• HOME
• TOPICS
  • Certificates & Records
  • Data & Statistics
  • Diseases & Conditions
  • Emergency Preparedness
  • Environments & Your Health
  • Facilities & Professions
  • Health Care & Coverage
  • Injury, Violence & Safety
  • Life Stages & Populations
  • Policy, Economics & Legislation
  • Prevention & Healthy Living
• ABOUT US

Minnesota Cancer Surveillance System

• MCSS Home
• About Us
• Minnesota Cancer Data
• MCSS Notes (bi-annual newsletter)
• HIPAA and Cancer Reporting
• Current Minnesota Statutes and Rules
• MCSS Rule Change
• Materials from Workshop on Brain Tumors and Data Quality
• Publications Containing Mesothelioma Rates
• Webinars

Overview of Cancer Data

Background and Data Sources

This report contains information on the incidence and mortality of cancer in Minnesota for the nine-year period 1988-1996. Cancer incidence and mortality provide two important measures of the impact of cancer in Minnesota. Incidence measures how many newly diagnosed cases of the disease occur, while mortality measures how many people die of the disease. Both measures are often expressed as the annual rate of occurrence per 100,000 people. The Minnesota Department of Health (MDH) collects and analyzes data on both incidence and mortality of cancer. Incidence data in this report are compiled by the Minnesota Cancer Surveillance System (MCSS), and mortality data are compiled by the Minnesota Center for Health Statistics (MCHS). According to common practice, all incidence and mortality rates presented in this report are age-adjusted to the 1970 United States standard population. Age-adjustment minimizes the effect of differences in age distributions when comparing rates among different populations.

The MCSS is an ongoing program within the Section of Chronic Disease and Environmental Epidemiology at the MDH. The MCSS systematically collects demographic and diagnostic information on all Minnesota residents with newly-diagnosed cancers.

The primary objectives of the MCSS are to: (1) monitor the occurrence of cancer in Minnesota and describe the risks of developing cancer, (2) inform health professionals and educate citizens regarding specific cancer risks, (3) answer the public's questions and concerns about cancer, (4) promote cancer research, and (5) guide decisions about how to target cancer control resources.

The need for accurate information about the occurrence of cancer was recognized by the Minnesota legislature in 1981, when legislation was introduced to establish a statewide cancer surveillance system. In 1987, following a six-year process which included consensus building, development of methods, and a feasibility study, legislation (Minnesota Statutes 144.671-144.69) was passed to establish the MCSS, and on January 1, 1988, the MCSS began operations.

This is the first MCSS report containing information on cancer stage at diagnosis. This information was not collected on cancers diagnosed before 1995 because insufficient funding was available. Funds for the enhancement of the MCSS became available in 1994 through the National Program of Cancer Registries (NPCR), which is administered by the U.S. Centers for Disease Control and Prevention (CDC). NPCR funding began in October 1994 and is scheduled to last at least through June 2000. The support of the NPCR has enabled the MCSS to collect additional information on each case of cancer, perform death clearance, perform quality control studies, and provide specialized training to Minnesota professionals who collect and code cancer data.

An attempt has been made to minimize the use of technical jargon in this report. However, because of the nature of the material and the diverse audience that this report must serve, some technical terms remain.

In order to minimize reproduction of discussion and materials presented in previous reports, liberal cross-referencing is employed. The four previous reports are:
The Occurrence of Cancer in Minnesota - 1988;
The Occurrence of Cancer in Minnesota 1988-1990: Incidence, Mortality, and Trends;
The Occurrence of Cancer in Minnesota 1988-1992: Incidence, Mortality, and Trends; and The Occurrence of Cancer in Minnesota 1988-1994: Incidence, Mortality, and Trends.
These reports will be referenced as MCSS 1991, MCSS 1993, MCSS 1995, and MCSS 1997, respectively; they are available from the MCSS at the MDH.

Data Sources and Operations

Incidence Data

The MCSS collects information on malignant tumors, as well as benign tumors occurring in the head and spinal cord, that are diagnosed in Minnesota residents. It does not collect information on the most common form of skin cancer (basal and squamous cell carcinomas). The MCSS has three important characteristics: it is population-based, it is pathology-based, and it is an active (vs. passive) system. Population-based surveillance means that data are collected from a defined population base (i.e., the state of Minnesota) so that incidence rates (risk) can be calculated. Pathology-based surveillance means that cancers are identified through the pathology laboratories where tissues are examined and the diagnosis of cancer is made. Active surveillance means that cancers are required to be reported to the MCSS by Minnesota statutes and that pathology reports in the state and surrounding border referral centers are reviewed by MCSS Field Service staff to ensure completeness of ascertainment.

Enough information is collected so that the MCSS can classify each new diagnosis by type of tumor (primary site, histologic cell type), tumor stage (how advanced the cancer is), and demographic characteristics of the patient (age, sex, and address) as of the date of diagnosis of the cancer, as well as a summary of the first course of cancer-directed treatment. Information about the patient, cancer, stage, and/or treatment that the pathology laboratory cannot provide is obtained from hospital-based cancer registries or the patient's hospital or clinic record.

Hospitals and pathology laboratories provide data to the MCSS in two main ways. Hospitals that have computerized cancer registries containing summaries for each cancer patient treated at the hospital submit these computerized case reports. The remaining cancer diagnoses are reported through pathology laboratories, which submit photocopies of the pathology report which contains information about the cancer, and the medical record face sheet or an equivalent form which contains the patient's demographic data. The MCSS has begun its twelfth year of operation, and a total of more than 400,000 reports of cancer representing approximately 240,000 different cancers have been registered with the MCSS as of April 1999. For the period covered by this report (January 1, 1988 to December 31, 1996), 177,007 newly diagnosed, invasive cancers were registered. Non-invasive (in situ) cancers of the urinary bladder are included with the invasive cancers of that site in order for Minnesota data to be consistent with the National Cancer Institute's Surveillance, Epidemiology, and End Results (SEER) program, the source of national cancer statistics.

The data upon which this report is based are dynamic. That is, they are always being updated and improved. For example, in the first legislative report on the MCSS (MCSS 1991), filed eight years ago, 17,728 cancers were included in the analyses of 1988 data. The current database for 1988 contains information on 18,853 cancers (some of the increase is because the data reported for 1988 did not include in situ cancers of the bladder). MCSS staff are constantly updating data for all years when new information becomes available. In this regard, all data are subject to change when appropriate. For purposes of analyses, the data are "frozen" (closed) in order that numbers and rates be consistent throughout the report. The date of closure for 1988-1996 data included in this report is February 8, 1999.

Mortality Data

Mortality data are obtained from death certificates. Death certificates are collected, coded, and computerized by the MCHS. Although the MCHS codes contributing causes of death as well as the underlying cause of death, only the underlying cause of death was used in calculating cancer mortality rates.

Population Data

Minnesota population estimates for the years 1988-1990 were developed from the United States Census Bureau's 1980 and 1990 censuses. Linear interpolation between the two census years provided age-sex-region estimates for 1988 and 1989. Age-, sex-, and county-specific intercensal population estimates obtained from the U.S. Census Bureau were used for the six years 1991-1996. Average annual rates for the most recent five-year period are the focus of this report. They are calculated by dividing the total numbers of cancers that occurred in the years 1992 to 1996 by the sum of the population estimates for the five years.

Completeness and Quality of Data

MCSS Field Service staff first identified 10.3 percent of all the cancer diagnoses reportable to the MCSS during their independent review of pathology reports. This review is an important feature of MCSS quality control in that it assures that virtually all eligible cancers are included in the data. For all of the individual cancers diagnosed during 1988-1996, 6.1 percent of these would have been missed without this review. It is estimated that more than 4.5 million pathology reports were reviewed during the nine-year period included in this report.

Death clearance is an important quality control activity which the MCSS began in 1995. Cancer-related deaths are linked with the MCSS database to identify cancer cases that were missed by the routine reporting processes. Potentially missed cancers are then followed back to determine if the cancer indeed should have been included in the MCSS database. Unresolved cancers are included in the database as "Death Certificate Only" (DCO) cases. Death clearance can identify sources where cancer reporting might be improved. Results to date indicate that MCSS case ascertainment is excellent. Only 1.2 percent of all cancers diagnosed in 1995 and 1996 (the two years for which death clearance has been performed) had the death certificate as the source of casefinding. A high-quality cancer registry should have between 1 percent and 3 percent of its cases as DCO.

MCSS data are very complete and of very high quality. This is documented by several measures of data quality which are available for the MCSS. First, in December 1998 the MCSS submitted a non-identified file of its provisional data for 1996 to the Registry Certification Committee of the North American Association of Central Cancer Registries (NAACCR). NAACCR is the organization in North America which develops standards and models for the collection of cancer data in central cancer registries. Table 1.1 contains the results of the certification process. The MCSS achieved the highest rating, the Gold Standard, for all criteria except the duplicate primary cases category. The data upon which the present report is based would meet the Gold Standard for duplicate cases as well. Second, NAACCR performed an external audit of the completeness and quality of MCSS data in 1996 (MCSS 1997). Case completeness was estimated at 99.7 percent. Data quality estimates varied by the data field and the type of cancer. The error rates for all audited data items were comparable to the median error rates found in other states audited by NAACCR, except for race -- a field for which reporting has improved since the audit took place. Third, the MCSS has also completed two studies of the accuracy of the data contained in the central registry. These indicate that MCSS data are of comparable quality to data of other central cancer registries in the United States (MCSS Quality Control Reports 97:2 and 99:1). Special attention has been paid to the data fields that are new to the MCSS - stage at diagnosis and the information on the first course of cancer therapy.

Another measure of data quality is the amount of missing information in all data collected to date. MCSS Field Service staff perform follow-up to obtain missing information and to resolve conflicting information on all cancers reported. During the process of preparing 1996 diagnoses for this report, contacts with hospitals, clinics, or physicians' offices were required for nearly 6,000 cases to ensure that the data were complete and accurate. The primary elements for surveillance are age at diagnosis, sex, date of diagnosis, county of residence, primary site (organ in which cancer first developed, e.g., lung), and histology (microscopic description of the cancerous cells). For the period 1988-1996, only 0.27 percent of the records contained missing information on one or more of those variables.

Race is another important variable for cancer surveillance. Unfortunately, this information is not always available in the medical records submitted to the MCSS, and prior to the 1995 diagnosis year the MCSS did not have the resources to perform the active follow-up to find the missing information. This is reflected in the fact that no indication of the patient's race was reported for one-sixth (16.1 percent) of the cancers incident during the period 1988-1996. The percentage can be improved by assuming that individuals are white if they live in counties that were more than 95 percent white in the 1990 census, and no other racial information is available. After making this assumption, race was "unknown" for only 3.7 percent of the cancers incident during the period 1988-1996. The effect of active follow-up is demonstrated by the fact that the percent with unknown race was much lower for cancers diagnosed in 1995 and 1996 (3.8 percent before and 1.0 percent after making an assumption based on county of residence).

Despite the recent improvement in the completeness of racial data, the quality of racial data available to the MCSS is still inadequate for meaningful surveillance. Active follow-up to find missing racial information has improved the completeness of the data, but the number of non-whites is quite small in Minnesota, so that several years of observations will be required to compute stable estimates of cancer rates among non-white Minnesotans. Furthermore, because there are no counties where a 95 percent accurate assumption could be made about race for people of color, a disproportionate number of people of color are coded as "unknown race."  Even when the race is reported for people of color, their race or ethnicity was usually recorded by a clerk based on the clerk's perception of the person's race. This process can create two errors.

First, errors in racial or ethnic designation can more easily occur if patients are not queried about their race and, thus, allowed to identify their own racial/ethnic designation. Second, the racial data collected without querying the patient are not comparable to racial data contained in the U.S. census. The census determination of race is through self-identification. The census data serve to define the Minnesota population by age, sex, and race. Cancer incidence is calculated using the census data, and if incidence data for people of color are to be meaningful, racial information on cancer patients must be collected in a similar manner (i.e., through self-identification). Because many of the communities of color represent very small populations, inconsistencies between the two types of racial data can lead to very large errors in the apparent cancer incidence in these populations. For these and other reasons discussed in Section 2 of the MCSS 1993 report, cancer incidence is reported only for all races combined. Another issue related to race is the fact that a person is assumed to be of only one race, even though an individual can be of mixed racial descent. The federal Office of Management and Budget has issued a directive that affects how racial data will be collected in the United States. In the Year 2000 Census, individuals will be allowed to report multiple races. The cancer registry community is preparing to collect data in a comparable manner, so that it will still be possible to calculate race-specific rates.

Although extensive efforts are directed at completeness and accuracy of MCSS data, the MCSS's total costs of operation are substantially less than those of other population-based cancer surveillance systems in the country. A major contributor to this efficiency is the sophisticated computerization of many tasks that are normally done manually elsewhere (MCSS 1991; Section 2).

Uses of MCSS Data

The MCSS has five primary objectives. The following is a brief summary of how the MCSS is accomplishing each objective.

Monitoring the occurrence of cancer in Minnesota and describing the risks of developing cancer. Sophisticated computer programs have been written to support the MCSS epidemiologists in describing the risks of developing cancer. These data have been carefully reviewed for 35 major cancer sites to identify important age, sex, or regional factors. Cancer mortality data have also been analyzed and included in this MCSS description of cancer occurrence in Minnesota.

MCSS staff conduct special analyses of the surveillance data. Emphasis is placed on integrating these findings into an epidemiologic and public health context. The thoroughness and rigor with which these reports are produced have been recognized in the academic community which uses them for courses on cancer epidemiology.

Informing health professionals and educating citizens regarding specific cancers. Since 1990, dozens of formal presentations before legislative and other policy making bodies have been made on the occurrence of cancer in Minnesota.

Answering the public's questions and concerns about cancer. The MCSS receives approximately 150 requests per year for information on cancer rates or cancer risks. These inquiries represent all geographic regions of the state. Although most of these inquires are from individual citizens, inquiries also frequently come from citizens' groups, schools, and workplaces, as well as the public health, scientific, and medical communities. Responses to these public and scientific inquiries range from provision of simple, descriptive statistics to detailed record linkage studies of a defined cohort. Routine requests about cancer concerns are now referred to the Cancer Response System (CRS) of the American Cancer Society (Minnesota Division, Inc.); members of the CRS have been trained by MCSS staff. Requests requiring further consideration are investigated by MCSS staff using a step-wise protocol until concerns are resolved. The decision to refer most inquiries to the CRS was due to the perception that the public is more likely to accept cancer and health messages from nongovernmental agencies. In addition, The Commissioner of Health convened the Minnesota Department of Health's Cancer Cluster Work Group, because the MDH had concluded that a technical response alone was not adequate to satisfy community concerns about cancer. The Work Group, which included prominent scientific, government, environmental activist, political, and media figures from throughout Minnesota, suggested increased use of health education experts, increased collaborative efforts, and networks with traditional and new partners of public health, particularly at the local level, and greater emphasis and coordination by the MDH and its partners in public education.

Promoting cancer research. The MCSS has assisted cancer researchers by providing information and data needed for the planning and support of grant applications. The MCSS has also received 20 data use applications since 1988, which are described in Table 1.3. The involvement of the MCSS in the approved studies has varied from providing information about the completeness of case finding to providing rapid identification of cases for case-control studies. In addition, MCSS data have been used to investigate concerns about cancer occurrence in the workplace.