In the present study we found that a high RETTS-P priority rate was significantly associated with hospitalization, a long hospital stay and referral to the PICU compared to patients with low priority. This association increased in strength according to higher priority level. The data indicated that more severe conditions were higher prioritized than less severe conditions for both medical and surgical patients. This is the first study to investigate the validity of RETTS-p. Two previous studies have shown that RETTS is a valid tool among adults in Sweden [7] and Denmark [12].
Approximately one of ten children had a life-threatening condition and were triaged red in need of urgent intervention, one third were triaged orange implying examination within 20 min, one third were yellow in need of evaluation within 2 h, and one fifth received the lowest urgency level green with until 4 h waiting time. Interestingly, triage ratings varied between disciplines. Medical patients were more likely to have a high triage priority compared to the other specialties. This could suggest that RETTS-p have systematic errors in the way it prioritizes children. However, our further analyzes suggest that this is not the case. As medical patients also were more likely to have a long hospital stay and to be referred to the PICU, it is likely that they more often were severely ill than the surgical patients, and hence that the triage was correct. Furthermore, for both medical and surgical patients, those with more severe diagnosis at discharge and those receiving advanced treatment had a higher priority when triaged at the PED. For example, those with a lower respiratory tract infection were higher prioritized than children with upper respiratory tract infection. Also, children with appendicitis were higher prioritized than children with constipation and abdominal pain. These factors indicate that in the PED RETTS-p prioritized the children correctly.
We found that approximately one third of the referred children was overtriaged; e.g. 31% of the pediatric patients with red priority were not hospitalized. This is probably explained by the fact that these patients predominantly represented children with airway complaints, which needs promptly medical attention although the distress often is reversible with treatment. In contrast to the medical patients, surgical patients were likely to be admitted to hospital regardless of priority, resulting in relatively high rates of undertriage. We interpret that this finding reflects the need to monitor surgical patients in case they need surgery or develop complications at a later stage. As such, it seems that there are different traditions and procedures in treating medical and surgical patients, but that RETTS-p prioritize patients appropriate in the PED according to their medical state. Considering these inconsistencies, the use of hospitalization and length of hospital stay for validation may be questionable, but the association between high priority and these proxy variables on a group level is undisputable.
Strengths of the present study are the population-based design with children referred from primary health care, and large cohort size with a broad specter of pediatric medical and surgical diagnoses. Weaknesses are a retrospective design, and the fact that the proxy variables we have used could be influenced by other factors than illness severity. E.g. we only found moderate sensitivities and specificities for hospitalization and length of hospital stay, proxy variables that may be influenced by living conditions, care takers and travel distance to hospital etc. [13]. A more robust proxy variable might be referral to the PICU, as it is in general less affected by these factors, and reflects severe illness with need of intensive care, continuous observation and high resource use [13]. Indeed, we found that RETTS-p identified need of PICU referral with a higher sensitivity of 83%. We do not have any information regarding whether RETTS-p prioritized children with chronic illnesses differently from otherwise healthy children. Seiger et al. [11] demonstrated that children with chronic illnesses were in greater risk of being undertriaged by the Manchester triage system (MTS). We would like to think that RETTS-p might be a more sensitive triage tool in this aspect, as it includes ESS scores which gives a higher triage if the child has certain chronic condition. However, we have not compared RETTS-p with a triage tool that does not include ESS scores. At last, we did not adjust for possible confounding factors other than the children’s age. We do not have data regarding chronic illness, travel distance to hospital and socioeconomic status of the parents, which might have influenced the results.
Comparing various triage priority systems is a challenge due to different populations and health care systems, and the use of various proxy variables for true medical urgency [14]. In Norway, children are referred to PEDs after evaluation at general practitioners and physicians at municipal EDs. Recently Engan and al. studied a Norwegian modified version of the South African triage scale (SATS), but this study only included children with pediatric medical complaints [15]. In this study more patients received a green triage score and fewer patients received an orange triage score than the present study [15]. Using hospitalization as the sole proxy variable for true medical urgency, they report a higher sensitivity of 74% but a lower specificity of 48%, compared with our current findings on RETTS-p [15].