Why the Reform Happened

SB 171 is a budget trailer bill, and the Legislature was explicit about why it acted: the Department of Industrial Relations projected that SIBTF liabilities, combined with a backlog of tens of thousands of pending claims, were on a trajectory toward roughly $30 billion by 2029, with employer assessments climbing toward $1.5 billion. The bill's stated intent is to restore the fund to its original purpose by narrowing the eligibility standard that had expanded under decades-old case law interpreting what counts as a qualifying pre-existing disability.

What Actually Changed

A narrower definition of "labor-disabling." SB 171 amends Labor Code §4750 to require that a pre-existing impairment actually caused lost earnings or interfered with the worker's ability to perform their job. Conditions that were being effectively managed with medication or a medical device — such that the worker was functioning without work incapacity — no longer qualify as labor-disabling on their own. This is a meaningful departure from the older standard, under which a pre-existing condition could support SIBTF eligibility even without any demonstrated effect on the worker's actual ability to work at the time.

A timing requirement for the medical evidence. The labor-disabling condition now has to be documented in medical evidence that existed at the time of the subsequent industrial injury — not reconstructed or inferred afterward. This has direct implications for how an evaluator frames causation and pre-existing impairment in a report: the record needs to show, contemporaneously, that the condition was both present and functionally limiting before the second injury occurred.

Combined values instead of simple addition. Where a worker has multiple impairment ratings, SB 171 requires them to be combined using the Combined Values Chart in the Permanent Disability Rating Schedule, rather than added together. Because the combined values method generally produces a lower total than straight addition, this changes the math evaluators and rating specialists need to apply when multiple impairments are in play.

Procedural changes to WCAB reconsideration. The bill also permanently adopts a 60-day deadline for the Workers' Compensation Appeals Board to act on a petition for reconsideration once it receives the case file — a change aimed at reducing the backlog rather than at the substantive eligibility standard itself, but one that affects how quickly SIBTF disputes move through the system.

What This Means for QME and AME Work Capacity Analysis

The practical effect on report writing is significant. A report that simply lists a pre-existing diagnosis and assumes it qualifies as labor-disabling is no longer sufficient. Under the new standard, the evaluator's work capacity analysis needs to affirmatively address:

• ·Functional impact at the relevant time. Was the pre-existing condition actually interfering with work capacity before the subsequent injury, or was it being managed without functional limitation? A diagnosis alone doesn't answer this question anymore — the analysis has to.
• ·Contemporaneous documentation. Does the medical record from the relevant time period actually support the labor-disabling finding, or is the evaluator being asked to infer backward from a current presentation? Reports built on retrospective inference are more vulnerable under the new evidentiary timing requirement.
• ·A defensible work capacity opinion, without a stand-alone vocational expert. SIBTF cases have always required this kind of analysis; SB 171 raises the bar for how well-supported it needs to be, since the eligibility threshold itself is now stricter.
• ·Correct application of the combined values methodology where multiple impairments are at issue, since simple addition of ratings will now overstate the combined result.

The Bottom Line for Attorneys and Claims Professionals

SIBTF cases that were viable under the old "labor-disabling" standard may not clear the bar under SB 171 — and reports written before the reform, or written without specific attention to functional impact and contemporaneous documentation, are more exposed to challenge. For attorneys and claims executives evaluating SIBTF exposure or preparing for an AME designation, the question worth asking early is whether the evaluator's methodology was built for the post-SB 171 standard, not the one that came before it.

The Problem Is the Workup, Not the Patient

The scale of the gap is well documented. One analysis of functional spine testing methodology put it plainly: despite decades of advances in imaging technology, the standard clinical approach to functional assessment of the lumbar spine has changed little, relies on uncontrolled, patient-directed bending, and produces high inter-patient variability — and as a direct consequence, up to 85% of low back pain patients evaluated by conventional means cannot be given a definitive diagnosis.1 That figure is not a statement about the prevalence of "unexplained" pain. It is a statement about the resolving power of the tools most commonly used to look for its source.

Static flexion-extension radiography — long treated as the clinical standard for detecting segmental instability — illustrates the gap directly. Comparative studies using computer-assisted vertebral motion analysis (VMA) against conventional static flexion-extension films found that VMA substantially outperformed the standard technique: markedly higher sensitivity (roughly 41–45% greater), higher negative predictive value, and higher specificity for detecting radiographic instability, with far greater measurement precision and reliability than manual digitization.2,3 Put differently, a meaningful proportion of segmental instability that static radiography misses — and that therefore gets folded into a "non-specific" diagnosis — is detectable with the right measurement technique. The pathology was there. The tool used to find it wasn't sensitive enough.

The same logic applies beyond instability. Molecular and functional imaging techniques such as SPECT-CT have been used to localize an active primary pain generator in patients whose CT, MRI, and flexion-extension films showed only diffuse, non-localizing degenerative change — identifying a single symptomatic level where anatomic imaging alone could not, and correlating that finding with resolution of the patient's pain after targeted treatment.4 The mechanism is worth spelling out: SPECT-CT pairs a radiotracer that concentrates at sites of elevated bone turnover and blood flow with a rotating gamma camera that maps where that activity is highest, then fuses that functional map onto a conventional CT image of the same anatomy.5 The result identifies not just what a structure looks like, but whether it is currently metabolically active — allowing a single "hot" level to be distinguished from several structurally similar but biologically quiet ones.

Two clinical scenarios illustrate why this matters. In multilevel degenerative disease, MRI or CT will often show degenerative change at three, four, or five levels simultaneously — but structural imaging cannot say which of those levels is actually generating the patient's pain. SPECT-CT can: a level with only modest structural degeneration but high tracer uptake is a more credible pain generator than an adjacent, worse-looking level with a metabolically quiet scan.4,5 In facet-mediated pain, SPECT-CT similarly identifies which facet joint is currently inflamed and biomechanically active — often before that activity is visible as structural change on CT — allowing injections, ablation, or surgical planning to be targeted to the joint that is actually driving symptoms rather than the one that simply looks worst on anatomic imaging.5 In both scenarios, a "non-specific" diagnosis reached without this functional information reflects an incomplete differential, not a genuinely unlocalizable pain source.

A caveat matters here, and it should be stated candidly: SPECT-CT's contribution to a ligamentous pain diagnosis is indirect rather than direct. The technique relies on a bone-seeking radiotracer that concentrates at sites of elevated osteoblastic activity and bone turnover; ligament tissue itself does not take up that tracer in a way that images the ligament directly, since SPECT-CT is fundamentally a study of bone metabolism, not soft tissue.6 What it can identify is the bony consequence of ligamentous dysfunction — reactive uptake at an enthesis where a ligament inserts into bone, or a secondary stress response in an adjacent facet joint or endplate absorbing the abnormal load created by segmental laxity. For directly visualizing the ligament itself — signal change, thickening, discontinuity — MRI remains the appropriate modality. Used together, the two are complementary: MRI characterizes the ligament, and SPECT-CT can confirm whether that ligament's dysfunction is producing an active, metabolically demonstrable mechanical consequence elsewhere in the segment.

The Neuroanatomic Basis: Ligaments and Discs Are Densely Innervated, Verifiable Pain Generators

The claim that ligamentous and discogenic structures are capable of generating genuine pain is not speculative — it is established at the level of neuroanatomy, and recent bench research has strengthened the case considerably.

The posterior longitudinal ligament (PLL) is supplied by branches of the sinuvertebral nerve carrying fibers immunoreactive for substance P and calcitonin gene-related peptide (CGRP) — both markers of nociceptive, pain-signaling innervation.10,11 Separately, dedicated mechanosensitive afferent units have been identified within the lumbar PLL, meaning the ligament carries both a nociceptive network and a distinct tension/position-sensing network.9 A structure with this dual innervation is not a passive restraint; it is a functioning pain sensor that responds directly to abnormal segmental motion or strain.

The anterior longitudinal ligament (ALL) likewise contains a dense supply of free nerve endings, with comparatively few mechanoreceptors — anatomy consistent with a structure whose primary innervated function is nociceptive rather than proprioceptive.12

The intervertebral disc is innervated by sinuvertebral nerve branches that normally reach only the outer third of the annulus fibrosus. In degeneration, that innervation pattern changes: nerve fibers extend inward into the inner annulus and nucleus pulposus alongside vascular ingrowth, tracking degenerative and inflammatory change (elevated phospholipase A2, TNF-α, and IL-6 have been documented in this setting).13 This is a well-established, biologically specific mechanism for discogenic pain — not an unexplained phenomenon.

The vertebral body itself carries PGP 9.5-positive nerve fibers concentrated posteriorly near the basivertebral foramen, following the basivertebral vessels to the posterior vertebral body, endplate, and PLL.14 This vertebrogenic pain pathway — the anatomic basis for basivertebral nerve ablation as a targeted intervention — is a further example of a specific, localizable pain generator that a standard workup focused on the disc and facet alone will not identify.

The facet joint capsule contains an extensive network of free and encapsulated nerve endings, including low-threshold mechanoreceptors, mechanically sensitive nociceptors, and silent nociceptors, along with substance P- and CGRP-containing fibers.15,16 Capsular strain from trauma — hyperextension or whiplash-type loading in particular — directly activates this nociceptive network, and inflammation has been shown to lower the activation threshold of these endings and increase their baseline firing rate.15,16

The remaining segmental ligaments of the spine — interspinous, supraspinous, intertransverse, and iliolumbar — are innervated as well, not inert. The interspinous ligament receives innervation from the medial branches of the lumbar dorsal rami, and histological work has identified Paciniform corpuscles, Ruffini end organs, and fine free nerve fibers in both the interspinous and supraspinous ligaments.19,20 The iliolumbar ligament has been mapped directly: immunohistochemical analysis of cadaveric specimens found mechanoreceptors and nociceptors distributed throughout the ligament, most densely concentrated at its iliac wing insertion.21 The sacroiliac joint's supporting ligaments follow the same pattern: substance P- and CGRP-immunoreactive nociceptive fibers have been demonstrated in both the anterior capsular ligament and the interosseous ligament of the SI joint.22

The ligamentum flavum is the instructive exception, and it strengthens rather than undermines the broader argument. In its normal, healthy state, the ligamentum flavum has comparatively little innervation.23 But that picture changes with pathology. More recent histological analysis of ligamentum flavum specimens from patients with lumbar spinal stenosis has demonstrated nociceptive nerve fiber sprouting into the ligament in the setting of chronic inflammation and degenerative change.24,25 A ligamentum flavum that appears unremarkable on a resting MRI is not, by itself, evidence against a ligamentous contribution; the innervation — and therefore the capacity to generate pain — is acquired specifically in the degenerated or hypertrophied state.

Perhaps most significant for the "non-specific" debate: recent bench research modeling capsular ligament injury has shown that biochemical or mechanical injury to the collagen matrix of the facet capsular ligament — using intra-articular collagenase or matrix metalloproteinase-1 (MMP-1) — produces sustained pain behavior and measurable dysregulation in the dorsal root ganglia and spinal cord, without any accompanying structural degeneration, cartilage change, or visible joint destruction.17,18 In other words, a ligament can be genuinely, mechanistically injured and actively generating pain while looking entirely normal on imaging. That is not evidence of "no pathology." It is evidence that the pathology occurs at a microstructural and neurochemical level below the resolution of conventional imaging.

Taken together, this body of neuroanatomic and mechanistic evidence undercuts the premise behind a default "non-specific" diagnosis. The ligamentous and discogenic structures of the spine are not inert connective tissue; they are richly innervated, mechanistically well-characterized pain generators. When pain is present and these structures have not been affirmatively evaluated — biomechanically, functionally, and neuroanatomically — the correct conclusion is that the pain generator has not yet been identified, not that one does not exist.

What a Complete Workup Actually Looks Like

Treating "non-specific" as the default outcome rather than the last resort inverts the diagnostic process. A rigorous evaluation — the kind that should precede any "non-specific" label — includes:

1. 1.The AOMSI (alteration of motion segment integrity) study — a digitized, computer-assisted biomechanical functional motion analysis, rather than manual measurement of static films — to detect segmental instability and translational/rotational abnormalities with the sensitivity and reliability that manual techniques lack.2,3
2. 2.Functional and provocative assessment, not merely static imaging, since many mechanical pain generators — facet-mediated pain, ligamentous laxity, segmental hypermobility — manifest only under load or motion and are invisible on a resting MRI.
3. 3.Systematic screening of every structure capable of generating spine pain — the disc and its innervated annular layers, the facet joint capsule, the anterior and posterior longitudinal ligaments, the segmental ligaments (interspinous, supraspinous, intertransverse, iliolumbar, and ligamentum flavum), the vertebral cartilaginous endplate, the sacroiliac joint and its supporting ligaments, myofascial tissue, and neural structures — rather than stopping the differential once fracture, tumor, infection, and frank radiculopathy have been excluded.9–25
4. 4.Correlation of mechanism of injury, clinical examination findings, and biomechanical data, so that a proposed pain generator is supported by more than one line of evidence.
5. 5.Assessment of central and peripheral sensitization as a contributing mechanism, not as a diagnosis of exclusion invoked only after everything else has failed to explain the pain.

The Medical-Legal Consequence

This matters directly in workers' compensation and personal injury evaluation. The AMA Guides' framework for rating spinal impairment — the Diagnosis-Related Estimates method — is built around documented objective findings: radiculopathy, loss of motion segment integrity, confirmed structural pathology.7 The Guides also contain a separate pathway (Chapter 18 in the 5th Edition) for pain-related impairment where no definable tissue pathology has been identified.7 That pathway exists as a fallback — but a fallback is not where a competent evaluation should routinely end up. If a "non-specific" diagnosis is reached without first exhausting biomechanical and functional diagnostic methods capable of identifying instability, a specific pain generator, or a mechanical contributor, the diagnosis reflects an incomplete record, not a genuinely unexplained condition.

For the medical-legal evaluator, the practical standard should be:

• ·A "non-specific" diagnosis is only defensible after the full available diagnostic armamentarium has been applied — not merely static imaging and a red-flag screen.
• ·AOMSI digitized motion measurements should be part of the record wherever instability is plausible on history or mechanism, given their demonstrated superiority over manual measurement of static films.2,3
• ·Every candidate pain generator should be affirmatively addressed and either implicated or excluded, with the reasoning documented, rather than left unaddressed on the way to a default "non-specific" conclusion.
• ·A causation or impairment opinion built on a genuinely comprehensive workup is far more defensible on cross-examination than one built on a truncated evaluation that simply stopped looking once the easy answers were exhausted.

A Note on Methodology and Foundation

It should be acknowledged candidly: some payer policies still classify certain computerized motion-analysis technologies as investigational for general use, and conventional guideline literature continues to use "non-specific low back pain" as its default nomenclature once red flags and radiculopathy are excluded.8 That is precisely the gap this article is addressing — the conventional nomenclature reflects the limits of the conventional workup, not a ceiling on what can actually be determined. An evaluator relying on the AOMSI study or other digitized, computer-assisted functional motion studies should be prepared to lay the appropriate foundation for the methodology, cite the comparative validation literature, and explain clearly why the finding changes the diagnostic picture.2,3,8 Done properly, that foundation is what turns "non-specific" into a specific, defensible diagnosis.

Conclusion

A diagnosis of "non-specific spine pain" is, in the great majority of cases, a description of what the evaluation failed to find — not a description of what is actually happening in the patient's spine. Genuine pain complaints correlate with specific, identifiable mechanical, neurological, or soft-tissue causes far more often than conventional workups suggest, once the diagnostic process includes the AOMSI study and a systematic canvass of every candidate pain generator. "Non-specific" should be the conclusion an evaluator reaches only after that full process has been exhausted — not the default label applied when it wasn't attempted.

The Problem

Flexion-extension films frequently show measurable ligamentous laxity — real, quantifiable instability — that falls short of the full AOMSI threshold but clearly exceeds a normal, uninjured segment. A strict either/or reading forces the evaluator to choose between two positions that both invite cross-examination: rate it Category II and ignore objective pathology on the films, or stretch it to Category IV and overstate an unmet threshold.

FDA-Cleared Biomechanical Analysis

Digital motion analysis using FDA-cleared diagnostic protocols and software tools quantifies the actual translation and angular displacement at each level, expressed as a precise percentage of the AOMSI threshold — not a subjective read of the films. Where that percentage falls short of full AOMSI but exceeds normal segmental motion, a prorated impairment rating interpolates a WPI value between the Category II and Category IV endpoints in proportion to the measured pathology, rather than defaulting to a category boundary that doesn't reflect what the films actually show.

Guides & Case-Law Basis for Proration

The AMA Guides, Chapter 1, page 11, acknowledge that the range and evolution of medical conditions means the Guides cannot supply a specific rating for every clinical presentation, and direct the evaluator to reasoned physician judgment — including analogy to comparable tables and criteria — to reach an accurate, reproducible result. The Tables 15-3/15-4/15-5 AOMSI criteria are exactly this kind of gap: they supply a single pass/fail threshold, with no intermediate provision for a motion segment that is measurably, objectively unstable but has not yet crossed the full translation/angular-motion line.

California's WCAB en banc authority in Almaraz/Guzman confirms that where a strict table-based rating does not fairly and accurately capture the documented impairment, the evaluator may depart from it — provided the departure stays within the four corners of the Guides and is grounded in objective, reasoned medical analysis rather than a desired outcome. A prorated AOMSI rating meets that standard on its face: it is not an outside methodology, but a proportional application of the Guides' own measured criteria to a presentation the Table itself does not address.

Note: page 11's "cannot provide an impairment rating for all impairments" language is most often invoked for conditions or body parts absent from the tables entirely (e.g., analogizing an unlisted muscle injury). This application extends it to a sub-threshold finding within an existing table rather than an absent condition — a reasonable and arguably stronger extension, but the distinction should be anticipated on cross.

Why It Holds Up

A prorated AOMSI rating is defensible across four dimensions:

• ·Traceable — the rating traces to a measured value on the films, not a judgment call
• ·Balanced — avoids both failure modes of the binary approach: under-rating real instability, or overreaching to justify Category IV
• ·Neutral — gives applicant and defense counsel a number they can evaluate against objective data rather than argue over categorically
• ·SB 171-aligned — matches the same evidentiary rigor now expected of QME/AME reporting: traceable, defensible, ratable

Conclusion

This approach does not replace DRE methodology — it disciplines it, using FDA-cleared biomechanical motion data, page 11's own acknowledgment of the Guides' limits, and Almaraz/Guzman's four-corners standard to justify where within, or between, the categories a given case actually belongs.

The Problem

In motor vehicle collision cases, the defense's most reliable argument is that a claimant's spine complaints are "soft tissue," "non-specific," and unsupported by objective findings — that pain is subjective and static imaging (plain films, MRI at rest) looks unremarkable. That argument works because a static image genuinely cannot distinguish a ligamentous sprain from a normal, asymptomatic spine.

Flexion-extension (dynamic) radiographs close that gap — but only if the reading is quantified. A qualitative "looks a little loose" read is exactly the kind of subjective impression that invites cross-examination on both sides: plaintiff's counsel needs more than an impression to prove a real structural injury occurred, and defense counsel needs more than an impression to rule one out.

FDA-Cleared Biomechanical Analysis

Digital motion analysis using FDA-cleared diagnostic protocols and software tools measures actual anterior-posterior translation (mm) and angular rotation (°) at each motion segment on flexion and extension views, and compares those measured values against validated segmental instability thresholds (White & Panjabi). The output is a number at each level, not an impression — and that number resolves the case into one of two evidentiary postures.

1. 1.Measurements within normal limits — confirms the diagnosis, doesn't inflate it. Where translation and angular rotation at every level fall within normal segmental limits, the analysis affirmatively rules out structural ligamentous disruption or instability. That is not a null result — it is an objective finding that supports a diagnosis of non-specific axial spine pain (myofascial strain/sprain) rather than a structural injury, which keeps the diagnosis and any damages claim anchored to what the evidence actually shows. It also forecloses a defense argument that the claimant's team is overstating a soft-tissue injury as a structural one, and it keeps the treating and evaluating physicians' opinions defensible on cross.
2. 2.Measurements exceeding threshold — objective evidence of ligamentous injury, level by level. Where measured translation or angular rotation exceeds the validated threshold at one or more levels, the finding is pathological segmental instability — a structural injury, not a subjective complaint. The direction and plane of the excess motion identifies which ligamentous structures failed, which lets the report speak to mechanism and causation rather than simply asserting "instability is present" in the abstract.

Reading the Direction of Instability

Naming the ligament, not just the segment, turns "instability at C5-6" into a mechanism-consistent finding a jury and opposing counsel can follow: a rear-impact hyperflexion/hyperextension mechanism producing posterior-complex or anterior-longitudinal-ligament failure at a specific, measured level.

• ·Excess flexion translation: excess anterior translation or hyperflexion on the flexion view implicates failure of the posterior ligamentous complex — the interspinous and supraspinous ligaments, ligamentum flavum, facet joint capsules, and posterior longitudinal ligament.
• ·Excess extension translation: excess posterior translation or hyperextension angulation on the extension view implicates the anterior longitudinal ligament and anterior annulus.
• ·Excess angular rotation, translation normal: rotation beyond threshold without a corresponding translation abnormality implicates facet capsule laxity specifically, distinct from a posterior-complex or annular injury.
• ·Upper cervical (C1-C2) findings: at C1-C2, translation/rotation findings are analyzed against the transverse and alar ligament complex rather than the standard subaxial framework — see the separate atlantoaxial instability citation set for the mechanism and threshold literature specific to that level.

Age-Dating the Injury: Acute Trauma vs. Pre-Existing Instability

Instability alone doesn't prove the collision caused it — defense counsel's next move is always to argue the laxity predates the crash. The motion analysis is coupled with a degenerative correlation review at the same level(s) to address that directly.

• ·No degenerative correlate at the unstable level: Degenerative markers that take years to develop — disc height loss (Pfirrmann grading; Wilke lumbar / van Eerd cervical scoring), Modic endplate changes, facet arthropathy, and marginal osteophyte formation — are graded at the unstable level. Measurable instability at a level with no meaningful degenerative correlate is inconsistent with a long-standing, pre-existing condition and consistent with an acute, trauma-onset ligamentous injury.
• ·Degenerative correlate present: Where degenerative changes are present alongside the instability, the analysis distinguishes aggravation of a pre-existing, previously asymptomatic condition from a new injury — relevant to causation and damages rather than to a workers'-compensation-style apportionment percentage.
• ·Prior imaging, if available: Prior imaging, when it exists, is the single strongest data point for age-dating and should always be requested and compared directly against the post-collision study before relying on degenerative correlation alone.

Value to Both Sides of the Case

Because the methodology is the same regardless of who orders it, the measured result gives plaintiff and defense counsel different but equally legitimate uses for the same underlying fact. Neither side is buying a conclusion — each is buying an early, reliable answer to a question their case turns on anyway.

For plaintiff/applicant counsel: a confirmed instability finding, with the implicated ligament(s) named, substantiates a structural injury early — before deposition, before an IME dispute — and anchors settlement demands and trial exhibits to a measured value rather than a subjective pain narrative. A normal-limits finding still has value: it takes a structural-injury dispute off the table so the case can be litigated honestly as what it is — a soft-tissue/non-specific-pain claim — without the credibility cost of an overstated structural theory.

For defense/carrier: a confirmed instability finding lets the adjuster or defense counsel set reserves accurately and make an informed early-settlement decision on a claim that is, in fact, real — rather than litigating a genuine injury for two years on the assumption it's soft tissue. A normal-limits finding supports setting reserves appropriately for a non-specific soft-tissue claim, and gives the carrier a defensible, objective basis — not just an adjuster's opinion — for resisting an inflated demand or an unsupported structural theory.

The measurement doesn't advocate for either side — it simply removes the guesswork each side is currently paying to litigate around. That is also why timing matters: whichever party commissions the analysis first controls when that evidentiary record is created, how it's framed in the initial report, and how the opposing side has to respond to it rather than originate it themselves.

Why It Holds Up

This does not replace the treating or evaluating physician's clinical judgment — it gives that judgment a quantified, reproducible foundation: a measured value at a named level, referenced against validated biomechanical thresholds, correlated against degenerative findings to address timing, and translated into the specific ligamentous structures a defense expert has to actually contest.

• ·Traceable — the finding traces to a measured translation/rotation value at a named level, not an impression of the films.
• ·Bidirectional — the same methodology confirms a non-specific diagnosis when instability is absent and documents structural injury when it's present — it isn't built to find damage on demand, which is what makes it credible to both sides.
• ·Mechanism-specific — naming the implicated ligament(s) ties the finding to the collision mechanism instead of leaving "instability" as an unexplained label.
• ·Causation-ready — pairing the instability finding with degenerative correlation gives the causation opinion — acute vs. pre-existing — an objective basis instead of a bare assertion.

The Party That Controls the Evidence Holds the Advantage

Whichever side commissions the motion analysis first sets the evidentiary record the other side has to respond to. Early retention is itself a strategic decision — not just a clinical one.