Skip to main content

Table 2 Variables of the final flare prediction model

From: Using real-world data to dynamically predict flares during tapering of biological DMARDs in rheumatoid arthritis: development, validation, and potential impact of prediction-aided decisions

Parameter

Hazard ratio (95% CI)

Linear time coefficient DAS28 trajectory latent class 1

1.04 (1.02–1.06)

Quadratic time coefficient DAS28 trajectory latent class 1

1.66 (0.42–6.55)

Linear time coefficient DAS28 trajectory latent class 2

1.14 (1.08–1.20)

Quadratic time coefficient DAS28 trajectory latent class 2

4.52 (3.83–5.33)

Time to reach stable low disease activity (weeks)a

0.97 (0.96–0.98)

DAS28 at baseline

1.18 (0.90–1.54)

Prescribed dose (% of standard dose) at baseline

1.21 (0.88–1.67)

SJ increase at baseline (yes/no)b

1.72 (0.94–3.17)

TJ increase at baseline (yes/no) b

2.07 (1.13–3.81)

Disease duration (years) at start of bDMARD

1.02 (0.99–1.05)

Seropositivity (RF and/or ACPA)

2.51 (1.39–4.53)

bDMARD TNFi type (yes/no)

0.90 (0.54–1.49)

bDMARD dose ≤50% of full registered dose (time-varying variable)

2.21 (1.73–2.82)

  1. ACPA anti-citrullinated protein antibody, bDMARD biological disease-modifying antirheumatic drug, DAS28 disease activity score based on 28-joint count, RF rheumatoid factor, TJ(C)/SJ(C) tender/swollen joint count, TNFi tumor necrosis factor inhibitor
  2. In development data, baseline is defined as the first DAS28 ≤ 3.2 (low disease activity)
  3. aIn development data: time from start biological until DAS28 < 3.2 for the first time. In DRESS data: time from start biological until baseline visit
  4. bAn increase in TJC or SJC (yes/no) at baseline, compared to the previous DAS28 measurement
Back to article page

Arthritis Research & Therapy

ISSN: 1478-6362

Contact us