63:581Y592. 10. Mortality measures overview. Publicly reporting risk-standardized, 30-day mortality measures for AMI, HF, and PN. Available at: http://www.qualitynet. org. Accessed August 21, 2013. 11. Cassel JB, Jones AB, Meier DE, Smith TJ, Spragens LH, Weissman D. Hospital mortality rates: how is palliative care taken into account? J Pain Symptom Manage. 2010;40:914Y925. 12. Franklin GA, Cannon RW, Smith JW, Harbrecht BG, Miller FB, Richardson JD. Impact of withdrawal of care and futile care on trauma mortality. Surgery. 2011;150:854Y860. 13. Adams SD, Cotton BA, Wade CE, Kozar RA, Dipasupil E, Podbielski JM, Gill BS, Duke JR, Adams PR, Holcomb JB. Do not resuscitate status, not
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
age, affects outcomes after injury: an evaluation of 15,227 consecutive trauma patients. J Trauma Acute Care Surg. 2013;74(5):1327Y1330. Sise MJ, Sise CB, Thorndike JF, Kahl JE, Calvo RY, Shackford SR. Withdrawal of care: a 10-year experience at a Level I trauma center. J Trauma Acute Care Surg. 2012;72:1186Y1193. Kahl JE, Calvo RY, Sise MJ, Sise CB, Thorndike JF, Shackford SR. The changing nature of death on the trauma service. J Trauma Acute Care Surg. 2013;75(2):195Y201. Thompson HJ, Rivara FP, Nathens A, et al. Development and validation of the mortality risk for trauma comorbidity index. Ann Surg. 2010;252: 370Y375. Hanchate A, Kronman AC, Young-Xu Y, Ash AS, Emanuel E. Racial and ethnic differences in end-of-life costs: why do minorities cost more than whites? Arch Intern Med. 2009;169(5):493Y501. Adames HY, Chavez-Duen˜as NY, Fuentes MA, Salas SP, Perez-Chavez JG. Integration of Latino/a cultural values into palliative health care: a culture centered model. Palliat Support Care. 2013;17:1Y9. Periyakoil VS, Stevens M, Kraemer H. Multicultural long-term care nurses’ perceptions of factors influencing patient dignity at the end of life. J Am Geriatr Soc. 2013;61(3):440Y446. Kelly KB, Koeppel ML, Como JJ, Carter JW, McCoy AM, Claridge JA. Continued rationale of why hospital mortality is not an appropriate measure of trauma outcomes. Am J Surg. 2012;203(3):366Y369; discussion 369. Brinkman S, Abu-Hanna A, de Jonge E, de Keizer NF. Prediction of longterm mortality in ICU patients: model validation and assessing the effect of using in-hospital versus long-term mortality on benchmarking. Intensive Care Med. 2013;39:1925Y1931. Bulger EM, May S, Kerby JD, Emerson S, Stiell IG, Schreiber MA, Brasel KJ, Tisherman SA, Coimbra R, Rizoli S, et al.; ROC investigators. Out-ofhospital hypertonic resuscitation after traumatic hypovolemic shock: a randomized, placebo controlled trial. Ann Surg. 2011;253(3):431Y441. Probst C, Zelle BA, Sittaro NA, Lohse R, Krettek C, Pape HC. Late death after multiple severe trauma: when does it occur and what are the causes? J Trauma. 2009;66(4):1212Y1217. Claridge JA, Leukhardt WH, Golob JF, McCoy AM, Malangoni MA. Moving beyond traditional measurement of mortality after injury: evaluation of risks for late death. J Am Coll Surg. 2010;210(5):788Y794, 794Y796. Davidson GH, Hamlat CA, Rivara FP, Koepsell TD, Jurkovich GJ, Arbabi S. Long-term survival of adult trauma patients. JAMA. 2011;305(10): 1001Y1007. Mosenthal AC, Weissmann DE, Curtis JR, Hays RM, Lustbader DR, Mulkerin C, Puntillo KA, Ray DE, Bassett R, Boss RD, et al. Integrating palliative care in the surgical and trauma intensive care unit: a report from the Improving Palliative Care in the Intensive Care Unit (IPAL-ICU) Project Advisory Board and the Center to Advance Palliative Care. Crit Care Med. 2012;40(4):11999Y11206. MacKenzie EJ, Rivara FP, Jurkovich GJ, Nathens AB, Frey KP, Egleston BL, Salkever DS, Scharfstein DO. A national evaluation of the effect of trauma-center care on mortality. N Engl J Med. 2006;354:366Y378. Mosenthal AC, Murphy PA, Barker LK, Lavery R, Retano A, Livingston DH. Changing the culture around end-of-life care in the trauma intensive care unit. J Trauma. 2008;64(6):1587Y1593. Elsayem A, Smith ML, Parmley L, Palmer JL, Jenkins R, Reddy S, Bruera E. Impact of a palliative care service on in-hospital mortality in a comprehensive cancer center. J Palliat Med. 2006;9(4):894Y902. Norton SA, Hogan LA, Holloway RG, Temkin-Greener H, Buckley MJ, Quill TE. Proactive palliative care in the medical intensive care unit: effects on length of stay for selected high-risk patients. Crit Care Med. 2007; 35(3):1530Y1535.
DISCUSSION Dr. Michael J. Sise (San Diego, California): Dr. Kozar’s well-presented study is accompanied by a well-written manuscript, which I have reviewed and have recommended for publication in the Journal of Trauma. It was helpful for the authors to share their thinking that led to this study in their manuscript: ‘‘We noted that trauma patients at our institution transferred to hospice care were not
* 2014 Lippincott Williams & Wilkins
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
639
J Trauma Acute Care Surg Volume 76, Number 3
Kozar et al.
being counted in our trauma registry as inpatient mortalities and, therefore, not undergoing our multidisciplinary peer review process.’’ This was a great introduction to their thinking, which, again, I think produced a very readable manuscript and obviously what we saw was a great presentation. This was not a small number of patients transferred to hospice at their center: 16 in 2010; 23 in 2011; and 51 in 2012, representing 8.5% to 18.5% of their trauma deaths. They further examined a variety of important administrative databases described in their presentation and they discovered the alarming discrepancies. When they examined the impact of these discrepancies across trauma centers that are compared to each other, comparative outcomes changed significantly. So this study addresses the important issue of often vexing limitations of administrative databases. The authors’ focus on post-discharge mortality identifies an important issue. If we are to truly have an inclusive trauma system, post-discharge death and disability outcomes are extremely important to include in any accounting of outcomes after injury. They are essential to the evaluation of the effectiveness of our practice guidelines and the organization of our clinical services. This is particularly true in tracking outcomes of our care of the injured elderly. Dr. Kozar and colleagues have appropriately identified disparities in calculating injury related to mortality between large and often quoted administrative databases. I congratulate the authors for identifying this problem and measuring the impact. We are left with some important questions. First, now that you have been swimming around in a variety of different data collection pools, tell us what you recommend for the practical elements of an administrative database to track outcomes. Specifically, what duration of survival is reasonable in order to count a patient as a survivor? And, secondly, what functional outcome variables in survivors should we also included in this tally? Although you discuss a variety of studies that have addressed these questions in your manuscript, what do you recommend, having become somewhat of a subject matter expert? And, finally, with what some have called the ‘‘silver tsunami’’ of rapidly increasing numbers of elderly, minimally injured trauma patients with severe comorbidities who we admit as trauma patients shortly before they die from their medical illnesses, how should we measure their outcomes? And how do we appropriately determine the quality of their care through our databases? I thank my colleagues of the AAST for the privilege of discussing the paper. Dr. Jay N. Collins (Norfolk, Virginia): Thank you for a good presentation. I think it is a very interesting topic as we all look at quality outcomes and how our mortality compares to other trauma centers. My question to you is how do you define hospice? Is this a failure to rescue? Is this a nonsurvivable injury? We have several patientsVolder patients with bad head injuriesVthat, after talking with the family and neurosurgeons, we all agree will have a terrible outcome. We often will withdraw care or make them comfort care only and they die 640
very soon after admission in our facility. Should this be counted as a death against us or as a failure to rescue? Along the same lines, in your study from Houston why does it matter where these patients die? If somebody has moved to hospice, that doesn’t count as a mortality, but if they actually die in your facility waiting to go to hospice, why do you count it against yourself? That seems contradictory as that could be related to many other factors beyond your control rather than your inability to rescue somebody. Dr. Steven Ross (Camden, New Jersey): In light of the ‘‘silver tsunami,’’ we see an increasing numberVand I’m sure everyone is seeing itVof patients who are in hospice and fall and suffer injuries and then are brought to the trauma center for care, contrary to their ‘‘do not hospitalize’’ orders. We return them fairly rapidly to their hospice, assuming they survive long enough for that. How would you count those in your mortality statistics? Dr. Karen J. Brasel (Milwaukee, Wisconsin): Rosemary, I wonder if you could comment on what your PI activity would trigger because this still is counting mortality as the important endpoint and I would argue that potentially those that survive but didn’t have a discussion about goals of care might more appropriately have been transferred to hospice and then captured in a mortality statistic. And I just wonder how you have reconciled that because you have a patient who survives but might have a devastating brain injury, and in another center, where goals-of-care discussions were had, the patient would have been transferred to hospice; in your scheme, this would then trigger PI, but if they did not have that goals-of-care discussion, it would not trigger anything in your current approach. I just wonder how you reconcile that and if you could comment on that. Dr. Rosemary A. Kozar (Houston, Texas): Thank you everyone for the questions and Dr. Sise, in particular. In terms of the questions by Dr. Sise, you asked me a difficult question, ‘‘What is the optimal time?’’ As you all know, it is not easy to get discharge data on trauma patients after they leave our trauma centers. We can say that we should get discharge data on hospice patients but that raises the question of follow up on other patient populations that we know have a high post-discharge mortality. Many of these patients, such as patients discharged to long-term acute care facilities, we know die although we don’t have good data on mortality. So it’s actually a really complicated question and not an easy one to answer. It is also very resource-intensive to suggest that trauma centers obtain post-discharge mortality data. When we tried to find out what happened to our patients discharged to outside hospices, it was not easy. We knew what institutions they were transferred to and still could not get complete data. Having said all that, however, I think if given the optimal choice, I like the definition of Medicare and Medicaid, deaths within 30 days. I think there is some data in the literature, at least from Eileen Bulger’s ROC study, looking at 30-day mortality, and it is fairly close to discharge figures. In terms of what functional outcomes, I think we first need a little more data. I don’t know if our center is unique in the fact that we send patients to outside hospices. * 2014 Lippincott Williams & Wilkins
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
J Trauma Acute Care Surg Volume 76, Number 3
I think at our center we have a lot of head injury patients going to hospice but there may be other centers that have more cancer patients, more elderly specific patients, etc, so I think in terms of the functional outcomes we probably need to get some more information on who these patients actually are. What would my specific recommendation be? I think as a first step for all of us, if we have patients under the hospice care at our institutions and they die at our institutions, we have got to say they are dead, look at them like deaths, and their care should undergo multidisciplinary peer review. I think that’s at least a first step. Dr. Collins, your question was how do you define hospice? The typical definition is that a patient should have less than six months to live. Does it matter where you die? It shouldn’t matter where you die. But if hospice patients are considered discharged alive, even though they are dying in our institutions, to me that just sounds like a huge discrepancy that we really shouldn’t tolerate at our institutions. Dr. Ross, I’m not sure how to answer your question. We sort of have a similar situation in that we’re across the street from M.D. Anderson and it’s not unusual to get a patient going to their cancer follow-up visit with Stage IV cancer and fall on the way then show up at our institution. If you’re sending them back to a hospice that they were at previously, I think that patient could be considered as discharged alive.
Kozar et al.
All I could say is right now I think as a first step we need to think about our own institution and then as an organization we have to really take a step back and think about this. It’s a pretty complicated issue. It’s got implications for verification visits as well. And, finally, Dr. Brasel’s question, you’re correct. I think it really depends on your institution. Some places will have a patient with a devastating head injury, withdraw care, and within a couple of days they are dead. Other places will transfer these patients to LTACs, SNFs and so forth, and they are never considered a death. And you are right. There isn’t an easy answer to that. The only thing I can say is that I think we need to think very carefully about our risk-adjusted calculations. Specific to the hospice patients, Dr. Nathens and I actually talked quite a bit about the different comorbidities we included in our analysis. TQIP uses a long list and a number of them would have implications for the elderly. Maybe, however, we need to come up with some elderly-specific criteria that we want to look at. Should we be recording frailty indexes, sarcopenia, dementia? Factors more specific to the elderly? Until we have those factors sorted out I don’t have a great answer for your question. I think for now we have to all think about what we want to look at and how we want to look at it and we should at least adjust our outcome measures and comorbidities in our calculations. I thank everyone.
* 2014 Lippincott Williams & Wilkins
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
641
