Medicare’s 2015 ICD-10 flexibility forgave imprecise codes in audits, not invalid codes at claim intake
ICD-10 became mandatory on October 1, 2015. Medicare’s flexibility covered specificity denials by review contractors on physician fee schedule claims; claims without a valid code were rejected, and Medicaid, commercial payers and prior authorization were outside it.
Looking Back revisits past weeks one at a time, up to the most recent, and asks what it meant for the banks, health systems and public agencies we serve. The series was written in 2026, with the benefit of hindsight, and each article says which week it looks back on.
Two-minute summaryThe dek and the first sentence of each section, quoted from the article
ICD-10 became mandatory on October 1, 2015. Medicare’s flexibility covered specificity denials by review contractors on physician fee schedule claims; claims without a valid code were rejected, and Medicaid, commercial payers and prior authorization were outside it.
- The week On October 1, 2015, ICD-10 became the required code set for diagnoses and inpatient hospital procedures for parties covered by HIPAA.
- The boundary: what the flexibility covered That scope excluded a great deal.
- The front door: what made a code valid The clarifying answers defined a valid code.
- The workforce: documentation as the constraint ICD-10 asked for more detail, and the detail had to come from the clinical record.
- The measures: knowing whether the transition worked CMS’s Next Steps toolkit, in a copy captured on March 25, 2016, began with measurement.
- How the switch landed across the system Hospitals faced both code sets at full weight.
- Objections, and what survives them For Part B physician claims under Medicare’s review contractors, the guidance did remove a risk. [2] But the CMS answers themselves said it did not delay implementation, did not change edits or coverage policies, did not reach prepayment review or prior authorization, and did not bind Medicaid or commercial payers. [3] The routine part was where claims stopped.
- Checks for revenue-cycle and IT leaders These checks are our suggestions, drawn from the CMS documents.
- With hindsight CMS kept adjusting its transition materials after the date.
The week
On October 1, 2015, ICD-10 became the required code set for diagnoses and inpatient hospital procedures for parties covered by HIPAA. A final rule published on August 4, 2014 had set that compliance date. [1] The same rule required continued use of ICD-9-CM through September 30, 2015. [1]
The date had moved twice. The rule recounted that a 2009 final rule adopted ICD-10 with an October 1, 2013 compliance date, and that a 2012 rule moved it to October 1, 2014. [1] Then Congress acted. Section 212 of the Protecting Access to Medicare Act of 2014, enacted on April 1, 2014, said the Secretary “may not, prior to October 1, 2015, adopt ICD–10 code sets” as the HIPAA standard. [1]
The 2014 rule said HHS read section 212 as requiring a delay and giving the Secretary discretion to choose October 1, 2015 or later. It chose October 1, 2015. [1] It said stakeholders had read the law in different ways, including as allowing either code set on October 1, 2014, and that HHS recognized growing apprehension about that uncertainty. [1]
HHS issued the rule without a proposed rule first. It found good cause to waive notice and comment, saying covered entities needed to know how to proceed and a full rulemaking would not finish in time. [1] The rule estimated the cost of a one-year delay to HIPAA covered entities at $1.1 to $6.8 billion. [1]
The rule also described the Medicare program’s position. It said Medicare had programmed its claims processing systems to accept and process ICD-10 codes on October 1, 2014, and would have to reconfigure them to process ICD-9-CM claims for another year. [1] It estimated the cost of the delay to Medicare at $21 to $32 million. [1]
In the months before the deadline, CMS published guidance on how Medicare would treat early ICD-10 claims. In a set of frequently asked questions, CMS said that “for 12 months after ICD-10 implementation,” Medicare review contractors would not deny certain physician claims based solely on the specificity of the diagnosis code. [2]
The condition was that the physician or practitioner used a valid code from the right family. The same answer said a valid ICD-10 code would be required on all claims starting on October 1, 2015. [2] It named the contractors that would adopt the policy: the Medicare Administrative Contractors, the Recovery Audit Contractors, the Zone Program Integrity Contractors and the Supplemental Medical Review Contractor. [2]
On July 27, 2015, CMS issued clarifying questions and answers about that guidance. The first answer said the ICD-10 Ombudsman would be in place by October 1, 2015. [3] The second said the guidance did not mean a delay in ICD-10 implementation for Medicare or any other organization. [3]
That answer said Medicare claims with a date of service on or after October 1, 2015 would be rejected if they did not contain a valid ICD-10 code. [3] It said Medicare systems could not accept ICD-9 codes for dates of service after September 30, 2015, or claims containing both ICD-9 and ICD-10 codes for any dates of service. [3]
In the week itself, CMS’s ICD-10 web page, in a copy marked as last modified on October 1, 2015, carried a welcome blog post from the head of its Center for Medicare. [4] It announced a new resource guide and contact list to help providers find answers to ICD-10 questions, and linked the flexibility guidance and the clarifying answers. [4]
The ICD-10 transition is remembered as a coding change. The July 2015 flexibility guidance is remembered as a grace period. Both memories are a little too broad. The guidance was narrow, and the part of the transition it did not cover fell on systems and staff on the first day. [2] [3]
Our argument is that Medicare’s flexibility forgave imprecision in later review, not invalidity at the point of entry. A claim with a valid code from the right family was protected from denial on specificity grounds by the named review contractors. A claim with an invalid code was rejected. [2] [3] In our reading, that left code validity, and the people and systems that produce it, as the work that could not wait.
The distinction is easy to state and was easy to miss. The CMS answers themselves had to say that the guidance was not a delay, that coverage policies were not changing, and that other payers were not bound by it. [3] In our reading, each of those clarifications marks a place where some readers may have assumed more relief than there was.
For operations leaders, the gap between assumed and actual relief is where planning could go wrong. Staffing, vendor deadlines and cash reserves set on the belief that Medicare would forgive early mistakes could prove too thin for the mistakes it would not forgive.
We take the transition in four parts, each a different layer of operations:
- The boundary: which claims, which reviews and which payers the flexibility covered.
- The front door: what a valid code meant, and how claims systems enforced it by date of service.
- The workforce: what coders, clinicians and billing staff had to change, and where documentation became the constraint.
- The measures: how an organization could tell whether the transition was working.
The parts do not overlap. The first is scope, the second systems, the third people and the fourth evidence. After them we look at how the change landed on hospitals, practices, health plans and state Medicaid agencies, set out the strongest objections and end with checks for revenue-cycle and IT leaders.
The boundary: what the flexibility covered
Part B physician fee schedule claims only
The flexibility applied to physician or other practitioner claims billed under the Medicare Part B physician fee schedule. [2] The clarifying answers confirmed that all services paid under that fee schedule were covered, including technical-component-only and global claims. [3]
That scope excluded a great deal. The answers said the guidance applied only to physician or other practitioner claims billed under that fee schedule. [3] Hospital inpatient claims, which carry ICD-10-PCS procedure codes, are facility claims rather than physician fee schedule claims, and fell outside it. [1] [3] In our reading, a hospital revenue-cycle team had little to gain from the guidance for its facility claims.
For physician practices, the scope was a real cushion. In our reading, for health systems that employ physicians and also bill facility claims, it meant two different risk profiles for the same patient encounter. The professional claim had the cushion; the facility claim did not. [2] [3]
Review contractors, not edits
The guidance spoke of denials by review contractors “through either automated medical review or complex medical record review.” [2] It did not change the claims processing edits.
The clarifying answers were direct about this. They said a claim may still be denied because the code was not consistent with an applicable policy, such as a Local or National Coverage Determination. [3] The reason they gave was that current automated claims processing edits were not being modified as a result of the guidance. [3]
The answers also said coverage policies that required a specific diagnosis under ICD-9 would continue to require a specific diagnosis under ICD-10. They noted that laterality, which ICD-9 lacked, was the one area where policies could require more, and that coverage policies with left, right or bilateral codes did not allow an unspecified side. [3]
Post-payment, not prepayment
A later answer narrowed the relief again. Asked whether the flexibilities extended to fee-for-service prior authorization requests, CMS said no: the audit and quality program flexibilities pertained only to post-payment reviews. [3]
It said ICD-10 codes with the correct level of specificity would be required for prepayment reviews and prior authorization requests. [3] In our reading, that answer moved some of the risk forward in time. Services needing prior authorization had to be coded precisely before they were delivered.
For a scheduling or utilization team, this mattered on the first day. An authorization request for an October procedure, submitted with an imprecise code, could not rely on the guidance. [3]
Medicare only
The clarifying answers said the guidance applied only to Medicare fee-for-service claims under the Part B physician fee schedule, and not to claims for beneficiaries with Medicaid coverage, primary or secondary. [3]
They said each commercial payer would have to decide whether to offer similar audit flexibilities. [3] For state Medicaid programs, they said Medicaid could deny claims based on system edits that indicate a diagnosis code is not valid. [3] They said federal matching funds would not be available for payments not supported by valid, billable ICD-10 codes. [3]
In our reading, a provider with a mixed payer base faced a patchwork. The same code on the same day could be protected from a specificity denial by Medicare’s review contractors, and receive no such protection from a commercial plan or a Medicaid program.
Quality reporting
The guidance had a second branch for quality programs. It covered program year 2015 and three programs: PQRS, the Value Based Modifier and Meaningful Use. [2] CMS said its quality data review contractors would not penalize physicians or eligible professionals during verification or auditing related to the added specificity of ICD-10 codes, if the code came from the correct family. [2]
It said an eligible professional would not be penalized if CMS had difficulty calculating quality scores because of the transition. [2] That branch addressed a different risk from claims payment. Quality measures depend on diagnosis codes to define who is in a measure’s population, and a change of code set can move patients in or out.
A cushion with an end
The claims flexibility was stated as lasting “for 12 months after ICD-10 implementation.” [2] For quality reporting, CMS said it would continue to monitor implementation and adjust the timeframe if needed. [2]
A cushion with a fixed term is a planning input. In our reading, the right use of it was to measure where imprecise codes were being protected and fix the documentation behind them before the term ran out. An organization that treated the cushion as permanent would meet the specificity risk later, with less warning.
That is why the measures in the fourth part of this article matter. Without a count of unspecified codes and specificity-related findings, a provider could not tell how much of its revenue the cushion was holding up. [5]
The front door: what made a code valid
Full length, every time
The clarifying answers defined a valid code. ICD-10-CM codes have three to seven characters. A three-character code may be used only if it is not further subdivided. To be valid, a code must carry the full number of characters required for that code, including a seventh character if applicable. [3]
They gave an example. E10, Type 1 diabetes mellitus, is a category title, not a valid code. E10.21 and E10.9 are valid codes within it, of five and four characters. [3] They said many people use the term billable codes to mean valid codes. [3]
They defined a family of codes as the three-character category, and said one must report a valid code and not a category number. [3] In our reading, the flexibility’s phrase “a valid code from the right family” therefore imposed two tests. The category had to be right, and the code had to be complete.
Rejection versus denial
The answers drew a line that matters for operations. A claim without a valid code would be rejected; a claim with a valid code that failed a coverage policy or edit could be denied. [3] They said submitters would know which had happened. [3]
The two outcomes run through different workflows. CMS told submitters to follow existing procedures for correcting and resubmitting rejected claims, and for issues related to denied claims. [3] CMS’s later toolkit drew the same distinction in its measures, separating claims rejected at payer front-end edits from claims accepted into adjudication and then denied. [5]
In our reading, the guidance protected only part of the denial path. It did nothing for rejections. A practice whose systems produced category codes or truncated codes would see rejections, not protected denials.
Date of service decides the code set
The clarifying answers said Medicare systems could not accept ICD-9 codes for dates of service after September 30, 2015, nor claims containing both code sets for any dates of service. [3] The date of service, not the date of billing, decided which code set applied.
CMS’s 2016 web page put it in two lines. [6] Code services provided on or after October 1 with ICD-10. Code services provided before October 1 with ICD-9, even if the claim is submitted after October 1. [6] It said separate guidance existed for claims that span the compliance date. [6]
For billing systems, that rule meant two code sets in production at once for weeks or months, selected by a date on each line. Work queues that held September services after October 1 needed ICD-9; new services needed ICD-10. In our reading, the switch was a date-driven branch in every system that produced or checked a claim.
Systems in the chain
The claim passes through several systems before it reaches a payer: the electronic health record, the practice management or patient accounting system, coding tools, a clearinghouse and the payer’s front-end edits. [5] Each had to handle the new codes and the date rule.
CMS’s toolkit told providers to verify that all systems had implemented available upgrades. [5] It said systems should generate only ICD-10 codes and qualifiers for services on or after October 1, and only ICD-9 for services before. [5] It warned that issues such as outdated qualifiers and patient problem lists could stem from glitches in practice management systems, health records or coding tools. [5]
It also said clearinghouses could not identify ICD-10 codes for a provider unless they offered billing or coding services. [5] Even then, it said, they could not translate ICD-9 to ICD-10 codes without the clinical documentation needed to select the right code. [5] In our reading, that sentence marks the limit of automation in the transition. A crosswalk could suggest; only the record could decide.
Problem lists and orders
Codes appear well before the claim. They sit on problem lists, orders, referrals and authorization requests, which may be entered earlier by different people. [5] The toolkit listed the use of ICD-10 codes on prior authorizations and referrals, and incomplete or missing diagnosis codes on orders, among the measures to track. [5]
An order placed in September for a test performed in October carried a code chosen under one set and billed under another. In our reading, the transition reached every upstream system that captured a diagnosis, and the cleanest claims came from organizations that cleaned those inputs as well.
Reports that cross the date
Claims were not the only output affected. Every report that compared periods across October 2015 compared data coded in two systems. CMS’s page in the week of the switch listed general equivalence mapping files for the 2016 code sets alongside the codes themselves. [4]
In our reading, mappings were the right tool for this job. A quality trend, a population count or a payer analysis can use a mapped series if its owners know where the mapping is approximate. The risk lay in reports that crossed the date without anyone deciding how to bridge it.
For data teams, the practical step was to flag every recurring report whose logic used diagnosis or procedure codes, and to decide for each whether to map, restate or break the series at the transition.
Testing with trading partners
The 2014 rule noted that trading partner agreements had been updated to begin using ICD-10 on the earlier date, and that entities had configured and tested systems for it. [1] The delay meant re-testing with each partner for the new date.
CMS’s 2016 toolkit said many clearinghouses continued to offer ICD-10 testing to help resolve post-implementation issues. [5] In our reading, testing did not end at go-live. Each payer and clearinghouse relationship was its own interface, and a provider could pass with one and fail with another.
The toolkit’s advice to track measures separately for each payer follows from this. It said doing so helps isolate the root cause of issues. [5] A rejection spike at one payer points to that payer’s edits or the provider’s mapping to it, not to coding in general.
The workforce: documentation as the constraint
Coders and the record
ICD-10 asked for more detail, and the detail had to come from the clinical record. The clarifying answers’ cataract example shows how: category H25 contains codes that capture the type of cataract and the eye involved. [3] A coder cannot choose the eye if the note does not say which.
CMS’s toolkit told providers to look at clinical documentation for services before and after the transition, and said issues might result from insufficient clinician training on ICD-10 coding concepts. [5] It said lag times in responses to coder questions could affect days to final bill and denials for lack of medical necessity. [5]
In our reading, this is where the transition became a workforce problem. Coding staff could be trained on the new code set in advance. The detail they needed depended on clinicians, whose documentation habits were harder to change and whose time was scarcer.
Clinicians and query volume
The toolkit listed the volume of coder questions as a measure: the number of records coders return to clinicians with requests for more documentation. [5] That measure counts a cost that falls on both sides. Coders wait; clinicians are interrupted.
It suggested identifying a physician champion to help understand documentation issues, advise on implementing change with physicians, and act as a liaison between health information managers, physicians and clinical documentation improvement staff. [5]
In our reading, the champion role recognizes that documentation change is a professional conversation, not a software setting. A health system that left documentation to templates alone would see the gaps show up in coder queries and unspecified codes.
Who chooses the code
The toolkit asked providers to understand their own process for selecting diagnosis codes. It asked who selects them and who ensures selection follows the guidelines: clinicians, billers, certified coders or vendor software. [5]
The answer differs by setting and decides where training and controls belong. Where clinicians pick codes in the health record, documentation and code selection happen together. Where coders abstract from notes, the note is the constraint. Where software suggests codes, the software’s logic is part of the control. [5]
In our reading, an organization that could not answer the toolkit’s question could not target its fixes. It would train everyone a little, when the problem sat with one step in one workflow.
Seventh characters and laterality
Two features of ICD-10 created specific training needs. The first was laterality, which the clarifying answers said ICD-9 did not have. [3] The second was the seventh character for some codes, which the answers named as part of validity. [3]
The toolkit gave an example of why the seventh character mattered for hospitals. It said using a subsequent encounter code instead of an initial encounter code, for injuries and musculoskeletal conditions, results in a different diagnosis-related group. [5]
In our reading, those details explain why the flexibility could not cover hospitals’ inpatient risk. A single character could change the payment grouping, and no review-contractor policy could make that character optional.
Billing and front-office staff
The toolkit suggested tapping accounts receivable staff to track payer issues, educate coders and work with front-office staff to resolve problems. [5] Front-office staff capture referrals and authorizations; accounts receivable staff see the rejections and denials.
Those roles connect the two ends of the claim. In our reading, an organization that routed rejection data back to the people who entered the codes closed the loop faster than one that fixed rejections in the billing office alone.
The toolkit also said some providers with system issues could benefit from contracting with a clearinghouse to process claims, within the limits it set out on what clearinghouses could do. [5] That is a staffing choice as well as a technical one: it moves work to a vendor’s staff.
Productivity as a planned cost
The toolkit listed coder productivity, measured as records coded per hour and reviewed by individual coder, among its measures. [5] Listing it acknowledges that the new code set could slow coding.
The 2014 rule had described the industry’s investment before the deadline in system upgrades, testing, training and changes to workflow processes and clinical documentation practices. [1] In our reading, the productivity measure was the post-transition counterpart: a way to see whether the training had translated into throughput.
Training that had to be repeated
The 2014 rule recorded that staff had been trained on new workflow processes ahead of the October 2014 date, before the law moved it. [1] Training delivered for one date and used at a later one can fade in between.
In our reading, the delay therefore carried a hidden workforce cost: refreshing skills that had been taught for a deadline that did not happen. The rule’s cost estimate covered the delay as a whole, not this component by itself, and we do not attempt to price it. [1]
The toolkit’s training suggestions, written after the switch, pointed to official CMS resources and to low-cost training from medical societies, trade associations and payers. [5] Training was still on the list months after the date.
The measures: knowing whether the transition worked
Baselines before the date
CMS’s Next Steps toolkit, in a copy captured on March 25, 2016, began with measurement. It told providers to establish a baseline for each measure from before October 1, 2015, and compare it with data after the transition. [5]
It suggested sources for baseline data: clearinghouses, practice management systems, health records and other IT systems, as well as third-party billers and system vendors. [5] It recommended comparing metrics by month against past years, to account for seasonality such as staff vacations and flu season. [5]
In our reading, the advice to set baselines was more useful before the date than after it. An organization that captured its pre-transition measures in September could see October’s changes clearly. One that started measuring in October had to reconstruct the past.
Measures by stage of the claim
The toolkit’s measures follow the claim from service to payment. Days to final bill measured the time from service to claim. Days to payment measured the time from submission to payment. [5]
Claims acceptance and rejection rates measured front-end edits; the denial rate measured adjudication, with separate tallies suggested for authorization and coding denials. [5] Payment amounts and reimbursement rates measured what came back. [5]
For hospitals it added discharged-not-final-billed counts and diagnosis-related group volumes under ICD-9 versus ICD-10. [5] In our reading, a set of measures arranged by stage lets an organization find where a problem sits, which a single cash figure cannot.
Unspecified codes as a signal
The toolkit listed the use of unspecified codes, by volume and frequency, as a measure. [5] Unspecified codes are valid, so they pass the front door. They can still fail coverage policies that require specificity, which the clarifying answers said were not changing. [3]
In our reading, the unspecified-code rate was the best early warning that the flexibility’s cushion was being used. While it lasted, Medicare’s review contractors would not deny on specificity alone. Afterwards, the same habits would carry the risk the cushion had covered. [2]
Where to take problems
The guidance described escalation paths. CMS said it would set up a communication and collaboration center to monitor implementation, with an ICD-10 Ombudsman to receive and triage physician and provider issues. [2] The clarifying answers said the Ombudsman would be in place by October 1, 2015. [3]
The guidance also described advance payments. It said that when Part B contractors could not process claims within established time limits because of administrative problems, such as contractor system malfunction, an advance payment might be available. [2] It said this was a conditional partial payment that required repayment, and that CMS lacked authority to make advance payments where a physician could not submit a valid claim. [2]
That last limit is the argument of this article in one sentence. Relief was available for problems on Medicare’s side. A provider that could not produce a valid claim had no such relief. [2]
Cash flow as an operations measure
The toolkit suggested tracking claims by aging bands and days in accounts receivable, by payer and by dollar value. [5] Those are finance measures, but in the transition they were early signs of operational trouble.
A rise in days to final bill points at coding or documentation. A rise in rejections points at systems or validity. A rise in denials points at coverage policies and specificity. In our reading, a revenue-cycle leader who could read the three apart knew which team to send in. [5]
The 2014 rule had warned that if health plans could not process claims, provider reimbursement would be delayed, and that many small and rural providers rely on timely payment to continue in business. [1] Cash flow was the channel through which a systems problem became a care problem.
How the switch landed across the system
Hospitals
Hospitals faced both code sets at full weight. ICD-10-PCS replaced ICD-9-CM procedure codes for inpatient procedures, and their inpatient facility claims sat outside the physician fee schedule flexibility. [1] [2]
CMS’s toolkit suggested hospital chart audits. [5] It named high-risk cases, cases with a diagnosis-related group shift between ICD-9 and ICD-10, the workflow for each record type, and surgical cases needing clusters of codes. [5] In our reading, those audits were the hospital’s substitute for the cushion it did not have.
Physician practices
Practices billing the physician fee schedule had the flexibility for Medicare and none from other payers unless those payers chose to offer it. [3] CMS’s page in the week of the switch pointed small practices to an online tool built with input from providers in small practices. [4]
For a small practice, the operational risk was concentrated in its practice management system and its clearinghouse. In our reading, the practice’s vendor relationships decided its first month more than its coding knowledge did.
Vendors and clearinghouses
The 2014 rule said its cost analysis did not include software vendors for practice management and health record systems, on the ground that they pass their costs to clients. [1] Their readiness still decided whether clients could bill.
CMS’s toolkit told providers to work with vendors to resolve systems issues affecting productivity, and noted that vendor upgrades to health records could have unintended effects on ICD-10, such as software incompatibility. [5] In our reading, the vendor’s release calendar became part of each client’s compliance calendar.
Health plans
Health plans were HIPAA covered entities and had to accept ICD-10 for services on or after the date. The 2014 rule recalled that, when HHS moved the 2013 date, it had described commercial health plans, medium and large hospitals and large physician practices as far along in their implementation. [1]
The clarifying answers left each commercial payer to decide whether to offer audit flexibilities like Medicare’s. [3] In our reading, that put plans’ provider-relations teams in the position of explaining their own policy, case by case.
State Medicaid agencies
State Medicaid agencies were both payers and public bodies running claims systems. The 2014 rule said they faced costs similar to commercial health plans from the one-year delay, including contract changes and renewed testing. [1]
The clarifying answers said Medicaid programs must process claims with ICD-10 codes for services on or after October 1 in a timely manner, and could deny claims with invalid codes. [3] They said federal matching funds would not be available for payments not processed through a compliant claims system with valid codes. [3] In our reading, that tied a state’s federal funding to the readiness of its own systems.
Objections, and what survives them
“The flexibility was the point”
A reader could argue that the flexibility was the main event: it removed the largest financial risk for physicians, and the rest was routine. [2]
For Part B physician claims under Medicare’s review contractors, the guidance did remove a risk. [2] But the CMS answers themselves said it did not delay implementation, did not change edits or coverage policies, did not reach prepayment review or prior authorization, and did not bind Medicaid or commercial payers. [3] The routine part was where claims stopped.
“Crosswalks handled the conversion”
A second objection is that general equivalence mappings and coding tools made the switch mechanical. CMS published mapping files alongside the code sets. [4]
The toolkit answered this directly. It said clearinghouses could not translate ICD-9 to ICD-10 codes unless they had the clinical documentation needed to select the right code. [5] In our reading, mappings helped with reports and analytics. They could not supply laterality or encounter type that the note did not record.
“The delays had already absorbed the cost”
A third objection is that the two delays gave the industry ample time, so October 2015 was an anticlimax. The 2014 rule itself described entities that had configured and tested systems for the earlier date. [1]
Systems readiness and documentation readiness are different. The 2014 rule described investment in systems, training and documentation practices. [1] The 2016 toolkit, written after the switch, still listed documentation review, coder queries and physician champions as work to do. [5] In our reading, the delays helped the systems more than the habits.
“Quality programs were protected”
A fourth objection is that CMS shielded quality reporting as well as claims, so the transition carried little risk to quality scores. [2]
The shield was specific. It covered program year 2015 and penalties arising from the added specificity of ICD-10 codes or from CMS’s own difficulty calculating scores. [2] It did not address whether a provider’s own measure logic, registries and dashboards were updated for the new codes.
In our reading, internal quality reporting carried the same risk as claims. A measure defined by ICD-9 codes would drop patients coded in ICD-10, without any error message, unless someone rebuilt its definition.
“This is history now”
A fifth objection is that the transition ended long ago. CMS’s current page still describes October 1, 2015 as the date the industry moved to ICD-10. [7]
The same page lists code updates effective October 1, 2026 and earlier April 1, 2026 updates. [7] In our reading, the date-of-service rule and the validity rule that defined October 2015 now apply at every update. Each release is a smaller version of the same switch.
Checks for revenue-cycle and IT leaders
These checks are our suggestions, drawn from the CMS documents. They are not requirements of HIPAA or of any payer, and organizations should confirm payer-specific rules with those payers and with counsel. [3] [5]
- Code-set calendar: every system that captures or checks diagnosis or procedure codes loads each new release before its effective date.
- Date-of-service logic: claims spanning a code-set update are tested before the update, not after.
- Validity edit in-house: the organization catches incomplete or category-level codes before the clearinghouse does.
- Rejection versus denial: reports separate front-end rejections from adjudication denials, and route each to the right team.
- Unspecified-code rate: tracked by clinician and service line, with a documentation follow-up.
- Coder query volume: tracked with the time clinicians take to answer.
- Baselines: captured before any code-set or payer-policy change.
- Payer matrix: each payer’s policy on specificity, prior authorization and audit is recorded, not assumed from Medicare’s.
The 2015 guidance gave Medicare physician claims a cushion on specificity, inside a boundary the CMS answers drew carefully. [2] [3] The checks above are about the ground outside that boundary, where valid codes, clean data and clinicians’ documentation decided whether claims moved.
With hindsight
CMS kept adjusting its transition materials after the date. Its ICD-10 page, in a copy captured on October 5, 2016, said the CMS-AMA guidance and FAQs had been most recently updated on August 18, 2016. [6] The same page described the ICD-10 transition as a mandate for all parties covered by HIPAA, whether or not they bill Medicare or Medicaid. [6]
It still linked the flexibility guidance, under a note that CMS and the AMA had announced their efforts on July 6, followed by guidance and FAQs. [6] In our reading, the page treated the guidance as a live reference rather than a closed episode. [6]
That 2016 page also offered an ICD-10 Next Steps toolkit, which it said gave providers tools to assess progress using key performance indicators. [6] The toolkit, in a copy captured on March 25, 2016, opened by saying ICD-10 became effective for all HIPAA-covered entities on October 1, 2015. [5]
The toolkit’s maintenance section looked ahead. It said ICD-10 updates take place annually on October 1, following the timeline used for ICD-9 updates, and told providers to keep systems and coding tools up to date. [5] It said the ICD-10 Coordination and Maintenance Committee, led by CMS and CDC, reviews proposals for code updates at its March and September meetings. [5]
As checked on 2026-11-16, CMS’s ICD-10 page lists new procedure and diagnosis codes effective October 1, 2026, and update files dated April 1, 2026. [7] The page says to use the October 2026 procedure codes for discharges from October 1, 2026 to March 31, 2027. [7] In our reading, updates now arrive on more than one date a year, so the date-of-service logic of 2015 recurs.
The same page still says ICD-10 applies to all parties covered by HIPAA, and that the industry moved from ICD-9 to ICD-10 on October 1, 2015. [7] The United States remains on ICD-10 for these purposes, as checked on that date.
The 2016 page also listed an ICD-10 Ombudsman and ICD-10 Coordination Center among its resources, alongside Medicare fee-for-service FAQs and guidance on claims that span the compliance date. [6] The escalation paths promised before the switch were still listed on the page in that October 2016 copy. [6]
Internationally, the classification moved on. On February 11, 2022, the World Health Organization said ICD-11 was now officially in effect for recording and reporting causes of illness and death. [8] It said Member States approved ICD-11 at the 72nd World Health Assembly in 2019, and that 35 countries were using it at that time. [8] It described the eleventh revision as containing around 17 000 unique codes and more than 120 000 codable terms, and as entirely digital. [8]
What held up from October 2015 is the shape of the switch: a hard date of service, a validity test at the front door and a narrow, time-limited cushion for one program’s reviews. [2] [3] What did not hold up is the idea of a single transition. Code sets now change on a schedule, and each change repeats the 2015 work at smaller scale. [7]
In our view, the lasting lesson of the flexibility guidance is in its fine print. The relief was real and narrow, and the organizations that read the boundary closely planned for the work it did not cover.
Spotted an error? Tell us and we will check it and correct the article where needed.