The main finding of the epidemiological part of the study was the wide spread between the different EMCCs, regarding both overall contact rates, specific acute rates and use of different Index criteria when assessing the situations. The questionnaire study showed a relatively high overall use of Index reported by the operators. It also showed large unexplained variations in the use of Start page and several symptom cards.
Strengths and weaknesses of the studies
The epidemiology study brings new knowledge about contacts made to 113, while the self-reported use of Index represents the first step towards determining use of Index. Both studies include all 19 EMCCs and hence are representative on a national level.
The short study period of 72 h limits the validity of extrapolation, and was mainly due to capacity issues in both ends, as data had to be exchanged in paper format at the time being. The questionnaire was not validated, as we wanted the main study population as large as possible, in terms of both individuals and centers.
Self-reporting allows for over- and underestimation. This was a calculated risk as we aimed to document how the operators themselves thought they used Index, and the results must be interpreted as subjective. We will address the problem with subjective contra objective use of Index in future audio-log study. The response rate of 63.4% could represent a selection bias, but the large range in ’use of Index’ score, both on an individual level and on EMCC level, together with all EMCCs being represented, indicates that we have a representative material. The response rate can to some extent be explained by all information, distribution and reminders depending on the management at each EMCC. The official policy by the EMCC managements on always using Index, combined with the questionnaires and reminders being provided by the management, raises the possibilities of an “eagerness to please” bias. This was attempted opposed by ensuring anonymity and allowing the filled out questionnaires to return directly to the main investigator, through prepaid preaddressed envelopes following each single questionnaire. Recall biases are likely to occur when trying to recall what one usually does in certain situations. The use of a non-validated questionnaire led to misinterpretation of some questions. Never and always are narrow categories, and the results on use of Index might have been more correct if the categories had been more evenly distributed.
Epidemiology
There is a wide spread in annual contact rates among the EMCCs, from 34 to 119/1 000 inhabitants. We have no national data on acute disease distribution to compare with, but one would not expect a similar spread in distribution. Accidents are shown to disperse a geographical pattern, a study from 2013 comparing a rural county with an urban/rural county found a higher accident rate in urban areas, but a higher mortality rate per accident in the rural areas [
16]. This dispersion of accidents does not account for the wide spread in contact rates, and this could indicate that the population uses 113 differently depending on location. A previous study of geographic variations in alerting, dispatch and response found that while severity of illness/trauma had no effect on the use of 113, the use was lower in rural areas compared with urban areas [
17]. This could be due to local differences in organisation of LEMCs and casualty clinics, and cultural differences for when to access the different levels of emergency health services.
When separating contacts assessed to be acute, we also find a large spread in rates between the different EMCCs. The overall acute contact rate of 21 is slightly lower than reported by a previous study based on data from three EMCCs. They found a three month acute rate of 6.2/1 000 inhabitants in 2007, giving an annual rate of 25/1 000 [
18]. As our study shows that the EMCCs differ with regards to both urgency distributions and contact rates, differences in inclusion of EMCCs will affect these outcomes.
A recently published study from Denmark studied Danish Index’ ability to triage patients according to severity [
19], following the implementation of the CBD guidelines in Denmark. They found a national acute (emergency level A) rate of 17 ambulance dispatches/1 000 inhabitants per year, which is somewhat lower than our 21 acute contacts/1 000 inhabitants per year. There were also differences between their included areas: with 13, 17 and 21 acute ambulance dispatches/1 000 per year. One possible explanation could be the large differences in EMCC population size, with five centres in Denmark and 19 in Norway covering 5.5 and 5 million inhabitants respectively. This difference could influence use of Index, and hence incidence of acute criteria codes. Cultural differences in what emergency medical level to access and differences in pre-hospital emergency medicine organization could influence this.
On primary health care level, a study of 85 000 contacts from 2007 investigated the distribution of urgency levels in the Norwegian emergency primary healthcare services: acute 2.3%, urgent 21.1% and non-urgent 76.6% [
20]. Compared to our findings (acute 37%, urgent 35%, non-urgent 26%), these differences show that the population as a whole know what level to address, depending on degree of medical emergency.
The most frequently used Index criteria, all urgencies included, was “6 - Unresolved issue”. This criterion covers a whole range of situations; from the unclear situations where the operator gets too little information to choose another criterion to well-defined situations where no other criteria match. The large differences between the EMCCs, both in terms of total use of this criterion and the variation in urgencies as it was used in, clearly indicates a variation in use of Index from center to center.
Use of Index
The operators reported a relatively high overall use of Index, corresponding to “in over 75% of real emergency calls”. The variation among the operators was quite large though. Although we found some factors associated with positive and negative effects on use of Index, we were only able to explain 23.4% of this variation, indicating that there are factors influencing use of Index that we were not able to uncover. The main reason given for not using Index was AMIS, and the confidence that this software program provided the necessary key words to properly assess the acute medical emergency situation. The operators’ background was to great extent as expected. It was an experienced group of employees, with median working time at the EMCC of 6 years. The results indicated decent routines for training new operators in use of Index, but not for repetition.
Our result of approximately 75% use of Index lies between the divergent findings from the 2005 study, with 99% self-reported Index guideline adherence and 64% adherence based on log-recordings [
11]. To our knowledge there are no other studies addressing adherence to criteria based guidelines. As algorithm based dispatch systems require a much stricter adherence to protocol they have systems developed to monitor and increase protocol compliance, among other quality markers [
6,
8,
21‐
24]. Although compliance is highlighted as such an important feature of MPDS, less than 3% of the registered users are Accredited Centers of Excellence [
25], which among other quality measurements include a minimum of 90–95% compliance with different parts of the protocol [
8].
Given the purpose of CBD guidelines, to assist health care personnel in decision making rather than defining specific questions or actions to be taken [
3], use of Index was not expected to reach full score. Looking at the different explanatory variables though, it was rather unexpected that time working at an EMCC had so little effect on use of Index. It was equally unexpected that the effect of what EMCC the operator works at was so negligible, as we held this to be the natural explanation for the variations in use of Index.
Working in rotation with ground ambulance is the variable strongest associated with decreased use of Index. This is not associated with educational level, as half the operators rotating with ambulance are registered nurses and there is no difference in use of Index between the different educational levels among those rotating with ambulance. Previous research has found that individual experience and professional autonomy affects guideline adherence [
10]. Our study finds no effect on use of Index for operators rotating with ER. A Dutch study on protocol adherence among emergency and ambulance nurses found the opposite result; ambulance nurses are more likely to hold protocol over experience while emergency nurses are more likely to hold experience over protocol [
26].
The low use of Start page is reflected in the reasons given for not using it. They know it, and find the key words they need in AMIS screen. The Start page (3
rd edition) includes the phrase: “Is the patient awake and can talk?” [
2]. The low percentage reporting to actually ask about this latter, confirms that neither AMIS nor memory equals Start page, and one may speculate that the reason for including this question is not fully understood among those using the tool.
The main reason for not using Index appears to be that the operators choose to use the AMIS screen above Index, and a belief that AMIS provides the necessary keys to assist them in properly assessing the situation. This might be of some concern, since AMIS does not supply any support in neither decision making nor advices for the public or health personnel at scene, and hence cannot replace Index. Furthermore, discarding the guidelines in favor of memory and own experience is a potential hazard of losing vital information or getting sidetracked due to unstructured interrogation. This risk is strongly advocated by critics of criteria based dispatch [
9,
22].
Based on our findings we would recommend increased focus on use of Index at each EMCC. The systematic review on guideline and protocol adherence in the pre-hospital and emergency care setting [
10] finds that tailored strategies towards identified barriers improve professional practice, as well as strategies aimed at influencing factors improve guideline adherence. Evidence based recommendations and a relationship between guideline adherence and clinical outcome are also mentioned as important motivational factors for guideline adherence.
In this study we estimated a mean value for use of Index among operators in Norway, based on self-report, which obviously represents a subjective view. Investigating the real practice based on objective data is therefore a natural next step in documenting use of Index.