Exam Logic and Red Flags

Most people who walk into a clinic with back or neck pain have a problem that is painful but not dangerous, and a small number have something that will not wait: a tumor, an infection, a fracture, a compressed cauda equina or a torn artery. The skill that separates a safe clinician from a risky one is not a memorized list of warning signs but a way of thinking: asking the right questions, choosing tests that actually change the odds, and knowing when to stop treating and send the patient elsewhere. This chapter teaches that way of thinking, from the first question in the history to the wording of a referral letter.

Section 1History and the questions that matter

Why the history comes first

The clinical history is the single most powerful diagnostic tool in spinal care. In general medicine, studies going back decades have found that the history alone points to the correct diagnosis in the large majority of cases, with the physical examination and tests adding confirmation rather than discovery. For spinal pain this is even more true, because most structures that can hurt (discs, facet joints, ligaments, muscles) cannot be reliably told apart by palpation or by imaging. What the history does extremely well is sort the patient into a category and estimate the probability of serious disease before a single test is performed. That estimate, called the pretest probability, governs how every later finding should be interpreted.

A good history has two goals running in parallel. The first is diagnostic: what is likely causing this pain, and is anything dangerous going on? The second is relational: who is this person, what do they fear, what do they expect, and what matters to them? Clinicians who neglect the second goal often miss the information that predicts how the patient will actually do, which is the subject of Section 4.

Diagnostic triage: three bins

International guidelines for low back pain, and increasingly for neck pain, ask the clinician to begin with diagnostic triage, sorting each patient into one of three groups:

1. Serious spinal pathology, also called specific or nonmechanical pathology: malignancy, infection, fracture, cauda equina syndrome, inflammatory arthritis such as axial spondyloarthritis, and visceral or vascular causes referring pain to the back. In primary care this group is small, generally estimated at around 1 percent of acute low back pain presentations, with osteoporotic vertebral fracture the most common member and spinal infection the rarest.
2. Radicular syndromes: pain, and often sensory or motor change, arising from irritation or compression of a nerve root, most often from disc herniation or foraminal or central stenosis. Estimates vary with definitions, but roughly 5 to 10 percent of patients with low back pain fall here.
3. Nonspecific low back pain (or nonspecific neck pain): pain for which no specific pathoanatomic cause can be reliably identified. This is about 90 percent of presentations. The word nonspecific does not mean imaginary or trivial; it means that current tools cannot assign the pain to one tissue with confidence, and that knowing the tissue would not change management much.

The value of triage is that each bin leads to a different plan. Serious pathology needs investigation or referral. Radicular syndromes need a neurological examination, monitoring for progression, and sometimes imaging or surgical opinion. Nonspecific pain is managed with reassurance, advice to stay active, and conservative care such as exercise and manual therapy, while watching for psychosocial barriers to recovery.

The structure of the history

A complete musculoskeletal history has a predictable skeleton. Students who learn the skeleton can adapt it to any complaint:

• Chief complaint in the patient's own words, and the reason they came today rather than last month.
• History of present illness, organized by a mnemonic such as OPQRST (below).
• Past medical history: prior episodes and how they resolved, prior spinal surgery, cancer, osteoporosis, diabetes, inflammatory disease, cardiovascular disease.
• Medications: especially long-term corticosteroids (fracture and infection risk), anticoagulants (bleeding risk with procedures, spinal epidural hematoma), immunosuppressants, and current pain medicines and their effect.
• Family history: inflammatory arthritis, early cardiovascular disease, cancers.
• Social history: occupation and physical demands, smoking, alcohol and drug use (including injection drug use, an infection risk), exercise, sleep, home situation.
• Review of systems: a quick survey of body systems for symptoms the patient may not connect to the back, such as fever, weight loss, urinary changes, abdominal pain or cough.
• Patient perspective: ideas, concerns and expectations (sometimes abbreviated ICE).

OPQRST and what each answer tells you

The history of present illness is easier to remember with a mnemonic. OPQRST is one of the most widely used, and each letter carries real diagnostic weight.

Many clinicians add associated symptoms (numbness, weakness, bladder or bowel change, fever, dizziness) and previous episodes. A useful habit is to finish the OPQRST sequence with an open question: "Is there anything else about this that worries you, or that I have not asked?"

Mechanical versus nonmechanical pain

The most useful single distinction in the spinal history is whether pain behaves mechanically. Mechanical pain varies with posture, movement and load: it is provoked by certain positions and eased by others, it is often worse with prolonged sitting or bending and better with a change of position, and it may be stiff in the morning for a short time before loosening. Nonmechanical pain does not follow these rules. It may be constant, unrelated to movement, progressive over weeks, present at rest and at night, or accompanied by systemic features such as fever, sweats, malaise or weight loss. Nonmechanical pain is not a diagnosis but a prompt to look harder for serious pathology or for pain referred from an internal organ.

Night pain is a classic example of a finding that needs careful interpretation. Many people with ordinary mechanical back pain report waking at night when they roll over; that pain eases when they change position and is mechanical. The worrying pattern is pain that is constant through the night, keeps the patient up regardless of position, and is getting worse over time. Asking "what happens when you change position?" separates the two.

Radicular pain versus somatic referred pain

Students commonly assume that any leg pain from the back is "sciatica" from a pinched nerve. In fact two quite different mechanisms produce pain in the limb. Somatic referred pain arises from a spinal structure such as a disc, facet joint or sacroiliac joint and is felt in a distant region that shares segmental innervation. It is deep, dull, aching and hard to localize, usually stays above the knee, and is not accompanied by neurological deficit. Radicular pain arises from the dorsal root or dorsal root ganglion itself, usually when an inflamed or compressed root fires ectopically. It is sharp, shooting or lancinating, travels in a narrow band, often extends below the knee (or below the elbow in the arm), and is commonly worse in the limb than in the back. When conduction along the root is actually blocked, the patient also has radiculopathy: objective loss of sensation, strength or reflexes in that root's territory. Radicular pain and radiculopathy often occur together, but either can occur alone.

Claudication: neurogenic or vascular?

Leg pain brought on by walking has two classic causes that must not be confused. Neurogenic claudication comes from lumbar spinal stenosis: narrowing of the central canal or lateral recesses squeezes the cauda equina roots, especially in extension, when the canal is smallest. Vascular claudication comes from peripheral artery disease: narrowed arteries cannot supply enough blood to working calf muscles. The history usually separates them before any test is done.

Asking well

The quality of answers depends on the quality of questions. Begin open ("Tell me about your back") and narrow later ("Does the pain go below your knee?"). Avoid leading questions that suggest the answer, and avoid medical jargon: patients often do not know what "radiating" or "incontinence" means, so ask directly and plainly ("Have you had any trouble passing urine, or any leaking you could not control?"). Sensitive questions, such as those about bladder, bowel and sexual function, are best introduced with a brief explanation: "I ask everyone with back and leg pain about this because, rarely, nerves to the bladder can be affected." Normalizing the question makes truthful answers more likely and avoids alarming the patient.

It also helps to ask what the patient thinks is going on. A person who believes their spine is "crumbling" or that a disc has "slipped out" will move and recover differently from one who believes they have strained something that will heal. The patient's own model of the problem is part of the diagnosis.

Section 2Physical examination logic

The examination tests hypotheses

A physical examination is not a ritual in which every test is performed on every patient. It is a set of experiments designed to test the hypotheses raised by the history. If the history suggests a nonspecific mechanical problem, the examination looks for movement patterns, provocation and a normal neurological screen. If the history suggests a nerve root, the examination maps strength, reflexes and sensation and tests neural tension. If the history raises serious pathology, the examination looks for fever, weight loss, a mass, focal bone tenderness or an abnormal neurological picture, and the next step may be a referral rather than further testing.

A typical spinal examination proceeds in a consistent order so nothing important is skipped:

1. General observation and vital signs: how the patient sits, stands and moves, facial expression, gait, temperature, pulse, blood pressure. A fever or tachycardia changes everything that follows.
2. Inspection: posture, alignment, muscle wasting, skin changes, scars, swelling.
3. Palpation: tenderness, temperature, muscle tone, and focal bony tenderness over a spinous process.
4. Range of motion: active first, then passive, noting the amount, the quality, and whether movement reproduces the familiar pain.
5. Neurological examination: strength (myotomes), reflexes, sensation (dermatomes), and screening for upper motor neuron signs.
6. Special tests: neural tension tests, provocation tests, joint stability tests and others chosen for the hypothesis.
7. Related regions: the hip for low back pain, the shoulder for neck pain, the abdomen and peripheral pulses when vascular or visceral causes are possible.

The numbers behind a test: sensitivity and specificity

Every diagnostic test, from a blood test to a straight leg raise, can be described by how it performs against a reference standard (the best available way of knowing the truth, such as surgical findings or MRI). The results fall into a two by two table. Sensitivity is the proportion of people who have the condition and test positive: true positives divided by everyone with the condition. Specificity is the proportion of people without the condition who test negative: true negatives divided by everyone without it.

Two rules of thumb follow. A highly sensitive test rarely misses the condition, so a negative result helps rule it out: SnNout. A highly specific test rarely fires in people without the condition, so a positive result helps rule it in: SpPin. Students often reverse these. The trick is to remember that sensitivity is about the sick (how many of them it catches), so its strength lies in what a negative result means.

Likelihood ratios and pretest probability

Sensitivity and specificity describe a test, but the clinician wants to know about the patient: given this result, how likely is the condition now? Likelihood ratios answer that. The positive likelihood ratio (LR+) is sensitivity divided by one minus specificity; the negative likelihood ratio (LR−) is one minus sensitivity divided by specificity. An LR+ above 10 or an LR− below 0.1 produces large, often decisive shifts in probability. Values of 5 to 10 or 0.1 to 0.2 produce moderate shifts. Values between about 0.5 and 2 barely move the estimate at all, and many orthopedic tests sit in that unhelpful zone.

The result of any test must be combined with the pretest probability, the estimated chance the patient had the condition before the test. This is Bayesian reasoning, and it explains why the same test means different things in different patients. Suppose a test has an LR+ of 5. In a patient with a 1 percent pretest probability, a positive result raises the probability only to about 5 percent: the condition is still unlikely. In a patient with a 40 percent pretest probability, the same positive result raises it to roughly 77 percent. This is why screening low-risk people with imperfect tests produces mostly false positives, and why the history, which sets the pretest probability, is so important.

Reliability and test clusters

A test also has to give the same answer when repeated. Reliability is the agreement between two examiners (inter-rater) or one examiner on two occasions (intra-rater), often reported as the kappa statistic, where 0 is chance agreement and 1 is perfect. Many traditional spinal examination procedures have poor reliability. Static palpation for vertebral position and motion palpation for segmental stiffness generally show low to fair agreement between examiners, while palpation for pain provocation (does pressing here reproduce the pain?) tends to be more reliable. This does not mean palpation is worthless: it guides where a clinician applies treatment and can reproduce symptoms. It means palpation findings should not be treated as precise diagnostic measurements.

Because individual tests are often weak, researchers combine them into clusters. For example, a cluster of sacroiliac joint provocation tests (distraction, thigh thrust, compression, sacral thrust and others) has better diagnostic accuracy when three or more are positive than any single test alone. Clusters also exist for cervical radiculopathy, combining Spurling's test, the upper limb tension test, distraction relief and reduced cervical rotation. Thinking in clusters reflects a general truth: confidence comes from several pieces of consistent evidence, not from one dramatic sign.

Common spinal special tests

The neurological screen

Whenever a patient has limb symptoms, or the history raises any neurological concern, the clinician performs a neurological screen. The logic is simple: each nerve root supplies a characteristic set of muscles (a myotome), a band of skin (a dermatome), and a deep tendon reflex. A lesion of one root produces weakness, sensory change and reflex change that line up with that root. A pattern that does not line up with any single root suggests a peripheral nerve lesion, several roots, or a central problem.

Strength is graded on the 0 to 5 Medical Research Council scale: 0 is no contraction, 1 a flicker, 2 movement with gravity eliminated, 3 movement against gravity, 4 movement against some resistance, and 5 normal strength. Reflexes are graded 0 (absent), 1+ (diminished), 2+ (normal), 3+ (brisk, may spread), and 4+ (very brisk, often with sustained clonus).

Remember that a lumbar disc herniation usually affects the root that exits one level below the disc. The common posterolateral L4 to L5 herniation compresses the traversing L5 root; an L5 to S1 herniation compresses S1. A far lateral herniation, by contrast, catches the root exiting at that same level. The L4 to L5 and L5 to S1 levels together account for the great majority of lumbar disc herniations.

Upper motor neuron signs

Nerve root and peripheral nerve lesions are lower motor neuron problems: they cause weakness, reduced or absent reflexes, reduced tone and, over time, muscle wasting. Lesions of the spinal cord or brain are upper motor neuron problems: they cause weakness with increased tone (spasticity), brisk reflexes, clonus, and pathological reflexes. In spinal practice the most important upper motor neuron condition is degenerative cervical myelopathy, compression of the cervical cord by disc, osteophyte and thickened ligament. It is the most common cause of spinal cord dysfunction in older adults and is frequently missed early. Patients describe clumsy hands, difficulty with buttons or handwriting, numb hands that do not follow a single dermatome, and a vague unsteadiness of gait.

Section 3Red flags: cancer, infection, fracture, cauda equina, vascular

What a red flag is, and what it is not

Red flags are features of the history or examination associated with a higher probability of serious pathology. They are prompts to think, not diagnoses. Research has shown two uncomfortable facts about them. First, red flags are very common: in primary care, most patients with low back pain have at least one, and nearly all of them do not have serious disease. Second, most individual red flags have poor diagnostic accuracy, with likelihood ratios close to 1. Treating every red flag as an alarm would send most back pain patients for imaging and blood tests, generating false positives, cost and anxiety. Ignoring them would occasionally miss a cancer or an abscess.

The solution is to weigh red flags in context. A few flags carry real weight on their own (a history of cancer is the strongest example). Most gain meaning only in combination, or when the pain is nonmechanical, or when the patient fails to improve as expected over several weeks. Good practice also includes safety netting: telling the patient which new symptoms should make them return or seek emergency care.

Malignancy

Spinal malignancy in a patient with back pain is usually metastatic, because the vertebral bodies are rich in red marrow and the vertebral venous plexus (Batson's plexus) is valveless, allowing tumor cells from the pelvis, abdomen and chest to seed bone. The cancers that most often spread to bone are prostate, breast, lung, kidney and thyroid. Multiple myeloma, a primary marrow cancer, also commonly presents with back pain and vertebral fractures in older adults.

• Previous history of cancer is the most informative single flag, with positive likelihood ratios reported in the range of roughly 7 to 15. A patient with new back pain and a history of breast, prostate or lung cancer needs a low threshold for imaging, even years after apparent cure.
• Age over 50, unexplained weight loss, failure to improve after about a month of appropriate care, and constant or progressive pain at night add modestly to suspicion, particularly in combination.
• Thoracic pain deserves extra respect, since the thoracic spine is a common site of metastasis and is less often the site of ordinary mechanical pain.

The main danger of a missed spinal metastasis is metastatic spinal cord compression, which can cause permanent paralysis. Back pain usually precedes neurological signs by weeks; new neurological symptoms in a patient with known cancer are an emergency.

Infection

Spinal infection includes vertebral osteomyelitis, discitis (infection of the disc space, often together as spondylodiscitis) and spinal epidural abscess. Bacteria reach the spine mainly through the blood, and Staphylococcus aureus is the most common organism. Tuberculosis of the spine (Pott disease) remains important worldwide.

Risk factors are the key to suspicion: injection drug use, immunosuppression (including long-term corticosteroids, diabetes, HIV and chemotherapy), a recent bacterial infection such as a urinary tract or skin infection, recent spinal surgery or injection, an indwelling catheter or hemodialysis. Fever is a helpful sign when present but is absent in a substantial minority, so its absence does not exclude infection. The classic triad of epidural abscess, namely back pain, fever and neurological deficit, is present in only a small proportion at first presentation; by the time all three are present, harm may already be done. Severe, unrelenting local pain with focal spinous process tenderness in a patient with risk factors should prompt urgent blood tests (an elevated erythrocyte sedimentation rate and C-reactive protein are typical) and MRI.

Fracture

Vertebral fracture is the most common serious pathology found in primary care back pain. Most are osteoporotic compression fractures of the thoracolumbar spine, which may occur with trivial loading such as bending, coughing or lifting a light object, and some occur with no remembered event at all. The flags that perform best are older age (especially over 70), prolonged corticosteroid use, significant trauma for the patient's age, and a visible contusion or abrasion. In combination, these raise the probability considerably. Women after menopause and anyone with known osteoporosis, a previous fragility fracture or a low body weight carry higher risk.

In the neck, high-energy trauma raises the question of cervical fracture or instability. Two well validated decision rules help decide whether imaging is needed in alert, stable trauma patients. The Canadian C-spine Rule requires imaging for any high-risk factor (age 65 or older, a dangerous mechanism, or tingling in the limbs), then uses low-risk features to decide whether range of motion can be safely assessed, and finally checks whether the patient can actively rotate the neck 45 degrees to each side. The NEXUS criteria allow clinicians to skip imaging only if all five are absent: posterior midline tenderness, intoxication, altered alertness, a focal neurological deficit, and a painful distracting injury. These rules are highly sensitive, meaning a negative rule reliably excludes important injury.

Cauda equina syndrome

The spinal cord ends at the conus medullaris, usually at about the L1 to L2 vertebral level in adults. Below that, the lumbar and sacral roots run down the canal as the cauda equina ("horse's tail"). Massive central compression of these roots, most often by a large central disc herniation at L4 to L5 or L5 to S1, produces cauda equina syndrome. Less common causes include tumor, epidural abscess, hematoma, fracture and severe stenosis. Because the sacral roots S2 to S4 supply the bladder, bowel, sexual organs and the skin of the perineum, the hallmark features are:

• Saddle anesthesia: numbness or altered sensation of the perineum, buttocks and inner thighs (the area that would touch a saddle). Patients may notice it when wiping after using the toilet.
• Bladder dysfunction: difficulty starting urination, a weak stream, loss of the sensation of a full bladder, or painless urinary retention with overflow incontinence. Urinary retention is the most consistent finding in established cases.
• Bowel dysfunction: loss of control or of sensation, reduced anal tone.
• Sexual dysfunction: new erectile dysfunction or loss of genital sensation.
• Bilateral leg symptoms: bilateral sciatica, or progressive weakness in both legs.

Clinicians distinguish an early, incomplete stage (CES-I, with altered urinary sensation or function but preserved voluntary control) from a later stage with retention (CES-R), when the bladder has become insensate and overflows. Outcomes are much better when decompression occurs before retention develops, so any suspicion warrants same day emergency referral for MRI and surgical assessment. Measuring the volume of urine left in the bladder after voiding (post-void residual) is often used in hospital: a large residual supports retention. No clinician should wait for every feature to appear.

Vascular causes

Two vascular emergencies are particularly relevant to spinal practice.

Abdominal aortic aneurysm (AAA) is a dilation of the abdominal aorta, most common in men over 65 who smoke or have smoked, with family history and hypertension adding risk. An expanding or leaking aneurysm can cause deep, constant back, flank or abdominal pain that does not change with movement. A pulsatile abdominal mass, low blood pressure, fainting or pallor suggests rupture, which is immediately life threatening. Back pain that is nonmechanical in an older smoker should prompt an abdominal examination.

Cervical artery dissection is a tear in the inner wall of the vertebral or internal carotid artery that allows blood to track into the wall, narrowing the vessel or forming a clot that can travel to the brain and cause a stroke. It is rare (a few cases per 100,000 people per year) and often occurs in younger adults, sometimes after minor neck trauma or for no clear reason. Early symptoms are usually unusual neck pain or headache, often described as unlike anything experienced before, which may precede stroke symptoms by hours to days. Signs of brainstem ischemia from vertebral artery involvement are often taught as the "5 Ds and 3 Ns": dizziness, diplopia (double vision), dysarthria (slurred speech), dysphagia (trouble swallowing) and drop attacks; plus nystagmus, nausea and numbness, with ataxia of gait. Internal carotid dissection may cause a partial Horner syndrome (drooping eyelid and small pupil), pulsatile tinnitus or cranial nerve signs.

The relationship between cervical manipulation and artery dissection has been debated for decades. Case reports describe dissection and stroke after neck manipulation. Population studies have found that people who later have a vertebrobasilar stroke are more likely to have seen a chiropractor in the preceding days, but similarly more likely to have seen a primary care physician, which suggests that many patients with an early dissection seek care for the neck pain and headache it causes. Current evidence does not establish that manipulation causes dissection, nor can it exclude a small risk. Professional guidance therefore emphasizes careful screening of every patient with neck pain or headache for features of vascular pathology, recognizing that the best protection is identifying a dissection in progress, and immediate emergency referral when it is suspected. Pre-manipulative positional tests of the vertebral artery have poor validity and are no longer recommended as a safety screen on their own.

Section 4Yellow flags and risk of chronicity

Why psychosocial factors predict outcome

Most acute low back pain improves substantially within about six weeks, but recovery is often incomplete and recurrences are common, and a minority of patients develop persistent pain and disability lasting beyond three months. Strikingly, the best predictors of who will develop long-term disability are usually not findings on imaging or examination but psychological and social factors. These are called yellow flags. Unlike red flags, they do not signal danger to the body; they signal risk to recovery. Chapter 6 explained the neuroscience: fear, threat and attention shape how the nervous system processes nociception, and avoidance behavior can turn a healing injury into a long-term problem.

The flag system

Clinicians have extended the color metaphor to describe different kinds of barriers to recovery.

The key yellow flags

• Fear avoidance beliefs: the belief that pain means harm and that activity should be avoided until pain is gone. Avoidance leads to deconditioning, loss of confidence and more pain on return to activity.
• Pain catastrophizing: an exaggerated negative focus on pain, with rumination ("I can't stop thinking about it"), magnification ("something terrible will happen") and helplessness ("there is nothing I can do").
• Kinesiophobia: an excessive fear of movement and reinjury.
• Low expectations of recovery: the patient's own prediction that they will not get better is one of the strongest single predictors of poor outcome.
• Depression, anxiety and distress, including poor sleep.
• Passive coping: reliance on rest, medication and treatments done to the patient rather than active self-management.
• Social and work factors: low job satisfaction, heavy physical work without control over pacing, time off work, and financial strain.

The clinician's own behavior can add or remove yellow flags. Telling a patient their spine is "out," "degenerating" or "unstable," or showing them an MRI full of age-normal findings without context, can create fear that outlasts the original injury. Imaging findings such as disc bulges and degeneration are extremely common in people without any pain and become more common with every decade of age. Language is a clinical tool, and Chapter 6 discussed how to use it well.

Screening tools

Several validated questionnaires help identify patients at risk of chronicity:

• The STarT Back Tool is a brief nine-item questionnaire that classifies patients with low back pain as low, medium or high risk of persistent disabling pain. It combines physical items (leg pain, comorbid pain, function) with a psychosocial subscale (fear, anxiety, catastrophizing, low mood, bothersomeness). In stratified care, low-risk patients receive advice and reassurance with minimal treatment, medium-risk patients receive physical therapy style care, and high-risk patients receive care that combines physical treatment with psychologically informed approaches. A large randomized trial found stratified care improved disability outcomes and was cost effective compared with usual care.
• The Örebro Musculoskeletal Pain Screening Questionnaire covers similar ground, with a strong emphasis on work.
• The Fear Avoidance Beliefs Questionnaire and the Tampa Scale of Kinesiophobia measure fear of activity and movement.
• The Pain Catastrophizing Scale measures rumination, magnification and helplessness.
• The PHQ-9 screens for depression, and its item on thoughts of self-harm must be acted upon if positive.

Section 5When and how to refer

Referral is part of good care

Every clinician works within a scope of practice and a range of competence. Referring a patient is not a failure but a core professional skill, and in spinal care it protects patients in three situations: when serious pathology is suspected, when the patient is not improving as expected, and when another professional can offer something the patient needs, such as surgery, medication, psychological care or specialist rehabilitation. Many patients benefit from co-management, in which the original clinician continues appropriate care while another provider manages a separate aspect of the problem.

Matching urgency to the problem

The most important decision in a referral is how fast it must happen.

An emergency referral means the patient goes directly to an emergency department or calls emergency services; the clinician should call ahead where possible and never send a patient with suspected stroke or cauda equina syndrome to "book an appointment." An urgent referral means contacting the patient's physician or appropriate specialist the same day and making sure the patient is seen within a day or a few days. A routine referral can go through normal channels. When in doubt between two tiers, choose the faster one.

Imaging and laboratory tests

Major guidelines agree that routine imaging is not recommended for acute nonspecific low back pain without red flags. Imaging in this group does not improve outcomes, exposes patients to radiation (for radiographs and CT), finds incidental abnormalities that may lead to further tests or surgery, and can increase fear and disability. Imaging is appropriate when red flags suggest serious pathology, when there are severe or progressive neurological deficits, or when a patient with persistent radicular symptoms is a candidate for an injection or surgery. MRI is the preferred study for suspected infection, malignancy, cauda equina syndrome and nerve root compression; plain radiographs are useful for suspected fracture, deformity and some forms of arthritis.

Common laboratory tests in spinal workups include a complete blood count, erythrocyte sedimentation rate and C-reactive protein (markers of inflammation that are usually raised in infection and often in malignancy), serum protein electrophoresis for myeloma, prostate specific antigen, and HLA-B27 when spondyloarthritis is suspected. These tests are ordered by clinicians whose scope includes them, which is another reason for clear referral pathways.

How to write a referral

A good referral is brief, specific and easy to act on. A widely used structure in health care is SBAR:

The referral should include the clinician's contact details and a clear statement of whether care will continue, pause or stop. Records must document the findings, the reasoning, the advice given and the safety netting instructions. If a patient declines an emergency referral, the conversation, the explanation of risk and the patient's decision should all be recorded, and the clinician should still contact appropriate services where duty of care requires it.

Talking to the patient about referral

Patients can find referral frightening. Explain why it is being made, what the other clinician will do, and how likely serious disease actually is, without either minimizing a real risk or catastrophizing a small one. For example: "Most back pain like yours is not serious, but because of your history I want a physician to check for anything we should not miss. If anything changes before then, here is what to do." Clear safety netting advice is an essential part of every plan, whether or not a referral is made.

Words worth knowing