Women and Health Research: Ethical and Legal Issues of Including Women in Clinical Studies: Volume 2: Workshop and Commissioned Papers (1994)

Ruth E. Zambrana

In research, legal and ethical issues are embedded in the scientific process, that is, in the current paradigms and methodological approaches used in the study of racial and ethnic groups in the United States. The purpose of this commentary is to discuss the methodological and conceptual areas relevant to the inclusion of Latino women in research and clinical studies. The main issues in this area have been the limited recognition of the importance of the use of racial and ethnic identifiers in investigations, narrow conceptual paradigms, and inappropriate data collection procedures.

The lack of adequate descriptions of the respondents in a study (the study sample) has violated the principles of scientific method. There is a current debate regarding how to measure racial and ethnic identifiers, what their importance and meaning is, and the potential negative implications in standardizing race and ethnic identifiers (King and Williams, 1993). There is clear evidence that race and ethnic identifiers are merely one set of indicators that are highly interrelated with socioeconomic indicators (which most investigators also are generally reluctant to carefully measure). There is a consensus in the health services research field that poverty or low-income status is the strongest predictor of the use of health services and health outcome. Yet there is limited understanding of the complex processes and factors which influence the pathways to unfavorable health outcome in Latino groups. In addition, socioeconomic status and race and ethnicity must be adequately measured as separate and independent variables to examine their effects on health behaviors, psychosocial factors, and institutional factors (access, cost, and quality

of services) (Williams, 1990).

The 1990 PHS/NIH policy ruling that required the inclusion of racial and ethnic groups and women forced the scientific community to examine issues of inclusion of these groups in studies, but did not provide guidelines for determining adequate sample size of these groups in a study. Thus there remains the central concern that current and future studies may not include sufficient numbers of Hispanic women for subgroup differences to be estimated, a methodological flaw rendering study data scientifically and practically unusable.

There has been an unprecedented increase in Latino and immigrant populations in the United States since 1970. In 1990, the number of Latinos in the United States reached 22 million (9.2 percent of the total population). It is expected that Latinos will continue to grow at the rate of 33 percent over the next decade. The Latino population is quite diverse, the largest group being Mexican Americans (63 percent), who are concentrated mainly in Texas, New Mexico, Arizona, and California; Puerto Ricans (15 percent), in New York, Boston, Chicago, and Washington, D.C.; Central and South American and other Hispanics (18 percent); and Cubans (6 percent), concentrated in Florida.

Immigration to the United States has been greatest among Mexican and Central American refugees owing to political and economic events in their country of origin. The Immigration and Naturalization Service (INS) recorded 2.7 million Latin American immigrants to the United States between 1960 and 1970 (Puerto Ricans, who are U.S. citizens, are not included in this count), with an increase of about 500,000 per year. At present one-third of all births in California are to foreign-born Hispanic women.

In 1990, the poverty rate for Hispanic persons was 28.1 percent and these rates were higher for families with a female head (48.3 percent). Current knowledge in the area of Latino health suggests that the health of Latino women, children, and their families seriously affects and is affected by their socioeconomic position in society. The following points provide a brief profile of Latinos in the United States and depicts subgroup differences.

1. The majority of Latino families tend to be larger than non-Hispanic white families (4.4 compared to 1.8), less educated (M = 8.8 years of schooling for Mexican immigrants, 11.9 for those of native Mexican origin, and 13.2 for Anglos), at a lower income level ($21,800 vs. $42,000 for non-Hispanic whites), younger (mean age 23 years vs. 31 years for non-Hispanic whites) and less likely to have medical insurance. It is estimated that 53 percent of Latinos have no health insurance in California, partly as a result of the types of work they do

as farm workers, domestic workers, and day workers.

2. Latino women have higher fertility, tend to begin childbearing at younger ages, and are three times more likely than non-Hispanic white or African American women to delay prenatal care, that is, initiate care in the third trimester. However, among all poor and racial and ethnic women only about 58 percent initiate care in the first trimester, compared to 80 percent of the general population.

3. Delayed use of health care, organizational and financial barriers, knowledge, attitudes, and beliefs have all been documented as a set of interrelated factors that seriously influence the health outcomes of Latino groups. These barriers are heightened by immigrant status.

4. The leading causes of death in Latino women are diseases of the heart, diabetes, and cancer (breast, lung). For example, females of Mexican origin account for 48 percent of all deaths from cancer in Texas. However, the prevalence and incidence of these diseases varies significantly across Latino subgroups (USDHHS, 1985; Frank-Stromberg, 1991; Desenclos and Hahn, 1992).

At present there are few data sets that include sufficient number by Latino subgroups to conduct any meaningful analyses (USPHS, 1992; Amaro, 1993). In 1990, extensive data were published from the Hispanic Health and Nutrition Examination Survey (H-HANES), undertaken by the National Center for Health Statistics (see JAMA, 1991; AJPH, 1991). H-HANES is the first large-scale health survey to target the Latino population in the United States, specifically Mexican Americans, mainland Puerto Ricans, and Cuban Americans. The total sample comprised 12,000 Latinos from the three ethnic subgroups, who lived in three geographic areas (namely, five southwestern states, New York City and its surrounding area, and Dade County, Florida). Although the value of the data obtained by H-HANES should not be underestimated, it is still crucial to note that these data were collected during 1982–1984. Thus, although this database is one of the largest comprehensive studies regarding the health status of Latinos, it has taken almost ten years for these data to become available.

In spite of the documented differences found in disease patterns among Latino subgroups, current plans for the H-HANES will include only Mexican Americans. In its most recent report, the PHS documented that most record-based surveys have no entry for "race and ethnicity," which severely limits the availability of tabulations about racial and ethnic populations in the United States (USPHS, 1992:17). With specific reference to Hispanics, vital statistics data are even more limited. Although there was a revision to the U.S. Standard Birth and Death Certificate in 1989 with clear race and specific ethnic identifiers, there is

no standard format established for the states to report these data (USPHS, 1992:27).

Heterogeneity within the Hispanic population clearly requires that data be collected by subgroup and analyzed separately. Data from the H-HANES and national birth data show distinct differences in patterns of clinical outcome by subgroup, that is, Puerto Ricans, Mexican Americans and Cuban Americans (JAMA, 1991). In addition, there are distinct differences between Mexican immigrants and Mexican Americans. These differences are related to place of birth (nativity), socioeconomic status, geographic context, and lifestyle behaviors. These variables must be incorporated and measured in future research studies.

The scientific approach to understanding the complex needs of Latino women, particularly in the field of health, has been driven by a rigid, narrow, and homogeneous set of principles which have not permitted the exploration of diversity in the health trajectories of these women. The fields of health services research, biomedical research, demography, and social epidemiology have narrowly and inconsistently defined the parameters of disease patterns as unidimensional, that is, occurring outside the context of a social environment, a psychosocial context, and a family context (Williams, 1990; Lillie-Blanton et al., 1993). The developmental life of a disease, the effectiveness of interventions and procedures used, and the patient's clinical outcome must be measured within the social, cultural, and psychological context of the individual's life. Thus the recognition that biologic, psychological, social, and cultural factors interact with clinical processes needs to inform the paradigm that seeks to understand the particular disease patterns, the effectiveness of interventions, and the course of illness in Latino groups.

Past and current research endeavors have exhibited several features: (1) they are generally individualistic, that is, guided by a principal investigator; (2) there is a limited value of the role of collaboration, either across disciplines or with the community or subjects; and (3) the experiences of many researchers are not practice- or community-based. Thus there is a need to improve our conceptual paradigms, methodological approaches, and data collection procedures (Zambrana, 1992).

There have been few guidelines and operational definitions to guide the investigations on low-income Latino groups (Zambrana, 1991, 1992). This has seriously impeded comparability across studies and across national data sets (USPHS, 1992). Research on Latinos, and women of this group, need to incorporate the following characteristics:

1. A sensitivity to and a real world knowledge of the problems experienced by the specific Latino group under study. For example, the socioenvironmental context and the availability and quality of health services varies by geographic region, socioeconomic status, rural/urban setting and immigration status among different subgroups of Latinos.

2. Adequate measurement of study variables, using a set of cross-culturally and socioeconomically appropriate instruments that have been pretested through a set of scientific and systematic steps to assure reliability and validity of the data.

3. Use of an analytic model that permits the examination of the direct effects of race, ethnicity, socioeconomic status, and nativity on health behaviors, psychosocial factors, and their interactions with institutional variables and their relationship to clinical outcomes of interest.

4. Measurement of the role of cultural beliefs with reference to health-seeking behaviors and the specific clinical conditions.

Thus a more comprehensive biopsychosocial model must inform and guide future studies which seek to better understand the intricate relationships among these variables and their influence on the health status of Latino women. Data collection procedures are integral to assuring reliable and valid data. Thus, in the design of a study focusing on Latino women, the investigators must understand that context and work closely with appropriate informants of that community so that they can develop appropriate and cross-culturally and linguistically sensitive instruments to collect data, develop appropriate procedures to involve the community, identify representative groups of women in conjunction with community experts, and involve community providers and experts as well as potential subjects in the development and conduct of the study (McGraw et al., 1992).

Efforts to develop mechanisms to protect the rights of research subjects have improved dramatically in the last two decades. The underlying principles guiding the legislative mandate were that potential research subjects be informed of the nature of the study and its potential risks and benefits to themselves and their community, and that they willingly and in full understanding of the research objectives participate and cooperate with the research protocol. The key elements of the informed consent process include disclosure of information, understanding of information, and decision making (Gray and Osterweis, 1986:548). These procedures did not provide any additional guidelines to investigators when dealing with racial and ethnic groups who had a history of

being uninformed experimental subjects in the past, or when conducting studies with language and cultural minorities and individuals with limited education. Thus there are three central areas that require attention in human subject protection procedures for Latino women: prior history as research subjects, language, and education with respect to appropriateness of measurement and instruments, and representation on institutional review boards to assure that ethical procedures are instituted.

In the study of Latino women, there is an extensive history of reproductive abuse. For example, Puerto Rican women were used as experimental subjects in early clinical trials of birth control pills, intrauterine devices, and Emko contraceptive cream in the 1950s. The long-range consequence of these experiments has been high rates of cervical cancer among Puerto Rican women. Furthermore, there is documented evidence of high rates of sterilization of Puerto Rican women, both in New York and Puerto Rico, and Mexican-American women in California and the Southwest (Lopez, 1987; Vasquez, 1988). These and other documented failures by the scientific community (Gamble, 1993) to protect the rights of poor and Latino women have contributed to a lack of trust of many in the Latino community and to a resistance to providing access for investigations.

However, equally important has been the lack of sensitivity of many investigators to issues of Spanish translation and literacy in the translation of instruments and informed consent procedures. A careful review of the Spanish translations of instruments and informed consent procedures immediately reveals several problems: almost all are direct translations of the English, which are generally neither culturally nor linguistically appropriate to low-income Latino groups. The language is often too sophisticated, and thus educationally and linguistically inappropriate. In terms of school completion, only about 60 percent of Latinos graduate from high school in the United States and many immigrants, especially from Mexico and Central America, have limited education.

The method most often used in translating instruments into Spanish is a back or direct language translation technique which makes a number of assumptions that may threaten the validity of the data: (a) the cultural meaning of the words exists in both cultures; (b) the grammatical structure of the language is the same; (c) the existing instruments are adequate for use among low-income and culturally or racially distinct populations; (d) anyone who knows ''Spanish'' can translate since Spanish is the mother tongue of all. This last assumption is the most damaging since it ignores different meanings of Spanish words in different subgroups, educational differences, and colloquialisms by region and group.

In effect, Spanish translation should be reviewed by individuals who are from the ethnic group under study and who have firsthand knowledge and experience of the target population. Methodologically, two approaches can be used: a panel of bilingual individuals of the particular Latino subgroup under

study, who have experience in working with the community, can participate in the review and translation of the instrument to assure correct colloquial words, symbolic meaning, word structure; and pretests of the instruments with subjects who have comparable characteristics to study population must be conducted. These approaches can replace or supplement back translation methodologies (Zambrana, 1992).

Institutional review boards (IRBs) at existing research institutions must judge the acceptability of the research protocol regarding risk, informed consent, confidentiality, and mechanisms to assure that the protocol is understood by the subjects and that subjects freely consent without coercion or perceived threat. In a study conducted by the National Commission for the Protection of Human Subjects in the late 1970s, it was found that IRBs did not appear effective in carefully monitoring forms for inclusion of important information. Forms "tended to be unrealistically difficult to read, and there was "great institutional variation in operation of IRBs." Although in 1983, the President's Commission "recommended that more active educational programs be instituted by the federal government" (p. 548), the central point made by these authors is that IRBs are an example of "the capture of the regulatory process by the regulated'' (p. 549).

Thus members of existing IRBs at research institutions are highly unlikely to have representatives who are familiar with particular Latino groups, with cultural and language issues related to these groups, or with appropriate research paradigms or cross-cultural instrumentation and measurement issues. Thus there may be significant barriers to the assurances of the inclusion of significant number of Latino women and most importantly to the development of ethically sound procedures for their inclusion.

Although there has been a clear recognition of the increased racial and ethnic diversity in our society, the scientific community continues to be entrenched in a scientific method that measures homogeneity and ignores the particular life context of communities in which individuals outside the middle class and dominant culture conduct their lives. There is a pressing need to shift our paradigms in the study of Latino women and to conduct research in ways that assure the ethical integrity of the subjects and the scientific findings (see King and Williams, 1993).

The following five recommendations are integrally related to assuring the ethical integrity and scientific validity of data when including Latino women in research studies:

1. Develop a more comprehensive biopsychosocial model which examines the influence of socioeconomic status, ethnicity, race, nativity (place of birth),

and cultural attitudes and beliefs on access to health services, compliance, and clinical outcome. A model can begin to guide systematic research in the area of Latino health, especially Latino women's health. Conceptual models are the cornerstone of a set of empirically valid research questions and an appropriate design and sampling procedures.

2. Design studies with adequate sample size of Latino women in relevant research studies to assure the analytic ability to examine data by age and ethnic subgroup, to conduct planned comparisons across Latino subgroups, and to establish baseline data for future investigations.

3. Develop mechanisms to foster a significant number of Latino investigators (who are familiar with the local cultural and geographic community) in this effort who will serve as principal investigators and members on IRBs to assure cultural, linguistic, and educational sensitivity in methodological procedures and instrument development.

4. Develop clear and required guidelines for the inclusion of the appropriate groups in the community to help identify appropriate Latino women for study, to involve the women in designing the study, such as in focus groups, and to better understand the study within the community context and in the pre- and pilot-testing of the instruments as subjects and as interviewers.

5. Assemble interdisciplinary teams so that both qualitative and quantitative methodological approaches can be appropriately used to insure contextual understanding of the community, cross-culturally sensitive instrument development, and scientifically useful data both in terms of scientific knowledge building and policy and program relevance to the community under study.

Presently the Latino population in the United States is growing at unprecedented rates and there is a compelling need to generate data that are usable in the community to develop programs and policies that can appropriately address the emerging health issues and problems of Latino women.

To this end, data must be generated which adequately captures their health status and those factors that differentially influence their clinical outcomes.

Amaro, Hortensia. 1993. Using National Health Data Systems To Inform Hispanic Women's Health. Paper presented at the National Center for Health Statistics Public Health Conference on Records and Statistics . Washington, D.C. July 19–21.

American Medical Association. 1991. JAMA, Hispanic Health Issue 265(2):161–296.

Desenclos, J. A. and Hahn, R.A. 1992. Years of Potential Life Lost Before Age 65, by Race, Hispanic Origin, and Sex—United States, 1986–1988. MMWR, 42 (November).

Frank-Stromborg, M. 1991. Changing Demographics in the United States. Cancer 67:1772–778.

Gamble, V.N., and Blustein, B.E. 1994. Racial Differentials in Medical Care: Implications for Research on Women. In A.C. Mastroianni, R. Faden, and R.D. Federman, eds. Women and Health Research: Ethical and Legal Issues of Including Women in Clinical Studies, Vol. 2. Washington, D.C.: National Academy Press.

Gray, B.H., and Osterweis, M. 1986. Ethical Issues in a Social Context. In Linda Aiken and David Mechanic, eds. Application of Social Science to Clinical Medicine and Health Policy. New Brunswick, New Jersey: Rutgers University Press.

King, G., and Williams, D.R. 1993. Race and Health: A Multidimensional Approach to African American Health. Department of Community Medicine and Health Care, School of Medicine, University of Connecticut Health Center. Farmington, CT. (unpublished manuscript)

Lillie, M., Martinez, R.M., Taylor, A.K., and Robinson, B.G. 1993. Latino and African American Women: Continuing Disparities in Health. International Journal of Health Services, 23(3):555–584.

Lopez, I.O. 1987. Sterilization among Puerto Rican Women in New York City: Public Policy and Social Constraints. In Leith Mullings (ed.), Cities of the United States: Studies in Urban Anthropology. New York: Columbia University Press.

McGraw, S.A., McKinlay, J.B., Crawford, S.A., Costa, L.A., and Cohen, D.L. 1992. Health Survey Methods with Minority Populations: Some Lessons from Recent Experience. Ethnicity and Disease, 2:273–287.

National Commission for the Protection of Human Subjects of Biomedical and Behavioral Research, 1978. Report and Recommendations Institutional Review Boards. Washington, D.C.: U.S. Government Printing Office.

Trevion, F.M., Falcon, A. P., and C.A. Stroup-Benham (eds.) 1990. Hispanic Health and Nutrition Examination Survey, 1982–84: Findings on Health Status and Health Care Needs. American Journal of Public Health 80: Supplement.

U.S. Department of Health and Human Services. 1986. Report of the Secretary's Task Force on Black and Minority Health. Infant Mortality and Low Birth Weight , 6.

U.S. Department of Health and Human Services. 1985. Report of the Public Health Service Task Force on Women's Health Issues. Women's Health, 2.

U.S. Public Health Service. 1992. Improving Minority Health Statistics. Report of the Public Health Service Task Force on Minority Health.

Vasquez-Calzada, J.L. 1988. La Populacion de Puerto Rico y su Trajectoria Historica. San Juan: Escuela Graduada de Salud Publica, Recinto de Ciencias Medicas, Universidad de Puerto Rico.

Williams, D.R. 1990. Socioeconomic Differentials in Health. Social Psychology Quarterly, 53(2):81–99.

Zambrana, R.E. 1992. The relationship between use of health care services and health status: dilemmas in measuring medical outcome in low-income and racial ethnic populations. In M.L. Grady and H.A. Schwartz (eds.), Medical Effectiveness Research Data Methods. Rockville, MD: U.S. Department of Health and Human Services. Pp. 103–114.

Zambrana, R.E. 1991. Cross-cultural methodological strategies in the study of low-income racial ethnic populations. In M.L. Grady (ed.), AHCPR Conference Proceedings—Primary Care Research: Theory and Methods. Rockville, MD: U.S. Department of Health and Human Services. Pp. 221–227.