Blog · Authorization Orchestration
Ninety Percent Upheld: What California's UR and IMR Numbers Are Telling Treating Physicians
Independent Medical Review volume is climbing again, nine out of ten utilization review denials survive appeal, one percent of physicians generate forty-two percent of the disputes, and new regulations took effect on April 1, 2026. Here is what a decade of DWC and CWCI data says about where providers actually win — and lose.
California publishes more data about treatment disputes than any other workers' compensation system in the country. The Division of Workers' Compensation (DWC) releases an annual Independent Medical Review report; the California Workers' Compensation Institute (CWCI) has tracked every IMR decision letter since 2015; RAND has now published two commissioned studies on utilization review and medical access. Most treating physicians never read any of it. They experience UR as a fax that comes back with a denial and IMR as a form someone in the office files and forgets. This article reads the data from the provider's side of the desk and draws out the lessons it supports.
Key takeaways
- IMR volume fell 31% from its 2018 peak and has risen every year since 2022; DWC issued 152,351 final determinations in 2025.
- Reviewers upheld UR denials 88–92% of the time in every year from 2015 to 2025, so the RFA — not the appeal — is where providers win.
- Evaluation and management, functional restoration and behavioral health are the categories most often overturned; opioids, hypnotics and muscle relaxants almost never are.
- One percent of physicians generate 42% of disputes, and the April 1, 2026 regulations add numeric pattern-and-practice triggers, a 5-business-day drug clock and a 10-day IMR window for formulary disputes.
- Track every RFA from submission to determination and cite the MTUS guideline that is in force today.
1. The volume is going back up
CWCI's March 2026 analysis of 1.68 million IMR decision letters shows a clear V-shape. Letters peaked at 184,735 in 2018, fell 31% to a low of 127,215 in 2022, and have risen every year since: up 2.9% in 2023, 8.2% in 2024 and 7.6% in 2025.1 DWC's own count — which tallies final determination letters from its own case system and therefore differs from CWCI's totals — tells the same story: 130,774 letters in 2023, 141,621 in 2024 and 152,351 in 2025, with 201,037 applications received last year.2,3,4
Two things explain the decline and the rebound. The decline was largely pharmaceutical. After the late-2017 opioid guideline and the January 2018 MTUS drug formulary, prescription-drug disputes fell from 50.7% of all IMRs in 2015 to 30.5% in 2025, and opioids fell from roughly a third of drug disputes to under a fifth.1,5 The rebound is everything else. Physical therapy rose from 8.3% to 14.0% of IMRs, injections from 6.9% to 12.8%, and durable medical equipment from 7.3% to 9.4%.1 Disputes did not go away when the formulary tamed the pharmacy; they migrated to the treatments that follow a denied prescription.
A note on the two data sets
CWCI counts "IMR decision letters" in its own research database. DWC counts "final determination letters" issued by the Independent Medical Review Organization. The series run parallel but are not interchangeable, and CWCI's own June 2025 release quoted the 2018 peak as 184,385 while its other releases say 184,735. We use CWCI for trend and mix, DWC for the official annual totals, and we never splice them together.
2. Nine out of ten denials are upheld — every single year
This is the number every treating physician should internalize. In every year from 2015 through 2025, IMR reviewers upheld the UR physician's modification or denial between 88% and 92% of the time. In 2025 the uphold rate was 89.8% by CWCI's count and, by DWC's, the overturn rate was 10.2%, down from 12.7% in 2024.1,2,3 A decade of regulation, guideline updates, formulary revisions and legislative hearings has moved that figure by only a few points.
The uniformity is partly structural. SB 863 built IMR to be patterned after the Department of Managed Health Care's model, with determinations that are binding on all parties "absent clear and convincing evidence of fraud or conflict of interest."6 Both the UR reviewer and the IMR reviewer are required by regulation to apply the same evidence hierarchy: first the MTUS, then the most current ACOEM or ODG guideline, then other nationally recognized guidelines, then peer-reviewed literature.7 Two physicians applying the same guideline to the same record will usually reach the same conclusion. IMR is a second reading of the file, not a second opinion about the patient.
IMR is a second reading of the file, not a second opinion about the patient. If the file did not support the request when UR read it, it will not support the request when Maximus reads it.
That is why the first lesson of the data is counterintuitive to physicians who think of IMR as "the appeal": the appeal is almost never where providers win. The request for authorization is.
3. Where the overturns actually happen
The 10% is not evenly distributed. DWC's 2025 report shows that the categories most likely to be overturned were specialized programs such as functional restoration (18.5%), behavioral and mental health services (18.3%) and evaluation and management visits (16.8%).2 CWCI's category view for the same year is consistent: evaluation and management disputes were upheld only 79.9% of the time, while acupuncture (93.9%), prescription drugs (91.2%), opioids (95.3%), hypnotics (97.7%) and musculoskeletal drugs (98.6%) were upheld at rates that approach certainty.1
| Category (2025) | Uphold rate | What it implies for the RFA |
|---|---|---|
| Evaluation and management | 79.9% | Denials of visits and consults are the most fragile; well-documented requests are worth appealing |
| Programs (functional restoration) | 81.5% | Multidisciplinary programs with documented goals and prior conservative care get a real hearing |
| Behavioral / mental health | 81.7% | Psych treatment is the category where documentation gaps are most often the reason for denial |
| Prescription drugs (all) | 91.2% | Formulary status decides most drug disputes before any reviewer reads the note |
| Acupuncture | 93.9% | Guideline limits on visits are applied mechanically; functional improvement must be shown |
| Opioids | 95.3% | The 2024 MTUS opioid guideline is applied strictly; weaning plans and risk tools must be in the record |
| Hypnotics / sleep aids | 97.7% | Effectively a closed category under current guidelines |
| Musculoskeletal drugs | 98.6% | Muscle relaxants beyond the acute window are nearly never overturned |
Uphold rates from CWCI's 2015–2025 series; program and behavioral-health rates derived from DWC's reported 2025 overturn percentages.1,2
The pattern is not random. The categories that get overturned are the ones where the medical-necessity question depends on the clinical narrative — functional status, prior treatment response, psychological factors, the reason this patient needs this visit now. The categories that are almost never overturned are the ones where the guideline has a bright line — a maximum number of sessions, a formulary status, a duration limit — and the record either meets it or does not. A treating physician's effort should be allocated accordingly: invest the documentation time in the narrative-dependent requests, and do not spend staff hours appealing a non-exempt drug that the formulary has already decided.
4. One percent of physicians, forty-two percent of disputes
CWCI has reported physician concentration in every annual release, and the number is remarkably stable. In 2025 the top 1% of requesting physicians — 81 doctors — accounted for 42% of all disputed treatment requests, and the top ten physicians alone accounted for 11%.1 In 2023 the top 10% of physicians generated 82.9% of disputes.8 Seven of the ten highest-volume physicians in one year appeared on the prior year's list; of the 2023 top ten, nine were pain-management specialists.5,9
The geography is concentrated too. CWCI's 2026 release highlights that the Bay Area accounted for 15.7% of statewide claims but 22.9% of IMR disputes.1 DWC's regional breakdown for the fourth quarter of 2025 shows Los Angeles with 8,008 of 37,872 regional cases, with 88.4% upheld.2
Why does this matter to a physician who is nowhere near the top 1%? Because the data is public, the same physicians recur year after year, and the April 2026 regulations now give claims administrators an explicit, numeric trigger for treating a physician as a pattern-and-practice problem (see section 6). Claims administrators and utilization review organizations profile requesting physicians the way lenders profile borrowers. A practice's RFA history — completeness, denial rate, how often it files IMR and how often it loses — is a reputation that follows every subsequent request from that office. Most physicians have no idea what theirs looks like, because the data lives in a dozen portals and fax logs.
5. The provider-side numbers look different from the payer-side numbers
The most-cited system-wide approval figure is CWCI's 2019 study, which found that 94.1% of medical services requested for California injured workers were approved or approved with modification, 93.9% before IMR, with approval by service type ranging from 78.6% for injections to 99.7% for evaluation and management.10 RAND's July 2026 study of SB 1160 reinforced that picture for early care: UR approval rates in its sample "consistently exceeded 90 percent during the first 30 days after injury both before and after SB 1160," with modest gains in time to first physical therapy (13.4 to 11.9 days) and a 13% increase in the odds of PT within 30 days.11,12
Provider-side data tells a less comfortable story. daisyBill, a billing platform that sees RFAs from the physician's side, reported that for January through July 2024 its clients submitted 135,308 RFAs containing 267,317 treatment requests, that only 42% of those requests had a UR decision posted at all, and that among posted decisions 71% were approvals and 22% were denials — with denial rates ranging from 3% to 46% depending on the claims administrator.13 Its 2023 sample showed the same 70% approve / 22% deny split.14 This is vendor data from one platform's clients, not a representative sample, and it should be cited as such. But the gap between a 94% approval rate measured from the payer's files and a 71% approval rate measured from the provider's files is not a rounding error. It is a measurement of how many requests never get a clean, trackable decision — the ones that are returned as incomplete, lost between fax and portal, deferred, or simply never answered.
RAND reached a related conclusion from a different direction. It recommended that DIR enforce public posting of UR plans and build a "systemwide UR database," because the data mandated by SB 1160 in 2016 and SB 537 in 2019 was never collected.11,15 Until that exists, the only complete record of what happened to a practice's requests is the one the practice keeps itself.
6. What changed on April 1, 2026
After two years of rulemaking — a 45-day comment period that closed in July 2024 and three 15-day periods in 2025 — the Office of Administrative Law approved DWC's overhaul of the utilization review regulations on December 30, 2025, effective April 1, 2026.16,17 The regulations implement SB 1160's 30-day exemption and AB 1124's exempt-drug formulary and rewrite sections 9792.6 through 9792.12 and the IMR sections that follow. The provider-relevant changes:18,19
- The 30-day exemption has an explicit exclusion list. Treatment in the first 30 days after injury by a network physician that is consistent with the MTUS is not subject to prospective UR — except non-exempt pharmaceuticals, non-emergency surgery, psychological treatment, home health, imaging other than X-ray, durable medical equipment over $250, electrodiagnostic testing and spinal injections. Requiring prospective UR inside the exemption window now carries a $3,000 penalty.
- Pattern-and-practice removal has numbers. A physician can be stripped of the exemption after 20 separate, unrelated recommended services across 10 or more injured workers in three months, or 8 separate services involving 2 or fewer workers within a month.
- Non-exempt formulary drugs get a hard five-business-day clock that cannot be extended, and formulary-only disputes must be filed for IMR within 10 days of the UR decision (reduced from 30), with records due within 10 days and an IMRO determination within five business days.
- "Complete RFA" is now defined by receipt. Receipt of the requested missing information renders an RFA complete; an incomplete RFA must be accepted or returned as not complete within five business days; the PR-1 form became optional; and claims administrators may no longer defer an RFA marked "Resubmission – Change in Material Fact."
- Denial letters must explain. A denial must give a "clear, concise and appropriate explanation in plain language" and identify the information that was missing.
- Timeliness penalties are real. Untimely prospective or concurrent UR is $250 per day up to $5,000; untimely expedited review is $250 per hour up to $18,000; a non-physician denial is $25,000; and the 85% investigation pass threshold that used to waive penalties has been eliminated.20
- Records and reports move electronically. Records must go to a new primary treating physician within 20 days, and PTP reports may be sent by encrypted email or electronic data interchange.
Read together, the new rules tighten both sides. Claims administrators face mandatory, non-waivable penalties for late or defective decisions. Physicians face a numeric pattern-and-practice trigger, a non-extendable drug clock, and a 10-day filing window that an office running on paper will miss.
7. The timelines that decide most disputes
Most RFA failures are procedural, not clinical, and the procedure is in 8 CCR §9792.9.1.21 A prospective or concurrent decision is due within five business days of receipt of a complete RFA; an expedited decision within 72 hours; a retrospective decision within 30 days. If the reviewer needs more information, the request must go out within five business days, and if the information is not received within 14 days the reviewer must deny "with the stated condition that the request will be reconsidered upon receipt of the information." A denial then stays effective for 12 months absent a documented change in material facts. A fax received after 5:30 p.m. is deemed received the next business day.
Two Appeals Board decisions shape how this plays out. Under the Dubon line of cases, an untimely UR determination is invalid, and the question of medical necessity goes to the workers' compensation judge rather than IMR — but the applicant still bears the burden of proving necessity with substantial evidence; late UR does not mean the treatment is "deemed authorized."22 And in Smith v. Plant Construction, the Board rescinded a UR denial issued on an RFA that lacked the physical examination, MRI results and prior-treatment history, holding that the UR process "requires a good faith effort by both the treating physician and the UR physician to assure that the necessary and appropriate information is available."23
That sentence is the whole provider-side lesson in one line. The physician who sends an RFA without the exam, the imaging and the prior-treatment history has not made a request that UR can approve, and has not made a record that IMR can overturn.
8. The guidelines keep moving
Because UR and IMR both begin with the MTUS, a physician whose template language was written against a guideline that has since been replaced is arguing from a document the reviewer is no longer allowed to use. The pace of change has accelerated. DWC adopted a new opioid guideline on March 27, 2024; a cannabis guideline on June 1, 2025; shoulder, elbow, hand/wrist/forearm and traumatic brain injury guidelines on January 2, 2026; PTSD and acute stress on April 1, 2026; eye disorders, chronic pain and initial approaches to treatment on June 1, 2026; and ankle/foot and hip/groin on August 1, 2026, with a knee update proposed in September 2026.24,25 The MTUS drug list reached version 14 on April 30, 2026.26 An orthopedic or pain practice has therefore had the guideline for most of its common requests replaced within the last 24 months.
9. Six lessons the data supports
Win at the RFA, not at IMR
With a 90% uphold rate for a decade, the economics are unambiguous. The marginal hour spent making the request complete — specific CPT codes, dates, the exam, the imaging, prior conservative care and its result, functional goals, the guideline citation — returns far more than the marginal hour spent on an appeal of a thin request.
Match the documentation to the category
Narrative-dependent categories (E/M, programs, behavioral health) are where a strong clinical story changes outcomes. Bright-line categories (formulary drugs, acupuncture counts, hypnotics) are decided by the rule; know the rule before requesting.
Know your own numbers before the payer does
Track every RFA from submission to determination: date received, completeness, days to decision, outcome, IMR filed, IMR result. Compare your denial rate and completeness rate by claims administrator and by request type. The concentration data shows that payers already know who the outliers are; the practice should know first.
Treat timeliness as evidence
A fax confirmation, a portal timestamp or an EDI acknowledgement is the only proof that the five-day clock started. daisyBill's missing-UR guidance — call, document the reason for the delay, keep the receipt, resubmit, report — is sound, and the new $250-per-day penalties make the record worth keeping.27
Run the clocks that the April 2026 rules created
A ten-day IMR window for formulary disputes, a five-day non-extendable drug decision and a pattern-and-practice counter are all dates that an office cannot manage on sticky notes. They need to live in the system that staff use every day, with the status returned to the chart.
Cite the guideline that is in force today
Every RFA template and every physician's standard language should be checked against the current MTUS adoption dates. The search sequence in §9792.21.1 is the reviewer's roadmap; a request that walks the same path arrives at the same destination.
10. A note on artificial intelligence in UR
Physicians increasingly ask whether their denials are being generated by software. The honest answer is that California workers' compensation UR has always required that only a physician may modify or deny a treatment request — the rule is in 8 CCR §9792.6.1 and was reinforced in the 2026 regulations.28,19 SB 1120, the 2024 "Physicians Make Decisions Act," added explicit limits on AI in utilization management, but it amends the health-plan and insurance codes governing DMHC-licensed plans; whether and how it reaches Labor Code UR is disputed among commentators, and we have found no DWC guidance specifically on AI in workers' compensation UR.29,30 SB 636, which would have required California-licensed UR physicians starting July 2026, was vetoed by the Governor, who cited "the lack of data to warrant the changes."31 DWC's 2025 IMR report notes that 85% of IMR cases are reviewed by California-licensed physicians.2
The practical implication for providers is the same regardless of how the legislation settles. Software on the payer side reads the request against the guideline in seconds. The only effective response is a request that was built to be read that way — complete, structured, guideline-cited, and timestamped — before it leaves the practice. That is the problem PhyPal's Authorization Orchestration is designed to solve: not replacing the physician's judgment, but making sure the request that carries it arrives complete, on time, and with its status returned to the workflow where the care team can see it.
This article is general information for clinicians and workers' compensation professionals, not legal advice. Statistics are quoted as published by the cited sources; regulations and decisions are cited as of October 2026. Confirm current text before relying on them.
References
- California Workers' Compensation Institute, "Independent Medical Review Trends and Outcomes: 2015–2025," press release, March 23, 2026. cwci.org
- California Division of Workers' Compensation, "Independent Medical Review Annual Report" (analysis of 2025 data), DIR Newsline 2026-43, May 19, 2026. dir.ca.gov · Report PDF
- DWC, IMR report for 2024 data, DIR Newsline 2025-36, April 8, 2025. dir.ca.gov
- DWC, IMR report for 2023 data, DIR Newsline 2024-48, June 20, 2024. dir.ca.gov
- CWCI, "CWCI Finds California Workers' Comp Independent Medical Reviews Are Trending Up," Bulletin 25-09, June 26, 2025. cwci.org
- Assembly Floor Analysis, SB 863 (De Saulnier), August 30, 2012. leginfo.ca.gov
- 8 Cal. Code Regs. §9792.21.1, Medical Evidence Search Sequence. dir.ca.gov
- CWCI, Quarterly Summary Q1 2024, July 16, 2024. cwci.org
- CWCI, "CWCI Finds IMR Volume Increased in First Half of 2023," Bulletin 23-11, July 31, 2023. cwci.org
- CWCI utilization review approval study, reported in Insurance Journal, "CWCI: 94% of California Workers' Comp Medical Services Approved," October 29, 2019. insurancejournal.com
- Rennane S., Heins S.E., Birnbaum J.S., Forbes M., Hubble M.W., Dworsky M., Examining the Impact of Senate Bill 1160 on Utilization Review and Medical Treatment in California Workers' Compensation, RAND RRA4256-1, July 20, 2026. rand.org · DIR Newsline 2026-67, August 19, 2026. dir.ca.gov
- Dworsky M., "Findings from Recent RAND Studies on Medical Access, SB 1160 (Utilization Review), and Qualified Medical Examiners," CHSWC meeting, August 12, 2026. dir.ca.gov (PDF)
- daisyBill, "daisyData: California RFA and Utilization Review 2024 Data," August 8, 2024 (vendor data). blog.daisybill.com
- daisyBill, "UR Data 2023," August 15, 2023 (vendor data). blog.daisybill.com
- daisyBill, "RAND to Study California UR," April 30, 2025 (commentary). blog.daisybill.com
- DWC, Utilization Review rulemaking index (2024–2025). dir.ca.gov
- DIR Newsline 2025-125, "Office of Administrative Law approves DWC's proposed utilization review regulations," December 30, 2025. dir.ca.gov
- Sullivan on Comp, "Special Report: DWC Overhauls Utilization Review Regulations," March 12, 2026. sullivanoncomp.com
- Enlyte, "2026 California Utilization Review Regulatory Update," January 12, 2026. enlyte.com
- 8 Cal. Code Regs. §9792.12, Administrative Penalty Schedule for Utilization Review Violations. dir.ca.gov · DIR Newsline 2026-06, UR investigation results for 2025, January 16, 2026. dir.ca.gov
- 8 Cal. Code Regs. §9792.9.1, Utilization Review Standards — Timeframe, Procedures and Notice. dir.ca.gov
- Dubon v. World Restoration, Inc. (2014) 79 Cal.Comp.Cases 1298 (WCAB en banc), as applied in Korn v. Entertainment Partners; LexisNexis case note. lexisnexis.com
- Smith v. Plant Construction (WCAB 2014); LexisNexis, "WCAB Provides Guidance on Addressing Insufficient Utilization Review Documentation." lexisnexis.com
- DWC, Medical Treatment Utilization Schedule, current guideline effective dates. dir.ca.gov
- DIR Newsline 2026-74, proposed ACOEM Knee Disorders Guideline, September 15, 2026. dir.ca.gov · DIR Newsline 2026-40, proposed Ankle/Foot and Hip/Groin guidelines, May 5, 2026. dir.ca.gov
- DIR Newsline 2026-27, MTUS Drug List v.14 effective April 30, 2026. dir.ca.gov
- daisyBill, "California Workers' Comp: What to Do About Missing UR," June 6, 2023 (updated May 14, 2025). blog.daisybill.com
- 8 Cal. Code Regs. §9792.6.1, Utilization Review Standards — Definitions. dir.ca.gov
- Sheppard Mullin, "California Limits Health Plan Use of AI in Utilization Management," November 7, 2024. sheppard.com
- Healthesystems, "California Enacts AI Disclaimer Bill and AI Utilization Review Bill," October 7, 2024. healthesystems.com · Ethos Risk, "AI in Utilization Review: Current Uses and Regulations," June 17, 2026. ethosrisk.com
- SB 636 (Cortese, 2024), veto message, September 2024; Boxer & Gerson summary. boxerlaw.com · daisyBill, "SB 636 Demands CA-Licensed UR Physicians," September 17, 2024. blog.daisybill.com
Newsletter
Get the next article in your inbox.
Occasional, substantive pieces on Workers' Compensation workflow, medical-legal evidence and provider operations. No marketing filler, and we never share your address.