Quality of Survival After Cardiopulmonary Resuscitation

Background  Outcome of cardiopulmonary resuscitation (CPR) can be poor, in terms of life expectancy and quality of life.

Objectives  To determine the impact of patient characteristics before, during, and after CPR on these outcomes, and to compare results of the quality-of-life assessment with published studies.

Methods  In a cohort study, we assessed by formal instruments the quality of life, cognitive functioning, depression, and level of dependence of survivors after in-hospital CPR. Follow-up was at least 3 months after discharge from the hospital (tertiary care center).

Results  Of 827 resuscitated patients, 12% (n = 101) survived to follow-up. Of the survivors, 89% participated in the study. Most survivors were independent in daily life (75%), 17% were cognitively impaired, and 16% had depressive symptoms. Multivariate regression analysis showed that quality of life and cognitive function were determined by 2 factors known before CPR—the reason for admission and age. Factors during and after resuscitation, such as prolonged cardiac arrest and coma, did not significantly determine the quality of life or cognitive functioning of survivors. The quality of life of our CPR survivors was worse compared with a reference group of elderly individuals, but better than that of a reference group of patients with stroke. The quality of life did not importantly differ between the compared studies of CPR survivors.

Conclusions  Cardiopulmonary resuscitation is frequently unsuccessful, but if survival is achieved, a relatively good quality of life can be expected. Quality of life after CPR is mostly determined by factors known before CPR. These findings may be helpful in informing patients about the outcomes of CPR.

CARDIOPULMONARY resuscitation (CPR) can only be called successful if patients survive and their quality of life is acceptable. Neurologic impairments caused by the cardiac arrest can be a threat to this quality of life.^1-3

Quality of life comprises at least physical, psychological, and social functioning.⁴ All these dimensions have been described in relation to CPR separately,^2,3,5-9 but not in relation to each other. In view of the risks of resuscitation, it is important to get an overall insight into the quality of life after CPR and how this relates to the procedure.

The objective of this study was to determine the outcomes of CPR in terms of quality of life and cognitive functioning, and to relate important patient characteristics before, during, and after CPR to these outcomes. We aimed to widen this perspective by comparing the quality of life between different groups of CPR survivors and between different patient populations as known from the literature.

This study was carried out at the Academic Medical Center in Amsterdam, the Netherlands (tertiary care hospital, 1000 beds, 26,000 admissions per year). Cardiopulmonary resuscitation was defined as either the application of artificial ventilation and external chest compression or immediate defibrillation. Survivors were identified through the database of calls for the resuscitation team between May 1988 and August 1995, in which patient and procedure variables were prospectively entered. The team is routinely called on to all wards and the emergency department. The coronary care unit calls if the resuscitation exceeds a few defibrillator shocks.

We included all consecutive survivors (age ≥18 years) discharged from the hospital, including those with an out-of-hospital onset of cardiac arrest. Patients with a survival of at least 3 months were studied. The study was approved by the hospital medical ethics committee.

The quality of life was evaluated by physical, psychological, and social functioning. Cognitive functioning, depression, and dependence in daily life were assessed in depth. Questionnaires were selected by their psychometric properties and use in other CPR studies. The quality of life was measured with the 136-item self-reporting Sickness Impact Profile (SIP) with 12 subscales and 3 aggregated scores (physical, psychosocial, and overall health).¹⁰ The SIP scores range from 0 to 100; higher scores indicate a worse quality of life. Perceived quality of life was expressed by the patients on a visual analog scale ranging from 0 (worst) to 10 (best) as an answer to the question "How do you rate your quality of life over the past 14 days?" Cognitive functioning was assessed with the 30-item Mini-Mental State Examination (MMSE).¹¹ Patients were classified as cognitively impaired at a cutoff value of 23 or less.¹² Depression was measured with the 20-item self-reporting Center for Epidemiologic Studies Depression Scale. Patients had depressive symptoms at a cutoff value of 16 or more.¹³ Dependence in daily life was assessed with the 6-point observer-rated Rankin Scale.¹⁴ The categories are no symptoms (0); minor symptoms that do not interfere with lifestyle (1); symptoms that lead to some restriction in lifestyle but do not interfere with the patient's capacity to look after himself or herself (2); symptoms that prevent a totally independent existence (3); symptoms that clearly prevent independent existence although constant attention is not needed (4); and a totally dependent patient requiring constant attention night and day (5).

We compared the quality of life of our study group with that of similar populations of CPR survivors, but after out-of-hospital cardiac arrest and cardiac arrest at the intensive care unit.^9,10 A population of elderly people visiting the general practitioner and an age-corrected subsample of patients with stroke^15,16 served as reference groups to indicate where the quality-of-life outcomes CPR survivors can be positioned between 2 relatively extreme health states of patients who are about the same age.

The characteristics before, during, and after CPR were associated with mortality as well as with the quality of life after survival calculating a χ² for nominal variables and using the Student t test or analysis of variance for continuous variables. In case of skewed data or ordinal categories, we used the Mann-Whitney or Kruskall-Wallis test. A P≤.05 was considered statistically significant. The probability of survival was calculated and presented by a Kaplan-Meier survival curve¹⁷ where patients were censored at the time of the interview. All clinical characteristics before, during, and after CPR were entered in multivariate linear regression models (forward selection) to assess the independent effect on the quality-of-life dimensions. The strength of the relationship was expressed in partially explained variances (partial R²). The total explained variance was calculated by summarizing the partial R².¹⁸

The MMSE and Center for Epidemiologic Studies Depression Scale scores were considered dichotomous according to their cutoff value. All characteristics before, during, and after CPR were entered in logistic regression analysis models (forward selection) to identify their independent effect on cognitive impairment and depression. The effect sizes of significant characteristics were expressed as odds ratios (95% confidence intervals).¹⁹

To compare the quality of life of the study group with a reference group, SIP scores were converted to mean standard scores ([mean scores of the study group − mean scores of the reference group]/SD of the reference group).²⁰ According to this method, the scores of the reference group were set at 0. Standard scores indicate the difference in SDs between the study and reference groups. Given the similarity with the effect size calculations, a mean standard score of 0.20 can be taken to indicate a small deviation from the reference data; 0.50, a moderate deviation; and 0.80, a substantial deviation.

During the study period 827 patients underwent CPR, 385 patients initially survived, 162 (20%) survived to discharge, of whom 5 were in a vegetative state. Of the 51 patients who died after discharge, 29 died in the first 6 months (Figure 1). Of 5 patients who were in a vegetative state at discharge, 4 died within 3 months. Ten patients were lost to follow-up. Death after discharge was significantly associated with age 70 years or older (P =.05) and dependency in daily life before CPR (P =.05).

Of 101 survivors, 90 (89%) participated in the interview: 50 men and 40 women with a mean (SD) age of 64 (13) years. The median time from CPR to interview was 15 months (range, 3-72 months). Survivors who refused to participate stated that they were in good health (n = 4) or bad health (n = 4), or would not answer any questions on their current health (n = 3). Of the participating survivors, 41 (46%) had ventricular fibrillation as initial rhythm and 84 (93%) had a circulatory arrest of cardiac origin and 6 (7%) of respiratory origin. Of 84 patients with a circulatory arrest of cardiac origin, 30 (36%) experienced an arrest because of a myocardial infarction and 32 (38%) because of a primary rhythm disturbance. Other causes of arrest were a variety of cardiac conditions, such as disturbances of the myocardial conduction system and acute heart failure.

All 90 patients completed the SIP, 4 of whom needed the assistance of a relative. The mean (SD) dimension score for the subscale physical functioning was 13.3 (15.7); for the subscale psychosocial functioning, 10.5 (11.5); and for the total SIP, 12.8 (12). All defined patient characteristics as known before CPR were significantly associated with physical functioning (Table 1). Physical functioning tended to be more impaired in case of in-hospital resuscitation, prolonged coma, and an interview 6 months or less after resuscitation. Psychosocial functioning was significantly more impaired with a noncardiac reason for admission and tended to be associated with dependence in daily life before CPR, but this was not statistically significant. The overall quality of life (total SIP) was significantly more impaired in the case of patients aged 70 years or older and those with a noncardiac reason for admission. Prolonged coma also tended to result in a more impaired overall quality of life. The mean (SD) score for perceived quality of life as rated by the patients was 7 (2) on a scale of 0 to 10.

The MMSE was completed by 88 patients, with a mean (SD) score of 27 (4.3); 2 refused to complete the MMSE. Fifteen patients (17%) were classified as cognitively impaired at the cutoff point of 23 or more. Cognitive dysfunctioning was significantly associated with age 70 years or older and with a noncardiac reason for admission. Duration of CPR and coma were not significantly associated with MMSE scores.

The Center for Epidemiologic Studies Depression Scale was completed by 86 patients. Four patients were not capable of completing the Center for Epidemiologic Studies Depression Scale. The mean (SD) score was 8 (8.1). At the cutoff point of 16 or more, 14 patients (16%) were classified as having depressive symptoms. No specific factor was significantly associated with depression.

The Rankin Scale was scored for all patients. Daily life was little or not limited in 66 patients (74%) (Rankin Scale score ≤2). Twelve (13%) had restrictions in daily life (Rankin Scale score, 3), but were able to look after themselves. Twelve patients (13%) were partially or totally dependent (Rankin Scale score, 4). Patients aged 70 years or older and with a noncardiac admission (P =.05) were significantly associated with dependence in daily life after CPR.

On entering all clinical characteristics in linear regression models, mainly the characteristics before CPR could explain the level of quality of life after CPR (Table 2). However, the variances explained were relatively low. An impaired physical functioning and dependence in daily life could be partially explained by multiple factors (noncardiac reason for admission, age, and duration of coma). An impaired psychosocial and total functioning could be explained by a single factor (noncardiac reason for admission). By logistic regression analysis, a noncardiac reason for admission was identified as the only independent risk factor for cognitive impairments (odds ratio, 10; 95% confidence intervals, 3.36). Multivariate analysis was not performed on perceived quality of life and depressive symptoms because no significant associated characteristic could be identified at the univariate level.

The comparison of the profile of our SIP scores with that of (1) survivors after out-of-hospital cardiac arrest¹⁰ and (2) survivors after cardiac arrest in the intensive care⁹ is shown in Figure 2. The quality of life of our study group did not importantly differ from the reference groups of survivors of out-of-hospital CPR and CPR while in the intensive care unit.

The comparison of the profile of SIP scores of our study group with that of (1) a population of 132 Dutch elderly people (age range, 61-75 years) visiting the general practitioner¹⁵ and (2) an age-corrected subsample of 441 Dutch patients with stroke (mean age, 64 years)¹⁶ is shown in Figure 3. The quality of life of CPR survivors was substantially worse (higher scores) compared with the elderly individuals, except for mobility, social interaction, alertness behavior, and communication, but better (lower scores) than that of patients with stroke, particularly in body care and movement, communication, and eating.

A commonly expressed fear is that CPR results in a poor quality of life, particularly the worst imaginable, a vegetative state. The results of our study indicate that quality of life after CPR in general is satisfactory. Of the survivors, 75% were independent in daily life and survivors themselves rated their quality of life at 7 on a scale of 10. The feared vegetative state occurred only in 3% of our patients and was followed by death within several months.

Comparing our quality-of-life outcomes with other studies is difficult because of methodological differences. Most studies involve a small number of patients,^6,20-28 with a few exceptions,^29-31 and outcomes are mainly assessed by clinical judgment^{21-23,26,27,30,31} and rarely by quantitative measurements.^9,10 When no important methodological differences exist,^9,10 our data suggest that across different studies, survivors after CPR have about the same quality of life, regardless of the conditions under which CPR took place. When the quality of life of our survivors of CPR is compared with a group of elderly patients who were studied with the same quantitative instruments, significant impairments in CPR survivors can be detected by all subscales of the SIP, but these impairments are of much smaller magnitude than the impairments of a group of stroke survivors. How much of the impairments after CPR can be attributed to the CPR itself and/or the underlying disease cannot be determined by our study. However, of all of the determinants of the quality of life studied, the admission diagnosis in the hospital was the most important factor that could explain the differences in the quality of life. This indicates that differences in disease leading to CPR, rather than differences in CPR and recovery itself, contribute to the quality of life after CPR. It is also important to notice that a prolonged duration of CPR did not have a negative influence on any of the dimensions of the quality of life after CPR. We expect that more information on the condition of the patient before CPR will lead to a more complete explanation of their quality of life after CPR.

The results of this study could have been biased by selection through mortality or patient participation. If only the most healthy patients survive or participate, it may be that the quality of life is overestimated and potential determinants of poor quality of life remain undetected. However, we have no clear evidence that participation biased our results—the few patients who refused to participate in our study were not characterized unanimously by a poor quality of life.

Patients who were interviewed within 6 months after discharge performed worse in physical and total functioning as measured with the SIP. This suggests a difference in physical functioning over time. It may imply that recovery may take more than 6 months, but also that the outcome quality of life after CPR depends on the timing of assessment. When investigating changes in the quality of life over time, a longitudinal study design is required, with repeated measures and control over mortality as a potential confounding factor.

We conclude that CPR is frequently unsuccessful, but if survival is achieved and the patient leaves the hospital, a fair to good quality of life can be expected. When deciding on a do-not-attempt resuscitation order, we recommend that one not only focus on a small probability of survival,³² but also consider the high probability of a good quality of life for the survivors.

Accepted for publication May 26, 1998.

This study was supported by grant 93.170 from the Netherlands Heart Foundation, The Hague.

Reprints: R. de Vos, RN, Academic Medical Center, University of Amsterdam, Division Operation Center H1-212, Meibergdreef 9, PO Box 22700, 1100 DE Amsterdam, the Netherlands (e-mail: r.vos@amc.uva.nl).