- Meeting abstract
- Open Access
Outcome prediction in patients after out-of-hospital cardiac arrest cannot be improved by prolonging observation
- Published: 1 March 1997
Keywords
- Cardiac Arrest
- Frequent Criticism
- Outcome Group
- Outcome Predictor
- Neurologic Recovery
Neurologic prognosis after out-of-hospital cardiac arrest has important implications for delivery of intensive care, with the aim to assure the best opportunities for recovery without prolonging futile treatment. The prognostic value of several neurologic signs has already been assessed in previous studies [1,2]. A too short period of full active treatment is, however, frequently reported as a possible important limitation of many study designs.
We investigated in a retrospective study outcome predictors for post-anoxic coma; all patients had homogeneous and complete treatment for at least 7 days. Only patients aged 18-80 years, admitted with post-anoxic coma without concomitant neurologic disorders, and who survived at least 24 h were included. At a follow up examination 6 months or more after ICU admission, the 23 patients were assigned to three outcome groups; almost complete neurologic recovery (A); moderate-severe disability (B); dead without regaining consciousness-vegetative state (C). In this analysis the outcome groups A (six patients) and B (five patients) are considered together and compared to patients of group C (12 patients).
Routinely performed neurologic tests, known to be potential outcome predictors were collected at four different times: at 6 h after successful CPR (T0), at 24 h (T1), on the 3rd day (T2) and on the 7th day (T3). Sensitivity and specificity of each test in predicting a good outcome and total correct classifications (TCC = efficiency of prediction) are reported in the table.
Consistent with the results of previous investigations [1] our results show that: (i) it is mostly impossible to state a correct prognosis from a very early evaluation (T0); (ii) data reported at 72 h support the prognosis obtained at 24 h without adding any new information. The single best predictor of good outcome is motor response to pain, also GCS > 4 is a good predictor, but with lower specificity. Conversely the persistence of seizures is associated in all cases with poor prognosis.
Table
T0 | T1 | T2 | T3 | |
---|---|---|---|---|
GCS > 4 | ||||
TCC (%) | 69 | 78 | 78 | 82 |
Sensitivity (%) | 73 | 100 | 100 | 100 |
Specificity (%) | 67 | 58 | 58 | 64 |
Motor response to pain | ||||
TCC (%) | 60 | 87 | 78 | 77 |
Sensitivity (%) | 56 | 100 | 100 | 100 |
Specificity (%) | 64 | 75 | 58 | 55 |
Spontaneous motility | ||||
TCC (%) | 61 | 83 | 83 | 83 |
Sensitivity (%) | 45 | 73 | 91 | 100 |
Specificity (%) | 75 | 92 | 75 | 67 |
Seizures | ||||
TCC (%) | 59 | 83 | 74 | 83 |
Sensitivity (%) | 30 | 0 | 0 | 0 |
Specificity (%) | 50 | 67 | 50 | 67 |
Cranial nerve reflexes | ||||
TCC (%) | 75 | 62 | 61 | 57 |
Sensitivity (%) | 100 | 90 | 100 | 100 |
Specificity (%) | 44 | 30 | 25 | 10 |
Authors’ Affiliations
References
- Edgren E, et al: . Lancet. 1994, 343: 1055-1059. 10.1016/S0140-6736(94)90179-1.PubMedView ArticleGoogle Scholar
- Levy DE, et al: . JAMA. 1985, 253: 1420-1426. 10.1001/jama.253.10.1420.PubMedView ArticleGoogle Scholar