Determining the level of cervical radiculopathy: agreement between visual inspection of pain drawings and magnetic resonance imaging

Key points

1. 01Overall agreement between pain drawing interpretation and MRI was 37.7% (range 36.7% to 44.9%)
2. 02Kappa values comparing clinicians to MRI ranged from -0.03 to 0.2, indicating no agreement beyond chance
3. 03Inter-clinician agreement was only fair to moderate (kappa 0.21 to 0.4)
4. 04C7 level was most accurately identified (47.7%) followed by C6 (37.7%); C4 and C5 showed near-zero agreement
5. 05Pain distributions in patients did not follow classic dermatomal patterns and showed marked overlap between levels

How it was conducted

Design

Secondary analysis of baseline data from a randomized controlled trial; post hoc agreement study

Participants

98 patients with MRI-confirmed single-level cervical radiculopathy (C4-C7), scheduled for surgical decompression, recruited from 4 neurosurgery clinics in southern Sweden (2009-2012)

Raters

4 experienced clinicians: 2 musculoskeletal physiotherapists (7 and 15 years experience) and 2 spine surgeons (15 and 35 years experience), blinded to all clinical data

Procedure

Each clinician independently viewed digitised pain drawings for all 98 patients and assigned a single cervical nerve root level (C4, C5, C6, or C7)

Primary outcome

Percentage agreement and Cohen's kappa between clinician-derived level from pain drawing and MRI-verified level, and inter-clinician kappa

What they found

• Overall mean percentage agreement between pain drawing interpretation and MRI-identified level was 37.7% (range 36.7% to 44.9%)
• Kappa values for MRI vs each clinician: PT1 kappa = -0.03, PT2 kappa = 0.17, S1 kappa = -0.06, S2 kappa = -0.04; none were statistically significant beyond chance
• Inter-clinician kappa values ranged from 0.21 (PT1 vs PT2) to 0.4 (S1 vs S2, PT1 vs S1, PT1 vs S2), indicating fair to moderate agreement
• Physiotherapists correctly identified the affected level 32.2% of the time vs 32.6% for spine surgeons
• Agreement for C7 level: mean 47.7% (range 40.8%-55.0%); for C6: mean 37.7% (range 32.9%-43.9%)
• C4 and C5 levels showed little to no agreement, with a range of 0%-25% across all clinicians
• Three of four clinicians performed worse than would be expected by chance alone when kappa was applied

Limitations

• Small subgroup sizes for C4 (n=4) and C5 (n=7) likely contributed to the poor agreement at upper cervical levels
• Clinicians had no access to patient history, physical examination findings, or symptom questionnaires, which does not reflect real clinical practice
• Only four clinicians from three countries were assessed, limiting generalizability to broader clinical populations
• Absence of additional reference standards such as nerve root blocks or electromyography limits confirmation of the true affected level

Why it matters

For patients

Patients with cervical radiculopathy should know that the location of their pain alone is unlikely to pinpoint which nerve is involved, so MRI and full clinical examination are essential for accurate diagnosis.

For clinicians

Pain drawings should not be used in isolation to determine the level of cervical radiculopathy; they should be integrated with imaging and full clinical assessment, as even experienced specialists cannot reliably identify the nerve root from pain distribution alone.

For readers

This study adds quantitative evidence that classic dermatomal maps poorly predict the actual nerve root affected in cervical radiculopathy, and highlights the need for multicomponent diagnostic approaches.