Factors associated with definitive observation unit care in patients with non-traumatic epistaxis: a retrospective observational study

2.1 Study design and population

This investigation was a retrospective, observational study performed via chart review at a single academic tertiary care center with an annual ED census of 45,000 patients. This study was conducted on patients with non-traumatic epistaxis who visited the ED between January 2016 and December 2020. The sample size was based on a priori power calculation [10, 11]. Adults over 18 years of age were included. Patients with bleeding due to malignant tumors and hematologic diseases were excluded from the study. Patients with missing medical records were also excluded.

2.2 Treatment protocols

All patients in this study were treated similarly. When a patient visited ED, the primary medical staff in the ED performed anterior nasal packing. After checking the vital signs (blood pressure, heart rate, oxygen saturation, body temperature), medical history was taken. Around the same time, blood sampling for the laboratory blood tests (complete blood count, prothrombin time, activated partial thromboplastin time) was performed while obtaining intravenous access. If the bleeding was uncontrolled with anterior nasal packing and vital signs were stable, the patient went to the ear, nose and throat (ENT) department to see an otolaryngologist. In the ENT department, patients underwent diagnostic naso-endoscopy to locate the bleeding focus. After checking the anterior nasal cavity using a rigid zero-degree endoscope, the endoscope was passed into the nose along the nasal floor, checking the appearance of the septum, inferior turbinate and eustachian tube orifice. The endoscope was then reintroduced above the inferior turbinate to check the middle turbinate and middle meatus and was again passed posteriorly to check the nasopharynx. If the bleeding focus was found, the otolaryngologist treated epistaxis using bipolar diathermy and monitored for 30 minutes. If the bleeding stopped, the patient was discharged. However, if we could not find the bleeding focus despite bleeding, or if the bleeding continued despite cauterization, the patients were transferred to DOU for evaluation and treatment.

2.3 Data collection

Data from the ED department and ENT department were collected. Data included patient variables such as age, sex, residence area, route of visit to the ED, state of consciousness, vital signs, comorbidities, drug history (aspirin, warfarin) and Korean triage and acuity scale (KTAS) score. To minimize bias and ensure objectivity, this data was collected by members of the study team, including physicians from each department, who were not directly involved in the patient’s care or treatment decisions. After the data were collected, researchers from each department discussed the appropriateness of the patient data collection and double-checked the treatment methods and outcomes. When visiting the ED, the severity of the patient was classified using the KTAS [12]. Patients were evaluated by classifying them into five levels: level 1, resuscitation; level 2, urgent; level 3, emergency; level 4, less urgent; and level 5, non-urgent. Bleeding at ED visit, revisit of ED or ENT department, laboratory data, bleeding site, electrocauterization, time from onset to ED visit, and stay time in the ED were investigated.

2.4 Statistical analysis

The characteristics of the study population are presented as means with standard deviations for continuous variables and frequencies with percentages for categorical variables. Continuous variables between independent groups were compared using a two-sample t-test with unequal variances. In contrast, categorical variables were compared using the chi-square or Fisher’s exact tests as appropriate. We constructed a series of multiple logistic regression models to determine the relevant risk factors for DOU care. Variables univariately predictive of DOU care at a 0.2 significance level were included. Multicollinearity was assessed, and none of the variables had a variance inflation factor >5. The results of the logistic regression analyses were reported as odds ratios (ORs) with corresponding 95% confidence intervals (CIs). The area under the receiver operating characteristic (AUROC) curve was analyzed to examine the prognostic performance of variables independently associated with DOU care. Dependent ROC curves were compared using the method described by DeLong et al. [13]. Data were analyzed using Stata/SE version 16.1 software for Windows (StataCorp, College Station, TX, USA). A two-sided p-value of 0.05 was considered significant.

3.1 Study population

A total number of 1197 patients with non-traumatic epistaxis were included in the study. Among these patients, 75 with incomplete medical records or blood test results were excluded from the analysis. Finally, 1122 patients were analyzed, with a mean age of 56.4 ± 19.5, 694 males and 429 females (Fig. 1). Among them, 1081 (96.3%) patients were discharged after simple management, and 41 (3.7%) patients needed DOU care. Detailed clinical characteristics are presented in Table 1.

Fig. 1.Flowchart of patients with non-traumatic epistaxis visiting the emergency department (ED) (2016–2020).

3.2 Factors associated with DOU care: univariate factor analysis

When we compared the clinical characteristics between the discharge group and the DOU care group, the univariate analysis showed no significant differences in age, mental status, vital signs or past medical history. However, male patients or patients with warfarin medication tended to need DOU care. In addition, visit route, revisit after prior discharge and delayed visit significantly differed between groups (Table 1). The patients transferred from other medical facilities, revisiting after an initial ED discharge, and patients who delayed visits to the ED for more than 180 minutes resulted in a higher DOU care rate (68.3%, p = 0.008) (Table 1).

Vital signs and mental status had no difference in the univariate analysis. Conversely, KTAS 3–5 patients needed DOU care. In the laboratory tests, the DOU care groups had lower Hb levels (13.5 g/dL vs. 11.1 g/dL, p < 0.001) (Table 1). When patients presented with ambiguous bleeding focus, it led to a significantly higher rate of DOU care than discharge (63.4% vs. 37.3%, p < 0.001). On the other hand, proper management of bleeding spots with electrocauterization led to less DOU care (44.3% vs. 19.5%, p = 0.019).

3.3 Factors in multiple analysis

Multiple logistic regression analyses were performed to identify the factors associated with the DOU care group (Table 2). The model included univariately predictive variables of DOU care at a 0.2 significance level. Male sex (OR = 3.606, p = 0.003), Hypertension (OR = 2.362, p = 0.020), Inter-hospital transfer (OR = 2.358, p = 0.039), Verbal mental status (OR = 29.436, p = 0.035), Time from onset to ED arrival (≥180 min) (OR = 2.451 p = 0.030), and Revisit after prior ED discharge (OR = 8.813, p < 0.001), Hemoglobin level (OR = 0.724, p < 0.001), were significant factors for DOU care.

3.4 Prognostic performance of multiple logistic regression model for predicting DOU care

Fig. 2 shows the results of the ROC curves of the multiple logistic regression models in predicting DOU care. The area under the curve for the multiple logistic regression model for predicting DOU care was 0.870 (95% CI: 0.819–0.922), indicating good prognostic performance.

Fig. 2.The area under the receiver operating characteristic curve in the multiple logistic regression model of non-traumatic epistaxis patient factors for predicting definite observational unit care (area under the curve: 0.870, 95% CI: 0.819–0.922). ROC, receiver operating characteristic.

4.1 Demographic risk factors

Male sex is a risk factor for DOU care in this study. Males required 3.606 (95% CI: 1.537–8.465) times more DOU care than females. This result was concomitant with the previous study. Hadar A et al. [14] and Li et al. [15] studies reported that men were male sex 2.07 (95% CI, 1.59–2.69) and 3.136 (95% CI, 1.50–6.55) times more likely to present with epistaxis than women. Daniell et al. [16] reported that nosebleeds are more common in males than females under the age of 49. Moreover, it has been suggested that the nasal mucosa in women is protected by estrogen, which may be the reason for the observed sex difference [17].

Another factor for DOU care was HTN. The ODD ratio for HTN was 2.358 (1.259–688.436, p = 0.020). Although there have not been any publications on the relationship between HTN and DOU care, many researchers have analyzed the relationship between HTN and epistaxis. A study by Byun et al. [18] showed a 1.47 (95% CI: 1.30–1.66) times higher risk of epistaxis in the hypertensive group. In addition, Lubianca et al. [19]’s and Liao Z et al. [20]’s studies also suggested that epistaxis may be associated with HTN. Our study focused on the relationship between HTN and DOU care. HTN has been considered a significant risk factor for DOU care in non-traumatic epistaxis patients [16, 17].

4.2 Prehospital and ED-related risk factors

Our results showed that patients who revisit ED after the initial discharge are more likely to require DOU care. The odd ratio of this patient group was 8.813 (95% CI: 3.317–23.411). A previous study also showed a similar result. Although the proportion of patients who revisit ED after being discharged from initial treatment was much lower at 4.1% compared to 37% in other studies [21]. It was higher in the DOU group at a rate of 22%. These patients were likely treated with temporary hemostasis, which could have led physicians to make a wrong decision to discharge.

In addition, inter-hospital transferred patients and the length of the time from onset to ED arrival (≥180 min) are also risk factors for DOU care. It is speculated that the probability of DOU care is high as the time is delayed because there are many cases where the treatment is not completed even after the primary treatment was done at another hospital. In addition, among laboratory values, Hb was a significant factor (95% CI: 0.630–0.832, p < 0.001). Our result was concomitant with previous results. Khan et al. [22] reported that the severity of the bleeding was associated with surgical intervention and hospitalization.

Level of consciousness is also a risk factor. Epistaxis patients with altered consciousness tend to receive more DOU care. Although we could not find any previous report which analyzed the correlation between the level of consciousness and DOU care in epistaxis, it is reasonable considering the loss of blood can cause a change in consciousness. In that situation, patients would need DOU care for their safety.

4.3 ENT-related risk factors

Two significant risk factors correlated to ENT care were identified in the univariate analysis: ambiguous bleeding focus and electrocauterization. Patients with ambiguous bleeding focus tend to need more DOU care. Most epistaxis occurs in the anterior site (85–90%), and approximately 10–15% occurs in the posterior site [14]. A previous studies reported that posterior bleeding is more related to hospitalization than anterior bleeding and requires twice as much packing [23, 24], making it more challenging to manage. However, in this study, instead of posterior nasal bleeding, ambiguous bleeding focus was associated with DOU care. In addition, patients with successful electrocautery tend to need less DOU care in univariate analysis. This result is consistent with Kallenbach et al. [25]’s study that reported the hospitalization period was increased by 2.810 (2.047–3.858) times in patients unable to undergo electrocautery. The location of the bleeding focus is vital in determining whether packing, cauterization or topical treatment is the most appropriate.

4.4 Limitation and strength

This study has several limitations. First, in previous studies and the literature, no particular scoring system to classify the severity of non-traumatic epistaxis in the ED were present. In addition, only the complete blood count (CBC), prothrombin time (PT), and activated partial thromboplastin time (aPTT) were mostly evaluated, and chemistry was not tested. Therefore, it was difficult to apply the severity scoring system used in patients with other bleedings such as traumatic brain injury [26]. As with general medical bleeding, it will be necessary to develop a comprehensive scoring system along with the patient’s vital signs and epidemiological factors by adding a chemistry test. Second, the use of pre-hospital emergency medical services was also included in the route of ED direct visits. Due to the nature of our data, the contents of emergency treatment by pre-hospital EMS were not known, and some results may have been affected. Third, this was a retrospective study performed at a single center. Therefore, the findings are not immediately generalizable to the entire population. Further multicenter studies with larger sample sizes and prospective designs are required to substantiate our findings.

Despite these limitations, this study elucidated that epistaxis is one of the leading causes of ED visits. However, the majority of the non-traumatic epistaxis cases are not emergent. In addition to the already-known risk factors, the newly identified risk factors in this study will help to more accurately and efficiently triage non-traumatic epistaxis patients.