# LOMN overlay — musculoskeletal / spine / pain > Specialty overlay for letter-of-medical-necessity-builder that applies stricter MSK, spine, orthopedic, and pain-management validations: conservative-care step-therapy rules per procedure (spine surgery, neurostimulator, RFA, ESI, facet injection, DME), ODI/NDI/VAS-pain trend analysis across baseline and post-conservative-care follow-up, and imaging-to-symptom concordance checks. Invoked from within the core LOMN workflow when the case specialty is MSK; adds requirements, does not weaken the core. Tags: Healthcare, Prior Authorization, Denial Appeals, Musculoskeletal, Orthopedics, Pain Management, Letter of Medical Necessity ## Example Prompts - Apply the MSK overlay to the lumbar fusion LOMN I am drafting — check the conservative-care documentation before I hand it to the physician - Run the MSK score trend check on ODI and VAS-pain at baseline and after 8 weeks of PT to build the escalation argument for a spinal cord stimulator trial - Check my imaging impression matches the clinical picture for this radiculopathy case before the payer reviewer catches the laterality mismatch - Validate conservative-care step-therapy for this radiofrequency-ablation request — prior injection, PT, and NSAID trial documentation URL: https://rakenne.app/skills/lomn-msk/index.md Try this skill: https://rakenne.app/a/?skill=lomn-msk ## Overview Spine, orthopedic, and pain-procedure prior authorizations are denied on patterns the core LOMN builder can't see: six weeks of PT instead of the payer's required twelve, a lumbar fusion requested on an L4-L5 symptom picture that the MRI impression actually locates at L5-S1, an SCS implant with no documented trial, an RFA with no prior diagnostic medial-branch block. The core LOMN builder enforces ICD-10 specificity, prior-therapy structure, and disease-score capture — it does not enforce MSK-specific sequencing or imaging concordance. The `lomn-msk` skill is a **specialty overlay** on top of `letter-of-medical-necessity-builder`. It is invoked from within the core LOMN workflow when the specialty hook is set to `msk` (lumbar) or `msk-cervical` (cervical). It never drafts prose. It returns deterministic validation results — conservative-care rule pass/fail per procedure, ODI/NDI/VAS-pain trend direction against published MCID thresholds, and imaging-to-symptom concordance on level and laterality — that the core skill's export gate refuses to release an LOMN without. ## What it covers - **Procedure-specific conservative-care step-therapy.** Eleven recognized procedures (`fusion`, `discectomy`, `laminectomy`, `adr`, `scs-trial`, `scs-implant`, `rfa`, `esi`, `facet-injection`, `tens-dme`, `back-brace-dme`). Shared baseline requires at least 6 PT sessions over 6+ weeks and a documented NSAID trial. Fusion and laminectomy additionally require a prior interventional option (ESI, facet injection, or MBB) with a documented outcome. SCS trial requires a psychological evaluation and prior interventional failure. SCS implant requires a successful trial (≥50% pain reduction or functional improvement) plus psych eval. RFA requires at least one prior diagnostic medial-branch block or facet injection. - **ODI / NDI / VAS-pain trend analysis.** Each instrument needs a baseline and a post-conservative-care follow-up. Trends are classified as `worsening`, `stable-insufficient`, `improving-undermines-necessity`, `improving`, or `insufficient-data` using published MCID thresholds (ODI 10 points, NDI 7 points, VAS-pain 2 points). An improvement that exceeds MCID is surfaced as a necessity-undermining signal — the LOMN must either continue conservative care or narrate why the improvement is insufficient before escalating. - **Imaging-to-symptom concordance.** Symptom laterality, distribution, and suspected level are cross-checked against the radiology impression's findings on level and laterality. Non-structural or unspecified finding types (`other`) do not count toward concordance. Concordance is required for spine-surgery (fusion, discectomy, laminectomy, ADR), SCS, ESI, and RFA requests. - **Required evidence bundle.** Imaging report with a concordant impression, PT notes documenting at least 6 sessions with SOAP findings and functional goals, ODI or NDI baseline and follow-up plus VAS-pain at both time points, medication trials with dose/duration/outcome/discontinuation, prior diagnostic block (RFA), successful SCS trial plus psych eval (SCS implant). - **Guideline citation pointers.** NASS coverage policies, ACP / APS "Noninvasive Treatments for Low Back Pain", ACR Appropriateness Criteria for low back pain imaging, AAOS positions, and ODI / NDI reference publications are named by pointer — the overlay does not inline guideline text; it tells the LOMN which citation `clinical-claim-evidence-mapper` must anchor to. - **Never weakens the core.** Overlay outputs feed into the same core LOMN export gate. An MSK LOMN must pass both the core gates (`disease_score_capture`, `prior_therapy_history_validator`, `lomn_completeness_check`) and the overlay gates. ## Example conversation > **You:** I'm drafting the LOMN for case 2026-0412-88174 — L4-L5 lumbar fusion. The core skill just finished prior-therapy validation. Hand it to the MSK overlay. > > **Agent:** Reading the case record and the in-progress LOMN at `drafts/lomn-2026-0412-88174.md`. Specialty hook is `msk` (lumbar). Requested procedure parsed from the LOMN's `requested_therapy` block as `fusion`. Running the overlay. > > **Agent:** Calling `msk_conservative_care_validator` with `procedure_requested: "fusion"` and the normalized conservative-care record — PT 4 sessions over 4 weeks, NSAID trial yes (ibuprofen 600 mg TID x 6 weeks, inadequate relief), muscle relaxant no, home exercise yes, one prior right L4-L5 transforaminal ESI on 2026-01-18 (< 2 weeks of relief), no psych eval (not required for fusion), imaging MRI lumbar spine 2026-02-03, imaging.concordant_with_symptoms flag not yet set. > > **Agent:** Two required rules failed. `pt-6-weeks-and-6-sessions`: **fail** — 4 sessions over 4 weeks, below the 6-and-6 threshold. `imaging-concordant-with-symptoms`: **fail** — concordance flag is not affirmed; run `msk_imaging_concordance_check` before this rule can pass. `prior-interventional-option-trialed`: pass (ESI recorded with outcome). `nsaid-trial`: pass. Returning `isError: true` — the LOMN cannot leave the workspace until the PT documentation and the concordance check clear. > > **You:** The clinic has more PT sessions — they just weren't on the initial upload. I'll add the continuation notes (sessions 5 through 14, weeks 5 through 14). And yes, run the concordance check — symptoms are right-sided L5 radiculopathy, distribution L5 dermatomal, suspected level L4-L5. > > **Agent:** Re-ingesting the continuation PT notes. PT now documented at 14 sessions over 14 weeks. Re-running `msk_conservative_care_validator`: `pt-6-weeks-and-6-sessions` now passes. > > **Agent:** Calling `msk_imaging_concordance_check` with `symptoms: { laterality: "right", distribution: "L5 dermatomal", level_suspected: "L4-L5" }` and the MRI impression's parsed findings. Impression lists two findings: (1) L4-L5 right paracentral disc extrusion, moderate; (2) L5-S1 mild facet arthropathy, bilateral. > > **Agent:** `concordant: true`. Finding 1 matches on level (`L4-L5` ↔ `L4-L5`), laterality (`right` ↔ `right`), and finding type (`extrusion`, structural). Finding 2 is flagged as non-concordant on laterality (bilateral facet arthropathy does not explain a right-sided L5 radiculopathy) but does not block — concordance is affirmed by Finding 1. Updating the LOMN's conservative-care block with `imaging.concordant_with_symptoms: true` and re-running `msk_conservative_care_validator`. All required rules pass. `ok: true`. > > **Agent:** Now trending the disease scores. Calling `msk_score_trend_check` with the ODI and VAS-pain entries. ODI: baseline 42 (moderate) on 2026-01-05, follow-up 56 (severe) on 2026-03-10 — delta +14, exceeds MCID of 10 — trend **worsening**. VAS-pain: baseline 6 (moderate) on 2026-01-05, follow-up 8 (severe) on 2026-03-10 — delta +2, meets MCID — trend **worsening**. Overall trend label: `worsening`. Returning suggested narrative sentences for the LOMN rationale block. > > **Agent:** Handing the overlay results back to the core skill. Core is re-running `disease_score_capture`, `prior_therapy_history_validator`, and `lomn_completeness_check` on the updated draft. All three pass. The LOMN is ready for `clinical-claim-evidence-mapper` citation and export. > > **You:** Add the NASS citation to the rationale block before export. > > **Agent:** Noted — NASS 2024 Lumbar Fusion Coverage Criteria pointer added to the rationale block. `clinical-claim-evidence-mapper` will anchor that claim to the guideline citation, not to the chart. ## Sample output excerpt The overlay produces three JSON payloads that the core LOMN workflow consumes and cites. Outputs from the L4-L5 fusion case after the conversation above. --- `msk_conservative_care_validator` (after PT documentation was updated and concordance was affirmed): ```json { "procedure": "fusion", "rules": [ { "rule": "pt-6-weeks-and-6-sessions", "required": true, "passed": true, "explanation": "Physical therapy documented: 14 sessions over 14 weeks." }, { "rule": "nsaid-trial", "required": true, "passed": true, "explanation": "NSAID trial documented." }, { "rule": "home-exercise-program", "required": false, "passed": true, "explanation": "Home exercise program documented." }, { "rule": "imaging-concordant-with-symptoms", "required": true, "passed": true, "explanation": "Imaging (MRI) flagged concordant with clinical picture." }, { "rule": "prior-interventional-option-trialed", "required": true, "passed": true, "explanation": "Prior interventional option documented (1 record(s))." }, { "rule": "muscle-relaxant-or-neuropathic-agent", "required": false, "passed": false, "explanation": "Muscle relaxant / neuropathic-pain agent not recorded; not strictly required but common reviewer expectation." } ], "ok": true } ``` `msk_score_trend_check`: ```json { "results": [ { "instrument": "ODI", "baseline": { "instrument": "ODI", "value": 42, "date": "2026-01-05" }, "follow_up": { "instrument": "ODI", "value": 56, "date": "2026-03-10" }, "delta": 14, "mcid": 10, "trend": "worsening", "narrative_sentences": [ "ODI worsened from 42 (severe) on 2026-01-05 to 56 (severe) on 2026-03-10, a change of +14 against an MCID of 10.", "This trajectory supports escalation beyond the current conservative-care regimen." ], "issues": [] }, { "instrument": "VAS-pain", "baseline": { "instrument": "VAS-pain", "value": 6, "date": "2026-01-05" }, "follow_up": { "instrument": "VAS-pain", "value": 8, "date": "2026-03-10" }, "delta": 2, "mcid": 2, "trend": "worsening", "narrative_sentences": [ "VAS-pain worsened from 6 (moderate) on 2026-01-05 to 8 (severe) on 2026-03-10, a change of +2 against an MCID of 2.", "This trajectory supports escalation beyond the current conservative-care regimen." ], "issues": [] } ], "overall_trend": "worsening", "suggested_narrative": [ "ODI worsened from 42 (severe) on 2026-01-05 to 56 (severe) on 2026-03-10, a change of +14 against an MCID of 10.", "This trajectory supports escalation beyond the current conservative-care regimen.", "VAS-pain worsened from 6 (moderate) on 2026-01-05 to 8 (severe) on 2026-03-10, a change of +2 against an MCID of 2.", "This trajectory supports escalation beyond the current conservative-care regimen." ], "ok": true } ``` `msk_imaging_concordance_check`: ```json { "concordant": true, "discordances": [ { "symptom": "right L5 dermatomal suspected L4-L5", "imaging": "mild bilateral facet-arthropathy at L5-S1", "reason": "Symptom laterality 'right' not explained by imaging laterality 'bilateral' at a non-suspected level." } ], "notes": [ "At least one imaging finding matches both the suspected level and the laterality of the clinical picture." ] } ``` ## Extension tools and validations The overlay registers three deterministic validators. None draft prose. The LLM narrates; the validators gate export when required rules fail. ### `msk_conservative_care_validator` - **Inputs:** `procedure_requested` (one of 11 literals — `fusion`, `discectomy`, `laminectomy`, `adr`, `scs-trial`, `scs-implant`, `rfa`, `esi`, `facet-injection`, `tens-dme`, `back-brace-dme`, `other`), and a `conservative_care` record (PT sessions + duration, NSAID / muscle relaxant / home-exercise flags, prior injections with type / date / outcome, psych evaluation flag, BMI counseling flag, imaging modality / date / concordance flag, SCS trial outcome with percent pain reduction / functional improvement). - **Shared baseline:** `pt-6-weeks-and-6-sessions` (required for clinical procedures), `nsaid-trial` (required for clinical procedures), `home-exercise-program` (recommended), `imaging-concordant-with-symptoms` (required for spine surgery, ADR, ESI, RFA, facet injection, SCS trial, SCS implant). - **Procedure-specific rules:** fusion / laminectomy → `prior-interventional-option-trialed` (required); discectomy → `radicular-pathway-documented` (recommended); ADR → `adr-structural-indication-acknowledged` (recommended); SCS trial → `psych-evaluation-done` + `failed-conservative-and-interventional` (both required); SCS implant → `scs-trial-documented-successful` (≥50% pain reduction or functional improvement, required) + `psych-evaluation-done` (required); RFA → `diagnostic-mbb-or-facet-block-prior` (required); TENS DME → `pt-and-pharmacologic-trial` (required); back brace DME → `structural-indication-for-brace` (recommended). - **Returns:** `{ procedure, rules[], ok }` with per-rule `{ rule, required, passed, explanation }`. - **`isError: true`:** when any `required: true` rule fails — the LOMN cannot leave the workspace until the chart or the narrative is updated. ### `msk_score_trend_check` - **Inputs:** `scores[]` — each entry `{ instrument: "ODI" | "NDI" | "VAS-pain", value: number, date: ISO-8601 }`. - **What it does:** groups entries by instrument, sorts by date, treats the earliest as baseline and the latest as follow-up. Computes delta and compares to published MCID thresholds (ODI 10, NDI 7, VAS-pain 2). Classifies each instrument's trend as `worsening` (delta ≥ +MCID), `improving-undermines-necessity` (delta ≤ -MCID), `stable-insufficient` (change below MCID in either direction), or `insufficient-data` (missing baseline or follow-up). Buckets scores into published severity labels (ODI minimal/moderate/severe/crippled/bed-bound; NDI none/mild/moderate/severe/complete; VAS-pain mild/moderate/severe). - **Returns:** per-instrument `{ baseline, follow_up, delta, mcid, trend, narrative_sentences[], issues[] }`, an `overall_trend` label (priority: worsening → stable-insufficient → improving-undermines-necessity → insufficient-data → improving), and a flat `suggested_narrative[]` the LOMN rationale block can cite. - **`isError: true`:** when any required instrument is missing a baseline or a follow-up (`issues[]` non-empty). VAS-pain is always required; ODI is required when any ODI or VAS entry is supplied; NDI is required when any NDI entry is supplied. ### `msk_imaging_concordance_check` - **Inputs:** `symptoms` (laterality `left|right|bilateral|midline|unspecified`, distribution string, suspected level string), `imaging_impression.findings[]` (each `{ level, type: herniation|stenosis|extrusion|protrusion|facet-arthropathy|other, laterality, severity }`). - **What it does:** normalizes level tokens (uppercased, dashes unified, substring match so `L5` matches `L4-L5`). Checks laterality compatibility — `bilateral` findings match `left`, `right`, or `bilateral` symptoms; `midline` only matches `midline`; `unspecified` is treated as compatible. A finding counts toward concordance only when level matches, laterality is compatible, and the finding type is structural (anything other than `other`). Non-matching findings are surfaced as `discordances[]` so the LOMN can narrate around an acknowledged mismatch. - **Returns:** `{ concordant, discordances[], notes[] }`. `concordant: true` requires at least one finding that matches on level, laterality, and structural finding type. - **`isError`:** not set — the tool surfaces discordances so the drafting agent can narrate around an acknowledged mismatch. The hard gate is enforced by `msk_conservative_care_validator`'s `imaging-concordant-with-symptoms` rule, which reads the concordance flag the LOMN sets after this check runs. ### Validation invariants - **The overlay strengthens, never weakens.** An MSK LOMN must pass both the core gates and the overlay gates; a failing overlay required rule blocks export regardless of core pass. - **Imaging concordance is required for spine surgery, SCS, ESI, and RFA.** Discordance does not automatically disqualify the request — but the narrative must address the mismatch. An unacknowledged discordance is a predictable denial. - **SCS implants require a documented successful trial.** Direct-to-implant with no trial record fails `scs-trial-documented-successful`. - **RFA requires a prior diagnostic block.** No prior medial-branch block or facet injection with a documented outcome → `diagnostic-mbb-or-facet-block-prior` fails. - **Score trends are not optional for escalation.** A spine-surgery or neurostimulator LOMN that does not show worsening or stable-insufficient ODI / NDI / VAS-pain after conservative care fails the escalation-rationale test and invites a step-therapy denial. - **MCID-exceeding improvement surfaces as `improving-undermines-necessity`.** The LOMN must either continue conservative care or narrate why the improvement is insufficient (residual functional deficit, symptom rebound, persistent red-flag findings). ## Getting started The `lomn-msk` skill is an **overlay**, not a standalone workflow. It assumes the case record, prior-therapy history, and disease scores are already structured by the core LOMN builder. 1. Install `letter-of-medical-necessity-builder` first. The overlay reads the core skill's `case.json`, `requested_therapy` block, prior-therapy history, and disease-score capture. Install `lomn-msk` alongside it — it will be auto-invoked by the core skill when the specialty hook resolves to `msk` or `msk-cervical`. 2. Install `denial-intake-normalizer` upstream so the case record exists before the LOMN is drafted. Install `clinical-claim-evidence-mapper` downstream so every clinical claim the LOMN asserts is cited back to a chart excerpt or a named guideline pointer. 3. Do not invoke this overlay standalone. It does not draft prose, does not read raw payer correspondence, and does not own the export gate. Trigger it through the core LOMN workflow with a prompt like "apply the MSK overlay to this LOMN" or "check conservative care for this spine request" — the core skill will pass the normalized inputs to the overlay tools and fold the results back into the draft. 4. Confirm `msk_conservative_care_validator` returns `ok: true` for the requested procedure. Fix any `required: true` rule that failed — add chart excerpts, extend PT, run the concordance check — before the core export gate can pass. 5. Confirm `msk_score_trend_check` returns a `worsening` or `stable-insufficient` overall trend for escalation cases. An `improving-undermines-necessity` trend requires either continued conservative care or a narrated rationale for why the improvement is insufficient. 6. Confirm `msk_imaging_concordance_check` returns `concordant: true` for spine-surgery, SCS, ESI, and RFA cases; when `concordant: false`, either update the symptom or imaging inputs or narrate the mismatch explicitly in the LOMN rationale. 7. Hand the updated LOMN back to the core skill. Core re-runs `disease_score_capture`, `prior_therapy_history_validator`, and `lomn_completeness_check`, then passes the draft to `clinical-claim-evidence-mapper` for citation and export. The overlay does not address oncology, rheumatology, GI biologics, dermatology, behavioral health, or pulmonary workflows. Use the corresponding specialty overlay for those cases. --- Back to [Skill Library](https://rakenne.app/skills/index.md)