# Medical record review statistics 2026 > 74% of 328 vendors won't say how long record review takes. EHR duplication, physician hours and adjuster caseload data, every figure sourced. Canonical page: https://medrecords.ai/reports/medical-record-review-statistics/ --- New — **Missing Records Detection:** flags every visit, provider, and date missing from the file. [See how →](https://medrecords.ai/product/missing-records-identification/) [Home](https://medrecords.ai/) ›[Reports](https://medrecords.ai/reports/) › — Medical record review statistics MEDICAL RECORD REVIEW STATISTICS ## 74% of medical record review vendors will not say how long the work takes Medical record review statistics from 328 vendor profiles and outside research: 74% of vendors do not publish a turnaround time, half the text in the average clinical note is duplicated from something written before, and a sustainable adjuster caseload runs 110 to 140 files of exactly that kind of record. 328 vendors read — ·EHR, adjuster and physician data cited — ·Recomputed on every build ### The numbers that matter most Four figures anchor this page: how transparent the vendor market is about speed, and why the underlying record got so large that speed became the question in the first place. **74%** — of 328 vendors publish no turnaround time **91%** — publish no export format **50.1%** — of a clinical note’s text is duplicated from an earlier note **110–140** — files: a commonly cited sustainable adjuster caseload Disclosure of a turnaround time (85 of 328) is rarer than disclosure of a processing model (287 of 328) or a human-QA stance (286 of 328). Vendors will say what they do before they will say how long it takes. ### How many medical record review vendors are there, and will they say how fast they are? 328 vendors carry a profile in the Content Hub: 176 software products and 152 service providers. Read each one’s own site for a stated turnaround time and the answer is usually no. #### 85 of 328 publish a turnaround 49 of 176 software listings state a processing time; 36 of 152 service listings do. Software is slightly more forthcoming, but neither half clears three in ten. #### 30 of 328 publish an export format 26 of 176 software vendors state what file formats you get back; only 4 of 152 service providers do. Export format is the single rarest disclosure in the corpus. #### 223 of 328 carry no public reviews No G2, Capterra, or comparable listing we could find for the majority of this market, software and services alike. Reputation here still runs on referral, not on a review site. ### Turnaround disclosure by category A vendor can carry more than one category tag, so it counts in each row it appears under. The gap between the most and least transparent category is wide. | Category | Vendors | Publish turnaround | Publish export format | Turnaround disclosure | | --- | --- | --- | --- | --- | | US Med-Legal Review / Legal Nurse Consulting | 26 | 5 | 3 | 19% | | Insurance Underwriting & Payer Chart Review | 27 | 5 | 2 | 19% | | Mass Tort / Settlement / Lien | 27 | 7 | 0 | 26% | | Medical Record Retrieval / ROI | 27 | 12 | 1 | 44% | | HIM / RCM Outsourcer & LPO at scale | 29 | 1 | 0 | 3% | | AI-native / Tech-enabled service | 14 | 8 | 1 | 57% | | Long-tail / Boutique | 38 | 12 | 6 | 32% | | India-delivery BPO / LPO | 20 | 7 | 0 | 35% | | Offshore (non-India): PH / LatAm / Other | 30 | 2 | 3 | 7% | HIM / RCM Outsourcer & LPO at scale discloses turnaround least (1 of 29); AI-native / Tech-enabled service discloses it most (8 of 14). 0 of the nine largest categories have at least one vendor stating a time, so a total blackout is rare; a near-total one is not. WHY THE RECORD IS SO LARGE ### The record duplicates itself faster than any vendor can summarize it Three outside data points explain why turnaround is the hardest thing in this market to promise. - **50.1% duplicated** Half the text in the average patient’s record was copied from something already written about that same patient, per [Lin, S.Y. et al., "Duplicate Information in Electronic Health Record Notes," JAMA Network Open, 2023](https://www.ama-assn.org/practice-management/digital-health/half-words-clinical-notes-copied-pasted-prior-visits) . The share was about a third in 2015 and 54.2% by 2020. - **95% adopt an EHR** [Office-Based Physician Electronic Health Record Adoption, ONC Health IT Quick Stats, 2024](https://www.healthit.gov/data/quickstats/office-based-physician-electronic-health-record-adoption) puts EHR adoption among US office-based physicians at 95% (any EHR) and 91% (a certified EHR) as of 2024, up from a fraction of that in 2008. Copy-forward is a feature of the tool nearly every physician now uses. - **13 hours a week on documentation** ["Doctors work fewer hours, but the EHR still follows them home," American Medical Association, 2024](https://www.ama-assn.org/practice-management/physician-health/doctors-work-fewer-hours-ehr-still-follows-them-home) found physicians spend 13 of a 57.8-hour week on indirect patient care such as order entry and documentation, plus another 7.3 hours on administrative tasks, against 27.2 hours of direct patient care. ### Who is stuck reading all of it? The people closest to the record are shrinking in number while the record itself gets more repetitive, not less. #### 376,100 claims adjusters, examiners and investigators Employed in the US per [Claims Adjusters, Examiners, and Investigators, Occupational Outlook Handbook, US Bureau of Labor Statistics](https://www.bls.gov/ooh/business-and-financial/claims-adjusters-appraisers-examiners-and-investigators.htm) , at a median wage of $78,000 (May 2025). Employment is projected to decline 5% through 2035, even as about 21,600 openings are projected each year from turnover. #### 110–140 files, sustainably Commonly cited industry benchmarks put a sustainable indemnity caseload at 110 to 140 claims per adjuster, with some administrators running past 300, per ["Ideal Caseloads vs. the Throughput of Claims, Part 1," WorkCompCollege](https://workcompcollege.com/ideal-caseloads-vs-the-throughput-of-claims-part-1/) . That source cites an unnamed 2019 study rather than naming it directly; treat the range as a secondary citation. A shrinking, similarly staffed workforce reading a record that is now more than half duplicate text is the arithmetic behind the caseload debate. Neither number moving on its own would be a crisis. Both moving together is the reason record review keeps coming up as the bottleneck, not the reading itself. ### Method and sources #### What comes from the Medrecords AI corpus The vendor-market figures on this page (328 vendors, the turnaround and export-format counts, the category table) are read from the same Content Hub corpus that backs [the pricing study](https://medrecords.ai/reports/medical-record-review-pricing/) , captured 13 Aug–1 Sep 2026. Every value is what that vendor published on its own site. These numbers are recomputed on every build, so this page and the underlying profiles cannot drift apart. #### What comes from outside research The EHR duplication, EHR adoption, physician documentation, and claims adjuster staffing figures are each quoted from a single named source with its own publish date, listed below. None of these are Medrecords AI research; we did not run these studies and are not affiliated with the organizations that did. Where a source is itself secondary (the WorkCompCollege caseload figure), we say so rather than presenting it as primary. #### Sources - 1 — [Lin, S.Y. et al., "Duplicate Information in Electronic Health Record Notes," JAMA Network Open, 2023](https://www.ama-assn.org/practice-management/digital-health/half-words-clinical-notes-copied-pasted-prior-visits) Penn Medicine, 104M+ notes, ~2M patients, 2015–2020. Summarized by the American Medical Association. - 2 — [Office-Based Physician Electronic Health Record Adoption, ONC Health IT Quick Stats, 2024](https://www.healthit.gov/data/quickstats/office-based-physician-electronic-health-record-adoption) National Electronic Health Records Survey, US Department of Health and Human Services. - 3 — ["Doctors work fewer hours, but the EHR still follows them home," American Medical Association, 2024](https://www.ama-assn.org/practice-management/physician-health/doctors-work-fewer-hours-ehr-still-follows-them-home) AMA Organizational Biopsy, ~18,000 physicians across 43 states. - 4 — [Claims Adjusters, Examiners, and Investigators, Occupational Outlook Handbook, US Bureau of Labor Statistics](https://www.bls.gov/ooh/business-and-financial/claims-adjusters-appraisers-examiners-and-investigators.htm) 2025 base-year employment and wage data, 2025–35 projections, accessed September 2026. - 5 — ["Ideal Caseloads vs. the Throughput of Claims, Part 1," WorkCompCollege](https://workcompcollege.com/ideal-caseloads-vs-the-throughput-of-claims-part-1/) Secondary source: cites an unnamed 2019 national claims-management study without linking it directly. Treat the range as a commonly cited industry benchmark, not a primary figure. ### Medical record review statistics: common questions 85 of 328, or 26%. We read every vendor's own site in the Content Hub and recorded what it states about speed. The other 243 say nothing about how long a file takes, which means the only way to find out is to send one and wait. 50.1%, per a 2023 study in JAMA Network Open covering six years of documentation at Penn Medicine (104 million-plus notes, nearly 2 million patients). The share of duplicated text rose from about a third in 2015 to 54.2% in 2020, so the record a reviewer receives today is more repetitive than it was five years ago, not less. Commonly cited industry benchmarks put it at 110 to 140 indemnity claims per adjuster, with some administrators running past 300. The figure traces to an unnamed 2019 national claims-management study cited by WorkCompCollege rather than a primary source we could locate directly, so we flag it as a secondary citation. 328 in the Medrecords AI Content Hub as of 1 Sep 2026: 176 software vendors and 152 service providers, spanning IME and peer review, legal nurse consulting, retrieval, payer chart review, mass tort, and HIM/RCM outsourcing. The corpus is public at /content-hub/. No. Medrecords AI does not retrieve records from providers or facilities. You bring the records you already have — review starts in minutes from upload. Retrieval vendors take days; you can keep yours and still cut the review to minutes. Vendor-market figures are computed from the same 328-profile corpus that backs the Content Hub, captured 13 Aug–1 Sep 2026, and recomputed every time this page is rebuilt. Every other figure is quoted from a named outside source with a publish date and a link, listed in full under Sources below. ### The record is half duplicate. We do not bill for the copy. Ten cents a page, duplicates not billed, and we say how long it takes: about 100 pages a minute. Send a file and compare it with what you are reading now. [Test a file](https://medrecords.ai/test-a-file/?src=reports-medical-record-review-statistics)