There have been a number of studies in the literature that have reported there to be no difference in the magnitude of risk reduction between intensive high-dose and non-high-dose intravenous PPI therapy and these include those conducted by the Hung and Yuksel groups [9
]. In addition, meta-analysis by Wang and Wu came to a similar conclusion, namely that low-dose intravenous PPI can achieve the same efficacy as high-dose PPI following endoscopic hemostasis [22
]. A more recent prospective study conducted by Songür and colleagues also showed that high-dose esomeprazole infusion therapy following endoscopic hemostasis treatment is not superior to low-dose PPI therapy in the terms of re-bleeding, need for surgery and mortality [24
]. Moreover, Chen and colleagues from Taiwan concluded in a randomized clinical trial that PPI dosage is not associated with rebleeding following combined endoscopic haemostasis of bleeding ulcers [25
]. Obviously, the findings remain somewhat unconvincing with respect to the optimal dosage of PPI for ulcer bleeding despite the publication of updated Vienna and Asian-pacific consensus statements [10
]. In parallel with the above findings, we still do not know whether the effect of intravenous non-high-dose PPIs is as good as high-dose PPIs in terms of reducing rebleeding rates among patients at low risk (Rockall score ≤ 6) or among those at high risk (Rockall score ≥ 6).
The overall rebleeding rate in current study was high at 21.5% (89/413) with Hung and Yuksel reported rebleeding rates of 3.9% (4/103) and 7.2% (7/97), respectively [9
]. The differences can be explained by the enrollment in the present study of more patients with concurrent illness resulting in a mean Rockall score of 6.1. The prevalence of CKD in Yuksel's study was only 2.06% (2/97), and the mean score of American Society of Anesthesiologist (ASA) criteria was only 1.5 to 1.6 in Hung's study. As is well known, severe concurrent illness may dilute the results attained when assessing the effect of high-dose PPI treatment. All these studies, including ours, observed that intravenous high-dose pantoprazole treatment did not appear to be more effective at reducing rebleeding compared to a non-high-dose regimen among patients with high-risk stigmata. The greatest limitation with a retrospective case-controlled study like the current study is that, despite the attempt to minimize the possible selection bias between the two treatment groups by conducting greedy matching, prejudice is still inevitable among the high-dose group. For instance, there were still more rebleeding patients in the high risk patients with Rockall score ≥ 6 (14.3% vs
. 40.2%, p
Most interestingly, after case-controlled matching, the low risk patient subgroup (Rockall score ≤ 6) analysis actually showed that the rebleeding rate was similar for both the high-dose and non-high-dose groups (3/27, 11.1% in each group) (Table ). This implies that non-high-dose intravenous PPI is enough to preventing rebleeding among low risk patient. Theoretically, there is no doubt about the effectiveness of the anti-secretory effect of high-dose PPI. Evidence points clearly towards the fact that continuous intravenous infusion of PPI is able to maintain an intragastric pH > 6 for 59% to 98% of the time during monitoring [26
]. In contrast, other studies have shown that low-dose intravenous PPI can be as effective as a high-dose regimen at maintaining a consistent pH of around 4-6 [29
]. Choi and colleagues reported that low-dose continuous infusion of pantoprazole (40 mg bolus followed by 4 mg/h) was able to maintain an intragastric pH > 6 in a manner similar to that of a high-dose group (80 mg bolus followed by 8 mg/h) among Korean patients with PU bleeding requiring endoscopic hemostasis [31
]. At least two other meta-analysis studies have revealed that high-dose PPIs are not superior to non-high-dose PPIs in terms of reducing the rate of rebleeding, neither with respect to the need for surgical intervention, nor in terms of mortality after endoscopic treatment among patients with bleeding PUs. This suggests that the non-high dose PPI treatment strategy may provide an equal clinical benefit among ulcer bleeding patients [24
]. Therefore, it makes sense that non-high-dose intravenous PPI might be sufficient to reduce the rebleeding of PUs after initial endoscopic hemostasis among Taiwanese, at least among the subgroup of low risk patients that have a Rockall score < 6.
Co-morbidities influence the rate of recurrent bleeding [33
]. Recurrent PU bleeding may be prolonged in individuals with co-morbidities and therefore Cheng and colleagues suggested a extended low dose infusion of intravenous PPIs for up to 7-days may result in better control of recurrent bleeding of PUs [13
]. For instance, the fact that patients have a more severe stage of kidney disease (CKD V) might be relevant to increased rebleeding [34
]. This is consistent with the present study in that CKD stage III to V was found to be an influencing risk factor for recurrent bleeding on univariate analysis in spite of the fact that all subjects with end stage renal disease (ESRD) received heparin-free dialysis at our hospital. Uremic platelet function impairment may be responsible for this higher risk of ulcer rebleeding [35
]. The reason behind uremic platelet dysfunction involving the interaction of von Willebrand factor (vWf) with various platelet membrane glycoproteins, namely Ib and IIb to IIIa, which do not normalize after dialysis [36
From the cost-effectiveness point of view, Leontiadis and colleagues reported that high-dose PPI therapy is more expensive, and that non-high dose is relatively inexpensive [37
]; however, Barkun and colleagues proved that high-dose intravenous esomeprazole strategy is more effective and less costly than a non-intravenous esomeprazole strategy [38
Several limitations of this study must be recognized. Firstly, this retrospective analysis is dependent on the completeness of the medical charts. If the chart report of ulcer pattern was not complete, we reviewed the image obtained by endoscopy or watching the recorded video in order to record the ulcer characteristics. Therefore, the investigation reliability might vary in this area. Secondly, selection bias definitely existed among high-dose group due to the clinicians' decision with respect to the dosage used for the more severely ill patients, even though we attempted to minimize this selection bias using the greedy matching method after controlling the baseline conditions of the subjects. Thirdly, some patients were treated by endoscopic epinephrine injection alone, which was suboptimal. Therefore, the intention to compare the overall efficacies between the non-high-dose and high-dose PPIs is hampered by these limitations. However, the interesting part of this study remains our observed in terms of the low risk subgroup and treatment strategy.