Clinical Investigations
Factors Associated With Two Different Protocols of Do-Not-Resuscitate Orders in a Medical ICU* Yen-Yuan Chen, MD, MPH, PhD1,2; Nahida H. Gordon, MS, PhD3; Alfred F. Connors Jr, MD4; Allan Garland, MA, MD5; Hong-Shiee Lai, MD, PhD6,7; Stuart J. Youngner, MD3
Objective: The State of Ohio in the United States has the legislation for two different protocols of do-not-resuscitate orders. The objective of this study was to examine the clinical/demographic factors and outcomes associated with the two different do-notresuscitate orders. Design: Data were concurrently and retrospectively collected from August 2002 to December 2005. The clinical/demographic factors of do-not-resuscitate patients were compared with those of non–do-not-resuscitate patients, and the clinical/demographic factors of do-not-resuscitate comfort care–arrest patients were compared with those of do-not-resuscitate comfort care patients. Setting: An ICU in a university-affiliated hospital located at Northeast Ohio in the United States. Patients: A sample of 2,440 patients was collected: 389 patients were do-not-resuscitate; and 2,051 patients were non–do-notresuscitate. Among the 389 do-not-resuscitate patients, 194
*See also p. 2299. 1 Department of Social Medicine, National Taiwan University College of Medicine, Taipei, Taiwan. 2 Department of Medical Education, National Taiwan University Hospital, Taipei, Taiwan. 3 Department of Bioethics, Case Western Reserve University, Cleveland, OH. 4 Department of Medicine, Case Western Reserve University School of Medicine at MetroHealth Medical Center, Cleveland, OH. 5 Department of Medicine, University of Manitoba Health Sciences Center, Winnipeg, MB. 6 Department of Surgery, National Taiwan University College of Medicine, Taipei, Taiwan. 7 Department of Surgery, National Taiwan University Hospital, Taipei, Taiwan. This work was performed at Case Western Reserve University, MetroHealth Medical Center, National Taiwan University College of Medicine, and National Taiwan University Hospital. The authors have disclosed that they do not have any potential conflicts of interest. Address requests for reprints to: Hong-Shiee Lai, MD, PhD, Department of Surgery, National Taiwan University College of Medicine, No. 1, Rd. Ren-Ai sec. 1, Chong-Cheng District, Taipei 100, Taiwan. E-mail: hslai@ ntu.edu.tw Copyright © 2014 by the Society of Critical Care Medicine and Lippincott Williams & Wilkins DOI: 10.1097/CCM.0000000000000411
2188
www.ccmjournal.org
were do-not-resuscitate comfort care–arrest patients and 91 were do-not-resuscitate comfort care patients. Interventions: None. Measurements and Main Results: The factors associated with donot-resuscitate were older age, race and ethnicity with white race, more severe clinical illness at admission to the ICU, and longer stay before admission to the ICU. Comparing do-not-resuscitate comfort care–arrest patients with do-not-resuscitate comfort care patients, those with more severe clinical illness, longer ICU stay before making a do-not-resuscitate decision, and being cared for by only one intensivist during ICU stay were significantly associated with do-not-resuscitate comfort care decisions. For 149 do-not-resuscitate patients who eventually survived to hospital discharge and 86 do-not-resuscitate patients who eventually did not, only eight (5.4%) and 23 (26.7%) had the order written within 48 hours before the end of ICU stay, respectively. Conclusions: Our study showed that some clinical/demographic factors predicted do-not-resuscitate comfort care orders. This study also suggested that Ohio’s Do-Not-Resuscitate Law, clearly indicating two different protocols of do-not-resuscitate orders, facilitated early do-not-resuscitate decision. (Crit Care Med 2014; 42:2188–2196) Key Words: cardiopulmonary resuscitation; comfort care; do not resuscitate
C
losed-chest cardiac massage was first introduced to clinical medicine in 1960 (1). In 1974, the American Heart Association approved the use of cardiopulmonary resuscitation (CPR) in clinical practice and also proposed that it is ethically appropriate to withhold or withdraw CPR if it is not anticipated to benefit the patient (2). When a do-not-resuscitate (DNR) order is written, the meaning of the order must be clear both to the healthcare workers and the patient/surrogate decision-maker. Nevertheless, the meaning of DNR is often too vague about medical care provided to DNR patients short of cardiac or respiratory arrest. Some studies showed that healthcare workers tend to provide less medical care to DNR patients (3–7). Others raised the concern that DNR patients may be psychologically and October 2014 • Volume 42 • Number 10
Clinical Investigations
emotionally abandoned (8). Although those studies did not specifically demonstrate that DNR patients are medically or psychologically abandoned, they may partly reflect that healthcare workers’ interpretations of medical care provided to DNR patients are inconsistent. Motivated in part to provide a clear direction to healthcare workers about medical care provided to DNR patients, the Ohio Department of Health established Ohio’s Do-NotResuscitate Law in 1998. This law sets forth two different protocols of DNR that allow patients/surrogate decision-makers to choose the medical care they wish to receive after DNR orders are written but before arrest occurs (9, 10). Ohio’s Do-Not-Resuscitate Law clearly describes how do-not-resuscitate comfort care–arrest (DNRCC-Arrest) and do-notresuscitate comfort care (DNRCC) orders should be executed. After a DNRCC-Arrest order is written, the patient is eligible to receive aggressive interventions to extend life short of arrest if the use of the interventions is ethically appropriate. CPR will not be performed at the moment of cardiac or respiratory arrest. DNRCC-Arrest is the same as “All except CPR.” After a DNRCC order is written, the patient only receives comfort care measures to relieve symptoms until the end of life or the withdrawal of the DNRCC order. The State of Ohio is the pioneer in the United States with regard to this type of legislation. To the best of our knowledge, no studies have been performed to explore the characteristics associated with the two different groups of DNR patients. The objective of this study was to examine the clinical/demographic factors and outcomes associated with DNRCC-Arrest/DNRCC patients.
MATERIALS AND METHODS The data were collected in a medical ICU in a 520-bed, countyowned, and university-affiliated tertiary teaching hospital located in West Cleveland. This is a closed-model ICU where, during the study interval, all patients were cared for by a single team of physicians comprising, at any given time, one boardcertified intensivist, one ICU fellow, and five house officers who took overnight calls on a one-in-five rotation. These intensivists did 14-day rotating blocks of time, but each weekend was cross-covered by a different intensivist from the same pool of nine intensivists. The individual intensivist on duty during any given weekday or weekend was independently responsible for all care decisions, including DNR decisions. Data were concurrently and retrospectively collected from August 2002 to December 2005, excluding March to May 2004 when data collection was suspended for personnel limitations. We only included the initial admissions to the ICU during the study period. For examination of factors associated with DNR decisions, DNR and non-DNR patients were included. For examination of factors associated with DNRCC-Arrest and DNRCC decisions, patients were excluded if they changed the status of DNR during ICU stay, either from DNRCC-Arrest to DNRCC or from DNRCC to DNRCC-Arrest. We collected variables associated with patient demographics, clinical data, and others (e.g., DNR decisions, prior DNR orders, cared for by only one intensivist during ICU stay). Critical Care Medicine
Two types of analyses were contemplated. The first was a preliminary analysis using exploratory data techniques to examine univariate characteristics (central tendency, dispersion, and distribution) and bivariate relationships (correlations) among covariates and covariates with outcomes. These exploratory techniques were based on proportions (in the case of categorical variables) and means (in the case of interval variables). Associations with a DNR order were analyzed by using Student t test and chi-square test whichever is appropriate depending on the scale of measure. As a preliminary step for the multivariate analyses, the characterization and estimation of the relationship of the outcome with each of the available variables serving as covariates was implemented. Any variable found to have a significance level less than 0.25 was retained for further consideration in the model building process (11). Two outcome variables for analyses were considered. The first was whether patients had a DNR order written (1: DNR; 0: non-DNR); and the second was whether the patient had a DNRCC-Arrest order or a DNRCC order (1: DNRCC-Arrest, 0: DNRCC). Because the outcomes were dichotomous, logistic regression models were implemented to examine the relationship of the independent variables with the outcomes. The discrimination of the multivariate logistic regression models was checked by the area under the receiver operating characteristic (ROC) curve (12). The calibration of the model was checked by the Hosmer-Lemeshow goodness-of-fit test (11). STATA 11.0 (StataCorp, College Station, TX) for Windows PC was used for all statistical analyses. This study was approved by the institutional review board in MetroHealth Medical Center (IRB07-01218).
RESULTS A sample of 2,440 patients was collected. Two thousand and fifty-one were non-DNR and 389 (15.94%) were DNR patients. Among the DNR patients, a total of 194 patients were DNRCC-Arrest: 188 had only DNRCC-Arrest during their ICU stay; and six had a DNRCC-Arrest order written and then an order “Withdraw” was added. In addition, a total of 91 patients were DNRCC: 88 had only DNRCC during their ICU stay, and three had a do-not-intubate order written initially and then the order was changed to a DNRCC order. DNR and Non-DNR Among the 2,440 patients, 193 (7.9%) died in the ICU, and 327 (13.4%) did not survive to hospital discharge. Among the 389 DNR patients, the ICU and hospital mortality rates were 37.3% and 57.3%, respectively. Table 1 shows the results of bivariate analysis for DNR and non-DNR patients. After controlling for confounding variables, each yearly increment of age was associated with 3% increased odds of having a DNR order written (p < 0.01). Black patients were less likely to have DNR than white patients (p < 0.01). Each unit increment of Acute Physiology and Chronic Health Evaluation (APACHE) II score at ICU admission was 1.12 www.ccmjournal.org
2189
Chen et al
Table 1. Comparison of Characteristics and Measures Between Do-Not-Resuscitate and Non–Do-Not-Resuscitate Patients p
Characteristics and Measures
DNR
Non-DNR
Total patients
389
2,051
Age at admission, yr
66.3 ± 16.0
54.2 ± 17.8
< 0.01
Gender (female) (%)
181 (46.5)
986 (48.1)
0.58
Demographics
Race/ethnicity (%)
< 0.01
White
292 (75.1)
1,277 (62.3)
Black
77 (19.8)
602 (29.4)
Others
20 (5.1)
172 (8.4)
Acute Physiology and Chronic Health Evaluation II score at ICU admission
28.4 ± 9.9
18.1 ± 8.6
< 0.01
Length of stay before ICU admission, hr
38.5 ± 115.7
18.0 ± 69.4
< 0.01
Mechanical ventilation (yes) (%)
202 (51.9)
552 (26.9)
< 0.01
Clinical data
Other data (%) Insurance type Private
< 0.01 166 (42.7)
611 (29.8)
Medicare only
67 (17.2)
257 (12.5)
Medicaid only
65 (16.7)
506 (24.7)
Medicare and Medicaid
65 (16.7)
316 (15.4)
None
26 (6.7)
361 (17.6)
Source of admission to ICU
< 0.01
Emergency department
254 (65.3)
1,513 (73.8)
Floor
116 (29.8)
413 (20.1)
11 (2.8)
49 (2.4)
Outside hospital
4 (1.0)
26 (1.3)
Miscellaneous
4 (1.0)
50 (2.4)
a
Other ICU
ICU admission diagnosis Respiratory diseases
< 0.01 150 (38.7)
587 (28.6)
Gastrointestinal diseases
40 (10.3)
321 (15.7)
Cardiovascular diseases
97 (24.9)
347 (16.9)
Neurological diseases
82 (21.1)
334 (16.3)
Others
20 (5.1)
462 (22.5)
Had a prior end-of-life decision (yes) (%)
21 (5.4)
43 (2.1)
Cared for by only one intensivist (yes) (%)
185 (47.6)
1,288 (62.8)
< 0.01 < 0.01
DNR = do-not-resuscitate. a “Floor” means that patients were admitted to the medical ICU from other medical or surgical floors in the same hospital. Continuous variables of DNR and non-DNR patients were compared using Student t test; categorical variables of DNR and non-DNR patients were compared using chi-square test.
(p
Factors Associated With Two Different Protocols of Do-Not-Resuscitate Orders in a Medical ICU* Yen-Yuan Chen, MD, MPH, PhD1,2; Nahida H. Gordon, MS, PhD3; Alfred F. Connors Jr, MD4; Allan Garland, MA, MD5; Hong-Shiee Lai, MD, PhD6,7; Stuart J. Youngner, MD3
Objective: The State of Ohio in the United States has the legislation for two different protocols of do-not-resuscitate orders. The objective of this study was to examine the clinical/demographic factors and outcomes associated with the two different do-notresuscitate orders. Design: Data were concurrently and retrospectively collected from August 2002 to December 2005. The clinical/demographic factors of do-not-resuscitate patients were compared with those of non–do-not-resuscitate patients, and the clinical/demographic factors of do-not-resuscitate comfort care–arrest patients were compared with those of do-not-resuscitate comfort care patients. Setting: An ICU in a university-affiliated hospital located at Northeast Ohio in the United States. Patients: A sample of 2,440 patients was collected: 389 patients were do-not-resuscitate; and 2,051 patients were non–do-notresuscitate. Among the 389 do-not-resuscitate patients, 194
*See also p. 2299. 1 Department of Social Medicine, National Taiwan University College of Medicine, Taipei, Taiwan. 2 Department of Medical Education, National Taiwan University Hospital, Taipei, Taiwan. 3 Department of Bioethics, Case Western Reserve University, Cleveland, OH. 4 Department of Medicine, Case Western Reserve University School of Medicine at MetroHealth Medical Center, Cleveland, OH. 5 Department of Medicine, University of Manitoba Health Sciences Center, Winnipeg, MB. 6 Department of Surgery, National Taiwan University College of Medicine, Taipei, Taiwan. 7 Department of Surgery, National Taiwan University Hospital, Taipei, Taiwan. This work was performed at Case Western Reserve University, MetroHealth Medical Center, National Taiwan University College of Medicine, and National Taiwan University Hospital. The authors have disclosed that they do not have any potential conflicts of interest. Address requests for reprints to: Hong-Shiee Lai, MD, PhD, Department of Surgery, National Taiwan University College of Medicine, No. 1, Rd. Ren-Ai sec. 1, Chong-Cheng District, Taipei 100, Taiwan. E-mail: hslai@ ntu.edu.tw Copyright © 2014 by the Society of Critical Care Medicine and Lippincott Williams & Wilkins DOI: 10.1097/CCM.0000000000000411
2188
www.ccmjournal.org
were do-not-resuscitate comfort care–arrest patients and 91 were do-not-resuscitate comfort care patients. Interventions: None. Measurements and Main Results: The factors associated with donot-resuscitate were older age, race and ethnicity with white race, more severe clinical illness at admission to the ICU, and longer stay before admission to the ICU. Comparing do-not-resuscitate comfort care–arrest patients with do-not-resuscitate comfort care patients, those with more severe clinical illness, longer ICU stay before making a do-not-resuscitate decision, and being cared for by only one intensivist during ICU stay were significantly associated with do-not-resuscitate comfort care decisions. For 149 do-not-resuscitate patients who eventually survived to hospital discharge and 86 do-not-resuscitate patients who eventually did not, only eight (5.4%) and 23 (26.7%) had the order written within 48 hours before the end of ICU stay, respectively. Conclusions: Our study showed that some clinical/demographic factors predicted do-not-resuscitate comfort care orders. This study also suggested that Ohio’s Do-Not-Resuscitate Law, clearly indicating two different protocols of do-not-resuscitate orders, facilitated early do-not-resuscitate decision. (Crit Care Med 2014; 42:2188–2196) Key Words: cardiopulmonary resuscitation; comfort care; do not resuscitate
C
losed-chest cardiac massage was first introduced to clinical medicine in 1960 (1). In 1974, the American Heart Association approved the use of cardiopulmonary resuscitation (CPR) in clinical practice and also proposed that it is ethically appropriate to withhold or withdraw CPR if it is not anticipated to benefit the patient (2). When a do-not-resuscitate (DNR) order is written, the meaning of the order must be clear both to the healthcare workers and the patient/surrogate decision-maker. Nevertheless, the meaning of DNR is often too vague about medical care provided to DNR patients short of cardiac or respiratory arrest. Some studies showed that healthcare workers tend to provide less medical care to DNR patients (3–7). Others raised the concern that DNR patients may be psychologically and October 2014 • Volume 42 • Number 10
Clinical Investigations
emotionally abandoned (8). Although those studies did not specifically demonstrate that DNR patients are medically or psychologically abandoned, they may partly reflect that healthcare workers’ interpretations of medical care provided to DNR patients are inconsistent. Motivated in part to provide a clear direction to healthcare workers about medical care provided to DNR patients, the Ohio Department of Health established Ohio’s Do-NotResuscitate Law in 1998. This law sets forth two different protocols of DNR that allow patients/surrogate decision-makers to choose the medical care they wish to receive after DNR orders are written but before arrest occurs (9, 10). Ohio’s Do-Not-Resuscitate Law clearly describes how do-not-resuscitate comfort care–arrest (DNRCC-Arrest) and do-notresuscitate comfort care (DNRCC) orders should be executed. After a DNRCC-Arrest order is written, the patient is eligible to receive aggressive interventions to extend life short of arrest if the use of the interventions is ethically appropriate. CPR will not be performed at the moment of cardiac or respiratory arrest. DNRCC-Arrest is the same as “All except CPR.” After a DNRCC order is written, the patient only receives comfort care measures to relieve symptoms until the end of life or the withdrawal of the DNRCC order. The State of Ohio is the pioneer in the United States with regard to this type of legislation. To the best of our knowledge, no studies have been performed to explore the characteristics associated with the two different groups of DNR patients. The objective of this study was to examine the clinical/demographic factors and outcomes associated with DNRCC-Arrest/DNRCC patients.
MATERIALS AND METHODS The data were collected in a medical ICU in a 520-bed, countyowned, and university-affiliated tertiary teaching hospital located in West Cleveland. This is a closed-model ICU where, during the study interval, all patients were cared for by a single team of physicians comprising, at any given time, one boardcertified intensivist, one ICU fellow, and five house officers who took overnight calls on a one-in-five rotation. These intensivists did 14-day rotating blocks of time, but each weekend was cross-covered by a different intensivist from the same pool of nine intensivists. The individual intensivist on duty during any given weekday or weekend was independently responsible for all care decisions, including DNR decisions. Data were concurrently and retrospectively collected from August 2002 to December 2005, excluding March to May 2004 when data collection was suspended for personnel limitations. We only included the initial admissions to the ICU during the study period. For examination of factors associated with DNR decisions, DNR and non-DNR patients were included. For examination of factors associated with DNRCC-Arrest and DNRCC decisions, patients were excluded if they changed the status of DNR during ICU stay, either from DNRCC-Arrest to DNRCC or from DNRCC to DNRCC-Arrest. We collected variables associated with patient demographics, clinical data, and others (e.g., DNR decisions, prior DNR orders, cared for by only one intensivist during ICU stay). Critical Care Medicine
Two types of analyses were contemplated. The first was a preliminary analysis using exploratory data techniques to examine univariate characteristics (central tendency, dispersion, and distribution) and bivariate relationships (correlations) among covariates and covariates with outcomes. These exploratory techniques were based on proportions (in the case of categorical variables) and means (in the case of interval variables). Associations with a DNR order were analyzed by using Student t test and chi-square test whichever is appropriate depending on the scale of measure. As a preliminary step for the multivariate analyses, the characterization and estimation of the relationship of the outcome with each of the available variables serving as covariates was implemented. Any variable found to have a significance level less than 0.25 was retained for further consideration in the model building process (11). Two outcome variables for analyses were considered. The first was whether patients had a DNR order written (1: DNR; 0: non-DNR); and the second was whether the patient had a DNRCC-Arrest order or a DNRCC order (1: DNRCC-Arrest, 0: DNRCC). Because the outcomes were dichotomous, logistic regression models were implemented to examine the relationship of the independent variables with the outcomes. The discrimination of the multivariate logistic regression models was checked by the area under the receiver operating characteristic (ROC) curve (12). The calibration of the model was checked by the Hosmer-Lemeshow goodness-of-fit test (11). STATA 11.0 (StataCorp, College Station, TX) for Windows PC was used for all statistical analyses. This study was approved by the institutional review board in MetroHealth Medical Center (IRB07-01218).
RESULTS A sample of 2,440 patients was collected. Two thousand and fifty-one were non-DNR and 389 (15.94%) were DNR patients. Among the DNR patients, a total of 194 patients were DNRCC-Arrest: 188 had only DNRCC-Arrest during their ICU stay; and six had a DNRCC-Arrest order written and then an order “Withdraw” was added. In addition, a total of 91 patients were DNRCC: 88 had only DNRCC during their ICU stay, and three had a do-not-intubate order written initially and then the order was changed to a DNRCC order. DNR and Non-DNR Among the 2,440 patients, 193 (7.9%) died in the ICU, and 327 (13.4%) did not survive to hospital discharge. Among the 389 DNR patients, the ICU and hospital mortality rates were 37.3% and 57.3%, respectively. Table 1 shows the results of bivariate analysis for DNR and non-DNR patients. After controlling for confounding variables, each yearly increment of age was associated with 3% increased odds of having a DNR order written (p < 0.01). Black patients were less likely to have DNR than white patients (p < 0.01). Each unit increment of Acute Physiology and Chronic Health Evaluation (APACHE) II score at ICU admission was 1.12 www.ccmjournal.org
2189
Chen et al
Table 1. Comparison of Characteristics and Measures Between Do-Not-Resuscitate and Non–Do-Not-Resuscitate Patients p
Characteristics and Measures
DNR
Non-DNR
Total patients
389
2,051
Age at admission, yr
66.3 ± 16.0
54.2 ± 17.8
< 0.01
Gender (female) (%)
181 (46.5)
986 (48.1)
0.58
Demographics
Race/ethnicity (%)
< 0.01
White
292 (75.1)
1,277 (62.3)
Black
77 (19.8)
602 (29.4)
Others
20 (5.1)
172 (8.4)
Acute Physiology and Chronic Health Evaluation II score at ICU admission
28.4 ± 9.9
18.1 ± 8.6
< 0.01
Length of stay before ICU admission, hr
38.5 ± 115.7
18.0 ± 69.4
< 0.01
Mechanical ventilation (yes) (%)
202 (51.9)
552 (26.9)
< 0.01
Clinical data
Other data (%) Insurance type Private
< 0.01 166 (42.7)
611 (29.8)
Medicare only
67 (17.2)
257 (12.5)
Medicaid only
65 (16.7)
506 (24.7)
Medicare and Medicaid
65 (16.7)
316 (15.4)
None
26 (6.7)
361 (17.6)
Source of admission to ICU
< 0.01
Emergency department
254 (65.3)
1,513 (73.8)
Floor
116 (29.8)
413 (20.1)
11 (2.8)
49 (2.4)
Outside hospital
4 (1.0)
26 (1.3)
Miscellaneous
4 (1.0)
50 (2.4)
a
Other ICU
ICU admission diagnosis Respiratory diseases
< 0.01 150 (38.7)
587 (28.6)
Gastrointestinal diseases
40 (10.3)
321 (15.7)
Cardiovascular diseases
97 (24.9)
347 (16.9)
Neurological diseases
82 (21.1)
334 (16.3)
Others
20 (5.1)
462 (22.5)
Had a prior end-of-life decision (yes) (%)
21 (5.4)
43 (2.1)
Cared for by only one intensivist (yes) (%)
185 (47.6)
1,288 (62.8)
< 0.01 < 0.01
DNR = do-not-resuscitate. a “Floor” means that patients were admitted to the medical ICU from other medical or surgical floors in the same hospital. Continuous variables of DNR and non-DNR patients were compared using Student t test; categorical variables of DNR and non-DNR patients were compared using chi-square test.
(p
