Jacob Hollenberg, Division of Cardiology, South Hospital, SE-118 83 Stockholm, Sweden. (fax: +46 8 6163040; e-mail: email@example.com).
Background. Dramatic differences in survival after out-of-hospital cardiac arrests (OHCA) reported from different geographical locations require analysis. We therefore compared patients with OHCA in the two largest cities in Sweden with regard to various factors at resuscitation and outcome.
Setting. All patients suffering an OHCA in Stockholm and Göteborg between 1 January 2000 and 30 June 2001, in whom cardiopulmonary resuscitation (CPR) was attempted were included in this retrospective analysis.
Results. All together, 969 OHCA in Stockholm and 398 in Göteborg were registered during the 18-month study period. There were no differences in terms of age, gender, and percentage of witnessed cases or percentage of patients who had received bystander CPR. However, the percentage of patients with ventricular fibrillation (VF) at arrival of the ambulance crew was 18% in Stockholm versus 31% in Göteborg (P < 0.0001). The percentage of patients who were alive 1 month after cardiac arrest was 2.5% in Stockholm versus 6.8% in Göteborg (P = 0.0008). Various time intervals such as cardiac arrest to calling for an ambulance, cardiac arrest to the start of CPR and calling for an ambulance to its arrival were all significantly longer in Stockholm than in Göteborg.
Conclusion. Survival was almost three times higher in Göteborg than in Stockholm amongst patients suffering an OHCA. This is primarily explained by a higher occurrence of VF at the time of arrival of the ambulance crew, which in turn probably is explained by shorter delays in Göteborg. The reason for the difference in time intervals is most likely multifactorial, with a significantly higher ambulance density in Göteborg as one possible explanation.
The majority of sudden cardiac deaths occur out of hospital . The ‘chain-of-survival’ concept  is nowadays well established for prehospital measures that increase survival after out-of-hospital cardiac arrest (OHCA), comprising early access, early cardiopulmonary resuscitation (CPR), early defibrillation and early advanced life support.
Survival after OHCA varies markedly in different studies, ranging from 1.4%  to 23% . Few analyses have so far been performed to explain these somewhat dramatic differences in survival between various ambulance organizations.
Previous experience indicates that survival after OHCA in Stockholm (Sweden's largest city) is markedly lower than that in Göteborg (Sweden's second largest city) [5, 6].
The aim of this retrospective study was to systematically investigate whether there was an actual difference in survival after OHCA between the two largest cities in Sweden, Stockholm and Göteborg, during an 18-month period, and to study underlying mechanisms behind a possible difference. No previous study has compared the outcome after OHCA between two national cities with almost identical rescue organizations.
Patients and methods
Ambulance registry and ambulance organizations
This study is a retrospective study based on data prospectively collected at the Swedish Ambulance Cardiac Arrest Registry. The registry was started in 1990 with a few ambulance services and has successively been joined by the majority of ambulance services in Sweden. The registry covers a large proportion of a total of 8.9 million inhabitants of Sweden. For several years, the ambulance organizations in Stockholm and Göteborg have been included in the registry.
The ambulance organizations in Stockholm and Göteborg work according to a two-tier system, i.e. for each call judged as a cardiac arrest one mobile coronary care unit (if available), as well as an ambulance are dispatched. Ambulances do not generally carry nurses but are equipped with specialized ambulance personnel, all trained in basic CPR and the use of defibrillators. The educational programme for the ambulance personnel is the same in both cities. The mobile coronary care units in both cities are equipped with a Registered Nurse trained in anaesthesiology with additional courses in advanced cardiac life support. All vehicles are equipped with defibrillators, both mobile coronary care units and ambulances. The ambulances are stationed throughout the two cities and are not predominantly situated at the hospitals. The prehospital (ambulance) pharmacological treatments do not differ between Göteborg and Stockholm or other parts of Sweden.
For each case of OHCA the ambulance crews in Stockholm and Göteborg complete a form (the same form is used in both cities) with relevant information such as age, place of arrest, bystander CPR (a bystander was defined as someone starting CPR before the arrival of the first ambulance, regardless of profession), witnesses, resuscitation procedure, probable cause of arrest, intervention times, defibrillation, intubation, drug treatment, type of initial rhythm, and clinical findings at first contact.
In ambulances with manual defibrillators, the rhythm was defined as ventricular fibrillation (VF), pulseless electrical activity (PEA) or asystole. If automated external defibrillators were available, the rhythm was defined as shockable rhythm [VF or non-shockable rhythm (PEA or asystole)]. In this study, VF includes patients with pulseless ventricular tachycardia (VT). The rhythm was analysed by the ambulance personnel in each case.
To estimate the time of cardiac arrest in witnessed cases, the ambulance crew was instructed to interview the bystanders about the delay from arrest to call. It was stressed in written instructions that maximum effort had to be made to obtain these times. The ambulance crew recorded their time of arrival at the patient's side, the time of starting CPR, the time of the first defibrillation, the time of a palpable pulse, the time of starting transport to hospital and arrival at hospital. The number of direct current shocks was recorded. The ambulance crew also classified the aetiology of the arrest in nine different diagnostic categories (heart disease, lung disease, trauma, drug overdose, suicide, drowning, suffocation, sudden infant death syndrome and other) based on clinical assessment and bystander information. Immediate outcome was reported by the ambulance crew as dead on arrival, dead in the emergency room (ER) or admitted alive to the hospital. The latter was defined as a patient admitted alive from the emergency departments to the hospitals and, thus, were not declared dead in the ER.
The form was completed during and immediately after the acute event. Each form was sent to the medical director and a copy was sent to the central registry in Göteborg. An extra copy of the protocol was completed 1 month after the arrest with additional information about subsequent outcome (dead or alive) for patients admitted alive to the different hospitals. In cases with incomplete information this was checked with the Swedish National Registry of Deaths. All data were computerized in a database in Göteborg.
The regular registry protocol described above was used for this study. This study was approved by the local ethics committee.
The municipality of Göteborg has a population of 471 000 inhabitants with a population density of 1050 inhabitants per km2 compared with a population of 1 640 000 inhabitants with a population density of 283 inhabitants per km2 in Stockholm.
The two cities are the largest in Sweden. The proportion of men in both Stockholm and Göteborg is equal, 49%. Age distribution is shown in Fig. 1. The mortality per 100 000 inhabitants in ischaemic heart disease and acute myocardial infarction during 2000–01 was slightly higher in Göteborg compared with Stockholm 215 vs. 170 and 120 vs. 98 respectively .
All cases of OHCA in Stockholm and Göteborg, where any type of resuscitation measure (airway assistance, chest compressions, administration of drugs, intubation, defibrillation) was implemented, between 1 January 2000 and 30 June 2001, were included. Ambulance personnel in Stockholm and Göteborg have used identical guidelines for exclusion, i.e. when not to start resuscitation. These cases include rigour mortis and marbleness. As a result, patients with cardiac arrest prior to the arrival of the ambulance, as well as during ambulance transport, were included in the survey. Furthermore, the criteria for the abortion of CPR are the same in both cities (asystole more than 30 min) and these decisions could only be taken by the mobile care unit personnel.
Distribution is given as the mean ± standard deviation, medians and percentages. For comparisons between groups regarding continuous variables, Fisher's nonparametric permutation test was used. For comparisons of dichotomous-dependent variables, Fisher's exact test was used. A two-tailed test was applied. Stepwise logistic regression was used to select independent predictors of dichotomous-dependent variables.
The following variables were entered: (i) age (continuous variable); (ii) gender (men or women); (iii) place of arrest (home versus elsewhere); (iv) aetiology (cardiac versus non cardiac); (v) region (Stockholm versus Göteborg); (vi) witnessed status (witnessed versus non witnessed), (vii) CPR performed prior to ambulance arrival (bystander CPR) (yes versus no); (viii) OHCA between 8 am and 8 pm (yes versus no); (ix) VF on initial ECG (yes versus no); (x) interval between cardiac arrest and arrival of ambulance (natural logarithm); (xi) interval between call for ambulance and arrival of ambulance (natural logarithm).
Patient characteristics and location of cardiac arrests
There were 969 OHCA in Stockholm and 398 in Göteborg. The incidence was 35 per 100 000 inhabitants per year in Stockholm and 56 per 100 000 inhabitants per year in Göteborg.
The two groups were similar in terms of age, gender and various factors at resuscitation with exception for VF, which was more common in Göteborg. (Table 1).
Table 1. Baseline characteristics (all patients)
Stockholm n = 969
Göteborg n = 398
aPercentage of patients with missing information. CPR, cardiopulmonary resuscitation. P-values denoted if <0.2.
Age (years), mean (SD) (9, 6)a
Gender (male), % (n) (6, 6)
Witnessed arrest by bystanders, % (n) (11, 8)
Witnessed arrest by ambulance personnel, % (n) (11, 8)
CPR started prior to arrival of ambulance crew, % (n) (15, 19)
Out-of-hospital cardiac arrest between 8 am and 8 pm, % (n) (13, 20)
VF on initial ECG, % (n) (16, 14)
Cardiac arrest at home, % (n) (1, 15)
Time intervals (min, median)
Call for to arrival of ambulance (13, 24)
Cardiac arrest to call (17, 22)
Cardiac arrest to start of CPR (22, 30)
Cardiac arrest to arrival of ambulance (19, 33)
Call for to arrival of ambulance (7, 14)
Call for ambulance to first defibrillation (13, 17)
The majority of cardiac arrests in both cities occurred at home (Table 1). Other places of cardiac arrest were: public places (15% in both cities), in nursing homes (5% in Stockholm, 6% in Göteborg), in ambulances (4% in Stockholm and 5% in Göteborg), at work (3% in both cities) and ‘other’ (13% in both cities).
The mean distance from the site of the cardiac arrest to the nearest hospital was 6.4 km in Stockholm and 5.0 km in Göteborg. The mean distance from an ambulance station to the place of the cardiac arrest was 11.7 km in Stockholm and 8.5 km in Göteborg.
The percentage of arrests that occurred <10 km from an ambulance station was 79% in Stockholm and 81% in Göteborg.
Cause of cardiac arrest
In Stockholm, 54% of the arrests had a cardiac aetiology versus 60% in Göteborg. In 33% in Stockholm and 31% in Göteborg, the cause was either ‘other or unknown’.
The time intervals were generally longer in Stockholm (Table 1). Almost 55% of victims in Göteborg were reached by an ambulance within 5 min after call versus 20% in Stockholm (Fig. 2). Less than 5% of patients found with VF in Stockholm were defibrillated ≤5 min after call versus 40% in Göteborg (Fig. 3).
The number of ambulances per million inhabitants differed between the two cities. During the daytime (7 am–10 pm), Stockholm had on average 31 ambulances per million inhabitants, versus 41 in Göteborg. During the night (10 pm–7 am), Stockholm had 19 ambulances per million inhabitants, versus 30 in Göteborg.
Information on survival was available for 98% of patients in Stockholm and 97% in Göteborg. In Stockholm, 158 patients (16%) were admitted alive to the seven major hospitals, whereas 119 patients (30%) were admitted alive to the two major hospitals in Göteborg. In Stockholm, 24 patients (2.5%) survived to 1 month, versus 26 patients (6.8%) in Göteborg. Survival in various subgroups is shown in Table 2.
Table 2. Alive after 1 month
Stockholm (15 missing)
Göteborg (13 missing)
aNumber of patients alive/number of evaluated patient values.
Values in parentheses are expressed as percentage.
Distance between cardiac arrest and ambulance station
Interval between call for and arrival of ambulance
≤7 min (median)
Time of cardiac arrest
8 am–8 pm
8 pm–8 am
Cardiac aetiology found in VF
When including all variables listed above in Statistical methods in the multivariate analysis only VF on initial ECG appeared as an independent predictor for an increased chance of survival (odds ratio 15.0; 95% CI 5.3–53.3). When only including patients found in VF in the model only bystander performed CPR appeared as an independent predictor for an increased chance of survival (odds ratio 3.4; 95% CI 1.1–13.0). When only including patients with a bystander witnessed cardiac arrest and an interval between cardiac arrest and arrival of ambulance of <10 min, VF on initial ECG appeared as the only predictor for an increased chance of survival (odds ratio 10.6; 95% CI 1.8–202.0). The model included 75% of all patients.
The rather dramatic difference in survival after OHCA between two cities with similar rescue organizations needs to be explained. This study analysed in depth the mechanisms behind a difference in survival after OHCA. We found that the 1-month survival was almost three times higher in Göteborg than in Stockholm. All time intervals that were analysed were longer in Stockholm. It is possible that the large difference in land area would make the comparison unfair. However, the percentage of arrests that occurred <10 km from an ambulance station was similar in Stockholm and Göteborg (approximately 80%). Differences in distances to the ambulance station did not appear to influence mortality data between the two cities.
The incidence of OHCA in which CPR was attempted was higher in Göteborg than in Stockholm. This can be explained by the fact that there is a higher mortality in ischaemic heart disease and acute myocardial infarction in Göteborg.
Another reason for the higher incidence of OHCA in Göteborg may be that this town is smaller and more densely inhabited than Stockholm. It is therefore possible that more patients with OHCA in more remote areas in Stockholm will not be candidates for CPR due to the large time interval between the occurrence of death and the arrival of the ambulance. These patients will not be included in the registry as CPR was not attempted.
Another possible explanation for this difference could be variations in the quality of including patients in the registry. Although this cannot be completely ruled out it seems unlikely as repeated audits of quality of the registry have not disclosed any such difference.
In Stockholm, 18% of the patients had an initial VF/VT. This is a low figure in comparison with both national and international reports [4, 8]. The observation that only half as many patients in Stockholm had a shockable rhythm on the first ECG recording than patients in Göteborg is likely to be due to the longer time intervals in Stockholm.
Reducing the time to CPR and to the first defibrillation increases the percentage of patients found in VF and, consequently, the overall survival [8–11]. The estimated time from cardiac arrest to calling for an ambulance, from calling for an ambulance to its arrival, from calling for an ambulance and the start of CPR and from calling for an ambulance to the first defibrillation were all longer in Stockholm than in Göteborg. These are probably the most important reasons for the large difference in the percentage of patients found in VF as well as overall survival.
The difference in time intervals can be explained by several factors. The number of ambulances per million inhabitants was higher in Göteborg. The traffic situation is probably worse in Stockholm (as it is a larger city). There was, however, no indication that survival differed more during the daytime than during night. Furthermore, distances between the place of cardiac arrest and the ambulance station tended to be larger in Stockholm. However, even when cardiac arrests close to the nearest ambulance station were compared with those occurring farther away there was still a higher survival rate in Göteborg. Finally, the difference in the estimated time from cardiac arrest to calling could perhaps be a consequence of a higher alertness to respond in the general population in Göteborg. For several decades, educational programmes and mass medial campaigns about cardiac disease, cardiac arrest and CPR has been offered to large parts of the population in Göteborg. This possible difference in higher alertness is, however, not supported by any data.
The fact that the distance from cardiac arrest to the hospital was longer compared with the ambulance station indicates that ambulances should be relocated to places were cardiac arrest most likely occur. The reason for this difference is most likely due to the fact that ambulances in both Stockholm and Göteborg generally tend to depart to new assignments from the hospitals (rather than from the ambulance stations) after completing their previous assignments.
A study from Stockholm in 1978 revealed a survival after OHCA of 3.8%  and a later investigation from 1987 showed similar figures with a survival of 3.6% . Thus, survival in Stockholm has not improved over the last decades; instead it has decreased. In addition, survival in Göteborg is lower than reported during the last 2 decades [4, 6, 13]. These figures may be compared with those from centres that present more favourable results [4, 14] as well as those with similar or even lower survival rates [3, 15].
Maybe, and perhaps as important, is the difference in the proportion of patients being hospitalized alive, i.e. 30% in Göteborg and 16% in Stockholm. Unlike the 1-month survival figures which also reflect the hospital course, these figures reflect the prehospital care in a more direct way.
Bystander CPR increases survival after OHCA [16, 17]. Our figures of 36% in Stockholm and 34% in Göteborg are consistent with previously reported national data from the past few years . However, the quality of bystander CPR is also important . Göteborg has been a pioneer city in Sweden with respect to teaching CPR to the lay population. The possibility that the awareness of OHCA and the quality of B-CPR performed in Stockholm were inferior to that in Göteborg cannot be excluded. We therefore compared survival between subgroups from Stockholm and Göteborg which were reached by the ambulance crew within 7 min after call (overall median). In this subgroup, survival was 4.2% in Göteborg versus 2.6% in Stockholm.
Based on analyses at the Swedish Cardiac Arrest Registry it can be estimated that reduction of the interval from call to arrival of ambulance from 7 to 5 min in Stockholm would save another five to 10 lives per annum.
As the present study focuses on the out-of-hospital events, information regarding hospital care has not been included in the evaluation. Accordingly, we can only speculate on the role of differences in hospital treatment on the overall outcome.
A number of strategies may improve survival after OHCA in metropolitan areas in future. These include public access defibrillation , defibrillation initiated by policemen , increased use of CPR guidance by telephone from dispatchers to bystanders , increased public awareness , increased ambulance density, more aggressive hospital treatment including use of hypothermia  and increased use of ICD among survivors . A recent study has shown that trained volunteers in a structured response system can increase the number of survivors to hospital discharge after OHCA in public places .
1The aetiology behind the cardiac arrest was based on the judgement of the ambulance crew. It is difficult in an OHCA situation to know the probable background of a cardiac arrest. However, it is important to extract as much information as possible from relatives and witnesses about what happened before cardiac arrest and this was emphasized in written instructions to the ambulance personnel.
2There was no information on medical history.
3Information on some variables was lacking in a proportion of patients, particularly in regard to time intervals for bystander-witnessed cardiac arrests. Therefore, these numbers should be interpreted with caution.
4We have compared two relatively large cities with a marked difference in size of the population being served by the two ambulance services and an even more marked difference in area in which they work. We cannot exclude the possibility that there are other differences which might explain some of the results that were found.
Survival after OHCA in Stockholm is almost three times lower compared with Göteborg and the percentage of patients with shockable rhythms is far lower in Stockholm than in Göteborg. This is most probably due to the longer time intervals in Stockholm from the cardiac arrests to critical resuscitation measures, such as the arrival of the ambulance, the start of CPR and defibrillation. The reason for the difference in time intervals is most probably multifactorial, with a significantly higher ambulance density in Göteborg as one probable explanation.