This study is part of a larger project called Non-conveyance - Go to Other Level of Care (No-Go), which has the overarching purpose of describing, analysing, and developing safe non-conveyance guidelines and decision-making support systems [3, 9].

This was a prospective analytical study that consecutively included patients who were not conveyed by the EMS. The current study follows the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines [20].

The study was conducted in Region Örebro county, Sweden. The region had a population of approximately 295,000. The regional EMS consisted of three ambulance departments that received approximately 30,000 assignments per year. Twelve ambulances operated 24 h per day seven days a week, and an additional four ambulances operated during the daytime and evening hours. In the studied region, there was one level-one trauma centre and two smaller hospitals. The region consisted of both rural areas and small- to mid-sized cities. Registered nurses, with three years of university education, and specialist ambulance nurses, with an additional year of university education and a master’s degree in emergency care, staffed the ambulances. The ambulances were staffed by two people. An emergency medical technician, with the equivalent of one and one-half to two years of basic education and an additional practical ambulance education, sometimes replaced one nurse or specialist ambulance nurse in the team. Only specialist ambulance nurses are employed full time on a permanent basis, making it uncommon for ambulances to be staffed with registered nurses.

The regional guidelines for non-conveyance, which were implemented in 2015, allow registered and specialist nurses to independently make the decision to not convey a patient. The guideline checklist is provided as Additional file 1 and additional exclusion criterias as Additional file 2. The guidelines were designed to be restrictive and to identify as many patients as possible who might risk deterioration. The guidelines state that patients cannot require the administration of drugs, supervision, or monitoring during transport to a health care facility. Patients (or their legal guardians) also must be able to communicate and understand the decision and information provided. The nurses have the option to contact a physician at the receiving hospital who can make a decision that contradicts the recommendation based on the guidelines. According to the guidelines, recommendations can be made for patients to seek different levels of care. The lowest level of care is self-care, followed by primary health care and care in the ED reached via personal or public transport.

When the decision is made to not convey a patient, a document containing information about the assessment, recommendations regarding further health care contact, and where the patient can seek assistance if the condition worsens is created and given to the patient. The nurse can non-convey a patient even if the patient prefers to go to the hospital by ambulance. Most often, the patient is involved in the decision.

Patients can be conveyed even if, according to regional non-conveyance guidelines, they are eligible for non-conveyance. Patients can, for instance, be conveyed due to humanitarian reasons or a lack of alternate options, or if the nurse, regardless of the triage outcome, believes that the patient needs additional assessment. During the study period, region-specific guidelines were used together with the Rapid Emergency Triage and Treatment System (RETTS) [21, 22]. For non-conveyance to be considered, all vital parameters must be within the normal ranges. Children must receive the lowest triage level colour (green), and adults (≥ 18 years) must receive the lowest (green) or next lowest (yellow) triage level in the four-colour system, in which patients can be assigned green, yellow, orange, or red triage colours.

The triage system combines the patient’s signs and symptoms and the main complaint, coded with an emergency signs and symptoms (ESS) code, with the patient’s vital signs to determine the urgency level. These ESS codes and vital sign cut-offs are different for children and adults. The different urgency levels indicate the time within which the patient needs to be seen by a physician. Green and yellow triage level colours indicate that the patient does not need immediate emergency care and could wait for longer than three hours to receive an additional assessment. The orange triage level colour means that the patient has urgent medical needs but could wait for up to 20 min before being assessed by a physician. The red triage level colour means that the patient needs immediate assessment and treatment. The RETTS was initially designed for use in the ED. Neither the regional non-conveyance guidelines nor the RETTS were validated for use when making prehospital non-conveyance decisions at the beginning of this study.

All paediatric and adult patients who were not conveyed by the EMS in the studied region between February 2016 and January 2017 were included. Patients were excluded if they refused care, examination or conveyance. Patients who were found dead at arrival were also excluded.

Data were obtained from handwritten medical records. All records produced during the study period and submitted according to the regional guidelines for non-conveyance documentation were included in the study. The regional non-conveyance guideline was implemented approximately one year before the start of data inclusion. Patients could be non-conveyed more than one time. A total of 163 patients were non-conveyed more than once, corresponding to 431 (16%) cases in the database. The handwritten medical records were manually entered into the study-specific database. Instructions for each variable guided database entry. The database fields had fixed boundaries making it impossible to enter ambiguous values, for example, numbers in text fields and numbers outside what is possible for a specific variable. The database also consisted of multiple-choice options from drop-down menus to guide database entry and ensure data quality. The variables sex, age, day of the week, time of day, ESS code, triage level colour, and destination were collected from the handwritten medical records. All database entries included personal identity numbers, which allowed correct linkage for every case. Follow-up data were both digitally and manually extracted from the hospital-specific digital medical record system (Klinisk Portal). The hospital medical record system automatically retrieves vital statistics, such as death, from the Swedish population register database. Chart review was performed by the first author.

The study focused on four outcome measures, ED visits, admissions to in-hospital care, intensive care units, and mortality, all within seven days, among patients who were not conveyed to the ED by the EMS.

Before analysis, 30% of the manually entered data were randomly checked by the first author and administrators at the University Health Care Research Centre. Each handwritten record was compared with the previously entered data in the digital database. This was performed to ensure that data from the handwritten non-conveyance documents had been correctly transferred to the database. Database entry errors were < 0.25%. The variable ‘sex’ had 30 missing values, and the variable ‘time of day’ had 60 missing values. Categorical variables were described as numbers and percentages and were analysed with the chi-square test and proportion test. Continuous variables were described as the medians and interquartile ranges and were assessed with Mann–Whitney U tests and two-sample t tests. Univariable logistic regression was used to analyse the relationships between the included variables and the outcome measures. All variables except destination and ESS triage level colour were included in the regression models. Destination and ESS triage level colour were not included due to the amount of missing data. All other variables were used individually to test for linear relationships. The outcome intensive care is not included in the tables since no one received intensive care during follow up.

The data were tested by both graphical and numerical methods to assess the normality of the distributions. A p-value threshold of < 0.05 was used to test for statistical significance. Bonferroni corrections for > 5 groups were used to adjust for statistical significance. All patients who visited the ED within seven days are included in the tables and analysis regardless of the non-conveyance destination. Hospitalised patients are also included in the “ED visit” column. In total, 32 different ESS codes were used. The data were analysed using STATA 15.1 software (College Station, Texas, USA: Stata Corporation).