• People are at high risk of serious nonsteroidal anti-inflammatory drug (NSAID)–induced gastrointestinal (GI) adverse events if they have one or more of the following risk factors:
• Older age and male. The risk of an upper GI complication, such as bleeding:
• Doubles with every decade above 55 years (for example, roughly 1 per 1000 persons per year at 55 years of age, 2 per 1000 persons at 65 years of age, 4 per 1000 persons at 75 years of age, and so on).
• Is twice as high in men as in women.
• History of gastroduodenal ulcer, GI bleeding, or gastroduodenal perforation.
• Concomitant use of medications that are known to increase the likelihood of upper GI adverse events (e.g. anticoagulants, aspirin used in low or standard doses, corticosteroids, and selective serotonin reuptake inhibitors [SSRIs], venlafaxine, or duloxetine).
• Serious comorbidity, such as cardiovascular disease, hepatic or renal impairment (including dehydration), diabetes, or hypertension.
• Requirement for prolonged NSAID use, including people with:
• Osteoarthritis or rheumatoid arthritis of any age.
• Chronic low back pain and 45 years of age or older.
• Use of the maximum recommended dose of an NSAID.
• Additional risk factors for NSAID-induced GI adverse events have also been identified, including:
• The NSAID used.
• The presence of Helicobacter pylori infection.
• Excessive alcohol use.
• Heavy smoking.

• For information on the risks associated with low-dose aspirin, see the CKS topic on Antiplatelet treatment.

The risk factors for GI adverse effects are based on recommendations in the National Institute for Health and Clinical Excellence (NICE) Technology Appraisal Guidance on the use of cyclo-oxygenase (COX) II–selective inhibitors for osteoarthritis and rheumatoid arthritis [NICE, 2001].

Increasing age, previous GI history, and presence of serious comorbidity:

• The consequences of GI bleeding are likely to be more serious in people who are otherwise in poor health, including those with the comorbidities identified by NICE.
• An RCT of 8843 people taking a NSAID for rheumatoid arthritis symptoms were randomized to misoprostol or placebo [Silverstein et al, 1995]. Analysis of 18 potential risk factors by logistic regression showed that older age (75 years or older), history of peptic ulcer, history of GI bleeding, and history of cardiovascular disease were all significant indicators of increased risk of serious upper GI complications.
• A systematic review and meta-analysis of 18 observational studies investigated NSAIDs and serious GI complications [Hernandez-Diaz and Rodriguez, 2000]. Pooled relative risks were calculated for different risk factors. NSAID users with advanced age or a history of peptic ulcer disease had the highest absolute risks for upper GI tract bleeding or perforation. Compared with people aged 25 to 49 years:
• People aged 60 to 69 years had 2.4 times the risk.
• People aged 70 to 80 years had 4.5 times the risk.
• People aged over 80 years had 9.2 times the risk.
• An RCT of 8076 people with rheumatoid arthritis taking either rofecoxib or naproxen for 9 months were assessed for the development of upper GI events [Laine et al, 2002]. A history of GI events, advancing age (in people 75 years of age or older, the risk was increased by 3.7 times, whilst for those aged 65 to 74 years the risk was increased by 2.4 times), and severe disability caused by rheumatoid arthritis, were found to be significant risk factors.

Concomitant use of medicines that increase the risk:

• A meta-analysis involving over 66,000 people found that the risk of GI bleeding for those taking long-term, low-dose aspirin was double the risk of those who did not take aspirin [Derry and Loke, 2000].
• A two-fold increased risk of GI bleeding was found in people taking low-dose aspirin for secondary prevention of cardiovascular disease who were not considered to be at high risk of GI complications [NICE, 2004].
• NSAID use can be influenced by drugs taken for other conditions. For example, people who are taking warfarin should ideally avoid taking NSAIDs [Baxter, 2006; BNF 55, 2008; National Collaborating Centre for Chronic Conditions, 2008].
• Selective serotonin reuptake inhibitors (SSRIs) are associated with an increased risk of bleeding, so should not normally be offered for people taking NSAIDs [NICE, 2009]. Venlafaxine and duloxetine as inhibit serotonin reuptake, so can be expected to interact similarly.

NSAID type, dose, and duration of use:

• A case-controlled study based on the GP Research Database estimated about a five-fold increase in the risk of bleeding or perforated peptic ulcer associated with taking NSAIDs but found higher risks with piroxicam (OR 18.0) and azapropazone (OR 23.4) [Garcia Rodriguez and Jick, 1994].
• A review of 12 epidemiological studies examined the performance of 14 NSAIDs relative to ibuprofen [Henry et al, 1996]. The review found that NSAIDs vary significantly in their GI toxicity. The review also found that the risk of GI injury increases for higher doses of the same NSAID. Doses in the region of ibuprofen 2.4 g daily may be no safer than intermediate-risk NSAIDs, such as diclofenac and naproxen.
• Nested case-controlled analysis of 2105 cases and 11,500 controls based in the UK found that high NSAID doses were associated with a much greater increased risk of bleeding compared with low NSAID doses [García Rodríguez and Hernández-Díaz, 2001]:
• For low-dose NSAIDs, relative risk (RR) 2.4 (95% CI 1.9 to 3.1).
• For high-dose NSAIDs, RR 4.9 (95% CI 4.1 to 5.8).
• Meta-analysis of data from three case-controlled studies (2472 cases of upper GI bleeding and 5877 controls) investigated NSAID dose-response relationships with GI risk [Lewis et al, 2002]. The risk of upper GI bleeding with NSAIDs varies up to 20-fold depending on the drug (ibuprofen at the lowest end, and piroxicam and ketoprofen at the highest end, of those drugs studied), and by three-fold to seven-fold depending on the dose chosen. Risk was maximal in the first week, decreased thereafter, and was reduced 1 week after discontinuing NSAID use.
• The Commission on Human Medicines (CHM; formerly Committee on Safety of Medicines) has created three categories of NSAID risk [CSM, 2002]:
• Lowest risk: ibuprofen (but serious and fatal GI adverse effects have still been reported).
• Intermediate risk: diclofenac, naproxen, ketoprofen, piroxicam, and indometacin.
• Highest risk: azapropazone.
• The CHM has confirmed that COX-2 selective NSAIDs have also been associated with serious and fatal GI adverse reactions [CSM, 2003]. However, COX-2 selective NSAIDs are thought to be associated with a lower risk of GI adverse events, although it is not clear whether this lower risk remains with their long-term use [Juni et al, 2003].
• An observational study of 5692 people with rheumatoid arthritis and 3124 people with osteoarthritis, representing 36,262 patient-years, was examined for rates of serious GI events from aspirin, paracetamol, and ibuprofen use [Fries and Bruce, 2003]. In people taking aspirin, paracetamol, or ibuprofen alone, only 1 event occurred in over 900 patient–years. People taking paracetamol in combination with other analgesics or corticosteroids had higher rates (borderline significance). With higher doses of all drugs, GI events tended to be relatively more frequent, but even in high-risk people on corticosteroids, the rate was around 20 per 1000 patient–years.
• The National Institute for Health and Clinical Excellence (NICE) recommend that a PPI is routinely co-prescribed with all NSAIDs (both standard NSAIDs and COX-2 selective NSAIDs) for all people with osteoarthritis, all people with rheumatoid arthritis, and anyone with chronic low back pain 45 years of age and older [National Collaborating Centre for Chronic Conditions, 2008; National Collaborating Centre for Chronic Conditions, 2009; National Collaborating Centre for Primary Care, 2009].
• For more information, see Prevention of GI adverse effects.

Presence of Helicobacter pylori infection:

• The relationship between H. pylori infection and NSAIDs in gastroduodenal pathology is complex, but H. pylori infection is believed to be an independent risk factor for peptic ulcer bleeding [Malfertheiner et al, 2007].

Excess alcohol use and heavy smoking:

• Alcohol has been thought to cause direct injury to gastric mucosa and cause non-ulcer dyspepsia [North of England Dyspepsia Guideline Development Group, 2004].
• Smoking increases gastric acid output and delays gastric emptying, which may be involved in the development of non-ulcer dyspepsia [North of England Dyspepsia Guideline Development Group, 2004].