ICD-10

ICD-10 is the tenth revision of the International Statistical Classification of Diseases and Related Health Problems, published by the World Health Organization (WHO). It assigns standardized alphanumeric codes to diseases, injuries, symptoms, and other health-related circumstances, providing a framework for recording and comparing morbidity and causes of death. Its categories organize health information for statistical purposes rather than supplying a complete description of every patient’s condition. The revision was approved by the World Health Assembly in 1990. (cdn.who.int)

As part of the International Classification of Diseases, ICD-10 supports epidemiological research, public health surveillance, health-service planning, and administrative systems. Its international core must be distinguished from national clinical modifications, such as the United States’ ICD-10-CM, and from separately developed procedure classifications. These adaptations apply the classification to particular healthcare settings and reporting requirements. (who.int)

ICD-10’s organization, coding conventions, and separate rules for mortality and morbidity explain how clinical documentation becomes comparable statistical information. Its history also encompasses continuing updates, national implementation, and the challenges of comparing data across revisions. Its successor, ICD-11, formally came into effect on January 1, 2022, subject to transitional arrangements; countries may continue using ICD-10 while preparing their transition. (cdn.who.int)

Development and international adoption

ICD-10 succeeded ICD-9, the revision of the International Classification of Diseases adopted in May 1976 and brought into effect internationally on January 1, 1979. ICD-9 had preserved the classification’s basic structure while adding diagnostic detail, but expanding use in hospital records, specialist statistics, and healthcare evaluation created demands for a more flexible successor. Preparation for the tenth revision had begun even before the 1975 ICD-9 revision conference. The World Health Organization (WHO) asked its collaborating centres to test alternative structures that could accommodate additional content without repeatedly reorganizing the classification. Because international consultation was lengthy and the existing ten-year revision cycle left insufficient time to evaluate each edition, member states agreed to postpone the next revision conference from 1985 to 1989. (who.int)

Preparation combined structural experiments, expert review, and circulation of draft proposals. A September 1983 preparatory meeting recommended alphanumeric coding to increase the number of available categories. WHO expert committees met in 1984 and 1987, alongside regular meetings of the heads of collaborating centres. An initial draft containing three-character category titles was circulated in August 1984 to member states, collaborating centres, nongovernmental organizations, and other interested parties; a second draft, incorporating four-character subdivisions, followed in July 1986. The second expert committee, meeting in Geneva on November 23–27, 1987, examined the revised categories, inclusion and exclusion notes, coding rules, and definitions. This consultation addressed both the allocation of space between chapters and the reservation of capacity for future expansion. (iris.who.int)

The International Conference for the Tenth Revision met at WHO headquarters in Geneva from September 26 to October 2, 1989, with delegates from 43 member states. Assistant Director-General J.-P. Jardel opened the conference, and Ron Wells of Australia was elected chairman. The proposed title, International Statistical Classification of Diseases and Related Health Problems, explicitly identified the classification’s statistical purpose and wider coverage while retaining the abbreviation ICD. The conference considered chapter content, accompanying instructions, arrangements for introduction, and relationships with other health classifications. (iris.who.int)

The principal structural innovation was the replacement of ICD-9’s numeric coding framework with categories consisting of one letter followed by two digits, with a further digit for four-character subdivisions. This more than doubled the available coding frame: each letter could accommodate 100 three-character categories. Twenty-five letters were initially used, while U was reserved for future additions and interim classifications; unused categories within chapters provided further expansion space. ICD-9’s supplementary classifications for external causes of injury and poisoning and for factors influencing health status and contact with health services became integral parts of the main classification. Separate chapters were also created for diseases of the eye and adnexa and diseases of the ear and mastoid process, bringing the initial chapter count to 21, compared with ICD-9’s 17 main chapters and two supplementary classifications. (iris.who.int)

The Forty-third World Health Assembly adopted ICD-10 in May 1990 through resolution WHA43.24; WHO records January 1, 1993 as its international coming-into-effect date. These milestones did not constitute a simultaneous changeover in every national reporting system. For example, the United States implemented ICD-10 for mortality statistics beginning with the 1999 data year, illustrating the distinction between international endorsement and operational national implementation. (who.int)

Scope and statistical purpose

ICD-10 classifies morbidity—the diseases and other health problems recorded in living populations—and mortality, principally the causes recorded at death registration. Its statistical purpose is to convert descriptions in individual health and vital records into standardized categories that can be counted and compared across places and periods. The World Health Organization (WHO), in its 2016 instruction manual, describes this process as translating diagnostic wording into alphanumeric codes for storage, retrieval and analysis. Records carrying the same classified condition can consequently be aggregated without requiring identical wording in the original documents. This provides a common framework for epidemiological statistics, rather than a separate description of each patient's illness. (cdn.who.int)

The essential mechanism is grouping. ICD-10 uses a finite set of mutually exclusive categories for classified conditions, not a separate category for every possible clinical expression. Frequently occurring diseases and conditions of particular public-health importance may receive individual categories; other related conditions are grouped together. Residual categories accommodate conditions that do not fit a more specific grouping. Mutual exclusivity concerns the allocation of a condition within the classification: it does not mean that a person can have only one recorded condition. Hierarchical subdivisions preserve some diagnostic detail while allowing records to be combined into broader, interpretable statistical totals. (icd.who.int)

Its coverage extends beyond named diseases. The 2016 classification includes symptoms, signs and abnormal clinical or laboratory findings in Chapter XVIII; injuries, poisoning and other consequences of external causes in Chapter XIX; and external causes of morbidity and mortality in Chapter XX. The last distinction separates what happened to the body from the circumstances producing the harm. Chapter XXI covers factors influencing health status and contact with health services, so a classified healthcare contact need not represent an illness. These groupings provide places for information that cannot appropriately be expressed as a confirmed disease diagnosis. (icd.who.int)

ICD-10 is a “variable-axis” classification: its organizing principle changes between groups. Cause, anatomical site and circumstances of onset all contribute to its arrangement. For example, infectious and parasitic diseases form a cause-oriented group, whereas respiratory and digestive diseases are grouped by body system. Pregnancy, childbirth and the puerperium, and conditions originating in the perinatal period, receive separate groupings based on reproductive circumstances or the period in which conditions originate. The structure retains elements of the scheme associated with William Farr, combining epidemic, general, site-specific and developmental diseases with injuries. These special groups keep statistically related conditions together rather than distributing them exclusively among anatomical chapters. (icd.who.int)

A statistical classification is therefore distinct from a comprehensive clinical terminology, a patient record and a diagnostic assessment. A terminology can distinguish clinical concepts more finely than the categories needed for statistical reporting; WHO explicitly notes that ICD does not supply sufficient detail for every specialty. A code represents selected information from documentation, not the complete history, examination, investigations or treatment contained in that documentation. Nor does assigning a category establish a diagnosis: symptoms, abnormal findings and reasons for consultation can themselves be classified. ICD-10 organizes recorded information for statistical use; the clinical assessment and the source record remain separate from that representation. (cdn.who.int)

Chapters, categories, and code structure

ICD-10 is organized hierarchically into chapters, blocks, three-character categories, and four-character subcategories. Chapters divide the classification into broad subject areas; blocks group related categories within them and are identified by code ranges rather than by a separate code assigned to an individual record. A three-character category consists of a letter followed by two digits, such as E11. Most categories are subdivided by adding a fourth, numeric character after a decimal point: E11.9 therefore contains four coding characters, with the punctuation excluded from the count. Categories may represent a single condition or a group of related conditions. The three-character level provides the core framework for international reporting, while the fourth character supplies more detailed distinctions. (icd.who.int)

The principal disease chapters combine several organizing approaches. Chapter I, A00–B99, covers infectious and parasitic diseases, while Chapter II, C00–D48, covers neoplasms, including malignant tumours and benign neoplasms. Chapter III, D50–D89, contains blood diseases and certain immune disorders; Chapter IV, E00–E90, contains endocrine, nutritional, and metabolic diseases. Chapter V, F00–F99, is devoted to mental and behavioural disorders. Chapters VI–XIV principally organize diseases by body system: nervous system (G00–G99), eye (H00–H59), ear (H60–H95), circulatory system (I00–I99), respiratory system (J00–J99), digestive system (K00–K93), skin (L00–L99), musculoskeletal system and connective tissue (M00–M99), and genitourinary system (N00–N99). Thus, a letter does not invariably identify one whole chapter: both D and H are shared between chapters. (cdc.gov)

Conditions associated with particular stages of reproduction and development occupy separate chapters. Chapter XV, O00–O99, covers pregnancy, childbirth, and the puerperium; Chapter XVI, P00–P96, covers conditions originating in the perinatal period; and Chapter XVII, Q00–Q99, covers congenital malformations, deformations, and chromosomal abnormalities. The distinction between the latter two is structural rather than simply age-based: a condition originating in the perinatal period can remain classified in Chapter XVI even when illness or death occurs later. Congenital abnormalities are assigned to their separate chapter rather than automatically being grouped with neonatal conditions. (cdc.gov)

The remaining major groups distinguish the condition recorded from its cause or the circumstances of contact with healthcare. Chapter XVIII, R00–R99, contains symptoms, signs, and abnormal clinical and laboratory findings not classified elsewhere. Chapter XIX, S00–T98, describes injuries, poisoning, and other consequences of external causes; Chapter XX, V01–Y98, classifies the external events, such as transport accidents or falls. An injury category and an external-cause category therefore describe different aspects of the same occurrence. Chapter XXI, Z00–Z99, records factors influencing health status and contact with health services, including examinations, immunization, and follow-up care, rather than necessarily identifying an active disease. (cdc.gov)

The diabetes hierarchy illustrates how these levels operate. Within Chapter IV, the block E10–E14 groups diabetes mellitus; E11 identifies type 2 diabetes mellitus; and E11.9 specifies that category without complications. The block’s shared fourth-character subdivisions distinguish, among other states, coma (.0), ketoacidosis (.1), and absence of complications (.9). These are subdivisions of the WHO classification, not examples requiring a national clinical modification. Although the basic hierarchy reaches the four-character level, WHO also provides supplementary fifth-character or subsequent subdivisions in selected areas, including anatomical sites in musculoskeletal disease and additional injury detail. Longer codes should therefore not automatically be treated as national extensions; their meaning depends on the particular classification and edition. (cdc.gov)

Reference volumes and coding conventions

The World Health Organization publishes ICD-10 in three complementary reference volumes. Volume 1, the tabular list, contains the classification: category and subcategory titles, inclusion and exclusion terms, and notes defining their scope. It also contains special statistical tabulation lists and associated definitions. Volume 2, the instruction manual, explains the classification’s structure and conventions, provides mortality and morbidity coding rules, and describes statistical presentation. Volume 3, the alphabetical index, connects diagnostic expressions with their classification positions, including many terms absent from the tabular list. Its three sections cover diseases and the nature of injuries, external causes of injury, and a table of drugs and chemicals. The volumes therefore serve different functions rather than offering interchangeable routes to a code. (icd.who.int)

Coding begins with the appropriate section of the alphabetical index, locating a lead term and checking its indented modifiers. Cross-references control the search: “see” requires consultation of another term, whereas “see also” directs further consultation when the recorded statement contains information not covered beneath the initial entry. The resulting code must then be verified in Volume 1, including applicable notes at chapter, block, category, and subcategory levels. An index entry ending in a point and dash, such as O03.–, indicates that a fourth character must be located in the tabular list; it is not a completed subcategory code. (cdn.who.int)

Inclusion terms are examples, synonyms, or boundary-defining expressions assigned to a rubric—a category or subcategory—not additional subdivisions within it. Their lists are not exhaustive. Exclusion terms identify conditions classified elsewhere despite an apparently suitable rubric title, and supply the destination code. For example, H01.0, blepharitis, excludes blepharoconjunctivitis, directing it to H10.5. Notes attached to a broader heading also govern the entries beneath that heading. (icd.who.int)

“Not otherwise specified” (NOS) means unspecified or unqualified; “not elsewhere classified” (NEC) signals that specified variants may have separate positions elsewhere in the classification. Consequently, NEC does not simply indicate missing diagnostic detail. For example, J16 covers pneumonia due to other infectious organisms not elsewhere classified, whereas J18 covers pneumonia whose organism is unspecified. Residual “other” categories accommodate specified conditions outside the named alternatives; “unspecified” categories accommodate descriptions lacking the distinctions required for a more precise allocation. At the fourth-character level, .8 generally denotes “other” and .9 “unspecified,” although the actual tabular wording governs assignment. (cdn.who.int)

The dagger-and-asterisk convention, introduced in ICD-9 and retained in ICD-10, represents two aspects of one diagnostic statement: an underlying disease and its manifestation in an organ or site. The dagger (†) marks the underlying-disease code; the asterisk () marks the manifestation code. For example, tuberculous meningitis is represented by A17.0† together with G01, placing the same condition in both its infectious-disease and neurological contexts. This is dual classification, not an ordinary subdivision or an extra character appended to one code. The dagger code is primary and must be retained; an asterisk code must never stand alone. For morbidity coding, their sequence may be reversed when the manifestation is the principal focus of care. The 2019 instruction manual identifies 83 special asterisk categories, listed at the beginnings of the relevant chapters. (icd.who.int)

Classification of causes of death

The World Health Organization (WHO) bases mortality coding on the International Form of Medical Certificate of Cause of Death. Its medical section separates the causal sequence in Part I from other significant contributing conditions in Part II. Part I places the immediate cause—the final disease, injury, or condition directly leading to death—on line (a), with the conditions that caused it on successive lines below. Intermediate causes occupy positions between the immediate cause and the condition that initiated the sequence. The certificate also records approximate intervals between each condition’s onset and death, helping to describe the chronology rather than merely listing diagnoses. Part II records conditions that contributed to the fatal outcome but were outside the direct sequence; it is not an inventory of every condition present at death. (cdn.who.int)

The underlying cause of death is the disease or injury that began the sequence leading to death, or the circumstances of an accident or violence that produced the fatal injury. It therefore differs from both the terminal condition and an unrelated contributory condition. WHO’s 2016 instruction manual illustrates a sequence comprising extensive haemorrhage, traumatic amputation of the right leg, and being run over by a streetcar. The haemorrhage is immediate, the traumatic injury is intermediate, and the accident circumstances initiate the sequence. A certificate containing only one sufficient cause needs no additional antecedent entries. Although the initiating condition normally appears on the lowest completed line of Part I, the underlying cause used in statistics is determined through coding rules, not simply by copying that line. (cdn.who.int)

WHO’s selection and modification rules standardize this determination for statistical purposes; they are not clinical diagnostic instructions. In the 2016 manual, steps SP1–SP8 establish a starting point, and steps M1–M4 examine modifications and special provisions. These procedures address accepted causal sequences, ill-defined entries, combinations of conditions, and more specific information elsewhere on the certificate. Modification can produce a statistical assignment different from the initially selected condition. For example, the manual’s special linkage instructions select myocardial infarction when it is reported as due to ischaemic heart disease. The certifier supplies the medical account; the coder applies a common classification framework to that account. (cdn.who.int)

Underlying-cause statistics assign each death to one underlying-cause category, allowing deaths to be counted without duplication across mutually exclusive categories. Multiple-cause-of-death statistics retain additional coded causes from both certificate parts, permitting analysis of conditions mentioned together and distinguishing deaths due to a condition from deaths involving it. As a concrete implementation, the United States CDC WONDER multiple-cause dataset for 1999–2020 provides one underlying cause and as many as 20 additional multiple-cause codes per death. Because one death can mention several conditions, counts of deaths involving different conditions overlap and cannot simply be added to obtain total deaths. (wonder.cdc.gov)

For published mortality tables, ICD-10 Volume 1 supplies special tabulation lists that aggregate detailed codes. Lists 1 and 2 concern general mortality; Lists 3 and 4 concern infant and child mortality. The condensed lists, 1 and 3, cover the classification through chapter groupings and residual “remainder” items. The selected lists, 2 and 4, emphasize conditions and external causes important for mortality monitoring without providing complete chapter totals. Their numbered items identify statistical groupings, not additional diagnosis codes. (cdn.who.int)

Morbidity coding and clinical documentation

For single-condition morbidity analysis, the World Health Organization defines the main condition as the condition diagnosed at the episode’s end that chiefly occasioned treatment or investigation. If several qualify, resource use determines selection. Other conditions coexist or arise during care and affect management; unrelated conditions from earlier episodes are excluded. The responsible practitioner records these distinctions. (cdn.who.int)

Documentation determines coding specificity: “acute appendicitis with perforation,” for example, conveys more than “appendicitis.” Without an established diagnosis, the main symptom, abnormal finding, or problem is recorded. WHO distinguishes this recording guidance from its coding rule: an unresolved suspected main diagnosis, without clarification, is coded as established. (cdn.who.int)

Inconsistent records should first be referred for clarification. If clarification is unavailable, rules MB1–MB5 permit reselection: MB1 addresses an incidental condition displacing the condition actually treated; MB2 addresses several main conditions; MB3 replaces a presenting symptom with its documented, treated cause; MB4 selects a more specific description; MB5 handles alternative diagnoses. WHO’s MB1 example reselects cervical carcinoma, C53.0, rather than sinusitis when the patient underwent hysterectomy. These rules select from recorded information, not new clinical diagnoses. (cdn.who.int)

For mental and behavioural disorders, WHO published The ICD-10 Classification of Mental and Behavioural Disorders: Clinical descriptions and diagnostic guidelines in 1992. Developed from Chapter V, this companion publication supplies clinical descriptions and diagnostic guidance beyond the general tabular classification. Each disorder’s entry describes its principal clinical features and less specific associated features, then indicates the number, combination, and duration of symptoms generally required for a confident diagnosis. Inclusion and exclusion criteria and differential diagnoses help distinguish neighbouring categories. The wording retains flexibility for clinical work, including situations in which information is incomplete or a diagnosis remains provisional. Its development included field testing by more than 700 clinicians and researchers at 110 institutes in 40 countries. Thus, the publication supports the clinical identification of conditions subsequently classified by ICD-10, rather than merely providing another list of category titles and codes. (who.int)

The companion Diagnostic criteria for research, published in 1993, serves a different purpose: defining sufficiently homogeneous study populations for comparable research findings. Covering more than 300 disorders derived from Chapter V, it uses deliberately restrictive criteria rather than the greater flexibility of the clinical guidelines. Lettered and numbered criteria distinguish requirements applying to an entire disorder group, obligatory criteria for a particular disorder, and additional characteristics of which only some are required; exclusion clauses appear where appropriate. The criteria were finalized after testing by clinicians and researchers in 32 countries. The clinical concepts underlying them are explained in the 1992 companion volume. These research requirements govern study inclusion, whereas the general tabular classification provides the categories used to aggregate diagnoses statistically; the two functions are related but not interchangeable. (who.int)

National modifications and implementation

National clinical modifications expand ICD-10 to meet domestic requirements for recording morbidity, adding diagnostic subdivisions and accompanying reporting rules while retaining a relationship to the international classification. They are not simply translations: the World Health Organization (WHO) authorized the United States National Center for Health Statistics (NCHS) to develop a clinical modification conforming to WHO’s classification structure and conventions, and licensed the Canadian Institute for Health Information (CIHI) to enhance ICD-10 for Canadian morbidity-data collection. National coding standards provide additional instructions for entering these more detailed classifications into local reporting systems. (cdc.gov)

In the United States, ICD-10-CM—the International Classification of Diseases, Tenth Revision, Clinical Modification—is the diagnosis classification administered by NCHS, part of the Centers for Disease Control and Prevention. Its codes contain three to seven alphanumeric characters. Two implementation dates distinguish different uses: ICD-10 entered US mortality reporting with the 1999 data year, whereas the transition from ICD-9-CM to ICD-10-CM and ICD-10-PCS for healthcare reporting took effect on October 1, 2015, and was mandatory for entities covered by the Health Insurance Portability and Accountability Act. The earlier mortality adoption therefore did not establish ICD-10-CM as the diagnosis classification for healthcare encounters. NCHS supplies fiscal-year-specific tabular lists, indexes and official coding guidelines; its browser permits selection of the release applicable to a particular date of service. (cdc.gov)

ICD-10-PCS, the Procedure Coding System, was developed separately by the Centers for Medicare & Medicaid Services for procedures performed in US hospital inpatient settings. It is neither the WHO core classification nor an extension of ICD-10 diagnosis categories. All PCS codes have seven characters; in its Medical and Surgical section, these identify the section, body system, root operation, body part, approach, device and qualifier. Thus, although ICD-10-CM and ICD-10-PCS share an implementation date and are commonly named together, they classify different aspects of an admission: diagnoses and procedures, respectively. Their structures and coding instructions are separate. (mmshub.cms.gov)

In Canada, CIHI administers ICD-10-CA together with the Canadian Classification of Health Interventions (CCI), a distinct intervention classification developed and maintained by CIHI rather than a diagnosis-code extension. Implementation began in April 2001 and proceeded by jurisdiction; all ten provinces and three territories had adopted the classifications by April 2006. Version 2022 took effect on April 1, 2022, accompanied by the Canadian Coding Standards for Version 2022 ICD-10-CA and CCI and tables identifying changes from the preceding release. That revision introduced 159 ICD-10-CA codes and disabled 74, while adding 94 CCI codes and disabling 103. The coding standards apply to datasets submitted to CIHI’s Discharge Abstract Database and National Ambulatory Care Reporting System. (secure.cihi.ca)

In Germany, ICD-10-GM—German Modification—is the official diagnosis classification for outpatient and inpatient care, issued by the Federal Institute for Drugs and Medical Devices (BfArM). It is revised annually through a proposal process and published in version-specific systematic and alphabetical directories. The 2026 edition, for example, introduced five-character subdivisions under B94.8 to distinguish sequelae of influenza and infectious mononucleosis from other specified infectious and parasitic diseases. Germany pairs its diagnosis classification with the separate Operationen- und Prozedurenschlüssel (OPS) for procedures, rather than with the US ICD-10-PCS or Canadian CCI. (terminologien.bfarm.de)

Health statistics, research, and administration

ICD-10 supplies common categories for comparing recorded causes of illness and death across populations and for organizing public health surveillance. The World Health Organization Mortality Database compiles national civil-registration data by country or area, year, age, sex, and cause of death; its detailed files include the underlying cause coded by national authorities. These categories permit cause-specific comparisons rather than comparisons based on locally worded diagnoses alone. The database contains registered deaths, however, and is distinct from estimates that compensate for incomplete registration or missing cause information. ICD categories standardize the classification of reported events; they do not themselves supply population denominators or correct registration gaps. (who.int)

Surveillance systems also select groups of ICD-10 codes to monitor particular health threats. In the United States, the Centers for Disease Control and Prevention uses multiple-cause death-certificate codes to identify deaths associated with influenza (J09–J11), pneumonia (J12–J18), and COVID-19 (U07.1). Including contributing causes captures a different set of deaths from selecting only the underlying cause. This is a coded-record surveillance definition, not a complete estimate of infection-associated mortality: influenza may contribute to a death without being recorded on its certificate, so CDC also uses statistical models to estimate its broader mortality burden. (cdc.gov)

For healthcare-utilization analysis and health-services research, diagnosis codes allow administrative records to be grouped into clinically interpretable categories. The Agency for Healthcare Research and Quality’s Clinical Classifications Software Refined (CCSR), developed within the Healthcare Cost and Utilization Project, aggregates more than 70,000 ICD-10-CM diagnosis codes into over 530 categories organized across 22 body systems. These groupings support analyses of costs, utilization, outcomes, and the diagnoses accounting for hospital activity, providing evidence relevant to service planning. A code may belong to more than one category when it represents multiple conditions or a disease with a manifestation. For analyses requiring mutually exclusive groups, such as ranking admissions by principal diagnosis, CCSR introduced default inpatient-category assignments in version 2020.2 and outpatient first-listed-diagnosis assignments in version 2021.1. These analytical groupings are additional tools, not subdivisions of WHO’s core classification. (hcup-us.ahrq.gov)

In reimbursement, ICD-coded diagnoses serve as inputs to diagnosis-related group and case-mix systems rather than constituting payment rules themselves. Under the US Medicare inpatient prospective payment system, Medicare Severity Diagnosis Related Groups use the principal diagnosis, up to 24 additional diagnoses, and up to 25 procedures; some groups also consider age, sex, or discharge status. Since October 1, 2015, these diagnosis and procedure inputs have been reported using ICD-10-CM and the separately developed ICD-10-PCS. The grouping software and reimbursement arrangements are administered outside the WHO classification: a diagnosis code supplies classified information but does not, by itself, specify a payment amount. (cms.gov)

Administrative applications in quality and patient safety similarly depend on measures constructed from coded records. AHRQ’s Patient Safety Indicators identify potentially preventable adverse events in inpatient administrative data and support internal monitoring and comparative reporting. An indicator requires specified numerator events, an eligible denominator, and often risk adjustment before observed rates can be compared across institutions treating different patient populations. Consequently, a coded complication, an indicator rate, and a finding of preventable harm are distinct outputs. The indicators identify potential safety problems; validation against clinical records remains important, and an outcome measure alone may not reveal the mechanism of harm or the organizational changes needed to address it. (psnet.ahrq.gov)

Updates, editions, and electronic access

The World Health Organization (WHO) maintained ICD-10 through updates to its tabular list, instruction manual, and alphabetical index, rather than only through wholesale revision of the classification. The ICD-10 Update Reference Committee, established in 1999, reviewed proposed modifications; recommendations following its establishment were assigned identifiers and classified as major or minor changes. WHO’s official update lists record the source of each recommendation, its approval date, and its intended implementation date. Under the documented updating procedure, changes were approved at the October meetings of the heads of WHO collaborating centres for the Family of International Classifications. Corresponding amendments to translations and related coding tools required dissemination by the responsible authorities. (who.int)

Approval, publication, and implementation are distinct events. Approval records acceptance of an amendment; publication makes the amendment available; the implementation date specifies when it is intended to enter use. For example, WHO’s publication page dates the “Official WHO ICD-10 Updates 2019 Package” to February 1, 2020: the package’s year therefore cannot simply be read as its publication date. Corrigenda correct errors in published material and are distinguished from approved classification changes; WHO’s lists omit approval dates for corrigenda. The archive includes annual amendment lists, cumulative documents covering multiple years, and separate corrections, including errata for the 2010 edition published on March 23, 2012. (who.int)

Electronic access preserves these version distinctions. WHO’s classification website lists the ICD-10 browser’s 2019 version alongside earlier online versions for 2016, 2015, 2014, 2010, and 2008, and links to the archived updating platform. The browser allows navigation by code and text: its quick search covers category titles, inclusion terms, and the alphabetical index, while advanced search provides additional options. It also links to the instruction manual and change lists. Downloadable releases provide machine-readable material; WHO’s download index explicitly labels its 2019 ClaML and plain-text tabular files as including COVID-19 updates. Thus, “2019 version” identifies a release framework, but does not necessarily mean that every element of an available electronic file was fixed in 2019. (who.int)

Historical databases consequently require more precise version identification than the label “ICD-10” alone. Relevant metadata include the WHO edition or release, any national modification, and applicable amendments and effective dates. A printed edition, a subsequently corrected electronic release, and a national code set may not contain identical material even when all are described as ICD-10. Reconstructing the classification used for a particular reporting period therefore involves consulting the corresponding archived release and update lists, rather than substituting the currently accessible browser. This is a version-control issue; its consequences for statistical time series also depend on the categories and coding rules changed. (who.int)

The emergency response to COVID-19 illustrates amendments made long after the tenth revision’s initial publication. WHO records the activation of emergency COVID-19 codes in February 2020: U07.1 identified COVID-19 with the virus identified, while U07.2 covered clinical or epidemiological diagnosis when laboratory confirmation was inconclusive or unavailable. Additional codes activated in September 2020 distinguished personal history of COVID-19, post-COVID-19 condition, and associated multisystem inflammatory syndrome. In January 2021, further emergency codes addressed immunization and adverse reactions to COVID-19 vaccines. These activations followed consultation with relevant WHO classification committees and reference groups. (who.int)

WHO amendments did not automatically determine national implementation schedules. In the United States, the National Center for Health Statistics initially announced October 1, 2020, as the implementation date for U07.1 in ICD-10-CM, then advanced it to April 1, 2020. The April release replaced the October 2019 release for healthcare services provided from April 1 through September 30, 2020. This national effective interval was separate from WHO’s emergency-code activation and was documented through ICD-10-CM addenda and reporting guidelines. (stacks.cdc.gov)

Comparability and limitations of coded data

Changes in the International Classification of Diseases can interrupt a statistical time series even when the underlying occurrence of disease has not changed correspondingly. Discontinuities arise both from altered category boundaries and from revised rules for assigning or selecting codes. In the United States, the introduction of ICD-10 for mortality statistics in 1999 changed the allocation of deaths among causes. For example, the revised interpretation of the direct-sequel rule treated pneumonia as a consequence of a wider range of conditions reported on death certificates, shifting some deaths previously assigned to pneumonia to other underlying causes. A fall in recorded pneumonia mortality across this boundary therefore cannot be interpreted entirely as a reduction in disease occurrence. (stacks.cdc.gov)

Bridge-coding studies measure these classification effects by coding the same death certificates independently under both revisions. For a specified cause $c$, the comparability ratio is

where the numerator and denominator count deaths assigned to that cause under each revision in the same study population. A ratio above 1 indicates a net increase attributable to the classification change; a ratio below 1 indicates a net decrease. Final US ratios included 0.6974 for influenza and pneumonia and 1.5812 for Alzheimer disease: approximately 30% fewer and 58% more deaths, respectively, were assigned to those underlying-cause groups under ICD-10. These are coding effects observed in a dual-coded population, not estimates of changes in biological risk. (cdc.gov)

An earlier mortality rate can be multiplied by the relevant ratio to produce a comparability-modified estimate for comparison with the new revision. This adjustment is specific to the cause grouping and coding comparison studied, rather than a universal conversion factor. National ratios may be unsuitable for age-, sex-, or race-specific analyses because the combinations of causes reported on certificates differ between demographic groups. Moreover, a ratio measures the net redistribution: movement into and out of a category can offset one another. Thus, a value near 1 does not establish that individual records received equivalent classifications. Bridge coding addresses revision-related discontinuities, not all differences in certification or data quality. (cdc.gov)

Datasets labelled “ICD-10” can also differ without crossing a major revision boundary. The WHO classification and national clinical modifications are related but not interchangeable: ICD-10-CM adds clinical detail for US morbidity reporting, whereas US mortality coding uses ICD-10. Release dates matter within a modification as well. For example, the ICD-10-CM update effective April 1, 2024, replaced the October 1, 2023, release for services during the remainder of that fiscal year. Consequently, identical column names or superficially similar code strings do not establish identical definitions. A reproducible comparison identifies the classification, national modification, applicable release, and any aggregation of detailed codes into broader groups. (cdc.gov)

Statistical categories preserve selected features of clinical documentation rather than its full content. WHO distinguishes classification from a nomenclature that separately names every known disease entity: less frequent conditions may share a category, while residual categories accommodate other specified or insufficiently specified conditions. Coding specificity also depends on what the record documents. For epidemiological research, the operational definition therefore includes the actual code list and exclusions, not merely a disease label. Mortality analyses must additionally distinguish deaths selected by underlying cause from deaths identified through any recorded cause; the latter retains contributing conditions that an underlying-cause-only table omits. These choices determine which records enter a numerator and must remain explicit when comparing populations or periods. (cdn.who.int)

Succession by ICD-11

The Seventy-second World Health Assembly adopted ICD-11 in May 2019, with entry into effect on January 1, 2022, subject to transitional arrangements. These dates distinguish international endorsement and commencement from operational replacement within individual countries. The World Health Organization (WHO) permits countries to continue using ICD-10 for as long as necessary during implementation; the 2022 commencement therefore did not simultaneously replace national mortality systems, hospital coding arrangements, or clinical modifications of ICD-10. National transitions involve separate decisions about readiness, reporting requirements, and the information systems in which the classification is embedded. (who.int)

ICD-11 changes the technical architecture of the International Classification of Diseases, rather than merely enlarging ICD-10’s tabular list and alphabetical index. Its Foundation Component is a structured database of health-related entities and their relationships, from which the classification for Mortality and Morbidity Statistics (MMS) is derived. The digital environment includes a browser, a term-searching Coding Tool, and an application programming interface for integration into local software; online and offline operation are supported. WHO described the February 2022 release as containing approximately 17,000 unique codes and more than 120,000 codable terms. These figures refer to that release, not to a permanently fixed inventory. (icd.who.int)

Expanded content includes separate chapters for diseases of the immune system, sleep–wake disorders, and conditions related to sexual health. ICD-11 also incorporates more detailed provisions for antimicrobial resistance, cancer registration, and patient safety. A significant coding mechanism is the use of extension codes to add attributes such as anatomical location, severity, or histopathology to a stem code. This combination, known as postcoordination, allows clinical detail to be represented without creating a separate precombined category for every possible combination of attributes. Consequently, software must accommodate code combinations as well as individual codes, while coders must learn the rules governing their construction. (who.int)

Transition work encompasses training clinical coders, clinicians, statistical personnel, and information-technology staff; adapting data-entry and storage systems to the new code structure; integrating coding tools; and revising reporting and case-mix arrangements. WHO’s 2019 implementation guide encouraged maintaining the existing classification during an overlap of at least 18 months, allowing continuity of reporting and identification of implementation problems. Its planning framework includes assessment of existing infrastructure, budgets, workforce requirements, and national reporting standards. WHO’s fact sheet estimated implementation periods of one to two years for a simple system newly introducing ICD-11, compared with four to five years for sophisticated systems already using an earlier revision; these were planning estimates, not universal deadlines. (icd.who.int)

Mapping tables, or crosswalks, support comparison between ICD-10 and ICD-11, but do not establish automatic equivalence between their categories. WHO explicitly distinguishes these cross-references from tools for directly converting existing records into ICD-11 codes: differences in medical content and coding structure can produce gaps or inaccuracies. The transition guide describes forward mappings from the older classification and backward mappings to it, together with dual-coding studies in which the same records are independently classified under both revisions. Such studies help assess the effects of the change on statistical time series and case-mix grouping, rather than assuming that mapped codes reproduce the information obtainable through direct ICD-11 coding. (who.int)

References

1. International Classification of Diseases (ICD)who.int
2. International Statistical Classification of Diseases and Related Health Problems, Tenth Revision, Volume 2: Instruction Manual, 2010 Editionicd.who.int
3. International Classification of Diseases—Tenth Revision: Report of a WHO Expert Committeeiris.who.int
4. International Statistical Classification of Diseases and Related Health Problemsiris.who.int
5. Report of the International Conference for the Tenth Revision of the International Classification of Diseases, A43/8iris.who.int
6. International Statistical Classification of Diseases and Related Health Problems, Tenth Revision, Volume 1iris.who.int
7. ICD-10 Injury Mortality Framework: Overview and Implicationscdc.gov
8. ICD-10: International statistical classification of diseases and related health problems, Volume 2, Fifth edition, 2016cdn.who.int
9. International statistical classification of diseases and related health problems, Volume 2, 2019icd.who.int
10. ICD-10 Version:2016icd.who.int
11. Instruction manual part 2e volume tabular index 2011cdc.gov
12. ICD-10-Mortality, Volume 1, 2022cdc.gov
13. Internationalcdn.who.int
14. Instructions for Classification of Underlying and Multiple Causes of Death – Section I – 2024cdc.gov
15. Underlying and Multiple Cause of Death Codescdc.gov
16. Underlying Cause of Death 1999–2020wonder.cdc.gov
17. Multiple Cause of Death 1999–2020wonder.cdc.gov
18. Cumulative Official WHO Updates to ICD-10, 1996–2012, Volume 1who.int
19. The ICD-10 Classification of Mental and Behavioural Disorders: Clinical descriptions and diagnostic guidelineswho.int
20. The ICD-10 Classification of Mental and Behavioural Disorders: Diagnostic criteria for researchwho.int
21. ICD-10-CM | Classification of Diseases, Functioning, and Disability | CDCcdc.gov
22. ICD-10-CA/CCI, (PDF) | CIHIsecure.cihi.ca
23. A Guide for Users of the Canadian Classification of Health Interventions (CCI)cihi.ca
24. International Classification of Diseases (ICD) | The Measures Management Systemmmshub.cms.gov
25. Public Health Transition to ICD-10-CM/PCSarchive.cdc.gov
26. ICD-10-PCS Official Guidelines for Coding and Reporting 2023edit.cms.gov
27. Version 2022 ICD-10-CA/CCI classifications, Canadian Coding Standards and related products | CIHIcihi.ca
28. Internationale statistische Klassifikation der Krankheiten und verwandter Gesundheitsprobleme - Zentraler Terminologieserverterminologien.bfarm.de
29. BfArM - FAQ/Ansprechpersonen - ICD-10bfarm.de
30. Kommentar zur ICD-10-GM Version 2026klassifikationen.bfarm.de
31. ICD-10-GM 2025: BfArM veröffentlicht endgültige Fassungbfarm.de
32. WHO Mortality Database - WHOwho.int
33. WHO Mortality Databasewho.int
34. MDBplatform.who.int
35. U.S. Influenza Surveillance: Purpose and Methodscdc.gov
36. Hospitalizations and Deathscdc.gov
37. Clinical Classifications Software Refined (CCSR) for ICD-10-CM Diagnoseshcup-us.ahrq.gov
38. MS-DRG Classifications and Softwarecms.gov
39. Patient Safety Indicatorspsnet.ahrq.gov
40. Making Adjustments to Health Care Quality Scoresahrq.gov
41. List of Official ICD-10 Updateswho.int
42. Official WHO ICD-10 Updates 2019 Packagewho.int
43. ICD-10 Version:2019icd.who.int
44. ICD-10 downloadable classificationsicdcdn.who.int
45. COVID-19 Emergency ICD Codeswho.int
46. International Guidelines for Certification and Classification (Coding) of COVID-19 as Cause of Deathwho.int
47. ICD-10-CM Filescdc.gov
48. New ICD-10-CM code for the 2019 Novel Coronavirus (COVID-19), April 1, 2020: effective March 18, 2020stacks.cdc.gov
49. Comparability of cause of death between ICD-9 and ICD-10: preliminary estimatesstacks.cdc.gov
50. National Vital Statistics Reports, Vol. 49, No. 2 (5/18/01)cdc.gov
51. NVSS - Mortality - Historical Data - Comparability Between ICD Revisionscdc.gov
52. Comparability ratio - Health, United Statescdc.gov
53. ICD-10-CM Official Guidelines for Coding and Reporting: FY 2026stacks.cdc.gov
54. The 72nd World Health Assembly Resolution for ICD-11 Adoptionwho.int
55. Eleventh revision of the International Classification of Diseasesapps.who.int
56. ICD-11 Implementationwho.int
57. ICD-11 Implementation or Transition Guideicd.who.int
58. ICD-11 2022 releasewho.int
59. ICD-11who.int