Article Text
PDF
Abstract
Background Non-steroidal anti-inflammatory drugs (NSAIDs) are one of the most commonly prescribed medications, but they are associated with a number of serious adverse effects, including hypertension, cardiovascular disease, kidney injury and GI complications.
Objective To develop a set of multidisciplinary recommendations for the safe prescription of NSAIDs.
Methods Randomised control trials and observational studies published before January 2018 were reviewed, with 329 papers included for the synthesis of evidence-based recommendations.
Results Whenever possible, a NSAID should be avoided in patients with treatment-resistant hypertension, high risk of cardiovascular disease and severe chronic kidney disease (CKD). Before treatment with a NSAID is started, blood pressure should be measured, unrecognised CKD should be screened in high risk cases, and unexplained iron-deficiency anaemia should be investigated. For patients with high cardiovascular risk, and if NSAID treatment cannot be avoided, naproxen or celecoxib are preferred. For patients with a moderate risk of peptic ulcer disease, monotherapy with a non-selective NSAID plus a proton pump inhibitor (PPI), or a selective cyclo-oxygenase-2 (COX-2) inhibitor should be used; for those with a high risk of peptic ulcer disease, a selective COX-2 inhibitor plus PPI are needed. For patients with pre-existing hypertension receiving renin-angiotensin system blockers, empirical addition (or increase in the dose) of an antihypertensive agent of a different class should be considered. Blood pressure and renal function should be monitored in most cases.
Conclusion NSAIDs are a valuable armamentarium in clinical medicine, but appropriate recognition of high-risk cases, selection of a specific agent, choice of ulcer prophylaxis and monitoring after therapy are necessary to minimise the risk of adverse events.
- aspirin
- bleeding peptic ulcer
- cardiovascular disease
- non-steroidal anti-inflammatory drugs
- portal hypertension
Statistics from Altmetric.com
- aspirin
- bleeding peptic ulcer
- cardiovascular disease
- non-steroidal anti-inflammatory drugs
- portal hypertension
Introduction
Non-steroidal anti-inflammatory drugs (NSAIDs), sometimes referred to as anti-inflammatory analgesic therapy, are one of the most commonly prescribed classes of medication in clinical practice and account for 5–10% of all drugs prescribed each year.1–4 NSAIDs are a group of pharmacological agents that target cellular cyclo-oxygenase (COX) and inhibit the synthesis of prostaglandins from arachidonic acid.5 6 Cyclo-oxygenase has two major isoforms: cyclo-oxygenase-1 (COX-1) is the major source of cytoprotective prostaglandins in the gastrointestinal (GI) tract, whereas cyclo-oxygenase-2 (COX-2) is the predominant isoform at sites of inflammation.6 7 NSAIDs are extensively used for the treatment of acute or chronic arthritis, headache, visceral pain, postoperative pain and various musculoskeletal problems.8–14 A good body of evidence confirms that NSAIDs effectively relieve pain, reduce local and systemic inflammatory response and improve musculoskeletal function and the quality of life.8–14
However, NSAIDs are associated with a number of serious adverse effects,15 including hypertension,16–18 cardiovascular disease,19–22 kidney injury18 23 24 and GI complications.22 25–27 In the past 20 years, a considerable amount of data has been published on the specific risk of individual adverse effects, relative risk with different NSAIDs and the efficacy of various prophylactic strategies to prevent individual adverse effects. However, most of the published literature deals with cardiovascular and GI complications,28–31 while the adverse effects on blood pressure (BP) and renal complications are less covered. Moreover, most of the guidelines usually focus on the complication of a single system and lack multidisciplinary considerations. Although adverse effects are similar between different ethnic groups, NSAIDs are extensively prescribed, both appropriately and inappropriately, in the Asia Pacific region, which has a growing population of people seeking treatments for acute and chronic pain, creating a rapidly growing analgesics market.32 33 This article aimed at developing a set of multidisciplinary recommendations on the safe prescription of NSAIDs.
It is important to note that although aspirin at a sufficient dose has the property of a NSAID, low dose aspirin is more commonly used in modern clinical practice for cardiovascular protection.34 Since low dose aspirin does not have a systemic anti-inflammatory effect and its adverse effects are probably different from those of other NSAIDs,34 35 our recommendations do not cover, and should not be applied to, patients using low dose aspirin for cardiovascular indications.
We strongly emphasise the importance of shared decision-making on the use of NSAIDs.36 Although our recommendations provide useful information on the risks of NSAIDs, the information on possible benefits should be provided separately. The role of doctors is to outline the benefits, risks and burdens of the treatment in a clear and understandable fashion. Our recommendations are not meant to be prescriptive, and it is the patient who must weigh up the information, together with other relevant concerns, and make the decision. Common clinical sense would also indicate that regardless of the requirement for a NSAID, the lowest effective dose should be used for only as long as clinically necessary.
Methods
Framework
The positional statements for the safe use of NSAIDs were supported by literature reviews with reference to the methods requirement in Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA). A steering committee meeting was set up to propose these positional statements and confirm the invitations to experts from different medical societies or associations. An expert panel was formed before the draft of the positional statements. A literature review was conducted to identify publications that reported adverse events due to NSAID use. Evidence from the literature was summarised and circulated among the panel members. At the expert panel meeting, the members met to discuss, finalise and endorse these position statements. Details of the steering committee meeting and the expert panel meeting are outlined in the online supplementary information.
Supplemental material
Literature review
A literature review was carried out in advance of the steering committee meeting (online supplementary figure 1). Search terms, including 'anti-inflammatory agent, 'non-steroidal', 'cyclo-oxygenase-2 inhibitors' and 'NSAID', were used to examine relevant literature from Medline, Embase and Cochrane Library. References of relevant publications were examined for additional potential papers. Eligible studies included randomised control trials and observational studies published before January 2018. A total of 22 683 papers were identified from the initial search. After removing 6791 duplicated records from the database, 15 892 abstracts were reviewed for relevance. Finally, a total of 329 papers were included after excluding papers with irrelevant topics, such as pathogenic mechanisms, psychological analysis, insurance plans, with children as subjects for recruitment, dose effectiveness, hearing loss as an adverse event or other topical analgesic treatments. The full text of the included articles was assessed, and a total of 104 papers were included on NSAID use for adults with arthritis. Other clinical guidelines or review articles were manually searched as additional references for the draft of statements. All positional statements were discussed in the expert panel meetings, and then prepared as an academic manuscript with endorsement from all members. Each statement was discussed for the level of evidence available and the strength of the recommendation using the classification system of the Grading of Recommendations, Assessment, Development and Evaluations (GRADE) (online supplementary table 1).37
Supplemental material
Supplemental material
Definitions
NSAIDs are conveniently classified into non-selective 'traditional NSAIDs', which inhibit COX-1 and COX-2 non-selectively, and 'selective COX-2 inhibitors'. It must be pointed out that the term 'cyclo-oxygenase-2 inhibitors' is commonly used to represent selective COX-2 inhibitors, but a number of relatively selective COX-2 inhibitors do not fall into the cyclo-oxygenase-2 inhibitors category (eg, nabumetone, meloxicam and etodolac). In this article, we use the term 'selective COX-2 inhibitors' unless the discussion specifically refers to 'cyclo-oxygenase-2 inhibitors'. The prescription of a NSAID could either be short term or chronic, and in both cases, could be used as regular or on-demand treatment. There is no generally accepted cut-off point for short-term versus chronic NSAIDs. As an arbitrary definition, we consider an intended duration of treatment for 4 weeks or more as chronic NSAID therapy, irrespective of a regular or on-demand dosing schedule.
Hypertension is defined according to the latest American College of Cardiology/American Heart Association guideline.38 Resistant hypertension is defined as failure to achieve target BP despite adherence to three appropriate antihypertensive agents, or BP controlled to the patient’s individualised target while receiving treatment with at least four antihypertensive drugs.39 40 An ischaemic vascular event is defined as a history of angina, myocardial infarction, stroke, transient ischaemic attack or intermittent claudication. Chronic kidney disease (CKD) is defined as abnormalities of kidney structure or function, present for at least 3 months, with implications for health.41 Risk factors of CKD include diabetes, hypertension, previous kidney disease, ischaemic vascular event, heart failure or age >80 years. CKD stage 3 refers to estimated glomerular filtration rate (eGFR) of 30–59 mL/min/1.73 m2; CKD stage 4 or 5 refers to eGFR of <30 mL/min/1.73 m2.41 Upper GI adverse events include symptomatic ulcer, perforation or obstruction, gastroduodenal bleeding and dyspepsia; lower GI adverse events include small or large bowel bleeding, clinically significant anaemia and perforation and complicated diverticular disease. Moderate risk of peptic ulcer disease is defined as possessing one or two of the following risk factors: age >65 years; previous history of uncomplicated ulcer; concurrent use of aspirin (low and high dose), antiplatelet drugs, corticosteroids or anticoagulants; and high risk of peptic ulcer disease in patients with a history of a previously complicated ulcer (especially recent ones).
Effects of NSAIDs on blood pressure
Guideline statements
1.1 We recommend BP be measured before initiation of NSAID. [1D]
1.2 We recommend BP be monitored after initiation of chronic NSAID therapy. [1C]
1.2.1 We suggest that BP should be measured in the clinic 4 weeks after initiation of NSAID (or earlier if clinically indicated). [2C]
1.2.2 Home BP monitoring is desirable. [2D]
1.2.3 Self-monitoring of fluid retention by body weight and checking for pedal oedema is desirable. [2D]
1.3 For patients with pre-existing hypertension, we recommend BP be monitored even with short-term NSAID therapy. [1C]
1.4 If a NSAID is necessary for patients with pre-existing hypertension, evidence does not allow the preferential recommendation of any specific drug. (not graded)
1.5 For patients with treatment-resistant hypertension, we recommend that a NSAID be avoided if possible. [2C]
1.6 For patients who are already receiving an angiotensin-converting enzyme (ACE) inhibitor or angiotensin receptor blocker (ARB), we recommend empirical addition (or increase in the dose) of an antihypertensive agent of a different class be considered when prescribing chronic NSAID therapy. [2C]
Background
NSAIDs have been shown to increase BP levels, cause new-onset hypertension and worsen the BP control of patients with pre-existing hypertension in multiple observational studies,42–44 randomised controlled trials,45–53 systematic reviews and meta-analyses.54–59 The absolute magnitude of the increase in BP is generally small. In patients without pre-existing hypertension receiving chronic NSAID therapy, the average increases in systolic and diastolic BP are around 2.0 and 0.5 mm Hg, respectively.47 A similar magnitude of increase in BP is seen in patients with pre-existing hypertension.42 43 47 53 A substantial minority group of patients, however, develop new-onset hypertension or worsening of BP control after a NSAID. Around 2–30% of patients treated with a NSAID require modification of antihypertensive treatment.44 47 50–52
Different NSAIDs have a substantial difference in their effect on BP. Most studies show that ibuprofen, etoricoxib and rofecoxib cause a greater increase in BP, whereas the effects of naproxen and celecoxib are relatively small.42 48 50–52 54–57 59 The increase in BP is probably more pronounced in patients receiving an ACE inhibitor or ARB.42 43 49 52
Rationale
Guideline 1.1
Hypertension is an important contributor to the global burden of disease and mortality.60 61 Good evidence supports screening for high BP in adults,62–64 and is recommended by the United States Preventive Services Task Force (USPSTF),65 the American College of Cardiology/American Heart Association (ACC/AHA),66 the Canadian Task Force on Preventive Healthcare,67 the European Society of Cardiology/European Society of Hypertension68 and the International Society of Hypertension.69 Because use of a NSAID may increase the BP, and because hypertension is frequently asymptomatic, and measurement of BP is simple and non-invasive, we recommend measuring BP before initiation of NSAID.
Guideline 1.2
As described above, NSAID consistently causes an increase in BP.44 47 50–52 Although the absolute magnitude of the increase in BP is usually small, 4–23% patients develop new-onset hypertension,45–47 70 and 2–30% of patients with pre-existing hypertension require modification of their drug treatment.44 47 50–52 Because there are no reliable predictors for the development of new-onset hypertension or worsening of BP control after NSAID therapy, we recommend BP be monitored in all patients, including those without a history of hypertension, after initiation of chronic NSAID therapy. However, the benefit of pre-emptive intervention to control the BP remains uncertain.
Guideline 1.2.1
The time course of BP increase after NSAID treatment remains incompletely defined. Most studies show that an increase in BP can be seen after 4–8 weeks of NSAID therapy, and the effect persists as long as the patients continue to receive treatment.43 44 48 52 53 For example, Sowers et al 53 found that systolic BP increased by 2.6 mm Hg after 4 weeks of celecoxib therapy. After 6 weeks of rofecoxib treatment, Whelton et al 52 noted that 14.9% of patients required dose adjustment of their antihypertensive regimen. Although there are few published studies on the relation between acute and delayed (eg, after 4 weeks) BP change, we suggest BP should be measured in the clinic 4 weeks after the initiation of a NSAID.
Guideline 1.2.2
Few studies have specifically compared the difference between office, home and ambulatory BP response in hypertensive patients receiving a NSAID. The rise in ambulatory BP is probably similar to that in clinic BP,48 50 51 but the application of home BP monitoring has not been specifically studied. The latest ACC/AHA guideline for the prevention, detection, evaluation and management of high BP in adults, however, put strong emphasis on the role of ambulatory BP monitoring (ABPM) and home BP monitoring (HBPM) for the diagnosis and monitoring of hypertension.66 Because ABPM and HBPM use different BP thresholds to define high BP from those in office-based methods, the best estimates for corresponding clinic BP, HBPM, daytime ABPM, night-time ABPM and 24 hours ABPM should be considered as a guide and interpreted with caution.66
Guideline 1.2.3
Fluid retention, manifesting as weight gain or pedal oedema, is common after NSAID therapy.71 72 Whelton et al 52 found that symptomatic ankle oedema developed in 4.7– 7.7% of patients after 6 weeks of COX-2 inhibitor therapy. Fluid retention may or may not coexist with elevation of BP, but may indicate worsening of kidney function.73 We suggest self-monitoring of fluid retention by self-checking for body weight change or the presence of pedal oedema.
Guideline 1.3
In patients with pre-existing hypertension, most studies show that worsening of BP control is seen after 4–8 weeks of NSAID therapy.44 47–51 However, Whelton et al 52 noted that an increase in BP could be seen after 2 weeks of selective COX-2 inhibitor treatment, while Sowers et al 51 found that 16–30% of patients with pre-existing hypertension require up-titration of their antihypertensive regimen following NSAID therapy.51 Taken together, it seems a safe practice to monitor BP with short-term (ie, less than 4 weeks) NSAID therapy in patients with pre-existing hypertension.
Guideline 1.4
There is no consistent difference in the effect on BP among different NSAIDs, and there is no consistent difference in the change in BP between traditional NSAIDs and COX-2 inhibitors.45–51 Although four meta-analyses or network analyses show that the effect on BP control depends on the specific NSAID,55–57 59 no single agent has been convincingly shown to be better than the others. We therefore conclude that current evidence does not allow the recommendation of any specific NSAID for patients with pre-existing hypertension.
Guideline 1.5
Resistant hypertension is defined as failure to achieve target BP despite adherence to three appropriate antihypertensive agents, or BP controlled to the patient’s individualised target while receiving treatment with at least four antihypertensive drugs.39 40 To the best of our knowledge, there are no published data on the effect of BP after NSAID therapy in patients with treatment-resistant hypertension. Although the magnitude and proportion of patients with raised BP seem similar between patients with and without pre-existing hypertension,44–47 50–52 70 up to 30% of patients had worsening of BP control after a NSAID and required up-titration of their antihypertensive regimen.47 50–52 The pharmacological options are limited for resistant hypertension, and thus it appears logical to recommend avoiding NSAID therapy in patients with treatment-resistant hypertension in order to avoid any inadvertent worsening of BP control.39
Guideline 1.6
Most studies that include subgroup analyses of different antihypertensive regimens report that the increase in BP (especially the systolic BP) following a NSAID is more pronounced in patients who are treated with an ACE inhibitor or ARB than other agents.42 43 49 52 74 Two observational studies, with a total of over 3800 patients receiving a NSAID, noted that hypertensive patients treated with ACE inhibitors had a greater increase in systolic BP than those treated with other classes of antihypertensive agents.42 43 In a randomised crossover study, patients receiving lisinopril had a 7–10% increase in BP following NSAID therapy, as compared with 1.1–1.6% when they were receiving amlodipine.74 Although the risk of worsening BP control is substantial, further increase in the dosage of ACE inhibitors or ARBs is probably not effective, and the risk of acute kidney injury is dose dependent.75 76 We therefore suggest that for patients who are receiving ACE inhibitors or ARBs, empirical addition (or increase in the dose) of an antihypertensive agent of a different class be considered when a NSAID is used.
Cardiovascular effects of NSAID therapy
Guideline statements
2.1 We recommend chronic NSAID therapy be avoided if possible in patients at a high cardiovascular risk. [2B]
2.2 For patients at a high cardiovascular risk in whom chronic NSAID therapy cannot be avoided, naproxen and celecoxib can be considered. [2B]
Background
The relation between cardiovascular disease and NSAIDs has been extensively examined in multiple observational studies,70 77–84 systematic reviews and meta-analyses.56 58 85–101 It is important to note that most meta-analyses that examine the absolute cardiovascular risk of NSAIDs extract data from randomised controlled trials (RCTs) reporting cardiovascular events as secondary outcomes. Almost all RCTs dedicated to cardiovascular safety compare different NSAIDs rather than comparing them with placebo or no treatment.102–108
NSAIDs are particularly associated with two types of cardiovascular disease: congestive heart failure and atherosclerotic cardiovascular events. The risk appears to be dose dependent.77 80 85 99 Although two meta-analyses reported that the risk of myocardial infarction is increased during the first 1 to 3 months of treatment,85 99 most studies reporting an excessive cardiovascular risk are referring to long-term treatment,78 80 84 and the duration of NSAID treatment has been found to be independently associated with the risk of cardiovascular events.109 Other cardiovascular effects of NSAIDs are discussed in the online supplementary information 1.
Rationale
Guideline 2.1
Although chronic NSAID treatment is well reported to be associated with an increased risk of heart failure, myocardial infarction, stroke and vascular deaths, the absolute risk is small in the general population77 78 82 85 or in patients with a low cardiovascular risk (eg, with chronic arthritis).80–82 85 91 106 In low risk patients, two RCTs showed that the risk of cardiovascular event was 0.3–0.4% at 6 months,103 and 2.3–2.7% at 34 months108 following NSAID therapy. Similarly, MacDonald et al 106 reported that the risk of a cardiovascular event was 0.86–0.95 per 100 patient-years of NSAID in patients with chronic arthritis. On the other hand, the risk of cardiovascular disease following NSAID therapy is considerable in patients with a high cardiovascular risk84 86 89 or pre-existing cardiovascular disease.70 83 In patients with chronic arthritis and a history of atherosclerotic cardiovascular events, a RCT showed that the risk of cardiovascular event after NSAID therapy for 18 months is 4.4–5.5%.102
Unfortunately, the definition of 'high cardiovascular risk' is variable across different studies,84 104 105 and the use of cardiovascular risk calculators (eg, ACC/AHA pooled cohort equations38) has not been validated in Asian populations. On the balance of available evidence and practicality, we recommend chronic NSAID therapy be avoided in patients with a history of an ischaemic vascular event70 83 89 99 or heart failure.110
Guideline 2.2
The excessive cardiovascular risk associated with NSAIDs is highly variable among different agents, both when they are individually compared, compared with no treatment or placebo70 77 81 84–86 89 94–96 98 99 or directly compared by head-to-head studies57 59 79 87 88 90 102 104 105 107 108 or network meta-analyses.86 89 93 94 97 Although it has been postulated that COX-2 inhibitors have a higher cardiovascular risk than traditional NSAIDs,111 112 the excess risk was largely confined to rofecoxib.111 Two recent randomised trials showed that celecoxib does not confer an excess cardiovascular risk as compared with conventional NSAIDs.106–108 In the SCOT study,106 the risk of a cardiovascular event was 0.95 versus 0.86 per 100 patient-years following treatment with celecoxib and conventional NSAIDs, respectively. In the PRECISION trial,107 108 the incidence of a cardiovascular event in 34 months was 2.3%, 2.5% and 2.7% following treatment with celecoxib, naproxen and ibuprofen, respectively. Among all NSAIDs, two network meta-analyses suggest that the cardiovascular risk of diclofenac is the highest, with a relative risk of 1.1 to 1.2 as compared with other agents.89 97 In contrast, most of the studies suggest that naproxen and celecoxib do not lead to an increase in the risk of cardiovascular events.58 70 79 84–86 90 94 95 98–101 A RCT that compared celecoxib and naproxen showed that the risk of a cardiovascular event in high risk patients is not significantly different between the two drugs.102 Taken together, we recommend that for patients with a history of ischaemic vascular event for whom chronic NSAID therapy is necessary, naproxen and celecoxib should be the preferred drugs.
Renal effects of NSAID therapy
Guideline statements
3.1 We recommend a NSAID be avoided in patients with severe CKD (eGFR <30 mL/min/1.73 m2), or patients with moderate CKD (eGFR 30–59 mL/min/1.73 m2) receiving ACE inhibitors, ARBs or diuretics. [1C]
3.2 If NSAID is to be given to patients at high risk of CKD, we recommend a renal function test be performed if it has not been done in the past 6 months. [2D]
3.3 If a NSAID is necessary for patients with CKD, evidence does not favour the recommendation of any specific drug. (not graded)
3.3.1 We recommend monitoring of renal function 1 week after initiation of a NSAID. [1C]
3.3.2 Regular monitoring of renal function during chronic NSAID therapy is desirable. [2C]
3.3.3 We recommend renal function be checked if fluid retention develops or there are clinical features of renal function deterioration during NSAID therapy. [1D]
Background
Good evidence on the adverse renal effects of NSAIDs is relatively limited, and most published studies are observational.113–121 A number of published systematic reviews and meta-analyses also explored renal safety as an outcome measure.58 73 110 122 123 Two RCTs provided the absolute risk of acute kidney injury (AKI) as secondary outcome measure
after NSAID treatment,45 124 but neither of them included a placebo or untreated group for comparison.
The most important concern about renal safety related to a NSAID is the development of AKI, usually in the form of acute tubular necrosis (ATN) as a result of haemodynamic disturbances and detrimental effects on intrarenal circulation. Acute allergic tubulointerstitial nephritis (TIN) is another classic cause of AKI secondary to a NSAID, but the absolute incidence is probably lower than that of ATN. A proton pump inhibitor (PPI), which is commonly used for peptic ulcer prophylaxis during NSAID therapy, is another common cause of allergic TIN.125 Without a kidney biopsy, it is not always possible to distinguish ATN from allergic TIN by clinical means. However, the management of the two conditions is often similar. Since most of the published studies on the renal risk of NSAIDs do not distinguish between the two pathological causes, we use AKI as a generic term to discuss the risk of acute renal function loss. Other renal effects of NSAIDs are discussed in the online supplementary information 1.
Rationale
Guideline 3.1
Although the use of NSAIDs is associated with the development of AKI in a normal population or in patients with low renal risk,73 110 114 118 123 the absolute risk is low. A meta-analysis of 114 RCTs, with a total of 116 094 patients, showed that the absolute risk of AKI is 0.2% in low risk patients.73 However, pre-existing CKD is an important risk factor of NSAID-induced AKI.123 126 In a meta-analysis of five observational studies, 12.1% of patients with CKD developed AKI following NSAID therapy, with a 63% excess risk as compared with no treatment.123 Moreover, superimposed AKI in patients with CKD is associated with more rapid progression to dialysis-dependent, end-stage renal disease.127–130 In a retrospective study of 618 critically ill patients with AKI, patients with underlying CKD were more likely to be dialysis dependent at hospital discharge.130 In another observational study of 6862 patients with CKD, superimposed AKI was associated with both a higher risk of death (adjusted relative risk 2.32) and dialysis (adjusted relative risk 2.33).127 As a result, we advise against the use of a NSAID in patients with pre-existing CKD.
The risk of NSAID-induced AKI is also substantial among patients receiving ACE inhibitor or ARB therapy.114 119 131 132 In a retrospective study of low risk patients, AKI developed in 1.9% patients receiving combined NSAIDs and ACE inhibitor therapy, but in 0% among patients receiving either drug alone.131 The risk of AKI is particularly high when the patient is also receiving diuretic agents. In a nested case–control study of 78 379 patients, the absolute increase in AKI risk was higher when NSAIDs were used with diuretic agents and an ACE inhibitor or ARB versus NSAIDs with ACE inhibitor or ARB alone but no diuretic agents.132 In another retrospective study of 487 372 hypertensive patients, concurrent use of this combination was associated with 31% increase in the risk of AKI, especially during the first 30 days of treatment.119
The severity of renal impairment for which NSAID should be avoided is not well defined. Möller et al 120 reported that a NSAID is an independent predictor for accelerated renal function decline when baseline eGFR was <30 mL/min/1.73 m2. Moreover, with moderate to severe pre-existing CKD, a mild AKI would be sufficient to tip the balance and mandate dialysis. We recommend a NSAID be avoided in patients with baseline eGFR <30 mL/min/1.73 m2 (ie, Kidney Disease: Improving Global Outcomes (KDIGO) stage 4 or 5 CKD), or in patients with baseline eGFR 30–59 mL/min/1.73 m2 (ie, KDIGO stage 3 CKD) who are receiving ACE inhibitors or ARBs. Although the literature is limited, we believe our recommendations also apply to kidney allograft recipients. The risk of AKI following NSAID is considerable in kidney allograft recipients with renal function impairment or those receiving a calcineurin inhibitor.133 Although AKI is generally not a concern in patients undergoing dialysis, residual renal function is an important predictor of mortality and morbidity.134 It is reasonable to avoid NSAIDs in patients undergoing dialysis to avoid worsening of residual renal function.135
Guideline 3.2
Since mild to moderate CKD is usually asymptomatic,136 137 it is important to screen for undiagnosed CKD before initiation of a NSAID. Screening for CKD for the general public or low risk patients is not necessary because the yield is low, and most published guidelines recommend targeted screening for CKD in high risk populations.137–141 Although the definition of 'high risk group' is different between guidelines, it generally encompasses elderly people, patients with underlying hypertension, diabetes, cardiovascular disease, heart failure, urological diseases, systemic autoimmune diseases and those with a family history of CKD.138–141 We recommend that if a NSAID is to be given to patients at risk of CKD, a baseline renal function test should be performed if it has not been done in the past 6 months. Since serum creatinine level may not rise above the upper limit of normal until over 50% of renal function has been lost, we suggest following the KDIGO recommendations to use a GFR estimating equation to derive GFR.139
Guideline 3.3
The relative risk of AKI among different NSAIDs has been studied in several observational studies and meta-analyses, but the results are conflicting.73 110 118 There is no consistent difference in renal risk between conventional NSAIDs and COX-2 inhibitors.110 118 In a RCT that compared celecoxib and diclofenac, the risk of AKI was 5.6% and 6.3%, respectively (P=0.8).45 In another study that compared celecoxib and naproxen, the risk of raised serum creatinine was 3.3% and 2.5%, respectively.124 We conclude that the evidence does not allow the recommendation of any specific NSAID for patients with CKD.
Guideline 3.3.1
Monitoring of renal function after initiation of a NSAID is logical, but few publications describe the optimal protocol for such monitoring. AKI is associated with the use of a NSAID in the past 1 to 30 days.114 Theoretically, ATN due to the inhibition of prostaglandin synthesis and haemodynamic effect of a NSAID, should develop soon after treatment, especially in patients whose kidney perfusion is dependent on local prostaglandin.142 143 On the other hand, it takes 7 to 10 days for the body to mount a cell-mediated immune response for allergic TIN in unsensitised patients.144 145 On balance, we suggest that renal function be monitored 3 to 7 days after the initiation of a NSAID in patients with CKD.
Guideline 3.3.2
Although a NSAID typically induces AKI shortly after the initiation of treatment, patients who receive long-term NSAID therapy (eg, for the treatment of chronic arthritis) continue to have an increased risk of AKI months or years later.58 73 123 Common precipitating factors include concurrent infections, dehydration, concurrent diuretic agent, ACE inhibitor or ARB therapy and radiocontrast exposure.146 In addition, a NSAID may also increase the rate of renal function decline without causing AKI in patients with pre-existing CKD (see online supplementary information 1). We suggest regular renal function monitoring by eGFR in patients receiving chronic NSAID therapy, particularly when there are other predisposing factors to AKI.
Guideline 3.3.3
Renal function should be checked when there are clinical features of AKI. However, oliguria and overt uraemic symptoms are late features. Fluid retention and pedal oedema are often present in AKI.73 147 Peripheral oedema itself is a risk factor for AKI development. In a cohort study of 12 778 critically ill patients, the presence of peripheral oedema on admission was associated with an increased risk of AKI within the first 7 days of critical illness.148 The presence of 1+, 2+ and 3+ oedema was associated with 17%, 47% and 57% higher risk of AKI, respectively, as compared with oedema-free patients.148 Raised BP is also commonly noted in patients with drug-induced ATN.149 150 We recommend renal function testing if fluid retention develops or there are clinical features of renal function deterioration during NSAID therapy.
Gastrointestinal effects of NSAID therapy
Guideline statements
4.1 For the treatment of NSAID-related dyspepsia:
4.1.1 We recommend the use of a PPI. [1B]
4.1.2 We do not recommend the use of antacids or H2 receptor antagonists. [1C]
4.2 For the prevention of upper GI complications:
4.2.1Wwe recommend a non-selective NSAID plus a PPI or selective COX-2 inhibitor monotherapy in patients with moderate risk of peptic ulcer disease. [1A]
4.2.2 We recommend a selective COX-2 inhibitor plus a PPI in patients with high risk of peptic ulcer disease. [1A]
4.3 In patients with unexplained iron-deficiency anaemia who require a NSAID:
4.3.1 We recommend gastroenterologist referral. [1D]
4.3.2 We recommend the use of celecoxib over other NSAIDs. [1A]
Background
NSAIDs are associated with an increased risk of upper and lower GI adverse events.45 46 53 58 90 97 102 103 124 151–181 The most common GI adverse event is dyspepsia. The most serious GI adverse event is peptic (gastric or duodenal) ulcer with bleeding. However, the correlation between dyspeptic symptoms and the presence of peptic ulcer is poor, and the need for ulcer prophylaxis depends on risk factors rather than symptoms. The absolute risk of ulcer and its complication following a NSAID is also substantially affected by the Helicobacter pylori status182–185 and concomitant use of aspirin or other antithrombotic drugs.153 186–188 Other gastrointestinal effects of NSAID are discussed in the online supplementary information 1.
Rationale
Guideline 4.1
Guideline 4.1.1
Use of a NSAID is associated with dyspepsia, abdominal pain and a variety of non-specific abdominal symptoms.152 165 168 However, the correlation between dyspepsia and endoscopically documented pathology is poor. Most clinical trials on NSAID-related upper GI problems focus on endoscopically confirmed ulcer or bleeding, and few data have been published on the treatment of NSAID-related dyspepsia. Functional dyspepsia, however, is common in the general population even without receiving a NSAID, and a subset of these patients is related to increased mucosal sensitivity to acid.189 A review of 15 RCTs that compared PPI therapy with placebo for the treatment of functional dyspepsia, showed a significant improvement in dyspepsia symptoms.190 Although head-to-head data comparing a COX-2 inhibitor with a traditional NSAID plus PPI are limited, indirect evidence suggests that PPI is the preferred treatment for NSAID-induced dyspepsia.190–192 A meta-analysis of 30 studies showed that a traditional NSAID plus PPI reduced the risk of dyspepsia by 66% as compared with a traditional NSAID alone, while COX-2 inhibitors alone reduced the risk by 12% only.191
Guideline 4.1.2
Treatment strategies other than PPI are commonly used for the treatment of NSAID-related dyspepsia. However, published data on their efficacy is scarce, and no head-to-head clinical trial has compared antacids or H2 receptor antagonists with PPI. A systematic review of six RCTs showed that PPIs were significantly more effective than either H2 receptor antagonists or antacids for the treatment of non-specific dyspepsia.193 Although antacids may alleviate abdominal symptoms, the risk of peptic ulcer and its complications is not reduced.156 Although the efficacy of H2 receptor antagonists is inferior to that of PPIs, prescription of H2 receptor antagonists remains a common practice, partly because of the reported risks of dementia, electrolyte disturbances, renal damage, infections and osteoporosis in chronic PPI users.194–196 However, the actual incidences of the above-mentioned complications are small. The pooled result of two RCTs showed that PPI is significantly more effective than prokinetics for the treatment of functional dyspepsia.197–199 On the balance of risk and efficacy, we conclude that antacids or H2 receptor antagonists are not recommended for the treatment of NSAID-related dyspepsia.
Guideline 4.2
Guideline 4.2.1
Many placebo-control trials have shown that a PPI reduces the risk of NSAID-induced peptic ulcer.30 In a combined publication of two RCTs, ulcer rates were 17.0%, 5.2% and 4.6% for the placebo, esomeprazole 20 and 40 mg groups, respectively.200 In another case–control study, PPI therapy was associated with a significant reduction in the risk of upper GI bleeding.188 201 Esomeprazole is effective in preventing ulcer recurrence in both H. pylori-positive and -negative NSAID users,199 and long-term esomeprazole treatment is well tolerated and efficacious for preventing ulcer recurrence in NSAID users.197 202 Most studies used omeprazole or esomeprazole for ulcer prophylaxis,45 46 102 160–162 164 171 176 but other PPIs are likely to be similarly effective.124 180 203 Vonoprazan, a potassium-competitive acid blocker that inhibits H+/K+-ATPase, is non-inferior to lansoprazole for the prevention of ulcer recurrence during long-term NSAID therapy,204 but more data are needed.
Other drugs have been used for ulcer prophylaxis, but are generally less effective than PPIs. H2 receptor antagonists may reduce the risk of peptic ulcer in NSAID users,159 169 but the evidence is controversial.156 A RCT showed that high-dose famotidine significantly reduced the incidence of both gastric and duodenal ulcers in chronic NSAID users.169 A pooled analysis showed that famotidine prophylaxis resulted in 44% reduction in upper GI ulcers in ibuprofen users, and was equally effective in patients who also received low-dose aspirin.159 Two economic modelling studies suggested that treatment with H2 receptor antagonists is a cost-effective strategy for the prevention of ulcer bleeding in NSAID users.205 206 However, another study showed that antacids or H2 receptor antagonists did not significantly lower the risk of serious GI complications.156 A meta-analysis of 14 trials found that H2 receptor antagonists were protective at high doses, but at usual doses they reduced the risk of duodenal, but not gastric, ulcers.207 Similarly, studies have shown that misoprostol reduces the risk of GI complications in NSAID users.208–210 A network meta-analysis of seven studies found that misoprostol plus non-selective NSAID was associated with a significantly lower risk of GI events.211
However, H2 receptor antagonists and misoprostol are generally less effective than PPIs for ulcer prophylaxis. Two RCTs showed that omeprazole significantly reduced the risk of NSAID-related ulcers in comparison with ranitidine212 or misoprostol.213 Another RCT showed that lansoprazole was more effective than gefarnate (a geranyl ester of farnesylacetic acid with mucosal protective properties) in reducing the risk of peptic ulcer recurrence in long-term NSAID users.199 A network meta-analysis showed that a H2 receptor antagonist plus a traditional NSAID is associated with a higher risk of ulcer complication than PPI plus a NSAID (relative risk 3.09, 95% CI 1.39 to 6.30), or selective COX-2 inhibitor alone (relative risk 3.52, 95% CI 1.38 to 7.56).211 Taken together, given the superior efficacy of PPIs, H2 receptor antagonist or misoprostol prophylaxis are not recommended for ulcer prevention in NSAID users.
Many RCTs and meta-analyses showed that selective COX-2 inhibitor monotherapy is as effective as a non-selective NSAID plus PPI for reducing the risk of ulcer in patients with a moderate risk of peptic ulcer disease.45 46 124 160 161 164 171 176 180 Most published evidence on selective COX-2 inhibitors has concerned celecoxib,45 46 124 160 161 164 171 but other COX-2 inhibitors probably have similar GI effects.58 176 179–181 As a result, we recommend monotherapy with either non-selective NSAID plus a PPI or a selective COX-2 inhibitor in patients with a moderate risk of peptic ulcer disease. However, the definition of 'moderate risk' could be arbitrary. In elderly patients with no history of ulcers, or those with a history of cardiovascular events, concomitant use of PPI was associated with a two-thirds reduction in the risk of serious NSAID ulcer complications.214 215 As a result, we define moderate risk of peptic ulcer disease as possession of one or two of the following risk factors: age over 65 years; previous history of uncomplicated ulcer; concurrent use of aspirin (low and high dose), antiplatelet drugs, corticosteroids or anticoagulant agents. It remains unclear whether low risk patients (ie, young patients with no history of ulcer and cardiovascular disease) require ulcer prophylaxis.
Guideline 4.2.2
Although monotherapy with a non-selective NSAID plus a PPI or a selective COX-2 inhibitor reduces the risk of peptic ulcers and their complications, the absolute risk (primarily GI bleeding) remains substantial in patients with a high risk of peptic ulcer disease, which is defined as patients with a history of a previously complicated ulcer. A RCT showed that combination treatment with omeprazole and celecoxib was more effective than celecoxib alone for prevention of recurrent ulcer bleeding in patients at high risk.162 In another RCT of patients at high risk of both cardiovascular and GI events, who required concomitant aspirin and NSAID, the cumulative incidence of recurrent ulcer bleeding at 18 months was 5.6% in the celecoxib plus omeprazole group and 12.3% in the naproxen plus omeprazole group.102 We conclude that a selective COX-2 inhibitor plus a PPI should be recommended for patients with a high risk of peptic ulcer.
Guideline 4.3
Guideline 4.3.1
For the general public, without the consideration of a NSAID, routine screening for iron-deficiency anaemia in asymptomatic pregnant women is recommended by several professional organisations.216–218 In contrast, routine screening is not recommended for asymptomatic men and postmenopausal women.219 However, testing for iron deficiency should be performed in patients with signs and symptoms of anaemia, and a complete evaluation should be performed if iron deficiency is confirmed.220 When abnormal uterine bleeding is excluded, most iron-deficiency anaemia is caused by GI pathologies.220–223 Referral to a gastroenterologist for endoscopic investigation is recommended.220 Specifically, all adult men and postmenopausal women with iron-deficiency anaemia should be screened for GI malignancy.220 223 224 The same recommendations should be applicable to patients with unexplained iron-deficiency anaemia and who need NSAID treatment, although this recommendation is based on extrapolation of evidence in general medical patients.
In addition, NSAID increases the risk of peptic ulcer bleeding,151–153 156 lower GI haemorrhage due to diverticulosis and other colonic pathologies,90 152–155 157 169 172 177 178 as well as bleeding from a small intestinal mucosal injury.225–227 Around 10–15% of patients have iron-deficiency anaemia related to long-term use of aspirin or other NSAIDs, presumably as a result of low-grade GI haemorrhage.220 222 223 A GI investigation would also seem appropriate if iron-deficiency anaemia develops in patients during NSAID therapy.
Guideline 4.3.2
Around 10–15% of patients have iron-deficiency anaemia related to long-term use of aspirin or a NSAID, probably due to low-grade GI haemorrhage.220 222 223 As discussed above, monotherapy with either a non-selective NSAID plus a PPI30 198–201 228 or a selective COX-2 inhibitor45 46 124 160 161 164 171 176 180 reduces the risk of upper GI bleeding. However, a PPI does not prevent NSAID-related lower GI complications, but instead increases the risk of microscopic colitis.229 Acid suppression with PPIs induces gut dysbiosis,229 which exacerbates NSAID-induced small intestinal injury.230–232 On the other hand, selective COX-2 inhibitors are associated with a lower risk of lower GI complications than traditional NSAIDs. In one RCT of 8076 patients with rheumatoid arthritis, the risk of a lower GI event was significantly lower in patients treated with rofecoxib than in those treated with naproxen (0.41 vs 0.89 per 100 patient-years).166 A systematic review showed that, in comparison with traditional NSAIDs, selective COX-2 inhibitors had a significantly lower risk of any lower GI events, endoscopic mucosal injury and overt lower GI bleeding.178 Although all selective COX-2 inhibitors probably have similar GI effects,58 176 179–181 most published evidence on selective COX-2 inhibitors are studies on celecoxib.45 46 124 160 161 164 171 We therefore recommend celecoxib in patients with unexplained iron-deficiency anaemia who require a NSAID.
Limitations
Although this set of guidelines is meant to be evidence based, many of our recommendations are derived from expert opinions supported by limited published data. We also pay attention to relevant published guidelines by other professional bodies (eg, the hypertension guidelines by the ACC/AHA), even though they may not be directly referring to the use of NSAIDs. It is important to note that for the effect on BP and renal risk, our recommendations focus on the initiation of NSAID and the short-term risk. On the other hand, the discussion on cardiovascular risk largely refers to long-term treatment.
One should be cautious with the quality of evidence. Although there are many large observational studies, often from well-established healthcare databases, on the risk of NSAID, the patient populations are substantially different between studies, so that a coherent and consistent conclusion is not always possible. There is also a paucity of observational data on the long-term effect on BP and GI risk. The relation between long-term NSAID therapy and CKD is another area of controversy because many of the comparators in observational studies (notably paracetamol) may themselves have long-term renal toxicity properties themselves.
Most published RCTs on NSAIDs compare two drugs (typically a COX-2 inhibitor, with a traditional NSAID as the control) rather than comparing with placebo or no treatment, and report relative risks rather than absolute ones. Many of the trials focused on efficacy and report adverse effects (particularly cardiovascular and renal events) as secondary or safety outcome measures.
Problems of implementation
Although cardiovascular, renal and GI complications of NSAIDs are increasingly recognised, and we have guidelines with reasonable scientific basis, there are major problems of implementation. In essence, the above-mentioned complications are generally taken care of by cardiologists, nephrologists and gastroenterologists, but NSAIDs are usually prescribed by primary care physicians, rheumatologists or surgeons. Guidelines on the appropriate use of NSAID and prevention of complications should be made known to these specialists. Appropriate education for the relevant specialists would be important, and the combined effort of various specialty societies and professional bodies would be necessary. With the availability of joint guidelines, multidisciplinary symposia in major conferences are potential ways to increase the awareness of cardiovascular, renal and GI complications and the appropriate use of NSAIDs. However, with the involvement of no fewer than six specialties and each having its own national society or professional college, the effort to coordinate and mediate the difference in priority between different specialties and various countries would be a Herculean task.
Perspective
The recommendations are summarised in box 1. It is important to realise that although we tried to keep an evidence-based approach while preparing the recommendation statements, published clinical trials are often patchy and may not directly answer a practical clinical question. Notably, there are many excellent RCTs and meta-analyses on cardiovascular and GI complications, so that our recommendations are often strong and confident. On the other hand, published trials are limited with regard to BP and renal complications of NSAIDs, and our recommendations in these areas are generally based on observational data supplemented with secondary outcomes of a few randomised trials. There is a particular knowledge gap with regard to the monitoring after the initiation of NSAIDs. Our current recommendations on BP and renal function monitoring are largely expert opinions. Although anaemia is a major manifestation of GI complications of NSAIDs, there is no published data on any strategy of haemoglobin monitoring. Further research is urgently needed to clarify the optimal strategy and schedule of patient monitoring in these areas.
Summary of recommendations
(A) Avoid NSAID if possible
treatment-resistant hypertension
high cardiovascular risk
patients with severe CKD (eGFR <30 mL/min/1.73 m2), or patients with moderate CKD (eGFR 30–59 mL/min/1.73 m2) receiving ACE inhibitors, ARBs or diuretic agents
(B) Before initiation of a NSAID
measure blood pressure
high risk of CKD: check renal function if not carried out in the past 6 months
unexplained iron-deficiency anaemia: refer patient to a gastroenterologist
(C) Choice of NSAID and concomitant therapy
high cardiovascular risk and NSAID use cannot be avoided: consider naproxen or celecoxib
pre-existing hypertension and receiving ACE inhibitors or ARBs: empirical addition (or increase in the dosage) of an anti-hypertensive agent of a different class
NSAID-related dyspepsia: PPI
moderate risk of peptic ulcer disease: non-selective NSAID plus PPI or selective COX-2 inhibitor monotherapy
high risk of peptic ulcer disease: selective COX-2 inhibitor plus PPI
unexplained iron-deficiency anaemia and NSAID use cannot be avoided: consider celecoxib
(D) Monitoring during NSAID treatment
blood pressure
no pre-existing hypertension: measure clinic BP 4 weeks after initiation of NSAID (or earlier if clinically indicated)
pre-existing hypertension: monitor BP even with short-term NSAID use
home BP monitoring, self-monitoring of fluid retention by body weight and checking for pedal oedema are desirable
renal function
all patients: check renal function if fluid retention or clinical features of renal function deterioration
pre-existing CKD: check renal function 1 week after initiation of a NSAID, follow by regular monitoring during chronic NSAID use
References
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Footnotes
C-CS and KS contributed equally.
Correction notice This article has been corrected since it published Online First. The corresponding author details have been updated and affiliations 14 amended.
Contributors KS, KF and FKLC are responsible for the literature review and statement preparation of the gastroenterology section. JGW, CHC and JBP are responsible for the literature review and statement preparation of the cardiovascular and hypertension sections. CCS, GKM and KV are responsible for the literature review and statement preparation of the renal section. SW and LST are responsible for overall literature review and inter-disciplinary statements. KT is responsible for primary literature search and final proof of the manuscript. CCS, KS and FKLC are responsible for manuscript writing.
Funding This work was supported by unrestricted educational grants from Pfizer Inc. and the Faculty of Medicine, the Chinese University of Hong Kong. The funders had no role in the study design, data collection and analysis, decision to publish or preparation of the manuscript.
Competing interests C-CS is an adviser or consultant for Baxter Healthcare and Gilead Sciences; speaker for Baxter Healthcare, AstraZeneca, Pfizer Inc. KS reports conflict of interest with Takeda Pharmacol Inc. and Astra-Zeneca Pharma. J-GW was supported by grants from the National Natural Science Foundation of China (91639203) and State Ministry of Science and Technology (2018YFC1704902), Beijing, China and the Shanghai Commissions of Science and Technology (15XD1503200) and Health (15GWZK0802 and a special grant for 'leading academics'), Shanghai, China. J-GW also reports receiving lecture and consulting fees from Astra-Zeneca, Bayer, Daiichi-Sankyo, MSD, Novartis, Omron, Pfizer, Sanofi, Servier and Takeda. C-HC is an adviser or consultant for Novartis Pharmaceuticals Corporation; speaker or member of a speakers bureau for AstraZeneca, Pfizer Inc., Bayer AG, Bristol-Myers Squibb Company, Boehringer Ingelheim Pharmaceuticals Inc., Daiichi Sankyo Inc., Novartis Pharmaceuticals Corporation, SERVIER, Merck & Co. Inc., Sanofi, TAKEDA Pharmaceuticals International. FKLC reports speaker's honoraria from AstraZeneca, Pfizer, Eisai and Takeda.
Patient consent for publication Not required.
Provenance and peer review Not commissioned; externally peer reviewed.
Data availability statement All data relevant to the study are included in the article or uploaded as supplementary information.