Poor adherence to gastroprotective agents (GPAs) is common among users of nonsteroidal anti-inflammatory drugs (NSAIDs) or low-dose aspirin (ASA). There are little data on the utilization of GPAs among NSAID and ASA users complicated by ulcer bleeding.
To study the utilization of GPA among NSAID and ASA ulcers before the onset of ulcer bleeding.
We conducted a cross-sectional study to determine the exposure to NSAIDs, ASA, and GPAs within 4 weeks before endoscopically confirmed ulcer bleeding. Sensitivity analysis was performed to study how improving adherence to GPA use would reduce the risk of ulcer bleeding in high-risk users.
Between 2000 and 2009, 1093 and 2277 patients had NSAID- and ASA-associated ulcer bleeding respectively. The incidence of NSAID-associated ulcer bleeding declined by 40%, whereas that of ASA-associated ulcer bleeding increased by 46%. Thirty-nine per cent of NSAID users and 75% of ASA users belonged to high ulcer risk category. Although GPA prescription rate has increased over time, only 41.6% and 30.6% of high-risk NSAID and ASA users received GPAs before ulcer bleeding respectively. Sensitivity analysis showed that if GPAs could reduce bleeding risk by 50%, improving adherence would prevent up to 35% of ulcer bleeding in high-risk users.
A substantial proportion of high-risk NSAID and ASA users had not received prophylaxis with gastroprotective agents before ulcer bleeding. These bleeding episodes may be preventable with better adherence to gastroprotective agent use.
Use of NSAIDs and ASA has emerged as the most important cause of peptic ulcer bleeding worldwide. A number of factors are known to increase the risk of ulcer bleeding with NSAID or ASA use, including old age, a history of ulcer or ulcer bleeding, severe comorbid illnesses, concomitant use of anticoagulants or corticosteroids, and use of high-dose or multiple NSAIDs.1 There is good evidence that co-prescribing NSAIDs or ASA with GPAs can effectively reduce the risk of ulcer bleeding in high-risk patients.[2-4] To date, a number of practice guidelines are available to guide clinicians on prescribing GPAs to NSAID and ASA users with high ulcer risk.[5-7]
However, physicians often manage high-risk patients discordant with the recommendations. Marked under-utilization of GPAs among high-risk NSAID or ASA users in Europe and United States has been well documented.[8, 9] It has been shown that the risk of peptic ulcer or upper gastrointestinal bleeding is significantly higher in NSAID users with poor adherence to GPAs.
Despite extensive literature on under-utilization of GPAs among NSAID and ASA users, one important clinical question remains unaddressed: among NSAID and ASA users complicated by ulcer bleeding, how many of these bleeding episodes were preventable? To address this question, one needs to know the proportion of NSAID and ASA bleeders belonging to high ulcer risk category, and the proportion of these high-risk users receiving GPAs before the bleeding episode. These questions are clinically relevant because ulcer bleeding in high-risk NSAID and ASA users could have been prevented if clinicians are adherent to practice guidelines. On the other hand, ulcer bleeding in average-risk NSAID and ASA users is unpredictable because current guidelines do not recommend routine prophylaxis in these patients. In this cross-sectional study, we sought to study the prevalence of NSAID- and ASA-associated ulcer bleeding, to assess the magnitude of the problem caused by under-utilization of GPAs in high-risk users before ulcer bleeding, and studying the potential impact of adherence to GPA use on the risk of ulcer bleeding.
This was a cross-sectional study conducted at the Prince of Wales Hospital that served a local population of 1.2 million (17%).
We screened consecutive patients with a clinical diagnosis of upper gastrointestinal (GI) bleeding to the Prince of Wales Hospital from January 2000 to December 2009. All patients underwent gastroscopy within 24 h of admission. Those with endoscopically confirmed ulcer bleeding were systemically scrutinized for any recent use of NSAIDs, ASA, or GPAs using a territory-wide electronic database and a standardized questionnaire as previously described.[10, 11] Briefly, we used a territory-wide electronic database to identify prescription drug use and concurrent medical conditions. This database belongs to the health authority that provides medical care to over 90% of the local population. Designated research staff was responsible for administering a locally validated, standardized questionnaire to capture patient demographics and screened for exposure to over-the-counter drugs.[10, 11] A collateral drug history was obtained from family members, friends, and family doctors if the patient was unable to provide a detail drug history. Exposure to NSAIDs, ASA or GPAs was defined as taking at least 1 dose of these drugs within the 4 weeks before ulcer bleeding. GPAs included proton pump inhibitors (PPIs), histamine type-2 receptor antagonists (H2RAs), and misoprostol. Patients using concomitant ASA and NSAIDs, who only accounted for about 10% of cases, were classified into the ASA group. Patients were classified as having high ulcer risk if they had a history of peptic ulcer/ulcer bleeding or carried two or more predefined risk factors: age >70 years, co-morbidity, and concomitant use of anticoagulants, other antiplatelet drugs, corticosteroids or multiple NSAIDs.[1, 5] Co-morbidity was defined as established coronary artery, cerebrovascular disease, cirrhosis or renal failure. Those who did not fulfil the criteria of high ulcer risk were classified into the average-risk group.
The patients' baseline characteristics were presented as descriptive data. We plotted the proportions of high-risk users of NSAID and ASA who had received GPAs in between 2000 and 2009 using yearly intervals. Calendar years were grouped as 2000–2002, 2003–2005 and 2006–2009 for comparisons. As ulcer bleeding in high-risk patients is potentially preventable, we performed a two-way sensitivity analysis to assess how varying the adherence to GPA prescription and the efficacy of GPAs might decrease the percentage of high-risk users complicated by ulcer bleeding. All statistical tests were 2-sided. Statistical significance was taken as P <0.05. Statistical analysis was performed by Statistical Package for Social Science (spss version 15.0; SPSS Inc, Chicago, IL, USA).
A total of 13 861 patients with a clinical diagnosis of upper GI bleeding underwent gastroscopy at the Prince of Wales Hospital between 2000 and 2009, of whom 5130 patients (47%) used NSAIDs or ASA. Among all NSAID/ASA users, 1760 did not have bleeding lesions found in the upper GI tract, 1093 had NSAID-associated ulcer bleeding and 2277 ASA-associated ulcer bleeding.
Table 1 listed demographic characteristics of patients with NSAID- and ASA-associated ulcer bleeding. The proportions of male and female patients were comparable in both groups. There were numerically more patients with a history of peptic ulcer or ulcer bleeding in the NSAID group (40.2%) than in the ASA group (26.9%). About 20% of patients in each group used concomitant drugs that can cause ulcer bleeding; multiple NSAIDs accounted for half of concomitant drug use. A high proportion of patients in the ASA group had co-morbidities (86.7%) because most patients used ASA for established coronary artery or cerebrovascular diseases instead of primary cardiovascular prevention.
Table 1. Demographic characteristics of patients with NSAID- and ASA- associated ulcer bleeding
Current smoker was defined as smoking daily irrespective of the quantity within 4 weeks of the event. Current drinker was defined as drinking daily irrespective of the quantity within 4 weeks of the event.
Patients using concomitant NSAIDs and ASA were classified as ASA users.
Overall, GPAs were prescribed to 1040 patients within 4 weeks before the bleeding episode, 392 (36%) in the NSAID group and 648 (29%) in the ASA group. The most frequently prescribed GPA in both groups were H2RA, 321 (29.4%) for NSAID users and 489 (21.5%) for ASA users. Thirty-nine per cent of patients in the NSAID group and 75% in the ASA group had high ulcer risk as defined a priori. Among these high-risk patients who should have prescribed GPAs, only 42% in the NSAID group and 31% in the ASA group had received GPAs before ulcer bleeding (Table 2). Figure 1 shows the percentages of high-risk ulcer patients receiving GPA prophylaxis in yearly intervals from 2000 to 2009.
Table 2. Demographic characteristics of patients with high ulcer risk
Risk factors include age >70 years, co-morbidity and concomitant use of anticoagulants, other antiplatelet drugs, corticosteroids or multiple NSAIDs. Co-morbidity was defined as established coronary artery, cerebrovascular disease, cirrhosis and renal failure.
The incidence rate of NSAID-associated ulcer bleeding declined by 40% over the study period (from 150 cases in 2000 to 90 cases in 2009) (Figure 2). Among these NSAID bleeders, the number of high-risk users decreased from 70 cases (46.7%) in 2000 to 19 (21.1%) in 2009. The prescription rate during 2006–2009 was 39% and 10% higher than that in 2000–2002 and 2003–2005, respectively.
The incidence rate of ASA-associated ulcer bleeding has increased by 46% over the study period (from 157 cases in 2000 to 230 cases in 2009) (Figure 2). There was a numerical trend towards more high-risk ASA users over the last 10 years (65% in 2000 vs. 76.1% in 2009). Using a modified Poisson regression model, high-risk ASA users were 122% more likely to receive GPAs before ulcer bleeding during 2006–2009 when compared with the GPA prescription rate during 2000–2002 and 2004–2005 (P < 0.001). The GPA prescription rate of high-risk ASA users before ulcer bleeding was 122% and 114% higher in 2006–2009 than in 2000–2002 and in 2003–2005, respectively.
Two-way sensitivity analysis
The high-risk group consisted of 428 NSAID and 1708 ASA users. Among them, 250 (58.4%) NSAID users and 1186 (69.4%) ASA users developed ulcer bleeding without receiving GPA prophylaxis. Two-way sensitivity analysis showed how varying the adherence to GPA prescription and the efficacy of GPAs might alter the percentage of high-risk users complicated by ulcer bleeding. In the best-case scenario, with GPA protective efficacy of 100% and all high-risk users receiving GPA, there will be 60% reduction in ulcer bleeding among NSAID users and 70% among ASA users (Figures 3 and 4). If the protective efficacy is 50% and all high-risk users received GPA, about 35% of ulcer bleeding in high-risk NSAID and ASA users can be prevented.
This study is among the first that sought to assess the utilization of prophylactic GPAs in users of NSAIDs and ASA who were complicated by ulcer bleeding. We found that among these patients with ulcer bleeding, 39% of NSAID users and 75% of ASA users had high ulcer risk (i.e. previous ulcer/ulcer bleeding or ≥2 risk factors). However, only 30–40% of high-risk users were prescribed GPAs before ulcer bleeding. Although the proportion of high-risk patients receiving GPA prophylaxis has risen steadily over the last 10 years, the overall adherence to guidelines on co-prescribing GPAs in NSAID or ASA users is still unsatisfactory as reflected by the current study.
Ulcer bleeding in high-risk NSAID and ASA users is avoidable because this complication can be prevented by improving adherence to GPAs. Our sensitivity analysis showed that the number of high-risk users with ulcer bleeding could be reduced by 60% if all of these patients received a highly effective GPA. This result has a substantial impact on patient outcome as well as healthcare cost. Although a cost-effective analysis is lacking, it is likely that the cost of prescribing expensive GPAs such as PPIs to high-risk patients will be offset by the cost saving associated with reduced morbidity and hospitalization. The importance of improving the adherence to prescription guidelines in high-risk users cannot be overemphasized.
Several reasons could account for the underutilization of GPAs in high-risk users. An American study reported that a lack of familiarity with NSAID prescribing guidelines in physicians was the most frequent barrier in guideline adherence. A combination of a computer alert and brief physician education on the risk factors for NSAID-related GI complications led to an increased GPA prescription. Misperception about the efficacy and safety of ASA is another plausible reason. Elnachef et al. showed that one in seven primary care practitioners preferred 325 mg to lower doses of ASA for cardioprotection. Current evidence indicates that lower doses of ASA are as effective and have lower risk of GI bleeding. Another confusion about ASA is that enteric-coated or buffered ASA reduces the risk on gastrointestinal adverse events. It has been shown that the risk of ulcer bleeding in coated or buffered ASA is similar to plain ASA. Education on appropriate use of ASA for cardioprotection is important because our result showed that the incidence of ASA-associated ulcer bleeding increased by 46% over the last 10 years.
Among patients who developed ulcer bleeding while receiving GPAs, we found that H2RAs accounted for about 80% of GPAs prescribed and PPI for approximately 20%. This finding is inconsistent with results reported in several other studies. Barozzi et al. found that PPI prescription have overtaken that of H2RAs in NSAID users in Australia. A Dutch study showed that PPI co-prescription was 77%, while H2RA accounted for only 22% in NSAID users. Similar results were observed in a Spanish study where PPI was the most frequently prescribed GPA in high-risk NSAID users. Our finding that a large percentage of high-risk NSAID and ASA users developed ulcer bleeding despite receiving H2RAs may suggest that H2RA provided insufficient gastroprotection.[7, 20, 21] It remains uncertain whether double-dose H2RA can reduce the risk of NSAID-related ulcer bleeding.[6, 22, 23]
The observation that over 60% of patients with NSAID-associated ulcer bleeding did not belong to the high ulcer risk category is unexpected. Current guidelines do not recommend GPA prophylaxis for these average-risk NSAID users. Ulcer bleeding in these patients was unpredictable. Further studies are needed to determine factors predicting ulcer bleeding in average risk users.
Our study had limitations. First, this was a single-centre study in Hong Kong. Our findings might not be generalizable to other healthcare systems. However, our findings should alert clinicians and health authorities worldwide to improving gastric protection in high-risk users. Second, we did not have a control group of NSAID or ASA users without ulcer bleeding. Therefore, we were unable to determine GPA use in this population. However, this should not undermine the importance of our finding that ulcer bleeding in many high-risk NSAID and ASA users was potentially preventable.
In conclusion, this study showed that although the prescription of GPA to high-risk NSAID and ASA users before the onset of ulcer bleeding has increased over the last decade, this was achieved in less than 50% of patients. Improving the adherence to prescription guidelines and patient education are essential in reducing the occurrence of ulcer bleeding in high-risk users.
Guarantor of the article: F. K. L. Chan.
Author contributions: Charita Ho was responsible for study design, data collection, data analysis, and drafting of the article. YK Tse provided statistical support. Brian Wu was responsible for data collection and data analysis. Chris Mulder was responsible for critical revision of the article. Francis Chan was responsible for study design and critical revision. All authors read and approved the final manuscript.
We thank Ms Jessica YL Ching for her support in data analysis. Declaration of personal interests: Francis Chan served as a consultant for Eisai, Pfizer and AstraZeneca. He has received lecture fees from Pfizer AstraZeneca, Takeda and Eisai.
Declaration of funding interests: This study was supported by a Focused Investment Scheme of The Chinese University of Hong Kong.