Does off-hours endoscopic hemostasis affect outcomes of nonvariceal upper gastrointestinal bleeding?

In this study, we used a retrospective study design, which was conducted in the collection in three hospitals of Tongji Hospital, Tongji Medical College, Huazhong University of Science and Technology, a large tertiary teaching hospital in China. The medical record management system was used to extract medical records of patients conformed to the following standard from 1 January 2016 to 31 January 2020. The inclusion criteria are as follows: patients diagnosed as NVUGIB; received endoscopic hemostasis treatment; after receiving endoscopic hemostasis, patients were hospitalized and managed with conventional therapies, including nil oral intake, intravenous fluids and acid secretion inhibitors. Those patients would be excluded: those who did not cooperate with treatment; those whose medical records were not complete and cannot be analyzed. This research was carried out following the guidelines of the Ethics Committee of Tongji Medical College, Huazhong University of Science and Technology.

Electronic medical records were employed to obtain the data. Data on baseline characteristics (such as age, sex, history of UGIB), clinical laboratory findings and outcomes of patients were recorded. Patients who received oral anticoagulants, antiplatelet drugs and NSAID were classified as medicine used. Comorbidities were reviewed from medical charts and included diabetes, hypertension, renal failure (creatinine >1.5 mg/dl), end-stage renal disease on dialysis, liver cirrhosis, coronary heart disease, cerebrovascular disease and neoplasms. The primary outcome was the composite outcome which included death, rebleeding after initial endoscopy, need for emergency surgery, angiographic embolization or endoscopic reintervention [19]. Secondary outcomes included length of hospital stay, the need for blood transfusion, volume of blood transfusion, postendoscopy adverse events (aspiration pneumonia, urinary tract infection, sepsis, cerebral infarction, hepatic failure, cardiac failure, renal failure). Rebleeding was defined as bleeding within 1 week that required secondary hemostasis or was associated with hematemesis; melena with an overt decrease in hemoglobin over 2 mg/dl; or endoscopic findings of recent bleeding, such as spurting, oozing or adherent clot [20,21]. Patients who received endoscopic hemostasis during regular-hours (08:00–17:30 on weekdays) were defined as the regular-hours group, and those who received endoscopic hemostasis during off-hours (17:30–8:00 the next day on weekdays or the whole day of weekends/holidays) were classified as off-hours group. The Glasgow–Blatchford Score (GBS) is based on the systolic blood pressure, pulse rate, hemoglobin, serum urea levels, whether the patient presented with melena or syncope and the presence or absence of hepatic disease or cardiac failure. Patients with GBS score of 12 or higher are thought to be at higher risk for further bleeding or death [22]. Patients with GBS<12 were defined as the lower-risk group, and patients with GBS ≥12 were defined as the higher-risk group. The predictors of the composite outcome in both groups were analyzed individually.

Data analysis was performed using SPSS 22.0 (IBM, NY, USA). Continuous variables are expressed as mean ± standard deviation, and categorical data are presented as absolute numbers and percent frequencies. Student t-test was performed to analyze continuous variables. Chi-square test or Fisher exact test were used for the analysis of categorical variables. A logistic regression model was used to analyze the effect of categorical variables and adjust for potential confounders. The results of logistic regression analyses were summarized by estimating the odds ratio (OR) and respective 95% CI. A two-tailed p-value < 0.05 was considered statistically significant.

A total of 301 patients was included in the retrospective study. Of these patients, 204 patients were classified as the regular-hours group, and 97 patients were included in the off-hours group. Parameters including sex, age, smoking, drinking, history of prescribed medicine, comorbidities, history of UGIB, heart rate, hemoglobin and GBS scores were not different between the two groups (p > 0.05), while patients in the off-hours group tended to have lower systolic pressure (112.3 ± 21.4 vs 120.7 ± 20.5 mmHg, p = 0.001), lower platelet count (163.8 ± 76.7 vs 185.4 ± 70.1 × 10⁹/l, p = 0.016), longer prothrombin time (PT) (16.3 ± 5.2 vs 15.2 ± 2.5s, p = 0.014), higher international normalized ratio (INR) (1.3 ± 0.6 vs 1.2 ± 0.3, p = 0.036), and higher AIMS65 score (1 [0, 2] versus 0 [0,1], p = 0.029). The etiology of NVUGIB in our study included peptic ulcer, Mallory–Weiss syndrome, vascular malformation, esophageal erosions/ulcers, diverticular bleeding and postoperative bleeding. Among those with NVUGIB, the most common cause was peptic ulcer (72.8%). More patients presented with peptic ulcers in the regular-hours group than off-hours group (76.5 vs 64.9%; p = 0.036). Treatment modalities, including norepinephrine spraying, hemostatic clips, epinephrine injection, were comparable between both groups (all p < 0.05) (Table 1).

It is worth noting that the off-hours group had a higher rate of the composite outcome (22.7 vs 9.8%; p = 0.003), a higher rate of recurrent bleeding (17.5 vs 7.8%; p = 0.012), a higher blood transfusion rate (68.0 vs 55.4%; p = 0.037), larger transfusion volume (p < 0.001), while the rate of endoscopic reintervention, need for emergency surgery, need for angiographic embolization, the rate of post endoscopy adverse events, mortality and length of hospitalization in two groups were not different significantly (Table 2).

Univariate and multivariate logistic regression analyses were performed to identify the predictors of the composite outcome. For this analysis, the following variables (potential confounders) were included in the model: age, sex, smoking, drinking, history of UGIB, history of prescribed medicine, comorbidities, systolic pressure, heart rate, hemoglobin, platelet count, prothrombin time, INR, GBS scores, AIMS65 scores, etiology, treatment modalities and off-hours endoscopic hemostasis. In univariate analysis, off-hours endoscopic hemostasis was a predictor of increased risk of the composite outcome (OR: 2.70; 95% CI: 1.39–5.23; p = 0.003). After controlling for confounding factors, the result was persistent (OR: 2.29; 95% CI: 1.08–4.81; p = 0.031). On the other hand, the hemoglobin level and AIMS65 scores were independent predictors of the composite outcome (OR: 0.98; 95% CI: 0.96–1.00; p = 0.034; OR: 1.58; 95% CI: 1.07–2.35; p = 0.022) (Table 3).

Of 216 patients in the lower-risk group, 29 patients (13.4%) reached the primary outcome. In the univariate logistic regression analyses, there were finally six parameters significantly correlated with the composite outcomes, namely hemoglobin (OR 0.96; 95% CI, 0.94–0.98; p < 0.001), albumin (OR 0.87; 95% CI, 0.82–0.93; p < 0.001), PT (OR 1.25; 95% CI, 1.04–1.50; p = 0.017), INR (OR 7.51; 95% CI, 1.17–4.81; p = 0.033), AIMS65 score (OR 2.25; 95% CI, 1.40–3.62; p = 0.001), epinephrine injection (OR 2.87; 95% CI, 1.02–8.06; p = 0.046). Only hemoglobin level was statistically significant after controlling for confounders (OR 0.96; 95% CI, 0.93–0.98; p < 0.001) (Table 4).

There were 85 patients in the higher-risk group, and 26 patients (30.6%) reached the composite outcome. Table 5 showed the results of univariate and multivariate logistic regression analyses on influencing factors of the composite outcome. Although the association of off-hours endoscopic hemostasis with the composite outcomes was reduced with adjustments, off-hours endoscopic hemostasis was still a significant risk factor for the higher-risk group (OR: 4.63; 95% CI: 1.35–15.90; p = 0.015).