Our study used a validated electronic health record system with a large sample size to determine whether the TTAS accurately triages the growing number of older patients presenting to EDs or whether a new or adjusted system is required. The predictions of the TTAS regarding in-hospital mortality were less accurate for older adults than for younger adults. Moreover, frailty was independently and positively associated with in-hospital mortality across TTAS levels. However, adding mobility status as a frailty indicator to the triage system did not lead to a considerable improvement in the ability of the TTAS to predict in-hospital mortality in older adults.
Triage systems are typically designed to screen all patients evenly; in other words, younger and older adults undergo the same screening process. This is potentially problematic because age is a potential modifier in triage; it is strongly associated with greater mortality risk in triage patients  and independently predicts hospital admission in patients aged > 65 years who are triaged as level 5 . Herein, the in-hospital mortality rate increased significantly with age. We also observed that the young adults had the lowest in-hospital mortality rate of 6.3%, which was almost one-fourth of the average in-hospital mortality rate of patients triaged as level 1 under the TTAS. This may be due to the faster recovery rates of this age group following immediate management.
Frailty greatly differs between older and younger patients and is a problematic manifestation of population ageing. It is a consequence of the age-related deterioration of multiple physiological systems, which renders patients vulnerable to sudden changes in health status triggered by relatively minor stressors . Blomaard et al. reported that the 30-day mortality rate in older high-risk individuals was three times higher than in low-risk individuals, suggesting that a modifier for the screening of older adults would improve the accuracy with which this population is triaged . By contrast, Mowbray et al. noted that triage acuity and frailty were independent but complementary measures with distinct clinical outcomes in patients aged ≥75 years . In the present study, mobility status as a frailty indicator was independently associated with in-hospital mortality but did not enhance the ability of the TTAS to predict in-hospital mortality in older patients. The present study used mobility as an indicator for the rapid determination of frailty status. Future studies may use more comprehensive frailty assessments such as the clinical frailty scale [21, 22] or geriatric screening tools such as the Acutely Presenting Older Patient screener , and to confirm their usefulness as triage tools in the ED.
As expected in our study, the older the patients, the higher the medical resource utilisation. However, Unlike the younger population, the elders often have more comorbidities and are usually weaker. Older patients and or their caregivers tend to have DNR or choose not to undergo unwanted or life-extending treatments, and this is a possible reason for the observation. In addition, the older adults cannot explain or express themselves clearly than young age group when coming into a critical illness, leading to the problem of under-diagnose and the under-estimate. Thus, current five-level triage system may have the problem of under-triage in the older populatuon. Therefore, we propose using mobility status as frailty indicator to improve the five-level triage system to better evaluate the older adults in EDs.
The capacity of triage systems to predict in-hospital mortality appears to decline with patient age. Such systems include the Japan Acuity and Triage Scale (AUROC: 0.74, 0.69, and 0.66 in older adults aged 65–74, 75–84, and ≥ 85 years, respectively)  and the Manchester triage system (AUROC: 0.79 and 0.71 in younger adults and older adults, respectively) . In our study, the AUROC corresponding to the TTAS’s prediction of in-hospital mortality was 0.84 in all ED adult patients, 0.86 in the young adults, and 0.79 in the older and very old adults. Although the TTAS presented an acceptable ability to predict this outcome (AUROC > 0.70), the AUROC decreased with patient age, suggesting that the room for improvement remains with regard to the prediction accuracy of the TTAS for older patients. We therefore recommend a revision to the current TTAS to triage older patients more accurately. However, whether the addition of a frailty modifier (as in the case of the CTAS) is advisable remains to be determined, as does which frailty assessment should be incorporated.
Our study has some limitations. First, the data were extracted from a single medical centre. Although our centre is one of the largest in Taiwan and has a high ED volume, our findings may not be generalisable to other settings, regions, or countries. Second, we acknowledge that the primary goal of triage is to determine which patients require urgent access to care rather than to predict in-hospital mortality. Nonetheless, using in-hospital mortality as a surrogate marker enables researchers to evaluate the possibility of undertriage. Third, aside from mobility status, we did not consider other determinants of frailty. Moreover, frailty was not examined by form or severity (e.g., chronic or acute); thus, bias may have been introduced. However, because triage in the ED aims to prioritise incoming patients and assign them a triage acuity level through a brief, focused assessment, a thorough frailty screening may not be ideal in such as context. Fourth, some data were missing; the available case analysis method was employed to handle the data not missing at random. Finally, the significant statistical differences across groups could be derived from the large sample size of the study; their clinical significance may be debatable.