Grade radiographic morphology only after fixing joint, side, projection and description lineage; expose feature-level disagreement and never use the number as a proxy for pain, clinical diagnosis, prognosis or treatment need.
{
"categories": [
{
"outcome_code": "0",
"label": "none",
"canonical_knee_description": "No radiographic osteoarthritis features.",
"osteophytes": "absent",
"joint_space_narrowing": "absent",
"subchondral_sclerosis": "absent",
"bone_end_deformity": "absent"
},
{
"outcome_code": "1",
"label": "doubtful",
"canonical_knee_description": "Doubtful joint-space narrowing and possible osteophytic lipping.",
"osteophytes": "possible",
"joint_space_narrowing": "doubtful",
"subchondral_sclerosis": "not_required",
"bone_end_deformity": "absent"
},
{
"outcome_code": "2",
"label": "mild_or_definite_radiographic_OA",
"canonical_knee_description": "Definite osteophytes with possible joint-space narrowing.",
"osteophytes": "definite",
"joint_space_narrowing": "possible",
"subchondral_sclerosis": "not_required",
"bone_end_deformity": "absent_or_not_required"
},
{
"outcome_code": "3",
"label": "moderate",
"canonical_knee_description": "Multiple or moderate osteophytes, definite joint-space narrowing, some sclerosis and possible deformity of bone ends.",
"osteophytes": "multiple_or_moderate",
"joint_space_narrowing": "definite",
"subchondral_sclerosis": "some",
"bone_end_deformity": "possible"
},
{
"outcome_code": "4",
"label": "severe",
"canonical_knee_description": "Large osteophytes, marked or severe joint-space narrowing, marked or severe sclerosis and definite bone-end deformity.",
"osteophytes": "large",
"joint_space_narrowing": "marked_or_severe",
"subchondral_sclerosis": "marked_or_severe",
"bone_end_deformity": "definite"
}
],
"applicability": {
"intended_use": "Descriptive and epidemiologic grading of radiographic osteoarthritis morphology on plain radiographs using a declared joint-specific implementation of the Kellgren-Lawrence framework.",
"classification_unit": "one_named_joint_side_and_declared_projection_or_a_prespecified_multiview_joint_assessment",
"common_knee_use": "The canonical 0-4 descriptions are most often applied to tibiofemoral knee osteoarthritis, commonly on a weight-bearing projection. Other joints require explicit joint-specific criteria or atlas provenance.",
"outside_scope": [
"clinical_diagnosis_of_osteoarthritis_from_the_radiograph_alone",
"localization_of_the_source_of_pain",
"direct_cartilage_measurement",
"MRI_or_ultrasound_osteoarthritis_scoring_without_a_validated_mapping",
"inflammatory_infectious_crystal_or_neoplastic_arthropathy",
"treatment_or_arthroplasty_selection_from_grade_alone",
"individual_progression_or_prognosis_prediction"
]
},
"acquisition_and_provenance_gate": {
"mandatory_metadata": [
"joint",
"side",
"compartment_or_whole_joint_scope",
"projection",
"weight_bearing_status",
"knee_flexion_or_position_when_relevant",
"atlas_or_grade_description_version",
"technical_adequacy",
"comparison_date_and_technical_comparability_when_longitudinal"
],
"knee_projection_rule": "Joint-space width changes with loading, beam angle and flexion. Do not compare a non-weight-bearing AP image directly with a weight-bearing flexed PA/Rosenberg view as if grade change were biologic progression.",
"compartment_rule": "An AP tibiofemoral view can underrepresent patellofemoral or compartment-specific disease. Describe the involved compartment and use the required additional views rather than assigning whole-knee normality from incomplete coverage.",
"image_quality_rule": "Rotation, flexion, beam centering, magnification and incomplete joint coverage can alter apparent narrowing and osteophyte conspicuity. Return technically limited when the grade boundary cannot be trusted."
},
"feature_model": {
"osteophytes": "Marginal bony proliferation is central to the classic threshold, especially the grade 1-versus-2 boundary.",
"joint_space_narrowing": "An indirect, load- and position-dependent marker influenced by cartilage, meniscus and alignment; radiography does not directly depict articular cartilage.",
"subchondral_sclerosis": "A supporting osseous response used in higher grades, not a standalone grade or pain marker.",
"bone_end_deformity": "Remodeling or contour deformity contributes to grades 3-4 and must not be confused with acute fracture, dysplasia or another arthropathy.",
"whole_pattern_rule": "Apply the declared description to the whole feature pattern. Do not average component severity or import OARSI component grades into a Kellgren-Lawrence number."
},
"boundary_rules": {
"grade_0_1": "Grade 1 requires only doubtful narrowing or possible lipping. When the possible feature cannot be distinguished from projection or normal variation, retain 0-versus-1 uncertainty.",
"grade_1_2": "A definite osteophyte is the usual canonical boundary for grade 2. Definite narrowing without a definite osteophyte exposes a known description/atlas ambiguity; report the feature and lineage rather than silently inventing a mapping.",
"grade_2_3": "Definite narrowing plus more extensive osteophytes and some sclerosis supports grade 3; possible narrowing with definite osteophytes remains grade 2 under the canonical description.",
"grade_3_4": "Marked narrowing, large osteophytes, marked sclerosis and definite bone-end deformity support grade 4. Do not call grade 4 from isolated severe narrowing without documenting the rest of the pattern and atlas rule.",
"epidemiologic_threshold": "Kellgren-Lawrence grade 2 or higher is a common research definition of definite radiographic knee osteoarthritis, not a universal clinical diagnostic threshold."
},
"description_lineage_and_variants": {
"historical_problem": "The 1957 paper described component features but later publications created multiple non-identical grade descriptions. A 2008 review identified at least five versions in use.",
"agent_rule": "Return the exact description or atlas used. Never resolve a discordant feature pattern by switching versions invisibly.",
"OARSI_boundary": "OARSI atlas component scores grade osteophytes and joint-space narrowing separately and are not numerically interchangeable with a Kellgren-Lawrence whole-joint grade.",
"cross_joint_boundary": "Do not export a knee threshold to hip, hand, spine or another joint without a source validating that joint-specific application."
},
"clinical_diagnosis_and_management_boundary": {
"current_guideline_context": "NICE NG226 recommends clinical diagnosis without routine imaging in typical adults aged 45 or older with activity-related pain and no morning stiffness or stiffness no longer than 30 minutes; image when atypical or an alternative/additional diagnosis is suspected.",
"symptom_function_rule": "Management is guided by symptoms and physical function, not the radiographic grade. Core care includes information/support, therapeutic exercise and weight management when appropriate.",
"follow_up_imaging_rule": "Do not routinely use imaging to follow or guide non-surgical osteoarthritis management.",
"arthroplasty_rule": "Referral for joint replacement depends on substantial symptom and quality-of-life impact plus ineffective or unsuitable non-surgical management. Use clinical assessment rather than a numerical imaging score alone.",
"safety_differential": "A hot swollen joint, rapid worsening, prolonged stiffness, recent trauma, destructive change or concern for infection, inflammatory arthritis or malignancy overrides routine KL grading and requires the appropriate pathway."
},
"symptoms_risk_and_reliability": {
"pain_discordance": "A systematic search found wide discordance: among people with knee pain, radiographic osteoarthritis prevalence ranged from 15 to 76 percent, and among those with radiographic osteoarthritis, pain prevalence ranged from 15 to 81 percent. These study ranges demonstrate heterogeneity, not an individual probability.",
"early_disease_limit": "Grade 0 or 1 does not exclude symptomatic, early, compartmental or non-radiographic disease.",
"severe_grade_limit": "Grade 4 is severe radiographic morphology but does not prove symptom severity, inevitability of surgery or a specific outcome.",
"current_reliability_example": "In one 2026 retrospective 576-patient knee/hip cohort, interspecialty kappa was 0.598 for knee and 0.552 for hip, with greater disagreement in subtle early features and systematic severity differences. Treat these as cohort-specific reliability evidence, not universal performance.",
"progression_rule": "Do not infer change from a one-grade difference unless projections and technique are comparable and the feature-level change is visible. A categorical change may reflect reader or acquisition variability."
},
"agent_output_contract": {
"always_return": [
"joint_side_compartment_and_projection",
"weight_bearing_and_position_status",
"grade_and_description_lineage",
"osteophyte_joint_space_sclerosis_and_deformity_findings",
"technically_limited_or_boundary_uncertainty",
"clinical_symptoms_and_function_as_unknown_unless_explicitly_supplied",
"alternative_or_urgent_differential_when_present"
],
"prohibited_inferences": [
"pain_generator_from_grade",
"clinical_OA_diagnosis_from_image_alone",
"arthroplasty_or_other_treatment_from_grade",
"individual_prognosis_from_grade"
]
},
"missing_input_behavior": [
"Without joint, side and projection, return insufficient provenance and do not emit a portable KL grade.",
"Without weight-bearing and position metadata for a knee, preserve acquisition uncertainty, especially at a joint-space boundary.",
"Without a declared grade description or atlas, use the canonical description only if explicitly labeled and expose any discordant feature pattern.",
"Without adequate compartmental views, do not call the entire joint grade 0 from an incompletely assessed compartment.",
"Without symptoms and function, report radiographic morphology only and withhold clinical diagnosis, pain attribution and treatment advice.",
"When features straddle adjacent grades, return the two-grade boundary and the deciding missing or discordant feature rather than guessing."
],
"supporting_sources": [
{
"role": "primary",
"citation": "Kellgren and Lawrence. Ann Rheum Dis. 1957;16:494-502",
"doi": "10.1136/ard.16.4.494"
},
{
"role": "classification_review",
"citation": "Kohn et al. Clin Orthop Relat Res. 2016;474:1886-1893",
"doi": "10.1007/s11999-016-4732-4",
"pmcid": "PMC4925407"
},
{
"role": "description_variants",
"citation": "Schiphof et al. Ann Rheum Dis. 2008;67:1034-1036",
"doi": "10.1136/ard.2007.079020",
"pmid": "18198197"
},
{
"role": "pain_discordance",
"citation": "Bedson and Croft. BMC Musculoskelet Disord. 2008;9:116",
"doi": "10.1186/1471-2474-9-116",
"pmcid": "PMC2542996"
},
{
"role": "current_clinical_guideline",
"citation": "NICE NG226",
"url": "https://www.nice.org.uk/guidance/NG226/chapter/recommendations",
"published": "2022-10-19"
},
{
"role": "current_reliability_example",
"citation": "Reliability of radiographic knee and hip osteoarthritis classification between radiologists and orthopaedic surgeons. 2026",
"doi": "10.1186/s13018-026-06695-6",
"pmcid": "PMC12964767"
}
],
"source_locator": "Kellgren and Lawrence 1957, DOI 10.1136/ard.16.4.494; Kohn et al. 2016, PMC4925407, Description and Validation; Schiphof et al. 2008, DOI 10.1136/ard.2007.079020; Bedson and Croft 2008, PMC2542996, Results; NICE NG226 recommendations 1.1.1-1.1.2, 1.2.2, 1.3.1, 1.5.4 and 1.6.1-1.6.2; 2026 reliability study PMC12964767, Results and Discussion."
}