Table 1

Clinical characteristics of the rheumatoid arthritis (RA) and non-RA comparator groups used to generate list of differentially expressed genes, which together comprise a training cohort for machine-learning (total n=111) and the independent undifferentiated arthritis (UA) validation cohort (n=62)

Training cohortTest cohort
RA (n=47)Non-RA (n=64)p*UA (n=62)
Age (years; mean, SD range)60 (46–74)48 (34–62)0.0152 (37–67)
% Female6561NS77
% White Caucasian9692NS90
Symptom duration (weeks; median, IQR)12 (8–24)21 (10.5–52)0.02614 (12–26)
Tender joint count (median, IQR)10 (4–15)7 (2–14)NS8 (3–16.5)
Swollen joint count (median, IQR)6 (2–10)0 (0–2)<0.0011 (0–3)
Morning stiffness (hours; median, IQR)1 (0.75–2)0.75 (0.1–2)0.0071 (0.5–2)
ESR (s; median, IQR)56 (30–78)24 (14–52)<0.00130 (18–60)
CRP (g/l; median, IQR)17 (9–62)5 (2.5–19)<0.0018.5 (0–17)
ACPA+ (number; per cent)29 (62)0 (0)<0.00113 (21)
RF+ (number; per cent)36 (77)3 (6)<0.00120 (32)
DAS28 (median)5.37NA
Leiden prediction score (median, IQR)NANA6.4 (5–7.6)
Outcome diagnoses (number, per cent)
 RA47 (100)0 (0)25 (40)
 Seronegative sponyloarthropathy22 (34)8 (13)
 Self-limiting inflammatory12 (19)9 (15)
 Other inflammatory3 (5)2 (3)
 OA/non-inflammatory27 (42)18 (29)
  • Values are mean (1 SD range), median (IQR) or % for normally distributed, skewed or dichotomous data, respectively.

  • * Statistical tests for significant difference between RA and non-RA groups; t-test, Mann–Whitney U or Fisher's exact test for normally distributed, skewed or dichotomous data, respectively.

  • Demographic, clinical and serological parameters are given for UA–RA and UA–non-RA subgroups in online supplementary table S2. ACPA, anti-citrullinated peptide antibodies; CRP, C reactive protein; DAS28, disease activity score (incorporating 28-swollen/tender joint counts); ESR, erythrocyte sedimentation rate; NS, not significant; OA, osteoarthritis; RA, rheumatoid arthritis; RF, rheumatoid factor; UA, undifferentiated arthritis.