Spinal fusion is a form of surgery intended to create a permanent bony union between two or more vertebrae. It eliminates movement across the treated spinal segment by placing bone graft or a bone-forming substitute between prepared skeletal surfaces. Screws, rods, plates, or interbody implants often provide immediate mechanical support while the biological fusion develops. These implants stabilize the spine but do not themselves constitute a fusion: the defining event is the formation of a continuous bridge of living bone.
The operation is used in selected cases of spinal instability, deformity, degenerative disease, trauma, infection, and tumor-related destruction. It may accompany decompression of the spinal cord or nerve roots, but decompression and fusion have different purposes: decompression creates space for neural structures, whereas fusion restricts movement and supports alignment. Procedures range from a single cervical disc level to extensive reconstructions extending from the thoracic spine to the pelvis.
Spinal fusion is a major component of orthopedic surgery and neurosurgery. Its results depend on the underlying disorder, the anatomical region, the number of levels treated, bone quality, surgical technique, and the outcome being measured. Successful bone union, relief of pain, neurological recovery, and improved function are related but distinct endpoints; achieving one does not necessarily ensure the others.
Anatomy and mechanical principles
The mobile spine consists of cervical, thoracic, and lumbar vertebrae, with the sacrum providing a connection to the pelvis. A typical motion segment includes two adjacent vertebrae, the intervening intervertebral disc, paired facet joints, and supporting ligaments. The disc distributes compression and permits movement, while the facets guide movement and resist rotation and shear. Stability therefore depends on several structures acting together rather than on the vertebral bodies alone.
Fusion changes this arrangement by converting a mobile segment into a relatively rigid unit. In an interbody fusion, bone is placed in the space formerly occupied by the disc, often within or around a cage. In a posterolateral fusion, graft is laid across prepared posterior or lateral bony surfaces, such as the transverse processes. Cervical fusion may use the disc space, posterior joints and lateral masses, or a combination of locations. Fusion involving the upper cervical spine requires different fixation strategies because its anatomy differs substantially from that of the lower cervical vertebrae.
Mechanical stability is particularly important during the early healing period. Excessive movement at the graft interface can prevent bone bridging and favor fibrous tissue formation. Spinal instrumentation reduces this movement and maintains the desired relationship between vertebrae. Pedicle screws obtain purchase through the pedicles into the vertebral bodies; cervical lateral-mass screws engage a different region of bone. Rods or plates connect the anchors. An interbody cage supports the disc space and transmits load between vertebral endplates. Its material and geometry influence stiffness, contact area, and visibility on imaging.
Alignment matters as well as local stability. The normal spine has cervical and lumbar lordosis and thoracic kyphosis. In adult reconstruction, lumbar curvature is considered in relation to pelvic anatomy and the position of the trunk over the pelvis. A technically solid fusion in an unfavorable position can leave substantial disability. For example, insufficient lumbar lordosis after a long fusion can produce a forward-stooped posture that requires compensatory bending at the hips and knees.
Eliminating motion at one level also changes loading elsewhere. Adjacent discs and joints may experience altered movement or stress, although subsequent degeneration also reflects aging and pre-existing disease. Longer constructs impose larger forces on their terminal anchors and on the transition between rigid and mobile regions. These principles explain why fixation strength, fusion biology, overall alignment, and the choice of upper and lower fusion levels are interdependent parts of surgical planning.
Conditions treated and limits of indication
Fusion is most clearly linked to a structural objective: stabilizing a mechanically compromised spine, maintaining a correction, or supporting reconstruction after tissue removal. It is not a uniform treatment for every cause of back pain. Similar symptoms can arise from discs, joints, muscles, nerve compression, fractures, or disorders outside the spine. Imaging abnormalities are also common in people without symptoms, so the presence of degeneration alone does not establish that fusion will improve a clinical problem.
Spondylolisthesis, the displacement of one vertebra relative to another, is an important indication in selected circumstances. Isthmic spondylolisthesis is associated with a defect in the pars interarticularis, whereas degenerative spondylolisthesis develops through disc and joint degeneration without that defect. Fusion may address instability, deformity, foraminal narrowing, or destabilization associated with decompression. However, many patients with degenerative spondylolisthesis and spinal stenosis can undergo decompression without fusion. A visible slip is therefore not, by itself, a universal indication.
In the cervical spine, anterior decompression and fusion is used for selected cases of nerve-root compression or spinal cord dysfunction caused by disc and bone abnormalities. Removal of a herniated disc or osteophytes creates space for neural structures; fusion then reconstructs and stabilizes the treated level. Multilevel cord compression can also be treated through posterior approaches, with or without fusion depending on alignment and stability. In contrast, a routine first-time lumbar disc herniation treated by discectomy generally does not require fusion.
Deformity surgery commonly incorporates fusion to preserve a correction. In adolescent idiopathic scoliosis, instrumentation corrects selected curves while fusion prevents progressive movement through the treated segments. Adult deformity may combine scoliosis, loss of lumbar lordosis, stenosis, and vertebral displacement. These operations can involve many levels and sometimes the pelvis. Fusion in a growing child also restricts growth in the treated region, which distinguishes it from growth-preserving strategies used for some early-onset deformities.
Trauma, infection, and tumors create other reconstructive indications. Unstable fractures or ligament disruption may require stabilization, although not every instrumented fracture treatment includes a formal fusion. Vertebral destruction from infection or cancer can require removal of diseased tissue and reconstruction of load-bearing support. Historical fusion procedures frequently treated deformity caused by spinal tuberculosis. For persistent nonspecific low back pain without a clear instability or neurological indication, the benefit of fusion is less predictable, and comparative evidence differs from that for structurally unstable disorders.
Assessment and operative planning
Assessment connects symptoms and functional impairment to a specific anatomical and mechanical problem. The medical history distinguishes predominantly axial pain from radiating limb pain, walking limitation, loss of hand dexterity, weakness, and symptoms of spinal cord dysfunction. A physical examination assesses strength, sensation, reflexes, gait, balance, and alignment. The duration and progression of neurological impairment can affect the potential for recovery even when decompression is technically successful.
Different imaging methods answer different questions. Standing radiographs show alignment under load and can document scoliosis, vertebral displacement, or loss of disc height. Flexion–extension images may demonstrate movement between vertebrae, although measurement techniques and definitions of instability vary. Full-length standing images are especially relevant to long deformity constructs because a regional image may not reveal compensatory pelvic or lower-limb positioning.
Magnetic resonance imaging depicts discs, neural structures, ligaments, and many infectious or neoplastic processes. Computed tomography provides detailed information about cortical bone, fractures, existing fusion masses, and hardware position. It is particularly useful in evaluating suspected nonunion after an earlier operation. Neither modality alone establishes the source of pain: anatomical findings require interpretation alongside symptoms and examination.
Bone quality and healing capacity influence both fixation and biological union. Osteoporosis can reduce screw purchase and increase the risk of vertebral fracture or cage subsidence. Smoking is associated with impaired fusion, particularly in lumbar procedures. Diabetes mellitus, poor nutritional status, chronic exposure to some medications, and prior operations can also affect perioperative risk or healing. The relevance of these factors differs across patients and procedures; they are not interchangeable predictors of failure.
Planning also defines what the operation is intended to achieve. Possible goals include neural decompression, correction of a curve, restoration of disc height, stabilization of an unstable segment, or prevention of further deformity. A treatment plan specifies the approach, levels, graft strategy, fixation method, and whether surgery will occur in one or more stages. Revision operations require additional attention to scar tissue, existing implants, the reason for previous failure, and the integrity of adjacent levels.
Shared decision-making in this setting concerns the expected benefits, alternatives, uncertainty, and burdens of a particular reconstruction. Pain relief, prevention of neurological deterioration, improvement in walking, and radiographic correction are not equivalent goals. Likewise, the risks of a one-level cervical procedure cannot be represented accurately by complication figures from a long adult deformity reconstruction.
Surgical approaches and reconstruction techniques
Fusion techniques are named according to anatomical region, route of access, and location of the intended bone union. An anterior approach reaches the spine from the front, a posterior approach from the back, and a lateral approach from the side. These routes expose different structures and permit different forms of decompression and reconstruction. A combined operation uses more than one route, either during the same anesthetic or in staged procedures.
Anterior cervical discectomy and fusion removes a diseased disc through the front of the neck. After decompression, an interbody graft or cage maintains the disc space while fusion develops. An anterior plate may be added, although some implants incorporate fixation within the interbody device. A cervical corpectomy removes part or all of a vertebral body to address pathology extending behind it; reconstruction then spans the resulting defect. Posterior cervical fusion uses fixation such as lateral-mass or pedicle screws and may accompany removal of the laminae.
Lumbar posterolateral fusion places graft over prepared posterior-lateral bone surfaces, commonly with pedicle-screw fixation. Posterior lumbar interbody fusion accesses the disc space from behind, usually requiring exposure on both sides of the neural elements. Transforaminal lumbar interbody fusion approaches the disc space more laterally through removal of part or all of a facet joint. Both aim to combine interbody support with posterior stabilization, but they differ in exposure and handling of the nerve roots and dural sac.
Anterior lumbar interbody fusion approaches the disc through the abdomen or retroperitoneal space. It permits placement of a relatively large implant and can restore disc height and segmental lordosis, but requires exposure near major blood vessels. Lateral approaches reach selected lumbar levels through or anterior to the psoas muscle. The iliac crest, vascular anatomy, and position of the lumbar plexus constrain access. These procedures may provide indirect decompression by restoring disc height and tensioning surrounding tissues rather than directly removing every compressive structure.
Minimally invasive surgery uses smaller access corridors, tubular retractors, or percutaneous screws to limit tissue disruption. It does not necessarily reduce the extent of the intended fusion, and it introduces its own technical and imaging demands. Fluoroscopy, three-dimensional navigation, and robot-assisted surgery can support implant placement. Their performance is assessed through measures such as placement accuracy, radiation exposure, operative time, and complications; improved technical accuracy does not automatically establish better long-term function.
Extensive deformity correction may require osteotomies, in which bone is cut or removed to permit realignment. More powerful corrections generally involve greater physiological and mechanical demands. Approach selection therefore reflects the pathology and reconstruction required rather than a universal hierarchy in which one technique is best for every spinal disorder.
Bone grafts and the biology of union
Fusion depends on the same broad biological processes that support fracture healing: cellular recruitment, vascular ingrowth, new bone formation, and remodeling. Prepared bone surfaces provide a vascular interface. Graft material occupies the intended fusion bed, and mechanical stabilization protects the developing tissue. Early repair tissue is gradually replaced by mineralized bone, which continues to reorganize under load. Immediate stability from implants and mature biological union occur on different timescales.
Bone grafting is commonly described through three properties. Osteogenesis refers to bone formation by living cells carried within the graft. Osteoinduction is the recruitment and differentiation of host cells toward bone-forming lineages. Osteoconduction is the provision of a scaffold along which new bone can grow. A graft material may contribute one or more of these properties, but biological activity also depends on processing, implantation conditions, and the surrounding tissue bed.
Autograft is obtained from the same individual. Bone removed during decompression can be reused locally, and the iliac crest is another established source. Autograft can provide living cells, matrix, and biological signals without the immune considerations associated with tissue from another person. Its limitations include available volume and, when a separate harvest is required, donor-site pain and other local complications. Cancellous graft offers a porous architecture favorable for incorporation, whereas cortical graft can provide greater initial structural support but generally incorporates more slowly.
Allograft comes from a human donor and is processed through tissue-banking systems. It avoids a separate harvest operation and can be supplied as structural pieces or particles. Processing reduces biological cellular activity and changes material properties. Demineralized bone matrix contains organic bone components after much of the mineral has been removed; its osteoinductive activity varies with product and processing. Synthetic materials, including calcium-phosphate ceramics, primarily provide an osteoconductive scaffold and are often used with other graft sources.
Bone morphogenetic proteins are signaling molecules involved in bone development and repair. Recombinant human bone morphogenetic protein-2 is used in selected spinal procedures with a carrier, but regulatory authorization is specific to products, approaches, and anatomical indications. Use outside an authorized indication is distinct from approved use. Reported adverse effects include inflammatory swelling, bone formation outside the intended site, and local bone resorption; in the cervical region, severe swelling can threaten the airway.
Fusion assessment is therefore not simply a question of which graft was chosen. The number of levels, vascularity, fixation, bone quality, nicotine exposure, prior surgery, and local anatomy all influence union. Studies comparing graft products are difficult to interpret when these factors differ or when investigators use different definitions of successful fusion.
Perioperative care and recovery
Spinal fusion is usually performed under general anesthesia. Positioning depends on the approach: posterior procedures commonly use the prone position, whereas anterior and lateral operations require different exposure arrangements. Positioning itself matters because prolonged pressure, impaired abdominal venous drainage, or excessive neck and limb positioning can cause injury or increase bleeding. Long reconstructions may require substantial physiological monitoring and coordinated anesthetic, surgical, and nursing care.
Intraoperative neurophysiological monitoring can include somatosensory evoked potentials, motor evoked potentials, and electromyography. These methods assess different aspects of neural function and may identify changes associated with traction, altered perfusion, or implant placement. Signals are affected by anesthetic drugs, temperature, blood pressure, and pre-existing neurological deficits. Monitoring supplements rather than replaces anatomical knowledge, direct inspection, and imaging.
Blood loss varies greatly between operations. A limited single-level procedure and a multilevel reconstruction with osteotomies have very different transfusion risks. Blood-conservation strategies may include meticulous hemostasis, cell salvage, and antifibrinolytic medication. Perioperative antibiotics reduce surgical-site infection risk, while measures against venous thrombosis balance clot prevention against bleeding concerns. An epidural hematoma is uncommon but potentially serious because blood accumulating within the spinal canal can compress neural structures.
Early postoperative care addresses neurological status, wound condition, breathing, swallowing when relevant, urinary function, mobility, and pain. Analgesia often combines several drug classes and non-drug measures rather than relying exclusively on an opioid. The surgical approach influences the pattern of recovery: anterior cervical procedures may cause swallowing difficulty, abdominal approaches may affect bowel function, and posterior operations can produce substantial muscle-related discomfort.
Rehabilitation concerns both recovery from surgery and adaptation to altered spinal mechanics. Mobilization often begins before biological fusion is complete because the construct provides early support. Physical therapy may address walking, endurance, strength, balance, and task performance, with timing and content differing by operation. The use of postoperative braces varies, and evidence does not establish a uniform benefit after every instrumented fusion.
Bone union develops over months rather than during the initial hospital stay. Clinical follow-up evaluates symptoms, function, alignment, and hardware, with imaging selected according to the question being investigated. Recovery of nerve function may be incomplete when compression has caused longstanding injury. Return to work and usual activities also depends on job demands, the extent of reconstruction, preoperative disability, complications, and social circumstances, not solely on whether an image shows a solid fusion.
Complications and long-term mechanical effects
Complications can arise from surgical exposure, neural injury, infection, impaired healing, or the altered mechanics of a fused spine. Their frequency depends strongly on anatomical region and procedural complexity. A single pooled complication percentage can be misleading when it combines short, uncomplicated fusions with extensive revision or deformity operations. Some complications occur immediately, whereas others become apparent only after months or years.
Neurological complications include injury to the spinal cord or nerve roots, postoperative weakness, sensory change, and persistent or new radiating pain. A tear in the dura can cause leakage of cerebrospinal fluid, sometimes producing positional headache or a postoperative fluid collection. Approach-specific injuries include vascular injury during anterior lumbar exposure, lumbar plexus injury during some lateral approaches, and swallowing or voice disturbances after anterior cervical surgery. Rare cervical complications include esophageal injury and airway-threatening hematoma.
Deep infection may involve graft and instrumentation. Organisms can form biofilms on implant surfaces, making implant-associated infection different from a superficial wound infection. Its management depends on timing, organism, implant stability, wound condition, and whether fusion has occurred. Because implants may be maintaining essential stability, their removal is not interchangeable with removal of a nonessential foreign body. Infection can also interfere with union and contribute to later implant loosening.
Pseudarthrosis, or nonunion, occurs when a solid bony bridge fails to develop across the intended fusion. It may be asymptomatic or associated with pain, motion, screw loosening, or implant breakage. The relationship is not one-to-one: a broken rod can indicate persistent mechanical loading across an ununited segment, but symptoms and imaging require combined interpretation. Revision may involve altered fixation, additional graft, treatment of alignment problems, or investigation for infection.
Other mechanical problems include cage subsidence into vertebral endplates, implant migration, screw pullout, and fracture near the construct. Poor bone quality increases several of these risks. In long deformity fusions, proximal junctional kyphosis describes increased angulation above the upper end of the construct. Proximal junctional failure denotes a more severe structural problem, such as fracture or disruption of supporting tissues, potentially accompanied by pain or neurological compromise.
Adjacent-segment disease refers to clinically important symptoms arising from degeneration next to a fusion and is distinguished from radiographic adjacent-segment degeneration without corresponding symptoms. Its causes include natural aging, pre-existing abnormalities, alignment, and altered loading. Persistent pain after fusion may also reflect residual nerve injury, an incompletely addressed pain source, or chronic pain mechanisms rather than failed union. These distinctions are central to interpreting an apparently unsuccessful outcome.
Clinical outcomes and comparative evidence
Outcome assessment separates radiographic union from effects that matter directly to patients. A fusion rate measures a biological and structural endpoint, but it does not fully describe pain, walking ability, neurological function, or participation in daily life. Patient-reported outcomes commonly include pain ratings and disability instruments such as the Oswestry Disability Index for lumbar disorders and the Neck Disability Index for cervical disorders. Health-related quality of life, complications, reoperation, and medication use provide additional perspectives.
Results are highly indication-dependent. Stabilizing a grossly unstable injury addresses a different clinical question from adding fusion to decompression for degenerative stenosis. Surgery for scoliosis may prioritize deformity control and balance, while cervical decompression may primarily aim to prevent deterioration from spinal cord compression. Combining these populations obscures both benefit and harm. Observational comparisons are additionally affected by treatment selection: patients receiving fusion may have more deformity, instability, or previous surgery than those receiving decompression alone.
Randomized controlled trials have clarified some disputed lumbar indications. The Swedish Spinal Stenosis Study, published in 2016, found no better clinical outcomes from routinely adding fusion to decompression for lumbar stenosis in its study population, which included patients with and without degenerative spondylolisthesis. The smaller SLIP trial, also published in 2016, reported an advantage for fusion on a physical-health measure in selected patients with stable grade I degenerative spondylolisthesis. The trials differed in size, eligibility, procedures, and outcomes, limiting direct equivalence.
The Norwegian NORDSTEN-DS trial, reported in 2021, found decompression alone noninferior to decompression with instrumented fusion for its primary disability outcome at two years in selected patients with degenerative spondylolisthesis. Five-year results published in 2024 supported that finding. These results concern defined trial populations; they do not establish that fusion is unnecessary in major deformity, severe instability, or every revision setting. They do show that degenerative vertebral slippage does not automatically imply an additional benefit from fusion.
For chronic low back pain attributed to disc degeneration, trials comparing fusion with intensive rehabilitation incorporating cognitive and exercise components have not demonstrated a consistent large advantage for fusion. Interpretation is complicated by selection criteria, crossover between treatments, and differences in the nonoperative comparison. In the cervical spine, disc replacement offers a motion-preserving alternative for selected patients, but eligibility and risk profiles differ from those of fusion.
Reoperation is informative but is not a simple measure of failure. Further surgery may address nonunion, infection, adjacent-level disease, or a planned stage of treatment. Differences in follow-up duration, access to care, and thresholds for revision influence reported rates. Consequently, robust evaluation combines structural findings, patient function, adverse events, and the durability of benefit rather than relying on a single endpoint.
Historical development and continuing research
Modern spinal fusion developed in the early twentieth century. In 1911, Russell Hibbs and Fred Albee independently described operations intended to stabilize the spine through bone union, particularly in patients with spinal tuberculosis. These early procedures relied on graft incorporation and prolonged external immobilization rather than modern internal fixation. Their biological objective—creating a stable bridge of bone—remains fundamental despite the transformation of surgical exposure, implants, and perioperative care.
Internal instrumentation expanded the capacity to correct and stabilize deformity. Paul Harrington developed a rod-and-hook system for scoliosis during the 1950s and reported his experience in 1962. Harrington instrumentation improved coronal curve correction but had limitations in controlling rotation and preserving sagittal alignment. Segmental systems subsequently distributed fixation across more points. Luque instrumentation and the Cotrel–Dubousset system were important stages in this transition, while pedicle-screw constructs later permitted stronger three-dimensional control in many spinal regions.
Interbody techniques also evolved. Ralph Cloward described posterior lumbar interbody fusion in the mid-twentieth century and contributed to the development of anterior cervical fusion. The anterior cervical techniques associated with Cloward and with Robert Robinson and Henry Bohlman’s predecessor collaborator William Smith established influential reconstructive approaches in the 1950s. Subsequent implant development introduced threaded devices, titanium and polymer cages, and expandable implants, each addressing particular access, support, or alignment goals.
Contemporary research examines the interaction between implant design and biology. Porous titanium surfaces and additively manufactured structures are intended to support bone attachment and ingrowth. Alternative graft materials and cell-containing products seek to improve union while reducing the need for autograft harvest. Comparisons require attention to product composition, viable cellular activity, regulatory status, and clinically meaningful outcomes; promising laboratory behavior alone does not demonstrate superior patient benefit.
Navigation, robotics, and automated planning are studied for their ability to improve implant placement and workflow. In adult deformity, research increasingly distinguishes alignment targets appropriate to different ages and anatomical characteristics rather than applying one correction goal uniformly. Mechanical modeling and machine learning are being investigated for prediction of complications, but the usefulness of a prediction depends on external validation and on whether acting on it improves outcomes.
Long-term study remains especially important because fusion creates a permanent change in spinal motion. Implant durability, junctional problems, adjacent-level symptoms, and function may evolve over many years. Registries and comparative studies therefore complement short-term trials by tracking uncommon complications and outcomes in broader populations, while randomized evidence remains important for determining when adding fusion produces benefit beyond decompression or other treatment alone.