Skip to main content
v2026.11,986 entries · CC-BY 4.0

NCDB: National Cancer Database Research Guide

NCDB research starts with the right access route: distinguish PUF files, institutional reports, cohort preparation and publication requirements.

Written and maintained by CASRAI Editorial Board

Published

On this page

NCDB, the National Cancer Database, is a hospital-based oncology database used to study cancer care and outcomes. For researchers, the key distinction is between institutional reporting tools and the Participant User File (PUF): a de-identified research extract available through an application process. This guide connects those access routes with file preparation and manuscript reporting requirements.

What NCDB is

NCDB brings together cancer registry records rather than journal articles or research profiles. Its records describe patients, tumors, treatment and outcomes. It is a clinical data registry, so using it means understanding how cases were collected and coded before interpreting an analysis. A search for publications mentioning NCDB is a different task from obtaining its underlying research data.

The American College of Surgeons (ACS) presents several ways to use the resource: comparative reporting, quality improvement applications, registry submission services and research files. These serve different tasks. A hospital’s comparison report is not interchangeable with an investigator’s patient-level extract, even though both draw on the NCDB infrastructure. Start by identifying which output your project actually requires.

The official NCDB landing page links to the Quality Portal, or QPort, and the Rapid Cancer Reporting System, or RCRS. It also identifies public benchmarking resources separately from PUF applications. Use those distinctions when describing access in a protocol: a public reporting page does not establish that the underlying patient-level file is openly downloadable.

Who runs NCDB and what it covers

NCDB is a joint program of the Commission on Cancer (CoC) of the American College of Surgeons and the American Cancer Society. ACS describes participation by more than 1,500 accredited cancer programs in the United States and Puerto Rico. Its overview says data include patient characteristics, cancer stage, tumor histology, first-course treatment and outcomes, collected using standardized registry definitions and transmission specifications.

Keep the overall database separate from a particular PUF release. The March 2026 Getting Started with the 2024 PUF Data document describes diagnosis years 2004–2024, organ-specific files defined through primary-site and histology combinations, and exclusions for Veterans Affairs facilities and facilities in Puerto Rico. Those extract rules matter more to a study’s denominator than a headline description of the whole database.

On its PUF page, ACS describes the files as de-identified patient-level data that do not identify patients, providers or hospitals. Only a subset of the fields in the underlying registry manuals appears in the research extract. Do not promise a collaborator a variable merely because it exists in a registry collection standard; establish that it is included in the PUF documentation for the release you will use.

How to use NCDB for a research project

The following workflow follows the official PUF application and preparation documents. It covers project planning and analysis preparation; the sign-in is available through the PUF page’s Application link. The documented research route is an application followed by file delivery and local analysis.

  1. Establish eligibility and program support. The June 2026 application instructions require association with a CoC-accredited program and a support letter from the Cancer Committee Chair. If the principal investigator is the chair, the Cancer Program Administrator supplies the letter.
  2. Match the question to the available file. Consult Organ Sites Available, the PUF Data Dictionary and Getting Started documentation. Define the disease site, age cohort and diagnosis years before drafting a detailed analysis plan. ACS asks applicants to name the variables central to feasibility.
  3. Prepare the application team and proposal. Required elements include the project title, investigator information, a data analyst at the principal investigator’s facility, research objectives, background and analysis plan. Use the PUF Application Manager for the application; do not substitute a general institutional data request.
  4. Complete the agreement before obtaining files. Accepted applicants receive download notification, and the principal investigator must electronically sign the Data Use Agreement before files become available. The June 2026 instructions specify a one-month download window following notification.
  5. Import with release-matched documentation. The Quick Start guide supplies item names, variable names, field lengths, character positions and diagnosis-year availability. Its SAS script reads the flat text and assigns names, labels and formats. Other statistical software can use the published layout.
  6. Describe the cohort before fitting models. Getting Started recommends preliminary frequency distributions, including primary site and histology, to establish inclusion and exclusion criteria. Document the selected years and definitions alongside the code that implements them.

Applications are open year-round according to the PUF page, and releases are named for their latest included diagnosis year. Record the release designation explicitly. A project called “2024 PUF” does not mean every record was diagnosed in 2024, and its supporting documentation should not be silently exchanged for a different edition.

Reading the file layout correctly

The 2024 Quick Start layout places YEAR_OF_DIAGNOSIS in character positions 79–82, PRIMARY_SITE in 83–86 and HISTOLOGY in 88–91. These are import positions for that release, not universal column numbers for every NCDB file. A shifted import can produce plausible-looking values assigned to the wrong variables.

Use a data dictionary and its variable definitions together with the physical layout. The layout establishes where a field sits; the dictionary establishes what its codes mean. As an editorial preparation check, keep the release label, import script and dictionary reference together so another authorized analyst can reconstruct the same starting dataset.

An illustrative feasibility check

Illustrative composite, not an actual study: a research team proposes to compare treatment patterns for a defined cancer site over several diagnosis years. Before specifying its model, the team writes down the requested organ file, eligible histologies, year range and required treatment variables. It then checks the release documentation against that list and runs initial frequencies.

This produces a concrete project artifact: a small inclusion-rule sheet linking each proposed criterion to a documented field. It also exposes an important planning question early: does the extract contain the outcome the team wants to analyze? Changing the question at this stage is more defensible than inventing a proxy after seeing results. This worksheet is an editorial planning aid, not an additional ACS application requirement.

Strengths and limits

NCDB’s strength is the connection between institutional cancer registration and structured outcomes research. The official program description emphasizes comparisons of care and quality improvement. Its research files support investigator-defined analyses, while its reporting applications provide established institutional outputs. That breadth is useful only when the researcher names the relevant product and understands its scope.

The May 2026 PUF FAQ states that NCDB is hospital-based, not population-based, and cannot be used to calculate incidence. It also says the PUF excludes recurrence data and salvage therapy. Those are constraints on what a project can ask, not shortcomings that a larger sample automatically fixes. For national surveillance questions, see the distinction explained in the U.S. Cancer Statistics guide.

The FAQ also warns against combining older and newer PUF versions: cases can be updated or duplicated, facility identifiers change, and variables can change across editions. It says multiple primary cancers belonging to the same patient cannot be linked. Treat the released file as a specific research product rather than an annual increment to append automatically to an existing dataset.

Data access is constrained too. The FAQ says people outside the investigator’s facility may not access the PUF. A multidisciplinary collaboration therefore needs to distinguish intellectual participation from permission to receive records. An analyst’s technical ability to open a file does not establish that the agreement permits sharing it.

Use this table to distinguish outputs rather than rank databases. Each route answers a different kind of question, and an institutional reporting application is not a substitute for a research extract.

Resource Documented output or scope Access context
NCDB PUF De-identified patient-level files for investigator analyses Application for investigators associated with CoC-accredited programs
Hospital Comparison Benchmark Reports Descriptive demographic, treatment and stage comparisons NCDB institutional reporting through QPort
NCDB Survival Reports Overall, unadjusted observed survival stratified by selected characteristics NCDB institutional reporting through QPort
NCI SEER data Population-based cancer incidence registry data Research and Research Plus products have distinct access requirements

NCI describes SEER as collecting incidence data from population-based registries. Its Research Data and Research Plus/NCCR products differ in included information and authentication requirements; analyses use the SEER*Stat access route described by NCI. SEER therefore merits consideration when the question concerns population-based incidence, without assuming its fields, coverage or permissions match an NCDB PUF.

Within NCDB, ACS also lists the Cancer Quality Improvement Program, an annual PowerPoint report drawing on several reporting tools. The downloadable data table below makes a separate crosswalk between PUF documents and their preparation and reporting controls. It is a compilation of official descriptions, not a table of patient records or a claim that all tools provide equivalent analyses.

Interpreting an institutional comparison

ACS describes Survival Reports as overall, unadjusted observed survival, while its description of CQIP includes risk-adjusted survival reports. Preserve those labels when presenting a result. An unadjusted survival display and a risk-adjusted output answer different analytical questions; do not describe one as the other. ACS also says the reporting tools use a “best stage” combining pathologic stage over clinical stage. Record the report’s stated staging basis rather than assuming it matches a single stage variable selected for a separate PUF analysis. This is a concrete reason to name the reporting product in a methods description.

Preparing an NCDB manuscript

The 2024 Quick Start guide requires acknowledgement that the data derive from a de-identified NCDB file, that ACS and CoC have not verified the investigators’ methods or conclusions, and that NCDB is their joint project with the American Cancer Society. Its opening page supplies the exact acknowledgement wording. Use that official text when preparing a submission rather than substituting a software-style citation.

Getting Started also identifies the institutional Business Associate Agreement and data use arrangement among the information to convey in peer-reviewed publications. Distinguish acknowledgement of the data provider from endorsement of the analysis. Naming NCDB does not transfer responsibility for coding decisions or statistical interpretation to its operators.

The PUF FAQ prohibits reporting cells with fewer than ten cases. Address disclosure control when designing result tables, including whether totals would reveal a suppressed value. As an editorial reporting checklist, identify the file edition, requested site, diagnosis years, selection rules and variable definitions, then check the manuscript against the agreement’s presentation conditions. Permission to analyze and permission to display a particular result are separate checks.

Sources

Data: NCDB PUF documentation crosswalk for research and reporting

An original crosswalk of four ACS NCDB documents, connecting each documented research task to its concrete control or requirement. Compiled from the application instructions, 2024 preparation and layout documents, and PUF FAQ; contains no patient data.

NCDB PUF documentation crosswalk for research and reporting
DocumentResearch taskDocumented control or requirementSource URL
How to Apply (June 2026)Prepare the applicationSupport letter and a data analyst at the principal investigator facilityhttps://www.facs.org/media/iwklez3q/instructions-to-potential-ncdb-puf-applicants.pdf
Getting Started with the 2024 PUFDefine the analytic cohortCheck primary site and histology frequencies and diagnosis-year availabilityhttps://www.facs.org/media/5fun25np/getting-started-with-the-2024-puf.pdf
Quick Start Data Structure (2024)Import the research fileUse release-specific character positions, lengths and variable labelshttps://www.facs.org/media/lx4njye1/ncdb-puf-quickstart-data-structure-2024.pdf
PUF FAQ (May 2026)Prepare publication tablesDo not report cells containing fewer than ten caseshttps://www.facs.org/media/5xukuwub/ndcb-puf-frequently-asked-questions.pdf

Sources

Compiled by CASRAI from the sources above. Free to reuse under CC BY 4.0 with attribution.

Follow CASRAI

Research-administration guidance, standards updates and independent tool reviews.

Ask CASRAI · free to try

Ask about NCDB: National Cancer Database Research Guide

2 free questions a day, no account.

An AI assistant for research: funder rules, compliance, methods and publishing. It links the sources behind its answers and labels answers that come from a web search.

Answers draw on CASRAI's guides and dictionary plus the federal and funder documents we index: Federal Register, Grants.gov, Regulations.gov and UKRI.

Works on this site and inside Claude, Cursor and the AI tools you already use.

Everything CASRAI publishes — this page, the dictionary, the guides and the news — stays open to read, with no account and no card.

Referenced across the research world

University of Cambridge logoColumbia University logoCrossref logoUniversity of Edinburgh logoHarvard University logoUniversity of Oxford logoPrinceton University logoStanford School of Medicine logoUniversity College London logoORCID logoUniversity of Cambridge logoColumbia University logoCrossref logoUniversity of Edinburgh logoHarvard University logoUniversity of Oxford logoPrinceton University logoStanford School of Medicine logoUniversity College London logoORCID logo
  • University of Cambridge logo
  • Columbia University logo
  • Crossref logo
  • University of Edinburgh logo
  • Harvard University logo
  • University of Oxford logo
  • Princeton University logo
  • Stanford School of Medicine logo
  • University College London logo
  • ORCID logo

View CASRAI adoption →