SystemDoc_2007_main_NPCRCSS.pdf

National Program of Cancer Registries

Cancer Surveillance System

System Documentation

Version 7.0

(Submission Year 2007)

August 2007 prepared for

Division of Cancer Prevention and Control

Centers for Disease Control and Prevention by

7315 WISCONSIN AVENUE • SUITE 400W • BETHESDA, MD 20814

NPCR-CSS August 2007

System Documentation, Version 7.0 i

CONTENTS

1. INTRODUCTION

1.1. Overview

1.2. Purpose of This System Documentation

1.3. System Architecture

2. SYSTEM OVERVIEW

2.1. Design Principles

2.2. Security and Confidentiality

2.2.1. Data Flow

2.2.2. Data Integrity

2.3. Data Submission System

2.3.1. Overview

2.3.2. Data Submission System Components

2.4. Pre-Edit Verification System

2.5. Cancer Database

2.5.1. Description

2.6. Variables for Reporting Years 1995–2005

2.7. Data Extraction Utility

2.8. Data Loading and Editing System

2.8.1. Description

2.8.2. Inter-Record Edits Utility

2.8.3. Data Edits and Corresponding Edit Codes

2.9. Monitoring Database

2.9.1. Description

2.9.2. Monitoring Database Screens

2.9.3. Submission Summary Report

2.9.4. Summary Variables

2.10. Data Management System

2.10.1. Process and Methods for Generating the Standard Status Report and Analytical File .. 68

2.10.1.1. Nonreportable Cases

2.10.1.2. Completeness of Case Ascertainment and Standard Status Reports

2.10.1.3. Analytic Data

2.11. SQL Subsystem

2.11.1. Source Tables

2.11.2. Report Tables

2.11.3. Data Views

2.11.3.1. Case Completeness View

2.11.3.2. Case Completeness Total View

2.11.3.3. Case Completeness Summary View

2.11.3.4. Standard Status View

2.11.3.5. Program Average View

2.12. SAS Parallel Subsystem

2.12.1. Standard Status Report

2.13. VB Subsystem

2.14. Additional Reports and Data Products

2.14.1. SEER*Stat Database

ii

APPENDIXES

A. Cancer Database Data Dictionary

B. Cancer Database Variable Codes

C. Core Edit Reports—Sample

D. Advanced Edit Reports—Sample

E. Inter-Record Edits Detailed Report—Sample

F. Inter-Record Edits Summary Report—Sample

G. Monitoring Database Data Dictionary

H. Cancer Database—Edit Flags

I. SAS-VB Process Flow Diagram

J. Data Evaluation Report—Sample

K. SEER Data Process

LIST OF DIAGRAMS

Diagram 1. National Program of Cancer Registries-Cancer Surveillance System data process flow Diagram 2. Flowchart illustrating the Pre-Edit Verification QA process Diagram 3. NPCR-CSS analytic file flow for the 2007 data submission Diagram 4. Automated system for the generation of Standard Status Reports Diagram 5. Standard Status Report SQL-VB process flow

LIST OF EXHIBITS

Exhibit 1. Login screen Exhibit 2. Welcome screen Exhibit 3. Your Profile screen Exhibit 4. Directory Contents screen Exhibit 5. Document Upload screen Exhibit 6. Extraction Utility main screen Exhibit 7. File Browsing dialog box Exhibit 8. Extraction Utility message box Exhibit 9. NPCR Inter-Record Edits Utility Welcome screen Exhibit 10. NPCR Inter-Record Edits Utility main screen Exhibit 11. NPCR Inter-Record Edits Utility Browse screen Exhibit 12. Inter-Record Edits Utility message box Exhibit 13. Login screen Exhibit 14. State Profile screen Exhibit 15. State Profile—Sample screen Exhibit 16. Submission Summary screen Exhibit 17. 2007 Submission Profile—Sample screen Exhibit 18. 2007 Edit Information—Sample screen Exhibit 19. 2007 Data Submission & Processing Summary—Sample screen Exhibit 20. Submission Tracking screen Exhibit 21. Data Standards screen Exhibit 22. 2007 Data Standards—NPCR 24 Month Standard screen Exhibit 23. Data Standards—Sample screen Exhibit 24. 2007 NPCR-CSS 24+ Month Standard: 2002 Diagnosis Year screen iii

Exhibit 25. Contacts Search screen Exhibit 26. Contacts—Search Results screen Exhibit 27. New Contact screen Exhibit 28. State Inquiries & Events screen Exhibit 29. Edit State Profile—Sample screen Exhibit 30. New Organization screen Exhibit 31. Request Tracking screen Exhibit 32. Make a Request screen Exhibit 33. Data Dictionary screen Exhibit 34. Data Dictionary Search screen Exhibit 35. Data Dictionary Results screen Exhibit 36. Item Detail screen Exhibit 37. NPCR-CSS Data Submission Forms Exhibit 38. Document Server Data Submission Forms—Sample Submission Exhibit 39. Data Items Transmitted—Sample screen Exhibit 40. Data Set Participation Agreement—Sample screen

LIST OF TABLES

Table 1. Patient-identifying variables not accepted by NPCR-CSS Table 2. Reportable variables for the 2007 CSS data submission Table 3. NPCR-CSS data edits and corresponding Commission on Cancer, NAACCR, and SEER edit codes Table 4. Summary variables posted to the Monitoring DB during the final phase of the Edit program run Table 5. Data sources for 2007 submission Table 6. Summary of mortality rates output file Table 7. Data Views

1. INTRODUCTION

1.1. OVERVIEW

The National Program of Cancer Registries Cancer Surveillance System (NPCR-CSS) is the result of an initiative by the Centers for Disease Control and Prevention (CDC) to create a system that collects and reports data from NPCR States. The CDC-funded NPCR is a population-based system of cancer registries established in 1992 by the Cancer Registries

Amendment Act (Public Law 102-515). When fully implemented, States funded by NPCR will collect data on cancer for 96 percent of the U.S. population. As of January 2007, 45 States, 2 territories, and the District of Columbia submitted data to NPCR-CSS.

NPCR-CSS allows NPCR-funded State central cancer registries to submit cancer incidence data annually to CDC. State-specific cancer registry data that are currently reported allow CDC to:

Assess the quality, completeness, and timeliness of data collected by the NPCR-CSS central cancer registries

Summarize and report cancer surveillance data that will be available to public health agencies, health advocacy groups, and researchers

Use enhanced data to generate State-specific data reports for the States for continuous quality improvement and program evaluation. The information from these assessments will identify for the individual States and CDC areas that require education, targeted technical assistance, and training.

NPCR-CSS also provides:

Greater access to cancer data for policymakers, scientists, and the general public

(e.g., public-use and restricted-access data files of cancer incidence)

More accurate and stable estimates of cancer incidence for population groups, including racial and ethnic minorities, and medically underserved groups

Information for NPCR-funded States, such as regional and national analyses to more accurately identify geographic variability in cancer treatment practices to assess the use of state-of-the-art cancer treatments

The data in the NPCR-CSS database are used to:

Identify areas in the United States with high proportions of women diagnosed with late-stage breast cancer or invasive cervical cancer

Provide State-specific cancer prevention and control management reports

Guide decisions about program resources and early detection efforts so that resources can be better allocated and targeted to areas of need

Expand efforts to assess the patterns and quality of cancer diagnosis and treatment provided to women participating in breast and cervical cancer screening programs

Identify regions or States with high rates of malignant melanoma in order to better target communities for prevention messages and education campaigns

Identify areas in the United States with high proportions of men and women diagnosed with late-stage colorectal cancer

Develop, implement, and evaluate new models of comprehensive, integrated approaches to cancer prevention and control

1.2. PURPOSE OF THIS SYSTEM DOCUMENTATION

The purpose of this system documentation is to provide a comprehensive overview of

NPCR-CSS and its technical and functional design and provide examples of system reports. The system documentation gives users the information they need to understand the system’s operation, including the process involved in collecting, processing, and reporting cancer data.

This document also details the procedures and precautions taken by Macro International Inc. to ensure the security of the databases and the confidentiality of the cancer data.

1.3. SYSTEM ARCHITECTURE

NPCR-CSS is a database-driven system designed to collect, record, and analyze patient cancer data and generate statistical outputs and reports on cancer incidence throughout the United

States. The system’s two main databases are:

NPCR-CSS Cancer Database (Cancer DB)—This is the main repository for all cancer data submitted by the NPCR-funded State cancer registries to CDC. All programming and processing routines and validation checks run on the data are performed in the Cancer DB.

NPCR-CSS Monitoring Database (Monitoring DB)—This houses the summary cancer statistics, which are derived from the data submissions stored in the Cancer DB. It also houses the contact and profile information for the NPCR-funded State cancer registries.

The system features a secure password-protected document server through which NPCR-funded

State cancer registries submit their data. Data received from State cancer registries are processed through the Pre-Edit Verification system, a SAS-based program that performs an initial validation check. This initial validation check determines the file record layout version, checks the consistency of data variables, identifies disqualifying data elements (such as the confidential data fields), and provides summary statistics on the contents of submitted data to ensure that the case counts across submission years are similar. These summary statistics are used by quality assurance (QA) specialists to visually check any anomalies. If submitted data contain disqualifying elements or if data variables are inconsistent, the data file is manually deleted and the State is asked to resubmit its data. If any confidential fields are included in the submitted data, the entire data file is deleted and the State is asked to resubmit its data. Submitted data that have been successfully validated are loaded into the Cancer DB.

NPCR-CSS runs several processing routines on the successfully validated data that have been loaded into the Cancer DB. These processing routines are performed by a series of command line executable programs and assess data quality and completeness and generate reports on cancer incidence and other statistics. The processing routines run on the Cancer DB records fall into two general categories: edits and validation, and output reporting. These processes are explained in detail in sections 2.8 and 2.10.

2. SYSTEM OVERVIEW

2.1. DESIGN PRINCIPLES

NPCR-CSS is designed with an emphasis on maintaining the security of cancer data submitted by State cancer registries and protecting patient confidentiality. The system uses a database-driven design to support its data collection, editing, processing, and reporting functions. Each of the system components was designed and implemented using the technical approach best suited for the intended output.

The NPCR-CSS system components consist of five secure servers on a local area network

(LAN)

located in Macro’s Bethesda, MD, offices:

Database Server—A Microsoft (MS) SQL Server 2000 that houses the Cancer, Data

Submission Monitoring, and Document Server databases

SAS Application Server—A server that houses the SAS software. SAS programs are executed on this server for data processing and data analysis tasks.

Web Server—A Web server that houses the NPCR-CSS secure hypertext transfer protocol

(HTTPS) Web sites (Document Server and Data Submission Monitoring System)

Firewall Server—A server that houses the firewall software

Log Server—A central repository for storing logs generated by all the servers on the CDC

LAN at Macro

The database components were designed and built using MS SQL Server 2000 database tools.

The user interfaces were programmed using a combination of hypertext markup language

(HTML), ColdFusion Markup Language, ColdFusion MX, and JavaScript. Access to both the databases and the interfaces is secured with user ID and password protection.

2.2. SECURITY AND CONFIDENTIALITY

NPCR-CSS is designed to ensure the security and confidentiality of all incoming cancer data.

Access to the databases and Web-based interfaces is user ID and password protected, and all data are encrypted during transmission. The system validates all incoming data using a SAS programming routine that checks the data records to ensure that they do not contain any confidential, patient-identifying data. Records containing personal identifiers do not pass validation and are deleted from the Cancer Server.

All data processing functions (e.g., validation, editing, report generation) are performed on the secure Database Server housing the Cancer DB. Results of the processing (e.g., summary statistics, contact data) are posted to the Monitoring DB, where they can be viewed through the

A geographically limited communications network allowing interconnection of computer terminals within adjacent physical locations Software that accesses, manages, analyzes, and presents data

A dedicated gateway with special security precautions, used to service outside network connections

Web-based interface. However, the data conduit between the Cancer DB and the Monitoring DB is strictly one-way; the Monitoring DB cannot send data back to the Cancer DB, thus minimizing the number of open channels to the Cancer DB, limiting opportunities for outside access to the secure server, and maintaining the security of the database and the confidentiality of patient data.

(For detailed information on system security and data security and confidentiality, see the

NPCR-CSS Project Security Plan Documentation.)

2.2.1. Data Flow

State cancer registries submit encrypted data to a secure document server designated for data submission. Within 1 hour of receipt, the submitted data are transferred to a temporary project server by project staff. Data are then evaluated by the Pre-Edit Verification process, a series of validation routines performed on the data to ensure their integrity before they are loaded into the

Cancer DB. This process is performed by running a SAS program that verifies the record layout version of the data file and the data format, identifies all patient-identifying data fields, and increasingly importantly, checks data quality to determine whether a resubmission is required

(see diagram 1).

Using a set of SAS programs, the Pre-Edit Verification process checks that all records in the data file conform to the data standards established by the North American Association of Central

Cancer Registries (NAACCR), an umbrella organization for cancer registries, Government agencies, professional associations, and private groups in North America interested in enhancing the quality and use of cancer registry data. It also identifies any personal, patient-identifying data that would compromise data security or patient confidentiality and checks data quality against a set of pre-established benchmarks. There are two main outcomes of the validation process:

Data files containing patient-identifying data (i.e., variables appearing in columns 1947 and after in NAACCR standards version 11) are deleted from the temporary server to maintain security and confidentiality; the submitting State is then instructed to remove the identifying information from the dataset and resubmit the data for processing. Any data inconsistencies

(such as invalid codes or data format) are also reconciled with the State, and a request is made for resubmission.

Data quality review outputs check whether all requested data items were included in the State submission and the reasonableness of count variations across submissions (see section 2.4 for details).

Data passing the Pre-Edit Verification process are loaded into the Cancer DB. (For detailed information on the data dictionary and data variables, see appendixes A and B, respectively).

Once the data are loaded into the Cancer DB, the Edits program checks each record to determine its status regarding quality and reportability. The Edits program runs sets of core, advanced, inter-record, and recommended edits against the reportable cases. Edit pass/fail flags are set in the Cancer DB indicating whether the records have passed the edit criteria. The Edits program generates sets of Detailed and Summary Edit Reports that are posted on the Production Server

(Summary Edit Reports are also displayed on the Monitoring DB on the Production Server).

Upon completion of the edit reports, the Submission Summary Reports are produced. These reports provide a detailed overview of the edit reports and an explanation of the use and appropriateness of override flags. The Submission Summary Reports:

Highlight trends in the quality of cancer data from the NPCR-funded State cancer registries across several statistical markers

Identify patterns in each State’s data submissions

The next step in data processing is the Post-Edit Verification process. A program is run against the data in the Cancer DB to examine inconsistencies identified by the edit flags. All edit flags are then reviewed against the summary statistics in the Monitoring DB. This process identifies reportable cases from all submitted cases. Additional edit flags are also set to identify those records with missing information in critical fields such as age, sex, race, and county; such records are excluded from the incidence rate calculation. Nonreportable cases are excluded from the data-quality reporting and further data analyses but are preserved in the database.

Nonreportable cases are assigned a label value of ―0.‖ Records that pass all edits are assigned a value of ―1‖ and corresponding flags are set in the Cancer DB.

Following completion of the Post-Edit Verification process, cancer incidence rates for each of the major sites for Black males, Black females, White males, and White females in each State are run using the SQL and SAS programs on the Cancer Server. These incidence rates are used to generate the following analytic reports and data products:

Standard Status Reports (SSR1)—These reports present statistics on data quality, completeness of case ascertainment, and timeliness. They detail the total number of reported cases and number of cases qualified to be included in the incidence rate calculation; the percentage of records with missing information in the fields of age, sex, race, and county;

rates of death certificate only cases; duplicate cases identified from data submission forms;

and adjusted completeness rates for each submitted year of data. See section 2.10 for details.

Submission Summary Reports (SSR2)—These reports summarize issues relating to data quality of the 2007 data submissions from each State, particularly the results from running the Edits program. See section 2.9.3 and appendix H for details.

Data Quality Indicator Report (DQI)—These reports provide feedback to the States assessing the quality and completeness of their cancer data as part of the continuing effort to improve national cancer surveillance through enhanced registry operations and research.

See appendix J for samples of SSR1, SSR2, and DQI.

DRAFT

System Documentation, Version 7.0

Diagram 1. National Program of Cancer Registries-Cancer Surveillance System data process flow

Data removed automatically from the Web Server to the Cancer Server upon submission

QA enter submission form information into MDB

DP runs Pre-Edit Verification program

QA contacts State for reconciliation

PD archives file to CD-ROM and loads into CDB

No

Yes

DP runs critical, noncritical, and inter-record Edits

DP posts appropriate flags to CDB

DP posts reports on the Production Server/MDB

QA reviews edit reports/checks flags in ACCESS

Yes

DP performs diagnostic test, fixes problems, and re-runs Edits

DP runs incidence rates

DP runs/QA completes SSR2 (Submission Summary

Reports)

DP runs SSR1 with Case Ascertainment Completeness & Program Comparison reports

DP runs reports on compliance with NPCR standards, frequencies, and data quality indicators

CDC

review

PD–Program Director CDB–Cancer Database

QA–Quality Assurance Team MDB–Monitoring Database

DP–Data Processing Team SSL–Secure Sockets Layer

QA verifies version, data format, columns, personal identifiers

DP runs post-edit verification on

CDB

No

DP runs Data Variable Frequency reports

DP runs Official Federal Cancer Statistics

DP runs SEER*Stat files

DP runs State Can*Ques files

QA contacts States/Provides ID and Password

Pass

Pass QA review

State registry sends data to the Web Server via the SSL

DP/PD run ad hoc reports

Archived CD-ROM saved in secure offsite storage

PD final review

Reports to

States

State Resubmits

Products and Deliverables

2.2.2. Data Integrity

As noted in section 2.2.1, State-submitted data that have been verified and loaded into the Cancer

DB are processed through the edits system, in which a series of edit routines check the data for missing elements, duplicate information, logical inconsistencies, and other anomalies that compromise data integrity. There are three basic types of edits performed on data that have been loaded into the Cancer DB:

Core edits address data issues that are used to assess compliance with NPCR quality standards. They have a direct impact on the incidence of cancer and include age, sex, race, site, stage, and State at diagnosis.

Advanced and recommended edits address data issues that are not designated as core.

Inter-record edits address the consistency of demographic, diagnostic, and other relevant data information among patients with multiple cancers in the same and/or across multiple diagnosis years.

Reportable data records containing errors that do not nullify validation are flagged by the system for further data quality and statistical reports but are not excluded from the Cancer DB (see section 2.8).

2.3. DATA SUBMISSION SYSTEM

2.3.1. Overview

NPCR-funded State cancer registries submit encrypted data files to Macro’s secure Document

Server using VeriSign’s certificate to provide 128-bit encryption during the file transfer. The

Data Submission System features an HTML-based user interface, which is accessible using any standard Web browser capable of 128-bit encrypted communication. The data are submitted in a

NAACCR version 11, fixed-width, flat file format that is stripped of all patient-identifying information. Upon receipt the data are automatically placed behind Macro’s CDC LAN firewall where they cannot be accessed from external sources (e.g., the Internet). Each record is validated or assessed using a SAS routine. Only those records that contain no confidential patient-identifying data pass validation; if patient-identifying information is found, the entire submission is rejected (not just those records with patient-identifying data) and deleted from the server. Patient-identifying data are those that contain variables that appear following column

1946 in the NAACCR Standards for Cancer Registries, Data Standards and Data Dictionary record layout version 11 (see table 1).

Table 1. Patient-identifying variables not accepted by NPCR-CSS

Column Number Item Number Item Name Section Name

1947–1971 2230 Name—Last Patient—Confidential

1972–1985 2240 Name—First Patient—Confidential

1986–1999 2250 Name—Middle Patient—Confidential

2000–2002 2260 Name—Prefix Patient—Confidential

2003–2005 2270 Name—Suffix Patient—Confidential

2006–2020 2280 Name—Alias Patient—Confidential

2036–2085 2290 Name—Spouse/Parent Patient—Confidential

2086–2096 2300 Medical Record Number Patient—Confidential

2097–2098 2310 Military Record No Suffix Patient—Confidential

2099–2107 2320 Social Security Number Patient—Confidential

2108–2147 2330 Addr at DX—No & Street Patient—Confidential

2148–2187 2335 Addr at DX—Supplementl Patient—Confidential

2188–2227 2350 Addr Current—No & Street Patient—Confidential

2394–2403 2352 Latitude Patient—Confidential

2404–2414 2354 Longitude Patient—Confidential

2228–2267 2355 Addr Current—Supplementl Patient—Confidential

2268–2277 2360 Telephone Patient—Confidential

2278–2283 2380 DC State File Number Patient—Confidential

2021–2035 2390 Name—Maiden Patient—Confidential

2314–2353 2392 Follow-Up Contact—No & St Patient—Confidential

2354–2393 2393 Follow-Up Contact—Suppl Patient—Confidential

2284–2313 2394 Follow-Up Contact—Name Patient—Confidential

2485–2494 2410 Institution Referred From Hospital—Confidential

2495–2504 2420 Institution Referred To Hospital—Confidential

2465–2474 2430 Last Follow-Up Hospital Hospital—Confidential

2475–2484 2440 Following Registry Hospital—Confidential

2555–2562 2460 Physician—Managing Other—Confidential

2563–2570 2470 Physician—Follow-Up Other—Confidential

2571–2578 2480 Physician—Primary Surg Other—Confidential

2579–2586 2490 Physician 3 Other—Confidential

2587–2594 2500 Physician 4 Other—Confidential

2.3.2. Data Submission System Components

States submit their data to NPCR-CSS using Macro’s Document Server through an encrypted

HTTPS link. The Document Server resides inside Macro’s firewall at its Bethesda, MD, offices;

it uses VeriSign’s certificate to provide 128-bit encryption during file transfer and is accessible only through HTTPS. The Document Server is also used to provide data quality and completeness reports for NPCR-CSS, NPCR data submission packets, NPCR utilities, and other

NPCR data reports and products to States. Each State has its own directory on the Document

Server and cannot see or download files from other States.

Once a data file is uploaded to the Document Server, the Macro data processing team is automatically notified via e-mail. The file is then moved to the Cancer Server. Only authorized

Macro project personnel have user IDs and passwords that give them access to the Document

Server.

To access the Document Server:

1. Go to http://www.npcrcss.org/. Depending on the browser version, one or more security alert messages may appear. Acknowledge the security alerts if necessary and click on Next.

2. In the Login screen, enter the State’s user ID and password (see exhibit 1).

Macro has provided each submitting State with a user ID and password. The user ID is the

State’s name. The password is randomly generated and contains at least one nonalphanumeric character. States can change their passwords after logging in. The user ID and password are not case-sensitive. All State passwords are kept on file, even those that have been changed by the user. If a user forgets the password for his or her State, the current password can be retrieved or a new password can be issued.

Exhibit 1. Login screen

3. Click on Login to continue to the Welcome screen (see exhibit 2).

On the left side of the Welcome screen is a directory tree containing a list of documents that are available for download, mostly data submissions or data quality and status reports.

Exhibit 2. Welcome screen

4. Click on Your Profile to continue to the Your Profile screen (see exhibit 3).

Complete the appropriate fields for the changes you want to make, then click on Update

Your Profile to apply the changes and continue to the Users screen to verify the changes.

Click on Cancel to discard the changes and keep the existing information.

You can change the State’s password, e-mail address, and other contact information on this screen.

Exhibit 3. Your Profile screen

5. On the left side of the screen, click on the name of the State to view the contents of that

State’s folder (see exhibit 4).

Exhibit 4. Directory Contents screen

6. To upload files to the State folder, click on Upload a File. This opens the Document Upload screen (see exhibit 5).

Exhibit 5. Document Upload screen

7. To upload a document:

Enter a title in the Document Title field.

Enter the name of the file in the Document Location field or click on Browse to locate the file to be uploaded from your hard drive.

Enter any text that you would like to send to the folder’s subscribers in the Enter accompanying note here (optional) field.

Click on Continue to upload the document.

8. When the file has been successfully uploaded, the Results screen will display the document profile.

9. After the file has been uploaded, click on Logout to return to the Login screen.

2.4. PRE-EDIT VERIFICATION SYSTEM

The Pre-Edit Verification system ensures that data submitted from NPCR States meet certain quality control standards. By determining whether a State’s submission is valid and acceptable or whether a resubmission is necessary, the Pre-Edit Verification system aims to prevent erroneous submissions from being processed further.

The Pre-Edit Verification system uses SAS tools to address three aspects of data reviews/checks during the pre-loading and pre-edit stage:

Data confidentiality: Macro developed a SAS program to determine whether the data submission contains any personal or patient-identifying information. If patient-identifying information is sent, the entire submission is rejected (not just those records with patient-identifying data) and deleted from the server to ensure security and confidentiality of patient data. Patient-identifying data are those variables in columns 1947 and after of the NAACCR record layout.

Data validity: The SAS program also checks the format of all data variables and determines whether the data records conform with NAACCR standards as specified by the NPCR Data

Submission Packet (e.g., version 11 standard). If the submission does not comply with the required version, the entire submission is rejected. The SAS program further checks the

NPCR-requested data elements for data evaluation, paying particular attention to the critical/required ones.

Data quality (reasonableness): A set of lookup tables have been generated based on data submissions in prior years. The SAS program generates tabular reports based on pre-defined quality control/QA criteria to see whether the submitted data show a systematic anomaly in case count between current and prior submissions, across all diagnosis years. The QA criteria are a set of the threshold values for anomalies based on historical trend data, including indicators such as cancer sites under evaluation, years of diagnosis, county of diagnosis, and other factors that are deemed necessary after consultation with CDC project officers.

All pre-edit reports are run soon after the data have been received, and the reports are reviewed by QA staff immediately to determine the acceptability of the submission. Any data or version issues detected by the Pre-Edit Verification system are discussed and reconciled with the submitting State. For those States with identified problems in their data, Macro QA staff coordinate with CDC program consultants to find a solution (e.g., resubmission). All actions and processes are documented in the Monitoring DB. Diagram 2 illustrates the interactive process for the Pre-Edit Verification system.

All data submitted from the State go through the Pre-Edit Verification process; no exclusion criteria are set at this stage. Although all cases submitted are checked, only reportable cases are used in the case count comparison. The previous year’s submission is used as a benchmark for specific year of diagnosis and site recode. Any abnormality exceeding the pre-defined threshold will be flagged. For any case count greater than or equal to 30, the record will be flagged if the new submission’s count increases more than 20 percent or decreases more than 10 percent compared with those from the previous year. For each year of diagnosis, the Pre-Edit

Verification system also checks the case distribution by quarter.

Data submissions that pass the Pre-Edit Verification checks will be loaded into the Cancer DB and processed through the Data Loading and Editing System.

Diagram 2. Flowchart illustrating the Pre-Edit Verification QA process

(SAS Automation + Human Interactions)

2.5. CANCER DATABASE

2.5.1. Description

The Cancer DB is the primary repository for the raw data submitted to CDC by the

NPCR-funded State cancer registries. It is actually a series of databases, one for each submission year; the Cancer DB is designed as a secure system for the collection and storage of cancer surveillance data.

The Cancer DB design is based on the record structure and data requirements designated by

NAACCR, which developed a set of record standards for organizations collecting and reporting cancer data in order to facilitate electronic sharing of data. These data exchange record layouts are designed to standardize data formats from reporting sources to and between central registries.

The NAACCR data standards consist of all data items required and recommended for use by the major cancer standard-setting organizations and encompass any data that an individual registry might collect. It is unlikely, however, that any single registry will collect data on all the items included in the record layout.

The full list of NAACCR requirements and registration standards, including a detailed data dictionary with full text descriptions of the data items and a rationale for their collection, can be found on NAACCR’s Web site at http://www.naaccr.org/index.asp?Col_SectionKey=7

&Col_ContentID=122.

2.6. VARIABLES FOR REPORTING YEARS 1995–2005

Table 2 contains a list of reportable variables that NPCR requires the States to include in their submissions. Variables in bold text are new data items for the 2007 CSS Data Submission. Not every record in a State-submitted data file will contain data for all these variables; however, a valid data file should contain only approved variables. Any fields in the NAACCR record layout version 11 containing nonreportable variables are left blank.

The Diagnosis Year Status 1995–2005 column contains all the variables approved for those diagnosis years. It should be noted that some of these data items will not be released to the public in any format or will be released only in a format that prevents direct or deductive identification of individual patients.

Table 2. Reportable variables for the 2007 CSS data submission

Data Items by Diagnosis Year Status

Record ID and Demographic Section—(Name and [Number]) Diagnosis Year Status

–2005

Record Type [10] Y

Patient ID Number [20] (unique) Y

Registry Type [30] Y

Registry ID [40] Y

NAACCR Record Version [50] Y

Address at DX—State [80] Y

County at DX [90] Y

Address at DX—Postal Code [100] Y

Census Tract 1970/80/90 [110] Y

Census Tract Coding Sys 1970/80/90 [120] Y

Census Tract 2000 [130]

Census Tract Certainty 1970/80/90 [364] Y

Census Tract Certainty 2000 [365]

Race 1 [160] Y

Race 2 [161] Y

Race 3 [162] Y

Race 4 [163] Y

Race 5 [164] Y

Spanish/Hispanic Origin [190] Y

NHIA

Derived Hispanic Origin [191]

IHS Link [192]

Computed Ethnicity [200] As Available

Computed Ethnicity Source [210] As Available

Sex [220] Y

Age at Diagnosis [230] Y

Birth Date [240] Y

Birthplace [250] Y

Primary Payer at Diagnosis [630] As Available

Cancer Identification Section—(Name and [Number])

Sequence Number—Central [380] Y

Date of Diagnosis [390] Y

Primary Site [400] Y

Laterality [410] Y

Grade [440] Y

Site Coding Sys—Current [450] Y

Morph Coding Sys—Current [470] Y

Diagnostic Confirmation [490] Y

Type of Reporting Source [500] Y

Histologic Type ICD-O-3 [522]

Behavior Code ICD-O-3 [523]

ICD-O-3 Conversion Flag [2116]

Treatment 1st Course Section—(Name and [Number])

Date of Initial Rx—SEER [1260]

Date of 1st Crs Rx—COC [1270]

RX Summ—Surg Primary Site [1290]

RX Summ—Scope Reg LN Sur [1292]

RX Summ—Surg Oth Reg/Dis [1294]

Reason for No Surgery [1340]

RX Summ-Radiation [1360]

RX Summ—Surg/Rad Seq [1380]

RX Summ—Chemo [1390]

RX Summ—Horm [1400]

RX Summ—BRM [1410]

RX Summ—Other [1420]

Rad—Regional RX Modality [1570]

RX Summ—Transplnt/Endocr [3250]

Stage/Prognostic Factors Section—(Name and [Number])

SEER Summary Stage 2000 [759]

SEER Summary Stage 1977 [760]

CS Extension [2810]

CS Lymph Nodes [2830]

CS Mets at DX [2850]

CS Site-Specific Factor 1 for C384 [2880]

CS Site-Specific Factor 3 for C619 [2900]

CS Version 1st [2935]

CS Version Latest [2936]

Derived SS2000 [3020]

Derived SS2000-Flag [3050]

Follow-up/Recurrence/Death Section—(Name and [Number])

Date of Last Contact [1750] Y

Vital Status [1760] Y

Follow-Up Source [1790] As Available

Cause of Death [1910] Y

ICD Revision Number [1920] Y

Over-rides/Conversion/System Admin. Section—(Name and [Number])

Over-ride Age/Site/Morph [1990] Y

Over-ride SeqNo/DxConf [2000] Y

Over-ride Site/Lat/Sequence Number [2010] Y

Over-ride Site/Type [2030] Y

Over-ride Histology [2040] Y

Over-ride Report Source [2050] Y

Over-ride Ill-define Site [2060] Y

Over-ride Leuk, Lymphoma [2070] Y

Over-ride Site/Behavior [2071] Y

Over-ride Site/Lat/Morph [2074] Y

Notes:

Shaded items denote variables that are considered advanced surveillance data.

Nonshaded variables denote core surveillance data.

Data items new to this submission or having a status change are noted in bold.

The data item names and numbers are those used in NAACCR Volume II, Version 11.

Status Key: Y=Yes

Leave blank for missing and not reported. Code ―999‖ for unknown and invalid. Do not include cases with code ―998‖ in the submission file. If State law precludes the registry from identifying specific counties on a file of individual records, the registry may recode all valid county codes to ―000.‖ Use of this field is recommended for reportable cases diagnosed in 1998 and later and required for reportable cases diagnosed in 2003 and later.

Report the results from the NAACCR Hispanic Identification Algorithm SAS Program. See the Submission Specifications document and/ or attachment 1 for more details. Please note that code 9, unknown, is not a valid code for this data item.

Report the results of cases submitted for Indian Health Service linkage using codes 0 or 1.

See the Submission Specifications document for further details.

This field is currently required as available.

For reportable cases diagnosed in 2001 or later, data should be coded using the

International Statistical Classification of Diseases (ICD)-O-3 manual, and data should be submitted with these original ICD-O-3 values. For reportable cases diagnosed before 2001, data should be coded using the ICD-O-2 manual. Prior to submission, convert these ICD-O-

2 codes to ICD-O-3 codes and use the ICD-O -3 Conversion Flag [2116] to document the conversion.

Submit either the SEER or COC field; for reportable cases diagnosed in 2003 or later, as available.

For reportable cases diagnosed in 2003 or later, as available.

For reportable cases diagnosed in 2000 or before, code using SEER Summary Staging

Guide 1977. For reportable cases diagnosed in 2001 and later, code using SEER Summary

Staging Manual 2000.

For reportable cases diagnosed in 2004 and later, this is used to derive, or is the derived, ―Summary Stage 2000‖ from the collaborative staging (CS) algorithm or extent of disease

(EOD) codes.

For reportable cases diagnosed 2004 and later, this item is the flag to indicate whether the ―Summary Stage 2000‖ was derived from the CS algorithm or EO D codes.

2.7. DATA EXTRACTION UTILITY

The NPCR-CSS Extraction Utility is a standalone program that States can use to extract stored data from their files for inclusion in their submission. The NPCR-CSS Extraction Utility program version 2.4 produces an extract file that contains only the variables required for reporting by NPCR-CSS. The result is a smaller, more manageable file that is easier to process and transmit. Any fields in the NAACCR record layout version 11 containing NPCR-CSS nonreportable variables will be filled with blank spaces.

Open the Extraction Utility main screen (see exhibit 6) using the icon on your computer’s desktop or the MS Windows Run command.

Exhibit 6. Extraction Utility main screen

Before running the program, you must select the input data file on which the program will run.

The Run button at the bottom left of the window will be inaccessible until a source file has been selected.

To select the input file:

Click on the Browse button next to the Input data file dialog box to open the file browsing dialog box where you will locate and select the input file (see exhibit 7). You can also enter the path and file name directly in the Input data file dialog box. Using the Browse button is recommended to avoid typing errors.

In the File name field of the file browsing dialog box, enter the name of the input file from which data will be extracted.

Exhibit 7. File Browsing dialog box

To select the output file:

Click on the Browse button next to the Output data file dialog box to open the file browsing dialog box where you will enter a name for the output file. You can also enter the path and file name directly in the Output data file dialog box. Using the Browse button is recommended to avoid typing errors. By default, the output file will be saved in the same directory that contains the input file. If you want to save the output data file to a different directory, you can use the file browsing window to choose a new directory (see exhibit 7).

After the input and output data files are selected, the utility will automatically return to the

NPCR-CSS Extraction Utility main screen. The Run button is now accessible.

Click on Run.

The number of extracted cases will be continuously displayed, and a blue progress bar will track the progress of the program as it processes each record.

When the process is complete, a message box appears stating how many records have been successfully extracted, as shown in exhibit 8.

Exhibit 8. Extraction Utility message box

Click on OK to close the message box. The data extraction is complete.

Click on Exit to exit the NPCR-CSS Extraction Utility.

2.8. DATA LOADING AND EDITING SYSTEM

2.8.1. Description

After the State cancer registries submit their data to Macro via HTTPS and the data are transferred to the temporary server, the Pre-Edit Verification program is run against the data, as illustrated previously in diagram 2.

All records that pass the Pre-Edit Verification process are loaded into the Cancer DB and processed through the Data Loading and Editing System, which consists of a series of MS Visual

Basic (VB) programming routines (collectively referred to as the Edits program) that are run against the data. The purpose of the Edits program is to verify the reportability and validity of the submitted cancer data and generate a series of summary statistics derived from those data. The subsequent Edits programs process the data and note any errors or inconsistencies within a data record (inter-field) and between data records (inter-record). Records containing errors are flagged by the system on all (core and advanced) data items.

The initial phase of the Edits program validates the results of the Pre-Edit Verification process by confirming that the records in the data file conform to version 11 of the NAACCR data standards. Any data items/records that do not conform to the NAACCR version 11 standards are flagged, and the flags are posted to the Data Submission Monitoring System DB. If the data records satisfy this requirement, they are passed on to the second phase of validation/edits.

At the beginning of the second validation phase, the Edits program checks each record to determine its reportability status. Nonreportable cases are excluded from the active dataset but are preserved in the Cancer DB.

Nonreportable cases are cases that meet any of the following criteria:

The State in which the cancer was diagnosed is different from the submission State.

The diagnosis year is earlier than the NPCR reference year for the State.

The cancer is benign or borderline.

The cancer is a carcinoma in situ of the cervix.

The cancer is a basal or squamous cell carcinoma of the skin.

The cancer is a prostatic intraepithelial neoplasia.

Records that satisfy any of these criteria are considered nonreportable. These records are flagged with a label value of ―0.‖ Records that do not meet these criteria are assigned a value of ―1‖ and corresponding flags are then set in the Cancer DB.

The Edits program then runs sets of core, advanced, and inter-record edits against the reportable cases. Edit pass/fail flags noting whether the records have passed the criteria for each edit are set in the Cancer DB.

Three types of edit checks are performed on the data:

Single-field edit—This class of edits applies to a single data field and serves as a validity check. For example, the Sex field should contain values only of ―1‖ (for male), ―2‖ (for female), ―3‖ (for hermaphrodite), ―4‖ (for transsexual), or ―9‖ (for unknown). If any other values are found in that field, the error will be detected by the single-field edit.

Inter-field edit—This class of edits checks for consistency. Inter-field edits address data issues that apply to more than one field but are within a single record. For example, if a record has a Sex field coded as ―male‖ and a Primary Site field coded as ―uterus,‖ the error will be detected by the inter-field edit.

Inter-record edit—This class of edits checks for internal consistency across multiple records for multiple sites for the same patient. For example, an inter-record edit will verify that the sequence numbers have been assigned in chronological order to the patient’s cancers.

The Edits program does not run edits on records that are marked with over-ride flags.

Edit over-ride flags are used for records that contain data that are contradictory or unusual but allowable. The inter-field edits identify possible errors that require manual review or resolution.

Once the contradictory or unusual data in a record are determined to be allowable, edit over-ride flags are set to eliminate the need to review the same record repeatedly. For example, if a State wanted to report a rare, but possible, case of prostate cancer in a 19-year-old man it must set an over-ride flag, since the inter-field edit routine would otherwise flag such a record as erroneous.

States set over-ride flags before data are submitted; the Edits program then verifies that valid codes were used to set the flags and counts the number of over-ride flags used for each error.

The edit routines run on reportable cases, generate the core, advanced, and inter-record edit reports that are posted on the production server, and produce the summary statistics that are passed on to the Monitoring DB.

2.8.2. Inter-Record Edits Utility

The Inter-Record Edits Utility is a program that allows cancer registries to run inter-record edits on their data. This utility reads a NAACCR version 11 data file and produces two inter-record edit reports—a Summary Report and a Detailed Report. The reports are produced in rich text format (RTF) files that can be opened and read in most standard word processing applications.

Currently the program accepts only numeric Sequence Number Central codes. The program will terminate with an error message if it encounters nonnumeric codes (e.g., AA). It is important to run this program on data that have already passed single-field edits. Inter-record edit results may be misleading for data that fail single-record edits.

The inter-record edits are based on COBOL programs written by staff from the National Cancer

Institute Surveillance, Epidemiology, and End Results (SEER) program and have been recoded into SQL.

To run the Inter-Record Edits Utility:

Launch the program using your computer’s desktop icon or the MS Windows Run command.

The NPCR Inter-Record Edits Utility Welcome screen opens. This screen contains instructions for running the program (see exhibit 9).

Exhibit 9. NPCR Inter-Record Edits Utility Welcome screen

Click on Continue to open the NPCR Inter-Record Edits Utility main screen (see exhibit 10).

Exhibit 10. NPCR Inter-Record Edits Utility main screen

Before running the program, you must select the source file on which the program will run.

Click on the Browse button next to the NAACCR V11 data file name field to select a source file. The Run button at the bottom left of the window will be inaccessible until a source file is selected.

In the Summary report file name field, enter the name under which you want to save the

Summary Report. By default, the file will be saved in the same directory as that containing the source file. If you want to save the Summary Report to a different directory, click on the

Browse button and select the path and file name to which you want to save the Summary

Report (see exhibit 11).

Exhibit 11. NPCR Inter-Record Edits Utility Browse screen

Repeat the previous step using the detailed report file name field for the Detailed Report.

Click on Run to begin the Inter-Record Edits Utility.

A popup box will appear, tracking the progress of the program as it processes each record.

When the processing is done, a message box appears, stating that testing is complete and noting how many records have been tested (see exhibit 12).

Exhibit 12. Inter-Record Edits Utility message box

The inter-record edits are complete. Click on OK to close the message box. To view the results, click on the View Summary Report or View Detail Report buttons.

2.8.3. Data Edits and Corresponding Edit Codes

Table 3. NPCR-CSS data edits and corresponding Commission on Cancer, NAACCR, and

SEER edit codes

Edit Set Edit Type Field Name Edit Name Edit Source

Advanced Inter-field SEERIF43 Age, Histologic Type, Cause of Death SEER IF43

Advanced Inter-field SRIF43_3 Age, Histologic Type, COD, ICDO3 SEER IF43

Advanced Inter-field SEERAPM Age, Primary Site, Morph ICDO3—Adult SEER

Advanced Inter-field NPCRAPM Age, Primary Site, Morph ICDO3—Pediatric NPCR

Advanced Inter-field SEERIF45 Census Tract 1970/80/90, Census Cod Sys SEER IF45

Advanced Inter-field SEERIF71 CompEthn, Date of Diag SEER IF71

Advanced Inter-field DXCNTYST County at DX, Addr at DX—State NAACCR

Advanced Inter-field NACRIF19 Date of Last Contact, Date of Diag NAACCR IF19

Advanced Inter-field SEERIF72 EthnSrc, Date of Diag SEER IF72

Advanced Inter-field FUP_VITL Follow-Up Source, Vital Status COC

Advanced Inter-field SEERIF37 ICD Revision Number, Cause of Death SEER IF37

Advanced Inter-field NPCRRVD ICD Revision, Vital Stat, Date Last Contact NPCR

Advanced Inter-field SEERI109 RX Summ—Scope Reg LN Sur, Primary Site SEER IF109

Advanced Inter-field SEERI108 RX Summ—Surg Prim Site, Primary Site SEER IF108

Advanced Inter-field SEERIF09 Type of Report Srce(DC/AO), COD SEER IF09

Advanced Inter-field SEERIF02 Type of Report Srce(DC/AO), Date of Dx SEER IF02

Advanced Inter-field RSRCVSTA Type of Report Srce(DC/AO), Vit Stat COC

Advanced Inter-record SEERIR11

Ver…