This study included all non-conveyed patients in the Central Denmark Region during a 2-year period and followed the patients for 48 h after their initial emergency call. Based on our findings, non-conveyance seems to be a safe practice, as most patients did not recontact the emergency healthcare system after being discharged on scene and mortality was low. Among non-conveyed patients, 4.7% (95% CI: 4.4–5.0%) were reassessed by the EMS, while 4.9% (95% CI: 4.6–5.3%) were admitted to the hospital within 48 h. Although it is challenging to determine an acceptable level of recontact that will balance the consideration of resource utilization and patient safety, [31] these findings are comparable to previously published studies from Sweden [20], Finland [19, 32], and Australia [18] with a similar study design. However, differences in follow-up periods, patient populations and definitions of outcomes complicate direct comparisons. These differences may also explain that the mortality was lower in the current study (0.08%, 95 CI: 0.04–0.13%) compared with the general literature (0.2–3.5%), e.g. some previous studies included patient-initiated non-conveyance, which may carry a greater risk of adverse outcomes [11].

While absolute risks were generally low across most patient groups, patient safety can be further enhanced by carefully considering non-conveyance decisions among certain patients with increased risk of reassessment and hospital admission. These includes older patients, males, those with abnormal vital signs and certain EMS dispatch codes. As in previous studies, older patients were more often reassessed by an EMS provider and admitted to the hospital after non-conveyance, when compared to younger patients [7, 16,17,18, 20]. This could be due to comorbidities, polypharmacy, behavioural changes, and different symptom presentation, contributing to the vulnerability of older patients [16]. More unexpectedly, we also found increased risk of both outcomes among men, when compared with women. Such gender differences have not previously been reported.

In line with previous studies, we found that having one or more abnormal vital signs were predictors of recontact to the emergency healthcare system [16, 18, 19, 23], suggesting that vital signs are helpful indicators of patient risk in relations to non-conveyance, although they should be seen in conjunction with other predictors, especially age and dispatch codes, that were stronger predictors in this study. We carried out additional analyses to gain detailed insights into the association between vital signs and patient outcomes across older age groups. These analyses showed the highest risk among patients aged 50–64 with low systolic blood pressure (< 90 mmHg).

Overall, the highest risk of both EMS reassessment and hospital admission was found among patients with the EMS dispatch code ‘ear, nose, throat’ with 17.6% of assignments ending in EMS reassessment and 16.0% in subsequent hospital admission within 48 h. Hence, EMS providers should critically consider non-conveyance decisions among such patients. Increased risk of both outcomes was also found among non-conveyed patients with the EMS dispatch code ‘seizures’ (e.g. epileptic seizures), which is in accordance with a previous study, where seizures were associated with the highest risk of ED attendance after non-conveyance [17]. The Central Denmark Region guidelines state that patients with seizures can be discharged on scene in cases of known epilepsy if the patient is considered stable after treatment. However, non-conveyance among patients with first-time seizures should be carefully considered. The current study could not differentiate between patients with known epilepsy and first-time seizures but points to greater caution when making non-conveyance decisions among these patients. Furthermore, increased risk was found among patients with psychiatric symptoms, which is in line with previous studies [18, 22]. Some evidence suggests that EMS personnel lack the required skills to manage such patients [33], indicating that new initiatives aimed at reducing the need for subsequent treatment for patients presenting primarily with psychiatric symptoms should be prioritized.

A key limitation of this study was the inability to differentiate between patients only attending the ED within 48 h and those admitted to other hospital departments, as in previous studies [11]. Both outcomes were categorized as hospital admissions, though prolonged admission may indicate a more severe outcome measure than ED attendance only. Additionally, there is potential misclassification in identifying the patient’s primary medical issue, as EMCC personnel assess the main presenting complaint during the emergency call based on brief patient descriptions, often during stressful circumstances. Therefore, the EMS dispatch code may not reflect the final diagnosis. Also, we could not confirm if hospital admission and EMS reassessment were directly related to the non-conveyance decision or a separate medical issue, introducing potential exposure misclassification.

Lastly, we categorized missing measurements of vital signs as normal, as we assumed the EMS personnel measured only relevant vital signs. For instance, most patients did not have their blood glucose measured. We believe that this was due to patients not having any complaints related to hypo- or hyperglycemia. Instead of categorizing the majority of patients with missing vital signs, we found this assumption reasonable for most patients. However, it cannot be ruled out that there was an overrepresentation of missing values among patients with an adverse outcome, hence introducing information bias. Nevertheless, we expect the vast majority of missing values to be attributable to a professional assessment of the patient leading to the conclusion that vital signs were stable.

Notwithstanding these limitations, our study had several strengths. The study had a large sample size and population-based design, including all EMS assignments within the region during a 2-year period. We followed patients for 48 h after first EMS contact by linking their registered personal identification number to in-hospital medical records within the region with less than 7% of assignments lacking such registration. Given that all patients are provided with free healthcare in Denmark, the risk of selection bias is substantially reduced as compared with healthcare system, where financial considerations play an important role in the selection of patients refraining from transportation to the hospital.

Results from the present study can guide EMS providers in making safe non-conveyance decisions on a population-based level. Most importantly, EMS providers should critically consider non-conveyance among patients with ear, nose and throat complaints, and to a lesser extent patients with seizures, non-traumatic bleeding and psychiatric complaints. Although absolute risks were low across most patient groups, patient safety could be further enhanced by carefully considering non-conveyance decisions among older patients and those with abnormal vital signs. Unexpectedly, the risk of both outcomes was higher for men compared with women. Given the exploratory nature of the study, novel findings from the current study should be confirmed by future research before they are integrated into non-conveyance guidelines, including the high risk found among patients with ear, nose and throat complaints. Moreover, we found that EMS providers spend a median 42.4 min (IQR: 31.2–55.2) on-scene, which raises questions of the cost-effectiveness of non-conveyance. While it is beyond the scope of this study, future studies should investigate the cost-effectiveness of non-conveyance, as longer time spend on-scene may warrant a different allocation of resources throughout the healthcare system, if prehospital EMS are increasingly treating and releasing patient on-scene, hence contributing to decreased patient load on EDs. However, releasing the patient on-scene saves the EMS vehicle the added time of transporting and handing over the patient to the hospital, thus freeing up time for new assignments. In Denmark, there is no upper limit on prehospital expenditures. Only in recent years has The Danish Medicines Council been established to assess the value of expensive treatments, however, not at this level.

Furthermore, future studies should aim to develop an algorithm to determine when non-conveyance is safe, preferably based on classification-based methods, like machine-learning, allowing for interpretations on an individual level rather than interpretations on a group-based level, like in the current study. On this basis, we believe our study could be a starting point for identifying relevant predictors that should be validated in future studies.