American Journal of Epidemiology Advance Access originally published online on March 8, 2006
American Journal of Epidemiology 2006 164(2):170-175; doi:10.1093/aje/kwj118
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Original Contribution |
Socioeconomic Status and Childhood Solid Tumor and Lymphoma Incidence in Canada
1 Environment Division, Electric Power Research Institute, Palo Alto, CA
2 Cancer Control Research Program, British Columbia Cancer Agency, Vancouver, British Columbia, Canada
3 Health Analysis and Measurement Group, Statistics Canada, Ottawa, Ontario, Canada
Correspondence to Dr. Gabor Mezei, Electric Power Research Institute, 3420 Hillview Avenue, Palo Alto, CA 94304 (e-mail: gmezei{at}epri.com).
Received for publication August 5, 2005. Accepted for publication January 20, 2006.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
The authors examined the relation between neighborhood income, as a measure of socioeconomic status, and childhood cancer. Incident cases of childhood solid tumor and lymphoma in 1985–2001 were identified from provincial cancer registries in Canada. Residential postal codes at the time of diagnosis were used to assign cases to census neighborhoods. Person-years at risk were determined from quintiles of population by neighborhood income, sex, and 5-year age group, constructed using census population data. Poisson regression was used to calculate incidence rate ratios across neighborhood income quintiles. Compared with the incidence rate in the richest income quintile, moderately lower rate ratios of 0.73 (95% confidence interval: 0.63, 0.86) and 0.84 (95% confidence interval: 0.69, 1.04) were observed, respectively, for carcinomas and renal tumors in the poorest income quintile. No association was found for other types of cancer. Although a potential relation between socioeconomic status and childhood cancer cannot be excluded, the overall pattern seems compatible with random variation.
adolescent; Canada; child; incidence; infant; lymphoma; neoplasms; social class
Abbreviations: ICCC, International Classification of Childhood Cancer
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
Socioeconomic status has been known to be associated with the development of chronic diseases (1
–4). Among these chronic diseases are several cancer types that show a gradient in incidence by levels of socioeconomic status among adults. Varying disease incidence by strata of socioeconomic status may provide important clues for the etiologic role of various environmental and behavioral exposures, which are not uniformly distributed among levels of socioeconomic status.
For childhood cancer in general, no strong socioeconomic status-related pattern is known. However, the published data available are quite sparse from which to draw firm conclusions concerning socioeconomic status-related trends of childhood cancers other than leukemia. For childhood leukemia, the socioeconomic status relation has been more extensively examined. Nevertheless, the direction of any relation between socioeconomic status and childhood leukemia incidence is still controversial (5–7). In a recent paper (8) linking data from Canadian provincial cancer registries and Canadian censuses, we reported a modestly decreased incidence of childhood leukemia in the poorest quintile of neighborhood income compared with the richest. In this paper, making use of the same data sets, we set out to explore the relation between socioeconomic status and the incidence of childhood solid tumors and lymphoma.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
All incident childhood (0–19 years) solid tumor and lymphoma cases diagnosed between 1985 and 2001 were identified from 10 Canadian provincial cancer registries in Alberta, British Columbia, Manitoba, New Brunswick, Newfoundland and Labrador, Nova Scotia, Ontario, Prince Edward Island, Quebec, and Saskatchewan, which also include data for the three Canadian territories. The registries cover the entire Canadian population, and it is estimated that they include at least 95 percent of all Canadian cancer cases (9
, 10). One province, New Brunswick, could provide data only for the years after 1988, and another province, Quebec, could provide data only for the years up to 2000. Cancer diagnosis was categorized according to the International Classification of Childhood Cancer (ICCC), 1996 Revision. This classification is based on the diagnostic morphology and topography codes of the second edition of the International Classification of Diseases for Oncology, with some additions for new codes. For each cancer case, the cancer registries provided data on the age at diagnosis, the year of diagnosis, histology (or morphology), site (or topography), sex, and postal code of the place of residence at the time of diagnosis. Population data by census year (1986, 1991, 1996, 2001), sex, and 5-year age groups were obtained from Statistics Canada for the smallest geographic area for which Canadian population data are released: enumeration areas for 1986–1996 and dissemination areas for 2001. Both kinds of areas have a typical population of approximately 400 persons and roughly correspond to US census block groups. For noncensus years, the population data for the closest census year to the year of diagnosis were used (e.g., for 1994–1998, the 1996 census data were used).
Neighborhood income quintiles were defined for enumeration or dissemination areas according to methods developed at Statistics Canada (8). Quintile values were determined for each census during the study period, as detailed below. We ascertained the postal code of the subject's usual place of residence at the time of diagnosis and assigned the neighborhood quintile value derived from the nearest census. Note that census income data in Canada are based on a 20 percent sample.
By use of Statistics Canada postal code conversion software, version 3J (11), the postal code of the subject's residence at diagnosis was linked to the appropriate 1996 census enumeration area. Additional files were used to determine the corresponding 1991 and 1986 census enumeration areas and the 2001 census dissemination areas, on the nearest centroids (latitude and longitude) of those areas with respect to the 1996 enumeration area centroids. Neighborhood income data were obtained from the census nearest in time to the diagnosis.
Neighborhood income quintiles were based on the average income per single-person equivalent in the enumeration area or dissemination area. This measure uses the weights from the Statistics Canada low-income cutoffs to derive multipliers for each household size. For example, for 1996 a single-person household received a multiplier of 1.0, a two-person household received a multiplier of 1.25, and a three-person household received a multiplier of 1.55. The total income of the enumeration area or the dissemination area (average household income times the number of households) was then divided by the total number of single-person equivalents, yielding the average income per single-person equivalent. This is a way of adjusting for household size, since more sophisticated variables (such as the percentage of the population under the low-income cutoff) are not available at the enumeration area or dissemination area level.
Quintiles of the population by neighborhood income were constructed within each area (census metropolitan area, census agglomeration, or residual areas within each province) and then pooled across areas. The reason for creating the quintiles within each area is that housing costs vary enormously across Canada. Rents and house prices in some places (such as most of Quebec and the Atlantic provinces) have historically been much lower than those in Toronto or Vancouver. Thus, area-based quintiles should better reflect differences in disposable income after shelter costs are accounted for, and quintiles calculated by area have in fact revealed greater disparities in effect measures (e.g., differences in life expectancy at birth) than quintiles by national standards (12). The medians of the single-person equivalent average incomes in the five neighborhood income quintiles were Can $22,228, $28,942, $32,880, $37,625, and $47,508.
For rural postal codes and for urban postal codes of outlying suburban and rural areas, the same postal code is generally used for multiple enumeration areas or dissemination areas (average number of areas = 3; range = 2–11). The selection of a single such area for coding purposes is random, but with probabilities respecting the proportions of population with that postal code in each of the possible small areas. Thus, the coding is far less precise than for centralized urban postal codes, which are usually linked to a single enumeration area or dissemination area only. For this reason, we also performed supplementary analyses of the data excluding rural and small-town areas (with a community population of <10,000, where rural postal codes predominate) from the numerator and denominator.
Of the 15,485 reported childhood solid tumor and lymphoma cases, we excluded 884 (5.7 percent) from our analyses for the following reasons: missing or invalid postal code (n = 711); the subject's residential postal code referred to a hospital, school, or university residence (n = 47); no area income quintile could be assigned (n = 125); or incorrect code for sex (n = 1). For 2,204 cases, only the first three digits of the six-digit Canadian residential postal code were available, thereby yielding less precise estimates of income quintile. Excluding them did not substantially affect our results, so they are included in all the results presented here.
For each ICCC category, 5-year age group, sex, calendar period (1985–1988, 1989–1993, 1994–1998, 1999–2001), and neighborhood quintile, age-standardized incidence rates were calculated by the direct method using the Canadian 1991 population as the standard. For each income quintile relative to the highest quintile, incidence rate ratios with 95 percent confidence intervals were computed by Poisson regression. For the Poisson regression analyses, the units of analysis are the 160 strata formed by age, sex, calendar period, and neighborhood quintile. The number of cases observed in the stratum is the outcome variable, and the expected number of cases based on the Canadian population's age-, sex-, and calendar-specific cancer rates was considered as an offset. Tests for heterogeneity in the incidence rate ratios for income quintiles were calculated by creating four dummy covariates. Tests for trend were computed by giving the values 1 through 5 for the income quintile and treating it as a continuous variable. Differences in income quintile rate ratios across sex, age, and calendar periods were tested by use of interaction terms in the Poisson regression model. Goodness of fit of the Poisson regression model was examined by comparing the model deviance with a chi-square distribution with 148 df.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
We could include 94.3 percent of all reported Canadian childhood solid tumor and lymphoma cases (n = 14,601) during the study period in our analyses. A somewhat larger number of male cases (n = 7,789) than female cases (n = 6,812) was reported. Of the included nonleukemia childhood cancers, tumors of the central nervous system (n = 3,426) and lymphomas (n = 3,307) were the most common.
The incidence of most tumor types exhibited a characteristic pattern by age groups. Tumors of the central and sympathetic nervous systems, retinoblastoma, and renal and hepatic tumors showed the highest incidence in the youngest age group (0–4 years), while lymphoma, bone tumors, germ cell tumors, and carcinomas showed the highest incidence in the oldest age group (15–19 years) (table 1).
|
No consistent association was observed across neighborhood income quintiles for group II (lymphomas and reticuloendothelial neoplasms), group III (central nervous system and miscellaneous intracranial and intraspinal neoplasms), group IV (sympathetic nervous system tumors), group VII (hepatic tumors), group VIII (malignant bone tumors), group IX (soft tissue sarcomas), and group X (germ cell, trophoblastic, and other gonadal neoplasms) (table 2). For group V (retinoblastoma) and group XII (other and unspecified malignant neoplasms), the rate ratio estimates were consistently above one for all income quintiles compared with the richest income quintile; however, all confidence intervals included one. For retinoblastoma, no consistent increasing pattern was observed. The elevated rate ratios observed for the group of other and unspecified malignant neoplasms (group XII) are difficult to interpret, given the group's diverse composition and small size. For group VI (renal tumors), the rate ratios were consistently below one for all income quintiles compared with the richest quintile, but with confidence intervals all including one. For group XI (carcinomas and other malignant epithelial neoplasms), rate ratio estimates demonstrated a trend for lower risk with poorer income quintile (ptrend < 0.001), and we observed the lowest relative rate estimate of 0.73 (95 percent confidence interval: 0.63, 0.86) in the poorest income quintile compared with the richest quintile.
|
No significant interaction was observed between income quintile and any of calendar period, sex, or age, and the data were found to fit the Poisson model for all ICCC categories. Repeating the analyses including only urban areas gave similar results (data not shown).
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
In the published literature, we have found reports only on socioeconomic status and its relation to adult cancers, but we have not found any systematic examination of the relation between socioeconomic status and childhood cancer incidence similar to our analysis. A recent review of childhood cancer epidemiology mentioned only a potential association, supported only with limited evidence, between low socioeconomic status and higher risk of soft tissue sarcoma (13
).
We explored the relation between a socioeconomic status surrogate, neighborhood income, and the incidence of childhood solid tumors and lymphoma among cases reported in Canada between 1985 and 2001. Overall, we found no consistent relation between socioeconomic status and various types of solid tumors and lymphoma. For two groups of cancers (retinoblastoma and other and unspecified tumors), lower income quintiles were associated with slightly increased risk. The number of cancers in these groups was, however, relatively small. For renal tumors and carcinomas, poorer income quintiles tended to be associated with lower relative rates. These associations, especially the association for carcinomas, were based on a larger number of tumor cases, and they may deserve further examination. The earlier finding (13) of a higher risk of soft tissue sarcoma in lower socioeconomic status groups was not confirmed by our results. Lack of an association between average neighborhood income and childhood cancer may provide some argument against the causal role of exposures with strong relation to socioeconomic status. It may also call into question the practice of "routine" adjustment for socioeconomic status as a confounder in epidemiologic studies of childhood cancers.
We recognize that our socioeconomic status measure was based on ecologic data, and that estimated neighborhood income may not be representative of individual income. However, as suggested by a recent review (7), neighborhood income and individual-level income may represent different aspects of socioeconomic status; therefore, independent assessment of both types of socioeconomic status measures is warranted. In addition, since socioeconomic status at diagnosis may not be the most etiologically relevant measure, there is a potential for bias if cases systematically experienced increasing or decreasing income over time. This potential for bias may be larger for cancer types with longer lag periods. Moreover, sampling error in the measurement of neighborhood income (from census 20 percent samples), plus any noise introduced in the translation across different vintages of census geographies, would tend to produce a bias toward null effects. The main advantage of using available data sets to study the relation of socioeconomic status and childhood cancer, as we did in our analysis, is that the results are not sensitive to the problems of subject recruitment and subject selection, which are a major concern in case-control studies (14).
In summary, although selective under- or overreporting of certain types of childhood tumors by socioeconomic status or a potential causal relation between some aspects of socioeconomic status and childhood cancers cannot be excluded, the overall pattern that we observed seems highly compatible with random variability.
|
ACKNOWLEDGMENTS |
|---|
This study was supported by the Electric Power Research Institute.
Conflict of interest: none declared.
|
References |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
- Feldman L, McMullan C, Abernathy T. Angina and socio-economic status in Ontario: how do characteristics of the county you live in influence your chance of developing heart disease? Can J Public Health 2004;95:228–32.
- Reynolds P, Hurley SE, Quach AT, et al. Regional variations in breast cancer incidence among California women, 1988–1997. Cancer Causes Control 2005;16:139–50.
- Mao Y, Hu J, Ugnat AM, et al. Socioeconomic status and lung cancer risk in Canada. Int J Epidemiol 2001;30:809–17.
- Mackillop WJ, Zhang-Salomons J, Boyd CJ, et al. Associations between community income and cancer incidence in Canada and the United States. Cancer 2000;89:901–12.
- Greenberg RS, Shuster JL Jr. Epidemiology of cancer in children. Epidemiol Rev 1985;7:22–48.
- Electric Power Research Institute. Selection bias in epidemiologic studies of EMF and childhood leukemia. Palo Alto, CA: Electric Power Research Institute, 2003. (EPRI report no. 1008149).
- Poole C, Greenland S, Luetters C, et al. Socioeconomic status and childhood leukemia: a review. Int J Epidemiol. Advance Access: November 24, 2005. (DOI: 10.1093/ije/dyi248).
- Borugian MJ, Spinelli JJ, Mezei G, et al. Childhood leukemia and socioeconomic status in Canada. Epidemiology 2005;16:526–31.
- Parkin DM, Whelan SL, Ferlay J, et al, eds. Cancer incidence in five continents. Vol VII. Lyon, France: International Agency for Research on Cancer, 1997.
- Turner D. Measuring completeness on the northern frontier. The Canadian Case Ascertainment Project. Presented at the annual meeting of the North American Association of Central Cancer Registries, Salt Lake City, Utah, 2004. (http://www.naaccr.org/filesystem/pdf/NAACCR Annual Meeting 2004 presentation TURNER.pdf).
- Wilkins R. PCCF+ version 3J user's guide. Automated geographic coding based on the Statistics Canada postal code conversion files, including postal codes to May 2002. Ottawa, Canada: Health Analysis and Measurement Group, Statistics Canada, 2002:61.
- Wilkins R, Berthelot JM, Ng E. Trends in mortality by neighborhood income in urban Canada from 1971 to 1996. Health Rep 2002;13(suppl):45–71. (Catalogue 82-003-SIE).
- Linet M, Wacholder S, Zahm SH. Interpreting epidemiologic research: lessons from studies of childhood cancer. Pediatrics 2003;112:218–32.
- Mezei G, Kheifets L. Selection bias and its implications for case-control studies: a case study of magnetic field exposure and childhood leukemia. Int J Epidemiol. Advance Access: November 22, 2005. (DOI: 10.1093/ije/dyi245).
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||