Data Changelog
This log lists material corrections and surface changes PlainDoctor applied after publication, including the 2026-08-09 removal of noindex /state, /hospital, /search, and DEAD research URLs from advertised sitemaps, the 2026-08-08 provider-page 503 fix, and the 2026-08-07 honest-org identity cleanup. Each entry is past-tense and tied to a repository commit; we do not invent rows to pad the list. How to report an issue is on the editorial & corrections policy and the typed corrections pathway. This public log starts on 2026-08-06. Data current as of 2026-09-28.
-
· Correction
Registry date: the footer, the home page, the statistics, taxonomy and guide pages and the dentist and pharmacist license pages called Aug 27, 2026 the NPPES file date. That is the day we loaded the file; the file itself is dated Aug 9, 2026, and every page now says so. The methodology page gives both dates.
-
· Data refresh
Provider records now show when the provider last updated their NPPES record, the middle name, the practice fax, Medicaid identifiers, Medicare enrollment and assignment, medical school and graduation year and telehealth (CMS NPPES and Doctors and Clinicians files). Every page now gives the registry date one way: the NPPES file of Aug 9, 2026. A new page explains how to check the OIG and SAM exclusion lists; PlainDoctor shows no individual exclusion entry. The NPI lookup shows how many providers match, a numbered pager, and says when a 10-digit number is not a valid NPI.
-
· Correction
Provider pages no longer show a statewide count of board disciplinary actions next to a named clinician, and a page whose provider type does not prescribe (for example a behavior technician) now says that Part D claims listed under its NPI may be misattributed, without comparing them. Each provider page links how to correct, restrict or remove it. Analytics now receive the previous page address without its search part, and the privacy policy says what advertising scripts and error monitoring can see.
-
· Correction
Named rankings and personal data, after a claims and privacy review. The lists of prescribers ranked by opioid share are gone: near the reporting floor every entry read 100%, and some named people whose registered taxonomy cannot prescribe; opioid figures stay at state and national level from CMS. Drug-cost totals that are mostly vaccines are now flagged, since pharmacies bill them under a standing order naming one clinician. The MIPS final score is no longer called a "quality score". Open Payments summaries now say the most common kind of payment is counted by number. The privacy policy now lists every service the pages load, says the site is run from the European Union, sets analytics cookies in the EU only after consent, keeps searched names out of analytics, and explains how a clinician can correct, object to or remove their page (answered within one month).
-
· Correction
The behavior technician study compared 2025 with 2015, but NUCC created the behavior technician code in July 2016, so the 2015 figure counted only people who switched to the code later. The headline now compares with 2017, the first full year of the code: 4.5% of new individual NPIs then, 22.5% in 2025.
-
· Data refresh
Two studies moved to the newest CMS year. The MIPS study now covers performance year 2024 (CMS published it on Aug 18, 2026; the page had kept 2023 and called it the latest) and adds every state and specialty; the Open Payments study now covers program year 2025 (published Jun 30, 2026) with 2024 beside it. The old addresses redirect. Provider pages still show MIPS 2023 and Open Payments 2024, labelled with their years, until the next database load.
-
· Correction
The behavior technician study gave two figures for 2015 (452 and 374; 452 is right) and 45,478 for a sum of 45,477. The Open Payments study quoted specialty totals from before the Sep 28 taxonomy correction while its table showed the corrected ones; its text now reads the same rows as the table. The statistics page called 979,136 paid clinicians "physicians"; they include nurse practitioners and physician assistants. The brand-name drug study now names its coverage (prescribers whose brand and generic split CMS reports) in its headline and lists all 52 jurisdictions.
-
· Correction
Page dates: the "modified" date in page data and the Last-Modified header followed our processing runs, so every page said it changed that day. They now show when the records last changed (the NPPES file of Aug 9, 2026, loaded Aug 27 and its Sep 28 primary-taxonomy correction), and trust footers show the data file date. The NPI lookup now asks the live NPI Registry when an exact NPI is newer than our copy, and state pages lead with physician specialties.
-
· Correction
The opioid prescriber study and the prescriber rankings counted prescribers with an opioid claim from the per-prescriber Part D file, where CMS leaves counts of 1 to 10 blank, so they reported 32.36% nationally instead of 57.09%. Both now use CMS's Part D Opioid Prescribing Rates by Geography file: 808,192 of 1,415,649 prescribers wrote an opioid in 2024; Mississippi is highest (67.05%) and Puerto Rico lowest (45.82%) among jurisdictions with 1,000 or more prescribers. Our earlier same-day correction naming Kansas was itself wrong.
-
· Correction
The opioid prescriber study showed figures from the 2023 Medicare Part D file under a 2024 label (for example Mississippi, 10,451 prescribers and 41.56%). It now reads the 2024 file: Kansas has the highest share, 41.79%, Mississippi 41.17% (10,881 prescribers) and the District of Columbia the lowest, 18.68%. Its links to state pages that did not exist were replaced.
-
· Index-surface
The research studies and the 52 state pages can be indexed again. Google is asked to index 154 pages (hubs, guides, studies, rankings and the state pages) that add comparisons or tools to the federal records; the other record pages stay available and indexable by Bing.
-
· Index-surface
Dropped hard-noindex /state and /hospital hubs from sitemap-static and sitemap-tier1 so crawlers stop being invited to empty URL_BLEED landing pages while disciplinary state pages stay advertised.
-
· Index-surface
Removed DEAD research hub and studies from sitemap-tier1 after they were already noindexed, so empty research URLs were no longer re-advertised through the tier-1 shard.
-
· Index-surface
Dropped the noindex /search route from sitemap-tier1 so the tools search surface stopped advertising a page Google was told not to index.
-
· Data refresh
Switched static and ranking sitemap lastmods to shard-aware page-vintage stamps so crawl dates reflect real page edits instead of a single portal-wide extract date.
-
· Reliability
Pre-warmed specialty peer cohorts on boot, persisted the query-cache volume across recreates with a stable PORTAL_SLUG key, and raised the Titan allotment to 2 CPUs / 2 workers after a /provider crawl storm load-shed homepage requests into 503s.
-
· Reliability
Reworked provider-detail specialty cohort queries to a single cached fetch after live provider pages returned HTTP 503 under CPU saturation.
-
· Editorial correction
Removed fabricated “Editorial Team” identity copy and aligned publisher attribution to the honest solo-operated disclosure on accountability pages (AR-29).
-
· Index-surface
Replaced soft-404 redirects with real HTTP 404 responses so empty or retired URLs stop looking like valid indexable pages.
-
· Index-surface
Noindexed the DEAD_CLASS states hub and emptied its sitemap after admission evidence showed the class was not earning valid indexation.
-
· Data integrity
Hardened ETL so a full rebuild refuses to destroy unrepopulatable rows, protecting Open Payments, affiliations, and other stages that cannot be rebuilt from a single NPPES extract.
Download the specialty-by-state provider counts cited on this page: plaindoctor-specialty-by-state.csv (CMS NPPES · public domain). When reusing the table, name PlainDoctor as the compiler.