 |
|
Derivation of a Clinical Decision Rule for the Discontinuation of In-Hospital Cardiac Arrest Resuscitations
Carl van Walraven, MD, FRCPC, MSc;
Alan J. Forster, MD, FRCPC;
Ian G. Stiell, MD, MSc, FRCPC
Arch Intern Med. 1999;159:129-134.
ABSTRACT
| |
Background Most patients undergoing in-hospital cardiac resuscitation will not survive to hospital discharge.
Objective To derive a decision rule permitting the discontinuation of futile resuscitation attempts by identifying patients with no chance of surviving to hospital discharge.
Patients and Methods Patient, arrest, and outcome data for 1077 adult patients undergoing in-hospital cardiac resuscitation was retrieved from 2 randomized clinical trials involving 5 teaching hospitals at 2 university centers. Recursive partitioning was used to identify a decision rule using variables significantly associated with death in hospital.
Results One hundred three patients (9.6%) survived to hospital discharge. Death in hospital was significantly more likely if patients were older than 75 years (P<.001), the arrest was unwitnessed (P = .003), the resuscitation lasted longer than 10 minutes (P<.001), and the initial cardiac rhythm was not ventricular tachycardia or fibrillation (P<.001). All patients died if there was no pulse 10 minutes after the start of cardiopulmonary resuscitation, the initial cardiac rhythm was not ventricular tachycardia or fibrillation, and the arrest was not witnessed. As a resuscitation rule, these parameters identified all patients who survived to hospital discharge (sensitivity, 100%; 95% confidence interval, 97.1%-100%). Resuscitation could have been discontinued for 119 (12.1%) of 974 patients who did not survive, thereby avoiding 47 days of postresuscitative care.
Conclusions A practical and highly sensitive decision rule has been derived that identifies patients with no chance of surviving in-hospital cardiac arrest. Prospective validation of the rule is necessary before it can be used clinically.
INTRODUCTION
CARDIOPULMONARY resuscitation (CPR) and advanced cardiac life support are used to resuscitate patients suffering cardiac arrest. With the exception of airway management and cardiac defibrillation, the effectiveness of interventions provided during resuscitation is uncertain.1 Physicians and other health care personnel often expend prolonged amounts of time and effort in attempting to resuscitate patients. Resuscitated patients may be subjected to prolonged, intensive, and invasive therapy. Throughout this time, the patient's family and members of the health care team often deal with difficult emotions and decisions.
Several clinical factors have been associated with poor outcome following in-hospital cardiac arrest. Death has been shown to be independently associated with several prearrest clinical factors including hypotension,2 renal failure,3-5 pneumonia,2 poor functional status,2, 6 and metastatic cancer.2, 5 Several factors identifiable during the arrest independently predict mortality including prolonged duration of resuscitation2, 6-7 and initial cardiac rhythms other than ventricular tachycardia or fibrillation.6, 8-9 Postarrest factors such as high Acute Physiology and Chronic Health Evaluation II or low Glasgow Coma Scale scores are associated with death after initially successful resuscitation.10 However, most of these studies were conducted within a single center, contained small numbers of arrests, or used data from unblinded and retrospective medical record review. Therefore, it is not surprising that some of the studies have conflicting results.
Several studies8, 10-12 have combined various prearrest, arrest, and postarrest factors into clinical decision rules for predicting survival following cardiac arrest. However, because the probability of survival from in-hospital cardiac arrest in most studies is less than 20%, a decision rule will not change management unless it can confidently predict the likelihood of survival to be nil. None of the published rules8, 10-12 have this level of accuracy. The rules can be impractical for common use because they require large amounts of clinical data,10-11 complex mathematical calculations,9 or involve clinical information not available to the resuscitation team.
In this study, we used prospectively collected data for patients entered into 1 of 2 randomized controlled trials to derive a clinical decision rule for in-hospital cardiac arrest. Specifically, we wanted to determine if readily available patient and arrest characteristics could be combined during the resuscitation to identify which patients had no chance of being discharged from hospital. This rule would therefore allow the treating team to discontinue futile resuscitation efforts.
PATIENTS AND METHODS
Patients were eligible for this study if they suffered in-hospital cardiac arrest and were enrolled in 1 of 2 randomized controlled trials. These negative trials determined that neither high-dose epinephrine13 nor active compression-decompression cardiopulmonary resuscitation (CPR)14 increased resuscitation from cardiac arrest. To be included in these studies, patients had to undergo CPR in the hospital. Consequently, patients whose resuscitation required defibrillation only were not included. Patients resuscitated in the emergency department had a spontaneous pulse when presenting to the department and subsequently underwent an arrest while in the department. Patients were excluded if they were younger than 16 years; had a terminal illness; had been without CPR for more than 15 minutes after their collapse; had acute trauma or exsanguination; were in the operating, recovery, or delivery rooms at the time of the arrest; had a recent sternotomy; or were judged (as determined by a blinded chart review) to have received inappropriate CPR (eg, a respiratory arrest with detectable pulse). These studies took place at 5 university-associated, tertiary care teaching hospitals in Ottawa and London, Ontario, between 1989 and 1995.
During the resuscitations, nurses, respiratory therapists, orderlies, and physicians administered CPR. These hospital staff members were fully trained in standard CPR and required regular testing. Advanced cardiac life support was directed by staff physicians or senior medical residents using standard protocols published by the American Heart Association.15 During the epinephrine trial,13 patients were randomized to receive either high (7-mg) or standard (1-mg) dosages of epinephrine. During the active compression-decompression–CPR trial,14 patients were randomized to receive either standard CPR or a device for active chest compression-decompression. Otherwise, resuscitation efforts were identical in the 2 studies. Neither study showed a significant difference between treatment and control groups.
Following each cardiac arrest, the resuscitation team completed data collection forms. Physicians documented the suspected cause of the arrest as well as the cardiac rhythm noted at the start of advanced cardiac life support. Resuscitation nurses recorded all other factors. The arrest was considered "unwitnessed" if the patient lost spontaneous circulation in the absence of another person. The duration of the arrest was defined as the time from the start of CPR to either return of spontaneous circulation or cessation of resuscitative efforts. A study nurse abstracted current and past medical conditions of each patient from the medical record. All resuscitated patients were followed up until death or discharge from hospital. Prior to discharge, the neurologic status of all patients were evaluated with the Modified Mini-Mental State Examination16 and a 5-point scale of cerebral performance.17
Data from both studies were combined and all clinical variables were assessed for association with discharge from hospital using the  2 test (for nominal data) or the unpaired separate-variance Student t test (for continuous data). Age and resuscitation duration was dichotomized at clinically reasonable cutoff points and analyzed as nominal variables. Variables that were significantly associated with survival-to-discharge (P<.05, 2-tailed) were analyzed by a 2 recursive-partitioning technique using a P value threshold of .05 for each split.18-19 Our goal was to generate a decision rule that would accurately identify patients with any chance of being discharged from hospital. For the decision rule, we chose the statistical model that offered a sensitivity of 1.0 (ie, no patients discharged from the hospital were predicted by the decision rule to die in hospital). Since several models meeting this standard were derived, that with the highest possible specificity was chosen as the final decision rule.
The performance of the derived decision rule for identifying patients with no chance of discharge from hospital was assessed by calculating sensitivity, specificity, and negative and positive predictive values with 95% confidence intervals (CIs).20 A receiver operating characteristics curve was not calculated since the prediction rule is binary. To determine what effect the decision rule could have if used clinically, the outcome of patients identified by the decision rule as having no chance of discharge was determined. All analyses were performed using SPSS for Windows, Version 7.0 (SPSS Inc, Chicago, Ill) and KnowledgeSEEKER for Windows 3.1 (Angoss Software, Toronto, Ontario).
RESULTS
During the clinical trials, 1472 patients were treated for in-hospital cardiac arrest. However, 376 of them were excluded from analysis because of exclusion criteria for the randomized trials. Nineteen (1.3%) were excluded because of incomplete data.
For derivation of the prediction rule, 1077 patients were eligible (Table 1). Patients were more commonly men and had a mean age of 68 years, with 35% being older than 75 years. Cardiovascular disease was more common than respiratory disease as a current diagnosis, a pattern also seen for chronic diagnoses. Most arrests were witnessed and occurred on the hospital ward. For witnessed cases in which the time of collapse could be determined, delay to CPR and advanced cardiac life support was short. In 31.4% of arrests, the initial cardiac rhythm was either ventricular tachycardia or fibrillation, with the remainder of the patients having asystole or pulseless electrical activity. Most arrests had a cardiac cause; 9.6% of the patients survived to discharge. Survivors had good cognitive function with a median Modified Mini-Mental State Examination score only slightly lower than normal. Patients from the 2 trials were similar for most characteristics.
|
|
|
|
Table 1. Description of Patient Characteristics and Resuscitation Outcome Following Cardiac Arrest*
|
|
|
The association of patient and resuscitation variables with survival-to-discharge is shown in Table 2. Although mean age was not associated with survival, patients older than 75 years were significantly less likely to survive. Neither sex nor any of the current or past diagnoses were associated with survival. Cardiac arrests of survivors were more likely to be witnessed, to have either ventricular tachycardia or fibrillation as the initially recorded rhythm, and were shorter in duration. Groups did not differ with respect to treatment delays, or location and cause of arrest.
|
|
|
|
Table 2. Univariate Association of Patient and Resuscitation Characteristics With Survival*
|
|
|
Using the variables significantly associated with survival-to-discharge, recursive partitioning identified 3 resuscitation factors that predicted no chance of survival (Figure 1). All patients died if there was no pulse 10 minutes after the start of CPR, the initial cardiac rhythm was not ventricular tachycardia or fibrillation, and the arrest was not witnessed. These factors were combined into the resuscitation rule, whereby continued resuscitation is unnecessary if all the following are true: (1) no pulse 10 minutes after the start of CPR; (2) initial cardiac rhythm is not ventricular tachycardia or fibrillation; and (3) the arrest was not witnessed. In our study population, this decision rule predicted 119 patients (11.0%) to die in hospital and correctly identified all patients who were discharged from hospital (sensitivity, 100%; 95% CI, 97.1%-100%) (Table 3). The low specificity (12.2%; 95% CI, 10.3%-14.4%) and positive predictive value (10.8%; 95% CI, 8.9%-12.8%) are the trade-offs for the high sensitivity.
|
|
|
|
Decision rule for identifying patients with no chance of being discharged from hospital. DC indicates discharged from hospital; CPR, cardiopulmonary resuscitation; VT, ventricular tachycardia; and VF, ventricular fibrillation. Decision nodes are represented by circles. Each node contains the number of patients to which it applies.
|
|
|
|
|
|
|
Table 3. Classification Performance of Decision Rule for Identifying Patients With No Chance of Discharge From Hospital*
|
|
|
Of the 119 patients identified by the decision rule to die in hospital, 99 (83.1%) could not be resuscitated with resuscitations lasting a mean of 25.6 minutes (range, 10-60 minutes). The remaining 20 patients survived a total of 1147 hours (mean, 57.4 hours; range, 1-240 hours), or 47.9 days in the intensive care unit. Details of the 7 patients identified by the rule to not survive who lived longer than 24 hours following the arrest are summarized in Table 4. All but 2 patients had multisystem disease prior to the arrest. Three of the patients (patients 3, 4, and 6) experienced new problems in hospital prior to the arrest and all except patient 4 experienced an arrest within the first week of hospitalization. Following the arrest, patients 1 through 4 had anoxic encephalopathy severe enough to preclude meaningful interaction with their environment; discussion with families resulted in life support being withdrawn. The remaining patients were able to participate in decisions regarding their treatment and all decided to not undergo further resuscitation in the event of cardiac arrest. Each patient died within 24 hours of extubation and only 1 (patient 6) was transferred out of the intensive care unit prior to death.
|
|
|
|
Table 4. Outcomes of Patients Identified by Decision Rule of Having No Chance of Discharge From Hospital Who Survived Longer Than 24 Hours Following the Arrest*
|
|
|
COMMENT
To our knowledge, this study represents the largest collection of prospectively gathered data for in-hospital cardiac arrests. With commonly available factors, our resuscitation rule identified all patients having any chance of being discharged from hospital following resuscitation. For patients identified by the rule as having no chance of survival, resuscitative efforts were often prolonged. At least 16.8% of patients predicted by the rule to die were treated in hospital for often extensive amounts of time following the arrest. If this decision rule is validated with prospectively collected data, it should be easy to use and avoid futile resuscitative efforts.
Decision rules can improve clinical decision making,21 decrease unnecessary investigations,22 and decrease the cost of health care delivery.23 This study meets most methodological standards for the development of clinical decision rules.24-25 The outcome predicted by the rule was clearly defined and clearly important. Since many characteristics of both the patients and study sites were described, the generalizability of the rule can be determined. The mathematical techniques used for the derivation of the rule were delineated and valid. Estimates for ramifications of the rule's application were given. The rule is clinically sensible since other studies have associated death after resuscitation with each variable in the rule including prolonged resuscitative efforts,2, 6-7,26 initial rhythms other than ventricular tachycardia or fibrillation,6-9,27-28 and unwitnessed arrest.7 Finally, the rule is easy to use, and suggests a course of action.
This study has several limitations. First, the rule has not been prospectively validated. The performance characteristics of the rule could change when applied to a different group of patients. Second, we did not determine the reliability of the variables included in the rule. Most likely, interobserver agreement for whether the arrest was witnessed is high. However, agreement may be lower for classification of the initial cardiac rhythm. Unreliable variables will decrease a rule's reproducibility. Prior to prospective validation, definitions for these and other parameters need to be developed and their interobserver agreement needs to be determined. Third, current and past diagnoses were grouped into relatively large categories (eg, respiratory disease), therefore making it impossible to determine the effect that particular diagnoses (eg, pneumonia) had on resuscitation outcome.
Fourth, the rule included factors concerning patient status at the start of the arrest and used them to predict prognosis 10 minutes later if resuscitation was unsuccessful. Therefore, the rule does not consider any factors arising within the first 10 minutes of the arrest. During this time, prognostically important cardiac rhythms and responses to therapy could occur. Also, the rule does not include recently described prognostic factors such as end-tidal carbon dioxide levels. Several studies29-30 have suggested this to be a powerful predictor of survival from cardiac arrest. Future studies need to determine if these and other data can be added to the present prediction rule to improve its performance.
Finally, it is uncertain how effective this rule could be, even if validated. Situations could arise where a physician decides that the rule should not apply to a particular patient because of what are considered to be special, individual circumstances. This is valid since clinical judgment must always be used when applying any clinical decision rule. Further research is needed to determine how frequently this would occur with a particular rule. Also, the rule identified only 12% of all people dying after in-hospital cardiac arrest. Most of these patients were not successfully resuscitated. Using the decision rule to predict patient death minutes prior to it actually happening is not terribly useful. However, we believe the rule would help physicians with limited experience in cardiac resuscitation decide when continued resuscitation is futile.
Despite these uncertainties, we believe that this decision rule, if validated, has the potential to be helpful. It will decrease the amount of time and effort that staff expend on futile cardiac resuscitations. While the rule will not protect families from difficult emotions associated with a loved one's sudden death, stressful days of watching and waiting could be avoided. Finally, the costs saved by using the rule to avoid prolonged stays in the intensive care unit could be used to improve other areas of the health care system. However, these potential benefits can only be realized if the rule is validated in a prospective, multicenter study.
In summary, we have derived a simple and accurate decision rule that identifies all patients with a chance of surviving in-hospital cardiac arrest. The rule is explicit, easy to apply, uses objective data that are readily available at all cardiac arrests, and could significantly decrease the amount of futile postresuscitation care. If validated, the rule could be combined with astute clinical judgment to more effectively use cardiac resuscitation.
AUTHOR INFORMATION
Accepted for publication May 28, 1998.
Dr van Walraven was an R. Samuel McLaughlin Foundation research fellow at the Institute for Clinical Evaluative Sciences, Toronto, Ontario, when this study was completed. Dr Stiell is a scientist of the Medical Research Council of Canada.
Corresponding author: Carl van Walraven, MD, FRCPC, MSc, F-6, Ottawa Civic Hospital, 1053 Carling Ave, Ottawa, Ontario, Canada K1Y 4E9.
From the Department of Medicine (Drs van Walraven and Forster) and the Division of Emergency Medicine (Dr Stiell), University of Ottawa, Ottawa, Ontario.
REFERENCES
| |
1. Niemann JT. Cardiopulmonary resuscitation. N Engl J Med. 1992;327:1075-1080.
ISI
| PUBMED
2. Bedell SE, Delbanco TL, Cook EF, Epstein FH. Survival after cardiopulmonary resuscitation in the hospital. N Engl J Med. 1983;309:569-576.
ABSTRACT
3. Nowak RM, Martin GB, Carden DL, Tomlanovich MC. Selective venous hypercarbia during human CPR: implications regarding blood flow. Ann Emerg Med. 1987;16:527-530.
FULL TEXT
| PUBMED
4. Beuret P, Feihl F, Vogt P, Perret A, Romand JA, Perret C. Cardiac arrest: prognostic factors and outcome at one year. Resuscitation. 1993;25:171-179.
FULL TEXT
|
ISI
| PUBMED
5. Ebell MH, Preston PS. The effect of the APACHE II score and selected clinical variables on survival following cardiopulmonary resuscitation. Fam Med. 1993;25:191-196.
PUBMED
6. Ballew KA, Philbrick JT, Caven DE, Schorling JB. Predictors of survival following in-hospital cardiopulmonary resuscitation: a moving target. Arch Intern Med. 1994;154:2426-2432.
FREE FULL TEXT
7. Bialecki L, Woodward RS. Predicting death after CPR: experience at a nonteaching community hospital with a full-time critical care staff. Chest. 1995;108:1009-1017.
FREE FULL TEXT
8. Marwick TH, Case CC, Siskind V, Woodhouse SP. Prediction of survival from resuscitation: a prognostic index derived from multivariate logistic model analysis. Resuscitation. 1991;22:129-137.
FULL TEXT
|
ISI
| PUBMED
9. Woodhouse SP, Case CC, Siskind V, Eller H. Prediction of hospital discharge in immediate survivors of ventricular fibrillation or asystole. Resuscitation. 1992;23:77-82.
FULL TEXT
|
ISI
| PUBMED
10. Niskanen M, Kari A, Nikki P, et al. Acute Physiology and Chronic Health Evaluation (APACHE Il) and Glasgow Coma scores as predictors of outcome from intensive care after cardiac arrest. Crit Care Med. 1991;19:1465-1473.
PUBMED
11. George AL Jr, Folk 3d BP, Crecelius PL, Campbell WB. Pre-arrest morbidity and other correlates of survival after in-hospital cardiopulmonary arrest. Am J Med. 1989;87:28-34.
ISI
| PUBMED
12. Burns R, Graney MJ, Nichols LO. Prediction of in-hospital cardiopulmonary arrest outcome. Arch Intern Med. 1989;149:1318-1321.
FREE FULL TEXT
13. Stiell IG, Hebert PC, Weitzman BN, et al. High-dose epinephrine in adult cardiac arrest. N Engl J Med. 1992;327:1045-1050.
ABSTRACT
14. Stiell IG, Hebert PC, Wells GA, et al. The Ontario trial of active compression-decompression cardiopulmonary resuscitation for in-hospital and prehospital cardiac arrest. JAMA. 1996;275:1417-1423.
FREE FULL TEXT
15. Guidelines for cardiopulmonary resuscitation and emergency cardiac care. Emergency Cardiac Care Committee and Subcommittees, American Heart Association, III: adult advanced cardiac life support. JAMA. 1992;268:2199-2241.
FREE FULL TEXT
16. Teng EL, Chui HC. The Modified Mini-Mental State (3MS) Examination. J Clin Psychiatry. 1987;48:314-318.
ISI
| PUBMED
17. A randomized clinical trial of calcium entry blocker administration to comatose survivors of cardiac arrest: design, methods, and patient characteristics: The Brain Resuscitation Clinical Trial II Study Group. Control Clin Trials. 1991;12:525-545.
FULL TEXT
|
ISI
| PUBMED
18. Vass GV. An exploratory technique for investigating large quantities of categorical data. Appl Stat. 1975;24:178-189.
19. Stiell IG, Greenberg GH, McKnight RD, Nair RC, McDowell I, Worthington JR. A study to develop clinical decision rules for the use of radiography in acute ankle injuries. Ann Emerg Med. 1992;21:384-390.
FULL TEXT
|
ISI
| PUBMED
20. Epi Info[computer program]. Version 6. Atlanta, Ga: Centers for Disease Control and Prevention; 1994.
21. Stiell IG, Greenberg GH, Wells GA, et al. Derivation of a decision rule for the use of radiography in acute knee injuries. Ann Emerg Med. 1995;26:405-413.
FULL TEXT
|
ISI
| PUBMED
22. Stiell IG, McKnight RD, Greenberg GH, et al. Implementation of the Ottawa ankle rules. JAMA. 1994;271:827-832.
FREE FULL TEXT
23. Stiell IG, Wells GA, Hoag RH, et al. Implementation of the Ottawa Knee Rule for the use of radiography in acute knee injuries. JAMA. 1997;278:2075-2079.
FREE FULL TEXT
24. Wasson JH, Sox HC, Neff RK, Goldman L. Clinical prediction rules: applications and methodological standards. N Engl J Med. 1985;313:793-799.
ABSTRACT
25. Laupacis A, Sekar N, Stiell IG. Clinical prediction rules: a review and suggested modifications of methodological standards. JAMA. 1997;277:488-494.
FREE FULL TEXT
26. Schultz SC, Cullinane DC, Pasquale MD, Magnant C, Evans SR. Predicting in-hospital mortality during cardiopulmonary resuscitation. Resuscitation. 1996;33:13-17.
FULL TEXT
|
ISI
| PUBMED
27. Rozenbaum EA, Shenkman L. Predicting outcome of in-hospital cardiopulmonary resuscitation. Crit Care Med. 1988;16:583-586.
ISI
| PUBMED
28. Skovron ML, Goldberg E, Suljaga-Petchel K. Factors predicting survival for six months after cardiopulmonary resuscitation: multivariate analysis of a prospective study. Mt Sinai J Med. 1985;52:271-275.
PUBMED
29. Callaham M, Barton C. Prediction of outcome of cardiopulmonary resuscitation from end-tidal carbon dioxide concentration. Crit Care Med. 1990;18:358-362.
ISI
| PUBMED
30. Cantineau JP, Lambert Y, Merckx P, et al. End-tidal carbon dioxide during cardiopulmonary resuscitation in humans presenting mostly with asystole: a predictor of outcome. Crit Care Med. 1996;24:791-796.
FULL TEXT
|
ISI
| PUBMED
 CiteULike  Connotea  Del.icio.us  Digg  Reddit  Technorati  Twitter
What's this?
RELATED ARTICLE
Do We Need a Clinical Decision Rule for the Discontinuation of Cardiac Arrest Resuscitations?
Arthur B. Sanders
Arch Intern Med. 1999;159(2):119-121.
EXTRACT
| FULL TEXT
THIS ARTICLE HAS BEEN CITED BY OTHER ARTICLES
|
Development and Application of Clinical Prediction Rules to Improve Decision Making in Physical Therapist Practice
Childs and Cleland
ptjournal 2006;86:122-131.
FULL TEXT
Cardiopulmonary Resuscitation in Critically Ill Neurologic-Neurosurgical Patients
Rabinstein et al.
Mayo Clin Proc. 2004;79:1391-1395.
ABSTRACT
CPR for Patients Labeled DNR: The Role of the Limited Aggressive Therapy Order
Choudhry et al.
ANN INTERN MED 2003;138:65-68.
ABSTRACT
| FULL TEXT
Cardiac Resuscitation
Eisenberg and Mengert
NEJM 2001;344:1304-1313.
FULL TEXT
Validation of a Clinical Decision Aid to Discontinue In-Hospital Cardiac Arrest Resuscitations
van Walraven et al.
JAMA 2001;285:1602-1606.
ABSTRACT
| FULL TEXT
Apolipoprotein E Polymorphism : Survival and Neurological Outcome After Cardiopulmonary Resuscitation
Schiefermeier et al.
Stroke 2000;31:2068-2073.
ABSTRACT
| FULL TEXT
Decision Rule for Discontinuing In-Hospital Cardiac Arrest Resuscitation
JWatch Emergency Med. 1999;1999:15-15.
FULL TEXT
Do We Need a Clinical Decision Rule for the Discontinuation of Cardiac Arrest Resuscitations?
Sanders
Arch Intern Med 1999;159:119-121.
FULL TEXT
|
