Leiden Clinical Prediction Rule (undifferentiated arthritis)

For adults with recent-onset undifferentiated arthritis. Enter examination and laboratory data as close as possible to presentation. Joint counts follow the rule's published bands (tender 0–68, swollen 0–66 in the original description).

Age (years) *

Points: age × 0.02

Sex *

Involvement of small joints of hands or feet? *

Symmetric joint involvement? *

Morning stiffness ≥60 minutes and/or stiffness VAS ≥26 mm? *

Tender joint count (0–68) *

1–3 → 0.5; 4–10 → 1; ≥11 → 1.5 pts

Swollen joint count (0–66) *

Same bands as tender joints

C-reactive protein (mg/L) *

≥5 mg/L → +1.5 pts (≈0.5 mg/dL)

Rheumatoid factor *

Anti-CCP (ACPA) *

Disclaimer: Educational support only. The Leiden rule supports risk estimation in UA; it does not diagnose RA or replace rheumatology assessment, repeat serology, imaging, or guideline-based care.

Leiden Clinical Prediction Rule — overview

The Leiden prediction rule (van der Helm-van Mil et al., 2007) estimates the likelihood of progression to rheumatoid arthritis (RA) within about one year in patients with recent-onset undifferentiated arthritis (UA). It combines demographics, joint pattern, morning stiffness, joint counts, CRP, rheumatoid factor (RF), and anti-citrullinated peptide antibodies (anti-CCP / ACPA).

Component scoring

Variable	Coding	Points
Age	Years	0.02 × age
Sex	Female / male	1 / 0
Small joints (hands/feet)	Involved vs not	0.5 / 0
Symmetric involvement	Yes vs no	0.5 / 0
Morning stiffness	≥60 min and/or stiffness VAS ≥26 mm vs not	1 / 0
Tender joints (each of tender and swollen uses the same bands)	1–3	0.5
	4–10	1
	≥11	1.5
Swollen joints	Same bands as tender	0 / 0.5 / 1 / 1.5
CRP	≥5 mg/L (≈0.5 mg/dL) vs below	1.5 / 0
RF	Positive vs negative	1 / 0
Anti-CCP (ACPA)	Positive vs negative	2 / 0

Total score = sum of all components. Because age is continuous, totals are not limited to integers. Many publications describe risk bands; this calculator labels <6 as lower, 6–7.9 as intermediate, and ≥8 as higher probability of RA progression—cutoffs vary in external validations and systematic reviews.

How to use the result

Support decisions about intensity of follow-up, repeat serology, and early DMARD discussion in UA—not as a stand-alone diagnostic test.
Compare performance to local cohorts; AUC and optimal thresholds differ across settings and eras (including anti-CCP assay generation).
Reassess if the clinical picture evolves or if new autoantibody or imaging data return.

Reference: van der Helm-van Mil AHM, le Cessie S, van Dongen H, et al. A prediction rule for disease outcome in patients with recent-onset undifferentiated arthritis: how to guide individual treatment decisions. Arthritis Rheum. 2007;56(2):433-440. doi:10.1002/art.22380

Background: undifferentiated arthritis and why risk stratification matters

Undifferentiated arthritis (UA) describes a common early presentation: inflammatory joint symptoms that do not yet meet classification criteria for a definite rheumatic diagnosis such as rheumatoid arthritis (RA). In many cohorts, a substantial proportion of patients improve without ever developing persistent inflammatory disease, while others evolve toward a chronic, erosive pattern for which early use of disease-modifying antirheumatic drugs (DMARDs) and tight control strategies can change long-term joint damage and function.

Because the first weeks to months are both therapeutically sensitive and diagnostically ambiguous, clinicians need tools that translate readily available history, examination, and laboratory data into an estimate of short-term outcome. The Leiden Clinical Prediction Rule was developed in the Leiden Early Arthritis Clinic setting to address that need: it produces a weighted score reflecting how strongly the initial presentation resembles the profile of patients who progressed to RA in the derivation work, without replacing careful clinical judgment, repeat assessment, or evolving serologic and imaging information.

What the rule estimates

The rule is framed around progression toward RA within approximately one year among adults presenting with recent-onset UA. It is not a stand-alone diagnostic test for RA, nor does a low score guarantee remission; it quantifies how closely a patient’s baseline features align with patterns that were associated with RA development in the original modeled cohort. As with any clinical prediction rule, performance—discrimination, calibration, and the performance of specific score cutoffs—varies when the tool is transported to populations with different referral pathways, ethnic backgrounds, prior treatment exposure, and laboratory assay characteristics.

Variables included and their clinical meaning

Age

Age enters the score as a continuous contribution: each year adds a small fixed increment. Inflammatory arthritis can occur at any adult age, but the statistical association between age and one-year RA progression in the derivation dataset motivated retaining this term. Practically, older patients can accumulate slightly higher totals even when joint counts and autoantibodies are modest, which is one reason the final score is not a simple integer sum of categorical items alone.

Sex

Female sex receives a fixed increment in the published scoring scheme, reflecting the higher prevalence and incidence patterns of RA and the sex distribution of progression in the development cohort. This component should be interpreted as a population-derived statistical weight, not as a statement about any individual patient’s certain trajectory.

Localization: small joints of the hands and feet

Involvement of the small joints of the hands or feet is weighted because RA disproportionately targets these areas over time, and early polyarticular small-joint inflammatory patterns overlap more strongly with eventual RA than do patterns confined to large joints alone. Clinical assessment should focus on synovitis or clear inflammatory arthropathy in metacarpophalangeal, proximal interphalangeal, metatarsophalangeal, or other small peripheral joints as judged at the index visit.

Symmetry

A symmetric pattern of affected joints—roughly similar joints involved on both sides of the body—is another feature that, in aggregate, tracks with diagnoses that behave like RA. Perfect mirror-image symmetry is not required; the intent is to capture whether the process is broadly bilateral rather than strictly unilateral or oligoarticular in a way more suggestive of alternative etiologies.

Morning stiffness

Prolonged morning stiffness is a classic inflammatory symptom. In the rule, the stiffness item is satisfied when stiffness lasts at least 60 minutes and/or reaches a threshold on a visual analog scale (VAS) for stiffness (≥26 mm), reflecting the original operational definitions. This dual pathway acknowledges that patients quantify stiffness differently in narrative time versus on a structured scale, and either can indicate clinically meaningful inflammatory morning symptoms.

Tender and swollen joint counts

The rule uses separate increments for tender and swollen joint counts, each mapped into ordered bands (for example, none; a few joints; a moderate number; and higher burdens). Tenderness captures patient-relevant pain on examination or palpation, while swelling emphasizes objective synovitis or soft-tissue inflammatory swelling. Both dimensions contribute independently because some patients have pain-predominant presentations early on, while others show more obvious effusion or synovial thickening with relatively modest tenderness scores.

When applying the calculator, enter whole-number counts consistent with your joint examination method. Small differences near band boundaries can move the score by a discrete step, so repeat examination at follow-up may reclassify risk even when serology is unchanged.

C-reactive protein (CRP)

CRP is a nonspecific but responsive marker of systemic inflammation. The rule assigns a substantial increment when CRP is at or above a conventional elevated threshold expressed in mg/L (commonly aligned with ≥5 mg/L, corresponding to about 0.5 mg/dL in alternate units). Normal or low CRP does not exclude inflammatory arthritis—especially if symptoms are early, localized, or partially treated—but higher values strengthen the inflammatory phenotype that tracked with RA progression in the derivation data.

Rheumatoid factor (RF)

Rheumatoid factor is an autoantibody associated with RA and with several other conditions. In this rule, seropositivity adds a fixed increment because RF-positive UA patients more often evolve toward persistent, classifiable RA in many cohorts. Laboratory cutoffs should follow your assay’s certified reporting; borderline or low-titer positives still count as positive for the purpose of the score but should be integrated with clinical context and repeat testing where appropriate.

Anti-citrullinated peptide antibodies (anti-CCP / ACPA)

Anti-CCP antibodies (also called ACPA) carry the largest single categorical weight in the rule. These antibodies are relatively specific for RA-associated autoimmunity and are among the strongest predictors of progression from UA to RA across multiple independent studies. A positive anti-CCP result therefore raises the score sharply, often shifting a patient into higher-risk strata even when joint counts are modest—again underscoring that the tool encodes population-level associations rather than a pathophysiologic explanation for any one person.

How the total score behaves

The total Leiden score is the sum of all weighted components. Because age contributes continuously, totals may include fractional values; this is expected and not an error. Many educational implementations and secondary analyses describe risk bands using cutoffs such as scores below 6 as comparatively lower probability, an intermediate gray zone around 6 to just under 8, and scores of 8 or higher as a higher-probability group in the original reporting and numerous validations. External studies and systematic reviews have discussed a range of alternative thresholds when optimizing sensitivity versus specificity, so local guidelines may emphasize different cut points depending on whether the goal is to minimize undertreatment, overtreatment, or resource use.

Integrating the score into care

Practical use of the Leiden rule typically parallels a broader early-arthritis pathway: ensure appropriate baseline investigations, arrange short-interval follow-up for patients in intermediate zones, and discuss earlier DMARD consideration and comorbidity review for patients at the higher end of the score distribution—always within the framework of shared decision-making, fertility and infection risk discussion, vaccination planning, and access to rheumatology. Imaging (such as ultrasound or magnetic resonance imaging of clinically affected joints) may refine synovitis assessment when examination is equivocal, and repeat serology can clarify initially negative or low-titer results.

The calculator on this site is provided for education and documentation support. It should not be interpreted as prescribing a specific medication, dose, or monitoring interval, and it does not replace specialist evaluation when symptoms are severe, rapidly progressive, or accompanied by systemic features that require alternative diagnoses to be excluded urgently.

CalcMD.

Leiden Clinical Prediction Rule for Undifferentiated Arthritis

Learn the Leiden Clinical Prediction Rule for recent-onset undifferentiated arthritis: scoring (age, sex, joint pattern, stiffness, tender/swollen counts, CRP, RF, anti-CCP), risk bands, and how to use the score alongside early RA pathways and shared decision-making.