ODS-C cheat sheet

Reportable case definition: a newly diagnosed malignancy (behavior code /3) in a resident of the registry catchment area, plus in-situ (/2) cases only where the standard setter requires them; benign and borderline tumors are reportable only for specified sites (e.g., brain/CNS, bladder).

Notifiable condition: case ascertainment relies on state legal mandate; reporting sources include hospitals, outpatient surgery centers, pathology labs, radiation oncology, and physician offices, with reporting deadlines set by state regulation (commonly 6 months from diagnosis).

Multiple primary rules: use the current solid tumor rules (site-specific, replacing the 2007 MP rules) for invasive and in-situ solid tumors; use the hematopoietic and lymphoid neoplasm rules for blood cancers, where histology and timing drive counting.

Date of diagnosis hierarchy: use the earliest date of histologic/cytologic confirmation; if none, the earliest date of clinical diagnosis (imaging, physical exam, or physician statement); never substitute the date of first treatment or accession date.

Abstracting sequence: identify the patient and facility, assign accession number, code patient demographics, then tumor-specific fields (site, histology, behavior, grade, laterality), then stage, then first-course treatment, then follow-up.

Primary site and histology are coded from ICD-O-3 (with current update tables); topography uses the 4-character site code and morphology uses 4-digit histology plus a one-digit behavior code, with grade in a separate field.

Class of case separates analytic (diagnosed and/or first course of treatment at the reporting facility) from non-analytic cases; only analytic cases enter survival and treatment analyses, so misclassification distorts facility comparisons.

Follow-up: active annual follow-up captures vital status, date of last contact, recurrence, and subsequent treatment; registries use death certificate clearance, NDI linkage, and hospital/clinic queries to reduce lost-to-follow-up.

Staging systems are not interchangeable: AJCC TNM (anatomic extent, edition-specific), SEER Summary Stage (localized/regional/distant), and collaborative-stage/SSDI rules each answer different questions; never convert one to another by assumption.

AJCC T describes primary tumor size/invasion, N describes regional lymph node involvement, M describes distant metastasis; prefix c = clinical, p = pathologic, y = post-neoadjuvant, r = recurrent, a = autopsy.

Pathologic stage requires resection of the primary (or sufficient tissue) to assess the highest T and N; clinical stage may be assigned from imaging and exam when no resection occurs, and must be coded even when pathologic stage is unavailable.

Summary Stage 2018 categories: in situ, localized, regional (direct extension, regional nodes, or both), distant, and unknown; benign/borderline CNS tumors use a separate schema.

Grade reflects differentiation and is site-specific: Gleason score/grade group for prostate, Nottingham (Bloom-Richardson) for breast, Fuhrman/WHO/ISUP for kidney, and differentiation for most carcinomas.

First course of treatment includes all treatment intended to modify the tumor, delivered within the first course window (generally the first 12 months or until progression/recurrence, whichever comes first); it includes surgery, radiation, systemic therapy, and immunotherapy.

Surgery codes distinguish diagnostic-only procedures from definitive resection; a biopsy that removes the entire tumor still may be diagnostic if the intent was diagnosis, so abstract intent rather than specimen size alone.

Biomarker and molecular fields (e.g., ER/PR/HER2 for breast, KRAS/NRAS for colorectal, EGFR/ALK for lung) are required data items because they drive treatment and stratified reporting.

Standard setters: the North American Association of Central Cancer Registries sets data standards and certification; state central registries set reporting regulations; facility cancer programs follow Commission on Cancer (CoC) standards for accreditation.

Data standards are versioned: each diagnosis year has a defined ICD-O-3 update, staging edition, and required data set; abstracting a case to the wrong diagnosis-year standard is a common audit finding.

Consent, confidentiality, and HIPAA: registry reporting is generally permitted as required-by-law public health activity, but release of identified data for research requires IRB approval and data-use agreements.

Quality control has three pillars: completeness (casefinding audits against pathology and discharge lists), accuracy (reabstracting/recode audits), and timeliness (reporting within required intervals).

NAACCR certification uses measures such as completeness (estimated percentage of reportable cases captured), timeliness, and data quality edits; failure to meet thresholds affects central registry certification status.

Edits: interfield and intrafield consistency checks (e.g., a male with cervical primary, or a localized stage with distant metastasis code) must be resolved before submission; unresolved errors block record acceptance.

Casefinding audit method: re-review a sample of pathology logs, radiation logs, and disease index listings against registry accession files; missed cases are counted and the miss rate is used to correct estimates.

Confidentiality breach response includes documentation, notification per state law and HIPAA, and root-cause review; registry staff access is role-based and audited.

Incidence rate = (new cases in period / population at risk) × 100,000; use the population matching the case definition (e.g., sex-specific denominators for sex-specific cancers).

Age adjustment: direct standardization applies a standard population's age-specific weights to the study population's rates; use the 2000 U.S. standard population for comparability with national statistics.

Crude rate reflects actual burden; age-adjusted rate allows comparison across populations with different age structures; both should be reported when describing burden.

Relative survival = observed survival / expected survival of a matched general population; it estimates the excess mortality attributable to cancer without requiring cause of death.

Observed survival requires accurate vital status follow-up; cases lost to follow-up bias survival upward if they are healthier, so completeness of follow-up must be documented.

Prevalence counts people alive with a cancer history at a point in time; it depends on incidence, survival, and time since diagnosis, so it is not a substitute for incidence.

Stage distribution, histology distribution, and treatment pattern analyses are descriptive; inferential comparisons require attention to small numbers (suppression thresholds, often <16 cases) and confidence intervals.

Public health reporting products (e.g., state cancer profiles, national incidence reports) lag diagnosis by years because of reporting, consolidation, and follow-up requirements; interpret trends with that lag in mind.