Table 3

Risk stratification and prediction models for RA

StudyDesign and primary outcomeFindings
van de Stadt et al 201327Prospective evaluation of 374 individuals who presented to several Dutch rheumatology clinics with arthralgia and ACPA and/or RF positivity and without IA at baseline. 131 (35%) participants developed IA after a median of 12 months.Nine features were included in the model:
  1. First degree relative with RA (1 point).

  2. Drinking alcohol (1 point if no).

  3. Symptoms start <12 months prior (1 point).

  4. Intermittent symptoms (1 point).

  5. Symptoms in upper and lower extremities (1 point).

  6. Pain scale ≥50 on Visual Analogue Scale (2 points).

  7. Morning stiffness ≥1 hour (1 point);

  8. Self-reported swelling in any joint (1 point).

  9. Autoantibodies:

    • RF-IgM pos/ACPA neg (0 points).

    • RF-IgM negative and ACPA positive <3×cut-off (2 points);

    • RF-IgM negative and ACPA positive ≥3×cut-off (3 points).

    • RF-IgM and ACPA positive (4 points).


All features if present (or positive) indicate point(s) with the exception of alcohol intake which if negative=1 point. Scores of 7–13 were associated with rates of development of IA of ~74% by 3 years, and ~81% at 5 years.
Karlson et al 201292Evaluation of the interactions between environmental factors, genetic risk scores and gene–environment interactions for the development of RA in the Nurses’ Health Study (NHS: 317 cases, 551 controls), and the Swedish Epidemiologic Investigation of RA cohort (EIRA: 987 cases and 958 controls).Primary models produced an AUC of 0.72 in NHS, 0.73 in EIRA women and 0.76 in EIRA men. A full environmental, genetic and gene–environment interaction model provided optimal predictive ability.
Sparks et al 201593Evaluation of epidemiological and genetic risk models incorporating autoantibody profiles, family history, genetics, smoking and body mass index (BMI) among women in the Nurses’ Health Study (NHS, 381 RA cases and 410 controls) and the Swedish Epidemiological Investigation of RA (EIRA, 1244 RA cases and 971 controls).Models demonstrated AUCs of 0.74 for seropositive RA in the NHS and 0.77 for ACPA+RA in EIRA. Discrimination was improved for women with a family history in the NHS (AUC 0.82) and in EIRA (AUC 0.83). When combining positive family history, high genetic susceptibility, smoking and increased BMI had an OR of 21.73 for ACPA-positive RA.
Matthijssen et al 2019 and 202094 95580 individuals with
CSA were followed for the development of IA; 18% progressed to IA within 2 years.
The model contained 4 variables: ACPA-positivity, RF-positivity, >2 locations of
subclinical inflammation on MRI and presence of MCP-extensor peritendinitis on MRI. PPVs were up to 86%. This model yielded an AUC of 0.79.
Duquenne et al 202328A single centre (Leeds, UK) prospective evaluation of 455 ACPA(+) individuals who initially presented to primary care with musculoskeletal complaints; 148 (32.5%) participants developed IA after a median of 255 weeks of follow-up.A ‘simple’ score method was recommended to be applied in primary care and included:
  • Morning stiffness 30 minutes or more (9 points)

  • Levels of ACPA (8 points if ≥3 to <10×ULN; 17 points if ≥10×ULN)

  • Positivity of RF (11 points)

  • An elevated ESR (6 points)

  • A ‘simple’ score ≥18 of was associated with a PPV of ~27% of developing clinical RA within 1 year.

  • A ‘comprehensive’ categorical scoring method was recommended to be applied in rheumatological practice and included ~13 variables (including ultrasound findings), and a certain score was associated with a PPV of ~70% of developing IA within 5 years.

Prajzlerová et al, 2021 and 202489 96The ARRA cohort (Prague) at-risk individuals is defined as having arthralgia without arthritis and being either ACPA or meeting the clinical EULAR CSA definition. Immune phenotyping of PBMC was undertaken by flow cytometry.An initial report identified a shift from classical (CD14++CD16) to non-classical peripheral blood monocytes (CD14+/−CD16++) in ACPA+ at-risk individuals. A follow-up predictive model identified risk factors associated with progression to arthritis including high ACPA IgG, higher % of B cells and lower % of NK cells (AUC 0.78). This was not the case in ACPA negative individuals.
  • ACPA, anticitrullinated protein antibody; ARRA, At-Risk of RA cohort; AUC, area under the curve; CSA, clinically suspect arthralgia; ESR, erythrocyte sedimentation rate; EULAR, European Alliance of Associations for Rheumatology; PBMC, peripheral blood mononuclear cells; PPV, positive predictive value; RA, rheumatoid arthritis; RF, rheumatoid factor; ULN, upper limit of normal.