Study protocol for a multicentre prospective cohort study to identify predictors of adverse outcome in older medical emergency department patients (the Risk Stratification in the Emergency Department in Acutely Ill Older Patients (RISE UP) study)

Older patients (≥65 years of age) constitute an increasing population in emergency departments (EDs). They experience more adverse outcomes than younger ED patients [1‐3] as their ED visits are often highly urgent and followed by hospitalization, Intensive Care Unit (ICU) admission, readmission, and mortality (up to 10% within 3 months). In older ED patients with internal medicine (medical) problems, mortality is even higher (23.8% in 3 months) [4].

Risk stratification can be used to identify ED patients who are at high risk of adverse outcomes. This may lead to improvement in medical care and outcome by early start of interventions [5]. Several risk stratification scores have been developed for the older ED population, such as the Identification of Seniors at Risk (ISAR) [6] or Triage Risk Stratification Tool (TRST) [7]. Unfortunately, these scores do not accurately identify those who experience adverse outcomes (areas under the curve (AUCs) range from 0.59–0.74) [8‐10]. Triage systems (e.g. Manchester Triage System (MTS) [11]) are also used in the ED population but they tend to undertriage older patients [12]. Furthermore, risk stratification scores either applicable to the general ED population (e.g. Acute Physiology and Chronic Health Evaluation II (APACHE II) score [13]) or to patients with specific diseases (e.g. Abbreviated Mortality ED Sepsis (abbMEDS) score [14] for sepsis) are used. However, these are not validated in the older ED population.

It may be possible that older patients at risk of adverse outcomes can be identified by assessing the disease perception of patients (or caregivers) or clinical intuition or impression of physicians and/or nurses. Indeed, both disease perception and clinical impression are shown to be associated with mortality and morbidity [15‐20]. Unfortunately, most studies were conducted in other clinical settings than the ED (e.g. admission units and ICUs) [4, 17, 20] and in younger patients [17, 18, 20]. A second method to assess clinical impression is to ask the ‘surprise question’ (SQ): ‘Would I be surprised if this patient died within the next 12 months?’. The SQ has been studied in cancer and renal failure patients but its diagnostic accuracy for one-year mortality varies considerably [21]. The predictive value of the SQ for short-term mortality in older medical ED patients is unknown.

Laboratory tests may also be useful for identification of older patients at risk of adverse outcomes [4, 22]. In addition, non-routine laboratory tests, such as lactate, high-sensitivity cardiac troponin T (hs-cTnT), N-terminal pro-B-type natriuretic peptide (NT-pro-BNP), procalcitonin (PCT) and d-dimer, may be valuable predictors of adverse outcome as well [23‐27]. Since these tests are indicators of serious conditions and diseases (e.g. tissue hypoperfusion, myocardial injury, bacterial infection and thromboembolism) and often present in the older ED population, we hypothesize that these tests can be useful as predictors. Until now, most studies regarding the predictive value of these tests were performed retrospectively [24, 28‐33] or in selected, mostly younger, patients [30‐38].

We hypothesize that in the early stage of an ED visit, when important treatment decisions have to be made, several factors can predict adverse outcomes. The aims of this multicentre, prospective study are to 1) identify early predictors of adverse outcome in older ED patients, and 2) develop a clinical prediction model for 30-day mortality.

Early risk stratification at the ED is extremely important to optimise treatment and improve outcomes in acutely ill older patients. To identify older patients with increased risk of adverse outcomes in an early stage, we need accurate predictors.

In the past decades several studies identified risk factors and predictors of adverse outcomes in the older ED population [4, 8, 49]. However, most of these studies were conducted in an unselected population of older patients, leading to conflicting results. For our study we chose medical ED patients because we assume that this group of patients represent a large group of patients who are highly at risk of adverse outcomes. Furthermore, we chose to evaluate more endpoints (including composite endpoints) because some adverse outcomes exclude others (i.e. dying in-hospital will prevent readmission). We are convinced that for quantification of the risk of adverse outcomes in the older population, more than one endpoint is needed for a reliable clinical appraisal. We consider patients older than 65 years to be old and we chose this cut off based on other well-known screening instruments such as the ISAR [6], the ISAR-HP [39] and TRST [7].

Development and validation of our prediction model for 30-day mortality will be according to the Stiell criteria [46]. Since reliable tools to predict short-term mortality are lacking there is need for such a prediction model (criterion 1). Before conduct of this study, internists, gastroenterologists and geriatricians were consulted on the need of a clinical tool, preferred outcomes to be studied and potentially meaningful predictors. The model will be derived according to methodologic standards and is intended to be easily implemented in routine ED care (criterion 2). Most of the predictors we will select will resemble the morbid state rather that the premorbid state because we hypothesize that that the severity of the disease for which the patient visits the ED (morbid state) may be more important than the premorbid state for prediction of short-term mortality. We aim to externally validate our prediction model in a different ED population (criterion 3). Once we have succeeded in developing and externally validating an accurate model, we intend to implement it into clinical practice by offering an online calculator, which may also be incorporated into a electronical file management system (criterion 4).

The main strengths of the RISE UP study are, in our opinion, its prospective multicentre study design in an ED setting and the use of composite endpoints. We aim to identify predictors of adverse outcomes in an early stage of presentation, in the ED setting, when important decisions need to be made. We will not only evaluate the predictive value of the clinical impression of both the patient, nurse and physician, but also that of routine and non-routine laboratory tests. A possible limitation is that we expect that we cannot include all potential candidates, due to crowding of patients at the ED or other logistic problems. Therefore, we will perform an additional analysis to investigate possible selection bias. In addition, this study will include internal medicine and gastroenterology patients only. If this study yields an accurate model, that model will have to be tested in the overall older ED population. Furthermore, we intend to implement the model into clinical practice by use of an online calculator. Possibly, this may also be incorporated into an electronical file management system or a mobile App.

In summary, the RISE UP study is a prospective multicentre cohort study that aims to identify predictors of adverse outcomes in older medical ED patients. The goal of this study is to develop a practical, feasible tool to identify older ED patients with an increased risk of adverse outcomes in an early stage, in order to improve their care in the future.