In this multicenter study, our treatment outcomes were generally favorable. The overall mortality rate due to peptic ulcer bleeding was only 1.8%, which is lower than previous studies that had reported mortality of 6% to 7%[1-3
]. In addition, the need for surgery and angiographic embolization was also low, at 1% and 1.2%, respectively. These outcomes, including mortality, might have been significantly affected by the severity of disease and comorbidities. However, mortality in the subgroup whose Rockall score was above 6 was also only 4.1% in our study (6/146), and mortality was only 0.3% in patients with Rockall scores under 6 (1/388); thus, our low mortality was not due to a milder presentation of bleeding. Such favorable outcomes should result from timely endoscopic hemostasis and appropriate intensive care. We provided early endoscopy within 24 h in most patients (97%), even for weekend and nighttime admissions. The mean time to endoscopy was only 333.8 ± 22.2 min overall. Therefore, we suggest that early intervention could reduce the mortality of peptic ulcer bleeding to 1.8%.
Regarding the so-called weekend effect, our study demonstrated no weekend effect on mortality, need for surgery, angiographic embolization, or length of stay. In fact, the patients in the weekend group showed favorable outcomes comparable to those of the weekday group. We expected to nullify the weekend effect since all four participating hospitals were teaching referral hospitals with well-organized duty systems and formal out-of-hours emergency endoscopy services, which allowed early endoscopy at any time of any day. Contrary to our expectations, the time interval between presentation to the ER and endoscopy was shorter for weekend and nighttime admissions. The reasons for shorter time to endoscopy at night and on weekends might be longer waiting time during weekdays due to previously appointed outpatients, and fewer traffic jams at night and on weekends in urban settings.
However, a higher rebleeding rate was noted in the weekend group, and it is possible that the weekend effect might exist in spite of early endoscopy. In multivariate logistic regression analysis, weekend presentation was not a significant risk factor for rebleeding, but nighttime presentation and a high Rockall score were independent risk factors for rebleeding. Therefore, statistically, the risk factor associated with rebleeding was not weekend but nighttime admission. The rebleeding rate in the weekday-nighttime group was also higher than weekday-daytime group, although the difference was not statistically significant (weekday-daytime 11.4% vs
weekday-nighttime 15.1%, P
= 0.487). The weekend-daytime group showed a similar rebleeding rate to that of the weekday-daytime group, but the weekend-nighttime group had a significantly higher rebleeding rate (rebleeding rate 34.8% in the weekend nighttime group vs
12.8% in the weekend daytime group, P
= 0.040). However, we cannot completely rule out a weekend effect because the rebleeding rate of the weekend-nighttime group was higher than that of the weekday-nighttime group. Thus, it is reasonable to infer that these findings represent a nighttime effect or a weekend-nighttime effect rather than a weekend effect. Therefore, we concluded that the nighttime effect represented a new risk factor for rebleeding and was more powerful on the weekend through a combination with the weekend effect. Rebleeding after endoscopic hemostasis could also be affected by various endoscopic and clinical factors, such as an active bleeding pattern, gastric location of the peptic ulcer, larger ulcer size[19-21
] and even by the level of experience of endoscpists[24
], but our finding of increased rebleeding in the nighttime group, especially the weekend nighttime group, was not associated with these factors. However, we were unable to identify specific factors that accounted for this nighttime effect; potential reasons include fatigue and decreased concentration of endoscopists at night and reduced staffing patterns of physicians, nurses, and other support staff at night. It is necessary to be more alert and particularly careful regarding hemostasis in patients with UGIB who present at night on the weekend.
Rockall scores and Charlson scores were predictors of mortality, rebleeding, and length of stay[16-18
]. The Rockall score was designed to predict mortality and can also be used to predict rebleeding[16,17,25,26
]. To allow for clinical application of the Rockall score and to verify our data (because we observed such a low mortality rate), we used a receiver operating characteristic (ROC) curve to determine the cut-off value predictive of mortality and rebleeding. According to the ROC curve, a Rockall score of 6 points could be chosen as the cut-off value. Rockall scores greater than six were significantly associated with rebleeding [odds ratio (OR) = 2.08] and an increased mortality (OR = 1.77) of 4.1% (6/146). Bessa et al[27
] also reported that the Rockall score indicated a risk of mortality up to 15% if the score was above six. Our data support a Rockall score of 6 points as a critical point.
Contrary to our expectation, length of stay was shorter in the weekend group than in the weekday group. The length of stay increased in proportion to the CACI, and linear regression analysis showed that length of stay was strongly associated with the type of comorbidity rather than weekend presentation, emphasizing the type and severity of illness.
One weak point in our data was that the mean age of the weekend group was 5 years younger than the weekday group. Through a review of the medical records, we observed that there were some weekend patients who postponed visiting the hospital due to a busy work schedule or other circumstances despite having experienced melena for several days. Younger patients are more likely to be employed, thus they are potentially more inclined to present on the weekend rather than on a weekday due to their work. Socio-environmental factors, such as employment, social status, and personal characteristics, might have an influence on visiting the hospital and could be a possible explanation for the younger age of the weekend group compared to the weekday group. However, such factors were not available in the majority of medical records due to the retrospective nature of our study, and we were unable to identify a satisfactory reason in our results. Despite this weakness, we are confident that the age discrepancy was not a problem because all analyses of outcomes were age-adjusted.
Additional limitations of our study were that the number of deaths was too small to allow for satisfactory analysis and that our study was retrospective; however, all of the endoscopic procedures were based on the same protocol, which should overcome this weakness. In addition, our study was not a nationwide study, but we think that the results of this multicenter study are sufficient to conclude that early endoscopic intervention can lead to favorable outcomes in peptic ulcer bleeding even on weekends and at night.
In conclusion, early endoscopy for peptic ulcer bleeding could reduce mortality to 1.8% and could prevent the weekend effect on the majority of outcomes in patients with peptic ulcer bleeding. However, we identified nighttime presentation as a new risk factor for rebleeding, despite early endoscopy. The Rockall score was also a useful predictor of rebleeding, and we should take this into consideration in the prognosis of peptic ulcer bleeding. Therefore, we need to be more careful and alert at night when dealing with peptic ulcer bleeding, especially in patients who present at nighttime and those with high Rockall scores (≥ 6).