Introduction

Patient safety, defined as “freedom from accidental injury”, [1] has been identified as an international priority [1], [2]. Adverse events (AE), defined here as unintended harms resulting from the care and services provided to the patient [3], represent a significant threat to patient safety and public health. Studies conducted in multiple countries have reported a prevalence of AE among hospitalized patients ranging from 2.9% to 16.6%, with 36.9% to 51% of events considered preventable [4], [5], [6], [7], [8].

The emergency department (ED) is considered particularly high risk for AE. Reasons include high patient volume, patient acuity and complexity, a work environment characterized by time constraints, multiple interruptions and disrupted sleep cycles for health care workers, as well as factors such as high risk diagnostic and therapeutic interventions and variable levels of physician training [8], [9], [10], [11]. While population based estimates from the seminal Harvard Medical Practice Study (HMPS) suggest that about 3% of AE occur in the emergency department (ED), this is likely an underestimation given that the study included only hospitalized patients [8] and only 9.5% of patients who visit an ED are admitted [12]. Research on AE in multiple care settings, has identified differences in the prevalence, types, severity and preventability of AE [4], [13], [14], [15], [16], [17]. This discrepancy in findings highlights the importance of setting specific research on AE. As a result of concerns that the ED is a high risk environment for patient safety and that setting specific research is important, we conducted a systematic review to summarize the best available evidence regarding the prevalence, preventability, severity and types of AE related to ED care.

Methods

The systematic review followed a prospective protocol that was developed a priori.

Assessment of Methodological Quality

We rated methodological quality of included studies using the Newcastle-Ottawa Quality Assessment Scale (NOS) [18]. We felt that the NOS did not adequately assess important aspects of AE measurement and reporting, including methods for identifying AE, methods to establish causality between health care and outcome, and methods for establishing severity and preventability of AE. As a result we also utilized a tool adapted from a previous systematic review on adverse drug reactions, referred to hereafter as the Smyth adapted AE tool (Table 1) [19]. Two reviewers (AS, AP or AW) independently assessed the methodological quality of each study using the NOS and the Smyth adapted AE tool.

Download:

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

Table 1. Summary of Quality Assessments.

https://doi.org/10.1371/journal.pone.0074214.t001

Results

Study Selection

The database search identified 14,454 potentially relevant citations with 11,623 citations after duplicate removal (Figure 1: PRISMA Diagram). Screening of titles and abstracts excluded 11,187 citations. We assessed 436 full text articles for eligibility and nine publications met the inclusion criteria. One article was identified in the hand search. A total of ten publications, representing eight studies (two studies reported results in two publications each [20], [21], [22], [23]) were included in the review.

Download:

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

Figure 1. PRISMA Diagram.

https://doi.org/10.1371/journal.pone.0074214.g001

Study Characteristics

The included articles (Table 2 and Table S3) were prospective, observational studies: seven prospective cohort studies [9], [10], [20], [21], [24], [25], [26], [27] and one before-and-after interventional design [22], [23], [27]. Methodological quality was low to moderate with weaknesses in study group comparability, follow-up, and outcome ascertainment and reporting (Table 1). Particular weaknesses in outcome ascertainment and reporting included failure to clearly report methods used to determine AE causality, preventability and severity. Seven studies took place in urban, tertiary care academic hospitals in either Canada, the United States or Australia [9], [10], [21], [24], [25], [26], [27] and one study took place in a rural base hospital in Australia [22], [23]. A total of 38,702 patients were included in the eight studies, with a median of 1219 patients (range 201 [25] to 20,500 [22], [23]). Four studies included only adults [10], [24], [25], [26] and four enrolled adults and children [9], [20], [21], [22], [23], [27].

Download:

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

Table 2. Summary of Studies.

https://doi.org/10.1371/journal.pone.0074214.t002

Discussion

Previous systematic reviews have been conducted on the incidence and nature of in-hospital AE [28] and patient safety in the EMS/pre-hospital setting [29], however, to the best of our knowledge this is the first systematic review on AE related to ED care. The proportion of patients with AE related to ED care varied widely between studies. The proportion of patients with AE was higher in studies that utilized flagged outcomes and/or chart review [10], [20], [21], [22], [23], [24], [25] compared to those that used voluntary reporting or active solicitation of error reports [9], [26], [27]. Previous research has suggested that voluntary reporting systems detect less than 10% of AE [4], [17], [30]. The two highest quality studies in our review (based on both the NOS and the Smyth adapted AE tool) reported the highest proportions of ED patients with AE at 6% [24] and 8.5% (5% attributable to ED care) [10].

Previous research has identified particular risk factors for AE, including patient age and health care setting. For example, the pediatric population seen in the ED has been identified as particularly vulnerable to harm from medical errors [31], [32], [33] and research among hospitalized children has shown a high prevalence of safety events, comparable to that among hospitalized adults [34]. While four of the studies in this review enrolled both pediatric and adult patients, there were no pediatric specific studies and minimal age specific data reported. Patient safety studies in hospitalized patients have also identified older patients (≥65) as high risk for AE [4], [7]. Recent ED specific research has suggested that in elderly patients a prolonged ED length of stay is associated with an increased risk of an in-hospital AE [35]. However, little is known about the occurrence of AE in the ED among elderly patients. Despite the finding of older age as risk factor in one of the studies reviewed [22], there was insufficient information overall to present AE data based on age.

With respect to health care setting, even after adjustment for co-morbid conditions, previous work in hospitalized patients has demonstrated a higher number of AE in teaching hospitals compared to community hospitals for both adults and children [4], [14]. Of note, in the pediatric setting, although the overall proportion of AE was higher in teaching hospitals, the proportion of AE attributable to the ED was higher in community hospitals (15.9% (n = 69) vs 5.7% (n = 210)) [14]. Although one study (two publications) [22], [23] in our review was conducted in a community hospital, the substantial variation in methods and reporting between studies prevented any comparisons based on study setting.

The results of this review suggest that a substantial proportion of AE related to ED care may be preventable. The preventability of AE occurring in the ED (36% to 71%) is at least comparable to that reported in previous studies of hospitalized patients (35% to 51%) [4], [6], [7], [8]. Previous work in hospitalized patients demonstrated that the largest proportion of AE were operative [4], [14], [28], [36] and medication related [4], [8]. In the ED setting, management, diagnosis and medication related events were common. One of the studies in this review also identified differences in the preventability and types of AE among discharged and admitted patients [10]. A greater proportion of AE were preventable among the discharged population (71.4%; n = 15) than among the admitted population (40.9%; n = 9). Among discharged patients, management issues (47.6%; n = 10), diagnostic issues (33.3%; n = 7), and unsafe dispositions decisions (19%; n = 4) were the most common causes of AE. In comparison, admitted patients more commonly suffered from procedural complications (50%; n = 11), management issues (36.4%; n = 8) and medication adverse effects (27.2%; n = 6). These findings suggest that the types of AE that occur in the ED may be different than in hospitalized patients, and different between patients who are discharged from the ED and those who are admitted. These results support the need for setting specific research and intervention.

Conclusions

The methodological flaws of the individual studies and the variability in definition of AE among the included studies precludes broad, generalized conclusions with respect to the prevalence, preventability and severity of AE related to ED care. The lack of large, high quality studies on AE in the high risk ED setting is, however, in itself an important finding and highlights the need for further research in this area. In particular, further research is required to better understand specific risk factors for AE in the ED, including patient age and ED setting. Future studies in this area should include an adequate duration of follow-up to capture AE attributable to ED care that occur after discharge, and should use rigorous and standardized outcome reporting, especially with respect to AE causality, preventability and severity.

Supporting Information

Acknowledgments

The authors would like to acknowledge Andrea Milne who conducted the literature search, Teodora Radisic who retrieved articles and Denise Thomson who coordinated the project.