Are 5-level triage systems improved by using a symptom based approach?—a Danish cohort study

We reported this study based on The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement and the STROBE explanation and elaboration [10, 11].

Study design

This is a retrospective cohort study based on a secondary analysis of data from “Syddanske Akutkohorte” (SAK-cohort) [12]. The cohort includes all adults who were seen in an emergency department in the Region of Southern Denmark from 1 April 2012 until 31 December 2015. Patients under 18 years and patients without a Danish personal identification number were excluded. We included all ED contacts despite individual patients having multiple visits. All patients were followed until discharge or death whichever came first.

Setting

The SAK-cohort included data from five EDs: Odense University Hospital (OUH), Odense University Hospital Department at Svendborg (OUHS), Hospital of South Jutland (SHS), Hospital of Lillebaelt (SLB) and Hospital of South West Jutland (SVS).

OUH is a university hospital and Level I trauma centre. Together with OUHS, they manage approximately 100,000 patients yearly [13]. SHS, SLB and SVS are regional teaching hospitals. They handle approximately 32,000 (SHS), 33,000 (SVS) and 50,000 (SLB) patients yearly [14,15,16] (Table 1). All regional hospitals offer 24-h emergency care and level-2 trauma including broad medical, surgical, neurological and ICU services.

Table 1 Baseline data on involving centers [13,14,15,16]DEPT triage

OUH, OUHS, SHS and SLB have implemented DEPT triage using vital signs and presenting symptoms since 2011 [17]. At the same time, SVS implemented a local modification of DEPT using only vital signs, VITAL-TRIAGE. The process of triage is carried out by trained staff prior to diagnostics and assessment by a physician. The triage category is based on vital signs (e.g., Glasgow Coma Scale, blood pressure, respiratory rate, heart rate, etc.) and a presenting symptom algorithm (e.g., chest pain, abdominal pain, trauma, unconsciousness etc.). The most urgent of either vital signs or presenting symptoms determines the final triage category.

In VITAL-TRIAGE, the triage category is based solely on vital signs and a visual analogue pain scale (VAS) [18] in case of surgical patients. The triage categories used in both systems are: Red (immediate evaluation by physician), Orange (emergent, evaluation within 15 min), Yellow (potentially unstable, evaluation within 60 min), Green (non-urgent, re-evaluation every 180 min), and Blue (minor injuries or complaints, re-evaluation every 240 min). In DEPT, secondary modifiers are included to increase the triage category of patients whose co-morbidity or medication might lead to an underestimation of the clinical condition at triage. Examples of secondary modifiers are existing ischemic heart disease or insulin demanding type II diabetes, primary immunodeficiencies or medication with immunosuppressants [4]. It is possible to be categorized higher than identified by DEPT and VITAL-TRIAGE if the triage nurse suspect a more critical condition, downward triage can only be done upon assessment by a physician [4]. The triage category is registered using the software package Cetrea Emergency (Getinge Cetrea A/S, Aarhus, Denmark).

Outcome and variables

The primary outcome was admission to intensive care unit (ICU) within 24 h after ED arrival. Secondary outcomes were 2-day mortality, patients subjected to surgery within 48 h, diagnosis of critical illness, length of stay (LOS) and discharge within 4-h.

Exposure variables were each of the triage categories provided by each triages system. In addition, a Grey category was formed containing all patients who were seen in the ED but was not registered with a triage category. Individual level variables of age, gender and Charlson comorbidity index (CCI) [19] was included.

Data sources

Information on sex, hospital, time of arrival to ED and triage category for each ED contact was retrieved from Cetrea. Cetrea data was merged with data from The Danish National Patients registry (DNPR). DNPR contributed with information on length of stay, Charlson co-morbidity index, surgical procedures, time of discharge and primary diagnosis [20]. Age was retrieved from The Danish Civil Registration System using the unique personal identification number in Cetrea [21].

For alle outcomes except critical illness, the time of registration on a Cetrea board in the ED was used as arrival time to ED. Twenty-four hour ICU admission was defined as a maximum of 24 h from arrival time to the ED until the time stamp of the first ICU admission retrieved by DNPR.

Forty-eight hour mortality was counted as death on the same day or the next day, after arrival to the ED using the date of death in The Danish Civil Registration System.

Surgery within 48 h was defined as a maximum of 48 h from arrival to the ED until time of operation as registered in DNPR.

Length of stay was measured from arrival time at ED until time of discharge as registered in DNPR, 4-h discharge was counted as an ED contact with an LOS less than 4 h.

The primary diagnosis of every ED contact was retrieved by DNPR at final hospital discharge despite the patient being submitted to hospitalization in a general ward or discharged directly from the ED.

Diagnoses of critical illness were obtained from registries of The Danish Clinical Quality Program—national clinical Registries (RKKP). RKKP asks expert clinicians to conduct clinical registries used for the improvement of quality, research and surveillance purposes [22]. We screened all databases and included those areas of critical illness where early intervention is important for optimal clinical outcomes. Four areas of disease and the corresponding ICD-10 codes were included: Stroke, Acute Coronary Syndrome, bleeding ulcer and Gastrointestinal perforation [23] (Additional file 1: Table S1).

Statistical methods

Baseline characteristics were presented as numbers, percentages, and medians. We used \(\upchi\)2-test to compare binary outcomes. p values below 0.05 were considered significant and 95% confidence intervals (CI) are presented when appropriate. To describe differences in length of stay, we calculated percentiles and interquartile range. Kruskal–Wallis H test and Wilcoxon rank-sum was used to test for significant differences between triage categories and triage systems.

Odds ratios (OR) were calculated to illustrate probabilities of 2-day mortality and ICU admittance and to stratify for potential confounders.

CCI were grouped into four: 0, 1, 2 and > 2, and age into age-groups: 18–49, 50–64, 65–79 and 79+ years of age. To calculate diagnostic sensitivity and specificity we dichotomised the DEPT categories into high-urgency categories (Red and Orange) and low-urgency categories (Yellow, Green, Blue).

All analyses were conducted using STATA V 16.1 (StataCorp, Texas, USA)

Ethics statement

The study was approved by the Danish data protection agency. Register-based studies are exempt from approval by an ethics committee according to Danish law. No further approval was required.

留言 (0)

沒有登入
gif