4.3 Objective Examination, Special Tests, and Diagnostic Accuracy
Key Takeaways
Choose examination steps for safety, function and the hypotheses being tested. Bring forward neurological assessment when the history warrants it.
Record end-feel, force and symptoms as examination findings. Interpret them with the rest of the assessment before assigning a lesion.
Combine bilateral sensation, strength and reflex findings with gait or long-tract testing as indicated. Sensory landmarks do not isolate one root in every patient.
Sensitivity and specificity use different denominators. Likelihood ratios multiply pretest odds, and the probability shift depends on the starting probability.
Apply diagnostic clusters using the studied test combination and population. Progressive cord symptoms require referral regardless of cluster count.
Objective Examination, Special Tests, and Diagnostic Accuracy
Use the examination to test the most important hypotheses, document function and identify findings that would change referral or treatment. Address urgent concerns first. When routine examination is appropriate, start with relevant observations and lower-demand tasks, then add more specific tests based on the history and the patient's response.
Movement, resistance and end-feel
Observe a meaningful task. Compare active motion, symptoms and control. Add passive movement or resistance if it answers a clinical question and is safe. Do not wait until the end of a sequence to assess neurological function if the history raises concern for nerve-root or cord involvement. Pain-free active motion alone does not establish that passive overpressure is appropriate.
End-feel describes the examiner's perception of resistance near the point where passive movement stops. Examples include soft tissue approximation in elbow flexion, firm resistance with tissue stretch, and an abrupt bony limit in some elbows at extension. If pain stops a movement before resistance can be assessed, record that observation. A springy block, guarding or an unexpected change in motion can direct further assessment. An end-feel alone cannot identify a tear, fracture, effusion or malignancy.
In a resisted isometric test, record the position, force or grade, familiar symptoms, and any substitution or movement. A load can provoke symptoms arising from several structures. The classic selective-tension approach compares pain and force during resistance with responses during active and passive movement. Interpret the comparison as evidence for or against a working hypothesis.
| Resisted finding | Interpretation to examine further |
|---|---|
| Strong and painless | No pain or apparent force loss during that task. A lesion could still be missed by the tested position or load. |
| Strong and painful | The maneuver reproduces pain while force is maintained. Check symptom location and which loaded structures are plausible. |
| Weak and painful | Pain can limit effort. Structural injury and other causes of reduced force are also possible. Use the rest of the assessment to distinguish them. |
| Weak and painless | Consider motor deficit, structural injury or another cause of weakness in the context of history and examination. |
Neurological screening
Compare sides and record motor, sensory and reflex findings together. Examine gait, coordination and long-tract signs when indicated. Radiculopathy can present with weakness, altered sensation or a reduced reflex. In suspected cord dysfunction, assess for hand dexterity changes, gait impairment and brisk reflexes or an extensor plantar response. Segmental and long-tract findings can coexist. A single absent or present sign does not establish the diagnosis.
The following are selected ISNCSCI sensory landmarks and key muscle functions used in standard spinal cord examination. They offer reproducible test sites. Clinical nerve-root territories overlap and the same findings can arise from peripheral nerve lesions. Common tendon reflex checks in a root examination include biceps and brachioradialis for C5–C6, triceps predominantly C7, patellar L3–L4 and Achilles predominantly S1. The reflex findings are separate from the ISNCSCI level scoring.
| Segment | Sensory landmark | Key motor function |
|---|---|---|
| C5 | Lateral antecubital fossa | Elbow flexion |
| C6 | Dorsal proximal phalanx of thumb | Wrist extension |
| C7 | Dorsal proximal phalanx of middle finger | Elbow extension |
| C8 | Dorsal proximal phalanx of little finger | Middle-finger distal interphalangeal flexion |
| T1 | Medial antecubital fossa | Little-finger abduction |
| L2 | Anteromedial thigh halfway between the midpoint of the inguinal ligament and the medial femoral condyle | Hip flexion |
| L3 | Medial femoral condyle | Knee extension |
| L4 | Medial malleolus | Ankle dorsiflexion |
| L5 | Dorsum of foot at third metatarsophalangeal joint | Great-toe extension |
| S1 | Lateral heel | Ankle plantarflexion |
Diagnostic accuracy and probability
Compare the index test with the study's reference standard. The reference defines what the study counted as the target condition and can be imperfect. Reliability concerns the consistency of a finding. Diagnostic accuracy concerns how well the test separates people with and without the target condition in the studied setting. Both matter when applying the result to a patient.
| Index test result | Target condition present | Target condition absent |
|---|---|---|
| Positive | True positive (TP) | False positive (FP) |
| Negative | False negative (FN) | True negative (TN) |
- Sensitivity = TP ÷ (TP + FN), the fraction of people with the condition who test positive.
- Specificity = TN ÷ (TN + FP), the fraction without the condition who test negative.
- Positive predictive value = TP ÷ (TP + FP). Negative predictive value = TN ÷ (TN + FN). These fractions describe the studied sample and change with the proportion of cases when other test properties are held constant.
- Positive likelihood ratio (LR+) = sensitivity ÷ (1 − specificity).
- Negative likelihood ratio (LR−) = (1 − sensitivity) ÷ specificity.
For example, suppose a study has 80 TP, 20 FN, 8 FP and 92 TN. Sensitivity is 80/100 = 0.80 and specificity is 92/100 = 0.92. LR+ is 0.80/0.08 = 10, and LR− is 0.20/0.92 ≈ 0.22. The sample's positive predictive value is 80/88 ≈ 91%. A new patient may have a different starting probability.
Update a probability with Bayes' rule
- Divide the pretest probability by one minus that probability to get pretest odds.
- Multiply the odds by the likelihood ratio for the observed result.
- Divide the resulting odds by one plus those odds to get the post-test probability.
With LR+ 10 and a 1% pretest probability, the post-test probability is about 9%. With the same LR+ and a 50% pretest probability, it is about 91%. Fagan's nomogram gives the same update graphically. An LR of 1 leaves odds unchanged for that result. The size of the probability shift and the clinical action depend on the starting estimate, uncertainty and consequences of a missed diagnosis.
SnNOut and SpPIn are memory aids for how a sensitive test's negative result or a specific test's positive result can be helpful. They are not guarantees. Use both test properties and consider the starting probability. A decision rule can also serve a specific purpose. For instance, the Ottawa Ankle Rules guide radiography after eligible acute injury. A positive rule supports imaging rather than confirming fracture.
Interpreting test clusters
Use the studied combination's likelihood ratio. Individual maneuvers in a cluster may be correlated. Multiplying their separate likelihood ratios assumes information about independence that is usually unavailable. Check the study reference, threshold, population and confidence intervals.
| Target and study | Findings and interpretation |
|---|---|
| Cervical radiculopathy, Wainner 2003 | Four items: Spurling A, median-biased upper limb tension test A (ULTT1), cervical distraction and active rotation below 60° toward the symptomatic side. The study compared examination with electrodiagnostic criteria in patients referred with suspected cervical radiculopathy or carpal tunnel syndrome. Three positive findings gave LR+ 6.1. Four gave LR+ 30.3 (95% CI 1.7–538.2). A later validation in a different population also found higher specificity with more positive findings and different estimates. |
| Cervical myelopathy, Cook 2010 | Five findings: gait deviation, Hoffmann sign, inverted supinator sign, Babinski response and age over 45 years. Compared with clinical diagnosis and imaging in a spine-center sample, at least three findings gave LR+ 30.9 (95% CI 5.5–181.8). At a positive threshold of at least one finding, zero findings gave LR− 0.18. A concerning course or signs of cord dysfunction warrant referral regardless of the cluster result. |
| Sacroiliac joint pain, Laslett 2005 | The four-test version uses distraction at the anterior superior iliac spines, thigh thrust with the hip flexed, side-lying pelvic compression and prone sacral thrust. Count reproduction of familiar pain. At least two of these four positive gave sensitivity 0.88, specificity 0.78 and LR+ 4.0 in the study using image-guided intra-articular anesthetic injection as reference. A positive cluster increases suspicion. It cannot by itself distinguish intra-articular from adjacent pain sources. |
In Wainner’s protocol, Spurling A combined side-bending toward the symptomatic side and controlled compression. Supine distraction was positive when presenting symptoms eased. Supine ULTT A progressed from scapular depression and shoulder abduction to forearm supination with wrist and finger extension, shoulder external rotation and elbow extension, followed by cervical side-bending. The study counted symptom reproduction, an elbow-extension asymmetry greater than 10° at the end of the sequence, or more symptoms with opposite-side neck bending or fewer with same-side bending. Use the named protocol when interpreting its study result.
Clinical case
After triage for trauma and urgent neurological or vascular signs, a 42-year-old with three weeks of neck and right arm symptoms undergoes a focused examination. Right rotation is 45°. Spurling A and the median-biased upper limb tension test reproduce familiar arm symptoms, and distraction eases them. Right wrist extension is mildly weak and sensation at the thumb differs from the left. The findings support a working hypothesis of cervical root involvement, with C6 among the levels to consider.
For an illustrative pretest probability of 25%, the odds are 0.25 ÷ 0.75 = 1/3. Multiplying by Wainner’s four-positive LR+ of 30.3 gives odds of 10.1, or a post-test probability of 10.1 ÷ 11.1 ≈ 91%. The wide interval and different patient populations limit precision. Use the history, neurological examination and course to choose the next step. If progressive gait and hand dysfunction or long-tract signs emerge, reassess for possible cord dysfunction and refer promptly.
A screened patient has pain and reduced force during resisted shoulder abduction. How should the clinician interpret that finding?
Use it alone to grade the size of a cuff tear
Record the painful weakness and compare other movement and strength findings before deciding which explanations fit
Use it to exclude pain-limited effort
Infer a peripheral nerve lesion from the single maneuver
A test has sensitivity 0.95 and specificity 0.40 in the applicable population. The pretest probability is 80%. Approximately what probability remains after a negative result?
5%
12.5%
50%
33%
Sensitivity is 0.80 and specificity is 0.92. At a pretest probability of 1%, what does a positive result imply?
LR+ is 10 and post-test probability is about 9%
LR+ is 10 and post-test probability is about 50%
LR+ is 0.22 and post-test probability is about 9%
LR+ is 10 and post-test probability is about 91%
Which four findings are used in the original Wainner cervical radiculopathy cluster?
Gait deviation, Hoffmann, Babinski and age over 45 years
Painful arc, Hawkins–Kennedy, empty can and resisted external rotation
Spurling A, median-biased ULTT A, cervical distraction and rotation below 60° toward the symptomatic side
Hoffmann, clonus, Lhermitte and Sharp–Purser
Sections you finish are checked off in the contents.