Prehospital Prediction of a Time-sensitive Condition Among Patients With Dizziness Assessed by the Emergency Medical Service

Background: Dizziness is a relatively common symptom among patients who call for the emergency medical service (EMS). Methods: All patients assessed by the EMS and triaged using the rapid emergency triage and treatment system for adults code 11 (=dizziness) in the 660,000 inhabitants in the Municipality of Gothenburg, Sweden, in 2016, were considered for inclusion. The patients were divided into two groups according to the nal diagnosis (a time-sensitive condition, yes or no). Results: There were 1,536 patients who fullled the inclusion criteria, of which 96 (6.2%) had a time-sensitive condition. The majority of these had an acute cerebrovascular disease. Eight predictors of a time-sensitive condition were identied. Three were associated with a reduced risk: 1) the dizziness was of a rotatory type, 2) the dizziness had a sudden onset and 3) increasing body temperature. Five were associated with an increased risk: 1) sudden onset of headache, 2) a history of head trauma, 3) symptoms of nausea or vomiting, 4) on treatment with anticoagulants and 5) increasing systolic blood pressure. Conclusion: Among 1,536 patients who were assessed by the EMS due to dizziness, 6.2% had a time-sensitive condition. On the arrival of the EMS, eight factors were associated with the risk of having a time-sensitive condition. They were linked to the type of symptoms, to clinical ndings on the arrival of the EMS and to the recent clinical history.


Background
Dizziness is a relatively common symptom among patients who seek emergency care. Many of these patients dial 112 for ambulance transport to hospital.
Dizziness is a unifying concept for a number of different experiences for the patient, such as being on a carousel, off balance, near syncope or motion of the sea. It has previously been reported that about three per cent of patients who visit the emergency department do so because of dizziness [1].
A number of conditions can be associated with the symptom of dizziness. Damage to central or peripheral parts of the vestibular nerve will generate an acute vestibular syndrome, which consists of several symptoms such as dizziness, nausea/vomiting, nystagmus and trouble maintaining body balance.
When the damage is localised to the inner ear or in the vestibular nerve, there is a peripheral aetiology. Examples of a peripheral aetiology of an acute vestibular syndrome are benign paroxysmal positional vestibular neuritis, Meniere's disease, bacterial labyrinthitis and herpes zoster oticus [2].
Dizziness of central origin damage is localised in the central parts of the vestibular system in the brain stem and/or cerebellum and the underlying aetiologies include TIA/stroke, migraine, tumour in the brain stem, encephalitis and multiple sclerosis.
However, it has been suggested that the majority of patients with acute dizziness have aetiologies other than damage to the peripheral or central parts of the vestibular system. In one large study, it was reported that 63% of all cases with acute dizziness had other aetiologies. The most common aetiology was an upper airway infection (35%), followed in order of frequency by hypertension (18%). Time-sensitive conditions such as bradycardia, AV block III, sepsis and acute coronary syndrome accounted for three per cent of all cases [3]. Similar ndings have been made by others reporting that more than half the patients with dizziness have an aetiology which is not related to the vestibular system, with the majority having an internal medicine disorder [4].
Time-sensitive conditions which are not related to the vestibular system but may still cause dizziness include cerebrovascular disease not speci cally affecting the vestibular system, water-electrolyte imbalance, arrhythmia, carbon monoxide poisoning and aortic dissection.
The variety of conditions that may exist behind symptoms of dizziness highlight the di culties healthcare providers experience when attempting to differentiate these symptoms into benign and malignant conditions when they meet patients with these symptoms.
The burden on the emergency medical service (EMS) has increased markedly during the last few decades. This is primarily explained by an elderly population with an often extensive comorbidity and the fact that people nowadays tend to dial 112 more often than before, despite not suffering from a timesensitive condition [5]. It has also been shown that not all these patients require transport to hospital [6] and that some could preferably be handled by a lower level of care [6].
This situation constitutes a demanding challenge for the EMS staff who, for the adequate utilisation of resources, must try to differentiate time-sensitive from non-time-sensitive conditions already at the scene.
A typical symptom where the EMS staff may have problems is dizziness. Although a large proportion of these patients appear to suffer from a relatively benign condition, there are time-sensitive conditions hidden among them which are important to identify.
The aim of this study was, among patients who are assessed by the EMS due to symptoms of dizziness, to attempt to identify clinical factors that may help to differentiate patients who are suffering from a time-sensitive condition from those that are not. We have recently reported on basic ndings from this study cohort [7].

Design
This study is a retrospective, observational study where patients with on-the-scene PEN triage of RETTS code no 11 (dizziness) were included for a manual review.

Recruitment of patients:
All the patients who were seen by the EMS in the Municipality of Gothenburg and given the rapid emergency triage and treatment system for adults (RETTS-A) code 11 = dizziness (n = 2,048) from 1 January to 31 December 2016 were included in the study. The EMS system in Sweden The healthcare provided in Sweden, including pre-hospital care, is tax funded. The EMS organisation uses national/regional guidelines. Each ambulance in Sweden should be staffed by at least one registered nurse (RN), responsible for the care. The RN often has an additional one-year postgraduate education in pre-hospital emergency care (PEN). However, ambulance crew set-ups can take the form of two nurses or one nurse and one emergency medical technician (EMT). The PEN assesses the patient at the scene and has approximately 40 different types of drug at his/her disposal. The PEN is responsible for deciding on the level of care, which means that not all patients are transported to hospital [8].
Ambulances are dispatched from a dispatch centre with three priorities as previously described [9].

Triage system
The RETTS-A [10] is a ve-level triage system currently in use in the majority of emergency departments (EDs) and EMS organisations in Sweden. The RETTS-A is developed, licensed and maintained by a Swedish company (Predicare AB). The RETTS-A contains 58 charts with common ED presentations. Each chart includes both emergency signs and symptoms (ESS) and vital signs (VS) as previously described [7]. The level of severity of both VS and ESS is divided into the colours of red, orange, yellow, green and blue, but blue is not used by the EMS. Triage level red is considered life threatening, resulting in immediate contact with a physician, orange is potentially life threatening, while yellow and green can wait in the ED without medical risk. Yellow is considered to be more urgent than green. The highest triage level of either VS or ESS becomes the nal triage level.

Data collection
Data were collected from digital case records in the EMS data system and the hospital medical records.
Final diagnosis at discharge was collected from the hospital medical records. Patients were then divided into two groups according to whether or not they had a time-sensitive condition. A time-sensitive condition was de ned if the nal diagnosis was any of those described by Wibring et al [11]. Of particular interest for this article was TIA/stroke.

Data analyses
Univariate comparisons between patients with and patients without a time-sensitive condition were performed using the Mann-Whitney U test for continuous/ordered variables and Fisher's exact test for dichotomous variables.
With the exception of the two variables of "non-transported, attended the ED within 72 hours" and "EMS on-the-scene time", which were regarded as outcomes rather than predictors, all variables with a univariate p < 0.20 for differences between the two groups were tested for inclusion in a multiple logistic regression model, using backward stepwise selection with p < 0.01 for staying in the model. This procedure was performed both using only complete cases and, due to the amount of missing data for several of the variables, using multiple imputations (primary analysis). For the latter, which was regarded as the primary analysis, missing data were assumed to be missing at random (MAR) and 50 imputed datasets were generated with the Markov Chain Monte Carlo (MCMC) method using the expectationmaximisation (EM) algorithm. Rubin's rules were used to pool the results from the imputed datasets. To identify independent predictors of time-sensitive conditions in the multiple imputation multivariable analysis, we started with a model including all the variables in Table 1 with a univariate p < 0.20. Collinearity was checked by association measurements between variables, as well as by inspection of the variance in ation factor, condition index and eigenvector proportions in a multiple linear regression model including all the candidate variables. One important collinearity, between systolic and diastolic blood pressure, was found and, as a result, diastolic blood pressure was excluded from the following analysis. Multiple logistic regression was performed in each of the 50 imputed datasets and the variable with the highest p-value in the pooled result was excluded from the model. A new regression analysis was then performed in each imputed dataset and, of the remaining variables, the one with the highest p-value in the pooled result was excluded. This procedure was repeated until all the remaining variables yielded a pvalue below 0.01 in the pooled result.

Results
Overall, there were 59,000 primary missions for the EMS in 2016. Of them, 2,048 (3.5%) were coded as dizziness according to the RETTS-A (code 11).
After applying inclusion and exclusion criteria, 1,536 patients remained. The clinical characteristics of the included patients are shown in Table 1. The overall median age was 73 years and 58% were women.

Univariable analyses
The results of univariable analyses of the association between clinical variables and time-sensitive conditions are shown in Table 1b and 1a.
Patients with a time-sensitive condition were signi cantly older. They also had higher systolic blood pressure, more often had ongoing anticoagulation, more often described a sudden onset of headache but less often described a sudden onset of dizziness and less often had dizziness of the rotatory type, nautical, or balance disturbing. Moreover, they more frequently had a history of head trauma. There were eight factors that were independently associated with the risk of a time-sensitive condition. Two factors, the rotatory type of dizziness and the sudden onset of the symptoms, were both associated with a threefold decrease in the risk of a time-sensitive condition. Furthermore, the risk of a time-sensitive condition was reduced by nearly 50% for each degree(C) of increase in body temperature.

Multivariable analyses (Table 2)
The following ve factors were associated with an increased risk of a time-sensitive condition: 1) sudden onset of headache with a ninefold increase in risk, 2) a history of head trauma with a fourfold increase in risk, 3) symptoms of nausea or vomiting with a twofold increase in risk, 4) on treatment with anticoagulants with a twofold increase in risk and 5) systolic blood pressure with a 1.5% increase in risk per mmHg increase in systolic blood pressure.

Discussion
We found that, among 59,000 primary missions for the EMS in the catchment area, 3.5% were reported as suffering from dizziness. Of them, about six per cent ful lled the criteria for having a time-sensitive condition, among which the majority had had an acute cerebrovascular disease. On the arrival of the EMS, there were eight factors that were associated with the risk of having a time-sensitive condition.
Three factors, i.e. having a rotatory type of dizziness, having a sudden onset of symptoms and increasing body temperature, were all associated with a decreased risk.
Five factors, i.e. sudden onset of headache, a history of head trauma, symptoms of nausea or vomiting, on treatment with anticoagulants and increasing blood pressure, were all associated with an increased risk.
Our nding that around three per cent of patients who were seen by the EMS had symptoms of dizziness is in agreement with previous research which states that about three per cent of patients who visit the emergency department have symptoms of dizziness [1]. Furthermore, Hjälte et al. found that three per cent of patients who called for the EMS did so because of dizziness [12].
The observation that around six per cent of the patients with dizziness had a time-sensitive condition is also within the range that has previously been reported [13][14][15][16].
The observation that rotatory dizziness is associated with a reduced risk of a time-sensitive condition is supported to some extent by previous research [14,17].
Moreover, the observation that the sudden onset of dizziness is associated with a reduced risk of a timesensitive condition has been reported [18]. However, the proportion of patients with a time-sensitive condition among those with a sudden onset of dizziness has been reported with a frequency varying from 0.7-11% [15,[19][20][21]. It is possible to speculate that there is di culty deciding how to describe the type of onset of dizziness in some cases.
The nding that the risk of a time-sensitive condition decreased with increasing body temperature is di cult to explain. It is possible to speculate that, among patients with fever and vertigo, the risk of an underlying infection (not de ned as a time-sensitive condition) is more marked.
Sudden onset of headache was strongly associated with an increased risk of a time-sensitive condition. Similar ndings were made by Kerber et al. [18]. Others [15,[22][23] did not report results that supported this statement.
Another factor that increased the risk of a time-sensitive condition was a history of head trauma. To the best of our knowledge, this has not been reported before. The mechanism behind the association between head trauma and a time-sensitive condition among patients with dizziness can only be speculated upon.
A third factor was the presence of nausea or vomiting. The fact that these symptoms are associated with an increased risk of a time-sensitive condition has previously been suggested [18,24]. There is no clear explanation of why nausea or vomiting should be a risk factor for a more alarming aetiology. Among patients with other symptoms such as chest pain, the presence of nausea or vomiting has been associated with an increased risk of an underlying acute coronary syndrome [24].
A fourth factor that increased the risk of a time-sensitive condition was whether the patient was on chronic treatment with anticoagulants. Although this was not reported by other researchers [15,23] there is a potential explanation for this nding. This may indicate that the patient has previously suffered from a thromboembolic event or suffers from a disease that is associated with an increased risk of such an event, for example, atrial brillation. Somewhat surprisingly, a history of atrial brillation did not appear as a risk factor for a time-sensitive condition in our survey.
The last risk factor was increasing systolic blood pressure on the arrival of the EMS. This nding is in agreement with a number of previous studies [15,16,18,23]. The nding that an elevation of blood pressure is a risk factor for the development of a cerebrovascular disease is well documented [13,[25][26][27][28].

Strengths And Limitations
This cohort of patients who were assessed by the EMS within the catchment area is large and representative. Since the data are based on a retrospective observational study, the results and the conclusion are dependent on the quality of the reporting. This quality most likely varies and is dependent on situational factors, as well as the experience and skills of the EMS staff. Information is missing for a large number of variables and this was adjusted for by multiple imputations.
Although the data are representative of the catchment area, they are collected from an EMS system within an urban area. For this reason, our results cannot be extrapolated to a national perspective where rural areas must be included as well.

Conclusion
Among 1,536 patients who were assessed by EMS due to dizziness, 6.2% had a time-sensitive aetiology.
On the arrival of the EMS, eight factors were associated with the risk of having a time-sensitive aetiology.
They were all linked to the type of symptoms or to clinical ndings on the arrival of the EMS but also to clinical history. Further studies should aim to develop a risk-stratifying instrument in the prehospital setting and the validation of such an instrument.

Declarations
Ethics approval and consent to participate This study was approved by the Regional Ethical Review Authority in Gothenburg, approval no. 970-15.
For the retrospective analysis of this register study, informed consent was waived. However, at the time of EMS assessment, patients who asked for their data to remain con dential were not included in the retrospective analysis.

Consent for publication
Not applicable Availability of data and materials I con rm that data and material are available if necessary.

Competing interest
The authors have no commercial associations or sources of support that might pose a con ict of interest.

Funding
The study was supported by the Swedish state under the ALF agreement (ALFGBG-922511). The funder costs for the statistical analysis of the data.
Authors contribution CM was responsible for the planning process, analysed and interpreted the data and was responsible for the writing process.
JG took part in the planning process, took part in the analysis and interpretation of the data and was active in the writing process.
EL took part in the planning process, took part in the analysis and the interpretation of the data and was active in the writing process.
JS took part in the planning process, took part in the analysis and the interpretation of the data and was active in the writing process.
RW took part in the planning process, took part in the analysis and the interpretation of the data and was active in the writing process.
CA took part in the planning process, took part in the analysis and the interpretation of the data and was active in the writing process.