1,2Stjepan Petričević
1,2Igor Pelaić
2 Božidar Veljković
Stjepan Petričević
Emergency Medical Institute of Zagreb County E-mail: stjepan.petricevic@hitna-zgz.hr
Results. A statistically significant association was found between the triage category “Altered con- sciousness/paralysis” and confirmed stroke diagnosis (χ² = 11.82; p < 0.001). Many stroke patients were initially categorized under non-specific symptoms. No significant sex difference was observed in triage allocation (p = 0.9508), while women were signifi- cantly older than men (p < 0.001).
The Medical Dispatch Unit (MPDJ) plays a critical role in the prehospital emergency care system, particu- larly in the recognition and triage of stroke cases. Stroke is a time-sensitive emergency in which every minute from the onset of symptoms to medical in- tervention can significantly impact patient outcomes
(1). Dispatchers in the MPDJ serve as the first point of contact with emergency medical services, and their ability to recognize symptoms and make accurate triage decisions determines the speed of medical response and the eventual clinical outcome (2). Rec- ognition of stroke during an emergency call is based on the assessment of symptoms described by the caller. The dispatcher must then quickly determine the urgency level and appropriate type of response
(1). The effectiveness of this process can greatly influence the availability of reperfusion therapies, such as thrombolysis and mechanical thrombectomy, which are time-sensitive and require immediate in- tervention (3).
Furthermore, research indicates that public aware- ness of stroke symptoms significantly affects the timeliness of seeking medical help. Studies conduct- ed in the United States and Canada have shown that public education campaigns can increase the pro- portion of individuals who recognize symptoms and promptly contact emergency services (4,5). Although early recognition of symptoms is considered crucial for timely hospital arrival, numerous challenges exist in practice regarding the identification of stroke signs among the general public. A British study showed that some citizens misinterpret or fail to recognize basic symptoms despite awareness campaigns such as the FAST test (6). A systematic literature review also highlights significant disparities in stroke knowl- edge among the general population, which affects response times and emergency call activation (7).
Therefore, the aim of this study was to examine the association between selected dispatch triage criteria and final stroke diagnoses in the Medical Dispatch Unit, to describe the distribution of stroke-related triage codes with respect to age, sex and caller type, and to identify potential areas for improving stroke recognition during emergency calls.
The Medical Dispatch Unit uses structured triage sys- tems to ensure standardized and efficient assessment of call urgency. Since 2011, Croatia has implemented the Croatian Emergency Call Reception Index, devel- oped based on the model of the Norwegian Index for Emergency Medical Assistance (8). This index defines the criteria and categories for emergency medical calls and ensures a uniform classification methodology across all counties (8,9). In contrast to strictly algorith- mic systems such as the Medical Priority Dispatch Sys- tem (MPDS), the criteria-based approach used in the Croatian Index allows dispatchers greater flexibility in decision-making based on symptoms reported by the caller (10). Studies conducted in Norway show that criteria-based triage systems can predict the need for emergency medical interventions with high specific- ity, though sometimes with limited sensitivity (11).
Various standardized protocols are used to recognize stroke during emergency medical calls. The most commonly used include the Advanced Medical Prior- ity Dispatch System (AMPDS) and the Medical Prior- ity Dispatch System (MPDS), while in Europe, national indexes such as the Danish Index and the Norwegian Index are also applied (3). Additionally, targeted stroke assessment tools such as the FAST test (Face-Arm- Speech-Time) and the Cincinnati Prehospital Stroke Scale enable rapid evaluation of neurological deficits (3,4). It is important to distinguish the criteria-based approach, such as the Croatian Emergency Call Recep- tion Index, from strictly protocol-driven systems like MPDS or AMPDS. While the criteria-based approach allows dispatchers more discretion in interpreting and making decisions based on the caller’s description of symptoms, protocol-driven systems are based on pre- defined algorithms that must be followed consistently without deviation. This flexibility in criteria-based sys- tems introduces greater subjectivity and variability in triage decisions, which may lead to under-recognition of certain conditions, including stroke. Research shows that using structured protocols can improve stroke rec- ognition rates (12). For example, the Madrid-Direct pro- tocol, specifically developed for stroke, demonstrated improved outcomes in early identification and routing of patients to appropriate healthcare facilities (12). However, these protocols are not flawless - their sen-
sitivity ranges from 41% to 83%, while positive predic- tive values vary between 24% and 88% (1,12).
Dispatchers face numerous challenges in their work, including limited information available during calls, var- iability in symptom presentation, and caller uncertain- ty. They often have very limited time for assessment, and uncertainty in symptom recognition may result in incorrect triage decisions (2). Moreover, research has shown that dispatchers without medical training have a lower stroke recognition rate compared to those with a healthcare background (10). Although classic stroke symptoms are well known (e.g., unilateral weakness, speech disturbances), it is important to emphasize that stroke may also present with less specific symptoms such as dizziness, nausea, confusion, or general weak- ness. These atypical symptoms pose a particular chal- lenge for dispatchers, especially when callers are not adequately informed or are unaware of the importance of accurately describing symptoms during the call. In practice, this highlights the necessity of educating not only dispatchers but also the general public, in order to reduce missed stroke cases during the early, critical phase of emergency calls. An additional problem is the lack of feedback to dispatchers regarding patient out- comes. Studies have shown that providing such feed- back can significantly improve triage accuracy (13).
An analysis of previous studies indicates that the sensitivity of stroke recognition among dispatch- ers is relatively low and ranges from 18% to 83%
(12). For example, a study conducted in Copenhagen found that only 66% of strokes were recognized dur- ing emergency calls, with a low positive predictive value (~30%) (14). Dispatcher education has proven to be one of the key factors in improving stroke rec- ognition. A study conducted in the United Kingdom showed that specialized training increased the stroke recognition rate from 63% to 80% (15).
Stroke triage within the Medical Dispatch Unit repre- sents a crucial step in the prehospital management of patients, and its effectiveness directly influences patient outcomes. Standardized protocols, dispatcher education, and public awareness campaigns about stroke are essential for improving symptom recogni- tion and optimizing the allocation of emergency medi- cal service resources. Nonetheless, further research is
needed to develop more accurate triage tools and en- hance the decision-making processes of dispatchers.
This study aimed to assess current stroke triage practices in the Medical Dispatch Unit (MPDJ) using the Croatian Emergency Call Reception Index, in or- der to identify areas for improvement in dispatcher training and caller awareness.
This study was conducted as a retrospective analysis of data collected via the e-Hitna system for the period from January 1 to December 31, 2023. This one-year period was chosen to enable a comprehensive analysis that accounts for potential seasonal variations in the incidence of emergency medical conditions, including fluctuations in the occurrence of stroke throughout the year. It is important to note that, as of March 27, 2024, a revised version of the Croatian Emergency Medical Call Reception Index has been implemented, with re- structured triage criteria that may affect data compa- rability in future analyses. In the revised Index, several criteria within chapters A.18 and A.25 were reworded, regrouped and assigned different urgency levels, which limits direct comparability of triage patterns before and after this change. For this reason, the year 2023 was selected for analysis to ensure consistency within a sin- gle triage framework, enabling uniform interpretation of the data. The study analyzed data from the Medical Dis- patch Unit (MPDJ) of the Emergency Medical Institute of Zagreb County, aiming to examine the association be- tween selected triage criteria and final diagnoses.
The data used in this research were extracted from the e-Hitna system and include all emergency calls received by the MPDJ during the observation period. General call statistics were analyzed (number, type, urgency level), along with assigned triage chapters
according to the Croatian Emergency Medical Call Re- ception Index. Particular attention was given to data within the following two triage categories:
Chapter 18: Headache (triage criteria A.18.01 – A.18.07)
Chapter 25: Altered consciousness/paralysis (triage criteria A.25.01 – A.25.09)
The analysis included all criteria within these prede- fined ranges (A.18.01–A.18.07 and A.25.01–A.25.09),
as they represent the core high-priority triage codes routinely used in MPDJ practice for the recognition of potentially time-critical neurological presentations. Criteria outside these ranges (A.18.08–A.18.10 and A.25.10–A.25.11) were not included, as they corre- spond to infectious, abdominal, metabolic, traumatic or other non-neurological presentations and there- fore fall outside the analytical scope of this study. A detailed overview of all included and excluded cri- teria, together with justifications, is provided in Sup- plementary Table S1.
For each observed triage criterion, the primary pa- tient diagnoses were analyzed in detail according to the International Classification of Diseases (ICD-10) for each call and correlated with the selected triage criteria. Cumulative data were also reviewed regard- ing the type of caller (family member, healthcare pro- fessional, bystander) in relation to the selected tri- age criteria, with the aim of evaluating the potential influence on triage card selection. As caller type data are not available at the individual ICD diagnosis level, this analysis was limited to the triage category level.
Subsequently, the patients’ primary diagnoses were analyzed based on the following ICD-10 codes:
G45–G45.9 – Transient ischemic attacks
I60–I69.9 – Cerebrovascular diseases (stroke, intracerebral and subarachnoid hemorrhage, stroke sequelae)
Patient diagnoses were reviewed using medical re- cords, and the triage criteria selected at the time of the emergency call, together with the intended purpose of the dispatch, were subsequently exam- ined. Although patients typically do not call for them- selves, and symptoms are interpreted by the caller, an analysis of age and gender differences among patients with stroke was conducted to identify po- tential variations in triage patterns. The focus of the research remains on the analysis of reported symp- toms and their influence on triage card selection.
The research was conducted in accordance with ap- plicable ethical principles and data protection regula- tions. The data used in the analysis were extracted from the e-Hitna system without access to any per- sonally identifiable patient information. All records were anonymized prior to analysis and processed ex- clusively in aggregate form.
According to Croatian national regulations and insti- tutional policy, retrospective analyses of fully an- onymised operational EMS data that do not involve direct contact with patients or any influence on pa- tient care are exempt from the requirement for for- mal ethics committee approval. The dataset used in this study meets all criteria for this exemption.
Written permission for the use of e-Hitna system data for scientific purposes was obtained from the Director of the Emergency Medical Institute of Za- greb County. The study was conducted in compliance with GDPR and the principles of the Declaration of Helsinki.
Statistical analysis was performed using descriptive statistical methods to present the distribution of tri- age categories, the frequency of specific symptoms, and the age distribution of patients. Differences between categorical variables were analyzed using the chi-square test (χ²), while differences in means between two independent groups were assessed using Welch’s t-test due to potential inequality of variances between groups. For contingency tables with small expected cell counts, Fisher’s exact test was additionally calculated to verify the chi-square results. The level of statistical significance was set at p < 0.05. Descriptive analyses were conducted using Microsoft Excel (with the Analysis ToolPak add-in), while inferential statistical tests (χ² test and Welch’s t-test) were calculated using the online tool Social Science Statistics (https://www.socscistatistics.com/ tests/).
Table 2. Distribution of urgency levels for intervention calls in 2023 | |||
Priority | Category (Color) | Number of Calls | Percentage (%) |
Priority I | Red | 8,471 | 31.46% |
Priority II | Yellow | 16,105 | 59.81% |
Priority III | Green | 2,351 | 8.73% |
Total | — | 26,927 | 100.00% |
During the observed period from January 1 to De- cember 31, 2023, the Medical Dispatch Unit (MPDJ) of the Emergency Medical Institute of Zagreb County received a total of 60,120 calls. These calls were cat- egorized into four main groups: interventions, con- sultations, other calls, and nuisance calls (as shown in Table 1).
Table 1. Categories of calls received by MPDJ ZZHM ZŽ in 2023 | ||
Type of Call | Number of Calls | Percentage (%) |
Interventions | 26 927 | 44.79% |
Consultations | 12 199 | 20.29% |
Other | 20 767 | 34.54% |
Nuisance | 227 | 0.38% |
Total | 60 120 | 100.00% |
The majority of calls resulted in emergency medical interventions (n = 26,927; 44.79%), while consul- tation calls totaled 12,199 (20.29%). Other calls— including administrative and non-urgent medical inquiries—accounted for 34.54% (n = 20,767). Nui- sance calls, such as prank or inappropriate calls, made up 0.38% of the total (n = 227).
In 2023, out of the 26,927 calls that led to medi- cal interventions, the highest proportion was clas- sified as Priority II (yellow category), accounting for 59.81% of all interventions (n = 16,105). Priority I (red category), which denotes life-threatening condi- tions, was assigned in 31.46% of cases (n = 8,471). Priority III (green category), which includes less ur- gent conditions, comprised 8.73% of all interventions (n = 2,351), as shown in Table 2.
Most calls were assigned Priority II (yellow category), indicating a high urgency level, though not immedi- ately life-threatening. At the same time, the signifi- cant share of Priority I (red category) calls suggests that over one-third of patients were assessed as being in a life-threatening condition already during emergency call triage.
The analysis of intervention call distribution according to selected triage chapters from the Croatian Emergen- cy Call Reception Index reveals significant differences in the frequency of individual categories, as shown in Table 3. The highest number of intervention calls was recorded under the category “Unclear Problem” (n = 6,586; 24.47%), which includes calls where symptoms are not clearly defined but still require medical assess- ment. This was followed by “Respiratory Disorders” (n = 3,995; 14.83%) and “Chest Pain / Heart Disease” (n = 2,508; 9.31%), highlighting a high proportion of patients presenting with potential cardiovascular and respiratory emergencies. Within the scope of particular interest for this study, the category “Altered Consciousness / Paraly- sis” included 1,974 calls, while “Headache” was the lead- ing symptom in 278 cases. These categories are crucial for analyzing the association between triage criteria and final diagnoses of cerebrovascular incidents.
Table 3. Distribution of intervention calls by triage chapter in 2023 | ||
Triage Chapter | Number of Calls | Percentage (%) |
Unclear Problem | 6,586 | 24.47% |
Respiratory Disorders | 3,995 | 14.83% |
Chest Pain / Heart Disease | 2,508 | 9.31% |
Abdominal / Back Pain | 2,330 | 8.65% |
Psychiatry / Suicide | 1,294 | 4.81% |
Altered Consciousness / Paralysis | 1,974 | 7.33% |
Wounds / Fractures / Minor Injuries | 1,521 | 5.65% |
Convulsions | 451 | 1.67% |
Headache | 278 | 1.03% |
Other Chapters | 5,990 | 22.24% |
Total | 26,927 | 100.00% |
The results highlight the need to improve stroke symptom triage. The relatively low proportion of calls categorized under “Altered Consciousness/Paralysis” and “Headache” in comparison to the actual inci- dence of cerebrovascular events indicates potential shortcomings in recognizing key stroke symptoms during the emergency call phase.
The association between triage criteria and primary stroke diagnoses was analyzed across a total of 278 calls categorized under the triage chapter “Head- ache” (A.18.01 – A.18.07) and 1,974 calls within the chapter “Altered Consciousness/Paralysis” (A.25.01 – A.25.09). From the total number of calls within these triage chapters, only those meeting the stroke-rele- vant criteria listed in Supplementary Table S1 were included in the analysis, while calls assigned to non- specific or non-neurological criteria within A.18 and
A.25 were excluded.
Within the “Headache” chapter (Table 4), 26 out of 278 calls met specific criteria (A.18.01 – A.18.07). Among these, a stroke diagnosis was confirmed in 2 cases (7.69%), while in the remaining 24 cases (92.31%) other diagnoses were established.
In the “Altered Consciousness/Paralysis” chapter (Table 4), 1,154 of the 1,974 received calls involved specific criteria (A.25.01 – A.25.09). Of these, 1,120 calls with assigned final diagnoses coded accord- ing to ICD-10 were analyzed. Stroke (ICD-10 codes: G45–G45.9, I60–I69.9) was confirmed in 461 cases (41.16%), while other diagnoses were established in 659 cases (58.84%). The difference of 34 calls (be-
tween the number of observed criteria and analyzed diagnoses) arose from cases for which no ICD-10 di- agnosis was subsequently recorded.
The higher prevalence of primary stroke diagnoses within the “Altered Consciousness/Paralysis” chapter compared to the “Headache” chapter suggests that headache alone is not a sufficiently specific predictor of stroke. The relatively low proportion of confirmed strokes in the “Headache” category (7.69%) com- pared to “Altered Consciousness/Paralysis” (41.16%) indicates the potential overestimation of headache severity as a primary indicator of stroke during emer- gency medical calls.
A Chi-square test, as shown in Table 5 and Table 6, was conducted to examine the association between the ini- tially selected triage chapters (“Headache” and “Altered Consciousness/Paralysis”) and the primary diagnosis of stroke. The results indicated a statistically significant difference between the two categories (χ² = 11.82; p < 0.001), confirming that patients triaged under “Altered Consciousness/Paralysis” were significantly more likely to receive a primary diagnosis of stroke compared to those initially categorized under “Headache”.
Yates’ correction, which reduces the likelihood of overestimating values in tests with small sample siz- es, also showed a statistically significant difference (χ² = 10.47; p = 0.001).
The association between triage category and final stroke/TIA diagnosis remained statistically signifi- cant when tested with Fisher’s exact test (p < 0.001).
Table 4. Distribution of stroke diagnoses according to triage criteria | |||||
Triage Criteria | Calls Meeting Triage Criteria | Transient Ischemic Attacks (G45–G45.9) | Cerebrovascular Diseases (I60– I69.9) | Other ICD-10 Diagnoses | Total ICD-10 Diagnoses Analyzed* |
Headache (A.18.01 – A.18.07) | 26 | 0 | 2 | 24 | 26 |
Altered Consciousness/ Paralysis (A.25.01 – A.25.09) | 1,154 | 101 | 360 | 659 | 1,120 |
*The difference between the total number of calls meeting triage criteria (1,154) and those with analyzed ICD-10 diagnoses (1,120) is due to missing diagnoses in some cases. | |||||
Table 5. Distribution of stroke diagnoses by triage criteria | |||
Triage Criteria | Stroke (G45–G45.9, I60–I69.9) | Other ICD-10 Diagnoses | Total |
Headache (A.18.01 – A.18.07) | 2 | 24 | 26 |
Altered Consciousness/Paralysis (A.25.01 – A.25.09) | 461 | 659 | 1,120 |
Total | 463 | 683 | 1,146 |
Taken together, these results confirm a strong asso- ciation between triage chapter and stroke diagnosis, highlighting challenges in recognizing stroke in pa- tients reporting headache as their primary symptom during emergency medical calls.
Table 6. Chi-square test results for association between triage criteria and stroke diagnosis | ||
Statistical Test | Value | p-value |
Chi-square (χ²) | 11.82 | < 0.001 |
Chi-square with Yates’ correction | 10.47 | 0.001 |
The primary diagnosis of stroke was confirmed con- siderably more often in patients triaged under the “Altered Consciousness/Paralysis” chapter, while it appeared significantly less frequently in the “Head- ache” category.
These results indicate that headache as a primary symptom of stroke is relatively rarely confirmed as an accurate indicator, whereas “Altered Consciousness/ Paralysis” proves to be a much more reliable predic- tor. This suggests that, in cases where headache is the main symptom, there may be an overestimation of its severity, while in other cases stroke may go un- recognized if additional neurological symptoms—such as paralysis or altered consciousness—are absent.
Education of dispatchers and the general public on recognizing atypical symptoms of stroke could con- tribute to more accurate urgency assessment and reduce the risk of stroke being overlooked during emergency medical calls.
The association between primary diagnoses in- dicative of possible stroke (ICD-10: G45–G45.9, I60– I69.9) and selected triage criteria within the chap- ters “Headache” (A.18.01 – A.18.07) and “Altered
Consciousness/Paralysis” (A.25.01 – A.25.09) was analyzed. The results were compared with other tri- age criteria not directly associated with suspected stroke (Table 7–9).
Among the calls resulting in a primary stroke-related diagnosis, the “Headache” and “Altered Conscious- ness/Paralysis” criteria were selected in 463 cases. Of these, 101 cases were diagnosed as transient is- chemic attack (G45–G45.9), and 362 as cerebrovas- cular diseases (I60–I69.9).
In the group of calls where other triage criteria were applied, a primary stroke diagnosis was assigned in 356 cases, of which 78 were classified as transient ischemic attack (G45–G45.9) and 278 as cerebrovas- cular diseases (I60–I69.9).
Table 7. Distribution of stroke diagnoses by selected and other triage criteria | ||
Stroke Diagnosis | Headache and Altered Consciousness/ Paralysis Criteria | Other Triage Criteria |
Transient Ischemic Attacks (G45–G45.9) | 101 | 78 |
Cerebrovascular Diseases (I60– I69.9) | 362 | 278 |
Total | 463 | 356 |
A chi-square test was used to examine the asso- ciation between the selected triage criteria (“Head- ache” and “Altered Consciousness/Paralysis”) and the final stroke diagnosis. The statistical signifi- cance of the difference between two groups of tri-
Table 8. Distribution results based on selected triage criteria and final stroke diagnosis | |||
Triage Criteria Used in MPDJ | Confirmed Stroke | No Stroke Diagnosed | Total |
Headache/Altered Consciousness/Paralysis | 463 | 683 | 1,146 |
Other Criteria | 356 | 25,359 | 25,715 |
Total | 819 | 26,042 | 26,861 |
age criteria was analyzed: selected criteria (“Head- ache” and “Altered Consciousness/Paralysis”) versus all other triage criteria.
The results of the chi-square test showed an ex- tremely high level of statistical significance (χ² = 5649.92; p < 0.001), clearly indicating a significant difference between the choice of triage criteria and the final diagnosis of stroke. Notably, 356 patients (43.5%) out of a total of 819 confirmed stroke di- agnoses were not initially identified by dispatch- ers using specific criteria (“Headache” or “Altered Consciousness/Paralysis”), but were assessed using other, less specific criteria.
Table 9 confirms a highly statistically significant dif- ference between the selected triage criteria (“Head- ache” and “Altered Consciousness/Paralysis”) and other criteria in relation to the assignment of a final stroke diagnosis (p < 0.05).
These findings highlight the urgent need for further dispatcher training on the early recognition of stroke symptoms, as well as a review and enhancement of existing triage criteria to improve stroke detection accuracy during emergency medical calls.
An additional analysis was conducted to examine the triage criteria selected in cases where the final prima- ry diagnosis was a stroke (ICD-10 codes: G45–G45.9, I60–I69.9). The aim of the analysis was to identify the most frequently used triage criteria and to explore po- tential discrepancies in symptom perception between callers and emergency medical dispatchers.
Table 9. Chi-square test results for association between selected triage criteria and stroke diagnosis | ||
Statistical Test | Value (χ²) | p-value |
Chi-square (χ²) | 5649.92 | < 0.001 |
Chi-square with Yates’ correction
5636.73 < 0.001
Table 10 presents the distribution of triage criteria used for patients who were ultimately diagnosed
Table 10. Most Frequently Used Triage Criteria in Confirmed Stroke Diagnoses | |||
Criterion | Description | Calls (I60–I69.9) | Calls (G45–G45.9) |
A.05.08 | Suspected red criterion (no additional data immediately available) | 44 | 15 |
H.25.03 | Sudden confusion/somnolence without known cause | 27 | 11 |
H.25.04 | Prolonged confusion/somnolence | 24 | 1 |
H.05.01 | Exhausted patient (unreliable/unclear data) | 22 | 1 |
A.01.03 | Unconscious adult, breathing | 20 | 4 |
H.25.05 | Sudden paralysis, rapidly resolves | 19 | 11 |
H.05.10 | Other yellow criterion (no appropriate criterion available in the Index) | 15 | 8 |
H.05.06 | Sudden dizziness with apparent physical weakness | 14 | 6 |
H.05.09 | Suspected yellow criterion (no additional data immediately available) | 9 | 1 |
A.27.02 | Severe difficulty breathing | 7 | 1 |
A.05.09 | Other red criterion (no appropriate criterion available in the Index) | 5 | 2 |
A.09.04 | Chest pain or discomfort – with breathing difficulty | 4 | 1 |
Table 11. Nonspecific Criteria Used in Confirmed Stroke Cases | |||
Criterion | Description | Calls (I60–I69.9) | Calls (G45–G45.9) |
A.05.02 | Conscious, weakness and near-syncope | 1 | 1 |
H.14.06 | Immunosuppression and fever | 2 | 1 |
H.23.09 | Back pain, partial loss of sensation in the legs | 0 | 1 |
H.05.12 | Communication difficulties and unclear situation | 1 | 1 |
A.27.03 | Barely able to speak due to breathing difficulty | 1 | 2 |
with either cerebrovascular disease (I60–I69.9) or transient ischemic attack (G45–G45.9).
The most frequently used criteria for patients with a primary stroke diagnosis include those related to loss of consciousness (A.01.03), general signs of severe medical condition (A.05.08), and nonspecific symptoms such as confusion or dizziness (H.25.03, H.05.06). These findings suggest that stroke symp- toms are often not recognized in their early stages and highlight the need for additional education of both callers and dispatchers.
The analysis also showed that in some cases, callers reported nonspecific symptoms such as exhaustion, shortness of breath, or chest pain, which may result in insufficient suspicion of stroke during triage. Cri- teria such as “Sudden dizziness” (H.05.06) or “Sud- den confusion/somnolence” (H.25.03) were found to be relatively common in cases of confirmed cerebro- vascular disease but are not necessarily specific to stroke (as shown in Table 11).
The results of the analysis highlight a discrepancy between the symptoms reported by callers and the confirmed stroke diagnoses. The most commonly used criteria relate to general signs of severe medical condi- tions, whereas specific stroke symptoms such as pa- ralysis or speech loss were reported in fewer cases.
Callers sometimes focus on symptoms that are most subjectively noticeable to them, while stroke-specific symptoms may remain unreported or unrecognized.
These findings indicate the need for additional edu- cation on stroke symptom recognition among indi- viduals calling emergency medical services. Timely recognition of key symptoms can improve patient outcomes and reduce the risk of delayed treatment.
Gender and age analyses were conducted exclusively on patients with a confirmed stroke diagnosis (ICD- 10: G45–G45.9, I60–I69.9). Within this subset of 819 patients, the triage criteria selected by dispatchers were examined to determine whether gender or age influenced the categorization of symptoms within the Croatian Emergency Call Reception Index. To assess the association between gender and assigned triage criteria, a chi-square (χ²) test was performed using ob- served and expected values. Data from 419 men and 400 women were analyzed, as presented in Table 12.
The p-value from the χ² test was 0.951, indicating no statistically significant difference between men and women in the allocation of triage criteria (p > 0.05). These results suggest that gender does not influence the assignment of triage codes within the Croatian Emergency Call Reception Index, pointing to consistency in dispatcher assessment.
Table 12. Distribution of triage criteria by gender in patients with confirmed stroke (n = 819) | |||
Triage category | Male (n = 419) | Female (n = 400) | Total |
Headache (A.18.01–A.18.07) | 1 (0.2%) | 1 (0.3%) | 2 (0.2%) |
Altered consciousness/paralysis (A.25.01–A.25.09) | 234 (55.9%) | 227 (56.8%) | 461 (56.3%) |
Other criteria | 184 (43.9%) | 172 (43.0%) | 356 (43.5%) |
Total | 419 | 400 | 819 |
Table 13. Age Differences by Gender | |||
Gender | Mean Age | Standard Deviation (SD) | Number of Cases (n) |
Male (M) | 71.10 | 11.61 | 419 |
Female (F) | 76.46 | 12.81 | 400 |
In addition to the gender distribution analysis, an age- related comparison was conducted (Table 13). It was found that female patients were, on average, older than male patients (M = 76.46, SD = 12.81 vs. M = 71.10, SD = 11.61). A Welch’s t-test confirmed that the difference was statistically significant (t = –6.26, p
< 0.001), indicating that female stroke patients were significantly older than their male counterparts.
Given the statistically significant age difference be- tween genders, the finding is clinically relevant from an epidemiological perspective; however, it is unlike- ly to directly influence dispatcher decision-making, as triage assessments are based solely on caller-re- ported symptoms rather than patient demographics.
The results of this study confirmed the impor- tance of timely and accurate stroke triage within the Medical Dispatch Unit (MPDJ), highlighting key challenges in recognizing stroke symptoms based on information provided by the caller. Given that patients often do not call for themselves, but rath- er a third party reports the symptoms, the role of the dispatcher becomes crucial in decision-making based on the Croatian Emergency Call Reception In- dex. The study showed that certain triage criteria more frequently correlate with stroke diagnoses. However, in as many as 43.5% of stroke patients, the initial triage did not fall under the “Headache” or “Disturbance of Consciousness/Paralysis” cat- egories but was classified under other, less specific criteria. This finding suggests the need to improve dispatcher assessment accuracy and optimize tri- age processes through further training for both dis- patchers and the general public.
Analysis of triage criteria distribution showed that stroke patients were most frequently categorized under the chapter “Disturbance of Consciousness/ Paralysis” (A.25), with some cases classified under “Headache” (A.18), even though headache has rarely been shown to indicate stroke. This may point to im- precise symptom interpretation by either the caller or the dispatcher. A similar pattern was reported by Wenstrup et al. (3), who noted that structured triage protocols offer high specificity but variable sensitiv- ity. This means that some symptoms may be misclas- sified, potentially resulting in missed stroke recogni- tion and delays in patient care.
It is particularly noteworthy that patients as- signed criteria such as “Sudden speech difficulties” (A.25.05) and “Sudden paralysis, rapidly returning to normal” (H.25.05) had a significantly higher like- lihood of being diagnosed with stroke compared to those initially reporting only a headache. This dis- tribution of triage criteria highlights the need for more precise communication during calls and addi- tional dispatcher education to identify subtle, less pronounced and noticeable signs of neurological deficit, even when not explicitly recognized by the caller as stroke symptoms.
One of the key findings of this research is that some patients later confirmed to have suffered a stroke were initially triaged under unrelated categories such as “Unclear problem” or “Headache.” This suggests that the information obtained during the call may be subjectively interpreted by the caller, making accurate dispatcher assessment more challenging. Additionally, inter-dispatcher variability in evaluating certain symp- toms may further influence triage decisions. Jamtli et al. (2) report that dispatchers often face limited infor- mation during emergency calls, and their interpreta- tion of symptoms can depend on their experience, training, and the clarity of the caller’s description. In
addition to dispatcher-related factors, caller charac- teristics - such as the clarity of symptom descriptions, emotional state and level of health literacy - may sub- stantially influence how stroke symptoms are commu- nicated and interpreted during the call.
Studies have shown that the sensitivity of stroke recognition by dispatchers ranges from 18% to 83% (14), which aligns with this study’s findings, where certain symptoms were not identified as indicative of stroke. A particular challenge lies in the fact that stroke symptoms can be atypical or non-specific, making correct triage decisions more difficult. For example, in some cases, callers reported symptoms such as dizziness, confusion, or chest pain—symp- toms not necessarily considered stroke-related by dispatchers but which ultimately led to a cerebrovas- cular diagnosis.
The results of the chi-square test (χ² = 0.005, p = 0.951) showed no statistically significant difference in the allocation of triage criteria between male and female patients. The notably high p-value further supports the conclusion that the patient’s gender does not influence the dispatcher’s decision when categorizing stroke within the Croatian Emergency Call Reception Index. These findings are consistent with previous research showing that dispatchers primarily base their decisions on the symptoms de- scribed by the caller rather than on the patient’s de- mographic characteristics (4).
However, it is known that women more frequently present with atypical stroke symptoms such as nau- sea, faintness, and general weakness, which can make early recognition in the prehospital phase more difficult (16). Since dispatchers lack visual contact with the patient, their assessment fully relies on the information provided by the caller. This can lead to variability in symptom descriptions and potentially to underestimation of atypical stroke presentations in women. Although the results of this study did not indicate a significant difference in triage categori- zation by gender, further research is needed to ex- amine whether unconscious bias may influence the triage assessment of symptoms in men and women.
The analysis of age differences revealed a statisti- cally significant disparity between male and female patients (t = -6.26, p < 0.001). The average age of
stroke patients was 71.10 years for men and 76.46 years for women, indicating that women tend to ex- perience stroke at an older age. This pattern aligns with previous research suggesting that women are more likely to suffer strokes later in life, a trend at- tributed to longer life expectancy, delayed exposure to vascular risk factors, and differing hormonal influ- ences (17). However, given that dispatchers do not see the patient and rely entirely on caller-reported symptoms, this age difference is unlikely to have a direct impact on real-time triage decision-making. In the context of emergency calls, the caller’s interpre- tation and communication of symptoms represent a far more influential factor than demographic charac- teristics of the patient.
Nevertheless, since patients most often do not call emergency services themselves, and symptoms are typically reported by a third party, the key factor in stroke recognition is not necessarily the patient’s age but the way in which the caller describes the symptoms. These findings are consistent with the claims of Jamtli et al. (2), who emphasize that the caller’s subjective interpretation of symptoms can have a greater impact on dispatcher decisions than the patient’s clinical presentation itself.
Due to the retrospective design, the analysis was lim- ited to data available within the e-Hitna system. This means it was not possible to control for the accuracy or completeness of symptom descriptions provided by callers during emergency calls. The subjective in- terpretation of symptoms by callers can significantly influence the dispatcher’s final triage decision, which constitutes an important methodological limitation. Furthermore, the study did not include follow-up of hospital diagnoses or clinical outcomes, which would allow for a more accurate assessment of triage deci- sion validity. Future studies should consider linking MPDJ data with final hospital diagnoses and patient outcomes to enable a more detailed evaluation and potential improvement of existing triage protocols. An additional limitation concerns the unequal size of the analysed groups, particularly the small number of cas- es in the Headache category compared with Altered Consciousness/Paralysis and other triage codes. This imbalance may reduce statistical power for some com- parisons and increase the uncertainty around effect estimates, despite the use of appropriate tests (e.g., Welch’s t-test and Fisher’s exact test) to account for
these differences. In addition, multivariable modelling techniques, including logistic regression, could not be applied. Essential predictor variables such as age, sex and caller-related characteristics were available only for patients with confirmed stroke diagnoses, and not for the full set of emergency calls. Applying regression under such constraints would compromise the valid- ity and interpretability of the results. For this reason, the analysis was limited to descriptive statistics and bivariate tests, which were methodologically appropri- ate given the structure of the available data.
The findings of this study may serve as a founda- tion for improving prehospital stroke triage within MPDJ operations. Identifying the most frequently used triage cards and comparing them with primary diagnoses provides insight into potential gaps in recognizing stroke during emergency medical calls. The results highlight the need to educate not only dispatchers but also the general public about stroke
mentation of additional diagnostic algorithms within medical dispatch systems, in order to reduce the number of unrecognized strokes. Additionally, rais- ing public awareness about stroke symptoms could decrease the number of inaccurate symptom presen- tations, thereby facilitating the work of dispatchers.
Stroke triage within the Medical Dispatch Unit is a complex process in which callers and dispatchers— not patients—play a central role. The findings of this study demonstrated that certain triage criteria are more frequently associated with confirmed stroke diagnoses, while others remain underrecognized. These insights underscore the need to improve dis- patcher education and communication strategies in order to enhance out-of-hospital stroke recognition and ensure timely intervention for affected patients.
symptoms and the importance of clear, timely com- munication during emergency calls.
The findings of this study point to several key areas for improving stroke recognition in Medical Dispatch Units. First and foremost, dispatcher training should include specialized modules focused on identifying atypical stroke symptoms such as dizziness, nau- sea, and transient confusion. In addition, enhanc- ing communication strategies—for example, through the implementation of specific questions regarding neurological deficits—could improve triage accuracy. In practice, dispatcher training should incorporate simulation scenarios based on real-life cases involv- ing atypical stroke presentations. For example, work- shops where dispatchers listen to real emergency call recordings and then discuss the appropriateness of the triage decisions made could significantly im- prove their ability to recognize atypical symptoms.
Furthermore, dispatchers should receive regular feed- back from hospital staff on the final diagnoses of pa- tients they triaged. This would enhance their clinical judgment and support continuous quality improvement.
Further research should examine the effectiveness of educational interventions and the potential imple-
The analysis provided insight into how stroke-related symptoms are categorized during emergency medi- cal calls and highlighted the crucial role of dispatch- ers in early recognition. Triage criteria within the chapter “Altered Consciousness/Paralysis” (A.25) were most strongly associated with confirmed stroke diagnoses, whereas headache-related criteria rarely indicated stroke. A considerable proportion of stroke cases were initially assigned to less specific triage chapters, underscoring the challenges posed by call- er-reported symptoms and the complexity of identi- fying neurological deficits in the prehospital phase.
These findings point to the need for enhanced dispatch- er training focused on recognising atypical and unspe- cific stroke presentations, as well as continued public education to improve the accuracy of symptom report- ing during emergency calls. Strengthening communica- tion strategies, together with systematic feedback from hospital outcomes, may contribute to more accurate tri- age and reduce the likelihood of missed strokes. Improv- ing dispatcher competencies and public awareness rep- resents a key step toward optimizing prehospital stroke pathways and supporting timely intervention.
Conceptualization and methodology (SP); data cura- tion and formal analysis (SP); investigation and pro- ject administration (SP); Supervision (BV); writing – original draft (SP); writing – review & editing (IP, BV). All authors have approved the final manuscript.
The authors declare no conflict of interest.
We thank the Emergency Medical Institute of Zagreb County for data access and institutional support.
This research did not receive any specific grant from funding agencies in the public, commercial or not for- profit sectors.
Gude MF, Valentin JB, Christensen HC, Mikkelsen S, Søvsø MB, Andersen G, et al. Associations between emergency call stroke triage and pre-hospital delay, primary hospital admission, and acute reperfusion treatment among early comers with acute ischemic stroke. Intern Emerg Med. 2023;18(8):2355–65. https://doi.org/10.1007/s11739-
023-03349-x
Jamtli B, Svendsen EJ, Jørgensen TM, Kramer-Johansen J, Hov MR, Hardeland C. Factors affecting emergency medi- cal dispatchers decision making in stroke calls – a qualita- tive study. BMC Emerg Med. 2024;24(1):214. https://doi. org/10.1186/s12873-024-01129-0
Wenstrup J, Hestoy BH, Sagar MV, Blomberg SNF, Chris- tensen H, Christensen HC, et al. Emergency Medical Services dispatcher recognition of stroke: A system- atic review. Eur Stroke J. 2024;9(2):283–94. https://doi. org/10.1177/23969873231223339
Patel A, Fang J, Gillespie C, Odom E, King SC, Lucheon C, et al. Awareness of Stroke Signs and Symptoms and Calling 9-1-1 Among US Adults: National Health Interview Survey, 2009 and 2014. Prev Chronic Dis. 2019;16:E78. https:// doi.org/10.5888/pcd16.180564.
Heart and Stroke Foundation of Canada. Stroke Aware- ness, Recognition, and Response [Internet]. Cana- dian Stroke Best Practices. Available at: https://www. strokebestpractices.ca/recommendations/acute-stroke- management/stroke-awareness-recognition-and-re- sponse Accessed: 04.03.2025.
Bietzk E, Davies R, Floyd A, Lindsay A, Greenstone H, Sy- monds A, et al. FAST enough? The UK general public’s
understanding of stroke. Clin Med (Lond). 2012;12(5):410–
5. Available at: https://doi.org/10.7861/clinmedi- cine.12-5-410
Jones SP, Jenkinson AJ, Leathley MJ, Watkins CL. Stroke knowledge and awareness: an integrative review of the evidence. Age Ageing. 2010;39(1):11–22. https://doi. org/10.1093/ageing/afp196
Hrvatski Zavod za Hitnu Medicinu. Predstavljeno dopun- jeno izdanje „Hrvatskog indeksa prijema hitnog poziva za medicinsku prijavno-dojavnu jedinicu“ [Internet]. Available at: https://www.hzhm.hr/predstavljeno-dopunjeno-izdan- je-hrvatskog-indeksa-prijema-hitnog-poziva-za-medicin- sku-prijavno-dojavnu-jedinicu/ Accessed: 04.03.2025.
Ellensen EN. Norwegian Index for Emergency Medical As- sistance. Studies on the Use and Precision of the Emer- gency Medical Dispatch Guidelines in Norway [Doctoral dissertation]. The University of Bergen; 2017. Available at: https://bora.uib.no/bora-xmlui/handle/1956/16736 Ac- cessed: 16.07.2025.
Ellensen EN, Naess H, Wisborg T, Hunskaar S, Zakariassen E. Stroke identification by criteria based dispatch – a register based study. Acta Anaesthesiol Scand. 2018;62(1):105–
15. https://doi.org/10.1111/aas.13032
Grusd E, Kramer-Johansen J. Does the Norwegian emergen- cy medical dispatch classification as non-urgent predict no need for pre-hospital medical treatment? An observational study. Scand J Trauma Resusc Emerg Med. 2016;24(1):65. https://doi.org/10.1186/s13049-016-0258-8
Wenstrup J, Blomberg SN, Christensen H, Folke F, Chris- tensen HC, Kruuse C. Dispatcher Stroke/TIA Recogni- tion in Emergency Medical Call Center and Out-of-Hours Service Calls in Copenhagen, Denmark. Neurol Clin Pract. 2023;13(6):e200197. https://doi.org/10.1212/ CPJ.0000000000200197
Abbas AY, Odom EC, Nwaise I. Association Between Dis- patch Complaint and Critical Prehospital Time Intervals in Suspected Stroke 911 Activations in the National Emer- gency Medical Services Information System, 2012–2016. J Stroke Cerebrovasc Dis. 2022;31(3):106228. https://doi. org/10.1016/j.jstrokecerebrovasdis.2021.106228
Viereck S, Møller TP, Iversen HK, Christensen H, Lippert
F. Medical dispatchers recognise substantial amount of acute stroke during emergency calls. Scand J Trauma Re- susc Emerg Med. 2016;24(1):89. https://doi.org/10.1186/ s13049-016-0277-5
Watkins CL, Leathley MJ, Jones SP, Ford GA, Quinn T, Sutton CJ. Training emergency services’ dispatchers to recognise stroke: an interrupted time-series analysis. BMC Health Serv Res. 2013;13(1):318. https://doi.org/10.1186/1472- 6963-13-318
Hrvatski zavod za javno zdravstvo. Moždani udar kod žena: simptomi, čimbenici rizika i prevencija [Internet]. Available at: https://www.hzjz.hr/aktualnosti/mozdani- udar-kod-zena-simptomi-cimbenici-rizika-i-prevencija/ Accessed: 10.03.2025.
Bushnell C, Howard VJ, Lisabeth L, Caso V, Gall S, Kleindor- fer D, et al. Sex differences in the evaluation and treatment of acute ischaemic stroke. Lancet Neurol. 2018;17(7):641–
50. https://doi.org/10.1016/S1474-4422(18)30201-1
Table S1. Included triage criteria from Chapters A.18 and A.25 (with justification) | ||||
Chapter | Code | Description | Included | Justification |
A.18 | A.18.01 | Ne reagira na protresanje i pozivanje | Yes | Included in analytical scope (A.18.01–A.18.07) |
A.18 | A.18.02 | Abnormalno ili otežano disanje | Yes | Included in analytical scope |
A.18 | A.18.03 | Iznenadna jaka, neuobičajena glavobolja | Yes | Included in analytical scope |
A.18 | A.18.04 | Glavobolja + mučnina | Yes | Included in analytical scope |
A.18 | A.18.05 | Glavobolja + paraliza | Yes | Included in analytical scope |
A.18 | A.18.06 | Glavobolja + otežani govor | Yes | Included in analytical scope |
A.18 | A.18.07 | Glavobolja + smušenost | Yes | Included in analytical scope |
A.18 | A.18.08 | Temp >38.5 °C + ukočen vrat | No | Infectious etiology; outside analytical scope |
A.18 | A.18.09 | Temp >38.5 °C + osip | No | Infectious etiology; outside analytical scope |
A.18 | A.18.10 | Konvulzije | No | Seizure-related; outside analytical scope |
A.25 | A.25.01 | Ne reagira na pozivanje | Yes | Included in analytical scope (A.25.01–A.25.09) |
A.25 | A.25.02 | Otežano disanje | Yes | Included in analytical scope |
A.25 | A.25.03 | Iskrivljenje lica | Yes | Included in analytical scope |
A.25 | A.25.04 | Gubitak snage u ruci/nogi | Yes | Included in analytical scope |
A.25 | A.25.05 | Poteškoće u govoru | Yes | Included in analytical scope |
A.25 | A.25.06 | Smetenost; sumnja na moždani udar | Yes | Included in analytical scope |
A.25 | A.25.07 | Nagla jaka glavobolja | Yes | Included in analytical scope |
A.25 | A.25.08 | Slabost pri svijesti | Yes | Included in analytical scope |
A.25 | A.25.09 | Hladni i oznojeni | Yes | Included in analytical scope |
A.25 | A.25.10 | Abdominalna bol | No | Non-neurological; outside analytical scope |
A.25 | A.25.11 | Ubrzan puls; slabost | No | Non-neurological; outside analytical scope |