Work through the ten groups in order, because later coding decisions assume the reportability, terminology, and anatomy foundations built earlier. Each concept pairs a rule with a resolved example and a common error, so quiz yourself by covering the example, deciding the answer, and checking your reasoning against the resolution. The references below each concept identify which manual to revisit for deeper study. This guide supports your preparation; it is not an official syllabus and does not replace the current SEER manual or the candidate handbook.
Registry Foundations and the ODS Professional Role
1. What a Cancer Registry Actually Collects
A cancer registry captures the complete history, diagnosis, treatment, and follow-up health status for every cancer patient, curating that data for researchers, healthcare providers, and public health officials. The registrar's product is a standardized data record, never clinical advice or treatment decisions.
Example: A registrar receives an oncology consult recommending chemotherapy; she records the documented plan and its dates rather than advising the patient, because registry work is data capture and curation.
Mistake to avoid: Treating the abstract as a clinical treatment document or altering physician statements to fit expected codes.
Reference: Certified Tumor Registrar (CTR) / Oncology Data Specialist (ODS); National Cancer Institute: What is cancer?
2. The Four Domains of ODS Practice
The credential tests legal and ethical aspects, registry operations, data identification, and coding and abstraction. The coding and abstraction domain permits use of references supplied inside the testing platform, while the other domains are closed-book, which shapes how each area must be mastered.
Example: A candidate maps reportability questions to the data identification domain and staging assignments to coding and abstraction, recognizing that only the latter permits reference lookups during testing.
Mistake to avoid: Assuming every domain allows reference lookups and skipping memorization of reportability terminology.
Reference: Certified Tumor Registrar (CTR) / Oncology Data Specialist (ODS)
3. Hospital Registries versus Central Registries
Hospital registries at CoC-accredited facilities abstract cases for their institution and flag analytic versus non-analytic class of case; central registries consolidate records from multiple reporting facilities into one tumor record. Some data items, such as geocodes and cause of death, are populated only at the central level.
Example: A hospital sets its CoC Accredited Flag to 1 for an analytic abstract; the central registry later sets its own flag to 1 if any contributing facility did, because consolidation follows the contributing values.
Mistake to avoid: Populating central-only items at the hospital level, or assuming hospital and central sequence numbers must match.
4. Casefinding Lists versus Reportable Lists
Casefinding is the process of screening many source documents, including pathology, radiology, surgery logs, and clinic lists, to identify every possible cancer case. A reportable list defines which diagnoses the registry must actually accession. Casefinding deliberately casts a wide net so no reportable case is missed.
Example: A registrar screens radiation therapy logs and finds a benign condition treated with radiation; she reviews it for casefinding but does not accession it because it is not on the reportable list.
Mistake to avoid: Accessioning every case found during casefinding, or limiting casefinding to pathology reports alone.
5. Standard Setters and the Data Exchange Layout
SEER, the CoC, and NAACCR set the rules registrars follow, and NAACCR defines the record layout used for electronic data exchange. The NAACCR record version communicates where each data item sits and how it is coded, and it should be entered when the record is created.
Example: A 2026 case is created under NAACCR Version 26, so the record version item is coded 260, telling the receiving registry which layout and item positions apply to the submission.
Mistake to avoid: Leaving the record version blank or assuming layout positions from an older NAACCR version still apply.
6. Changing Abstracted Data When Better Information Arrives
An abstract may be changed to correct abstracting errors, to apply rule changes that retroactively affect codes, when better information emerges later such as consults or pathology addenda, or when the diagnosis is confirmed in retrospect to be earlier. Text fields should document what changed and the supporting source.
Example: A 2025 uterine specimen originally called benign is re-read in 2026 as a malignant mesenchymal tumor; the registrar changes the diagnosis date to 2025 and updates histology and behavior, documenting the re-read.
Mistake to avoid: Refusing to update a completed abstract, or changing coded values without documenting the supporting source.
Cancer Fundamentals and Classification Systems
7. Benign versus Malignant Behavior
Benign tumors do not invade nearby tissues and usually do not grow back after removal, while malignant tumors invade and can metastasize. ICD-O behavior codes express this distinction: /0 benign, /1 borderline, /2 in situ, and /3 invasive malignant, and the code drives reportability decisions.
Example: A pathology report describes a tumor invading surrounding muscle; the registrar assigns behavior /3 because invasion is documented, regardless of how slow-growing the tumor appears clinically.
Mistake to avoid: Coding behavior from tumor size, growth speed, or clinical impression instead of documented invasion.
Reference: National Cancer Institute: What is cancer?; NCI SEER Program Coding and Staging Manual 2026
8. The Hyperplasia to Carcinoma in Situ Spectrum
Hyperplasia is an overabundance of normal-appearing cells; dysplasia adds abnormal cells and disorganized architecture; carcinoma in situ shows abnormal cells that still do not invade. Carcinoma in situ is sometimes called stage 0 but is not invasive cancer, although it is often treated because it may progress.
Example: A cervical biopsy reports severe dysplasia without invasion; the registrar recognizes this sits below invasion on the spectrum and checks reportability rules for that site and grade.
Mistake to avoid: Describing carcinoma in situ as invasive cancer in text fields, or assuming every dysplasia is reportable.
Reference: National Cancer Institute: What is cancer?
9. Metastatic Tumors Keep the Primary's Identity
Cancer that spreads retains the primary's name and cell type; melanoma forming a tumor in the liver is metastatic melanoma, not liver cancer. Registry primary site coding therefore follows the documented origin, and metastatic sites are recorded separately in extent-of-disease items.
Example: A liver biopsy shows metastatic melanoma from a known skin primary; the registrar codes the primary site to skin and records liver only as a metastatic site.
Mistake to avoid: Coding the primary site as the organ where a metastasis was found when the origin is documented.
Reference: National Cancer Institute: What is cancer?; NCI SEER Program Coding and Staging Manual 2026
10. Carcinoma Subtypes and Their Cell Origins
Carcinomas arise from epithelial cells and are named for the cell of origin: adenocarcinoma from glandular, mucus-producing cells; squamous cell carcinoma from flat lining cells; transitional cell carcinoma from urothelium lining the bladder and ureters; basal cell carcinoma from the skin's basal layer.
Example: A bladder tumor is called transitional cell carcinoma; the registrar recognizes urothelial origin and applies urinary-site coding rules rather than generic carcinoma instructions.
Mistake to avoid: Treating all carcinomas as one histology family and skipping the site-specific histology coding rules.
Reference: National Cancer Institute: What is cancer?
11. Hematopoietic Neoplasms Follow Separate Rules
Leukemias, lymphomas, and plasma cell neoplasms arise in blood-forming and immune tissue and generally do not form solid tumors. They are abstracted under the Hematopoietic and Lymphoid Neoplasm Coding Manual and Database, which supplies its own reportability, multiple primary, and treatment rules.
Example: A patient has plasma cell myeloma; the registrar uses the hematopoietic manual for reportability, staging, and treatment rules instead of the solid tumor site schemes.
Mistake to avoid: Applying solid tumor multiple primary or staging rules to lymphoma and leukemia primaries.
Reference: National Cancer Institute: What is cancer?; NCI SEER Program Coding and Staging Manual 2026
12. ICD-O-3 for Registries versus ICD-10-CM for Encounters
ICD-O-3 pairs a topography code with a morphology code to describe the tumor itself for cancer registration, whereas ICD-10-CM is a morbidity classification for coding diagnoses and reasons for visits in all health care settings. The official ICD-10-CM guidelines state that coding guidelines for inconclusive diagnoses such as probable, suspected, or rule-out conditions were developed for inpatient reporting and do not apply to outpatients, so those inpatient conventions cannot be assumed when coding an outpatient encounter. Registry reportability follows its own rules: a case is not accessioned on suspicious cytology alone, follow-back on cytology diagnoses using ambiguous terminology is strongly recommended, the case is accessioned when a reportable diagnosis is confirmed later, and the date of the suspicious cytology is used as the date of diagnosis.
Example: A sputum cytology report from an outpatient clinic in March reads 'suspicious for malignancy,' and a bronchoscopy biopsy confirms squamous cell carcinoma in April; the registrar accessions the case using the March cytology date as the date of diagnosis, while the coder for the outpatient encounter cannot apply the inpatient inconclusive-diagnosis conventions, which the guidelines state do not govern outpatient reporting.
Mistake to avoid: Using ICD-10-CM neoplasm codes in the registry abstract, assuming the inpatient inconclusive-diagnosis conventions govern outpatient encounters, or accessioning a case on suspicious cytology alone before confirmation.
Reference: NCI SEER Program Coding and Staging Manual 2026; CDC ICD-10-CM Official Guidelines, FY 2027
Confidentiality, Legal, and Ethical Practice
13. Protected Health Information Covers Every Medium
Protected health information is individually identifiable health information held or transmitted by a covered entity in any form, electronic, paper, or oral, relating to health, care, or payment. Registry abstracts, pathology reports, and even a hallway conversation identifying a patient's cancer all fall within this definition.
Example: A registrar verbally confirms a patient's diagnosis to an unidentified caller without authorization; this oral disclosure involves PHI because the rule covers any medium.
Mistake to avoid: Assuming privacy rules apply only to electronic records or written reports.
Reference: HHS Summary of the HIPAA Privacy Rule
14. Who Counts as a Covered Entity
Health plans, health care clearinghouses, and providers who transmit health information electronically in connection with standard HIPAA transactions are covered entities. Merely using email does not make a provider covered; the transmission must relate to a standard transaction such as a claim or eligibility inquiry.
Example: A small practice bills electronically using standard claim formats, so it is a covered provider; the registry receiving its case reports handles that data under permitted public health disclosures.
Mistake to avoid: Assuming every organization touching health data is covered, or that casual email use creates coverage.
Reference: HHS Summary of the HIPAA Privacy Rule
15. The Minimum Necessary Standard
Covered entities must limit uses, disclosures, and requests of PHI to the minimum needed, supported by role-based access policies. The standard does not apply to disclosures for treatment, to the individual, under an authorization, to HHS enforcement, or when the use is required by law.
Example: A registry clerk requests an entire medical record when only the pathology and treatment summaries are needed; the request exceeds minimum necessary and should be narrowed.
Mistake to avoid: Sending whole records for routine requests without justification, or applying minimum necessary to treatment disclosures.
Reference: HHS Summary of the HIPAA Privacy Rule
16. Public Health Reporting Permits Registry Disclosure
Covered entities may disclose PHI without individual authorization to public health authorities legally authorized to collect or receive it for preventing or controlling disease, injury, or disability. This permitted disclosure is the legal foundation for mandatory cancer reporting to central registries.
Example: A hospital transmits case reports to the state cancer registry under the public health provision; no patient authorization is required because the registry is an authorized authority.
Mistake to avoid: Delaying legally required cancer case reporting while waiting for patient authorization.
Reference: HHS Summary of the HIPAA Privacy Rule
17. De-identified Data versus Limited Data Sets
De-identified information has specified identifiers removed or statistical certification and carries no use restrictions. A limited data set removes direct identifiers but may retain dates and geographic detail, and it may be used for research, operations, or public health only under a data use agreement promising safeguards.
Example: A researcher needs survival dates by town; the registry supplies a limited data set under a data use agreement rather than de-identified data, because de-identification strips those elements.
Mistake to avoid: Releasing a limited data set without a data use agreement, or believing removing names alone de-identifies a record.
Reference: HHS Summary of the HIPAA Privacy Rule
18. Registry Restriction Flags for Data Release
The No Patient Contact Flag blocks research contact requests at the patient level, while the Reporting Facility Restriction Flag marks cases reported only by facilities such as VHA or DoD whose data may not be releasable. Neither flag blocks routine surveillance reporting to NCI, CDC, or NAACCR.
Example: A patient asks not to be contacted for studies; the central registry sets the contact flag to 1 across all her tumor records yet still submits them for routine surveillance.
Mistake to avoid: Suppressing routine surveillance reporting because a restriction flag is set, or applying a patient-level flag to only one tumor record.
Casefinding and Reportability Decisions
19. The Basic Reportability Standard for Solid Tumors
Report all histologies with behavior /2 or /3 in ICD-O-3.2, including newly reportable terms and high-grade intraepithelial neoplasia grades II and III at many sites. Reportability is determined by registry rules and the reportable diagnosis list, never by whether a case can be staged under AJCC.
Example: Anal intraepithelial neoplasia III is documented; the registrar accessions it because high-grade intraepithelial neoplasia at that site carries a reportable behavior code.
Mistake to avoid: Skipping accession because a lesion is in situ, or using AJCC staging to decide reportability.
20. Skin Primaries That Are Not Reportable
For skin sites C440-C449, malignant neoplasm NOS, epithelial and papillary squamous cell carcinoma, and basal cell carcinoma histologies are not reportable to SEER. If a registry collects them by agreement, such cases are sequenced in the 60-87 range and are not submitted.
Example: A pathology report shows basal cell carcinoma of the cheek; the registrar does not report it to SEER, though her registry may collect it by agreement with a 60-range sequence number.
Mistake to avoid: Reporting routine basal or squamous skin cancers to SEER, or forgetting the special sequence range when collecting locally.
21. Benign and Borderline CNS Tumors Are Reportable
Primary intracranial and CNS tumors with behavior /0 or /1 are reportable for sites including brain lobes, meninges, spinal cord, cranial nerves, pituitary, craniopharyngeal duct, and pineal gland. Benign and borderline tumors of the cranial bones and peripheral nerves are excluded from this requirement.
Example: An MRI report describes a cerebellar tumor; the registrar accessions because tumor is a reportable term for intracranial sites even without a tissue diagnosis.
Mistake to avoid: Limiting CNS reportability to malignant tumors, or reporting benign cranial bone lesions.
22. Cervical In Situ and Prostatic Intraepithelial Neoplasia
In situ carcinoma of the cervix, CIN III, and cervical SIN III are not reportable, with collection stopped for diagnoses from 1996 forward. Prostatic intraepithelial neoplasia PIN II and PIN III are likewise not reportable for diagnoses from 2001 forward.
Example: A 2026 cervical biopsy shows CIN III; the registrar does not accession it, and any locally collected cervix in situ cases are sequenced in the 60-87 range.
Mistake to avoid: Accessioning cervical in situ disease diagnosed after the cutoff, or mixing it into the malignant sequence series.
23. Clinical Diagnoses and the Pathology Precedence Rule
A case is reportable when a recognized practitioner clinically states cancer, even without tissue confirmation, and a pathology report normally takes precedence over a clinical diagnosis. Two exceptions override a negative biopsy: when cancer treatment is given, and when the physician still calls the disease reportable six months or longer afterward.
Example: A discharge summary lists lung cancer clinically; a later biopsy is negative, yet the physician still affirms the diagnosis seven months on; the registrar accessions the case.
Mistake to avoid: Accessioning every clinical mention of cancer while ignoring a subsequent negative biopsy.
24. In-Utero Diagnoses Require a Live Birth
Diagnoses made before birth are reportable only when the pregnancy results in a live birth; absent documentation of stillbirth, abortion, or fetal death, assume a live birth and report. When disease regresses before birth, the case is still accessioned based on the pre-birth diagnosis.
Example: A fetal ultrasound at 36 weeks shows an adrenal mass consistent with neuroblastoma and a live birth follows; the registrar reports the case using the prenatal diagnosis date.
Mistake to avoid: Using the birth date as the diagnosis date, or dropping prenatal cases lacking explicit live-birth documentation.
Ambiguous Terminology and Diagnosis Dates
25. The Reportable Ambiguous Terms
Terms such as apparent, appears, comparable with, compatible with, consistent with, favors, malignant appearing, most likely, presumed, probable, suspect, suspicious for, and typical of are reportable when attached to a cancer term. Do not substitute unlisted synonyms, such as using likely in place of most likely.
Example: A report describes a gastric lesion compatible with adenocarcinoma; the registrar accessions because compatible with is a reportable term paired with a malignancy.
Mistake to avoid: Treating unlisted synonyms like possible or likely as reportable equivalents of listed terms.
26. Phrases That Are Not Reportable
Phrases equivalent to not diagnostic, including highly suspicious for but not diagnostic of, high probability for, most compatible with a non-reportable diagnosis, and differential considerations, are not reportable when no other information exists. Rule-out statements likewise express uncertainty rather than a diagnosis.
Example: The final diagnosis reads rule out lymphoma; the registrar does not accession, because rule-out language expresses uncertainty rather than diagnosis.
Mistake to avoid: Accessioning rule-out or differential-consideration statements as cancer diagnoses.
27. Resolving Mixed Ambiguous Terms
When a single report contains both reportable and non-reportable ambiguous terms, accept the reportable term and accession the case. When the original source document used a non-reportable term, later references to a history of cancer do not convert the case into a reportable one.
Example: A CT discussion calls a renal lesion compatible with renal cell carcinoma while the impression says possible carcinoma; the registrar accessions, giving weight to the reportable term within the same report.
Mistake to avoid: Letting a later history-of-cancer note override an original non-reportable term in the source document.
28. Suspicious Cytology Never Stands Alone
Never accession a case based only on cytology reported with ambiguous terminology, even when the term itself is reportable. Follow back and accession when a definitive diagnosis follows; the date of the suspicious cytology then serves as the date of diagnosis. Cytology positive for malignant cells is reportable.
Example: A sputum cytology reads suspicious for malignancy in March, and an April lung biopsy confirms squamous carcinoma; the registrar accessions with March as the diagnosis date.
Mistake to avoid: Accessioning on ambiguous cytology alone, or using the confirmation date instead of the suspicious cytology date.
29. Choosing the Date of Diagnosis
Code the month, day, and year the tumor was first diagnosed clinically or microscopically by a recognized practitioner, including physicians, physician assistants, nurse practitioners, residents, and fellows. Use the procedure date rather than the specimen receipt or report date, keep the clinical date when later confirmed, and never establish the date from positive tumor markers alone.
Example: A biopsy is performed on the 3rd and read positive on the 7th; the registrar records the 3rd, because the procedure date governs when pathology is the only evidence.
Mistake to avoid: Recording the dictation, transcription, or specimen receipt date as the diagnosis date.
30. Estimating Unknown Diagnosis Dates
When exact dates are unavailable, estimate systematically: spring codes to April, summer to July, fall to October, early in year to January, and late in year to December; a couple of years means two years earlier and a few years means three. Otherwise use the month or year of admission, and never leave the year blank for SEER-transmitted cases.
Example: A record states the diagnosis was a few years before a 2026 admission; the registrar estimates 2023 and documents the basis for the estimate.
Mistake to avoid: Defaulting unknown months to July without using available information, or leaving the diagnosis year blank.
Demographics and Patient Identification
31. Residency at Diagnosis Follows Usual Residence
Residency at diagnosis follows Census-style rules: code the usual residence, the place where the patient lives and sleeps most of the time. For seasonal residents, migrant workers, and people staying with family during treatment, code the usual residence rather than the temporary address; boarding-school students code to the parents' residence.
Example: A snowbird diagnosed in January at a Florida hospital lists a Michigan home; the registrar codes Michigan because that is where she spends most of the year.
Mistake to avoid: Coding the treatment facility's location or a temporary mailing address as the patient's residence.
32. Address Sources in Priority Order
Use the stated street address of the usual residence first; a post office box is unreliable for geography and is used only when no street address exists after follow-back. Use death certificate residency only when residency from all other sources is unknown, since certificates may list a former home rather than the diagnosis-time residence.
Example: A death certificate lists a childhood home, but the chart shows the patient lived in an assisted-living facility at diagnosis; the registrar codes the facility address.
Mistake to avoid: Geocoding from a PO box when a street address exists, or letting a death certificate override known diagnosis-time residency.
33. Race Coding Priorities and Self-Report
Self-reported race takes precedence over medical record documentation and death certificates, keeping registry numerators comparable to census denominators. When multiple races apply, Native Hawaiian code 07 outranks all others, specific codes outrank non-specific ones, and unused Race 2 through Race 5 fields are filled with 88 rather than left blank.
Example: A patient describes herself as Samoan and White; the registrar codes Race 1 as 27 and Race 2 as 01, filling the remaining fields with 88.
Mistake to avoid: Coding race from the patient's name, photographs, or social media, or leaving secondary race fields blank.
34. Inferring Race from Birthplace, With Limits
When race is unrecorded but birthplace is known, nationality descriptions may suggest a race code. However, when the patient's name is incongruous with the nationality-based inference and no further information exists, code race as 99 unknown rather than forcing the inference. Multi-racial with no detail also codes 99, not 98.
Example: Birthplace is listed as Nigeria, suggesting Black per nationality tables, but the recorded name conflicts and nothing else is known; the registrar codes 99.
Mistake to avoid: Overriding unknown race with nationality inference when the name clearly conflicts, or coding 98 for multi-racial.
35. Spanish and Hispanic Origin Coding
Hispanic origin is coded independently of race, since Hispanic persons may be of any race. Code Brazilian birthplace as 0; when multiple Hispanic origins apply, use code 6; code 7 when a Hispanic surname is the only evidence and nothing contradicts it. When consolidating records, prefer specific country codes over the NOS code.
Example: A patient born in Cuba with no ethnicity statement anywhere in the record is coded 3, Cuban, based on birthplace.
Mistake to avoid: Assigning Hispanic origin based on race, or coding Brazilian birthplace as Hispanic.
36. Sex Assigned at Birth in the 2026 Manual
Sex Assigned at Birth replaces the former Sex item for 2026 diagnoses. Assign 1 or 2 when the patient is documented male or female with no information indicating a different assignment at birth. When sex is unknown, infer from primary site, C600-C639 for male and C510-C589 for female; for other sites assign 9.
Example: A record documents a non-binary patient with no sex-assigned-at-birth information and a lung primary; the registrar codes 9 because the site permits no inference.
Mistake to avoid: Coding 9 when sex is clearly documented, or inferring sex from sites outside the specified ranges.
Coding the Tumor Description
37. Primary Site Means Site of Origin
Code the site where the tumor arose, even when it extends onto adjacent subsites; the biopsy site and metastatic sites do not define the primary. For sarcomas of unknown primary, the default is soft tissue C499 rather than the general unknown primary code, and transplanted-organ malignancies are coded where the organ now resides.
Example: A pancreatic tumor originates in the tail but extends into the spleen and stomach; the registrar codes pancreas tail, the site of origin, not the invaded organs.
Mistake to avoid: Coding every involved site, or defaulting sarcoma unknowns to the general unknown primary code.
38. Overlapping Lesions versus Multiple Subsite Tumors
A single tumor overlapping adjacent subsites of one organ, with the point of origin undetermined, takes the overlapping-lesion code ending in 8. When multiple tumors in different subsites of the same organ form a single primary and the origin cannot be determined, code the organ NOS with a final digit 9.
Example: A single colon tumor involves the hepatic flexure and the transverse colon with the point of origin undetermined; the registrar codes C188, the overlapping-lesion code for colon, rather than forcing one subsite. Had two separate tumors arisen in those subsites as a single primary, the code would be C189, colon NOS.
Mistake to avoid: Using the 8 ending for multiple separate tumors, or the 9 ending for a single tumor spanning two subsites.
Reference: NCI SEER Program Coding and Staging Manual 2026; CDC ICD-10-CM Official Guidelines, FY 2027
39. Whose Statement Wins for Primary Site
The surgeon generally determines the site of origin because of direct visualization, but a pathologist examining an entire organ may pinpoint the origin within it. For site, histology, and behavior, consult pathology reports are preferred over the original reports because specialists' re-reads are considered more accurate.
Example: During a pancreatectomy the surgeon calls the tumor tail, while whole-organ pathology localizes it to the body; the registrar weighs the pathologist's whole-organ assessment for that organ.
Mistake to avoid: Automatically preferring the surgeon for every organ, or ignoring a consult re-read that changes the site.
40. Behavior Coding Rules and Edge Cases
Assign /3 if any portion of the primary tumor is invasive, however limited, and when lymph nodes are positive despite an in situ appearance. In situ behavior requires pathologic examination, with rare imaging-based exceptions. When the pathologist states a behavior that differs from the ICD-O listing, the pathologist's description governs.
Example: A colon polyp specimen contains mostly in situ disease but one focus of submucosal invasion; the registrar codes /3 because any invasion makes the tumor malignant.
Mistake to avoid: Coding /2 because invasion is minimal, or assigning metastatic /6 behavior to a SEER-reported case.
41. Laterality Coding for Paired Sites
For paired sites, code 1 right and 2 left origin; code 3 when only one side is involved but the side is unstated; code 4 for bilateral involvement of a single primary such as both ovaries with one histology; code 5 for midline tumors of listed sites like cerebral meninges and trunk skin; code 9 when laterality is simply unknown.
Example: A testicular tumor is confirmed in a single testis without naming right or left; the registrar codes 3 because one-side involvement is documented without the side.
Mistake to avoid: Coding 9 when the record clearly confines disease to one side, or assigning midline 5 to unlisted sites.
42. Cancer PathCHART Combination Standards
Cancer PathCHART designates each site-morphology combination as valid, unlikely, or impossible. Valid combinations abstract without issue; impossible ones cannot be abstracted and require a different combination; unlikely ones trigger an edit requiring thorough record review and, if the coding is confirmed correct, the Site/Type interfield override flag.
Example: An edit rejects the coded site-histology pair as impossible; the registrar re-reviews the record and corrects the combination, since impossible entities cannot be abstracted.
Mistake to avoid: Setting the override flag without reviewing the record, or assuming an unlikely combination can never be abstracted.
Staging, Extent of Disease, and Tumor Characteristics
43. Summary Stage 2018 Categories
Summary Stage 2018 codes run 0 in situ, 1 localized, 2 regional by direct extension only, 3 regional lymph nodes only, 4 regional by both direct extension and nodes, 7 distant site or node involvement, 8 benign or borderline disease within the Brain, CNS Other, Intracranial Gland, and Medulloblastoma chapters, and 9 unknown or unstaged disease as well as death certificate only cases. Derived Summary Stage 2018 is calculated from the EOD items through a defined algorithm, so directly assigned and derived values should agree.
Example: A tumor extends into neighboring tissue with no nodal or distant involvement; the registrar assigns Summary Stage 2, regional by direct extension only.
Mistake to avoid: Confusing the distant-metastasis code 7 with the benign or borderline use of 8 in the CNS chapters, or treating 9 as DCO-only when it also covers unstaged and unknown disease.
44. Extent of Disease Items and Derived Stage
Extent of Disease Primary Tumor, Regional Nodes, and Mets are collected at diagnosis, each carrying special codes such as 000 in situ, 800 no evidence of primary tumor, and 999 unknown. Derived Summary Stage is computed from these three items through an algorithm that may incorporate other data, and unknown information takes the designated codes rather than blanks.
Example: A pathology-only abstract lacks extension detail; the registrar codes the EOD items to their specified unknown values rather than guessing extent from histology.
Mistake to avoid: Leaving EOD items blank instead of using the specified unknown codes for missing information.
45. Mets at Diagnosis Site Items
Six items record bone, brain, liver, lung, distant node, and other metastatic involvement at diagnosis. Code 0 when metastases exist but that site is not mentioned; code 9 when metastases are known but involvement of the specific site is unclear. Carcinomatosis takes code 2 in the Other item and outranks code 1 there.
Example: A colon cancer patient has lung metastases and carcinomatosis; the registrar codes Lung as 1 and Other as 2, since carcinomatosis takes priority in the Other item.
Mistake to avoid: Coding 9 when metastases are documented but a site is simply not involved, or recording contiguous extension as metastasis.
46. Tumor Size Summary Priorities
Record size from the resection specimen when surgery is the first definitive treatment, preferring the synoptic report when measurements conflict. With neoadjuvant therapy, use the largest pretreatment size instead of the specimen. Record the invasive component's size when given, and convert less-than statements by subtracting a millimeter and greater-than by adding one.
Example: A 3.1 cm rectal mass shrinks to 1.8 cm after chemoradiation; the registrar records 031 because the largest pretreatment size governs after neoadjuvant therapy.
Mistake to avoid: Recording post-neoadjuvant pathologic size, or coding the polyp or cyst size instead of the tumor's.
47. The Four Grade Data Items
Grade Clinical records the pretreatment grade; Grade Pathological applies to resections without neoadjuvant therapy; Grade Post Therapy Clin (yc) and Grade Post Therapy Path (yp) apply after neoadjuvant or primary systemic or radiation therapy. Derived Summary Grade uses the highest clinical or pathological grade, with breast an exception because behavior affects priority.
Example: A rectal cancer is biopsied and graded, then resected after chemoradiation; the registrar completes the clinical, post therapy pathologic, and derived grade items per the grade tables.
Mistake to avoid: Copying the pathological grade into the clinical field, or ignoring the breast exception in derived grade.
48. Lymphovascular Invasion Coding
Lymphovascular invasion is coded from pathology, with the CAP synoptic report as the preferred source. Codes run 0 not present, 1 present, 2 lymphatic or small vessel only, 3 venous or large vessel only, 4 both, 8 not applicable, and 9 unknown, indeterminate, or not mentioned. Perineural invasion is not coded in this item.
Example: A synoptic report notes venous invasion only; the registrar codes 3 rather than 1, capturing the specific vessel type documented.
Mistake to avoid: Coding perineural invasion as LVI, or assigning 1 when the report specifies only one vessel type.
First Course of Therapy Coding
49. Defining the First Course of Therapy
First course of therapy includes all treatment given to destroy or modify cancer tissue after the original diagnosis. It ends when the documented treatment plan is completed, however long that takes, or earlier upon documented disease progression, recurrence, or treatment failure; with no documented plan, it ends one year after diagnosis.
Example: A prostate plan includes radiation followed by years of androgen deprivation; all of it remains first course until the plan completes, absent documented progression.
Mistake to avoid: Cutting off first course at an arbitrary time limit when a documented plan is still in progress.
50. Surveillance, Refusal, and Reasons No Treatment Was Given
For active surveillance or watchful waiting, code treatment items 0 and Treatment Status 2; treatment begun after progression is second course. Refused codes remain even if the patient later accepts treatment after one year or progression. Distinguish code 1, surgery not part of the planned first course, from code 7, recommended surgery the patient refused.
Example: A laryngeal cancer patient offered laser excision or chemoradiation chooses chemoradiation; the registrar codes Reason for No Surgery as 1, not 7, because surgery was an unchosen alternative.
Mistake to avoid: Calling a choice among treatment options a refusal, or overwriting refused codes when treatment begins later.
51. Surgery of Primary Site 2023
Code the most invasive, extensive, or definitive surgery performed, reflecting the cumulative effect of all first-course procedures to the primary site. The operative report takes precedence over pathology for determining what was done, and the code reflects the procedure actually carried out, not the one planned, including aborted cases.
Example: A partial mastectomy is followed by a completion mastectomy showing no residual tumor; the registrar codes the total removal as the cumulative, most definitive result.
Mistake to avoid: Coding the intended procedure when surgery was aborted or altered, or ignoring the cumulative effect of prior surgeries.
52. Scope of Regional Lymph Node Surgery
Distinguish sentinel lymph node biopsy, code 2, from regional dissection using the operative report, never the node count alone. Code 6 combines sentinel biopsy and regional dissection in the same surgical event or unknown timing; code 7 covers separate events. Failed mapping followed by dissection codes as 6, or 2 when no dissection followed.
Example: A melanoma case maps poorly, so the surgeon proceeds directly to a full groin dissection in the same session; the registrar codes 6 because mapping failed and dissection followed.
Mistake to avoid: Deciding sentinel versus dissection from the number of nodes examined, or coding failed mapping with dissection as 2.
53. Regional Nodes Positive and Examined
Regional nodes positive and examined are cumulative, pathology-based counts across first-course procedures. Exclude isolated tumor cells except for cutaneous melanoma and Merkel cell carcinoma; do not double-count an aspirated node from the same chain as dissected nodes. Code 95 covers aspiration-only positivity; code 97 applies when the positive count cannot be determined.
Example: A positive cervical node core biopsy precedes a separate supraclavicular dissection showing 4 of 10 positive; because the basins differ, the registrar codes 05 positive and 11 examined.
Mistake to avoid: Counting isolated tumor cells as positive nodes for most sites, or adding an aspirated node from the same chain.
54. Classifying Systemic and Radiation Therapy
SEER*Rx determines whether a drug is chemotherapy, hormone therapy, or immunotherapy, regardless of the primary sites listed for that agent. Six monoclonal agents moved from chemotherapy to immunotherapy for diagnoses on or after January 1, 2013, decided by diagnosis date. Radioembolization codes as radioisotope radiation, and chemoembolization codes as chemotherapy.
Example: A patient diagnosed in 2026 receives trastuzumab in 2027; the registrar codes it under immunotherapy because the diagnosis date, not the treatment date, controls the classification.
Mistake to avoid: Coding the reclassified agents as chemotherapy based on treatment date, or moving a SEER*Rx chemotherapy drug to Other Therapy.
Follow-up, Death Clearance, and Data Quality
55. Cancer Status and Its Date
Cancer Status records whether this tumor is clinically evident as of the Date of Last Cancer Status, using official sources such as providers or the death certificate, never family or friends. Long-term maintenance therapy by itself does not establish evidence of tumor, so such patients take code 1; deceased patients are coded from all death certificate entries.
Example: A thyroid cancer patient takes suppressive thyroid hormone for years with no other findings; the registrar codes 1, no evidence of this tumor.
Mistake to avoid: Coding tumor present solely because maintenance therapy continues, or using a relative's report as the source.
56. Recurrence Type Coding
Recurrence Type codes the first recurrence after a documented disease-free period: local, trocar, regional by adjacent tissue, nodes, or both, distant by site, or systemic. Code 70 applies when the patient was never disease-free, including metastasis at diagnosis. Codes 06, 16, 17, 26, 27, 36, and 46 are reserved for tumors originally diagnosed as in situ and must not be used for invasive primaries.
Example: A patient presents with carcinomatosis at diagnosis and never becomes disease-free; the registrar codes 70 and keeps checking for any later disease-free interval.
Mistake to avoid: Recording second and later recurrences after the first is coded, or using in-situ-specific codes such as 06 for an invasive primary.
57. Vital Status and Date of Last Contact
Vital Status, 0 dead and 1 alive, and Date of Last Contact are patient-level items, so they must be identical across all of one patient's tumor records. Code the date the patient was actually seen or contacted, not the date a follow-up report was received, and record the death date for deceased patients.
Example: A clinic visit occurs in March but the follow-up letter arrives in May; the registrar records March as the date of last contact.
Mistake to avoid: Recording different vital status values across one patient's tumor records, or using report receipt dates.
58. SEER Death Clearance Requirements
SEER death clearance exceeds the general NAACCR manual in two ways: all entries on the death certificate must be matched at the patient level, not just the underlying cause, and death certificates must be linked at the tumor level to identify possible second primaries that were never reported during life.
Example: A certificate mentions a second cancer in its comment text; the registrar checks the registry for a matching tumor and abstracts a new primary if none exists.
Mistake to avoid: Matching only the underlying cause of death and missing additional cancers noted elsewhere on the certificate.
59. Coding the Underlying Cause of Death
Use the state health department's coded underlying cause of death even when it appears questionable; coded causes from other sources such as NDI Plus follow. Code 7797 when the certificate is available but the cause is uncoded, and 7777 when neither a certificate nor a coded cause can be obtained from any source.
Example: A death is reported through a data exchange but no state certificate can be located; the registrar codes 7777 rather than assigning an ICD-10 cause.
Mistake to avoid: Recoding a cause you believe the state got wrong, or letting untrained staff assign ICD-10 cause codes.
60. Interfield Edits and Override Flags
Edits flag unusual combinations such as site with histology, histology with behavior, age with site and histology, and in situ disease without microscopic confirmation. Override flags document that a flagged case was reviewed and confirmed as reported, which prevents the same cases from requiring review at every data submission.
Example: A lymphoma is diagnosed clinically without any microscopic confirmation; the registrar verifies the diagnosis is legitimately documented in the record and sets the leukemia or lymphoma diagnostic confirmation override, since the edit identifies lymphomas confirmed only by clinical diagnosis.
Mistake to avoid: Setting overrides to silence edits without review, or leaving reviewed confirmed cases unflagged so they draw review at every submission.
References and further reading
Source links checked on Oct 2, 2026.