a b s t r a c t

Objectives: To explore the determinant factors and prognostic significance of emergency department do-not-re- suscitate (ED-DNR) orders for patients with spontaneous intracerebral hemorrhage (SICH).

Methods: Consecutive adult SICH patients treated in our ED from January 1, 2012 to December 31, 2016 were se- lected as the eligible cases from our hospital’s stroke database. Patients’ information was comprehensively reviewed from the database and medical and nursing charts. ED-DNR orders were defined as DNR orders written during ED stay. Multiple logistic regression analysis was used to identify significant determinants of ED-DNR or- ders. Thirty- and 90-day neurological outcomes were analyzed to test the prognosis impact of ED-DNR orders. Results: Among 835 enrolled patients, 112 (12.1%) had ED-DNR orders. Significant determinant factors of ED-DNR orders were age, ambulatory status before the event, brain computed tomography findings of Midline shift, intra-

ventricular extension, larger hematoma size, and ED arrival GCS <=8. Patients with and without ED-DNR orders

had a similar 30-day Death rate if they had the same initial ICH Score point. During 30 to 90 days, patients with ED-DNR orders had a significantly increased mortality rate. However, the rate of improvement in Neurological status between the two groups was not significantly different.

Conclusions: Older and sicker SICH patients had higher rate of ED-DNR orders. The mortality rates between pa- tients with and without ED-DNR orders for each ICH score point were not significantly different. During the 30-to-90-day follow-up, the rates of neurological improvement in both groups were similar.

(C) 2017

Introduction

Spontaneous intracerebral hemorrhage (SICH) is an acute devastat- ing event. Despite advanced medical or surgical care, half of patients die within one year, and most of the survivors suffer from significant long- term Neurological sequelae [1-3]. Given this poor prognosis, treating physicians may raise concerns about care aggressiveness, and some- times provide instead a palliative approach during an early stage in ac- cordance with the preference of the patient or family, in order to avoid exposing patients to unnecessary suffering [1,4-9]. Accordingly, previ- ous studies had reported that do-not-resuscitate (DNR) orders are com- mon following intracerebral hemorrhage (ICH) [5-7,9,11].

Many studies have defined early DNR orders as DNR orders written within 24 h [1-3] (or within 48 h in a; smaller number of studies) [4] after hospitalization. However, major decisions on care aggressiveness including endotracheal intubation, mechanical ventilation, and surgical

* Corresponding author at: Department of Emergency Medicine, Taipei-Veterans General Hospital, 201 Sec 2, Shih-Pai Rd., Taipei, Taiwan, ROC.

E-mail address: [email protected] (D.H.-T. Yen).

intervention, etc. are often made earlier in the emergency department (ED) in order to fulfill the patient’s or family’s expectations for care and for proper disposition [4,5,6], and therefore, some local routine pro- cedures even require the emergency physician to initiate the dis- cussion of DNR orders before admission in specific circumstances [4]. In addition, with increasing emphasis on the issue of patient autonomy and dying with dignity in the ED, more end-of-life care has been initiat- ed in the ED [5-9]. Emergency physicians have been requested to be sensitive to the patient’s or family’s preference and be familiar with pro- viding objective information upon limiting care [5-9]. As a result, a sub- stantial percentage of DNR orders for ISCH patients are written in the ED [1,2,4]. Although, many studies have delineated the factors related to early DNR orders and reported their impacts on prognoses [1-4,10], the associated factors and exact prognostic significance of DNR orders written during ED stay, ED-DNR orders, have not been previously eluci- dated. We conducted a retrospective analysis of a prospectively collect- ed database to elucidate factors associated with ED-DNR orders and their impact on prognosis, and we also tried to clarify the interaction of initial severity on the association between ED-DNR orders and prognoses.

http://dx.doi.org/10.1016/j.ajem.2017.06.016

0735-6757/(C) 2017

Methods

Participants

The current retrospective cohort study was conducted in the ED of Taipei General Veterans Hospital (TVGH), a 3000-bed tertiary referral center that treats more than 200 SICH cases each year in a protocol- based way [11-13], From January 1, 2012 to December 31, 2016, consec- utive adult SICH patients treated in our ED were initially identified using the hospital’s Stroke registry databank, which was a prospectively de- signed database having prognosis information of Modified Rankin scale (mRS) on hospital discharge, 30-day, and 90-day, set up in 2001 for the initial purpose of quality assurance [11]. To avoid missing poten- tial participants, charts were cross-checked for coding with Internation- al Classification of Diseases, Ninth Revision, Clinical Modification (ICD- 9-CM) 431 and 432.9 (for SICH). Patients with prior a DNR order, or ICH caused by trauma, tumor, cerebral infarction, Brain abscess, or bleeding tendency, or incomplete data collection were excluded. The study was approved by Institutional Review Board (IRB) of TVGH and informed consent was waived by the IRB.

Patient management

In our ED, the diagnosis and treatment plans for SICH patients basi- cally follows the suggestions of the latest guideline and were standard- ized [14,15]. ED personnel were asked to be familiar with the standardized procedures because the protocol was also the ED quality initiative, which aimed to treat stroke patients safely and effectively. At the beginning when there was a patient suspected of having stroke, the physician on duty and the laboratory and radiological technicians are notified immediately by phone in order to speed up the evaluation process. The triage and registration station receive the medical informa- tion including DNR status with National Health Insurance (NHI) card through a computer chart system. After SICH was diagnosed based on brain computed tomography (CT), a neurologist and neurosurgeon are consulted for further discussion of treatment strategies. The DNR order is discussed at that time and members of the DNR discussion in- clude the patient, the patient’s family or surrogates, ED chief nurse, emergency physician, and neurospecialists. If a DNR order is deter- mined, informed consent and a DNR paper are written immediately and then the information is inputted into the NHI card and electronic chart system of our hospital so that all the physicians and nurses can see it through the patient’s page in our electronic information system. In addition to a DNR order, written informed consent is also needed be- fore the performance of invasive procedures such as endotracheal intu- bation, central line placement, Foley tube insertion, and nasogastric tube insertion and is discussed at that time. Patients or their surrogates can obtain information regarding a DNR and the written paper at any time during the ED stay. Hypertension is controlled below 160/ 110 mm Hg using labetalol or perdipine. Patient prognosis (death or survival; good or bad neurological outcome) are determined by com- prehensively reviewing the medical records, and in some cases, by tele- phone consultation. poor neurological outcome among SICH survivors is defined as >= 3 on the mRS.

Definition of emergency department do-not-resuscitate (ED-DNR) order

The hospital definition of a DNR order is no resuscitation, i.e., endo- tracheal intubation, chest compression, or epinephrine injection in the case of cardiac arrest, without other limitations of care. ED-DNR orders are defined as DNR orders written during ED stay. In usual practice in our system, once a DNR order is written, the information (time, place, name of involved specialist, and name of surrogates) should be entered into the NHI card system through the electronic chart system of our hos- pital, and the details can be seen later when opening the patient’s page

in that system. Patients, family members, or treating physicians can ini- tiate a review of the DNR order and change the decision at any time dur- ing hospitalization. Information on DNR orders for the study cases was retrieved by review of the NHI card records and patients’ journal in the hospital, with a special focus on the records during the stay in the ED.

Data collection

The basic information for SICH patients such as age, sex, underlying disease, insurance status, religion, marriage status, onset time, ED arriv- al time, initial vital signs, Glasgow Coma Scale score, results of blood survey, and treatments in the ED were retrieved from stroke da- tabase, emergency management technician records, ED triage data, and medical and nursing records. A single neuroradiologist reviewed each patient’s first brain CT and recorded hematoma size (in milliliters using the ABC/2 method), location (supratentorial or infratentorial), presence or absence of midline shift, and intraventricular hemorrhage (IVH), and recorded the data in the electronic data base. Cases were ex- cluded if there was clear evidence of secondary ICH or if there were in- complete clinical data including absence of initial CT scan, not discharged from TVGH, or incomplete follow-up data.

Data analysis

The rates of the factors were compared between the two patient groups (patients with and without ED-DNR orders) using various tests including t-test (continuous data), chi-squared test (nominal data), and Mann-Whitney U test (ordinal data with skewed distribution) to assess any differences in baseline characteristics. Using SPSS, “Statistical Package for the Social Sciences version 19.0 (SPSS Statistics, Chicago, IL, USA),” the association between each variable and ED-DNR orders was analyzed. Factors showing significance in the univariate analysis were included in multiple logistic regression analysis. The 30-day mortality rates for patients with and without ED-DNR who had the same initial ICH score [16] (similar severity) were compared. The distribution of 30-day and 90-day mRS scores of patients with and without ED-DNR was assessed to find the difference in neurological change during this period.

Results

In total, 1023 cases were coded as eligible cases during the study pe- riod and 188 patients were excluded because of secondary ICH (n = 160) or incomplete data (n = 28). A total of 835 consecutive SICH pa- tients were included in the analysis. Of the 835 patients, 112 (13.4%) had a written DNR order during their ED stay and 181 (21.7%) died within 30 days. Two ED-DNR patients cancelled the DNR order during hospitalization (on the 18th day and 32nd day). Of those with an ED- DNR order, 54 (48.2%) patients had expressed their preferences for end-of-life care to family members before the event and 26 (23.2%) pa- tients had made Advance directives. The baseline characteristics of the study population are outlined in Table 1. An ED-DNR order was highly correlated with 1-week mortality, 30-day mortality, and 90-day poor neurological outcome. In patients with an ED-DNR order, more than half of the mortality cases died within 1 week. After adjusting covariates in multiple logistic regression analysis (Table 2), factors independently correlated with an ED-DNR decision were age (P = 0.024, odds ratio [OR] = 1.049, 95% confidence interval [CI] = 1.006-1.093 by 1-year in- crement in age), fully ambulatory before the event (P b 0.001, OR = 0.320, 95% CI = 0.166-0.595), midline shift (P b 0.001, OR = 9.299,

95% CI = 5.118-16.897), intraventricular extension (P b 0.023, OR = 2.533, 95% CI = 1.137-5.645), hematoma size (by 1-mL increment of hematoma, P b 0.001, OR = 1.038, 95% CI = 1.021-1.056), and ED ar- rival GCS <= 8 (P b 0.001, OR = 8.115, 95% CI = 4.515-14.583). In the

subgroup analysis, the 30-day mortality rates of patients with and

Table 1

Demographic and clinical characteristics of spontaneous intracerebral hemorrhage (ICH) patients with and without ED-DNR orders.

severe liver cirrhosis, and the mortality rate was not significantly higher than for those without an ED-DNR order. During the 30-day to 90-day period, patients with an ED-DNR order had a significantly increased

Patient characteristics Patients with ED-DNR order, n = 112

Patients without ED-DNR order, n = 723

P

value

mortality rate (Fig. 3). However, the rate of improvement in neurologi- cal status between two groups was not significantly different (Fig. 3).

Age 76.1 +- 8.1 71.2 +- 8.4 b0.001

Gender (female) 63 (56.3) 386 (53.4) 0.611

Insurance status 0.352

National Health Insurance only	89 (79.5)	600 (83.0)
With Medicaid	23 (20.5)	123 (17.0)
MarriAge SIngle	21 (18.8)	0.233 119 (16.5)
Married	39 (34.8)	300 (41.5)
Divorced	23 (20.5)	169 (23.4)
Widow or widower Religion	29 (25.9)	135 (18.7) 0.514
Taoism	23 (20.5)	170 (23.5)
Buddhism	34 (30.4)	224 (31.0)
Catholic/Christian	19 (17.0)	87 (12.0)
None/other	36 (32.1)	242 (33.5)
level of education			0.169
Higher than senior high school	76 (67.9)	536 (74.1)
Lower than senior high school	36 (32.1)	187 (25.9)
Referred from other hospital	25 (22.3)	153 (21.2)	0.804
nursing home residence	28 (25.0)	102 (14.1)	0.005
Number of chronic diseases >= 4	55 (49.1)	184 (25.4)	b0.001
alcohol consumption	13 (11.6)	47 (6.5)	0.073
Current smoker	19 (17.0)	89 (12.3)	0.174
Fully ambulatory before the event	66 (58.9)	600 (83.0)	b0.001
ED arrival GCS <= 8	74 (66.1)	132 (18.3)	b0.001
ED arrival time			0.416
0 am-8 am	7 (6.3)	69 (9.5)
8 am-16 pm	46 (41.1)	310 (42.9)
16 pm-0 am	59 (52.7)	(47.6)
ED stay time ? 4 h	21 (17.2)	146 (20.2)	0.524

elapsed time ? 6 h 25 (22.3) 133 (18.4) 0.364

ICH size (mL) 35.9 +- 16.3 24.1 +- 14.8 b0.001

Infratentorial ICH 25 (22.3) 114 (15.8) 0.101

Midline shift 68 (60.7) 89 (12.3) b0.001

Intraventricular extension 62 (55.4) 124 (17.2) b0.001 Prognoses

1-week mortality 31 (27.7) 48 (6.6) b0.001

30-day mortality⁺ 59 (52.7) 137 (18.9) b0.001

Discussion

This study is the first to investigate the associated factors and prog- nostic impact of ED-DNR orders in patients with spontaneous ICH, and it was found that 1) poor ICH outcome indicators were also the determi- nants of ED-DNR orders and 2) no significant difference on rate of neu- rological improvement was observed between patients with and without ED-DNR order during follow-up from 30 to 90 days.

In our study, 31.0% of the patients signed a DNR during hospitaliza- tion, which is similar to the percentage reported from studies conducted in North America and Europe [1,2,4,17]. However, the rates of ED-DNR orders in those studies could not be compared with our results because they were not clearly delineated in those studies. A study from China re- ported a very low DNR order rate because of cultural factors and inade- quate trust in the physician-patient relationship [3]. It is reasonable to infer that the ED-DNR order rate should be much lower in China. We found a significant higher rate of ED-DNR orders (13.4%) which might be the result of the huge difference in health care system and advocacy, and implementation of the Hospice palliative care Act across the strait [18-20]. Although ED-DNR order is a complex issue for patients and families, they made their decision in just 3-4 h during the ED stay. Since there was an extremely low cancellation rate (2 cases, 1.8%), we believe it was not a rushed decision for the patients and families to make with regard to ED-DNR orders in the ED. With the expanded cov- erage of the Hospice Palliative Care Act after it was amended in 2011 [19,21], patients and families concerned about this issue were encour- aged to make Advance directives [19,21,22]. This might have led to ini- tiation of a discussion on DNR orders which could be reflected in the relative high rate of patients preferring at end of life to be involved in the DNR decision-making process. As a result, family members could have continued to largely agree on this issue and make a decision

Modified Rankin Scale >= 4 at 90 days⁺

94 (83.9) 340 (47.0) b0.001

about an ED-DNR order within hours.

Strong predictors of ICH mortality were highly correlated with ED-

Results expressed as number (percentage) for categorical variables and mean (+-SD) for numerical variables. Abbreviations, ED, emergency department; DNR, do-not-resuscitate; GCS, Glasgow Coma Scale.

⁺ P b 0.05.

without an ED-DNR order were compared for each ICH score point (Fig. 1) and no significant difference was found. In the distribution of 30-day and 90-day mRS score, the most obvious finding was very high mortal- ity rate in the ED-DNR group (Fig. 2). In this cohort, seven patients with very mild disease (ICH score = 0-1) signed a DNR order for their long- term bedridden status caused by old ischemic stroke (n = 4), long bone fracture (n = 2), and spinal cord injury (n = 1). Of these seven patients, only one patient died, from esophageal varices bleeding caused by

Table 2

Multiple logistic regression analyses for associated factors of ED-DNR order for spontane- ous intracerebral hemorrhage (SICH) patients.

Patient characteristics P value OR (95% CI of OR)

Age (by 1 year increment) 0.024 1.049 (1.006-1.093)

Nursing home residence 0.071 1.846 (0.949-3.592)

Number of chronic diseases ? 4 0.125 1.518 (0.826-3.825) Fully ambulatory before the event b0.001 0.320 (0.166-0.595) Midline shift b0.001 9.299 (5.118-16.897)

Intraventricular extension 0.023 2.533 (1.137-5.645) Hematoma size (by 1 mL increment) b0.001 1.038 (1.021-1.056) ED arrival GCS <= 8 b0.001 8.115 (4.515-14.583)

Abbreviations, ED, emergency department, DNR, do-not-resuscitate, OR, odds ratio; CI, confidence interval.

DNR orders, indicating that the primary purpose of ED-DNR orders was to decrease the suffering of those who were expected to die in the near future, and this seemed to be the case in this cohort. ED-DNR patients came to the ED with higher ICH scores, and half of the mortality cases died within 1 week, which was consistent with previous studies [1-4]. It is noteworthy that 92.0% of ED-DNR patients also signed the paper for refusing invasive treatment such as surgery (craniectomy or ventricular drainage), adjunct airway device placement (nasal or oral airway), non-invasive positive pressure ventilation device, or preven- tive endotracheal intubation (for airway protection in comatose pa- tients), and even nasogastric tube insertion, to avoid extra suffering. Although these might be the main causes of early mortality in ED-DNR patients (not protecting the airway in the deep comatose condition or not receiving Decompressive surgery for brainstem compression), it also means that the overall decline of care aggressiveness might not just be the result of the physician’s self-fulfilling prophecy, but more im- portantly, the actual preference of the patients or families for decreasing extra suffering by the patients.

Patients with and without an ED-DNR order had the same initial ICH score and did not have a statistically significant difference in 30-day mortality rate (Fig. 1). Indeed, ED-DNR patients with an ICH score of 3-4 points had a slightly lower death rate, indicating that initial severi- ty, instead of care aggressiveness played the major role in 30-day mor- tality in ED-DNR patients. More ED-DNR patients had higher ICH scores and consequently a higher overall mortality rate. The trend showed higher that a mortality rate was noted in the ED-DNR group in patients with a high score (ICH score 5-6), suggesting that the effect of increased

Fig. 1. Comparisons of mortality rates between patients with and without ED-DNR orders in each ICH score. Abbreviations, ED, emergency department; DNR, do-not-resuscitate; ICH, intracerebral hemorrhage.

mortality was more evident in the higher severity group. And it seemed to continue to impact the mortality rate during the period of 30 to 90 days (Fig. 3). Fortunately, during period of the 30 to 90 days the over- all rate of neurological improvement in patients with an ED-DNR order was similar to those without an ED-DNR order. A previous study focus- ing on an early DNR order and functional outcome also showed a similar result [4], suggesting ED-DNR orders might not preclude the potential improvement of ICH patients.

Many studies have mentioned the potential negative impacts of early DNR orders in ICH patients including decreased survival and neu- rologic improvement [1,2,4,10] and few of them even advocated not to sign a DNR order in the early stage [1,14]. The concerns were mainly based on the statistical association between early DNR order and poor neurological outcome. However, the underlying physical and mental co- morbidities might not have been adequately assessed in the statistical

analysis, and high collinearity between DNR and other prognostic pre- dictors might bias the results of the multiple regression analysis, weakening the direct link of early DNR order and poor prognosis [2-4,10]. In addition, different prognostic endpoints might be affect- ed by early DNR orders differently [4,10]. Therefore, the causation between early DNR orders and prognosis might be far more complex than we think. Moreover, an acceptable neurological outcome for a meaningful life varies among patients and families. From the stand- point of providing quality care and respecting a patient’s autonomy, giving comprehensive prognostic information to patients or surro- gates with a discreet attitude rather than avoiding a DNR order might be more appropriate in the ED [10]. With the increase of new methods for improving the accuracy of Prognostic predictions in ICH patients [15,23], making DNR decisions might involve less likelihood of a dilemma in the future.

Fig. 2. Distributions of 30-day and 90-day Modified Rankin Scale in patients with and without ED-DNR orders. Abbreviations, ED, emergency department; DNR, do-not-resuscitate.

Fig. 3. Changes in rates of Good neurologic outcome and mortality during 30-day to 90-day follow-up in patients with and without ED-DNR orders. Abbreviations, ED, emergency department; DNR, do-not-resuscitate.

Limitations

This study had some limitations. First, its retrospective nature pre- cluded the standardized diagnostic and treatment procedures which might mean some data in the study population were missed. Fortunate- ly, management of patient with suspicious stroke had been protocolized in our ED, and only 28 patients were excluded due to incomplete data and were believed distributed randomly in the patient groups with and without ED-DNR orders. Moreover, in comparison with previous studies focusing on this issue, a larger study population in our cohort might increase the strength of the study results. Second, there are large differences in DNR policy between different healthcare systems and even different hospitals, decreasing the generalizibility of the study, which means that the rate of ED-DNR orders and even their im- pact on prognosis could vary significantly among hospitals [3,10,24]. But we did provide a new concept because this is the first report on ED-DNR orders in ICH patients and might have the potential to initiate further study. In addition, we provided information regarding the im- pact of ED-DNR orders on the period of 30 to 90 days, which was also the first time such information has been provided. Third, due to the lim- itation of the database there were no data on religion or educational and economic level of the surrogates, which might have impacts on making DNR decisions. However, the surrogates almost all were the closest fam- ily member of the patients, and therefore it is reasonable to infer that to some extent they were similar with regard to those variables.

Conclusions

The rate of ED-DNR orders for patients with SICH who were treated in our ED was 12.1% (112 in 835), and was highly correlated with early mortality. Older and sicker SICH patients had an increased rate of ED- DNR orders. Although overall mortality rate was increased in the group with ED-DNR orders, mortality rates for each ICH score point were not significantly different. During the 30-day to 90-day follow- up, mortality rate increased significantly in the ED-DNR group compared with the group without ED-DNR but the rate of neurological improvement was similar.

References

1. Hemphill 3rd JC, Newman J, Zhao S, Johnston SC. Hospital usage of early do-not-re- suscitate orders and outcome after intracerebral hemorrhage. Stroke 2004;35: 1130-4.
2. Zahuranec DB, Morgenstern LB, Sanchez BN, Resnicow K, White DB, Hemphill 3rd JC. Do-not-resuscitate orders and predictive models after intracerebral hemorrhage. Neurology 2010;75:626-33.
3. Yang TC, Li JG, Guo W. Do not resuscitate orders for patients with intracerebral hem- orrhage: experience from a Chinese tertiary care center. Eur Neurol 2015;73:144-9.
4. Brizzi M, Abul-Kasim K, Jalakas M, Selariu E, Pessah-Rasmussen H, Zia E. Early do-not- resuscitate orders in intracerebral haemorrhage; frequency and predictive value for death and functional outcome. A retrospective cohort study. Scand J Trauma Resusc Emerg Med 2012 May 25;20:36. http://dx.doi.org/10.1186/1757-7241-20-36.
5. Wrenn K, Brody SL. Do-not-resuscitate orders in the emergency department. Am J Med 1992;92:129-33.
6. Cheng YH, Wang JJ, Wu KH, Huang S, Kuo M, Su CH. Do-not-resuscitate orders and related factors among family surrogates of patients in the emergency department. Support Care Cancer 2016;24:1999-2006.
7. Fox E, Siegler M. Redefining the emergency physician’s role in do-not-resuscitate de- cision-making. Am J Med 1992;92:125-8.
8. Iserson KV. Withholding and withdrawing medical treatment: an emergency medi- cine perspective. Ann Emerg Med 1996;28:51-4.
9. Leon MD, Wilson EM. Development of a statewide protocol for the prehospital iden- tification of DNR patients in Connecticut including new DNR regulations. Ann Emerg Med 1999;34:263-74.
10. Patel AA, Mahajan A, Benjo A, Jani VB, Annapureddy N, Agarwal SK, et al. A national perspective of do-not-resuscitate order utilization predictors in intracerebral hem- orrhage. Neurohospitalist 2016;6:7-10.
11. Fan JS, Huang HH, Chen YC, Yen DH, Kao WF, Huang MS, et al. Emergency depart- ment Neurologic deterioration in patients with spontaneous intracerebral hemor- rhage: incidence, predictors, and prognostic significance. Acad Emerg Med 2012; 19:133-8.
12. Fan JS, Chen YC, Huang HH, How CK, Yen DH, Huang MS. The association between on-scene blood pressure and early Neurological deterioration in patients with spon- taneous intracerebral haemorrhage. Emerg Med J 2015;32:239-43.
13. Fan JS, Chen YC, Huang HH, Yen DH, How CK, Huang MS. Interhospital transfer neu- rological deterioration in patients with spontaneous intracerebral haemorrhage: in- cidence and risk factors. Postgrad Med J 2017;39:349-53. http://dx.doi.org/10.1136/ postgradmedj-2016-134463.
14. Morgenstern LB, Hemphill 3rd JC, Anderson C, Becker K, Broderick JP, Connolly Jr ES, et al. American Heart Association Stroke Council and Council on Cardiovascular Nursing, guidelines for the management of spontaneous intracerebral hemorrhage: a guideline for healthcare professionals from the American Heart Association/Amer- ican Stroke Association. Stroke 2010;41:2108-29.
15. Hemphill 3rd JC, Greenberg SM, Anderson CS, Becker K, Bendok BR, Cushman M, et al. American Heart Association Stroke Council; Council on Cardiovascular and Stroke Nursing; Council on Clinical Cardiology. guidelines for the management of spontaneous intracerebral hemorrhage: a guideline for healthcare professionals from the American Heart Association/American Stroke Association. Stroke 2015; 46:2032-60.
16. Hemphill 3rd JC, Bonovich DC, Besmertis L, Manley GT, Johnston SC. The ICH score: a simple, reliable grading scale for intracerebral hemorrhage. Stroke 2001;32:891-7.
17. Jain A, Jain M, Bellolio MF, Schears RM, Rabinstein AA, Ganti L. Is early DNR a self-ful- filling prophecy for patients with spontaneous intracerebral hemorrhage? Neurocrit Care 2013;19:342-6.
18. Peabody JW, Yu JC, Wang YR, Bickel SR. Health system reform in the Republic of China. Formulating policy in a market-based health system. JAMA 1995;273:1-777.
19. Lai YL, Su WH. Palliative medicine and the hospice movement in Taiwan. Support Care Cancer 1997;5:348-50.
20. Ross MM, Dunning J, Edwards N. Palliative care in China: facilitating the process of development. J Palliat Care 2001;17:281-7.
21. Chiu TY. End-of-life decision making in Taiwan. End-of-life decision making: a cross-

    national study; 2005 169-75.

    Yang CL, Chiu TY, Hsiung YF, Hu W. Which factors have the greatest influence on be- reaved families’ willingness to execute advance directives in Taiwan? Cancer Nurs 2011;34:98-106.

22. Szepesi R, Szell IK, Hortobagyi T, Kardos L, Nagy K, Lanczi LI, et al. New prognostic score for the prediction of 30-day outcome in spontaneous supratentorial cerebral haemorrhage. Biomed Res Int 2015;2015:961085. http://dx.doi.org/10.1155/2015/ 961085 (Epub 2015 Jan 8).
23. Dzeng E, Colaianni A, Roland M, Chander G, Smith TJ, Kelly MP, et al. Influence of in- stitutional culture and policies on do-not-resuscitate decision making at the end of life. JAMA Intern Med 2015;175:812-9.