Are Benign and Borderline Brain Tumors Underreported?

BACKGROUND: Primary benign and borderline (BB) brain tumors have been reportable since 2004 by population-based cancer registries in the United States. Because these tumors often are diagnosed clinically at nonhospital settings, underreporting is a big concern. Despite this, the magnitude and geographic variations in underreporting are unknown. The objectives of this study are to assess variations in BB brain tumor incidence rate by registry and trend in comparison to malignant brain tumors, as well as to identify the factors associated with the completeness of BB brain tumor reporting. METHODS: North American Association of Central Cancer Registries (NAACCR) Cancer in North America (CINA) Deluxe 1995­2012 Analytic File, which included data from 47 US population-based cancer registries, was used. Age-adjusted incidence rate and average annual percent change (APC) were calculated. Correlation coefficients were used to assess the relationships between incidence rates and clinical factors. RESULTS: The overall age-adjusted incidence rate was 14.2 per 100,000 for BB brain tumors and 6.6 per 100,000 for malignant brain tumors. The age-adjusted incidence rates of BB brain tumors varied by registry from 9.8 per 100,000 to 19.9 per 100,000, whereas the variations in malignant brain tumors were much smaller from 4.1 per 100,000 to 7.7 per 100,000. BB brain tumor cases were more likely than malignant brain tumors to be diagnosed through radiography without microscopic confirmation or surgery. Overall, the BB brain tumor incidence rate significantly increased by 2.3% per year from 2004 to 2012. In contrast, incidence rates of malignant brain tumors significantly decreased by 0.9% per year in the same period. Higher BB brain tumor incidence rates were significantly associated with higher proportions of cases without microscopic confirmation or surgery. These associations were not observed for malignant brain tumors. CONCLUSIONS: Incidence rates of BB brain tumors varied substantially across 47 US registries and were higher than those of malignant brain tumors in the United States. The variations in incidence rate of BB brain tumors may be largely attributable to difference in identifying clinically diagnosed cases. The increasing incidence rate of BB brain tumors may reflect improved case ascertainment rather than a biological trend. Key words:

Linkage of 1986-2009 National Health Interview Survey with 1981-2010 Florida Cancer Data System.

BACKGROUND: National survey data linked with state cancer registry data has the potential to create a valuable tool for cancer prevention and control research. A pilot project-developed in a collaboration of the Centers for Disease Control and Prevention's National Center for Health Statistics (NCHS) and the Florida Cancer Data System (FCDS) at the University of Miami -links the records of the 1986-2009 National Health Interview Survey (NHIS) and the 1981-2010 FCDS. The project assesses the feasibility of performing a record linkage between NCHS survey data and a state-based cancer registry, as well as the value of the data produced. The linked NHIS-FCDS data allow researchers to follow NHIS survey participants longitudinally to examine factors associated with future cancer diagnosis, and to assess the characteristics and quality of life among cancer survivors. METHODS: This report provides a preliminary evaluation of the linked national and state cancer data and examines both analytic issues and complications presented by the linkage. CONCLUSIONS: Residential mobility and the number of years of data linked in this project create some analytic challenges and limitations for the types of analyses that can be conducted. However, the linked data set offers the ability to conduct analyses not possible with either data set alone.

Using the National Death Index to identify duplicate cancer incident cases in Florida and New York, 1996-2005.

INTRODUCTION: Cancer registries link incidence data to state death certificates to update vital status and identify missing cases; they also link these data to the National Death Index (NDI) to update vital status among patients who leave the state after their diagnosis. This study explored the use of information from NDI linkages to identify potential duplicate cancer cases registered in both Florida and New York. METHODS: The Florida Cancer Data System (FCDS) and the New York State Cancer Registry (NYSCR) linked incidence data with state and NDI death records from 1996 through 2005. Information for patients whose death occurred in the reciprocal state (the death state) was exchanged. Potential duplicate cases were those that had the same diagnosis and the same or similar diagnosis date. RESULTS: NDI identified 4,657 FCDS cancer patients who died in New York and 2,740 NYSCR cancer patients who died in Florida. Matching identified 5,030 cases registered in both states; 508 were death certificate-only (DCO) cases in the death state's registry, and 3,760 (74.8%) were potential duplicates. Among FCDS and NYSCR patients who died and were registered in the registry of the reciprocal state, more than 50% were registered with the same cancer diagnosis, and approximately 80% had similar diagnosis dates (within 1 year). CONCLUSION: NDI identified DCO cases in the death state's cancer registry and a large proportion of potential duplicate cases. Standards are needed for assigning primary residence when multiple registries report the same case. The registry initiating the NDI linkage should consider sharing relevant information with death state registries so that these registries can remove erroneous DCO cases from their databases.

Muddy water? Variation in reporting receipt of breast cancer radiation therapy by population-based tumor registries.

PURPOSE: To evaluate, in the setting of breast cancer, the accuracy of registry radiation therapy (RT) coding compared with the gold standard of Medicare claims. METHODS AND MATERIALS: Using Surveillance, Epidemiology, and End Results (SEER)-Medicare data, we identified 73,077 patients aged ≥66 years diagnosed with breast cancer in the period 2001-2007. Underascertainment (1 - sensitivity), sensitivity, specificity, κ, and χ(2) were calculated for RT receipt determined by registry data versus claims. Multivariate logistic regression characterized patient, treatment, and geographic factors associated with underascertainment of RT. Findings in the SEER-Medicare registries were compared with three non-SEER registries (Florida, New York, and Texas). RESULTS: In the SEER-Medicare registries, 41.6% (n=30,386) of patients received RT according to registry coding, versus 49.3% (n=36,047) according to Medicare claims (P<.001). Underascertainment of RT was more likely if patients resided in a newer SEER registry (odds ratio [OR] 1.70, 95% confidence interval [CI] 1.60-1.80; P<.001), rural county (OR 1.34, 95% CI 1.21-1.48; P<.001), or if RT was delayed (OR 1.006/day, 95% CI 1.006-1.007; P<.001). Underascertainment of RT receipt in SEER registries was 18.7% (95% CI 18.6-18.8%), compared with 44.3% (95% CI 44.0-44.5%) in non-SEER registries. CONCLUSIONS: Population-based tumor registries are highly variable in ascertainment of RT receipt and should be augmented with other data sources when evaluating quality of breast cancer care. Future work should identify opportunities for the radiation oncology community to partner with registries to improve accuracy of treatment data.

Evaluating factors associated with unknown SEER Summary Stage 2000 derived from collaborative stage at central registry level.

BACKGROUND: Cancer stage is critical for treatment planning and assessing disease prognosis. The percentage of unknown staged cancer cases varies considerably across state cancer registries; factors contributing to the variations in unknown stage have not been reported in the literature before. The purpose of this study was to examine whether these variations were influenced by demographic and/or clinical factors as well as the type of reporting facility. METHODS: Invasive colorectal, lung, female breast, and prostate cancers diagnosed between 2004 and 2007 were obtained from the North American Association of Central Cancer Registries (NAACCR); 47 population-based cancer registries in the United States were included. The unknown stage was based on Summary Stage 2000 codes derived from Collaborative Stage Version 1 (CSv1). Relative importance analysis was used to identify variables that were essential in predicting unknown stage. Using state central registries as analytical units, multiple linear regression was used to evaluate factors associated with the percentage of unknown stage by cancer site; potential outlier registries with a high percentage of unknown stage cases were identified using boxplots and standardized residuals. RESULTS: Overall, lung cancer had the highest percentage of unknown stage (8.3%) and prostate cancer had the largest variation of unknown stage among registries (0.6%-18.1%). The percentages of neoplasms not otherwise specified (NOS) histology, non-microscopic confirmation, and non-hospital reporting source were positively associated (p less than 0.05) with percentage of unknown stage for all studied cancer sites before adjustment. Variables that retained a positive association with unknown stage including all demographic and clinical variables, year of diagnosis, and type of reporting source were black race, metropolitan area less than 1 million population, histologies of neoplasms NOS or epithelial neoplasms NOS, diagnosis year 2005, and non-hospital reporting source for colorectal cancer; metropolitan area less than 1 million population, neoplasms NOS histology, and non-hospital reporting source for female breast; and diagnosis year 2005 and non-hospital reporting source for prostate. After adjustment, none of the predictors were significant for lung cancer. We observed 1 potential outlier registry each for colorectal, lung and female breast cancers. CONCLUSIONS: Factors associated with unknown stage differ by cancer site; however, the type of reporting source is an important predictor of unknown stage for all cancers except lung after adjustment. Central registries with high percentage of unknown stage should be made aware of their data quality issue(s). As a result, these registries can investigate those factors and provide training to registrars to improve their cancer data quality.

Toward the identification of communities with increased tobacco-associated cancer burden: Application of spatial modeling techniques.

INTRODUCTION: Smoking-attributable risks for lung, esophageal, and head and neck (H/N) cancers range from 54% to 90%. Identifying areas with higher than average cancer risk and smoking rates, then targeting those areas for intervention, is one approach to more rapidly lower the overall tobacco disease burden in a given state. Our research team used spatial modeling techniques to identify areas in Florida with higher than expected tobacco-associated cancer incidence clusters. MATERIALS AND METHODS: Geocoded tobacco-associated incident cancer data from 1998 to 2002 from the Florida Cancer Data System were used. Tobacco-associated cancers included lung, esophageal, and H/N cancers. SaTScan was used to identify geographic areas that had statistically significant (P<0.10) excess age-adjusted rates of tobacco-associated cancers. The Poisson-based spatial scan statistic was used. Phi correlation coefficients were computed to examine associations among block groups with/without overlapping cancer clusters. The logistic regression was used to assess associations between county-level smoking prevalence rates and being diagnosed within versus outside a cancer cluster. Community-level smoking rates were obtained from the 2002 Florida Behavioral Risk Factor Surveillance System (BRFSS). Analyses were repeated using 2007 BRFSS to examine the consistency of associations. RESULTS: Lung cancer clusters were geographically larger for both squamous cell and adenocarcinoma cases in Florida from 1998 to 2002, than esophageal or H/N clusters. There were very few squamous cell and adenocarcinoma esophageal cancer clusters. H/N cancer mapping showed some squamous cell and a very small amount of adenocarcinoma cancer clusters. Phi correlations were generally weak to moderate in strength. The odds of having an invasive lung cancer cluster increased by 12% per increase in the county-level smoking rate. Results were inconsistent for esophageal and H/N cancers, with some inverse associations. 2007 BRFSS data also showed a similar results pattern. CONCLUSIONS: Spatial analysis identified many nonoverlapping areas of high risk across both cancer and histological subtypes. Attempts to correlate county-level smoking rates with cancer cluster membership yielded consistent results only for lung cancer. However, spatial analyses may be most useful when examining incident clusters where several tobacco-associated cancer clusters overlap. Focusing on overlapping cancer clusters may help investigators identify priority areas for further screening, detailed assessments of tobacco use, and/or prevention and cessation interventions to decrease risk.

Comparisons of directly coded SEER Summary Stage 2000 and Collaborative Staging Derived SEER Summary Stage 2000.

This study assessed comparability of the directly coded Summary Stage 2000 and the Collaborative Stage (CS) Derived Summary Stage 2000 (SS2000) using 2001-2004 data from 40 population-based cancer registries in the United States that met the high quality criteria. The likelihood ratio test was employed to determine whether stage differences between 2003 (pre-CS) and 2004 (CS) were attributable to 2001-2004 linear trends, decreases in percentage of unknown stage cases, or both. Statistically significant differences in stage distribution between 2003 and 2004 were observed for 30 out of the 34 cancer sites. For 4 cancer sites, the differences were attributable to 2001-2004 linear trends of stage distribution. For 8 cancer sites, the differences were attributable to decreases in percentage of unknown stage cases alone or in combination with the temporal trends of stage distribution. For the remaining 18 cancer sites, either (1) no linear trends of stage distribution were identified or (2) the combination of the decline in cases with unknown stage plus linear trends did not explain the stage differences between 2003 and 2004. By comparing the SS2000 and CS manuals, we found differences in coding definitions for direct extension and/or lymph node involvement for all cancer sites except cancers of the breast, cervix, and cranial nervous and other nervous system. Evidence showed that the stage differences between 2003 and 2004 may be attributable in part to the implementation of the CS System for some cancer sites.

Gender- and race-specific comparison of tobacco-associated cancer incidence trends in Florida with SEER regional cancer incidence data.

OBJECTIVE: Analysis of state and national tobacco-associated cancer trends is critical for the identification of high-risk regions of the country that require the attention of the public health community. This study compares Florida race- and gender-specific cancer trends with pooled data obtained from nine Surveillance, Epidemiology, and End Results (SEER-9) registries. METHODS: Age-adjusted, race- and gender-specific cancer incidence trends were evaluated using joinpoint regression analysis. Pooled, age-adjusted incidence rates and standardized incidence rate ratios were computed for each cancer for the years 1999-2003 to compare Florida to SEER-9. RESULTS: Relative to SEER-9 whites and irrespective of gender, lung cancer rates in white Floridians were elevated through the 1990s. However, lung cancer rates have recently declined at a steeper rate among white Floridians than among SEER-9 whites. For years 1999-2003, black Floridians had significantly lower rates of lung, bladder, pancreas, and kidney cancer relative to SEER-9 blacks. The opposite pattern was evident for white Floridians with significantly higher rates of lung and laryngeal cancer relative to SEER-9 whites. CONCLUSION: Progress in the reduction of tobacco-associated cancers among white Floridians lags behind the progress noted in SEER-9 registries suggesting that additional state-directed smoking prevention and smoking cessation measures are needed.

Prevention myths and HIV risk reduction by active drug users.

Drug injectors and crack users (526) in South Florida responded to a survey questionnaire that was designed to examine belief in the effectiveness of various strategies, other than condom use, employed to reduce personal risk of contracting HIV during sexual acts. Each strategy was believed to be effective by at least one quarter of the study participants. Factor analysis was used to group these strategies. Subsequent multivariate analysis indicated that the participants who believed in the effectiveness of HIV prevention strategies other than condom use were also less likely to report using condoms. These findings highlight the need for prevention interventions to elicit prevention myths and the full range of risk reduction strategies practiced.