Articles

Use of oral prednisolone or naproxen for the treatment of : a double-blind, randomised equivalence trial

Hein J E M Janssens, Matthijs Janssen, Eloy H van de Lisdonk, Piet L C M van Riel, Chris van Weel

Summary Lancet 2008; 371: 1854–60 Background Non-steroidal anti-infl ammatory drugs and colchicine used to treat gout arthritis have gastrointestinal, See Comment page 1816 renal, and cardiovascular adverse eff ects. Systemic corticosteroids might be a benefi cial alternative. We investigated Department of General Practice equivalence of naproxen and prednisolone in primary care. (H J E M Janssens MD, E H van de Lisdonk PhD, Methods We did a randomised clinical trial to test equivalence of prednisolone and naproxen for the treatment of Prof C van Weel PhD) and Department of Rheumatology monoarticular gout. Primary-care patients with gout confi rmed by presence of monosodium urate crystals were (Prof P L C M van Riel PhD), eligible. 120 patients were randomly assigned with computer-generated randomisation to receive either prednisolone Radboud University Nijmegen (35 mg once a day; n=60) or naproxen (500 mg twice a day; n=60), for 5 days. Treatment was masked for both patients Medical Centre, Nijmegen, and physicians. The primary outcome was pain measured on a 100 mm visual analogue scale and the a priori margin Netherlands; and Department of Rheumatology, Rijnstate for equivalence set at 10%. Analyses were done per protocol and by intention to treat. This study is registered as an Hospital, Arnhem, Netherlands International Standard Randomised Controlled Trial, number ISRCTN14648181. (M Janssen PhD) Correspondence to: Findings Data were incomplete for one patient in each treatment group, so per-protocol analyses included 59 patients Dr Hein J E M Janssens, Radboud in each group. After 90 h the reduction in the pain score was 44·7 mm and 46·0 mm for prednisolone and naproxen, University Nijmegen Medical Centre, PO Box 9101, 6500 HB respectively (diff erence 1·3 mm; 95% CI –9·8 to 7·1), suggesting equivalence. The diff erence in the size of change in Nijmegen, Netherlands pain was 1·57 mm (95% CI –8·65 to 11·78). Adverse eff ects were similar between groups, minor, and resolved by [email protected] 3 week follow-up.

Interpretation Oral prednisolone and naproxen are equally eff ective in the initial treatment of gout arthritis over 4 days.

Funding Rheumatology Research Fund Arnhem, Netherlands.

Introduction window,10,11 its disadvantage in the case of renal failure,12 Gout is a painful infl ammatory arthritis with a sudden and the availability of NSAIDs. The fi rst choice of drug onset and tendency to recur.1 The gold standard for the treatment is now NSAIDs,13,14 despite their diagnosis of gout is the microscopic identifi cation of gastrointestinal15,16 and cardiovascular risks.17 About 40% of monosodium urate crystals in synovial fl uid or tophi. upper gastrointestinal bleeding events are attributable to Acute severe pain prompts most patients to seek medical NSAIDs (risk is highest during the fi rst week of use),16,18 care immediately. and the cardiovascular risks prompted the American The history of gout is marked by famous patients, Heart Association recently to recommend restricted use.19 including Desiderius Erasmus, Holy Roman Emperor Loss of renal function, fl uid retention, and interaction Charles V and Immanuel Kant; by impressive scientifi c with anticoagulants, further discourage the use of researchers like Antoni van Leeuwenhoek, who fi rst NSAIDs. The drawbacks of the NSAIDs are especially identifi ed monosodium urate crystals from a tophus important in the case of gout,13 because patients are at with his hand-held microscope in the 1600s, Anton Von high risk for the gastrointestinal side-eff ects, most of Störck, who laid the foundations for the modern use of them are middle aged or elderly,20,21 and many have colchicine in the 1800s, and Sir Alfred Baring Garrod, comorbid renal and cardiovascular diseases.6,12,21–23 who fi rst postulated in the 1900s that deposits of serum As a consequence, safer therapeutic approaches are urate were the cause not the consequence of gout; by needed.14 Systemic corticosteroids might be a safe various prescientifi c treatments, such as diets, plasters, alternative, particularly in elderly people.13,14,21 Although baths, herbal remedies, opium; and by many disputes they are associated with side-eff ects such as osteoporosis, about the triggering factors, including socioeconomic fl uid retention, and hyperglycaemia when used long status, alcohol, food, and drugs.2–7 Gout has a substantial term, systemic corticosteroids do not have important eff ect on the population. An estimated 1–2% of adults in drawbacks in the short term.24,25 However, the only developed countries have the disorder,4,8 which creates a evidence on the eff ectiveness of systemic corticosteroids substantial burden of work-related and medical costs.9 in the treatment of gout is from three studies included in The most widely used drugs for gout arthritis are a recent systematic Cochrane review:26 two small, colchicine and non-steroidal anti-infl ammatory drugs unblinded, non-randomised studies,27,28 and another in (NSAIDs). Colchicine is the most longstanding treatment, patients with gout-like arthritis not confi rmed with tests but its use has declined because of its narrow therapeutic for monosodium urate crystals.29

1854 www.thelancet.com Vol 371 May 31, 2008 Articles

We aimed to investigate whether oral prednisolone is equivalent to naproxen in patients with confi rmed gout 381 patients with monoarthritis diagnosed arthritis. by family doctors

Methods 165 patients without MSU crystals: 81 arthritis unknown cause Patients 84 arthritis other causes* We designed a randomised, double-blind, active- comparator, controlled trial to test for equivalence of 216 patients with gout confirmed by prednisolone and naproxen for the treatment of gout presence of MSU crystals assessed for eligibility arthritis. The study took place in the eastern part of the 96 patients excluded† Netherlands in a population of about 330 000 inhabitants. In the Netherlands, patients are listed with a personal family doctor who provides primary care for about 120 patients randomly assigned to trial treatments 2250 people and refers patients for secondary medical care if needed. From March 24, 2004, until July 14, 2006, family doctors were asked to send all patients with monoarthritis to the 60 assigned to 60 assigned to trial centre, even if gout was not the most likely diagnosis, prednisolone (ITT) naproxen (ITT) to minimise the risk of missing potential study participants. Patients were assessed in the trial centre for 1 patient had 1 patient had inclusion eligibility within 1 day after they visited their incomplete data‡ incomplete data‡ family doctor. Participants were patients with a monoarticular gout arthritis confi rmed by identifi cation of monosodium 59 patients 59 patients (per protocol) (per protocol) urate crystals in the synovial fl uid of the aff ected . All patients provided written informed consent. The study was done according to good clinical practice Figure 1: Trial profi le guidelines and approved by the regional ethics review *Tested positive for cyclic citrullinated peptide antibodies or rheumatoid factor positive or had progression towards rheumatoid arthritis (12); psoriatic arthritis (six); pseudogout (14); reactive arthritis (one); committee (CMO Arnhem-Nijmegen). Because of post-streptococcal reactive arthritis (fi ve); arthritis secondary to infl ammatory bowel disease (4); Lyme-arthritis ethical considerations, the study had no placebo control (three); bacterial arthritis (three); palindromic rheumatism (fi ve); osteoarthritis (19); no arthritis (12). †86 men, group. ten women, mean age 60·5 years (12·8). Reasons for exclusion: contemporary use of NSAID or colchicine (49); Exclusion criteria were unstable condition (prevalent history of gastrointestinal ulcer or bleeding (six); contemporary use of acenocoumarol (24); serious renal problems (eight); serious other comorbidity (three); refusal (six). ‡Both drop outs had arthritis of the leg or foot. angina pectoris, myocardial infarction, manifest heart ITT=intention to treat. failure, severe renal failure, renal transplant, or cancer); chronic rheumatic diseases; current use of anticoagulants; and medical history of upper gastrointestinal diseases. One of the investigators (MJ) reviewed all patients by Patients were not excluded on the basis of age. We did means of a standardised interview and physical not allow the use of NSAIDs or other analgesics examination. The following data were recorded: current (including colchicine) within 24 h before baseline joint pain (eg, intensity, duration); other current assessments or for the duration of the trial. complaints; medical history (former arthritis, urolithiasis, cardiovascular morbidity); lifestyle Procedures (smoking, use of alcohol); use of medication before Patients were randomly assigned to one of two oral study inclusion (eg, analgesics, urate-lowering drugs, treatments (prednisolone or naproxen) given for 5 days. diuretics); family history (gout); the status of the An independent trial monitor used a computer-generated aff ected joint (redness, swelling); the status of the other randomisation procedure to create an allocation ; presence of tophi; cardiovascular parameters sequence list with a block randomisation of four (blood pressure, body mass index); laboratory fi ndings treatments, each treatment being given twice. (eg, , C-reactive protein, ESR, glomerular Consecutive patients received drugs following the fi ltration rate). allocation list: either prednisolone 35 mg once a day and The aff ected joint of each patient was aspirated. look-alike placebo naproxen twice a day, or naproxen Aspirated fl uid was analysed for monosodium urate 500 mg twice a day and a placebo capsule prednisolone crystals under a polarisation microscope. (double-dummy design). Patients took their fi rst dose in When monosodium urate crystals could not be the presence of one of the investigators. Patients and identifi ed the patient was excluded from the trial, but was investigators were unaware of drug during the whole assessed for other joint diseases (like rheumatoid study, including the analyses. arthritis, pseudogout, and reactive arthritis). www.thelancet.com Vol 371 May 31, 2008 1855 Articles

Prednisolone n=60 Naproxen n=60 Prednisolone n=60 Naproxen n=60 General (Continued from previous column) Men 54 (90%) 53 (88%) Diastolic blood pressure (mm Hg) Age (years) Mean (SD) 85 (13) 84 (11) Mean (SD) 57·3 (12·2) 57·7 (13·4) Median (IQR) 80 (80–95) 80 (80–92) Median (IQR) 58·0 (51·0–64·0) 57·5 (48·5–67·0) >90 mm Hg 18 (30%) 15 (25%) Gout features Serum concentrations Joints Uric acid (mmol/L)* Metatarsophalangeal 38 (63%) 38 (63%) Mean (SD) 0·46 (0·10) 0·48 (0·08) joint 1 Median (IQR) 0·45 (0·42–0·52) 0·46 (0·43–0·52) Other foot joints, ankle, 17 (28%) 18 (30%) >0·35 mmol/L 56 (93%) 58 (97%) or knee Calcium (mmol/L)* Elbow, wrist, or hand 5 (8%) 4 (7%) Mean (SD) 2·47 (0·10) 2·45 (0·09) Tophus 7 (12%) 4 (7%) Median (IQR) 2·45 (2·42–2·52) 2·45 (2·41–2·52) Patients with previous 41 (68%) 44 (73%) attack in life-time >2·55 mmol/L 10 (17%) 5 (8%) Time since last attack (months) Glucose (mmol/L)* Mean (SD) 7·7 (11·1) 9·5 (12·4) Mean (SD) 5·9 (1·3) 6·0 (1·3) Median (IQR) 3·2 (1·1–10·2) 4·2 (0·9–12·5) Median (IQR) 5·7 (5·1–6·2) 5·7 (5·3–6·5) Patients with attack in 33 (55%) 33 (55%) >12·2 mmol/L 0 (0%) 0 (0%) previous year Cholesterol (mmol/L)* Time since last attack (months) Mean (SD) 5·3 (1·0) 5·1 (1·2) Mean (SD) 3·1 (3·0) 3·1 (3·0) Median (IQR) 5·3 (4·6–6·1) 4·9 (4·3–6·0) Median (IQR) 1·8 (0·9–4·0) 2·0 (0·7–4·5) >6·5 mmol/L 6 (10%) 7 (12%) Number of attacks in past year Creatinine (μmol/L)* Mean (SD) 3·5 (3·1) 3·3 (2·2) Mean (SD) 91·2 (26·9) 94·3 (30·9) Median (IQR) 3 (1–5·0) 3 (1·5–4·0) Median (IQR) 84·0 (75·0–98·0) 86·5 (76·0–103·8) Urate-lowering therapy 1 (2%) 0 (0%) >105 μmol/L 12 (20%) 14 (23%) Family history of gout 17 (28%) 18 (30%) Glomerular fi ltration rate (mL/min/1·73 m²)† Infl ammation Mean (SD) 79 (22) 76 (23) Joint redness 54 (90%) 54 (90%) Median (IQR) 80 (67–91) 76 (59–94) C-reactive protein (mg/dL) Medication Mean (SD) 31 (38) 24 (34) Diuretics (all types) 14 (23%) 20 (33%) Median (IQR) 16 (9·0–37·0) 14 (6·0–22·0) Antihypertensive drugs 26 (43%) 32 (53%) >10 mg/L 42 (70%) 34 (57%) other than diuretics ESR (mm/h) Cardiovascular drugs other 13 (22%) 11 (18%) than diuretics Mean (SD) 24 (22) 22 (20) Lifestyle Median (IQR) 18 (7·0–27·0) 14 (9·0–31·0) Alcohol Men >20 mm/h, women 23 (38%) 22 (37%) >30 mm/h No 16 (27%) 25 (42%) Comorbidity <7 units/week 12 (20%) 8 (13%) Hypertension 30 (50%) 34 (57%) 7–21 units/week 19 (32%) 19 (32%) One or more cardiovascular 11 (18%) 12 (20% ) >21 units/week 13 (22%) 8 (13%) diseases Smoking 10 (17%) 11 (18%) Diabetes 2 (3%) 9 (15%) Working disability‡ History of renal stones 5 (8%) 4 (7%) Not applicable 34 (57%) 30 (50%) Other clinical fi ndings Yes 17 (28%) 18 (30%) Body-mass index (kg/m²) Data are number (%), unless otherwise stated. Data missing for erythrocyte Mean (SD) 29 (4) 30 (5) sedimentation for one patient in each group and for C-reactive protein for one Median (IQR) 28 (27–31) 29 (27–32) in each group and for calcium (one patient), glucose (one), cholesterol (one), Systolic blood pressure (mm Hg) and creatinine (one), in the prednisolone group. *Normal ranges: uric acid 0·12–0·35 mmol/L; calcium 2·10–2·55 mmol/L; glucose 5·0–12·2 mmol/L; Mean (SD) 143 (22) 145 (23) cholesterol 3·0–6·5 mmol/L; creatinine 70–105 μmol/L. †Modifi cation of diet Median (IQR) 140 (130–158) 140 (130–157) in renal disease equation four variables: age, ethnicity, sex, plasma creatinine. > 140 mm Hg 24 (40%) 25 (42%) ‡Patient had to stop work because of gout arthritis. (Continues in next column) Table 1: Baseline characteristics of patients

1856 www.thelancet.com Vol 371 May 31, 2008 Articles

The primary outcome was the pain in the aff ected joint Pain General disability Walking disability* as experienced by the patient. Secondary outcomes were the general disability caused by the aff ected joint, and the Prednisolone Naproxen Prednisolone Naproxen Prednisolone Naproxen (n=59) (n=59) (n=59) (n=59) (n=54) (n=55) walking disability if the arthritis was in the lower limb. Pain was scored on a visual analogue scale from 0 mm Baseline 61·5 (22·4) 58·9 (20·8) 59·2 (26·9) 55·1 (25·7) 70·9 (22·2) 67·4 (19·8) (absence of pain) to 100 mm (the most severe pain ever After 90 h 16·8 (24·0) 12·9 (18·1) 17·2 (25·0) 12·7 (18·1) 17·4 (25·4) 13·0 (18·3) experienced), a validated method for the assessment of Reduction 44·7 (25·0) 46·0 (21·2) 42·1 (29·6) 42·4 (26·4) 53·5 (28·1) 54·4 (22·3) pain.30 Disability related to use of the aff ected joint was Interval scored on a visual analogue scale from 0 mm (absence of 1 (0–6 h) –9·9 (13·2) –8·4 (16·9) –5·9 (15·0) –4·7 (14·9) –12·5 (15·8) –12·0 (16·6) disability) to 100 mm (completely unable to do 2 (6–18 h) –10·5 (16·0) –12·3 (11·7) –10·6 (15·6) –12·1 (14·5) –13·6 (16·1) –15·6 (15·7) something). Walking disability was also scored on a 3 (18–30 h) –6·9 (11·3) –1·5 (15·0) –6·8 (10·7) –3·2 (12·2) –6·2 (10·5) –0·5 (12·7) visual analogue scale, from 0 mm (walking without any 4 (30–42 h) –1·4 (15·9) –9·0 (17·1) –2·1 (16·8) –7·1 (15·9) –3·0 (19·3) –10·2 (14·9 problem) to 100 mm (completely unable to walk). Neither 5 (42–54 h) –6·9 (10·3) –2·8 (13·5) –6·6 (10·3) –3·5 (11·0) –7·6 (10·6) –2·5 (10·6) disability scale has been validated. 6 (54–66 h) –1·8 (10·5) –4·0 (12·1) –2·1 (10·2) –4·2 (12·0) –2·8 (10·3) –6·6 (7·5) After the baseline assessment in the trial centre, 7 (66–78 h) –4·2 (9·6) –4·0 (12·1) –4·7 (8·0) –3·1 (11·6) –3·6 (9·8) –1·0 (10·7) patients scored their pain and disability in a trial diary on 8 (78–90 h) –3·2 (8·1) –4·0 (7·2) –3·3 (7·4) –5·3 (10·6) –4·2 (7·7) –6·1 (12·5) eight occasions, starting on the evening of the inclusion Mean –5·6 (12·5) –5·8 (13·9) –5·3 (12·4) –5·4 (13·2) –6·7 (13·6) –6·8 (13·8) day, thereafter at intervals of 12 h, and ending on day 4. Data are mean distance in mm (SD) on 100-mm visual analogue scales, unless otherwise stated. *If the arthritis was in Each patient was contacted by telephone at 3 weeks after the leg or foot. inclusion to assess pain and disability. Having registered the last outcome scores, patients Table 2: Score on visual analogue scale at the baseline and after 90 h, and mean change of for each were asked to select in the trial diary one or more of fi ve follow-up interval categories of side-eff ects: none; gastric pain, abdominal pain, or both; itch, dizziness, or both; dyspnoea, and analysed for the mean diff erence (95% CI) between palpitations, or both; other. the two treatments. We did a repeated measurement Patients were asked to return their trial diaries and model (PROC MIXED, SAS 8.2 for Windows). As the drug packages within 1 week of ending the treatment by dependent variable in this model we chose the change in prepaid post. Compliance was assessed by counting and millimetres in the score on the visual analogue scale assessing the returned diaries, drug packages, and between two consecutive measurement points (T2–T1, unused drugs, for discontinuation of treatment, drug T3–T2, to T9–T8): per patient eight values from eight intolerance, incompleteness of data, and drop-out rate. intervals. We assumed an autoregressive correlation In the telephone follow-up at 3 weeks patients were also structure. The estimated eff ect size in this model was the assessed for incidence of new illnesses or recurrent diff erence in the size of change of the score per interval. gout. To estimate eff ect size for the whole period (eight intervals) we multiplied the estimated eff ect size per Statistical analysis interval by eight. Because we aimed to investigate the equivalence of two Analyses were done primarily for the per-protocol treatments, the alternative hypothesis that there is a group for sensitivity reasons (ie, not missing diff erence) diff erence between the treatments should be rejected if and repeated for the intention-to-treat group. the 95% CI of the diff erence in treatment eff ect was Except for the repeated measure model, all analyses within a clinically relevant range of equivalence. We were done with SPSS 14.2 for Windows. arbitrarily predefi ned the margin of equivalence as 10%. With a 1–β of 0·8 and α of 0·05 we calculated that Role of the funding source 59 patients per group were needed, assuming an The study sponsor had no role in the study design, data arbitrary relevant treatment eff ect (a mean reduction of collection, data analysis, data interpretation, or writing of pain of 55% on a visual analogue scale of 100 mm, an the report. All authors had full access to the study data absolute reduction of 30 mm,31,32 or both) in 95% of the and all had fi nal responsibility for the decision to submit patients. Because extra measurements per patient were for publication. done with the raw scores at each follow-up, we expected a greater power than we had calculated. Results Baseline characteristics were analysed by descriptive 93 family doctors sent 381 patients with suspected methods (mean, SD, median, IQR, and frequencies). monoarthritis to the trial centre. Of these 216 (57%) met Baseline characteristics and categories of adverse eff ects the inclusion criterion of gout after identifi cation of were investigated for statistical diff erences (p<0·05) with monosodium urate crystals (fi gure 1). 96 (32%) were Fisher’s exact test (two sided). excluded for several reasons. Excluded patients were older The reduction of the mean scores on the visual analogue (mean age 60·5 years, SD 12·8) than patients who scales between baseline and last follow-up was calculated underwent randomisation (57·5, 12·7). 120 completed the www.thelancet.com Vol 371 May 31, 2008 1857 Articles

trial. Two patients, one in each treatment group, recorded 47 participant (80%) in the prednisolone group and their visual analogue scale in an inappropriate way so were 52 (88%) in the naproxen group had clinically signifi cant not suitable for analysis; we classifi ed these as drop outs improvement (mean reduction of pain of 55% on the and they were removed from the per-protocol analysis. visual analogue scale, and absolute reduction of 30 mm, There were no important diff erences between the or both) within the observation period. treatment groups at baseline (table 1). Most patients were The reduction of mean score on the visual analogue men (107, 89%). Mean age was 57 years. Metatarso- scales for the primary and secondary outcomes between phalangeal joint 1 was the most aff ected joint (76, 63%). baseline and the last observation moment (on day 4, after 11 patients (9%) had tophi. 64 (53%) had hypertension, 90 h) is presented in table 2. The decrease of the pain was and 23 (19%) had one or more cardiovascular disease. 44·7 mm for prednisolone and 46·0 mm for naproxen The results of the intention-to-treat analyses agree with (diff erence 1·3 mm; 95% CI –9·8 to 7·1). The mean those of the per-protocol analyses. In the report we diff erence in reduction of general disability was 0·3 mm present the latter. (–10·5 to 9·9) and of walking disability was 0·8 mm (–10·5 to 8·8). The mean change in scores on the visual analogue scales per observation interval is also shown in A table 2 and fi gure 2. The diff erence between the two 80 Prednisolone groups in the size of the change of scores for pain for the Naproxen 70 whole observation period (944 intervals used for 60 calculation) was 1·57 mm (–8·65 to 11·78). The 50 diff erences were 0·93 mm (–9·51 to 11·37) for general disability and 0·72 mm (–11·3 to 12·7) for walking 40 disability. In all cases these diff erences non-signifi cantly 30 favoured naproxen. 20 There were no adverse eff ects reported in 39 (66%) of 10 patients taking prednisolone and in 37 (63%) taking 0 naproxen (p=0·42). There were no signifi cant diff erences between the two treatment groups in the numbers of B adverse events (table 3). 80 After 3 weeks, all patients reported complete relief of 70 signs and symptoms. No patient reported a recurrent

60 attack. 50 Discussion 40 Both oral prednisolone 35 mg once a day and naproxen 30 500 mg twice a day seemed eff ective in the treatment of 20 gout arthritis. The 95% CI for the small and

10 non-signifi cant, diff erence between the two treatments in pain reduction after 90 h (–9·8 to 7·1) was within the 0 predefi ned 10% margin of equivalence. The 95% CI of the diff erence in the size of change in pain during 90 h C (–8·65 to 11·78 mm) was outside our a priori margin but 80

70 Prednisolone Naproxen 60 Gastric or abdominal pain 9 (15%) 9 (15%) 50 Itch or dizziness 4 (7% ) 4 (7%)

40 Dyspnoea or palpitations 3 (5%) 3 (5%) Other adverse eff ects 13 (22%) 12 (20%) 30 Data are number (%).Other adverse eff ects: “1 day calf-muscle spasm”, “sick”, 20

Visual analogue scale (mm) Visual analogue scale (mm) Visual analogue scale (mm) “sweating and more tired”, “1 day tired back and legs”, “increased bodyweight of 10 2·5 kg”, “trembling eyelid”, “sleep disturbance”, “corporal warmth and tingling feelings in one foot”, “red head in the morning”, “feeling of swollen face and legs”, 0 “morning hiccup”, “1 day slight headache and glowing cheeks”, “headache”, each 0 6 12 18 24 30 36 42 48 54 60 66 72 78 84 90 once in the prednisolone group; “nocturnal muscle spasm”, “headache”, “little Time (h) heaviness in the head”, “diarrhoea”, “one swollen foot”, “less cycle condition”, “tiredness”, “sleep disturbance”, “oesophagus pain”, “slightly depressive”, each once, and “low back pain” twice in the naproxen group. Figure 2: Reduction in pain, general disability, and walking disability from baseline to 90 h follow up Scores on visual analogue scales for pain (top), general disability (middle), and walking disability (bottom) Table 3: Adverse eff ects recorded at baseline and eight follow-ups.

1858 www.thelancet.com Vol 371 May 31, 2008 Articles

was within the 13% margin used as the non-negligible within 1 day of referral by family doctors. The study had a diff erence in another recent gout trial.29 There was also successful randomisation and allocation concealment equivalence within the 13% margin, but not the throughout, including the analyses. Only one research 10% margin, for general disability and walking disability. centre was involved in study enrolment and follow-up. Hence, we conclude that prednisolone was, in this trial, The period of assessment was limited to 90 h because clinically equivalent to naproxen in the treatment of gout. we were primarily interested in the rate of improvement Participants reported no important side-eff ects. during the fi rst few days of treatment, excluding possible Safety of patients is an important reason for the bias by patients’ expectations about complete recovery at reappraisal of routine treatments of gout.14 We chose to the end of treatment. All patients were free of complaints investigate oral prednisolone as a potentially safe equivalent after 3 weeks (evaluated by telephone). At the end of the to the NSAID naproxen. Systemic corticosteroids have period for the assessment (on day 4, when patients had inhibiting eff ects on the infl ammation process in 1 day more of treatment), 47 (80%) in the prednisolone rheumatoid arthritis resulting in eff ective pain reduction.25 group and 52 (87%) in the naproxen group had clinically Similar eff ects have been seen in a few studies on signifi cant improvement, and 13 (22%) and 10 (17%) had gout.27–29,33,34 Two of these studies compared systemic complete relief of symptoms (results not shown). In the corticosteroids with a NSAID: one unblinded and power calculation a 95% clinical improvement had been non-randomised trial in 27 patients, showed that postulated. We compensated for this loss of statistical intramuscular triamcinolone acetonide resulted in robustness with a repeated measurement model. The complete relief of symptoms after a mean of 7·5 days.27 assessment of complete relief of symptoms at 90 h was The other study, a double-blind, randomised trial in solely based on patients’ self-reporting at 3 weeks. 90 patients showed that a combination of oral prednisolone Furthermore, disability outcomes were assessed with and paracetamol resulted in a mean decrease of pain with non-validated scales. Also, the study population was activity of 2·9 mm on the visual analogue scale per day limited to white Dutch people. during 2 weeks.29 However, gout was diagnosed clinically The present study confi rmed the risk profi le of and not confi rmed by monosodium urate crystals. The patients with gout for NSAIDs (eg, mean age more than results of these two studies are in agreement with our 57 year, hypertension in 53%, and cardiovascular fi ndings. morbidity in 19%). Although no patients were excluded Our insistence on the identifi cation of monosodium because of the risks from a prednisolone treatment, a urate crystals as a criterion for patient inclusion is strongly quarter of the eligible patients had to be excluded recommended for prospective trials, to prevent the because of direct safety risks if they would have been inclusion of false-positive cases or the exclusion of treated with naproxen (serious renal diseases, serious false-negative cases.35 After reviewing research and a recent other comorbidity, a history of upper gastrointestinal systematic review on the eff ectiveness of interventions for ulcer or bleeding, and the current use of acenocoumarol). the treatment of acute gout,14 we are aware of only four For them a 5-day treatment with prednisolone would studies that used this diagnostic criterion for gout: one have been no problem. robust study on colchicine,10 and three small ones on In addition to better safety, the direct drug costs would intramuscular triamcinolone acetonide,27 intramuscular also be less if systemic corticosteroids (such as adrenocorticotropic hormone,28 and topical ice therapy,36 prednisolone) were fi rst-line drug choice. In the context but none on NSAIDs. of economic benefi ts, the additional costs of We did not fi nd any trial that selected patients with gastroprotective drugs added to NSAID-treatment, gout in a primary-care source population, despite the fact should also be taken into account.37 that most patients are diagnosed and managed in primary In conclusion, the present study provides a strong care.4 All the gout trials of which we are aware recruited argument to consider prednisolone as a fi rst treatment their participants from hospital-based populations, which option in patients with gout. might include patients with diff erent characteristics (eg, Contributors age, gender, joint locations, comorbidity, use of urate- HJ was the principal investigator. He conceived and coordinated the lowering drugs) to ours, who were probably representative study, wrote the manuscript, and edited the fi nal paper. MJ and EvdL of gout patients in the community. Only one patient was contributed to the conception and the design of the study. MJ reviewed all the study participants, collected the study data, and was responsible on urate-lowering therapy; although this might be fewer for the intervention. MJ, EvdL, PvR, and CvW reviewed the manuscript than expected, there are at least two reasons why such and contributed to the editing of the fi nal paper. therapy might not have been started in these patients: 35 Confl ict of interest statement (29%) had a debut attack, and the median number of We declare that we have no confl ict of interest. attacks in the previous year was low. Acknowledgments Ours is one of the fi rst controlled trials on systemic We are thankful to the employees of the Pharmacy Riet (Rotterdam) and corticosteroids to include patients who all had a defi nite the drug dispensing Primary Care Centre Lobede (Lobith-Tolkamer) who prepared the study drugs; the 93 family doctors working in diagnosis of gout and who were recruited in a general Arnhem/Liemers and surroundings who selected the 381 patients; the population. All patients were assessed for eligibility patients who were willing to participate the study; Tonnie Berends, www.thelancet.com Vol 371 May 31, 2008 1859 Articles

Carla De Gendt, Alphons de Jong, and Henk Visser (Department 19 Antman EM, Bennett JS, Daugherty A, Furberg C, Roberts H, Rheumatology, Rijnstate Hospital Arnhem) for facilitating the data Taubert KA. Use of nonsteroidal antiinfl ammatory drugs: an update collection; Twanny Jeijsman-Rouwhorst, Jan van Doremalen, Hans Bor, for clinicians: a scientifi c statement from the American Heart and Henk van den Hoogen (Department General Practice, Radboud Association. Circulation 2007; 115: 1634–42. University Nijmegen) for helping us with the data collection and 20 Lawrence RC, Felson DT, Helmick CG. Estimates of the prevalence analysis; and John Macfarlane for critically reading the manuscript. of arthritis and other rheumatic conditions in the United States, part II. Arthritis Rheum 2008; 58: 26–35. References 21 Fam AG. Gout in the elderly: clinical presentation and treatment. 1 Terkeltaub RA. Gout. N Engl J Med 2003; 349: 1647–55. Drugs Aging 1998; 13: 229–43. 2 Porter R, Rousseau GS. Gout, the patrician malady. New Haven, 22 Janssens HJEM, van de Lisdonk EH, Bor H, van den Hoogen HJ, London: Yale University Press, 2000. Janssen M. Gout, just a nasty event or a cardiovascular signal? A 3 Ordi J, Alonso PL, de Zulueta J, et al. The severe gout of Holy study from primary care. Fam Pract 2003; 20: 413–16. Roman Emperor Charles V. N Engl J Med 2006; 355: 516–20. 23 Krishnan E, Baker JF, Furst DE, Schumacher HR. Gout and the risk 4 Zhang W, Doherty M, Bardin T, et al. EULAR evidence based of acute myocardial infarction. Arthritis Rheum 2006; 54: 2688–96. recommendations for gout, part I: diagnosis—report of a task force 24 Rowe BH, Spooner C, Ducharme FM, Bretzlaff JA, Bota GW. Early of the Standing Committee for International Clinical Studies emergency department treatment of acute asthma with systemic Including Therapeutics (ESCISIT). Ann Rheum Dis 2006; corticosteroids. Cochrane Database Syst Rev 2001; 1: CD002178. 65: 1301–11. 25 Gotzsche PC, Johansen HK. Short-term low-dose corticosteroids vs 5 Choi HK, Atkinson K, Karlson EW, Willett W, Curhan G. Alcohol placebo and nonsteroidal anti-infl ammatory drugs in rheumatoid intake and risk of incident gout in men: a prospective study. Lancet arthritis. Cochrane Database Syst Rev 2005; 1: CD000189. 2004; 363: 1277–81. 26 Janssens HJEM, Lucassen PLBJ, Van de Laar FA, Janssen M, 6 Janssens HJEM, van de Lisdonk EH, Janssen M, Van de Lisdonk EH. Systemic steroids for acute gout. van den Hoogen HJM, Verbeek ALM. Gout, not induced by Cochrane Database Syst Rev 2008; 2: CD005521. diuretics? A case-control study from primary care. Ann Rheum Dis 27 Alloway JA, Moriarty MJ, Hoogland YT, Nashel DJ. Comparison of 2006; 65: 1080–83. triamcinolone acetonide with indomethacin in the treatment of 7 Choi HK, Curhan G. Soft drinks, fructose consumption, and the acute gouty arthritis. J Rheumatol 1993; 20: 111–13. risk of gout in men: a prospective cohort study. BMJ 2008; 28 Siegel LB, Alloway JA, Nashel DJ. Comparison of 336: 309–12. adrenocorticotropic hormone and triamcinolone acetonide in the 8 Bieber JD, Terkeltaub RA. Gout: on the brink of novel therapeutic treatment of acute gouty arthritis. J Rheumatol 1994; 21: 1325–27. options for an ancient disease. Arthritis Rheum 2004; 50: 2400–14. 29 Man CY, Cheung IT, Cameron PA, Rainer TH. Comparison of oral 9 Brook RA, Kleinman NL, Patel PA, et al. The economic burden of prednisolone/paracetamol and oral indomethacin/paracetamol gout on an employed population. Curr Med Res Opin 2006; combination therapy in the treatment of acute goutlike arthritis: a 22: 1381–89. double-blind, randomized, controlled trial. Ann Emerg Med 2007; 10 Ahern MJ, Reid C, Gordon TP, McCredie M, Brooks PM, Jones M. 49: 670–77. Does colchicine work? The results of the fi rst controlled study in 30 Todd KH, Funk KG, Funk JP, Bonacci R. Clinical signifi cance of acute gout. Aust N Z J Med 1987; 17: 301–04. reported changes in pain severity. Ann Emerg Med 1996; 27: 485–89. 11 Schlesinger N, Schumacher HR, Catton M, Maxwell L. Colchicine 31 Stahmer SA, Shofer FS, Marino A, Shepherd S, Abbuhl S. Do for acute gout. Cochrane Database Syst Rev 2006; 4: CD006190. quantitative changes in pain intensity correlate with pain relief and 12 Petersel D, Schlesinger N. Treatment of acute gout in hospitalized satisfaction? Acad Emerg Med 1998; 5: 5851–57. patients. J Rheumatol 2007; 34: 1566–68. 32 ten Klooster PM, Drossaers-Bakker KW, Taal E, van de Laar MA. 13 Underwood M. Diagnosis and management of gout. BMJ 2006; Patient-perceived satisfactory improvement (PPSI): interpreting 332: 1315–19. meaningful change in pain from the patient’s perspective. Pain 14 Sutaria S, Katbamna R, Underwood M. Eff ectiveness of 2006; 121: 151–57. interventions for the treatment of acute and prevention of recurrent 33 Groff GD, Franck WA, Raddatz DA. Systemic steroid therapy for gout—a systematic review. Rheumatology 2006; 45: 1422–31. acute gout: a clinical trial and review of the literature. 15 Rostom A, Muir, K, Dubé C, et al. Gastrointestinal safety of Semin Arthritis Rheum 1990; 9: 329–36. cyclooxygenase-2 inhibitors: a Cochrane Collaboration systematic 34 Werlen D, Gabay C, Vischer TL. Corticosteroid therapy for the review. Clin Gastroenterol Hepatol 2007; 5: 818–28. treatment of acute attacks of crystal-induced arthritis: an eff ective 16 Lewis SC, Langman MJ, Laporte JR, Matthews JN, Rawlins MD, alternative to nonsteroidal antiinfl ammatory drugs. Wiholm BE. Dose-response relationships between individual Rev Rhum Engl Ed 1996; 63: 248–54. nonaspirin nonsteroidal anti-infl ammatory drugs (NANSAIDs) and 35 Schumacher HR Jr, Edwards LN P-RF, Becker M, et al. Outcome serious upper gastrointestinal bleeding: a meta-analysis based on measures for acute and chronic gout. J Rheumatol 2005; individual patient data. Br J Clin Pharmacol 2002; 54: 320–26. 32: 2452–55. 17 Kearney PM, Baigent C, Godwin J, Halls H, Emberson JR, 36 Schlesinger N, Detry MA, Holland BK, et al. Local ice therapy Patrono C. Do selective cyclo-oxygenase-2 inhibitors and traditional during bouts of acute gouty arthritis. J Rheumatol 2002; 29: 331–34. non-steroidal anti-infl ammatory drugs increase the risk of 37 Lanza FL, Members of the Ad Hoc Committee on Practice atherothrombosis? Meta-analysis of randomised trials. BMJ 2006; Parameters of the American College of Gastroenterology. A 332: 1302–08. guideline for the treatment and prevention of NSAID-induced 18 Laporte JR, Ibanez L, Vidal X, Vendrell L, Leone R. Upper ulcers. Am J Gastroenterol 1998; 93: 2037–46. gastrointestinal bleeding associated with the use of NSAIDs: newer versus older agents. Drug Saf 2004; 27: 411–20.

1860 www.thelancet.com Vol 371 May 31, 2008