The main results of this study show that OHCA recognition rates in cases were no CPR was initiated prior to call varied from 71 to 96% between three capital city EMDCs in Scandinavia. Time to recognition varied by over half a minute, and time to first chest compression varied by over a minute. We question whether this is due to differences in dispatcher performance or differences in systems as system variables were difficult to compare objectively.
International literature shows extensive variations in recognition sensitivity of OHCA. 13]. Reasons for the significant difference in OHCA recognition in this study is multifaceted. Oslo had 1 ½ years prior to this study undergone a targeted intervention to improve performance in recognition rates and DA-CPR . ‘Attention to appropriate handling of cardiac arrest calls was an ongoing priority in the EMDC in 2016, which may have affected the high level of recognition in Oslo. The ERC guidelines 2015 state that dispatchers can improve recognition by focusing on “unresponsiveness” and “not breathing normally” . All three sites addressed consciousness and breathing in most cases, but Oslo was the only site also addressing abnormal breathing in most cases. This might be the main reason for differences in recognition rates between the three countries.
Differences in educational levels or professional background of the dispatchers might also be factors impacting on recognition rates. In Stockholm, EMDs were not necessarily health care personnel, and only 20% were nurses. In Oslo and Copenhagen, EMDs were nurses or paramedics.
Dispatcher performance can directly affect OHCA recognition rates by including false OHCA cases in the cardiac arrest registry. Inclusion criteria in the cardiac arrest registries in the Scandinavian countries are the same; all cases where anyone at scene (bystander or EMTs) have started CPR, is included in the registry. But if callers are instructed to start CPR in unclear cases where the patient is actually not in cardiac arrest, they are still included in the registry. Hence, pro-active dispatchers can increase the incidence numbers in the cardiac arrest registries by providing unnecessary CPR instructions. Although not reported in this study, these cases might also affect outcome data, possibly including both cases were the patient was never in cardiac arrest, as well as cases where no treatment was started by EMT due to futility.
Despite a high proportion of recognised cases in all three sites, and AEDs mentioned in the protocol, AEDs were rarely addressed in the NO-CPRprior group. Reasons for this is unknown, but a study from Sweden exploring this specifically found that AEDs were not nearby in 93% of the cases. Other reasons might be inaccessible AEDs and caller being alone . There were significant differences between sites on AED addressed in this study (11% in Copenhagen, 4% in Stockholm and 3% in Oslo). Longstanding efforts from the Danish AED registry might have affected this. Oslo did not have a functional AED registry at the time of data collection.
There was seemingly a relatively long time interval from OHCA recognition to start of first chest compression instruction in all sites (00,54 in Copenhagen, 01:27 in Stockholm, 01:05 in Oslo). This was due to the Cares definition of time interval for “chest compression instruction”, which states that “Instructions to get a patient to a hard, flat surface should not be considered the start of CPR instructions. Instructions begin when a call-taker or dispatcher tells the rescuer to “kneel by the patient’s side.”” .
CPR initiated prior to the emergency call
There were great variations in proportions of cases were CPR was initiated prior to call in the three study sites (69 (Copenhagen) versus 9 (Stockholm) versus 38 (Oslo)). Reasons for this might be differences in basic life support (BLS) programs or organisational differences, in Copenhagen all calls are handled by the police before transferred to a medical dispatcher, providing more time for qualified bystanders to initiate CPR. Results from this study indicate that dispatchers are less likely to provide CPR instructions to callers when CPR is initiated prior to the emergency call. Early CPR is associated with increased survival [22,23,24] but favourable outcome depends on high quality CPR performance . Studies comparing DA-assisted CPR and bystander initiated CPR prior to the call have found no significant difference in survival between the two groups [13, 26]. However, Takei et al. showed significantly more good quality CPR compared to low quality CPR (OR 2.67) in bystander-initiated CPR prior to the call to the EMDC . It is fair to assume that bystanders who start CPR without instructions are at least willing to perform CPR. Their skills and abilities are uncertain at the time of the emergency call and might be clarified by the medical dispatchers who in addition can support the bystander in performing high quality CPR. Lack of CPR instructions to bystanders performing CPR prior to call indicates a knowledge gap. There is a need for further exploration of the consequences when dispatchers do not provide CPR instructions to a large group of bystanders in a group mostly excluded from studies on DA-CPR.
A high proportion of health care personnel as callers might explain why dispatchers to a lesser extent provide CPR instructions in cases where CPR is initiated prior to the call. Health care professional bystanders are more likely to initiate CPR prior to the call . CPR performed by health care professionals has been shown to have increased patient survival compared to bystander initiated CPR performed by laypeople [28, 29], but a recent study showed no such difference . It can be difficult for dispatchers to assess callers’ competence in CPR based on the fact that they are health care providers. For example, a common situation in OHCA is that the patient is discovered by representatives from the home care services, and the caller is perceived by the dispatcher to be a health care provider. The home care services can be staffed by people with no or limited medical training  and potentially no experience in handling a cardiac arrest patient, hence in as much need of CPR instructions as lay people.
When CPR is initiated prior to the call to the EMDC, cases are recommended to be excluded from review in studies reporting OHCA recognition rates, [13, 17]. These cases are rarely described, even though this seems necessary in order to give proper recommendations for dispatchers on handling these cases. In the future it is reasonable to include more information on the interaction in the first resuscitation team in BLS-programs. Likewise, terminology should be standardised between national CPR councils providing course curriculums and dispatch organisations decision and prioritisation tools, also in cases where CPR is initiated prior to emergency calls. To avoid unnecessary complications and ensure optimal treatment of all OHCA patients, we suggest that dispatchers always provide CPR instructions (also to health care personnel and when CPR is initiated prior to call), make quality assessments during the call and stay on the line until EMS arrival.
This study has several limitations. There are few cases where CPR is initiated prior to the call, and the numbers vary between the three study sites. More data would have made comparisons possible. These data should therefore be regarded as explorative findings providing hypothesis generating knowledge on a group rarely described in other studies. A low number of cases from each site and potential differences in BLS-programs may interfere with the analysis. Inclusion and outcome data is affected by dispatcher performance between sites, and differences in system organisations make comparisons between sites difficult. For example OHCA incidence (per 100,000 inhabitants) in the registries differ from 45 (Stockholm), 61 (Oslo) and 85 (Copenhagen), and CPR by EMT is lower in Oslo than Stockholm and Copenhagen. There is no reason to believe that there are great variations in morbidity between countries, and explanations to these variations might be found in inclusion criteria and reporting rate to the cardiac arrest registries. Time differences between sites are also likely to differ due to differences in call handling. In Denmark the emergency call is initially answered by the police who then refers the call to the EMDC, time variable is measured from EMDC taking the call. In Stockholm the EMDC answer all emergency calls, and 70% of calls are not medical emergencies. In Oslo there is a specific telephone number for medical emergencies only.
High recognition rate may be associated with a higher false positive rate. This is important because of scarce resources which influences management decisions. We have not been able to establish a false positive rate in this study. When reviewing audio files there will always be some interpretation by the reviewer. We tried to minimize this issue by a detailed data extraction protocol. Despite meticulous preparations, there are cases not possible to determine accurately from audio files, resulting in more unknown cases than expected (Table 3). Reasons for this were most often if callers were excessively distraught, (in quite a few cases the caller left the phone) or there were language barriers. Reviewers from respective countries did not have access to other countries’ audio files, and no interrater agreement across countries could be performed.