American Journal of Epidemiology Advance Access originally published online on September 7, 2007
American Journal of Epidemiology 2007 166(10):1210-1219; doi:10.1093/aje/kwm192
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
PRACTICE OF EPIDEMIOLOGY |
Recruiting Hispanic Women for a Population-based Study: Validity of Surname Search and Characteristics of Nonparticipants
1 Division of Clinical Epidemiology, Department of Internal Medicine, School of Medicine, University of Utah, Salt Lake City, UT
2 Department of Epidemiology and Population Health, School of Public Health and Information Sciences, University of Louisville, Louisville, KY
Correspondence to Dr. Carol Sweeney, Division of Clinical Epidemiology, Department of Internal Medicine, School of Medicine, University of Utah, 30 North 1900 East, Room AC230, Salt Lake City, UT 84132 (e-mail: carol.sweeney{at}hsc.utah.edu).
Received for publication December 14, 2006. Accepted for publication May 31, 2007.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
Conducting research on the health of Hispanic populations in the United States entails challenges of identifying individuals who are Hispanic and obtaining good study participation. In this report, identification of Hispanics using a surname search and ethnicity information collected by cancer registries was validated, compared with self-report, for breast cancer cases and controls in Utah and New Mexico. Factors influencing participation by Hispanics in a study interview in 2000–2005 were evaluated. The positive predictive value of identification as Hispanic by cancer registry records and surname search was 82.3% for cases and 73.2% for controls. Hispanics who were correctly classified differed from those who were misclassified, reporting lower language acculturation and educational attainment. Older age was positively associated with success in contacting Hispanic controls (ptrend < 0.0001) but negatively associated with cooperation with the interview (ptrend < 0.0001). Community characteristics described by US Census data, including income, education, and urban/rural residence, did not significantly influence participation by Hispanic cases or controls. The authors conclude that a surname search efficiently identifies Hispanics, although individuals identified using this method are not completely representative. Recruitment of Hispanic cases and controls does not appear to be affected by selection bias related to community characteristics.
case-control studies; data collection; Hispanic Americans
Abbreviations: GUESS, Generally Useful Ethnic Search System; PPV, positive predictive value
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
The Hispanic population in the United States experiences differences in incidence and outcomes of a number of diseases, notably diabetes and cancer, compared with the non-Hispanic population (1, 2). Health research focusing on Hispanics is needed, but methodological challenges exist, including identifying individuals who are Hispanic and obtaining good study participation.
Administrative databases may have no information or incomplete information to distinguish individuals of Hispanic ethnicity. Individuals with Hispanic surnames can be identified from existing databases by use of algorithms or matching (3–9). The success of these approaches varies by geographic region in the United States and differs for males and females (3–5). Data describing the incidence of disease in the Hispanic population may not be representative if individuals correctly classified by these methods differ from those who are misclassified.
Hispanics have been reported to have low rates of participation in studies (4, 10–15), and differences between participants and nonparticipants can introduce selection bias in case-control studies. Few studies have examined the characteristics influencing study participation among Hispanics (16). Hispanics are more likely than non-Hispanic Whites in the United States to have low income and low educational levels (17), factors which are recognized to be associated with nonparticipation in health research (18). Older age negatively affects participation in studies; the extent to which this holds true in US Hispanics, who have a younger average age than the non-Hispanic population, has not been examined in detail. Language barriers can affect study participation, and lifestyle- and health-related behaviors are thought to vary according to language acculturation among US Hispanics (19–21).
Education, income, and language use in communities are described by US Census data (22). These community characteristics can inform comparisons of study participants with nonparticipants. In this report, we evaluate issues of identifying and recruiting Hispanic women for a case-control study in the US Southwest. Our objectives were to examine the validity of identification of Hispanics based on information obtained by cancer registries and a Hispanic surname search; to describe associations of age, ethnicity, and community characteristics with study nonparticipation for Hispanic and non-Hispanic White women; and to investigate whether factors influencing participation differ by case-control status.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
Study subjects were selected for a population-based case-control study of breast cancer. The selection, recruitment, and interview process have been described in detail in previous reports (23, 24) and are summarized below. The study included four centers, Arizona, Colorado, New Mexico, and Utah; this analysis considers subjects from New Mexico and Utah, who represent 53 percent of subjects selected for the study. (Arizona and Colorado were excluded from this analysis because these centers used different methods for control selection, and because ZIP codes for nonparticipants in Arizona and Colorado were not available.) This research was conducted in accordance with protocols approved by institutional review boards for human subjects' research at the University of Utah and the University of New Mexico.
Subject selection
Women with incident breast cancer were ascertained through the Utah Cancer Registry or the New Mexico Tumor Registry, both of which are US National Cancer Institute Surveillance, Epidemiology, and End Results Program registries. All Hispanic women aged 25–79 years diagnosed with primary breast cancer during October 1999–September 2002 in New Mexico and March 2000–September 2004 in Utah were selected for the study. An age-matched random sample of non-Hispanic White cases was selected. Control subjects, frequency matched on age and ethnicity, were randomly selected from computerized drivers' license lists for persons aged 64 years and younger and from Center for Medicare Studies' lists for those aged 65 years and older. A case or control was selected as Hispanic if her surname was selected by the Generally Useful Ethnic Search System (GUESS) Hispanic algorithm (4, 25) or matched by the Census Spanish Surname lists (3, 26), or, for cases, if Hispanic ethnicity was recorded in cancer registry records from other information, for example, hospital charts.
Screening and recruitment
Each woman selected for the study received a letter informing her of the study, followed by a telephone call to screen for eligibility. During the screening, conducted in English or Spanish, each woman was asked, "How would you describe your racial or ethnic group?" (New Mexico) or "Are you Spanish/Hispanic or Latina?" followed by "What is your race?" (Utah). Women whose response was Hispanic or Latina, Native American, American Indian, White, Caucasian, or non-Hispanic White were considered eligible. Additional eligibility criteria included residence in the state at the reference date and being able to answer interview questions in English or Spanish.
Interview
Participants took part in an in-person interview in English or Spanish. (The questionnaire is available at https://zorro.hrc.utah.edu/.) Information was obtained about history of exposure to breast cancer risk factors, including diet, medical history, reproductive history, cigarette smoking, alcohol consumption, and physical activity. Women were also asked about their educational attainment, and Hispanic subjects were asked about their ability to read and speak English and Spanish. Interview data were used to create scaled variables for analysis, including a lifetime physical activity score, number of pack-years of smoking, and language acculturation score. Body mass index was calculated (weight (kg)/height (m)2).
Community characteristics
We used data from the 2000 US Census (22) to describe community characteristics. We accessed publicly available census information about populations residing in each ZIP Code in New Mexico and Utah. The characteristics selected for analysis were as follows: percentage of households with income below the poverty level, percentage urban area, percentage of women aged 25 years or more who did not complete high school, percentage of women who are homemakers, percentage of the population who live in the same house as 5 years ago, percentage of the population who are Hispanic, percentage of Hispanic women aged 25 years or more who did not complete high school, and percentage of Hispanics who speak Spanish and do not speak English or do not speak English well. We matched these variables to each woman selected for the study by ZIP Code of residence. We were unable to geocode subjects to match census data at the more precise census tract or block group units because cancer registries did not release addresses for nonparticipants. Subjects whose residential ZIP Code did not match a 2000 US Census ZIP Code were excluded from analyses using census data. We created categorical variables using quartile cutpoints from the distribution in controls.
Data analysis
We assessed validity of identification as Hispanic from the surname search and cancer registry data by cross-tabulating ethnicity from these sources with self-report and calculating positive predictive value (PPV), the percentage of subjects classified as Hispanic by our initial methods (surname search and cancer registry data) who self-identify as Hispanic. The validity of surname search for cases could not be evaluated separately because the "initial ethnicity" variable provided by the cancer registries incorporated ethnicity data from all sources. We calculated sensitivity, the percentage of self-identified Hispanics correctly classified as such by the initial classification methods, and specificity, the percentage of self-identified non-Hispanics correctly classified as such by the initial classification. Sensitivity and specificity were calculated by use of weighting to account for sampling fractions, that is, to adjust for the fact that Hispanic and non-Hispanic participants were selected in different proportions from their representation in the population. We compared self-reported characteristics (education, body mass index, parity, and so on) of two groups of controls who self-identified as Hispanic during the screening interview: those who were selected by surname search as Hispanic and those initially selected as non-Hispanic, testing for differences using regression models, adjusting for age and study center.
Contact and cooperation rates were calculated as defined by the American Association for Public Opinion Research (27). We calculated contact rates as the number of women whom we were able to contact and invite to participate in the study, excluding those determined to be ineligible, divided by the sum of the number of women known to be eligible plus those selected for whom eligibility was not determined. The result of this calculation is American Association for Public Opinion Research Contact Rate 1, a conservative definition of the contact rate in which subjects for whom eligibility can not be determined are retained in the denominator. This treatment of subjects of undetermined eligibility is commonly used in reporting participation from epidemiologic studies (28). We considered a woman to have been contacted if study staff were able to reach her by telephone or if she returned a reply form declining participation. Noncontacts were women who made no reply to any study mailing and who could not be reached by telephone despite repeated attempts and tracing for new address and phone number information, plus women who moved out of the study area or were deceased after the selection date. Participants were women who took part in the study interview. We calculated cooperation rates as the number of eligible participants divided by the number of women who were contacted and invited to participate in the study, excluding women determined to be ineligible (27). Contact rates and cooperation rates were summarized by age, case-control status, and ethnicity. The initial classification of ethnicity by the cancer registry or surname search was used for this analysis and subsequent analyses because self-report of ethnicity was not obtained for subjects who did not complete the screening. We assessed associations of age and ethnicity with contact and cooperation rates using multivariable logistic regression models, with adjustment for study center and age.
We assessed the validity of census variables as a proxy for individual characteristics in this population by calculating correlations between community characteristics, described by census variables, and self-reported characteristics of interviewed control subjects. The Spearman rank correlation coefficient (theta) and its associated p value were calculated for correlation between the census variable (percentages, treated as a continuous variable) and the interview variable (ordered categories).
We analyzed the associations between community characteristics, as described by census data, and nonparticipation. In this analysis, nonparticipants included both women who were not contacted and those who refused; women determined to be ineligible were excluded. Risk ratios for nonparticipation in the study were calculated by proportional hazards regression models, treating the population as a closed cohort, assigning an equal observation time for all subjects. Likelihood ratio tests of ordered categorical variables were used to evaluate trends in associations. To address possible selection bias, we used likelihood ratio tests of multiplicative interaction terms to test for case-control differences in characteristics associated with participation. SAS, version 9.1, software (SAS Institute, Inc., Cary, North Carolina) was used for statistical analysis.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
Among controls and cases who were initially identified as Hispanic by surname search or cancer registry records and who completed the screening, 73.2 percent and 82.3 percent, respectively, identified themselves as Hispanic (table 1). These percentages represent the PPV of identification as Hispanic by the surname search or by the combination of surname search and cancer registry records. PPVs were higher in New Mexico than in Utah. Identification as Hispanic using these methods had high specificity, greater than 95 percent for both controls and cases, but sensitivity was lower, 59.8 percent for controls and 71.6 percent for cases.
|
We compared self-identified Hispanic control subjects who were initially identified as Hispanic by surname search with self-identified Hispanic women who were initially misclassified, that is, whose surnames had not been identified as Hispanic (table 2). The women identified by surname search had lower levels of English language acculturation, less education, and higher body mass index.
|
We next examined study contact and cooperation rates in relation to initial identification for the study as Hispanic or non-Hispanic White and in relation to age. A total of 2,063 Hispanic controls, 1,014 Hispanic cases, 3,923 non-Hispanic White controls, and 1,846 non-Hispanic White cases were selected for the study. Of these, 287 Hispanic controls, 71 Hispanic cases, 371 non-Hispanic White controls, and 52 non-Hispanic White cases were determined to be ineligible; outcomes for the remainder are shown in table 3. Hispanic women were less likely to be contacted and, if contacted, less likely to participate in the study than non-Hispanic White women (table 3). The contact rate for Hispanic controls was 13 percent lower than that for non-Hispanic White controls (age-adjusted p < 0.0001), and the difference between the groups in cooperation rates for controls was 9 percent (p = 0.0005). Older age among Hispanic women was associated with higher contact rates but lower cooperation rates. The decline in cooperation with age was somewhat stronger for Hispanics than non-Hispanics (in a test for interaction, p = 0.10).
|
Community characteristics from census data were matched by ZIP Code to each individual selected for the study. ZIP codes for 124 Hispanics (4.6 percent) and 234 non-Hispanic Whites (5.0 percent) were not found in census data; these represent either post office box addresses, new ZIP codes created after 2000, or invalid ZIP codes. We observed that educational status in the community as represented in the census data, specifically the census variable for percentage of women in the community who completed high school, correlated with several characteristics reported by control participants (data not shown in tables). The census education variable showed positive correlations with educational attainment reported by Hispanics (Spearman's correlation coefficient, theta = 0.282, p < 0.01), age at first birth for non-Hispanic White women (theta = 0.180, p < 0.01), alcohol consumption for both ethnic groups (theta = 0.122 for Hispanics, p = 0.01; and theta = 0.172 for non-Hispanic Whites, p < 0.01), and physical activity for both groups (theta = 0.130 for Hispanics, p < 0.01; and theta = 0.061 for non-Hispanic Whites, p = 0.05), as well as an inverse correlation with parity for Hispanic women (theta = –0.144, p < 0.01). The census variable for percentage of Hispanics in the community who speak English well was correlated with self-report of language acculturation by Hispanics (theta = 0.276, p < 0.01).
Community characteristics of women who were selected but did not participate in the in-person interview for the study were similar to those of participants (table 4). There were no statistically significant trends in associations between community characteristics and participation among Hispanics overall nor among Hispanic controls or cases separately. The risk ratios that fell farthest from the null, for both Hispanics and non-Hispanic Whites, were for variables describing education and language. Hispanic controls and cases who lived in communities in the quartile with the highest percentage who do not speak English well were estimated to be 9 percent and 23 percent, respectively, more likely to be nonparticipants. Hispanic controls and cases living in communities that fell in the highest quartile of Hispanic women not graduating from high school were estimated to be 14 percent and 18 percent, respectively, more likely to be nonparticipants. These nonsignificant trends were in the same direction for cases and controls. Tests for interaction revealed no evidence of case-control differences in community characteristics associated with participation among Hispanics. Trends in association between community characteristics and participation were stronger for non-Hispanic White cases than for controls, with the case-control difference in trend reaching statistical significance for the percentage of Hispanic women who did not complete high school (p = 0.04) and the percentage of Hispanics who did not speak English well (p = 0.02).
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
The GUESS algorithm and census Hispanic surname search, with or without cancer registry data, had an overall PPV of greater than 70 percent for identifying Hispanic women in New Mexico and Utah. The PPVs in this study were higher than the 47 percent PPV reported for identification of Hispanic breast cancer cases (5) and the 55.7 percent PPV for Hispanic female health plan members (6) using similar methods in northern California. The PPVs were highest in New Mexico, which can be attributed in part to the high proportion of Hispanics, 42 percent, in the state's population (22) and to the development of the GUESS program in New Mexico (4). High racial and ethnic diversity in northern California, particularly the presence of many Filipino individuals with surnames identified as Hispanic, has been noted to adversely affect the validity of Hispanic surname searches in that region (7).
Surname searches improve the identification of Hispanics in administrative databases, such as those of cancer registries (5), health management organizations (6), and Medicare (7). Our results are in agreement with those of previous studies reporting sensitivities in the range of 58–82 percent for identification of females as Hispanic by computer surname search, alone or in combination with other records (3–6). Stewart et al. (5) estimated that breast cancer incidence for Hispanics in northern California would have a negative relative bias (calculated as the ratio of sensitivity to PPV, minus one) of –20 percent if based on ethnicity reported to the cancer registry alone, but that adding the GUESS search resulted in a positive relative bias of 28 percent. GUESS is not widely used outside New Mexico, and data reported by Stewart et al. (5) from northern California indicate low specificity in that population. However, because of the high PPV of census Spanish surname match and GUESS in New Mexico and Utah, we estimate that there is not a positive relative bias for the application of these methods to cancer registry data in these states.
At the sensitivities reported by us and others, a significant portion (18–42 percent) of self-identified Hispanic women will not be identified as such by surname searches. Our comparison of Hispanic control subjects who were correctly identified as Hispanic by surname search with Hispanic women who were initially misclassified revealed differences in language acculturation, educational attainment, and body mass index. Our data and those of others (8) indicate that misclassification of Hispanic ethnicity by surname search is nonrandom, so that Hispanic women identified by surname searches will not be entirely representative.
Researchers working in New Mexico in prior studies (4) and in other parts of the United States (11–15) have reported low participation rates in health studies for Hispanics compared with non-Hispanic Whites, although some researchers in northern California have obtained high cooperation rates for Hispanics (29, 30). In our study, the patterns of noncontact and noncooperation differed by age. Young Hispanic controls were hard to find but, if contacted, were almost as likely as young non-Hispanic Whites to participate in the interview. A high proportion of older Hispanic women were unwilling to be interviewed. Similar trends of higher contact rates, but lower cooperation rates, with age were reported for Black female controls in a study in North Carolina (31), and success in contacting or tracing younger Hispanics has been low in other studies (15, 32). Greater residential stability of older adults may contribute to better contact rates.
We found no significant associations between any community characteristic described by census variables and study participation. The strongest, although nonsignificant, influences on nonparticipation were low educational levels or a high percentage of Hispanics who do not speak English well. The nonsignificant trends that we observed were qualitatively in agreement with prior observations that a high proportion of non-English speakers in a community (33) or a low literacy level among Latinos (16) adversely influences participation. However, our approach was distinct from that of Link et al. (33), who evaluated participation by all racial and ethnic groups as one outcome, and from that of Frack et al. (16), who reported on follow-up in a longitudinal study. We found that no characteristic affected participation by Hispanic cases differently from controls. The education and language of Hispanics were associated with a case-control difference in study participation by non-Hispanics. This finding may be due to chance, given that a number of comparisons were made. Given lower overall participation by controls, differences between participants and nonparticipants might be expected among controls, but in this example they were observed for cases. Education and language use among Hispanics may be indicators of community social structure characteristics that affect study participation for non-Hispanics.
Census measures of education and language were significantly correlated with self-reported language and education of controls and with exposures that influence breast cancer risk. Thus, correlation between community characteristics measured by the census variables and characteristics of individuals in our sample supports the validity of using census characteristics to evaluate potential selection bias. However, the correlation coefficients were low, and therefore our analysis of the influence of community characteristics on nonparticipation may underestimate individual-level differences between participants and nonparticipants. Even if the effects of factors influencing participation are understated because of the ecologic nature of the census data, several qualitative conclusions likely hold true: The direction and magnitude of trends for factors influencing study participation were similar for Hispanic cases and controls; education and language are the strongest identifiable selection factors for Hispanic and non-Hispanic subjects; and, exposures relevant to breast cancer risk correlate with education and language in the community. Thus, despite disappointing participation rates, there is no evidence of serious differential selection bias in recruitment of Hispanics for this case-control study. Adjusting case-control comparisons for education and language characteristics of individuals and communities would be a good practice, to take into account residual case-control differences introduced by nonparticipation.
Incomplete identification of Hispanics and low rates of study participation by Hispanics are obstacles to research on the health of diverse populations in the United States. Our data confirm a previous report (8) that Hispanics who are identified by surname differ from those who are not. We found no evidence of significant differences, other than age, between Hispanic study participants and nonparticipants.
|
ACKNOWLEDGMENTS |
|---|
This study was funded by National Institutes of Health grants CA078682 and CA078762. The Utah Cancer Registry is funded by contract N01-PC-67000 from the National Cancer Institute, with additional support from the Utah Department of Health.
Conflict of interest: none declared.
|
References |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
- Miller BA, Kolonel LN, Bernstein L, et al. Racial/ethnic patterns of cancer in the United States 1988 –1992. (1996) Bethesda, MD: National Cancer Institute. (NIH publication no. 96-4104).
- Burrows NR, Valdez R, Geiss LS, et al. Prevalence of diabetes among Hispanics—selected areas, 1998 –2002. MMWR Morb Mortal Wkly Rep (2004) 53:941–4.[Medline]
- Perkins RC. Evaluating the Passel-Word Spanish Surname List: 1990 decennial census post enumeration survey results. (1993) Washington, DC: US Bureau of the Census. (Population Division working paper no. 4). (http://www.census.gov/population/www/documentation/twps0004.html 1993).
- Howard CA, Samet JM, Buechley RW, et al. Survey research in New Mexico Hispanics: some methodological issues. Am J Epidemiol (1983) 117:27–34.
[Abstract/Free Full Text] - Stewart SL, Swallen KC, Glaser SL, et al. Comparison of methods for classifying Hispanic ethnicity in a population-based cancer registry. Am J Epidemiol (1999) 149:1063–71.
[Abstract/Free Full Text] - Perez-Stable EJ, Hiatt RA, Sabogal F, et al. Use of Spanish surnames to identify Latinos: comparison to self-identification. J Natl Cancer Inst Monogr (1995) (18):11–15.
- Morgan RO, Wei II, Virnig BA. Improving identification of Hispanic males in Medicare: use of surname matching. Med Care (2004) 42:810–16.[CrossRef][Web of Science][Medline]
- Swallen KC, West DW, Stewart SL, et al. Predictors of misclassification of Hispanic ethnicity in a population-based cancer registry. Ann Epidemiol (1997) 7:200–6.[CrossRef][Web of Science][Medline]
- Swallen K, Guend A. Data quality and adjusted Hispanic mortality in the United States, 1989 –1991. Ethn Dis (2003) 13:126–33.[Web of Science][Medline]
- Yancey AK, Ortega AN, Kumanyika SK. Effective recruitment and retention of minority research participants. Annu Rev Public Health (2006) 27:1–28.[CrossRef][Web of Science][Medline]
- Ashing-Giwa KT, Padilla GV, Tejero JS, et al. Breast cancer survivorship in a multiethnic sample: challenges in recruitment and measurement. Cancer (2004) 101:450–65.[CrossRef][Web of Science][Medline]
- Des Jarlais G, Kaplan CP, Haas JS, et al. Factors affecting participation in a breast cancer risk reduction telephone survey among women from four racial/ethnic groups. Prev Med (2005) 41:720–7.[CrossRef][Web of Science][Medline]
- Kolonel LN, Henderson BE, Hankin JH, et al. A multiethnic cohort in Hawaii and Los Angeles: baseline characteristics. Am J Epidemiol (2000) 151:346–57.
[Abstract/Free Full Text] - Rowland ML, Forthofer RN. Investigation of nonresponse bias: Hispanic Health and Nutrition Examination Survey. (1993) Hyattsville, MD: National Center for Health Statistics. 1–75. (Vital and health statistics, series 2: data evaluation and methods research, no. 119).
- Vernon SW, Roberts RE, Lee ES. Ethnic status and participation in longitudinal health surveys. Am J Epidemiol (1984) 119:99–113.
[Abstract/Free Full Text] - Frack SA, Woodruff SI, Candelaria J, et al. Correlates of compliance with measurement protocols in a Latino nutrition-intervention study. Am J Prev Med (1997) 13:131–6.[Web of Science][Medline]
- Ramirez RR, de la Cruz GP. The Hispanic population in the United States: March 2002. In: Current population reports, P20-545. (2003) Washington, DC: US Census Bureau. (http://www.census.gov/prod/2003pubs/p20-545.pdf).
- Armstrong BK, White E, Saracci R. Chapter 11: response rates and their maximization. In: Principles of exposure measurement in epidemiology. (1992) Oxford, United Kingdom: Oxford University Press. 294–317.
- Lara M, Gamboa C, Kahramanian MI, et al. Acculturation and Latino health in the United States: a review of the literature and its sociopolitical context. Annu Rev Public Health (2005) 26:367–97.[CrossRef][Web of Science][Medline]
- O'Malley AS, Kerner J, Johnson AE, et al. Acculturation and breast cancer screening among Hispanic women in New York City. Am J Public Health (1999) 89:219–27.
[Abstract/Free Full Text] - Slattery ML, Sweeney C, Edwards S, et al. Physical activity patterns and obesity in Hispanic and non-Hispanic white women. Med Sci Sports Exerc (2006) 38:33–41.
- Census 2000. American FactFinder. (2005) Washington, DC: US Census Bureau. (http://factfinder.census.gov/). Accessed April 2005.
- Rogers A, Murtaugh MA, Edwards S, et al. Contacting controls: are we working harder for similar response rates, and does it make a difference? Am J Epidemiol (2004) 160:85–90.
[Abstract/Free Full Text] - Slattery ML, Sweeney C, Edwards S, et al. Body size, weight change, fat distribution and breast cancer risk in Hispanic and non-Hispanic white women. Breast Cancer Res Treat (2007) 102:85–101.[CrossRef][Web of Science][Medline]
- Buechley RW. Generally Useful Ethnic Search System: GUESS. (1976) Albuquerque, NM: Cancer Research and Treatment Center, University of New Mexico.
- Word DL, Perkins RC. Building a Spanish surname list for the 1990 's—a new approach to an old problem. (1996) Washington, DC: US Bureau of the Census. (Population Division working paper no. 13). (http://www.census.gov/population/documentation/twpno13.pdf).
- American Association for Public Opinion Research. Standard definitions: final dispositions of case codes and outcome rates for surveys. (2004) Lenexa, KS: American Association for Public Opinion Research.
- Olson SH, Voigt LF, Begg CB, et al. Reporting participation in case-control studies. Epidemiology (2002) 13:123–6.[CrossRef][Web of Science][Medline]
- Marin G, Vanoss Marin B, Perez-Stable EJ. Feasibility of a telephone survey to study a minority community: Hispanics in San Francisco. Am J Public Health (1990) 80:323–6.
[Abstract/Free Full Text] - John EM, Phipps AI, Davis A, et al. Migration history, acculturation, and breast cancer risk in Hispanic women. Cancer Epidemiol Biomarkers Prev (2005) 14:2905–13.
[Abstract/Free Full Text] - Moorman PG, Newman B, Millikan RC, et al. Participation rates in a case-control study: the impact of age, race, and race of interviewer. Ann Epidemiol (1999) 9:188–95.[CrossRef][Web of Science][Medline]
- Psaty BM, Cheadle A, Koepsell TD, et al. Race- and ethnicity-specific characteristics of participants lost to follow-up in a telephone cohort. Am J Epidemiol (1994) 140:161–71.
[Abstract/Free Full Text] - Link MW, Mokdad AH, Stackhouse HF, et al. Race, ethnicity, and linguistic isolation as determinants of participation in public health surveillance surveys. Prev Chronic Dis (2006) 3:A09.[Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  K. B. Baumgartner, T. J. Schlierf, D. Yang, M. A. Doll, and D. W. Hein N-acetyltransferase 2 Genotype Modification of Active Cigarette Smoking on Breast Cancer Risk among Hispanic and Non-Hispanic White Women Toxicol. Sci., November 1, 2009; 112(1): 211 - 220. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. S. Hamilton, T. P. Hofer, S. T. Hawley, D. Morrell, M. Leventhal, D. Deapen, B. Salem, and S. J. Katz Latinas and Breast Cancer Outcomes: Population-Based Sampling, Ethnic Identity, and Acculturation Assessment Cancer Epidemiol. Biomarkers Prev., July 1, 2009; 18(7): 2022 - 2029. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
L. M. Hines, B. Risendal, M. L. Slattery, K. B. Baumgartner, A. R. Giuliano, and T. Byers Differences in Estrogen Receptor Subtype According to Family History of Breast Cancer among Hispanic, but not Non-Hispanic White Women Cancer Epidemiol. Biomarkers Prev., October 1, 2008; 17(10): 2700 - 2706. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. A Murtaugh, C. Sweeney, A. R Giuliano, J. S Herrick, L. Hines, T. Byers, K. B Baumgartner, and M. L Slattery Diet patterns and breast cancer risk in Hispanic and non-Hispanic white women: the Four-Corners Breast Cancer Study Am. J. Clinical Nutrition, April 1, 2008; 87(4): 978 - 984. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||