Prognostication for Older People at Emergency Department Presentation

The ageing demography represents both an immense success to society and a challenge to healthcare systems including Emergency Departments (EDs). These are tasked with the provision of acute care and the responsibility to understand patients’ trajectories both in the ED and post-ED and plan accordingly. This requires an understanding of patients’ prognosis for both young and old, both to inform treatment choices, but also to recognize when diagnosis, treatment or cure may be futile. The ageing demography thus represents an opportunity to review the formal prediction tools we use for individualized risk prediction and their applicability to older patients. It also prompts us to consider how such tools may be informed by knowledge about vulnerability of older people by virtue of the frailty paradigm.

Assessment of prognosis in the ED is partly based on vital signs which provide a measure of illness or injury severity. Abnormalities of multiple vital signs are summarized with aggregate scores that provide a measure of illness or injury severity at baseline and aid in detection of subsequent clinical deterioration. However, these aggregate scores are generic to the entire adult population, and the predictions they provide may be ill-suited to the older population resulting in undertriage or delayed treatment.

Frailty is a phenotype that characterizes older people with decreased resistance to illness or injury, whose physiological age is higher than their chronological age suggests. Accordingly, the prevalence of frailty is one in three in the ED, while only one in ten in among community dwelling older people. Frailty is strongly associated with adverse outcomes for older people, and it may supplement the aggregate vital sign score when assessing older patients in the ED. However, only few measures of frailty have seen transdisciplinary acceptance and even fewer are applicable to the ED setting. Additionally, acutely ill older frail patients often need interventions that span from hospital to community and are often subject to fragmented continuity of care. Hence, a frailty measure used in the ED needs to demonstrate reliability when measured by different health care providers involved in cross-sectoral collaborations.

The purpose of this thesis was 1) To examine if a widespread aggregate vital sign score was equally accurate in different age-categories, and to develop and compare the accuracy to an adjusted risk score that retained the potential for generic use in all adults, 2) To test whether a frailty measure with wide transdisciplinary acceptance and validated applicability in the ED was reliable when used by healthcare providers involved in cross-sectoral collaborations, and 3) To develop and examine the accuracy of a risk score that combines frailty and aggregated vital signs.

In a retrospective study of two independent cohorts of ED patients aged 18 or above with vital signs measured at ED presentation, we found that the National Early Warning Score (NEWS), an aggregated vital sign score, was miscalibrated for age categories 18 to 64 and 80 or above, with oppositely directed bias of risk estimates when compared to observed rates of inhospital mortality. For the oldest category, those aged 80 or above, risks were underestimated, and worsened with increasing values of NEWS. When the NEWS was adjusted using chronological age in multivariable logistic regression, the miscalibration was eliminated.

In a cross-sectional study, the Clinical Frailty Scale (CFS), a frailty assessment tool, containing pictographs and case vignettes was translated and adapted to Danish and displayed excellent interrater reliability both within and across four groups of healthcare professionals (10 community nurses, 10 primary care doctors, 10 intensive care doctors and 10 hospital doctors from internal medicine). They rated 15 clinical case vignettes designed to represent all nine levels of the CFS. Inter-rater reliability within the four groups differed, with the lowest lower bound confidence interval suggesting that the intra class correlation (ICC) may be at least good (ICC ≥ 0.6), in a larger population of raters.

In a prospective study of one cohort of older ED patients aged 65 or above with vital signs and CFS measured at ED presentation, we found that aggregation of vital signs using NEWS and frailty status enabled accurate and early identification of patients at increased risk of 30-day mortality. Good fit was indicated in both age groups 65 to 79 and 80 or above, and internal validation indicated limited risk of overfitting.

In conclusion, our understanding of older patients’ prognosis based on vital signs could be improved to better accommodate the ageing demography in EDs. Adjusting the NEWS aggregated vital-sign score with chronological age may enable continued use of one generic vital-sign score on all adult patients to predict inhospital mortality. The CFS was reliable when comparing assessments from four key groups of health care professionals involved in cross-sectoral collaborations and paired with what was previously known about applicability for the ED setting, construct validity to the frailty index and transdisciplinary acceptance, the CFS seems suitable for use in EDs. Finally, it was possible to accurately predict 30-day mortality for older patients at ED presentation using CFS in combination with NEWS. This operationalized early assessement of the clinical interaction between the severity of the illness or injury and the ability to resist and recover in older patients in EDs.