This article was downloaded by: [The University of Manchester Library] On: 08 October 2014, At: 15:08 Publisher: Routledge Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK
Death Studies Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/udst20
Factors Affecting Code Status in a University Hospital Intensive Care Unit a
Lauren Jodi Van Scoy & Michael Sherman
a
a
Department of Internal Medicine and Division of Pulmonary, Critical Care and Sleep Medicine , Drexel University College of Medicine , Philadelphia , Pennsylvania , USA Accepted author version posted online: 01 Apr 2013.Published online: 09 May 2013.
To cite this article: Lauren Jodi Van Scoy & Michael Sherman (2013) Factors Affecting Code Status in a University Hospital Intensive Care Unit, Death Studies, 37:8, 768-781, DOI: 10.1080/07481187.2012.699908 To link to this article: http://dx.doi.org/10.1080/07481187.2012.699908
PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content.
Downloaded by [The University of Manchester Library] at 15:08 08 October 2014
This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://www.tandfonline.com/page/terms-and-conditions
Death Studies, 37: 768–781, 2013 Copyright # Taylor & Francis Group, LLC ISSN: 0748-1187 print=1091-7683 online DOI: 10.1080/07481187.2012.699908
Downloaded by [The University of Manchester Library] at 15:08 08 October 2014
FACTORS AFFECTING CODE STATUS IN A UNIVERSITY HOSPITAL INTENSIVE CARE UNIT LAUREN JODI VAN SCOY and MICHAEL SHERMAN Department of Internal Medicine and Division of Pulmonary, Critical Care and Sleep Medicine, Drexel University College of Medicine, Philadelphia, Pennsylvania, USA
The authors collected data on diagnosis, hospital course, and end-of-life preparedness in patients who died in the intensive care unit (ICU) with ‘‘full code’’ status (defined as receiving cardiopulmonary resuscitation), compared with those who didn’t. Differences were analyzed using binary and stepwise logistic regression. They found no differences in demographics, comorbidities, ventilator, hospital, or ICU days between groups. No-code patients were more likely to have higher APACHE-II scores (p < .0001), gastrointestinal=hepatic conditions (p < .01) and an advanced directive (p ¼ .03). Patients dying with full code status were more likely to have previously coded (p < .0001), and had more central lines (p ¼ .03). Implications are discussed.
With the geriatric population growing rapidly, predictions are that by the year 2050, 20.4% of the United States population will be over the age of 65 (Hobbs, 2008). Today’s physicians must therefore develop a skill set that will allow them to facilitate the inevitable end-of-life care issues that will arise as these patients age. Previous studies of patient preferences, expectations, and satisfaction with end-of-life care found a general preference for home-based, symptom-guided care (Davison, 2010; Lamont, 2005; Received 27 September 2011; accepted 9 May 2012. This study was funded by the Drexel University College of Medicine Aging Initiative Grant. We acknowledge Jesse Chittams, Dominique Williams and Cindy Liao of the Biostatics Service Center at the Drexel University School of Public Health Department of Epidemiology and Biostatistics for their assistance with data analysis. We also acknowledge Ms. Pamela Fried of Drexel University’s Academic Publishing Services for her editorial assistance. Address correspondence to Lauren Jodi Van Scoy, Department of Internal Medicine, Division of Pulmonary, Critical Care and Sleep Medicine, Drexel University College of Medicine, 245 N. 15th Street, Mail Stop 107, Philadelphia, PA 19102. E-mail: lvanscoy@ drexelmed.edu
768
Downloaded by [The University of Manchester Library] at 15:08 08 October 2014
Code Status
769
Steinhauser et al., 2000; The SUPPORT Principal Investigators, 1995). However, despite this well-established preference, most patients receive invasive, expensive medical interventions in the last days of their lives that often do not correspond with their stated preferences (Bell, Somogyi-Zalud, & Masaki, 2010; Lamont, 2005; The SUPPORT Principal Investigators, 1995). These studies and others suggest that there may be a gap between patients’ preferences and actual outcomes. Consequently, the need is great for physicians to understand what lies behind the wishes, differences, and biases that may exist in their patients concerning end-of-life planning and care. Because the majority of these end-of-life studies look at reported patient preferences rather than at actual outcomes, it is unclear if the stated preferences contribute directly to a patient’s level of care (i.e., code status) at the end of his or her life. We therefore conducted a retrospective chart review of 100 patients who died in the intensive care unit (ICU) in our urban university hospital to determine relevant factors that might impact the level of care provided at the end of life. We compared patients receiving aggressive measures at the end of life (full code) with those opting for ‘‘do not resuscitate’’ (no-code) status at the time of death. We investigated patient demographics, diagnosis, aggressiveness of hospital course, end-of-life preparedness, and relationships of proxies to determine if any of these factors affect code status at the time of death. Method Patient Selection Our protocol was reviewed and approved by the university’s Institutional Review Board. Hospital mortality lists that were generated by hospital administration mortality databases were reviewed from November 2006 to March 2008 in chronological order by date of death. Participants were eligible for inclusion in the study if they died in an ICU, including medical, neurological, cardiac, cardiothoracic, and surgical ICUs, after a minimum 72-hr stay. Patients were excluded if they were less than 18 years of age at death. Two groups of patients were identified. The first group consisted of patients who died in a ‘‘no-code’’ status, which was defined as not receiving cardiopulmonary resuscitation or advanced cardiac life
770
L. J. Van Scoy and M. Sherman
Downloaded by [The University of Manchester Library] at 15:08 08 October 2014
support protocol and as having do-not-resuscitate (DNR) or ‘‘withdrawal of life support’’ orders at the time of death. The second group was the ‘‘full code’’ group, defined as having had cardiopulmonary resuscitation and advanced cardiac life support protocol. Eligible patients were included in the study until we had approximately 50 patients per group who met the inclusion criteria. Variables We collected data in three broad categories: patient demographics, characteristics of hospital course, and end-of-life care preparedness. The first category, patient demographics and disease, included factors such as age, sex, race, diagnosis at death, and number of comorbid conditions. The second category, hospital course characteristics, included measures to quantify the aggressiveness of care while the patient was in the hospital. The number of central lines (central venous catheters), invasive procedures, units of blood products transfused, and surgical procedures were counted for the entire hospital stay. Each of these variables require an informed consent by the patient or proxy and therefore each measure represents a decision to pursue aggressive care. The number of central lines was divided by the length of stay to control for variable hospital durations. Units of blood products was chosen as a measure because the most critically ill patients often require blood product transfusion prior procedures due to the severity of their illness. The final category, end-of-life preparedness, identified the presence or absence of advance directives in the medical record. Advance directives included living wills and the presence of documents formally assigning proxies (such as a durable power of attorney) as well as the relationship of the proxy to the patient. We hypothesized that a more severely ill patient undergoing a long or aggressive hospital course, or their proxy decision makers witnessing the prolonged course, would have more time to come to terms with dying and thus decisions would be made to shift care toward nonlife-sustaining treatments by the time of death. Statistical Analysis Chi-square tests were used to compare categorical demographic variables, and t-tests were used to compare continuous and ordinal
Downloaded by [The University of Manchester Library] at 15:08 08 October 2014
Code Status
771
variables. For categorical variables, statistics included frequencies and percentages. For continuous variables, statistics included estimates of central tendency (means, medians) and measures of variability (standard deviations, interquartile ranges, ranges). Correlation matrixes were constructed for all continuous variables, and cross tabulations were run on all categorical variables. For all statistical tests the level of significance was set at the alpha level of .05. Summary statistics and cross tabulations were generated to assess the effect of a number of potential predictors on code status. These initial evaluations were followed by a formal statistical assessment using logistic regression. A series of univariate logistic regression analyses were conducted to assess the effect of each potential covariate independently on the code status. The univariate analyses were followed by a stepwise logistic regression to develop the final multiple logistic regression model consisting of the variables that were significant at the univariate level. A logistic regression was also used to assess the effect of measures of aggressiveness of hospital course on coding status. All analyses were performed using the SAS software package (SAS Institute Inc, 2009; Stokes, Davis, & Koch, 1995). Results A total of 100 study patients were analyzed. The full code group contained 48 patients and the no-code group contained 52 patients. All but one patient was designated as full code on arrival at the hospital. Of the patients in the full code group, there were no changes in code status at any time during their hospitalization. For the patients in the no-code group, there was a mean of 1.44 (0.63) changes in code status throughout the hospital stay, and all of the changes involved a de-escalation in the level of care. Patient Demographics and Disease Table 1 shows the characteristics of the study participants. There were no significant differences in age, gender, race, comorbidities, APACHE II disease severity score at hospital arrival, hospital length of stay, and intensive care length of stay, unit type, or number of ventilator days between the code and no-code groups. The
772
L. J. Van Scoy and M. Sherman
TABLE 1 Description of Study Subjects
Downloaded by [The University of Manchester Library] at 15:08 08 October 2014
Demographics Age (years), M (SD) Demographic factor Gender Male Female Race African American White Other Disease type responsible for deatha Sepsis Cardiovascular Pulmonary Gastrointestinal or hepatic Cancer Neurological Other Unit type Medical=medical step-down Cardiothoracic surgery, surgical, and trauma unit Cardiac care unit Hospital course characteristics Comorbidity APACHE II at hospital arrival Hospital days Unit days Ventilator days
All groups (n ¼ 100)
Code group (n ¼ 48)
No-code group (n ¼ 52)
p value
62.5 (14.4) n, (%)
60.8 (14.6) n, (%)
64.1 (14.2) n, (%)
.25
58 (58) 42 (42)
28 (58) 20 (42)
30 (58) 22 (42)
43 (43)
23 (48)
20 (38)
49 (49) 8 (8) n (%)b
20 (42) 5 (10) n (%)b
29 (56) 3 (6) n (%)b
41 17 6 11
(41) (17) (6) (11)
24 10 2 1
(59) (59) (33) (9)
17 7 4 10
(41) (41) (67) (91)
.27 .47 .41
Death Studies Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/udst20
Factors Affecting Code Status in a University Hospital Intensive Care Unit a
Lauren Jodi Van Scoy & Michael Sherman
a
a
Department of Internal Medicine and Division of Pulmonary, Critical Care and Sleep Medicine , Drexel University College of Medicine , Philadelphia , Pennsylvania , USA Accepted author version posted online: 01 Apr 2013.Published online: 09 May 2013.
To cite this article: Lauren Jodi Van Scoy & Michael Sherman (2013) Factors Affecting Code Status in a University Hospital Intensive Care Unit, Death Studies, 37:8, 768-781, DOI: 10.1080/07481187.2012.699908 To link to this article: http://dx.doi.org/10.1080/07481187.2012.699908
PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content.
Downloaded by [The University of Manchester Library] at 15:08 08 October 2014
This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://www.tandfonline.com/page/terms-and-conditions
Death Studies, 37: 768–781, 2013 Copyright # Taylor & Francis Group, LLC ISSN: 0748-1187 print=1091-7683 online DOI: 10.1080/07481187.2012.699908
Downloaded by [The University of Manchester Library] at 15:08 08 October 2014
FACTORS AFFECTING CODE STATUS IN A UNIVERSITY HOSPITAL INTENSIVE CARE UNIT LAUREN JODI VAN SCOY and MICHAEL SHERMAN Department of Internal Medicine and Division of Pulmonary, Critical Care and Sleep Medicine, Drexel University College of Medicine, Philadelphia, Pennsylvania, USA
The authors collected data on diagnosis, hospital course, and end-of-life preparedness in patients who died in the intensive care unit (ICU) with ‘‘full code’’ status (defined as receiving cardiopulmonary resuscitation), compared with those who didn’t. Differences were analyzed using binary and stepwise logistic regression. They found no differences in demographics, comorbidities, ventilator, hospital, or ICU days between groups. No-code patients were more likely to have higher APACHE-II scores (p < .0001), gastrointestinal=hepatic conditions (p < .01) and an advanced directive (p ¼ .03). Patients dying with full code status were more likely to have previously coded (p < .0001), and had more central lines (p ¼ .03). Implications are discussed.
With the geriatric population growing rapidly, predictions are that by the year 2050, 20.4% of the United States population will be over the age of 65 (Hobbs, 2008). Today’s physicians must therefore develop a skill set that will allow them to facilitate the inevitable end-of-life care issues that will arise as these patients age. Previous studies of patient preferences, expectations, and satisfaction with end-of-life care found a general preference for home-based, symptom-guided care (Davison, 2010; Lamont, 2005; Received 27 September 2011; accepted 9 May 2012. This study was funded by the Drexel University College of Medicine Aging Initiative Grant. We acknowledge Jesse Chittams, Dominique Williams and Cindy Liao of the Biostatics Service Center at the Drexel University School of Public Health Department of Epidemiology and Biostatistics for their assistance with data analysis. We also acknowledge Ms. Pamela Fried of Drexel University’s Academic Publishing Services for her editorial assistance. Address correspondence to Lauren Jodi Van Scoy, Department of Internal Medicine, Division of Pulmonary, Critical Care and Sleep Medicine, Drexel University College of Medicine, 245 N. 15th Street, Mail Stop 107, Philadelphia, PA 19102. E-mail: lvanscoy@ drexelmed.edu
768
Downloaded by [The University of Manchester Library] at 15:08 08 October 2014
Code Status
769
Steinhauser et al., 2000; The SUPPORT Principal Investigators, 1995). However, despite this well-established preference, most patients receive invasive, expensive medical interventions in the last days of their lives that often do not correspond with their stated preferences (Bell, Somogyi-Zalud, & Masaki, 2010; Lamont, 2005; The SUPPORT Principal Investigators, 1995). These studies and others suggest that there may be a gap between patients’ preferences and actual outcomes. Consequently, the need is great for physicians to understand what lies behind the wishes, differences, and biases that may exist in their patients concerning end-of-life planning and care. Because the majority of these end-of-life studies look at reported patient preferences rather than at actual outcomes, it is unclear if the stated preferences contribute directly to a patient’s level of care (i.e., code status) at the end of his or her life. We therefore conducted a retrospective chart review of 100 patients who died in the intensive care unit (ICU) in our urban university hospital to determine relevant factors that might impact the level of care provided at the end of life. We compared patients receiving aggressive measures at the end of life (full code) with those opting for ‘‘do not resuscitate’’ (no-code) status at the time of death. We investigated patient demographics, diagnosis, aggressiveness of hospital course, end-of-life preparedness, and relationships of proxies to determine if any of these factors affect code status at the time of death. Method Patient Selection Our protocol was reviewed and approved by the university’s Institutional Review Board. Hospital mortality lists that were generated by hospital administration mortality databases were reviewed from November 2006 to March 2008 in chronological order by date of death. Participants were eligible for inclusion in the study if they died in an ICU, including medical, neurological, cardiac, cardiothoracic, and surgical ICUs, after a minimum 72-hr stay. Patients were excluded if they were less than 18 years of age at death. Two groups of patients were identified. The first group consisted of patients who died in a ‘‘no-code’’ status, which was defined as not receiving cardiopulmonary resuscitation or advanced cardiac life
770
L. J. Van Scoy and M. Sherman
Downloaded by [The University of Manchester Library] at 15:08 08 October 2014
support protocol and as having do-not-resuscitate (DNR) or ‘‘withdrawal of life support’’ orders at the time of death. The second group was the ‘‘full code’’ group, defined as having had cardiopulmonary resuscitation and advanced cardiac life support protocol. Eligible patients were included in the study until we had approximately 50 patients per group who met the inclusion criteria. Variables We collected data in three broad categories: patient demographics, characteristics of hospital course, and end-of-life care preparedness. The first category, patient demographics and disease, included factors such as age, sex, race, diagnosis at death, and number of comorbid conditions. The second category, hospital course characteristics, included measures to quantify the aggressiveness of care while the patient was in the hospital. The number of central lines (central venous catheters), invasive procedures, units of blood products transfused, and surgical procedures were counted for the entire hospital stay. Each of these variables require an informed consent by the patient or proxy and therefore each measure represents a decision to pursue aggressive care. The number of central lines was divided by the length of stay to control for variable hospital durations. Units of blood products was chosen as a measure because the most critically ill patients often require blood product transfusion prior procedures due to the severity of their illness. The final category, end-of-life preparedness, identified the presence or absence of advance directives in the medical record. Advance directives included living wills and the presence of documents formally assigning proxies (such as a durable power of attorney) as well as the relationship of the proxy to the patient. We hypothesized that a more severely ill patient undergoing a long or aggressive hospital course, or their proxy decision makers witnessing the prolonged course, would have more time to come to terms with dying and thus decisions would be made to shift care toward nonlife-sustaining treatments by the time of death. Statistical Analysis Chi-square tests were used to compare categorical demographic variables, and t-tests were used to compare continuous and ordinal
Downloaded by [The University of Manchester Library] at 15:08 08 October 2014
Code Status
771
variables. For categorical variables, statistics included frequencies and percentages. For continuous variables, statistics included estimates of central tendency (means, medians) and measures of variability (standard deviations, interquartile ranges, ranges). Correlation matrixes were constructed for all continuous variables, and cross tabulations were run on all categorical variables. For all statistical tests the level of significance was set at the alpha level of .05. Summary statistics and cross tabulations were generated to assess the effect of a number of potential predictors on code status. These initial evaluations were followed by a formal statistical assessment using logistic regression. A series of univariate logistic regression analyses were conducted to assess the effect of each potential covariate independently on the code status. The univariate analyses were followed by a stepwise logistic regression to develop the final multiple logistic regression model consisting of the variables that were significant at the univariate level. A logistic regression was also used to assess the effect of measures of aggressiveness of hospital course on coding status. All analyses were performed using the SAS software package (SAS Institute Inc, 2009; Stokes, Davis, & Koch, 1995). Results A total of 100 study patients were analyzed. The full code group contained 48 patients and the no-code group contained 52 patients. All but one patient was designated as full code on arrival at the hospital. Of the patients in the full code group, there were no changes in code status at any time during their hospitalization. For the patients in the no-code group, there was a mean of 1.44 (0.63) changes in code status throughout the hospital stay, and all of the changes involved a de-escalation in the level of care. Patient Demographics and Disease Table 1 shows the characteristics of the study participants. There were no significant differences in age, gender, race, comorbidities, APACHE II disease severity score at hospital arrival, hospital length of stay, and intensive care length of stay, unit type, or number of ventilator days between the code and no-code groups. The
772
L. J. Van Scoy and M. Sherman
TABLE 1 Description of Study Subjects
Downloaded by [The University of Manchester Library] at 15:08 08 October 2014
Demographics Age (years), M (SD) Demographic factor Gender Male Female Race African American White Other Disease type responsible for deatha Sepsis Cardiovascular Pulmonary Gastrointestinal or hepatic Cancer Neurological Other Unit type Medical=medical step-down Cardiothoracic surgery, surgical, and trauma unit Cardiac care unit Hospital course characteristics Comorbidity APACHE II at hospital arrival Hospital days Unit days Ventilator days
All groups (n ¼ 100)
Code group (n ¼ 48)
No-code group (n ¼ 52)
p value
62.5 (14.4) n, (%)
60.8 (14.6) n, (%)
64.1 (14.2) n, (%)
.25
58 (58) 42 (42)
28 (58) 20 (42)
30 (58) 22 (42)
43 (43)
23 (48)
20 (38)
49 (49) 8 (8) n (%)b
20 (42) 5 (10) n (%)b
29 (56) 3 (6) n (%)b
41 17 6 11
(41) (17) (6) (11)
24 10 2 1
(59) (59) (33) (9)
17 7 4 10
(41) (41) (67) (91)
.27 .47 .41
