Prognosis Fibrillation

of Late Versus Early Ventricular in Acute Myocardial Infarction

Gunnar V.H. Jensen, MD, Christian Torp-Pedersen, MD, Lars Kprber, MD, Frank Steensgaard-Hansen, MD, Yvonne H. Rasmussen, MD, Jens Berning, MD, Knud Skagen, MD, and Asger Pedersen, MD

To determine the prognosis of late ventricular fibrillation (VF) after acute myocardial infarction (AMI), the length of the monitoring period after AMI was extended. All patients in this series were continuously monitored in a coronary care unit to ensure observation of all VF within 18 days of AMI. From 1977 to 1966,4,269 patients were admitted with AMI and 413 (9.6%) had in-hospital VF. Df these 261 (6.8%) had early VF ( 10 seconds. VF was treated promptly (within 1 minute) with direct current conversion (320 watts/second) and a IOO-mg lidocaine bo-

lus followed by an infusion of 180 mg/hr for 24 hours. The infusion was stopped if ventricular ectopic activity ceased. Further antiarrhythmic treatment was given only if recurrent ventricular arrhythmias occurred (VF, ventricular tachycardia or ventricular ectopic beats [Lown class III or IV]). Mexiletine was the first choice and if clinically unsuccessful other antiarrhythmics were tried (quinidine, p blockers, amiodarone). Efficacy was tested by continuous electrocardiographic monitoring during the entire hospitalization period and by exercise testing. Heart failure was consideredpresent when basal bilateral rales were present on lung stethoscopy or when chest x-rays showed pulmonary congestion or pulmonary edema requiring diuretic treatment. Cardiogenic shock, defined as blood pressure 580 mm Hg for at least 30 minutes accompanied by clinical signsof shock, oliguria and absence of other explanations for shock, was included as a sign of heart failure. The presence of heart failure was usually determined at the time of first VF. Long-term survival was determined by using the Danish Central Person Register, which immediately registers all deaths in Denmark. One patient, a foreigner, could not be traced and was consideredcensored at discharge. The search was performed in August 1987 ensuring at least 1.8 years of follow-up and >5 years of follow-up for most patients. Statistical analysis: Median and range were used to describe continuous variables. Chi-square tests were used for contingency tables. For use in regressionmodels logarithmic transformation of age and maximal value of creatine kinase-MB were used. The distribution was more symmetric and also these transformed variables fitted better in proportional hazard models. Influence of possible risk factors for in-hospital mortality was estimated by stepwise linear logistic regression using the BMDP-LR program and the log-likelihood method.‘) For long-term survival proportional hazard models were used using the BMDP-2L program. Bivariate variables were assignedthe value of zero when the factor was not present and the value of 1 when present. The assumption of proportional hazard was checked graphically id and found reasonable with all variables usedexcept diabetes. In this caseanalysis indicated (basedon only 13 cases)a grossly increasing hazard with time. Therefore the importance of diabetes was estimated by the log-rank test and diabetes was otherwise entered in the proportional hazard model in strata. Kaplan-Meier estimators of the survivor function were calculated at various times including 95% confidence limits. Relative risks (odds ratio) and their confidence limits were calculated from logistic regressionand Cox models. RESULTS The cumulated distribution of VF on each day after AM1 is shown in Figure 1. Sixty-two percent of first VF occurs on the first day and 80% within the first week. Recurrent VF, which in this context is VF on a separate day, was much rarer. According to the first case of VF THE AMERICAN

JOURNAL

OF CARDIOLOGY

JULY 1. 1990

11

TABLE

I Characteristics

of 413 Patients

with Acute

Myocardial

Infarction

All Pts (n = 413)

Median age (yrs) Female (%) Previous CAD* (%) Previous AMI (%) Anterior AMI (%) Heart failure (%) Pulmonary edema (%) Cardiogenic shock (%) Systemic hypertension (%) Diabetes mellitus (%) Valvular disease (%) Thyroid disease (%) Serum-K (mmol/liter) median Serum-K (mmol/liter) (25-75X

63 (range 118(29) l@JW) 140 (34) 184 (45) 195 (47) *(II) 55 (13) ~(19) 36(9 7 (2) ll(3) 3.8 3.5-4.1

32 to 87)

Early VF is defined as VF within 48 hours of AMI. late as >48 hours after AMI. *Previous CAD is prewous known angina pectons or AMI. AMI = acute myocardial Infarction; CAD = coronary artery disease: VF = ventricular

and Ventricular

Fibrillation

Early VF (n = 281)

Late VF (n = 132)

62 80 (29) 121(43) 96 (34) 115(41) 110 (39) 33(11) 36 (13) 57 (20) 23 (8) 4(l) 8 (3) 3.8 3.4-4.1

65 38 (29) 59 (45) 44 (34) 69 (52) 85 (64) 13(10) 19(14) 23(17) 13(10) 3 (2) 3 (2) 4.0 3.7-4.3

0 Value NS NS NS NS 18

16

AMI

:HE

AMERICAN

JOURNAL

OF CARDIOLOGY

JULY

1. 1990

13

value of signs of heart failure has to some extent been exhausted because the analysis is restricted to those surviving hospital stay (p = 0.0018) and the relative risk is 1.8 (95% confidence limits 1.1 to 2.8). This it not unexpected. Patients with the most severe heart failure will be expected to die in the hospital. DISCUSSION The present investigation clearly demonstrates that among patients with AM1 complicated by VF prognosis is determined by the presence or absence of signs of heart failure. Late and early VF has the same prognosis when not associated with heart failure. Clinical signs of heart failure were more common among patients with late VF. The tradition of dividing VF into early and late (occurring < or >48 hours of AMI) probably stems from early traditions of monitoring patients with AMI. The choice of 48 hours might not be optimal, but detailed analysis of occurrence of VF with time after AM1 did not demonstrate any high- or low-risk periods. The present study is to our knowledge the only one in which the importance of late VF can be estimated without selection bias introduced by early discharge or early termination of monitoring. Use of validated strict procedures for discharge is rarely used elsewhere. Goldberg et al* demonstrated a better prognosis for patients with cardiac arrest