- Email: [email protected]
Analysis of limited resuscitations in patients suffering in-hospital cardiac arrest
Resuscitation 80 (2009) 985–989
Contents lists available at ScienceDirect
Resuscitation journal homepage: www.elsevier.com/locate/resuscitation
Clinical paper
Analysis of limited resuscitations in patients suffering in-hospital cardiac arrest夽 Kristofer Dosh a,b,c,∗ , Abhijeet Dhoble a,b , Rudolph Evonich a,b,d , Amit Gupta a,b,e , Ibrahim Shah a,b,d , Joseph Gardiner b,f , Francesca C. Dwamena a,b a
Department of Internal Medicine, Michigan State University, East Lansing, MI, United States College of Human Medicine, Michigan State University, East Lansing, MI, United States c Gill Heart Institute and Division of Cardiovascular Medicine, University of Kentucky, Lexington, KY, United States d Division of Cardiology, William Beaumont Hospital, Royal Oak, MI, United States e Division of Pulmonary and Critical Care Medicine, University of Iowa, Iowa City, IA, United States f Department of Epidemiology, Michigan State University, East Lansing, MI, United States b
a r t i c l e
i n f o
Article history: Received 12 March 2009 Received in revised form 17 April 2009 Accepted 4 May 2009 Keywords: Limited code Full code Survival Cardiac arrest
a b s t r a c t Background: Although clinicians are expected to help patients make decisions about end-of-life care, there is insufficient data to help guide patient preferences. The objective of this study was to determine the frequency of patients who undergo ‘limited code’ and compare survival to discharge with those who undergo maximum resuscitative efforts (‘full code’). Methods: We performed a retrospective analysis of all adult in-hospital cardiac arrests (IHCA) at a tertiary care teaching hospital from January 1999 to December 2003 to compare survival in patients with limited code to survival in patients with a full code. We collected data on demographic and clinical variables known to influence survival in IHCA. Logistic regression was used to assess the association of code status with subsequent survival through the code and to hospital discharge after adjusting for potential confounding factors. Results: Of the 309 patients having IHCA, there were 17 (5.5%) patients with limited code status and 292 (94.5%) with full code status. Among full code patients, 171 (58.6%) survived the code compared to five patients (29.4%) who had a limited code (p = 0.023). After adjusting for demographic variables and prearrest co-morbidities, patients with full code status compared to limited code status had an odds ratio for return of spontaneous circulation of 3.69 (95% CI: 1.13–14.34). Conclusions: Patients who opt for limited code have a significantly lower probability of survival compared to patients who choose full code. Patients who choose limited code should be informed of the likely negative outcome as compared to full resuscitation. © 2009 Elsevier Ireland Ltd. All rights reserved.
1. Introduction At many institutions, including ours, discussions about advance directives are expected to occur at the time of admission. In these discussions, patients choose to accept, limit, or decline cardio-pulmonary resuscitation (CPR). These choices, recorded as ‘full code’, ‘limited code’, or ‘do not resuscitate’ respectively, are expected to dictate physician behavior in the event of in-hospital cardiac arrest (IHCA). Yet, few analytical studies have been performed to address these options and its outcome, particularly the choice of limited code.1,2
夽 A Spanish translated version of the abstract of this article appears as Appendix in the final online version at doi:10.1016/j.resuscitation.2009.05.011. ∗ Corresponding author at: Gill Heart Institute and Division of Cardiovascular Medicine, 900 S. Limestone, 326 CTW Bldg., Lexington, Kentucky, 40536-0200, United States. Tel.: +1 859 323 8040; fax: +1 859 323 6475. E-mail address: [email protected] (K. Dosh). 0300-9572/$ – see front matter © 2009 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.resuscitation.2009.05.011
Patient preferences regarding CPR have been shown to be influenced by inadequate information.3,4 For example, 41% of 371 elderly patients endorsed CPR in the event of demise during an acute illness.3 After being informed of the low probability of survival (10–17%), the rate of endorsement dropped to 21%. Patients will benefit from descriptive data about both choices of full codes and limited codes. The ethics of limited codes, particularly in the context of physician decision-making, has long been debated in the medical literature.5,6 However, only one study1 published almost a decade ago has looked at outcomes of limited codes in cases where patients have endorsed them in their advance directives. In that single institution, 6 (16%) of 37 patients who chose limited codes survived the initial event, but none survived to discharge. This study has never been replicated. Moreover, there have been no studies of the frequency of limited codes relative to full codes or of the correlates of survival in patients who opt for limited codes. The purpose of our study was to determine the frequency of limited codes in a tertiary care hospital, and to compare survival of
986
K. Dosh et al. / Resuscitation 80 (2009) 985–989
patients who choose limited codes with survival of patients who do not direct any limitations of CPR. We also sought to identify clinical correlates of return of spontaneous circulation (ROSC) and survival to discharge in all patients who underwent CPR during the study period. 2. Methods 2.1. Study setting and patient population The study was conducted in a 687-bed tertiary care communitybased teaching hospital in mid-Michigan. CPR in this hospital is performed by a designated code team. The code team consists of one to three residents (from internal medicine, family medicine, and/or emergency medicine), one anesthesiologist, a critical care nurse, respiratory therapist, and laboratory personnel). All are certified in advanced cardiac life support (ACLS). The code team is activated by voice and overhead paging after the code button is pushed by anyone witnessing a cardio-pulmonary arrest. Using a standardized form, a member of the code team, usually a nurse, records demographic and clinical data including date, time, location, type of arrest (respiratory, cardiac, or both), start and end time, patient survival (yes/no), and all interventions, including drugs. This form is filed in the patient’s chart. A duplicate copy is subsequently submitted to the hospital code committee for review. In addition, hospital operators log all codes announced over the intercom system in a dedicated log book and the central supply office bills appropriate patients for all code supplies utilized. To identify all patients who suffered IHCA from 1 January 1999 to 31 December 2003, we reviewed records of the hospital code committee, the hospital operator’s log, and central supply billings. In addition, we queried the hospital database for a discharge diagnosis of cardiac arrest during the study period. Patients were included in the study if they had documentation in the medical record of an IHCA defined as (1) absence of palpable pulse, (2) unresponsiveness due to any cause and (3) apnea, agonal respirations, or mechanical ventilation.7 Exclusion criteria were: (1) IHCA in the emergency department, operating room, or cardiac catheterization lab, where codes are activated and run by different personnel than in the rest of the hospital; (2) less than 18 years of age; (3) repeat IHCA. All consecutive patients with IHCA meeting inclusion criteria during the study period were included. 2.2. Study design A retrospective study was performed using medical record data of patients undergoing CPR for IHCA. The study was approved by the institutional review board at the Michigan State University. This manuscript is written in accordance with the ‘Strengthening the Reporting of Observational Studies in Epidemiology’ (STROBE) statement.8 2.3. Data collection Demographic data were obtained from the patient’s chart and included date of birth, gender, race, marital status, and residence. Admission data included the following: date of admission, bed type (critical care bed/telemetry bed/general medical bed), diagnosis at admission, past medical history, tobacco use, functional status (independent with activities of daily living or dependent), admitting attending physician’s specialty, and whether the patient was a resident teaching case. We defined limited code as resuscitation with advance directives prohibiting one or more of the following: intubation, chest compression, electrical defibrillation, or ACLS medications.1,5,6 If no advance directive was noted in the chart, the resuscitation was considered to be full code.
Physiologic variables recorded at time of admission were: height, weight, body mass index (BMI), systolic blood pressure, diastolic blood pressure, mean arterial pressure, temperature, heart rate, respiratory rate, arterial pH, pCO2, pO2, white blood count, hematocrit, BUN, creatinine, sodium, potassium, bicarbonate, glucose, albumin, bilirubin, whether or not mechanically ventilated, FiO2, Glasgow Coma Scale, prognosis-after-resuscitation (PAR) score9 and APACHE II score.10 Hospitalization data included: diagnosis using Diagnosis Related Group (DRG), any surgery, and the above-mentioned physiologic parameters within 24 h of cardiac arrest. The following information was obtained from the code team’s standardized form: date and time when code started and ended, location (critical care bed, telemetry bed, general medical bed without telemetry monitoring), physician directing code, witnessed or un-witnessed arrest, initial rhythm, pulse, blood pressure, subsequent rhythms, medications administered (time and dose), and procedures (defibrillation, ventilation, intubation, chest compressions, pacing, central line placement, pericardiocentesis, sternotomy, thoracotomy). Outcome of the code was recorded as return of spontaneous circulation (ROSC) or death along with the time when the code was called off. We used a modified code point scoring system as described by Saklayen et al.,2 to assess the intensity of resuscitative effort; one point was given for each attempt at defibrillation or intubation, and another point for each dose of medication given. In addition, one point each was given for any attempt to place a central line, pericardiocentesis, and thoracotomy or sternotomy. The hospital chart was then reviewed to determine the date of death or hospital discharge. The data were recorded by seven reviewers who were residents in Internal Medicine. Ten percent of the total charts were randomly reviewed by two independent reviewers to check for inter-rater accuracy. Accuracy in the recording exceeded 90%. 2.4. Outcome measures Main outcome measures for the study were ROSC and survival to hospital discharge. 2.5. Statistical analysis Descriptive statistics were computed as means and standard deviations for continuous variables, and as frequencies for categorical variables. Comparisons of baseline characteristics between patients under full code and limited code were performed using t-tests and 2 -tests as appropriate. Logistic regression was used to assess the association of code status with subsequent survival and adjust for potential confounding factors. In addition to variables with known association with ROSC and survival, we adjusted for factors that differed between the two patients groups (p < 0.05). The former included intubation, defibrillation, medications, and procedures (central access attempted, pericardiocentesis, thoracotomy, or sternotomy). Results are reported as odds ratios with 95% confidence intervals. Exact statistical tests (exact likelihood ratio tests) and methods (exact logistic regression) were used where applicable. A p-value < 0.05 was considered statistically significant. Analyses were conducted with SAS Software, version 9.1 (SAS Institute Inc., Cary, NC, USA). 3. Results During the period under review, January 1999 to December 2003, 408 patient records were reviewed based on initial search criteria. Ninety-nine of these were excluded including 34 not meeting the definition of cardiac arrest, 22 with absent or grossly insufficient code documentation in the chart, 10 due to code occurring in the
K. Dosh et al. / Resuscitation 80 (2009) 985–989
987
Table 1 Primary hospital diagnoses for all patients suffering cardiac arrest.
Table 3 Characteristics of patients by code status.
Primary diagnosis
Patients, no. (%) (N = 309)
Characteristic
Limited code
Full code
p-value
Cardiovascular disease Cardiac surgery Vascular disease (peripheral vascular disease, aortic aneurysm, aortic dissection, aortic rupture) Congestive heart failure, myocardial infarction, unstable angina, coronary artery disease, valvular heart disease Arrhythmia (atrial fibrillation, ventricular tachycardia) Other (chest pain, cardiac tamponade, pericarditis)
141 (100) 25 (17.7) 7 (5.0)
Male gender (%)
35.3
49.7
0.32
Race (%) White Other Missing
82.4 5.9 11.8
63.7 21.2 15.1
0.19
Marital status (%) Married Other Missing
64.7 35.3 0
48.6 50.5 1.4
0.35
Functional status (%) Independent Dependent Missing
52.9 23.5 23.5
54.8 27.1 18.2
0.90
Age, mean (SD), years
75.2 (16.1)
67.8 (14.6)
0.08
BMI, mean (SD) (kg/m2 )
21.9 (6.6)
29.1 (10.2)
0.002
PAR, mean (SD)
9.1 (5.1)
5.6 (4.7)
0.01
APACHE II, mean (SD)
16.5 (6.7)
12.1 (6.0)
0.03
Witnessed code (%) Yes No Missing
82.4 11.8 5.9
74.3 18.2 7.5
0.77
Location (%) Unit Tele Other Missing
29.4 41.2 23.5 5.9
31.5 39.7 27.1 1.7
0.87
11.8 41.2 47.1
21.6 43.8 33.9 0.7
0.55
98 (69.5)
7 (5.0) 4 (2.8)
End-stage disease Metastatic cancer, hematologic malignancy Chronic obstructive pulmonary disease, acute respiratory failure Liver failure Renal failure
37 (100) 20 (54.1) 4 (10.8)
Infectious disease Abscess, cellulitis, endocarditis, pancreatitis, pneumonia Sepsis Other (fever of unknown origin)
65 (100) 34 (52.3)
Neurological disease Closed head injury Seizure Intracranial bleed Stroke, transient ischemic attack
16 (100) 4 (25.0) 3 (18.8) 5 (31.2) 4 (25.0)
Miscellaneous Gastrointestinal bleed Genitourinary surgery Orthopedic surgery Bowel obstruction General trauma Pulmonary embolism Other blood dyscrasia (anemia, thrombocytopenia, pancytopenia) Other
50 (100) 11 (22.0) 4 (8.0) 5 (10.0) 5 (10.0) 2 (4.0) 2 (4.0) 4 (8.0)
4 (10.8) 9 (24.3)
30 (46.2) 1 (1.5)
17 (34.0)
catheterization laboratory, operating room, or emergency department, 6 due to code occurring prior to hospital admission, 22 in which no chart could be found, and 5 as a result of family stopping the code early. This left 309 patients that met inclusion and exclusion criteria, of which 292 (94.5%) were full code and 17 (5.5%) were limited code. Primary hospital diagnoses for the entire group and limited code group are listed in Tables 1 and 2 respectively. Patient characteristics are compared between limited code and full code groups in Table 3. There were no significant differences between groups on age, race, gender, marital status and functional status. On average, full code patients had larger body size than limited code patients (mean BMI 29.1 kg/m2 vs. mean BMI 21.9 kg/m2 , p = 0.002). Compared to limited code patients, full code patients had significantly Table 2 Primary hospital diagnoses for all limited code patients. Primary diagnosis
Patients no. (%) (N = 17)
Cardiovascular disease: congestive heart failure, myocardial infarction End-stage disease Metastatic cancer Chronic obstructive pulmonary disease, acute respiratory failure Renal failure Infectious disease Pneumonia Sepsis Other (vertebral fracture)
7 (41) 4 (24) 1 1 2 5 (29) 2 3 1(6)
Initial rhythm (%) Pulseless, VT/VF Pulseless, non-VT/VF Other Missing Code points, mean (SD)
5.9 (6.2)
8.5 (5.9)
0.11
Code duration in minutes, mean (SD)
23.6 (22.8)
27.8 (49.8)
0.51
lower PAR score (5.6 vs. 9.1, p = 0.01) and lower APACHE II score (12.1 vs. 16.5, p = 0.03). Overall, 171 (58.6%) full code patients had ROSC compared to only 5 (29.4%) of the limited code patients (p = 0.023). For survival to discharge, 67 (23.0%) patients survived in the full code group versus only 1 (5.9%) patient in the limited code group. This difference was not statistically significant (p = 0.095). Significant correlates of ROSC were functional status (p = 0.012), initial rhythm (p = 0.005) and code points (p < 0.0001). Pulseless VT/VF and any other rhythm were each associated with higher odds of survival relative to other pulseless rhythms (non-VT/VF). The unadjusted association of code status with ROSC was obtained using exact logistic regression. The unadjusted estimated OR for survival (full code vs. limited code) was 3.38 (95% CI: 1.07–12.57, p = 0.035). Adjusted and unadjusted correlates of survival during the code are shown in Table 4. 4. Discussion Our study confirms the results of Dumot et al.1 and suggests the futility of limited codes. Of the 17 patients who opted for limited code, only 5 survived the code with ROSC, and of these, only one survived to discharge. Overall survival in full code patients in this study was comparable at 23% with other studies.1,2 This is the first study to compare survival in full code patients with limited code after controlling for clinical confounders. Comparison of clinical variables between the full and limited code patients revealed a higher PAR9 and APACHE II10 score and lower BMI for limited code
988
K. Dosh et al. / Resuscitation 80 (2009) 985–989
Table 4 Correlates of return of spontaneous circulation. Characteristic
Odds ratio (95% CI) [unadjusted]
Odds ratio (95% CI) [adjusted]
Code status Limited code Full code
1.00 [Reference] 3.38 (1.07, 12.57)
3.69 (1.13, 14.34)
Functional status Dependent Independent
1.00 [Reference] 1.95 (1.11, 3.45)
1.79 (1.03, 3.11)
Initial rhythm Pulseless, non-VT/VF Pulseless, VT/VF Other
1.00 [Reference] 2.23 (1.16, 4.36) 2.07 (1.19, 3.62)
2.11 (1.06, 4.32) 1.97 (1.09, 3.60)
Code points
0.86 (0.82, 0.90)a
0.84 (0.79, 0.89)
PAR
1.01 (0.97, 1.06)a
APACHE II
0.98 (0.95, 1.02)a
BMI
1.02 (0.99, 1.05)a
a
OR for 1 unit increase; PAR = prognosis-after-resuscitation, BMI = body mass index (kg/m2 ).
patients. The significance of this finding is uncertain but might suggest a generally more feeble condition of those in the limited code group. However, after adjusting for confounding variables, limited code status remained a predictor of poor outcomes. We found a significant disadvantage of limited code in ROSC and a similar trend in survival. Prior literature has demonstrated that explaining the probability of survival in CPR clearly affects patient preferences about code status.3,4 The significant differences in survival between limited and full code patients as determined by our study may impact patient preferences and would have a likely effect of patients opting for a full or no code rather than a limited code. The results of our study and previous literature should stimulate further discussion with the patient regarding their motivations and expectations of resuscitation in the event of in-hospital cardiac arrests. We expect this will lead to better decision-making. It is unclear whether the outcomes in limited code patients may have been affected by generally sub-standard treatment of their other medical conditions. Although literature on limited code patients is lacking, some correlates may be drawn from studies on patients selecting do-not-attempt-resuscitation (DNAR) status. Wenger et al.11 performed a retrospective review of 12,821 patients and compared in-hospital and 6-month mortality for those with and without DNAR orders. After adjusting for baseline demographic variables, co-morbid conditions, and hospital characteristics, patients with DNAR orders had an increased in-hospital (40% vs. 9%, p < 0.001) and 6-month (61% vs. 26%, p < 0.001) mortality compared to those without DNAR orders. More recently, Cohen et al. found that patients with DNAR orders are less likely to be admitted to an intensive care unit irrespective of baseline functional status and severity-of-illness.12 Ethical arguments support the notion that a DNAR order should not be interpreted as an order to withhold other standard therapies for a patient.13 But these results suggest that physicians and other healthcare providers may treat DNAR patients differently than full code patients. In this trial, we did not assess deviations from standard-of-care treatment of the patients’ admitting diagnoses. But it is possible that the limited code patients received less than standard care during their hospitalization prior to cardiac arrest. This effect might bias our results toward a worse prognosis for the limited code patients. Our analysis was structured to control for baseline and within-code variables that have been previously shown to predict outcomes in cardiac arrest.9,14–22 Although we believe this was
a reasonable method to control for confounding factors, this and other unrecognized bias may have had an effect on our outcomes. Moreover, application of this study’s results to patient care will require consideration of physicians’ personal preferences and opinions about CPR in general23 and limited codes in particular. Research has shown that physicians are often unaware of their patients’ resuscitation preferences24 and that advance directives may have limited to no effect on the documentation and implementation of DNAR requests.25 One prior study demonstrated that life-sustaining therapy in critically ill patients is often applied, withdrawn, or withheld in spite of patients’ or surrogates’ wishes.26 Such conflicts between physician decisions and patient preferences occur in spite of good intentions because physicians are often poor at predicting patients’ desires with regards to CPR27,28 and also because of differences between physician and public knowledge about life-sustaining measures.29 In this study, overall code intensity as measured by code points and code duration (Table 3) was found to be similar between limited code and full code patients. This would suggest that physicians put equal effort into running codes in each respective group. But physician notions about the futility of limited codes or their personal preferences might also have had an impact on outcomes that we were unable to measure. 4.1. Limitations Retrospective analysis may be felt to be a limitation by some, but a randomized controlled study of limited versus full codes could never be performed for obvious reasons. Furthermore a prospective study would require informed consent at time of admission and informed consent would likely introduce bias. Our goal was to determine patient preferences for code status when elicited by a clinician, not a researcher, and how those preferences affect survival. In fact, we believe that retrospective studies of existing data are better suited for this type of study to produce unbiased results. 5. Conclusions This study has proved that the ‘limited code’ status has a survival disadvantage to ‘full code’ status. We do respect patients’ preferences regarding their decisions about end-of-life directives but suspect that many of them choose limited code status without understanding the full implications of their decision. We hope the data presented in this paper will provide more evidence for physicians to counsel patients about this important issue. We did not measure physicians’ confidence in running limited resuscitation codes versus full codes, and believe that it may affect outcomes to a certain extent. Further research is warranted at multiple high volume centers for further investigation of this issue. Future studies are also needed to determine if patients who initially request a limited resuscitation change their directives after knowledge of outcome data. Conflict of interest None of the authors have any conflict of interest with regards to this manuscript. Acknowledgments We are very thankful for the help provided by Julie Bey, Suzanne Leialoha, and Stacy Near as well as Arman Raza, MD, Amir Azeem, MD, Kwsai Al-Rahhal, MD, and Dwarakraj Soundarraj, MD who assisted by collecting the data for this study. We also thank Lin Liu, PhD for her help with data analysis. All authors had full access to all data in the study and take responsibility for the integrity
K. Dosh et al. / Resuscitation 80 (2009) 985–989
of the data and the accuracy of the data analysis. This study was funded by an unrestricted educational grant from Wyeth Pharmaceuticals. Wyeth Pharmaceuticals had no involvement in the study design, data collection, analysis and interpretation of data, writing of the manuscript, or decision to submit the manuscript for publication. References 1. Dumot JA, Burval DJ, Sprung J, et al. Outcome of adult cardiopulmonary resuscitations at a tertiary referral center including results of “limited” resuscitations. Arch Intern Med 2001;161:1751–8. 2. Saklayen M, Liss H, Markert R. In-hospital cardiopulmonary resuscitation. Survival in 1 hospital and literature review. Medicine (Baltimore) 1995;74:163–75. 3. Murphy DJ, Burrows D, Santilli S, et al. The influence of the probability of survival on patients’ preferences regarding cardiopulmonary resuscitation. N Engl J Med 1994;330:545–9. 4. Schonwetter RS, Walker RM, Kramer DR, Robinson BE. Resuscitation decision making in the elderly: the value of outcome data. J Gen Intern Med 1993;8:295–300. 5. Fowler MD. Slow code, partial code, limited code. Heart Lung 1989;18:533–4. 6. Ross JW, Pugh D. Limited cardiopulmonary resuscitation: the ethics of partial codes. QRB Qual Rev Bull 1988;14:4–8. 7. Cummins RO, Chamberlain D, Hazinski MF, et al. Recommended guidelines for reviewing, reporting, and conducting research on in-hospital resuscitation: the in-hospital ‘Utstein style’. A statement for healthcare professionals from the American Heart Association, the European Resuscitation Council, the Heart and Stroke Foundation of Canada, the Australian Resuscitation Council, and the Resuscitation Councils of Southern Africa. Resuscitation 1997;34:151–83. 8. Vandenbroucke JP, von Elm E, Altman DG, et al. Strengthening the Reporting of Observational Studies in Epidemiology (STROBE): explanation and elaboration. PLoS Med 2007;4:e297. 9. O’Keeffe S, Ebell MH. Prediction of failure to survive following in-hospital cardiopulmonary resuscitation: comparison of two predictive instruments. Resuscitation 1994;28:21–5. 10. Knaus WA, Draper EA, Wagner DP, Zimmerman JE. APACHE II: a severity of disease classification system. Crit Care Med 1985;13:818–29. 11. Wenger NS, Pearson ML, Desmond KA, Brook RH, Kahn KL. Outcomes of patients with do-not-resuscitate orders. Toward an understanding of what donot-resuscitate orders mean and how they affect patients. Arch Intern Med 1995;155:2063–8. 12. Cohen RI, Lisker GN, Eichorn A, Multz AS, Silver A. The impact of do-notresuscitate order on triage decisions to a medical intensive care unit. J Crit Care 2009;24:311–5.
989
13. Martin RD, Cohen MA, Weiss Roberts L, Batista SM, Hicks D, Bourgeois J. DNR versus DNT: clinical implications of a conceptual ambiguity: a case analysis. Psychosomatics 2007;48:10–5. 14. Andreasson AC, Herlitz J, Bang A, et al. Characteristics and outcome among patients with a suspected in-hospital cardiac arrest. Resuscitation 1998;39:23–31. 15. Cohn EB, Lefevre F, Yarnold PR, Arron MJ, Martin GJ. Predicting survival from inhospital CPR: meta-analysis and validation of a prediction model. J Gen Intern Med 1993;8:347–53. 16. de Vos R, Koster RW, De Haan RJ, Oosting H, van der Wouw PA, LampeSchoenmaeckers AJ. In-hospital cardiopulmonary resuscitation: prearrest morbidity and outcome. Arch Intern Med 1999;159:845–50. 17. Ebell MH. Prearrest predictors of survival following in-hospital cardiopulmonary resuscitation: a meta-analysis. J Fam Pract May 1992;34:551–8. 18. Ebell MH. Artificial neural networks for predicting failure to survive following in-hospital cardiopulmonary resuscitation. J Fam Pract Mar 1993;36:297–303. 19. Ebell MH, Smith M, Kruse JA, Drader-Wilcox J, Novak J. Effect of race on survival following in-hospital cardiopulmonary resuscitation. J Fam Pract Jun 1995;40:571–7. 20. Ebell MH, Kruse JA, Smith M, Novak J, Drader-Wilcox J. Failure of three decision rules to predict the outcome of in-hospital cardiopulmonary resuscitation. Med Decis Making 1997;17:171–7. 21. George Jr AL, Folk IIIrd BP, Crecelius PL, Campbell WB. Pre-arrest morbidity and other correlates of survival after in-hospital cardiopulmonary arrest. Am J Med Jul 1989;87:28–34. 22. Marwick TH, Case CC, Siskind V, Woodhouse SP. Prediction of survival from resuscitation: a prognostic index derived from multivariate logistic model analysis. Resuscitation 1991;22:129–37. 23. Marik PE, Varon J, Lisbon A, Reich HS. Physicians’ own preferences to the limitation and withdrawal of life-sustaining therapy. Resuscitation 1999;42:197–201. 24. A controlled trial to improve care for seriously ill hospitalized patients. The study to understand prognoses and preferences for outcomes and risks of treatments (SUPPORT). The SUPPORT Principal Investigators. JAMA 1995;274:1591–8. 25. Teno J, Lynn J, Wenger N, et al. Advance directives for seriously ill hospitalized patients: effectiveness with the patient self-determination act and the SUPPORT intervention SUPPORT Investigators. Study to Understand Prognoses and Preferences for Outcomes and Risks of Treatment. J Am Geriatr Soc 1997;45:500–7. 26. Asch DA, Hansen-Flaschen J, Lanken PN. Decisions to limit or continue lifesustaining treatment by critical care physicians in the United States: conflicts between physicians’ practices and patients’ wishes. Am J Respir Crit Care Med 1995;151:288–92. 27. Uhlmann RF, Pearlman RA, Cain KC. Physicians’ and spouses’ predictions of elderly patients’ resuscitation preferences. J Gerontol 1988;43:M115–121. 28. Bedell SE, Delbanco TL. Choices about cardiopulmonary resuscitation in the hospital. When do physicians talk with patients? N Engl J Med 1984;310:1089–93. 29. Carmel S. Life-sustaining treatments: what doctors do, what they want for themselves and what elderly persons want. Soc Sci Med Nov 1999;49:1401–8.
Contents lists available at ScienceDirect
Resuscitation journal homepage: www.elsevier.com/locate/resuscitation
Clinical paper
Analysis of limited resuscitations in patients suffering in-hospital cardiac arrest夽 Kristofer Dosh a,b,c,∗ , Abhijeet Dhoble a,b , Rudolph Evonich a,b,d , Amit Gupta a,b,e , Ibrahim Shah a,b,d , Joseph Gardiner b,f , Francesca C. Dwamena a,b a
Department of Internal Medicine, Michigan State University, East Lansing, MI, United States College of Human Medicine, Michigan State University, East Lansing, MI, United States c Gill Heart Institute and Division of Cardiovascular Medicine, University of Kentucky, Lexington, KY, United States d Division of Cardiology, William Beaumont Hospital, Royal Oak, MI, United States e Division of Pulmonary and Critical Care Medicine, University of Iowa, Iowa City, IA, United States f Department of Epidemiology, Michigan State University, East Lansing, MI, United States b
a r t i c l e
i n f o
Article history: Received 12 March 2009 Received in revised form 17 April 2009 Accepted 4 May 2009 Keywords: Limited code Full code Survival Cardiac arrest
a b s t r a c t Background: Although clinicians are expected to help patients make decisions about end-of-life care, there is insufficient data to help guide patient preferences. The objective of this study was to determine the frequency of patients who undergo ‘limited code’ and compare survival to discharge with those who undergo maximum resuscitative efforts (‘full code’). Methods: We performed a retrospective analysis of all adult in-hospital cardiac arrests (IHCA) at a tertiary care teaching hospital from January 1999 to December 2003 to compare survival in patients with limited code to survival in patients with a full code. We collected data on demographic and clinical variables known to influence survival in IHCA. Logistic regression was used to assess the association of code status with subsequent survival through the code and to hospital discharge after adjusting for potential confounding factors. Results: Of the 309 patients having IHCA, there were 17 (5.5%) patients with limited code status and 292 (94.5%) with full code status. Among full code patients, 171 (58.6%) survived the code compared to five patients (29.4%) who had a limited code (p = 0.023). After adjusting for demographic variables and prearrest co-morbidities, patients with full code status compared to limited code status had an odds ratio for return of spontaneous circulation of 3.69 (95% CI: 1.13–14.34). Conclusions: Patients who opt for limited code have a significantly lower probability of survival compared to patients who choose full code. Patients who choose limited code should be informed of the likely negative outcome as compared to full resuscitation. © 2009 Elsevier Ireland Ltd. All rights reserved.
1. Introduction At many institutions, including ours, discussions about advance directives are expected to occur at the time of admission. In these discussions, patients choose to accept, limit, or decline cardio-pulmonary resuscitation (CPR). These choices, recorded as ‘full code’, ‘limited code’, or ‘do not resuscitate’ respectively, are expected to dictate physician behavior in the event of in-hospital cardiac arrest (IHCA). Yet, few analytical studies have been performed to address these options and its outcome, particularly the choice of limited code.1,2
夽 A Spanish translated version of the abstract of this article appears as Appendix in the final online version at doi:10.1016/j.resuscitation.2009.05.011. ∗ Corresponding author at: Gill Heart Institute and Division of Cardiovascular Medicine, 900 S. Limestone, 326 CTW Bldg., Lexington, Kentucky, 40536-0200, United States. Tel.: +1 859 323 8040; fax: +1 859 323 6475. E-mail address: [email protected] (K. Dosh). 0300-9572/$ – see front matter © 2009 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.resuscitation.2009.05.011
Patient preferences regarding CPR have been shown to be influenced by inadequate information.3,4 For example, 41% of 371 elderly patients endorsed CPR in the event of demise during an acute illness.3 After being informed of the low probability of survival (10–17%), the rate of endorsement dropped to 21%. Patients will benefit from descriptive data about both choices of full codes and limited codes. The ethics of limited codes, particularly in the context of physician decision-making, has long been debated in the medical literature.5,6 However, only one study1 published almost a decade ago has looked at outcomes of limited codes in cases where patients have endorsed them in their advance directives. In that single institution, 6 (16%) of 37 patients who chose limited codes survived the initial event, but none survived to discharge. This study has never been replicated. Moreover, there have been no studies of the frequency of limited codes relative to full codes or of the correlates of survival in patients who opt for limited codes. The purpose of our study was to determine the frequency of limited codes in a tertiary care hospital, and to compare survival of
986
K. Dosh et al. / Resuscitation 80 (2009) 985–989
patients who choose limited codes with survival of patients who do not direct any limitations of CPR. We also sought to identify clinical correlates of return of spontaneous circulation (ROSC) and survival to discharge in all patients who underwent CPR during the study period. 2. Methods 2.1. Study setting and patient population The study was conducted in a 687-bed tertiary care communitybased teaching hospital in mid-Michigan. CPR in this hospital is performed by a designated code team. The code team consists of one to three residents (from internal medicine, family medicine, and/or emergency medicine), one anesthesiologist, a critical care nurse, respiratory therapist, and laboratory personnel). All are certified in advanced cardiac life support (ACLS). The code team is activated by voice and overhead paging after the code button is pushed by anyone witnessing a cardio-pulmonary arrest. Using a standardized form, a member of the code team, usually a nurse, records demographic and clinical data including date, time, location, type of arrest (respiratory, cardiac, or both), start and end time, patient survival (yes/no), and all interventions, including drugs. This form is filed in the patient’s chart. A duplicate copy is subsequently submitted to the hospital code committee for review. In addition, hospital operators log all codes announced over the intercom system in a dedicated log book and the central supply office bills appropriate patients for all code supplies utilized. To identify all patients who suffered IHCA from 1 January 1999 to 31 December 2003, we reviewed records of the hospital code committee, the hospital operator’s log, and central supply billings. In addition, we queried the hospital database for a discharge diagnosis of cardiac arrest during the study period. Patients were included in the study if they had documentation in the medical record of an IHCA defined as (1) absence of palpable pulse, (2) unresponsiveness due to any cause and (3) apnea, agonal respirations, or mechanical ventilation.7 Exclusion criteria were: (1) IHCA in the emergency department, operating room, or cardiac catheterization lab, where codes are activated and run by different personnel than in the rest of the hospital; (2) less than 18 years of age; (3) repeat IHCA. All consecutive patients with IHCA meeting inclusion criteria during the study period were included. 2.2. Study design A retrospective study was performed using medical record data of patients undergoing CPR for IHCA. The study was approved by the institutional review board at the Michigan State University. This manuscript is written in accordance with the ‘Strengthening the Reporting of Observational Studies in Epidemiology’ (STROBE) statement.8 2.3. Data collection Demographic data were obtained from the patient’s chart and included date of birth, gender, race, marital status, and residence. Admission data included the following: date of admission, bed type (critical care bed/telemetry bed/general medical bed), diagnosis at admission, past medical history, tobacco use, functional status (independent with activities of daily living or dependent), admitting attending physician’s specialty, and whether the patient was a resident teaching case. We defined limited code as resuscitation with advance directives prohibiting one or more of the following: intubation, chest compression, electrical defibrillation, or ACLS medications.1,5,6 If no advance directive was noted in the chart, the resuscitation was considered to be full code.
Physiologic variables recorded at time of admission were: height, weight, body mass index (BMI), systolic blood pressure, diastolic blood pressure, mean arterial pressure, temperature, heart rate, respiratory rate, arterial pH, pCO2, pO2, white blood count, hematocrit, BUN, creatinine, sodium, potassium, bicarbonate, glucose, albumin, bilirubin, whether or not mechanically ventilated, FiO2, Glasgow Coma Scale, prognosis-after-resuscitation (PAR) score9 and APACHE II score.10 Hospitalization data included: diagnosis using Diagnosis Related Group (DRG), any surgery, and the above-mentioned physiologic parameters within 24 h of cardiac arrest. The following information was obtained from the code team’s standardized form: date and time when code started and ended, location (critical care bed, telemetry bed, general medical bed without telemetry monitoring), physician directing code, witnessed or un-witnessed arrest, initial rhythm, pulse, blood pressure, subsequent rhythms, medications administered (time and dose), and procedures (defibrillation, ventilation, intubation, chest compressions, pacing, central line placement, pericardiocentesis, sternotomy, thoracotomy). Outcome of the code was recorded as return of spontaneous circulation (ROSC) or death along with the time when the code was called off. We used a modified code point scoring system as described by Saklayen et al.,2 to assess the intensity of resuscitative effort; one point was given for each attempt at defibrillation or intubation, and another point for each dose of medication given. In addition, one point each was given for any attempt to place a central line, pericardiocentesis, and thoracotomy or sternotomy. The hospital chart was then reviewed to determine the date of death or hospital discharge. The data were recorded by seven reviewers who were residents in Internal Medicine. Ten percent of the total charts were randomly reviewed by two independent reviewers to check for inter-rater accuracy. Accuracy in the recording exceeded 90%. 2.4. Outcome measures Main outcome measures for the study were ROSC and survival to hospital discharge. 2.5. Statistical analysis Descriptive statistics were computed as means and standard deviations for continuous variables, and as frequencies for categorical variables. Comparisons of baseline characteristics between patients under full code and limited code were performed using t-tests and 2 -tests as appropriate. Logistic regression was used to assess the association of code status with subsequent survival and adjust for potential confounding factors. In addition to variables with known association with ROSC and survival, we adjusted for factors that differed between the two patients groups (p < 0.05). The former included intubation, defibrillation, medications, and procedures (central access attempted, pericardiocentesis, thoracotomy, or sternotomy). Results are reported as odds ratios with 95% confidence intervals. Exact statistical tests (exact likelihood ratio tests) and methods (exact logistic regression) were used where applicable. A p-value < 0.05 was considered statistically significant. Analyses were conducted with SAS Software, version 9.1 (SAS Institute Inc., Cary, NC, USA). 3. Results During the period under review, January 1999 to December 2003, 408 patient records were reviewed based on initial search criteria. Ninety-nine of these were excluded including 34 not meeting the definition of cardiac arrest, 22 with absent or grossly insufficient code documentation in the chart, 10 due to code occurring in the
K. Dosh et al. / Resuscitation 80 (2009) 985–989
987
Table 1 Primary hospital diagnoses for all patients suffering cardiac arrest.
Table 3 Characteristics of patients by code status.
Primary diagnosis
Patients, no. (%) (N = 309)
Characteristic
Limited code
Full code
p-value
Cardiovascular disease Cardiac surgery Vascular disease (peripheral vascular disease, aortic aneurysm, aortic dissection, aortic rupture) Congestive heart failure, myocardial infarction, unstable angina, coronary artery disease, valvular heart disease Arrhythmia (atrial fibrillation, ventricular tachycardia) Other (chest pain, cardiac tamponade, pericarditis)
141 (100) 25 (17.7) 7 (5.0)
Male gender (%)
35.3
49.7
0.32
Race (%) White Other Missing
82.4 5.9 11.8
63.7 21.2 15.1
0.19
Marital status (%) Married Other Missing
64.7 35.3 0
48.6 50.5 1.4
0.35
Functional status (%) Independent Dependent Missing
52.9 23.5 23.5
54.8 27.1 18.2
0.90
Age, mean (SD), years
75.2 (16.1)
67.8 (14.6)
0.08
BMI, mean (SD) (kg/m2 )
21.9 (6.6)
29.1 (10.2)
0.002
PAR, mean (SD)
9.1 (5.1)
5.6 (4.7)
0.01
APACHE II, mean (SD)
16.5 (6.7)
12.1 (6.0)
0.03
Witnessed code (%) Yes No Missing
82.4 11.8 5.9
74.3 18.2 7.5
0.77
Location (%) Unit Tele Other Missing
29.4 41.2 23.5 5.9
31.5 39.7 27.1 1.7
0.87
11.8 41.2 47.1
21.6 43.8 33.9 0.7
0.55
98 (69.5)
7 (5.0) 4 (2.8)
End-stage disease Metastatic cancer, hematologic malignancy Chronic obstructive pulmonary disease, acute respiratory failure Liver failure Renal failure
37 (100) 20 (54.1) 4 (10.8)
Infectious disease Abscess, cellulitis, endocarditis, pancreatitis, pneumonia Sepsis Other (fever of unknown origin)
65 (100) 34 (52.3)
Neurological disease Closed head injury Seizure Intracranial bleed Stroke, transient ischemic attack
16 (100) 4 (25.0) 3 (18.8) 5 (31.2) 4 (25.0)
Miscellaneous Gastrointestinal bleed Genitourinary surgery Orthopedic surgery Bowel obstruction General trauma Pulmonary embolism Other blood dyscrasia (anemia, thrombocytopenia, pancytopenia) Other
50 (100) 11 (22.0) 4 (8.0) 5 (10.0) 5 (10.0) 2 (4.0) 2 (4.0) 4 (8.0)
4 (10.8) 9 (24.3)
30 (46.2) 1 (1.5)
17 (34.0)
catheterization laboratory, operating room, or emergency department, 6 due to code occurring prior to hospital admission, 22 in which no chart could be found, and 5 as a result of family stopping the code early. This left 309 patients that met inclusion and exclusion criteria, of which 292 (94.5%) were full code and 17 (5.5%) were limited code. Primary hospital diagnoses for the entire group and limited code group are listed in Tables 1 and 2 respectively. Patient characteristics are compared between limited code and full code groups in Table 3. There were no significant differences between groups on age, race, gender, marital status and functional status. On average, full code patients had larger body size than limited code patients (mean BMI 29.1 kg/m2 vs. mean BMI 21.9 kg/m2 , p = 0.002). Compared to limited code patients, full code patients had significantly Table 2 Primary hospital diagnoses for all limited code patients. Primary diagnosis
Patients no. (%) (N = 17)
Cardiovascular disease: congestive heart failure, myocardial infarction End-stage disease Metastatic cancer Chronic obstructive pulmonary disease, acute respiratory failure Renal failure Infectious disease Pneumonia Sepsis Other (vertebral fracture)
7 (41) 4 (24) 1 1 2 5 (29) 2 3 1(6)
Initial rhythm (%) Pulseless, VT/VF Pulseless, non-VT/VF Other Missing Code points, mean (SD)
5.9 (6.2)
8.5 (5.9)
0.11
Code duration in minutes, mean (SD)
23.6 (22.8)
27.8 (49.8)
0.51
lower PAR score (5.6 vs. 9.1, p = 0.01) and lower APACHE II score (12.1 vs. 16.5, p = 0.03). Overall, 171 (58.6%) full code patients had ROSC compared to only 5 (29.4%) of the limited code patients (p = 0.023). For survival to discharge, 67 (23.0%) patients survived in the full code group versus only 1 (5.9%) patient in the limited code group. This difference was not statistically significant (p = 0.095). Significant correlates of ROSC were functional status (p = 0.012), initial rhythm (p = 0.005) and code points (p < 0.0001). Pulseless VT/VF and any other rhythm were each associated with higher odds of survival relative to other pulseless rhythms (non-VT/VF). The unadjusted association of code status with ROSC was obtained using exact logistic regression. The unadjusted estimated OR for survival (full code vs. limited code) was 3.38 (95% CI: 1.07–12.57, p = 0.035). Adjusted and unadjusted correlates of survival during the code are shown in Table 4. 4. Discussion Our study confirms the results of Dumot et al.1 and suggests the futility of limited codes. Of the 17 patients who opted for limited code, only 5 survived the code with ROSC, and of these, only one survived to discharge. Overall survival in full code patients in this study was comparable at 23% with other studies.1,2 This is the first study to compare survival in full code patients with limited code after controlling for clinical confounders. Comparison of clinical variables between the full and limited code patients revealed a higher PAR9 and APACHE II10 score and lower BMI for limited code
988
K. Dosh et al. / Resuscitation 80 (2009) 985–989
Table 4 Correlates of return of spontaneous circulation. Characteristic
Odds ratio (95% CI) [unadjusted]
Odds ratio (95% CI) [adjusted]
Code status Limited code Full code
1.00 [Reference] 3.38 (1.07, 12.57)
3.69 (1.13, 14.34)
Functional status Dependent Independent
1.00 [Reference] 1.95 (1.11, 3.45)
1.79 (1.03, 3.11)
Initial rhythm Pulseless, non-VT/VF Pulseless, VT/VF Other
1.00 [Reference] 2.23 (1.16, 4.36) 2.07 (1.19, 3.62)
2.11 (1.06, 4.32) 1.97 (1.09, 3.60)
Code points
0.86 (0.82, 0.90)a
0.84 (0.79, 0.89)
PAR
1.01 (0.97, 1.06)a
APACHE II
0.98 (0.95, 1.02)a
BMI
1.02 (0.99, 1.05)a
a
OR for 1 unit increase; PAR = prognosis-after-resuscitation, BMI = body mass index (kg/m2 ).
patients. The significance of this finding is uncertain but might suggest a generally more feeble condition of those in the limited code group. However, after adjusting for confounding variables, limited code status remained a predictor of poor outcomes. We found a significant disadvantage of limited code in ROSC and a similar trend in survival. Prior literature has demonstrated that explaining the probability of survival in CPR clearly affects patient preferences about code status.3,4 The significant differences in survival between limited and full code patients as determined by our study may impact patient preferences and would have a likely effect of patients opting for a full or no code rather than a limited code. The results of our study and previous literature should stimulate further discussion with the patient regarding their motivations and expectations of resuscitation in the event of in-hospital cardiac arrests. We expect this will lead to better decision-making. It is unclear whether the outcomes in limited code patients may have been affected by generally sub-standard treatment of their other medical conditions. Although literature on limited code patients is lacking, some correlates may be drawn from studies on patients selecting do-not-attempt-resuscitation (DNAR) status. Wenger et al.11 performed a retrospective review of 12,821 patients and compared in-hospital and 6-month mortality for those with and without DNAR orders. After adjusting for baseline demographic variables, co-morbid conditions, and hospital characteristics, patients with DNAR orders had an increased in-hospital (40% vs. 9%, p < 0.001) and 6-month (61% vs. 26%, p < 0.001) mortality compared to those without DNAR orders. More recently, Cohen et al. found that patients with DNAR orders are less likely to be admitted to an intensive care unit irrespective of baseline functional status and severity-of-illness.12 Ethical arguments support the notion that a DNAR order should not be interpreted as an order to withhold other standard therapies for a patient.13 But these results suggest that physicians and other healthcare providers may treat DNAR patients differently than full code patients. In this trial, we did not assess deviations from standard-of-care treatment of the patients’ admitting diagnoses. But it is possible that the limited code patients received less than standard care during their hospitalization prior to cardiac arrest. This effect might bias our results toward a worse prognosis for the limited code patients. Our analysis was structured to control for baseline and within-code variables that have been previously shown to predict outcomes in cardiac arrest.9,14–22 Although we believe this was
a reasonable method to control for confounding factors, this and other unrecognized bias may have had an effect on our outcomes. Moreover, application of this study’s results to patient care will require consideration of physicians’ personal preferences and opinions about CPR in general23 and limited codes in particular. Research has shown that physicians are often unaware of their patients’ resuscitation preferences24 and that advance directives may have limited to no effect on the documentation and implementation of DNAR requests.25 One prior study demonstrated that life-sustaining therapy in critically ill patients is often applied, withdrawn, or withheld in spite of patients’ or surrogates’ wishes.26 Such conflicts between physician decisions and patient preferences occur in spite of good intentions because physicians are often poor at predicting patients’ desires with regards to CPR27,28 and also because of differences between physician and public knowledge about life-sustaining measures.29 In this study, overall code intensity as measured by code points and code duration (Table 3) was found to be similar between limited code and full code patients. This would suggest that physicians put equal effort into running codes in each respective group. But physician notions about the futility of limited codes or their personal preferences might also have had an impact on outcomes that we were unable to measure. 4.1. Limitations Retrospective analysis may be felt to be a limitation by some, but a randomized controlled study of limited versus full codes could never be performed for obvious reasons. Furthermore a prospective study would require informed consent at time of admission and informed consent would likely introduce bias. Our goal was to determine patient preferences for code status when elicited by a clinician, not a researcher, and how those preferences affect survival. In fact, we believe that retrospective studies of existing data are better suited for this type of study to produce unbiased results. 5. Conclusions This study has proved that the ‘limited code’ status has a survival disadvantage to ‘full code’ status. We do respect patients’ preferences regarding their decisions about end-of-life directives but suspect that many of them choose limited code status without understanding the full implications of their decision. We hope the data presented in this paper will provide more evidence for physicians to counsel patients about this important issue. We did not measure physicians’ confidence in running limited resuscitation codes versus full codes, and believe that it may affect outcomes to a certain extent. Further research is warranted at multiple high volume centers for further investigation of this issue. Future studies are also needed to determine if patients who initially request a limited resuscitation change their directives after knowledge of outcome data. Conflict of interest None of the authors have any conflict of interest with regards to this manuscript. Acknowledgments We are very thankful for the help provided by Julie Bey, Suzanne Leialoha, and Stacy Near as well as Arman Raza, MD, Amir Azeem, MD, Kwsai Al-Rahhal, MD, and Dwarakraj Soundarraj, MD who assisted by collecting the data for this study. We also thank Lin Liu, PhD for her help with data analysis. All authors had full access to all data in the study and take responsibility for the integrity
K. Dosh et al. / Resuscitation 80 (2009) 985–989
of the data and the accuracy of the data analysis. This study was funded by an unrestricted educational grant from Wyeth Pharmaceuticals. Wyeth Pharmaceuticals had no involvement in the study design, data collection, analysis and interpretation of data, writing of the manuscript, or decision to submit the manuscript for publication. References 1. Dumot JA, Burval DJ, Sprung J, et al. Outcome of adult cardiopulmonary resuscitations at a tertiary referral center including results of “limited” resuscitations. Arch Intern Med 2001;161:1751–8. 2. Saklayen M, Liss H, Markert R. In-hospital cardiopulmonary resuscitation. Survival in 1 hospital and literature review. Medicine (Baltimore) 1995;74:163–75. 3. Murphy DJ, Burrows D, Santilli S, et al. The influence of the probability of survival on patients’ preferences regarding cardiopulmonary resuscitation. N Engl J Med 1994;330:545–9. 4. Schonwetter RS, Walker RM, Kramer DR, Robinson BE. Resuscitation decision making in the elderly: the value of outcome data. J Gen Intern Med 1993;8:295–300. 5. Fowler MD. Slow code, partial code, limited code. Heart Lung 1989;18:533–4. 6. Ross JW, Pugh D. Limited cardiopulmonary resuscitation: the ethics of partial codes. QRB Qual Rev Bull 1988;14:4–8. 7. Cummins RO, Chamberlain D, Hazinski MF, et al. Recommended guidelines for reviewing, reporting, and conducting research on in-hospital resuscitation: the in-hospital ‘Utstein style’. A statement for healthcare professionals from the American Heart Association, the European Resuscitation Council, the Heart and Stroke Foundation of Canada, the Australian Resuscitation Council, and the Resuscitation Councils of Southern Africa. Resuscitation 1997;34:151–83. 8. Vandenbroucke JP, von Elm E, Altman DG, et al. Strengthening the Reporting of Observational Studies in Epidemiology (STROBE): explanation and elaboration. PLoS Med 2007;4:e297. 9. O’Keeffe S, Ebell MH. Prediction of failure to survive following in-hospital cardiopulmonary resuscitation: comparison of two predictive instruments. Resuscitation 1994;28:21–5. 10. Knaus WA, Draper EA, Wagner DP, Zimmerman JE. APACHE II: a severity of disease classification system. Crit Care Med 1985;13:818–29. 11. Wenger NS, Pearson ML, Desmond KA, Brook RH, Kahn KL. Outcomes of patients with do-not-resuscitate orders. Toward an understanding of what donot-resuscitate orders mean and how they affect patients. Arch Intern Med 1995;155:2063–8. 12. Cohen RI, Lisker GN, Eichorn A, Multz AS, Silver A. The impact of do-notresuscitate order on triage decisions to a medical intensive care unit. J Crit Care 2009;24:311–5.
989
13. Martin RD, Cohen MA, Weiss Roberts L, Batista SM, Hicks D, Bourgeois J. DNR versus DNT: clinical implications of a conceptual ambiguity: a case analysis. Psychosomatics 2007;48:10–5. 14. Andreasson AC, Herlitz J, Bang A, et al. Characteristics and outcome among patients with a suspected in-hospital cardiac arrest. Resuscitation 1998;39:23–31. 15. Cohn EB, Lefevre F, Yarnold PR, Arron MJ, Martin GJ. Predicting survival from inhospital CPR: meta-analysis and validation of a prediction model. J Gen Intern Med 1993;8:347–53. 16. de Vos R, Koster RW, De Haan RJ, Oosting H, van der Wouw PA, LampeSchoenmaeckers AJ. In-hospital cardiopulmonary resuscitation: prearrest morbidity and outcome. Arch Intern Med 1999;159:845–50. 17. Ebell MH. Prearrest predictors of survival following in-hospital cardiopulmonary resuscitation: a meta-analysis. J Fam Pract May 1992;34:551–8. 18. Ebell MH. Artificial neural networks for predicting failure to survive following in-hospital cardiopulmonary resuscitation. J Fam Pract Mar 1993;36:297–303. 19. Ebell MH, Smith M, Kruse JA, Drader-Wilcox J, Novak J. Effect of race on survival following in-hospital cardiopulmonary resuscitation. J Fam Pract Jun 1995;40:571–7. 20. Ebell MH, Kruse JA, Smith M, Novak J, Drader-Wilcox J. Failure of three decision rules to predict the outcome of in-hospital cardiopulmonary resuscitation. Med Decis Making 1997;17:171–7. 21. George Jr AL, Folk IIIrd BP, Crecelius PL, Campbell WB. Pre-arrest morbidity and other correlates of survival after in-hospital cardiopulmonary arrest. Am J Med Jul 1989;87:28–34. 22. Marwick TH, Case CC, Siskind V, Woodhouse SP. Prediction of survival from resuscitation: a prognostic index derived from multivariate logistic model analysis. Resuscitation 1991;22:129–37. 23. Marik PE, Varon J, Lisbon A, Reich HS. Physicians’ own preferences to the limitation and withdrawal of life-sustaining therapy. Resuscitation 1999;42:197–201. 24. A controlled trial to improve care for seriously ill hospitalized patients. The study to understand prognoses and preferences for outcomes and risks of treatments (SUPPORT). The SUPPORT Principal Investigators. JAMA 1995;274:1591–8. 25. Teno J, Lynn J, Wenger N, et al. Advance directives for seriously ill hospitalized patients: effectiveness with the patient self-determination act and the SUPPORT intervention SUPPORT Investigators. Study to Understand Prognoses and Preferences for Outcomes and Risks of Treatment. J Am Geriatr Soc 1997;45:500–7. 26. Asch DA, Hansen-Flaschen J, Lanken PN. Decisions to limit or continue lifesustaining treatment by critical care physicians in the United States: conflicts between physicians’ practices and patients’ wishes. Am J Respir Crit Care Med 1995;151:288–92. 27. Uhlmann RF, Pearlman RA, Cain KC. Physicians’ and spouses’ predictions of elderly patients’ resuscitation preferences. J Gerontol 1988;43:M115–121. 28. Bedell SE, Delbanco TL. Choices about cardiopulmonary resuscitation in the hospital. When do physicians talk with patients? N Engl J Med 1984;310:1089–93. 29. Carmel S. Life-sustaining treatments: what doctors do, what they want for themselves and what elderly persons want. Soc Sci Med Nov 1999;49:1401–8.