Background Medical management of adults with osteoarthritis (OA) who require non-steroidal anti-inflammatory drugs (NSAIDs) must be decided after assessing prevalent gastrointestinal (GI) and cardiovascular (CV) risks in the individual patient.
Objective To evaluate the GI and CV risk profile of patients with OA who require NSAIDs.
Methods A transversal, multicentre and observational study was conducted in consecutive patients with OA who were considered candidates for NSAID treatment and were visited by 374 unselected rheumatologists throughout the National Health System. Patients were classified into three risk groups (low, moderate and high) for their GI and CV characteristics. These were defined by considering the presence of a number of well-established GI risk factors or by application of the Systematic Coronary Risk Evaluation model for assessing the overall risk for CV disease, respectively.
Results Of 3293 consecutive patients, most (86.6%) were at increased GI risk and a considerable number, 22.3%, were at high GI risk. The CV risk was high in 44.2% of patients, moderate in 28.5% and low in 27.3%. Overall, 15.5% of patients presented a very high-risk profile, having high GI and CV risks. The type of NSAID prescription was similar regardless of the associated GI and CV risk profile.
Conclusion Most patients with OA requiring NSAIDs for pain control showed a high prevalence of GI and CV risk factors. Over half of the patients were at either high GI or CV risk, or both, such that the prescription of OA treatments should be very carefully considered.
Statistics from Altmetric.com
Osteoarthritis (OA) is the most common disease of the joints. It is a chronic condition that has periods of acute pain and affects quality of life owing to impairment of normal functional activity. Environmental risk factors for OA include trauma, nutritional factors, obesity, genetic factors and specific occupational and other activities.1 Therapeutic intervention focuses on alleviating pain and improving activity by pharmacological and non-pharmacological approaches.2 3
One of the most common types of pharmacotherapy used to relieve pain and inflammation in OA is non-steroidal anti-inflammatory drugs (NSAIDs). NSAIDs can be divided into the so-called non-selective NSAID (ns-NSAIDs) and cyclo-oxygenase 2 (COX-2) selective inhibitors (coxibs). Both are effective painkillers and anti-inflammatory agents widely used in the treatment of OA, although they are also associated with adverse effects, including gastrointestinal (GI) and cardiovascular (CV) damage. It is well established that ns-NSAIDs increase the risk of serious GI adverse events.4 Coxibs have a better GI safety profile than ns-NSAIDs,5 but data from long-term clinical trials show that they can be associated with CV risk, which prompted the need to assess whether ns-NSAIDs may also have a similar deleterious CV effect. Epidemiological studies and meta-analyses of randomised controlled trials and observational studies suggest that this is the case,6,–,9 adding to the current knowledge of the effects of ns-NSAIDs on CV parameters, such as blood pressure. In this respect, whereas the US Food and Drug Administration clearly states that the CV risk is associated with all NSAIDs (excluding aspirin),10 the European Medicines Agency is more cautious and indicates that although the overall benefit–risk balance for ns-NSAIDs remains favourable when used appropriately, the possibility of CV adverse events cannot be excluded.11 However, the two agencies agree that the CV risk associated with NSAIDs and coxibs increases with high dose and long-term treatment and therefore recommend that the lowest effective dose for the shortest period of time should be prescribed.
The European League Against Rheumatism and the American College of Rheumatology offer guidelines for the treatment of OA with similar objectives but different recommendations. However, they both agree that medical management of adults with OA who require NSAIDs must be decided after assessing GI and CV risks in individual patients. It has been suggested that the appropriate selection of NSAIDs must be driven by the assessment of GI and CV risks of the individual patient, rather than by a drug's reported rate of adverse events in heterogeneous populations.12 Initial evaluation of GI risk must include determination of the presence of well-known risk factors such as age, previous history of GI bleeding, peptic ulcer and/or GI symptoms (dyspepsia), the use of high-dose NSAIDs, the combination of two NSAIDs including low-dose aspirin, or the concomitant use of NSAIDs and anticoagulants or corticosteroids. The presence of Helicobacter pylori infection is also widely accepted as another factor increasing the risk of upper GI complications, but it is rarely assessed in clinical practice. Likewise, the CV risk profile must be determined before prescribing any drugs. To unify criteria when assessing CV risk, the Systematic Coronary Risk Evaluation (SCORE) project proposed a risk scoring system aimed at estimating the total CV risk rather than the risk of coronary heart disease. The model is based on European studies using parameters that distinguish between areas with high or low incidence of CV disease, with Spain in the low incidence zone.13 Current European guidelines on CV disease support the use of this system as an appropriate tool for evaluating the CV risk of individual patients.14 However, some updating may be required after a recent report suggested that the changing pattern of CV mortality within Europe called for reconsideration of such classification.15
Although GI and CV risk factors for severe events in patients with OA using NSAIDs are well established and firm recommendations have been made as to which type of drugs should be prescribed according to the presence of these risk factors, information about how frequent or prevalent these factors are in clinical practice is very limited. Therefore, our study aimed at evaluating the GI and CV risk profiles in patients with OA who are candidates for NSAID treatments.
Samples and design
We conducted a transversal, multicentre, observational study in patients with OA who needed NSAID treatment. Data from patients were collected from the 2 October through to the 8 October in 2006. A total of 374 rheumatologists from the National Health System throughout the Spanish territory collaborated in the recruitment of patients and collection of data. The large number of doctors and geographical separation of medical centres ensured that the study was homogeneous and representative of the Spanish OA population. The sample comprised 3293 consecutive patients who were considered candidates for NSAID treatment for their underlying disease. To obtain all the necessary data without interfering with normal clinical practice, the study was designed to collect all the information in a single visit. Patients were given treatment and/or medical care considering exclusively the criteria of their doctor, and particular therapeutic strategies were determined according to their normal clinical or assistant practice. Patients accepted for the study had to meet the following inclusion criteria: (1) female/male patients with OA aged >18 years; (2) not responding to peripheral analgesics (eg, paracetamol, metamizole); (3) having pain interfering with normal-day activities (ie, visual analogue scale >50 or verbal numeric rating scale >5); (4) having an indication for NSAID treatment; (5) with a minimum 6-month-old clinical history record; (6) with recorded blood chemistries within the previous 6 months; (7) having verbal consent from the patients or their legal representatives. Exclusion criteria were age <18 years and a patient's refusal to participate in the study.
Outcome measure and GI/CV risk classification
The primary objective was to evaluate the GI and CV risk profile of patients who had an indication for NSAID treatment. Based on available evidence, the following GI risk factors were considered16,–,21: age >60, concomitant acetylsalicylic acid (ASA), corticosteroid or anticoagulant use, previous ulcer history, previous ulcer bleeding history, history of dyspepsia, use of two NSAIDs or a high dose of one NSAID (a dose was considered high if the treatment with the chosen NSAID was the maximum dose recommended for the symptomatic treatment of arthritic pain—for example, diclofenac ≥150 mg/day, aceclofenac ≥100 mg/day, meloxicam ≥15 mg/day, naproxen ≥1000 mg/day, piroxicam ≥20 mg/day; ibuprofen >1800 mg/day).22 Owing to the nature of the study, H pylori infection was not considered here, since the proportion of patients with this information was very low. ASA was considered a low dose for any CV prescription at ≤300 mg/day.
For the purpose of this study and based on previous reports,20 23 24 we classified patients into three GI categories. Patients at low GI risk were considered to be those without any of the above-mentioned clinical conditions. Patients at moderate GI risk included patients with at least one of the following GI risk factors: (1) age >60 years; (2) concomitant use of low-dose ASA; (3) concomitant use of corticosteroids; (4) history of symptomatic peptic ulcer; (5) history of dyspepsia; (6) current high-dose of NSAID; (7) use of two NSAIDs. Patients at high GI risk were considered to be those with either a GI bleeding history, concomitant use of NSAIDs and anticoagulants or the presence of three of the risk factors described for moderate GI risk. These combinations were based on the overall relative risk obtained from combining different risk factors that might put patients at a similar risk to that seen for patients with a bleeding peptic ulcer history.20 23 In this way, for patients with low GI risk, the expected rate of upper GI complications should not exceed 1.5 events per 100 patients/year, whereas for those at moderate GI risk, the rate should be between 1.5 and 10 and for those at high GI risk, the rate should be greater than 10 events per 100 patients/year. It was considered that patients being treated with anticoagulants (warfarin or coumadin) and NSAIDs should be considered high risk, since bleeding events in discoagulated patients can be associated with more severe bleeding.24
CV risk was also grouped into low, moderate and high, according to the SCORE marking system, which provides an estimation of the 10-year risk of fatal CV disease stratification in the primary prevention of CV disease.13 Briefly, a six-step method is used that combines a calculation of the 10-year risk for coronary heart disease and for non-coronary CV disease, together with the presence of CV risk factors, such as smoking, cholesterol and high systolic blood pressure. A numerical value is obtained whereby CV risk is estimated as low when <2%, moderate 2–5% and high >5%. Diabetes and previous CV diseases (acute myocardial infarction, angina and ictus) are not included in the SCORE, these patients being categorised directly in the group at high CV risk. Prescribing doctors who wish to calculate the total CV risk in their patients can use simple charts.14 In order to calculate SCORE, we used the most recently available laboratory data between the day of the visit and the previous 6 months, whereas other clinical data were collected from the patient's chart (eg, disease history) and/or from the current clinical visit (eg, smoking). Blood pressure levels were obtained at the clinical visit and with those registered in the chart, if available. If the values were not within the expected range of those obtained at the clinical visit, or were not explained by the antihypertensive drug treatment, new blood pressure measurements were performed to confirm values at the clinical visit.
Patients were considered to receive a prescrition of COX-2 selective NSAIDs if they received either celecoxib or etoricoxib alone; all other prescribed NSAIDs were considered non-selective.
The descriptive analysis of the patients included demographic and clinical characteristics, GI and CV risk factors, pharmacological and non-pharmacological treatments. Quantitative and qualitative variables were analysed using measurements of central tendency (mean, median) and of dispersion (95% CI). Further, qualitative variables were defined according to their absolute and relative frequencies. The mean and SD are reported for continuous variables and percentages for qualitative variables. Data were analysed with SAS 8.2 statistical software (SAS Institute, Cary, North Carolina, USA).
The study complied with all ethical considerations involving human subjects, as adopted in the 18th World Medical Assembly, Helsinki, Finland. All information recorded was obtained following standard clinical guidelines, and patients were not subjected to any therapeutic or diagnostic experimentation. The study followed standard security and confidentiality measures, complying fully with Spanish legislation on data protection (Ley Orgánica de 15/99). The work was presented to, and approved by, the Regional Ethics Committee for Clinical Research, Hospital San Carlos (Madrid). The name of the patients remained confidential, with an identification number use instead.
Characteristics of patients and pharmacological treatments
The sample comprised 3293 subjects whose demographic and clinical characteristics are summarised in table 1. Their mean age was 64.7 years and the majority of patients were women (73.2%). The mean time from diagnosis was 5.4 years, with 66.6% of patients diagnosed with axial OA of the spine and 81.1% with peripheral OA of the hand, hip and knee.
Of the pharmacological treatments, ns-NSAIDs or coxibs, alone or combined with other OA drugs, were taken by 2716 (82.5%) of patients, whereas 577 (17.5%) did not receive NSAID medication, but paracetamol alone (6.7%) or combined with other drugs (opioids, symptomatic slow-acting drugs in OA (SYSADOA), metamizole) (3.6%). OA medication, including opioids, SYADOA and metamizole, without paracetamol was received by 7.2% of patients. Paracetamol was the most frequently prescribed drug, followed by ibuprofen and diclofenac (table 2). Concomitant treatments included gastroprotective agents (837/3293; 25.4%), antihypertensive drugs, treatments for diabetes, diet supplements (data not shown), and low-dose ASA, which was used by 14.1% of the patients.
GI and CV risk profile of patients
Age (>60 years), history of dyspepsia, high-dose NSAID and co-therapy with aspirin use were the most common risk factors, whereas 8.5% of patients had a history of uncomplicated peptic ulcer and smaller percentages of patients presented other GI risk factors (table 3). More than 86% of the patients had one or two GI risk factors, and a remarkable 20% had three or more. The overall GI risk profile was established for every individual patient according to the number of GI risk factors; and subjects were classified into low, moderate or high GI risk groups. The majority of patients (86.6%) fitted into the increased GI risk group, with a significant 22.3% within the high GI risk group, and only 13.4% were classified as having low GI risk (table 3).
CV risk factors were defined according to age, smoking habits, CV history and biochemical and CV parameters. Thus, the median age was 65 years, 25% of the patients were smokers or ex-smokers, 23% were diabetic and 20.5% had a history of CV events (table 1). The mean±SD values for a number of risk factors and biochemical CV parameters such as systolic blood pressure, diastolic blood pressure, cholesterol, low-density lipoprotein, high-density lipoprotein, triglycerides, glucose and HbA1c are summarised in table 4. The risk of death based on the SCORE model is shown in table 4. Stratification combining CV risk factors and the SCORE value placed 44% of the patients at a high CV risk, and only a 27% at low risk (table 4). Aspirin was prescribed in 14% of patients, but most of them with either moderate or high CV did not receive aspirin, and only 35.6% of those with a previous CV event did receive aspirin (table 5). In addition, among patients treated with low-dose ASA, 23% also received ibuprofen treatment.
GI and CV risk combined
When combining GI and CV risk into low, moderate or high risk, it must be noted that only 10% of patients fell within the low GI and CV risk group, whereas the vast majority were moderate or high GI and CV risk. Of note, 15.5% of patients receiving NSAIDs had high GI and CV risk (tables 6 and 7). The proportions and types of NSAIDs (ns-NSAIDs, coxibs) or analgesics prescribed for each subcategory did not change significantly based on the presence of either GI or CV risk.
The criteria to recommend drug treatment for symptomatic musculoskeletal pain or anti-inflammatory drugs in patients with OA depend as much on the safety profile of the drug as on their individual characteristics, with particular attention to their GI and CV risk profile. Thus, clear guidelines as to how to measure these risk profiles are essential before recommending any particular therapeutic strategy, especially when pharmacological approaches are required.16 18 25,–,30 However, very limited or no information is available about the proportion of patients with OA who carry GI and CV risk factors in clinical practice.
This study found that the vast majority of patients with OA who had an indication for NSAID treatments were at increased risk of having a GI complication, since >86% of them had one or more GI risk factor and almost one quarter of them should be considered at high risk. This figure is even higher when considering CV risk, since over 40% of patients included in the study were at a high risk of developing CV events. This is in contrast with current estimations that place Spain within a low-incidence area for CV disease according to the SCORE system of classification.13 These differences, however, must not be seen as contradictory, since our study refers exclusively to the Spanish OA population (commonly found among older people and therefore with an increased probability of CV risk due to age) and cannot be taken as representative of the whole country. Still, we found that the percentage of patients with a high CV risk was remarkable, which was probably the consequence of using the SCORE system. We chose to use this method following the European recommendations on CV disease prevention,14 and because it was developed as an integrated system that represents more accurately the total risk of CV disease, as opposed to the traditional list of CV factors.
The challenge with these patients is to recommend the appropriate therapeutic strategy after considering their individual GI and CV risk profile. Guidelines agree that non-pharmacological measures should be central to the management of OA, with pharmacological agents used as adjunctive treatments.27,–,29 Drugs, however, are usually prescribed in normal clinical practice. In our study, >80% of patients were prescribed ns-NSAIDs, or coxibs, with paracetamol, opioids and other drugs completing the list of pharmacological treatments. Paracetamol is usually recommended as the initial oral drug to treat OA.2 However, paracetamol is still questionable for managing OA symptoms in the long term, owing to the lack of anti-inflammatory activity, the lower efficacy compared with NSAIDs and because it may not be devoid of CV31 and GI toxicity, which includes dyspepsia, liver damage and even GI complications, especially at a high dose.30 32
We report that almost 90% of patients with OA in this study were found to have GI risk factors, with age as one of the main factors, which is consistent with previous reports.33 We also report that almost one quarter of patients should considered at high risk of complications owing to the existence of previous ulcer complication, anticoagulant treatment or a cumulative number of risk factors. These proportions are larger than expected, which may have important clinical implications for the recommendation of prevention strategies, as the vast majority of patients who need NSAIDs are often considered to be at low or moderate GI risk.
In addition to increasing the risk of adverse GI effects, coxibs and most, if not all, ns-NSAIDs have also been linked with CV risk.10 Current guidelines consider that CV risk factors should be assessed before any therapeutic decision and that strategy should be guided by these factors in combination with GI risk factors. This study has shown that more than half of patients with OA seen in rheumatology clinics have increased CV risk. Furthermore, a large proportion of patients were classified within the high CV risk group, and yet, a small proportion of them received concomitant cardioprotective ASA treatment, mostly in combination with ns-NSAIDs; this may be explained, at least in part, as being more specific to European Mediterranean countries, since low-dose ASA is prescribed in most cases for the secondary prevention of CV risk. The high proportion of patients with high CV risk might be due to the SCORE model being used in this study, which may not be used in clinical routine. This suggests that many patients with high CV risk are not detected appropriately and that prescription routines are preferentially instigated according to GI risk factors. Choosing an appropriate therapeutic strategy for patients with OA with CV risk who require anti-inflammatory drugs and/or are taking low-dose aspirin is a daunting task for most doctors. Regulatory agencies in Europe, but not in the USA, do not recommend the use of coxibs in such a situation, but we have found that prescriptions used are very similar regardless of the presence/absence of these GI or CV risk factors.
Finally, this study shows, for the first time, that >15% of patients with OA, which is a remarkable proportion, and visiting rheumatology clinics were considered to have high GI and CV risk. This has enormous clinical implications, since current guidelines indicate that these patients should not be receiving either NSAID or coxib treatments unless strictly needed. Still, our study shows that most of them are prescribed NSAIDs or coxibs in a similar manner to other patients with less severe GI or CV risk, which should prompt a reconsideration of our normal habits of drug prescription and patient evaluation in clinical practice and the tools used to uncover the underlying risk factors. A future study looking at the appropriateness of drug treatments based on these guidelines has been planned. On the other hand, it is also obvious that chronic pain has a serious negative impact on the quality of life, which means it is a delicate balance for patients and clinicians when considering the benefits and risks of these treatments for the individual patient.
The limitations of our study are consistent with the intrinsic nature of observational studies and the unavoidable bias on patient selection, which we tried to correct by selecting a large number of participants from the whole country and enrolling consecutive patients. Also, some CV and GI risk factors and data on drug prescription relied on information reported in charts and on participants' self-reports, which we controlled by maintaining data consistency across participants.
In conclusion, this study shows the baseline profile of CV and GI risk factors in patients with OA in clinical practice. The findings are relevant, since it demonstrates that the vast majority of patients with OA attending rheumatology clinics have either GI or CV risk factors, or both, and that more than half of them can be considered as high GI or CV risk, which should considerably limit the use of NSAIDs, based on current guidelines. Since NSAID prescription habits are similar regardless of the presence of these risk factors, new strategies should be implemented to facilitate the recognition of these GI and CV risk factors in clinical practice and the appropriate therapeutic approach to these patients.
This study has the scientific support of the Sociedad Española de Reumatología. Thanks to Beatriz Armada for facilitating the authors' work.
Funding Sponsored by Pfizer Spain Laboratories, Avenida de Europa 20-B, Parque Empresarial La Moraleja, 28108, Alcobendas, Madrid, Spain.
Competing interests AL is an advisor to Pfizer, AstraZeneca and Nicox and is member of steering committees of studies sponsored by Bayer (ARRIVE), AstraZeneca (ENERGIB) and Pfizer (CONDOR). JT and JLZ declare no conflicts of interest.
Ethics approval The work was presented to, and approved by, the Regional Ethics Committee for Clinical Research, Hospital San Carlos (Madrid).
Provenance and peer review Not commissioned; externally peer reviewed.
The manuscript has been written by the authors and represents the interpretation of data by the authors of the study and not by Pfizer. The authors of this article have had full access to the data, which have been analysed based on predefined criteria and guided by the authors.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.