Risk Factors for Revisiting After Emergency Department Discharge: a Longitudinal Retrospective Chart Review


 Background: Emergency department (ED) patients are often discharged after treatment for minor illnesses. However, many of these patients revisit the ED. This study aimed to describe ED revisit rate and investigate the risk factors for 3-day/30-day ED revisit. Methods: A longitudinal retrospective chart review was used to extract data regarding predisposing factors (e.g. gender, age), enabling factors (e.g. economic status, ambulance use) and need factors (e.g. diagnosis at ED discharge, history of diseases), from the medical charts of patients discharged from an ED in Tokyo, Japan, from February to December 2013. Multiple logistic regression analyses were used to predict 3-day/30-day ED revisit based on initial visit information. Results: During the studied period, 8,754 patients were discharged from the ED, of whom 48.8% were men, and the mean age was 39.7 years. Of these, 338 (3.9%) revisited the ED within three days and 682 (7.8%) revisited the ED within 30 days of discharge. In the multiple logistic regression analysis, risk factors for 3-day ED revisits were: having a history of same hospital use, arriving by ambulance, having a common problem (e.g. fever), and history of asthma. Risk factors for 30-day ED revisits were using health care services and having a history of same hospital use. Conclusions: In this study, the rate of ED revisit was less than that reported by other studies, which might be because most patients included in this study usually visited the hospital. Our results show that the predictive factors for 3-day/30-day ED revisit in Japan were similar to those in other countries. Patients with the abovementioned factors should be carefully screened on their initial ED visit, and some follow-up care in clinical and community settings should be arranged.

This study was conducted in a university hospital, in a critical care center ED in Tokyo, Japan. The hospital was a special function hospital with 1,163 general beds, which received 16,552 patients in 2018 [11]. Participants Inclusion criteria were: 1) ED visit from February 1 to December 31, 2013; 2) no ED hospital visit during previous month; 3) discharged home or to another homecare setting (but not other hospitals) after the initial ED visit.
Recruitment occurred in the following ve steps: 1) collecting patient ID (patient unit data) by ED visit system from January 1, 2013 to March 31, 2014; 2) matching the data to Patient ID; 3) assigning an anonymized study ID to each patient; 4) identifying the initial ED visit date of each patient; 5) excluding patients whose initial ED visit was not in the selected study range.
Outcome measurements ED revisit was measured in three categories: patients who revisited the study hospital ED within three days were set to the "A: 3-day" group, 4-30 days ED discharge patients were placed in the "B: 30-day" group, and patients not revisiting within 30 days formed the "C: no revisit" group.

Variables & Data cleaning methods
We organized the items based on Andersen's [12] behavioral model of health services use: predisposing characteristics, demographic factors, social factors, health beliefs, enabling characteristics; nancing, economic status, organization, need characteristics, perceived need and evaluated need, including the Japan Triage and Acuity Scale (JTAS) [13]. An additional table le shows complete list of codes and data cleaning methods in more detail [see Additional le 1].

Data sources, Data access, & Linkage
We collected the medical chart information of ED patients through a "ED patient visit system" which included an ED visit date and patient ID. Using the patient ID, we retrieved 11 CSV les by the Department of Healthcare Information Management of the hospital. The data set included three databases: 1) the ED patient visit system: age, gender, ED visit information, diagnosis, outcome, among others; 2) the electronic medical record system describing data by ED doctors; 3) the ordering system, including medical examination, injection ordered and treatment. All data were linked through patient ID.

Bias & Study size
There was no bias in this study's outcome "ED revisit," because of the complete data surveyed during study period. On the other hand, information quality could vary because of input differences from ED doctors.

Statistical methods
Descriptive statistics, frequency counts, percentage rates, means, standard deviation [SD], and range were calculated for each variable at the time of the initial ED visit and transition of ED revisits within 90 days after ED discharge. ED revisit cumulative incidence was described by frequency count until 90 days in those discharged from the ED after the initial visit. We compared the three groups of data at the time of initial ED visit by chi-square tests, and one-way ANOVAs. A multiple logistic regression analysis (backward elimination) was used for initial ED visit to examine the risk factor of group A, and to examine the characteristics of B, compared with C. The p-value <0.01 was de ned as signi cant, because of big sample size and many analyses. All data were analyzed using IBM SPSS Statistics 24 (IBM Corp., Armonk, NY, USA).

Participants
The ow of participants included in this study is shown in Figure 1. There were 17,437 cases (13,180 patients) who visited the ED from February to December 2013.

Figure. 1 Flow of participants
Patients who were hospitalized, transferred to another hospital, or died were excluded. We also excluded patients whose initial ED visit occurred in January 2013, and nally, analyzed 11,322 cases (8,754 patients) who were discharged from the ED after an initial visit. Table 1 shows the characteristics of ED patients in this study.  Figure. 2 Transition of emergency department revisits within 90 days after discharge Among ED revisit patients, patients who revisited within three days may differ from those who revisited after four days, given how groups were disposed. Table 2 shows the results of comparison between the three groups. Group C had less women (50.8%, p = 0.04) than A or B, whereas B had an older mean age of 47.2 years (p <0.01) and more older adults (34.6%, p <0.01) than A or C. 1. a) Chi-squared test b) One-way ANOVA ICD-10: I Certain infectious and parasitic diseases, II Neoplasms, III Diseases of the blood and blood-forming organs and certain disorders involving the immune mechanism, IV Endocrine, nutritional, and metabolic diseases, V Mental and behavioral disorders, VI Diseases of the nervous system, VII Diseases of the eye and adnexa, VIII Diseases of the ear and mastoid process, IX Diseases of the circulatory system, X Diseases of the respiratory system, XI Diseases of the digestive system, XII Diseases of the skin and subcutaneous tissue, XIII Diseases of the musculoskeletal system and connective tissue, XIV Diseases of the genitourinary system, XV Pregnancy, childbirth, and the puerperium, XVI Certain conditions originating in the perinatal period, XVII Congenital malformations, deformations, and chromosomal abnormalities, XVIII Symptoms, signs and abnormal clinical and laboratory ndings not elsewhere classi ed, XIX Injury, poisoning, and certain other consequences of external causes, XXI Factors in uencing health status and contact with health services

The risk factors for 3-day ED revisit
To examine the risk factor of group A, a multiple logistic regression analysis was conducted in those discharged from the ED after the initial visit ( Table 3). The following factors were related to more frequent 3-day ED revisit after discharge from the same ED: having a history of same hospital use (odds ratio [OR] = 1.60, 95% con dence interval [CI] :1.20-2.13), arriving at the ED by ambulance (OR = 1.62, CI: 1.23-2.12), having a common problem (e.g., fever, problems with medical devices, and consultations with referrals) (OR = 2.02, CI: 1.45-2.81), and history of asthma (OR = 1.77, CI: 1.25-2.49). Note: Adjusted by ED month visit and diagnosis of ED discharge.

The risk factors for 30-day ED revisit
To examine the risk factor of group B, a multiple logistic regression analysis was conducted for those who were discharged from the ED after the initial visit ( Table 4). The following factors were related to more frequent 30-day ED revisit after discharge from the same ED: having a history of same hospital use (OR = 2.88, CI: 2.05-4.04) and using health care services (OR = 2.48, CI: 1.73-3.55).

Discussion
This study investigated the rate of ED revisits its related factors in Japan for the general population among patients discharged from a university hospital ED. Among 8,754 patients who were discharged from February to December 2013. The 3-day ED revisit rate was 3.9%. Risk factors of 3-day ED revisit were a history of same hospital use, arriving at the ED by ambulance, presenting a common problem (e.g., fever, problems with medical devices, and consultations with referrals), and history of asthma. And the 30-day ED revisit rate was 7.8%. The characteristics of 30-day ED revisit group compared to no revisit group were having a history of same hospital use and using health care services.

ED revisit rate
The ED revisit rate in this study was 0.6% on the same day, 3.92% within three days, 7.8% within 30 days, and 11.7% within 90 days. In previous studies, ED revisit rate was 3.1%-13.1% within 3 days, 15.8%-22.4% within 30 days, 22%-34.0% within 90 days [4,14]. Overall, the ED revisit rate in this study was lower than those commonly reported.
A previous study reported that the small size of the ED and the absence of a social worker were predictive factors for ED revisits [15]. In addition, according to a previous study, among patients who revisited the ED within three days after discharge, 32% visited a different ED from their initial visit [16].
The low rate of ED revisits might be affected by the fact that the hospital was a special function hospital, so some ED discharge patients might visit other hospitals, general outpatient clinics, or might have died. Additionally, since it was a university hospital, many specialists available at any given time might have collaborated between departments.
According to previous studies, the revisit rate within three days was considered to be an indicator of the quality of emergency medical care in other countries [4]. It is important to consider a 3-day ED revisit rate as a quality indicator of ED care in Japan, and it is necessary to explore the possibility that a 30-day ED revisit rate might be a quality indicator of collaboration of the ED with other community primary care providers.
The risk factors for 30-day ED revisit after discharge This study identi ed risk factors for 3-day ED revisit, and the characteristics of the 30-day ED revisit group compared with no revisit group by using data at the time of initial ED visit.
In the Comprehensive Care System in Japan, the number of patients with high medical needs at home is increasing [17]. For an aging society, transitional care is generally given to inpatients in Japan [17]. The current recommended steps of transitional care are: (1) screening assessment, (2) deciding on the appropriate level of support, and (3) service adjustment to inpatients upon hospitalization [18]. Thus, discharged patients may also be considered eligible for such transitional care, but there is no such exact system in Japan. Therefore, it is strongly recommended taking steps to provide such care for ED patients in order to identify the patient with the information at the time of initial ED visit to involve them in some intervention.
In what follows, we discuss the data based on Andersen's behavioral model [12].

Predisposing Characteristics
In this study, having a history of same hospital use was more frequently associated with 3-day ED revisit and 30-day ED revisit. As a background, the medical insurance system guarantees free access for patients to any domestic hospitals, specialists, and clinics in Japan [19]. According to a patient behavioral survey, patients select hospitals for reasons such as "introduction by a doctor," "good accessibility for transportation," and "offering highly specialized medical care" [20]. When patients have a history of same hospital use, they can easily choose to visit the same hospital's ED, even if the symptoms are unrelated to their chronic diseases. Therefore, even though patients are more inclined to visit larger hospitals, they should be encouraged to visit the primary emergency institutions of their local government rst and be ensured that they will be referred to larger hospitals if required.

Enabling Characteristics
In this study, arriving at the ED by ambulance was associated with higher 3-day ED revisit. A previous study found that arriving at the ED by ambulance at the initial ED visit, can increase revisit [21]. One possible reason for this is that the experience of ED visit with the support of an ambulance lowered the psychological hurdle to using ED.
Using health care services (e.g., long-term insurance services and disability certi cate) was also associated with higher 30-day ED revisit. A previous study reported that patients who need home care services tend to revisit the ED [22]. Activities of daily living (ADL) in these patients are often low, they are highly dependent on medical care and use long-term care insurances. A handicapped person should be put in the highrisk group for ED revisits. Moreover, these people have contact points with some home care providers, who can be trained to be more attentive and perform regular follow-ups for patients to discover the symptoms earlier. This seems very important for ensuring proper communication channels between the ED staff and home care providers.

Need Characteristics
In this study, presentation during the ED visit with common problems (e.g., fever, problems with medical devices, and consultations with referrals) was highly associated with 3-day ED revisit. In previous studies, fever was considered a risk factor for ED revisits [23]. According to a previous medical records survey in Japan, about half of the pediatric patients visited ED for fever, 20% of which had visited another pediatrician or another hospital's ED [24]. We observed that fever was a risk factor even after adjusting for age. As fever is a frequentlyoccurring symptom in the ED and many patients are discharged home with fever, it is necessary for ED staff to explain to patients and their families how to manage a fever at home and when to seek further medical advice.
Having a history of asthma was also signi cantly associated with more 3-day ED revisit. In a previous study, asthmatic attacks were considered to be predictors of ED revisit within 72 hours [25]. In this study, the patients not only had asthmatic attack at initial ED revisit, but also other diagnosis. Patients with a history of asthma in this study might have been part of the group of patients who would more frequently visit ED.

Limitations
There are two wider limitations to this study. First, this was a retrospective study of old data, implying a limit to the accuracy of data and that it might not be applicable to the current situation of these years. Second, this was a single-center study, thus, the result may not be generalized for nationwide or rural areas. Future studies should be designed to follow-up of the actual situation of ED revisits in community based, including other hospitals.

Contribution to Emergency Nursing Practice
The current state of scienti c knowledge on patients' risk factors for 3-day/30-day ED revisit. The main ndings of this research revealed that after ED discharge in Japan, factors related to a 3-day ED revisit were: history of same hospital use, arriving by ambulance, having a common problem, or history of asthma. The factors related to 30-day ED revisit were: using health care services and having a history of same hospital use. Key implications for emergency nursing practice from this research are related to the importance of taking care of patients with these factors during their initial ED visit in clinical and community settings to reduce the high rates of ED revisits.

Further Study
As revealed in this study, ED revisit patients had various predisposing, enabling, and need-based characteristics. Hence, in future studies, it would be necessary to discover a pattern for reasons for ED revisit among patients, clarifying associated factors with each ED revisit reason.
ED revisit rate should be further assessed by analyzing medical claim data to link them to individuals and clarify patient medical/homecare services use behavior.

Figure 2
Transition of emergency department revisits within 90 days after discharge

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.