Sample data

During the survey period (12/5/2014 to 29/5/2014), 1,345/2,041 emergency department attendees were eligible; 1,212/1,345 surveys were completed. Breathlessness most days over the past month was self-reported by 424/1,212 and acute-on-chronic breathlessness was a reason for presentation in 245/424; 177/245 consented to clinical record review and data were extracted between 30/5/2014 and 31/7/2014. 171 complete cases were analysed.

The 245 with acute-on-chronic breathlessness had an average age of 65 (+/-19) years, 117/245 were men and most described moderate (72/236 mMRC grade 3) or severe (87/236 mMRC grade 4) breathlessness over the previous month. About half (112/237) had breathlessness for over 2 years.

Selection of potential variables associated with emergency department departure status: Unifactorial analysis

Forty-eight independent clinical and demographic variables (extracted from both the survey and clinical record) were assessed for significant (p<0.05) univariate association with emergency department departure status (data available on request). Seven candidate patient characteristic predictors were identified as being very strongly significantly associated with emergency department departure status (see Table 1). Three related to a diagnosis of heart disease. In view of the likely collinearity “Case-record history of heart disease” was chosen because of little difference in statistical significance, a larger odds ratio, and biological plausibility to include ischaemic heart and other diseases as well as heart failure.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Predictor variables associated with admission to hospital in patients presenting to the emergency department due to breathlessness.

https://doi.org/10.1371/journal.pone.0289263.t001

Predictors of emergency department departure status: Multifactorial analysis

The five most strongly correlated included variables were: age, oxygen saturation at presentation, having talked to a specialist doctor about breathlessness, a documented heart condition and the Charlson comorbidity score.

In the final model, the odds of admission were increased with every extra year of age [OR 1.041 (95% CI: 1.016 to 1.066)], having talked to a specialist doctor about breathlessness [9.262 (1.066 to 10.386)] and a known heart condition [4.177 (1.680 to 10.386)] and were decreased with every percentage increase in oxygen saturation [0.826 (0.701 to 0.974)] (see Table 1).

Summary

Our exploratory study showed the odds of hospital admission resulting from emergency department presentation due to acute-on-chronic breathlessness increased with increasing age, decreasing oxygen saturation, having talked to a specialist doctor about breathlessness and having a heart condition.

Strengths and limitations

Our consecutive survey sample, a 90% response rate and minimal missing data are strengths, although seasonal variation was not addressed. However, of the 245 presentations due to acute-on-chronic breathlessness, only 177 consented to clinical record review with 171 complete records obtained introducing selection bias. The study was conducted in one site, but produced findings similar to a multi-site study [7]. These findings are not representative of all patients presenting to the emergency department with acute-on-chronic breathlessness; especially those with very severe breathlessness requiring immediate resuscitation, who lacked capacity to complete the survey.

Our study is exploratory, was not designed to develop a prediction model, and data were collected post conveyance to the ED. As we could only investigate a few strongly associated variables, other important ones may have been missed. We did not include variables relating to treatment or laboratory investigations. However, the tested variables can be assessed in the community and are therefore relevant for conveyance triage. The data were collected six years ago however, the predictors identified remain relevant.

Comparison with existing literature

Around two-thirds of people attending the emergency department with breathlessness are admitted to hospital [1, 7, 9–11]. A disproportionately high admission rate of 86% was also noted in the heart failure substudy of the AANZDEM cohort [12]. Half of our sample were admitted for only one day. Collins and colleagues [13] also noted that over half of those admitted with heart disease from the emergency department were discharged within a brief period. The high proportion of admissions for people with heart disease, compared with lung disease, probably reflects the greater need for hospital-based investigations and intensive medical management.

A number of prediction models exist regarding admission from the emergency department. Older age is a consistently important variable [14–17], consistent with increased likelihood of multi-morbidity and frailty, coupled with lack of social support and greater care needs. The chief complaint is also a recognised predictor, and older adults are more likely to present to the emergency department with breathlessness [18]. The importance of oxygen saturation in triage is well recognised [19], although pulse oximetry may miss some who require admission [20].

Tachycardia is associated with admission in people with chronic heart failure and chronic obstructive pulmonary disease [5, 10, 14], although less helpful in older adults [18], but we did not record this variable. Arrival by ambulance to the emergency department, tachycardia and very severe self-reported breathlessness (numerical rating scale severity of ≥8/10) predicted hospital admission, as did increased distress due to breathlessness [3]. With regard to breathlessness severity, Saracino et al. [3] assessed this on entry to the emergency department, whilst we assessed patients sufficiently stable to be offered the survey, and their breathlessness had settled whilst waiting.

Self-management of acute-on-chronic breathlessness using cognitive [21] and other non-pharmacological [22] interventions reduces hospital admission, and some patients become expert [23, 24]. Interventions to foster a feeling of safety at home and control of breathlessness may reduce presentations to the emergency department; sometimes the mere arrival of a paramedic reduces anxiety-driven acute-on-chronic breathlessness [6]. This is the premise of a current feasibility trial of a paramedic-delivered non-pharmacological intervention (ISRCTN number 80330546) [25].

Implications for research and/or practice

The identified predictors are simple and non-invasive, measurable by community clinicians. Our findings may be useful to aid clinical decision-making regarding conveyance to the emergency department and help prevent attendance by those who settle quickly and would be discharged back home. Additionally, these predictors might be useful at triage in the emergency department to expedite the admission of those most in need. Confirmation of our findings with a larger dataset from community-collected data is needed before being applied in clinical practice. Designs could include quasi-experimental approaches based on propensity score matching to address confounding to test different possible interventions and the usefulness of candidate predictors in clinical practice.