Changing Incidence of Out-of-Hospital Ventricular Fibrillation, 1980-2000 FREE

The number of patients in Seattle, Wash, treated for out-of-hospital cardiac arrest with ventricular fibrillation (VF) as the first recorded rhythm has decreased since 1980.¹ Such a change might be anticipated in view of the overall reduction in coronary heart disease mortality in the United States.² Since out-of-hospital deaths contribute substantially to coronary heart disease mortality,³ it follows that the incidence of out-of-hospital cardiac arrest may be reduced concomitantly.

Kuisma and colleagues⁴ recently described a marked decrease in the incidence of out-of-hospital VF from 1994 to 1999 in Finland. Herlitz et al⁵ reported a modestly decreased incidence of VF cases during a 17-year period in Sweden but no appreciable change in the total number of treated cardiac arrests.

The purpose of this report is to analyze the changing pattern of out-of-hospital cardiac arrest in Seattle during the past 20 years in terms of incidence, relationship to initially recorded cardiac rhythm, and sex and race of patients.

The Seattle Fire Department's emergency medical services (EMS) system, Medic One, has been described previously.^1,6 This system is the sole provider within the city for medical emergency response dispatched for 911 calls and it is 2-tiered, with mean initial responses by emergency medical technicians of 3.5 to 4.0 minutes and paramedic responses of 7.5 to 8.0 minutes following dispatch. Since 1970, details of all cardiac arrests attended by Medic One in Seattle have been entered into a comprehensive database that includes age, sex, race, initial cardiac rhythm, treatment, and EMS parameters relevant to patient outcome.

The patients described in this report were consecutive cases of cardiac arrest who received advanced life support by Seattle Fire Department paramedics, including patients (9.2%) whose cardiac arrests were witnessed by EMS personnel. In this report, advanced life support is defined as the application by a paramedic of 1 or more of the following: tracheal intubation, precordial shock for VF or ventricular tachycardia, intravenous line placement, and administration of medication. Patients whose arrest was apparently not due to cardiac causes (eg, trauma, exsanguination, drug overdose) were excluded from this analysis. Three patients with VF, all in 1990, received 1 or more precordial shocks but no further EMS interventions. Because of end-of-life instructions or cardiac arrest in a hospital, these 3 patients were not included as treated and were not considered in the present analysis.

The primary source of these data was the medical incident report submitted by fire department paramedics for all patients attended. Each report was reviewed to identify treated cardiac arrests and edited for omissions or inconsistencies. Any missing data were obtained from death certificates or hospital admission forms. During the years covered in this report, we adopted several procedures to ensure identification of all cases, primarily formal periodic checks against fire department records for complete ascertainment. After 1990, we accounted for cases by tabulation of all sequentially numbered reports and after 1998 by additional electronic verification. Use of automated external defibrillators (AEDs) by first responders, begun in 1984 and phased in completely by 1990, provided another cross-check. Receipt of AED tapes and telephone notification of each instance of AED use were checked against the medical incident reports identified as cases of cardiac arrest. Beginning in 1994, 2 screeners identified cases by reviewing duplicate sets of forms. Such efforts have ensured that few, if any, cases were missed and would be expected, if anything, to enhance the identification of cardiac arrests in later years. The data are presented in 3 periods, each including a US Census year: 1979-1980, 1989-1990, and 1999-2000.

During the earliest period, the determinations of initial cardiac rhythm were obtained from defibrillator monitors used by paramedics who typically arrived on the scene approximately 4 minutes after the first responders. During the latter 2 periods, the initial rhythm was usually based on tracings obtained with AEDs used by the first responding emergency medical technicians.⁷ In view of the small number of patients with ventricular tachycardia and cardiac arrest (ie, 1% of those found with VF), patients with ventricular tachycardia are counted with those whose first recorded rhythm was VF.

Because of the small number of cardiac arrests among persons younger than 20 years, they were excluded; accordingly, the census information was also restricted to those aged 20 years or older. Race-specific incidence rates were calculated for whites, blacks, and Asians/Pacific Islanders; the latter were combined to provide comparable racial classification across the 3 census periods. The numbers of cases and population size were limited for other racial groups, so they were not considered separately in this report. In calculating age- and sex-adjusted annual incidence rates, we counted the events during periods of 2 years and halved the resultant (2-year) rates. The incidence rates in persons aged 20 years or older in the earlier 2 periods were adjusted for age and sex to those of the 2000 Seattle census and the US 2000 census for all periods.⁸

The significance of linear trends for incidence over the 3 periods was estimated according to the Mantel-Haenszel test of linear association, and 2-tailed P values are reported.⁹ Relative risks (RRs) and their 95% confidence intervals (CIs) are reported for the rates in 1999-2000 using 1979-1980 rates as the baseline.¹⁰ We used 1-way analysis of variance and the resulting linearity significance level to examine trends in mean ages. P<.05 was considered significant for all analyses.

Since 1980, the number of VF cases treated annually has declined, with an approximate 50% reduction in the number treated each year (Figure 1). Although the population of Seattle aged 20 years or older increased modestly from 1980 to 2000, the total number of treated cardiac arrests per year decreased by 31%, related to a substantial decline in patients whose first recorded rhythm was VF (Table 1). The overall adjusted incidence rate for all treated arrests with presumed cardiac etiology declined by 34% (RR, 0.66; 95% CI, 0.58-0.75). The rate declined for men (43%), but only to a minor extent for women (10%).

After age and sex adjustment to the 2000 US population, the annual incidence of VF in Seattle declined by 56% (RR, 0.44; 95% CI, 0.37-0.53) during this 20-year span. Reductions in VF incidence rates were evident in both men and women (57% and 51%, respectively). Among patients with VF, declining incidence rates were evident in virtually all age strata (data not shown). The incidence of VF in men far exceeded that of women, and the ratio of male-female incidence rates decreased only from 4.0 to 3.5 in 20 years.

The incidence of asystolic arrests and the overall incidence of patients treated for pulseless electrical activity (PEA) did not demonstrate consistent changes over the study period (Table 1). In men, the incidence of treated cardiac arrests with asystole or PEA tended to decrease, whereas in women, these rates tended to increase. The overall changes in rhythm-specific incidence rates in Seattle over the 3 periods are shown in Figure 2.

The incidence of VF cardiac arrests decreased significantly in blacks and whites but not in Asians/Pacific Islanders over the study interval. Whites experienced a decrease of 53% (RR, 0.47; 95% CI, 0.38-0.58), from 0.85 to 0.40 per 1000. Blacks had a similar 54% decrease in VF incidence (RR, 0.45; 95% CI, 0.26-0.79), but that decline was offset by an increase in the incidence of arrests with asystole (0.45-0.81 per 1000) (Table 2). Reduction in the overall incidence of treated cardiac arrests was apparent in whites but was less prominent or not observed among blacks and Asians/Pacific Islanders, the latter group having relatively few cases. However, the proportion of patients with cardiac arrest discharged alive from the hospital did not change whether the initial rhythm was VF, PEA, or asystole.

In Seattle, a city with a population of about 560 000, there has been a marked reduction in the incidence of treated out-of-hospital cardiac arrest, particularly in cases in which the initial recorded rhythm was VF. The decline began in about 1980 and persisted to 2000. This is not an isolated observation; decreases in the incidence of VF cardiac arrests have also been reported in Göteborg, Sweden,⁵ and Helsinki, Finland.⁴

Although the incidence of VF cardiac arrests was reduced among men, women, blacks, and whites, the basis for the declining incidence of VF is only speculative. It seems likely, however, that this reflects the general reduction in age-adjusted mortality attributed to coronary heart disease.^2,11 Further clarification is required to determine the relative influences of primary and secondary preventive measures as well as improved medical/surgical outcomes.¹²

In searching for alternative explanations for the declining incidence of treated VF, we considered the possibility that compared with 10 or 20 years earlier, more patients in recent years were found dead and without indication for attempted resuscitation. We were able to examine this possibility with 339 untreated patients aged 20 years or older from a 12-month case series in 1989 and 730 patients from 1999-2000. All were declared dead on arrival without obvious cause, except for a small group with end-of-life instructions. Slightly more than half of patients in these groups were men, deaths were unwitnessed in 98%, and 98% occurred at home or in a nursing home. The age- and sex-adjusted (US Census, 2000) annual incidence of these events was 0.85 per 1000 aged 20 years or older in 1989 and 0.87 in 1999-2000. Thus, the limited available data do not support this alternative explanation.

An additional explanation might be that patients with symptoms of acute myocardial infarction in latter years of the study incurred less delay in seeking care than those in earlier years. In that scenario, the site of VF might be transferred from the out-of-hospital setting to the hospital. However, there is little reason to believe that this has occurred, particularly in the past decade. Efforts to reduce patient delay in reacting to symptoms of myocardial infarction, including a program in the Seattle area, have met with little or no success.^13- 14 Additionally, national trends over a recent 4-year period showed only minor, nonsignificant decreases in delay in seeking care and unchanged median delay times at 2.1 hours.¹⁵

Although we noted a reduction in the incidence of treated cardiac arrest with VF as the initial rhythm, we did not observe a concomitant decrease in the incidence of asystole or PEA. The reasons for this discrepancy are uncertain; however, we believe it is likely that an appreciable proportion of those with nonshockable rhythms developed cardiac arrest from noncardiac causes or developed circulatory arrest from cardiac pump failure without an arrhythmic event. Whereas the vast majority of VF cardiac arrests are related to an underlying cardiac disorder, our impression is that cardiac causes are less frequent in persons with asystole or PEA, even after excluding obvious noncardiac causes such as drug overdose, trauma, exsanguination, and primary respiratory failure.^16- 18 Patients with asystole or PEA were, on average, older than those with VF; many could well have died from conditions not affected by the improvement in nationwide coronary mortality.

There are areas for potential inaccuracy in the reporting of these trends. First, US Census data are not free of error.¹⁹ Race as tabulated in 2000 is not directly comparable with prior enumerations because persons may be counted as combinations of 2 or more races rather than according to the first race listed, as in prior census years.²⁰ The number of black Seattle residents aged 20 years or older who were counted as of "single race" in 2000 declined in comparison with the 1990 population (33 725 vs 34 730), while 22 880 persons aged 20 years or older were counted in 2000 as being of 2 or more races. This may provide evidence that potential underreporting is not proportional across the 3 racial groupings. The effect of underreporting by race in 2000 would be to diminish a decline in incidence by that race; this possibility may play a role in the apparent increase in asystole reported for black patients in 2000 (Table 2).

Second, the incidence rates may be spuriously elevated by including cardiac arrests among nonresidents in Seattle or lowered by not including city residents who developed cardiac arrest elsewhere. For cardiac arrest cases in 1999 and 2000, 92 (12%) of the 744 were nonresidents of Seattle. Nonresident men with treated cardiac arrest were appreciably younger than residents: mean, 59.1 (SD, 12.8) years vs 70.1 (SD, 14.5) years. However, the mean ages of female residents and nonresidents were similar (72.5 and 71.9 years, respectively). The 1990 census described a net influx of about 89 000 commuters into Seattle (113 000 in 1980), and it is likely that our calculated incidence rates are modestly overestimated by using the Seattle population denominators.^21- 22 Comparable census data are not yet available for 2000. However, it also seems most probable that our reported trends over time are minimally, if at all, affected by travel into or out of Seattle.

Although unlikely to significantly affect our findings and conclusions, we note that the means for recognizing VF in Seattle changed after about 1985. At that time, the determination of initial rhythm began to shift to review of recordings from AEDs⁷ rather than from paramedic observations, which had been carried out a mean of 4 to 5 minutes later in the course of resuscitation. However, tracings obtained later in the course of the cardiac arrests, as in 1979-1980, most likely would serve to identify fewer cases of VF compared with those in which first-arriving EMS personnel applied AEDs (as in the latter 2 periods), thereby tending to obscure any decrease in incidence over a span of time. Of course, any changes in the identification of VF could not affect the overall incidence of cardiac arrest.

Extrapolation of the 2000 incidence rates of cardiac arrest for persons aged 20 years or older in Seattle to a national level suggests that approximately 184 000 treated cardiac arrests due to presumed cardiac causes could be anticipated in the United States each year. Of these, about 76 000 would have VF as the initially recorded rhythm. These numbers contrast with the estimated 400 000 to 460 000 sudden cardiac deaths in the United States.³ This difference is related to the substantial number of all-cause out-of-hospital deaths that were not treated and to the considerable likelihood of overestimation of sudden cardiac deaths when relying on data from death certificates.^23- 24 Furthermore, suddenness of death does not equate treatable arrhythmic events. From an EMS perspective, the number of treatable VF cardiac arrests are most relevant; these may represent only about one fifth of so-called sudden cardiac deaths. A caveat in this extrapolation is to recognize the potential limitations of our use of a population sample from a single geographic site that represents only about 1/500 of the US population.

In summary, the incidence of cardiac arrest due to VF has substantially declined in Seattle. Since 2 European communities have also reported declining numbers of cases, we conclude that arrhythmogenic cardiac arrests are less commonly encountered now than 10 to 20 years ago.