Brain Tumor Registry Reporting Training Materials
Other Materials

Surveillance of Primary Intracranial and Central Nervous System Tumors: Recommendations from the Brain Tumor Working Group
Discussion

Surveillance of Primary  Intracranial and Central  Nervous System Tumors Preface Executive Summary Introduction Background Methods Results Discussion Recommendations References Tables Appendix A Appendix B Appendix C Appendix D

The findings for this report indicate that collection and reporting of incidence data for primary malignant intracranial and other CNS tumors are well standardized. However, substantial variation exists in the processes of collecting and reporting nonmalignant tumor data. One obstacle is that not all hospitals have cancer registries. Another problem is that not all hospitals with registries accession cases of nonmalignant tumors. Data from Minnesota (Table 13) suggest that approximately 85 percent of cases of nonmalignant intracranial tumors are accessioned by hospital registries. However, nonaccessioned CNS cases are less likely to be routinely reported to the central registry (Table 13). These differences in collection and reporting practices among the registries make it difficult to assess the burden of primary intracranial and CNS tumors.

Terminology for describing intracranial and CNS tumors is also not standardized. For example, although the pituitary gland and pineal glands are not technically part of the CNS, tumors that involve these organs are often included in the term "CNS tumors." On the other hand, some consider the eye to be intracranial; however, "CNS tumors" generally do not include the eyes. Although the World Health Organization (WHO) has published a list of histologic groupings, at present no single comprehensive list of histologic groupings exists for intracranial and CNS tumors. Since the International Classification of Childhood Cancer (ICCC) is based on histologic type rather than site, certain benign histologies of sites other than brain/CNS (e.g., non-CNS ependymomas) would not be included in the BTWG's primary intracranial and CNS histology groupings (Appendix D). Thus, the variation in incidence estimates of primary intracranial and CNS tumors may be attributable, in part, to the variation in the definitions of these tumors. In addition, differences in cancer registry training and procedures may contribute to nonreporting and inconsistent reporting of the tumors.

The BTWG discussed the feasibility of conducting a random sampling procedure as an alternative to complete enumeration and ongoing surveillance of nonmalignant brain tumors. It was agreed that this technique would not be valid or feasible for the following reasons:

1. If a random sample of incident brain tumors were desired, a complete list would be needed first (a sampling frame is required).
2. If a survey method were used to identify brain tumors in the general population (e.g., random digit dialing or other general population survey techniques), the information needed could not be obtained for several reasons:
   • Either a prohibitively large numbers of individuals would have to be surveyed, or else only a very small number of brain tumors could be identified, given the rarity and sometimes short survival time.
   • The brain tumors identified would not be representative of all brain tumors that occur, because people who had survived their brain tumors would be more likely to be identified.
   • The incidence rate would still not be known, since cross-sectional surveys only identify disease prevalence.
3. If a sample of hospitals were selected to contribute data, the tumors included would not represent the entire spectrum of brain tumors because those that were diagnosed and treated outside of the hospital setting would be excluded.

An alternative approach to sampling, and the least troublesome, would be to conduct population-based surveillance only in certain geographic areas of the country. If one assumes that the etiology of brain tumors does not differ from one area of the country to another, then the scientific validity of studies based on this type of sampling would not be a problem. However, given the rarity of some types of brain tumors, sufficient numbers for study may not be available unless the vast majority of the nation's population were covered by population-based data collection. Also, geographic areas without population-based surveillance would have no data to use for assessing local trends or variations in brain tumor occurrence.

In the absence of standard tumor registration procedures and training, special case finding efforts by central registries may be necessary to ensure that complete data are collected for nonmalignant intracranial and CNS tumors. Data from MCSS (Table 10), which has a legal requirement for reporting nonmalignant intracranial tumors and performs 100 percent case finding audits in all laboratories, indicate that only 80 percent of nonmalignant tumors were routinely reported. In contrast, routine reporting accounted for more than 95 percent of malignant intracranial and CNS tumors. Nonmalignant cases in Minnesota that were not accessioned by hospital registries had a 50 percent chance of being routinely reported to the central registry; however, these were only microscopically confirmed cases.

Completeness of reporting is critical to cancer registries. Accurate case counts are necessary to assess the burden of cancer, to guide cancer control program planning, to prioritize the allocation of health resources, and to facilitate epidemiologic research. Most central cancer registries have state laws that mandate reporting of cases by physicians, and by hospitals, laboratories, and other facilities that provide screening, diagnostic, or therapeutic services. Complete reporting of nonmalignant intracranial and CNS tumors would be greatly improved if reporting requirements of the COC, central cancer registries, SEER, and the National Program of Cancer Registries (NPCR) were changed to require the collection of information for these tumors. However, this requirement would have many implications, including an increase in work load as well as associated costs for reporting facilities. An estimated 1.4 percent of all new cancer cases diagnosed in 1998 will involve invasive brain and other nervous system tumors;⁷ since the numbers of benign and invasive brain and other CNS tumors diagnosed annually are similar, facilities that presently do not accession nonmalignant cases could expect an approximately 1.4 percent increase in the total number of CNS cases collected by the registry. Some estimates of the percentage of brain tumors have been lower (0.5 percent)³⁶ and others have been higher (9 percent);³⁷ therefore, the extent to which the workload would be affected may vary. The data in this report suggest that if reporting of nonmalignant intracranial and CNS tumors were required, the total number of cases of tumors for these anatomic sites would double for facilities that presently report only malignant tumors. This is consistent with the findings of Davis et al,³⁶ who reported that the incorporation of benign brain tumors into the cancer-reporting systems of four central registries increased the overall incidence of brain cancer by 49 percent. Forty percent (913) of the hospital registries that submitted data to the NCDB did not report nonmalignant brain tumors. These registries could expect up to a 50 percent increase in the number of intracranial and CNS tumors reported; nonregistry facilities, which represent approximately 20 percent of cases reported to central cancer registries, could expect a similar increase. Also, central cancer registries that currently do not require reporting of nonmalignant intracranial and CNS tumors could expect increases in their workloads as a result of the additional time spent on processing and quality control procedures and training. The lack of standard definitions and collection and reporting guidelines would make these tasks more time-consuming as well.

For some central cancer registries that want to expand their reporting requirements to include nonmalignant intracranial and CNS tumors, a change in current legislation and/or regulations may be needed. These changes could involve several months of lead time if public hearings or other legal procedures are necessary. In other states, the reporting of these tumors is not required in either their legislation or regulations; in these situations hospitals have been asked to voluntarily report the cases. For example, in Massachusetts the reporting of benign brain tumors is not required by law, but hospitals have been reporting these cases since 1982. However, the completeness of data reported on a voluntary basis is difficult to assess. For public health surveillance systems, a mandate to report is the basic requirement of a comprehensive, higher quality system.

If COC, SEER, NPCR, and central cancer registries changed their reporting requirements to include nonmalignant intracranial and CNS tumors, the registry manuals disseminated by these programs and registries would also need to be modified to include new definitions of reportable diagnoses. All reporting facilities would have to be notified of the new requirements. To ensure complete reporting from registries, additional training of registrars should be expected. Case finding methods may need to be modified in order to identify the new cases and sources of information. Central cancer registries would also need to increase their case finding audits. The inclusion of nonmalignant tumors would also necessitate changes to other data items such as sequence number. Hospital and central cancer registries would need to consider adding WHO brain tumor grade as an additional data item.

Finally, the software used by reporting facilities and by central registries would need to be modified to accept nonmalignant morphology and behavior codes. Also, edits programs would need to be modified to include these additions and to accept a nonmalignant sequence numbering procedure.

Before reporting requirements for primary intracranial and CNS tumors could be changed, feasibility studies would have to be conducted to determine whether such changes should be recommended. Two SEER special studies are under way to evaluate the impact of requiring the collection of benign brain tumors on case finding, cost, quality control, and training. However, additional studies may be required.