Jessop 1976.

Methods	Design: Cross‐over study Number of centres: 1 Treatment duration: each treatment period was 4 weeks Flare design: No Wash‐out period: No Time point of assessments: BL, 4, 8 weeks
Participants	Inclusion criteria: AS for which patient was currently receiving treatment in an outpatient department. Exclusion criteria: 1. Renal, hepatic, or cardiac failure; 2. Severe dyspepsia or previous intolerance to phenylbutazone; 3. Sacro‐ileitis associated with ulcerative colitis, regional ileitis, Reiter's disease or psoriasis. Classification: Bennet and Wood 1968 Ketoprofen (200 mg) first: Number of participants: 15 Number of completers: 12 Age (mean): 46.0 (range 20 to 59) Male (%): 83 Symptom duration: NA Disease duration: 2 participants 1 to 5 years, and 10 participants > 5 years HLA‐B27 positive (%): NA Phenylbutazone (300 mg) first: Number of participants: 11 Number of completers: 8 Age (mean): 37.3 (range 28 to 54) Male (%): 88 Symptom duration: NA Disease duration: 1 participant 1 to 5 years, and 7 participants > 5 years HLA‐B27 positive (%): NA
Interventions	Ketoprofen (200 mg) first vs Phenylbutazone (300 mg) first Co‐medication: Only paracetamol tablets (500 mg) from a measured supply could be taken as a 'rescue' analgesic. Physiotherapy was allowed, provided it had been started at least 4 weeks prior to start of study and was allowed to continue unchanged throughout the trial period.
Outcomes	Extracted outcomes: 1. Pain on Likert scale 2. Patient's global assessment of disease activity 3. Duration of morning stiffness 4. Severity of morning stiffness 5. Chest expansion (BL, post‐treatment after 4 weeks) (in cm, higher is better) Ketoprofen 200 mg: 4.2 (N = 9), 3.9 (N = 9) Phenylbutazone 300 mg: 4.5 (N = 7), 5.2 (N = 7) (P = not significant) 6. Tragus‐to‐wall distance (BL, post‐treatment after 4 weeks) (in cm, higher is worse) Ketoprofen 200 mg: 20.0 (N = 12), 19.6 (N = 12) Phenylbutazone 300 mg: 20.0 (N = 8), 20.1 (N = 8) (P = not significant) 7. Occiput‐to‐wall distance (BL, post‐treatment after 4 weeks) (in cm, higher is worse) Ketoprofen 200 mg: 8.0 (N = 12), 8.3 (N = 12) Phenylbutazone 300 mg: 8.8 (N = 8), 8.6 (N = 8) (P = not significant) 8. Intermalleolar distance (BL, post‐treatment after 4 weeks) (in cm, higher is better) Ketoprofen 200 mg: 85.2 (N = 12), 85.9 (N = 12) Phenylbutazone 300 mg: 102.0 (N = 8), 105.0 (N = 8) (P = not significant) 9. Schober's test (BL, post‐treatment after 4 weeks) (in cm, higher is better) Ketoprofen 200 mg: 2.0 (N = 12), 1.9 (N = 12) Phenylbutazone 300 mg: 2.4 (N = 6), 2.4 (N = 6) (P = not significant)
Notes	Outcomes that were not included in the meta‐analysis, because no measure of variance (SD, SE or CI) was reported for these outcomes: chest expansion, tragus‐to‐wall distance, occiput‐to‐wall distance, intermalleolar distance, Schober's test. Available results are described in this table. For the outcomes pain on Likert scale, patient global assessment of disease activity, duration of morning stiffness and severity of morning stiffness individual patient data that were presented in the paper were combined for the meta‐analysis to a mean and SD. For the outcome tragus‐to‐wall distance, "straight tragus‐to‐wall distance" was extracted, and not "turning to right" or "turning to left" (which was also presented in the paper). For the outcome occiput‐to‐wall distance, "greatest displacement tolerated" was extracted, and not "when pain first appears" (which was also presented in the paper. Presented demographics are those of the participants that completed the trial. Only results of first part of cross‐over trial are used in analysis. Funding source: May & Baker Ltd. provided the drugs and record cards used in this trial.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	"The patients were randomly allocated..." Probably done, but no further information provided on sequence generation.
Allocation concealment (selection bias)	Unclear risk	No information provided on allocation concealment.
Blinding of participants and personnel (performance bias) All outcomes	Low risk	"in identical capsules".
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	"double‐blind". Probably done, but no further information provided on blinding outcome assessors.
Incomplete outcome data (attrition bias) All outcomes	High risk	Dropouts (6) were excluded from all analysis, no information provided on whether dropouts differed from analysed participants. Although number of dropouts per group was equal (2x3), the reason for dropout differed between treatment groups, possibly introducing bias.
Selective reporting (reporting bias)	Low risk	All pre‐specified outcomes according to the methods section, are reported in the results section.
Other bias	Unclear risk	Crossover design, no information provided on possible carry‐over effect. BL inequality cannot be assessed because demographic information was only provided on completers.