To set the stage for the rest of the workshop, Howard Koh (Harvard University T.H. Chan School of Public Health) presented on the importance of health research on small populations. Scarlett Lin Gomez (University of California, San Francisco) discussed data issues in studying small populations, followed by Lisa Signorello (National Cancer Institute), who talked about her experience in fielding studies in underrepresented populations. The session was moderated by steering committee chair Graham Colditz (Washington University in St. Louis).
THE IMPORTANCE OF HEALTH RESEARCH ON SMALL POPULATIONS
Howard Koh opened by referring to how the World Health Organization (WHO) defines and describes health in its Constitution:1 “A state of complete physical, mental, and social wellbeing and not merely the absence of disease or infirmity” and “the enjoyment of the highest attainable standard of health is one of the fundamental rights of every human being.” Koh said these quotes capture the public health goals to promote and achieve the highest standard of health for ourselves, our families, our community, and for society at large.
While Koh was the assistant secretary for health (ASH) in the Department of Health and Human Services (HHS), he oversaw the unveiling of
Healthy People 2020, an initiative founded in 1979 by ASH/Surgeon General Dr. Julius Richmond. It represents the nation’s public health agenda and offers goals and objectives for the country for each decade. Some of the overarching goals of Healthy People 2020 are attaining high-quality, longer lives for all members of society; achieving health equity and eliminating disparities; and improving the health of all groups and promoting quality of life, healthy development, and healthy behaviors across all life stages. These goals are particularly pertinent and challenging for small populations, when size, dispersion, or accessibility makes it difficult to obtain adequate sample sizes and relevant data.
Koh said the need for more specific, consistent, and disaggregated data grows more important as the nation becomes more diverse. Demographers project that the United States will become a majority-minority country (more than 50% nonwhite) in the next several decades. For example, the Asian American population, currently at about 6 percent, is expected to more than double by 2065.2 As the country grows more diverse, this diversity brings forth new health challenges and opportunities. Of note, dimensions of diversity extend well beyond race and ethnicity to include socioeconomic position, level of disability, and sexual orientation and gender identity among others. Major surveillance tools that can help the health community obtain the necessary data include cancer registries and major national health surveys, such as the National Health Interview Survey (NHIS), the National Health and Nutrition Examination Survey, the Medical Expenditure Panel Survey, the Behavior Risk Factor Surveillance Survey (BRFSS), the National Survey on Drug Use and Health, and the Current Population Survey. With these and other data sources, the question remains about the capacity to capture the data necessary to address critical health disparities.
Koh then specifically addressed the data and public health issues in three critical areas: (1) lesbian, gay, bisexual, and transgender (LGBT) health; (2) Asian American, Native Hawaiian, Pacific Islander (AANHPI) health; and (3) the health of homeless populations.
Until recently, LGBT health issues drew on few nationally representative surveys and very limited data. Thanks to the efforts of then Secretary of HHS Kathleen Sebelius, an LGBT workgroup at HHS, and other activities during Koh’s tenure, NHIS added sexual orientation questions (lesbian, gay, bisexual) in 2013. In addition, Healthy People 2020 added as goals an increase in the number of population-based data systems and the number of states and territories that include questions on sexual orientation and gender
2 Pew Research Center. (2015). The Changing Face of America, 1965-2065. Available: http://www.pewresearch.org/fact-tank/2016/03/31/10-demographic-trends-that-are-shaping-the-u-s-and-the-world/ [March 2018].
identity in the BRFSS. The NHIS numbers reported in the 2014 National Health Statistics Reports were low (less than 3% reported being lesbian, gay, or bisexual),3 but issues of underreporting may have complicated outcomes. The most recent Gallup estimate (2016)4 reports that 4.1 percent of the population (just over 10 million people) self-identify as LGBT.
The Youth Risk Behavior Surveillance System (YRBSS), managed by the Centers for Disease Control and Prevention, started asking about sexual orientation on its survey several years ago. Zaza and colleagues (2016)5 have estimated that about 1.3 million U.S. high school students self-identify as LGB. Compared with their heterosexual peers, they report significantly higher prevalence and adverse outcomes such as being bullied in person or electronically, experiencing violence (both physical and sexual), and other adverse outcomes. The report also noted that approximately 500,000 high school students reported being unsure about their sexual identity, which Koh said underscores the need to be supportive to youth about these topics, particularly at this life stage.
Koh noted recent advances in cancer research in the LGBT community, including efforts by the National Cancer Institute (NCI) and the Food and Drug Administration. A study by Boehmer and Elk (2016)6 summarizes issues and risk factors that affect cancer rates and prevention control for this subgroup. The National Academies of Sciences, Engineering, and Medicine (2017)7 also looked at health disparities, including for the LGBT population, documenting the status of areas such as health insurance coverage, mammogram rates, HIV infection rates, stigma, bullying, and violence.
Koh noted that the AANHPI population comprises more than 50 ethnicities and more than 100 different languages. Two-thirds are foreign born, and one-third have low English proficiency.8 The White House Initiative on Asian Americans and Pacific Islanders is working to advance research data collection, analysis, and dissemination on these populations. Section 4302 of the Affordable Care Act (ACA) required that all HHS-sponsored health
3 Department of Health and Human Services. (2014). Sexual orientation and health among U.S. adults: National Health Interview Survey, 2013. National Health Statistics Reports, 77. Available: https://www.cdc.gov/nchs/data/nhsr/nhsr077.pdf [March 2018].
4 See http://news.gallup.com/poll/201731/lgbt-identification-rises.aspx [March 2018].
5 Zaza, S., Kann, L., and Barrios, L.C. (2016). Lesbian, gay, and bisexual adolescents: Population estimate and prevalence of health behaviors. Journal of the American Medical Association, 316(22):2355-2356.
6 Boehmer, U., Elk, R., eds. (2015). Cancer and the LGBT Community. Heidelberg, Germany: Springer International.
7 National Academies of Sciences, Engineering, and Medicine. (2017). Communities in Action: Pathways to Health Equity. Washington, DC: The National Academies Press. Available: https://www.nap.edu/catalog/24624.
surveys involving self-reporting offer seven subgroups for Asian Americans: Asian Indian, Chinese, Filipino, Japanese, Korean, Vietnamese, and Other Asian. Pacific Islander was further subdivided into four subgroups: Native Hawaiian, Guamanian or Chamorro, Samoan, or other Pacific Islander. This level of granularity allowed for more accurate reflection of the diversity of Asian American culture for the first time.
Koh noted several health outcomes for which the AANHPI population fares worse than non-Hispanic whites, including higher rates of hepatitis B infection and diabetes. There appears to be a higher diabetes risk at each body mass index (BMI) level for Asian Americans than for other groups, which led the American Diabetes Association to recommend diabetes testing for Asian American adults with a BMI of 23 (versus 25 for the general population).9
The Cancer Prevention Institute of California’s Cancer Registry is also utilizing a similar set of AANHPI subgroups to track incidences of various cancers. For example, Torre et al. (2016)10 show how cancer rates vary by such subgroups in comparison to non-Hispanic whites. Liver and stomach cancers are more prevalent among AANHPI males than non-Hispanic white males, with considerations that increased rates of the former were driven by hepatitis B among other factors.
Koh said that despite many investigations concerning the impact of the ACA on health insurance coverage, virtually no data existed relative to AANHPI groups. In an effort to locate more data and improve analysis, Koh, John Park, Graham Colditz, and others are collaborating to document potential reductions in overall uninsured rates and disparities among AANHPI subgroups. Results of this work are pending.
Koh said lack of health care for the homeless population presents one of the largest health disparity challenges. The population is in many ways disadvantaged, but data are difficult to obtain. An estimate from the Department of Housing and Urban Development (HUD) placed the homeless person count in 2015 at more than 560,000; however, other estimates place the count as high as 3.5 million.11 Koh explained that accurate counts are elusive because the definitions of homelessness vary, the population is peripatetic, and robust surveillance systems are lacking. Many cities, for
10 Torre, L.A., Sauer, A.M., Moon, S.C., Kagawa-Singer, M., Jemal, A., and Siegel, R.L. (2016). Cancer statistics for Asian Americans, Native Hawaiians, and Pacific Islanders 2015: Convergence of incidence between males and females. CA: A Cancer Journal for Clinicians, 66(3):182-202.
example, employ point-in-time measurements, literally counting the number of homeless people on the street on a given night of the year.
Koh noted Stories from the Shadows (O’Connell, 2015)12 as a gripping portrayal of the issues that homeless people endure, including high rates of cancer fueled by excessive tobacco use, high rates of communicable diseases such as tuberculosis, HIV/AIDS, and hepatitis, and high prevalence of violence in shelters. One study by Baggett and colleagues (2017)13 documented a higher burden of cancer for the homeless compared to the general population by linking client lists from a 28,000-person cohort of homeless people with outcomes in the Massachusetts Cancer Registry.
Koh then reviewed the debate of whether to treat underlying issues (e.g., substance use and mental health challenges) in people who are homeless before providing housing versus first providing housing. Many have endorsed what is known as the Housing First approach that provides housing as well as supportive services (“supportive housing”), but debates focus on the cost-effectiveness of such an approach. The National Academies of Sciences, Engineering, and Medicine has convened a committee to address many of these issues; its report is due to be released in 2018.
The veteran homeless population has received special attention. The Department of Veterans Affairs announced a goal to end veteran homelessness and has partnered with HUD to distribute housing vouchers to veterans and provide grants to service providers. Several states and communities have announced that they have ended, or are about to end, veteran homelessness. Using specific data to tailor programs and sustain progress should be a priority for the future.
DATA ISSUES IN STUDYING SMALL POPULATIONS
Scarlett Gomez presented statistics from a 2012 Pew Report14 on the rapidly growing Asian population. She said that while some researchers took issue with the way the report seemed to confirm the “model minority” myth by concluding that the Asian American population was well assimilated and doing quite well in the scope of the larger American society, the report proved useful by disaggregating data on Asian American groups. Gomez emphasized the importance of disaggregation for seeing the heterogeneity between the groups. For example, the percentage of those with
12 O’Connell, J.J. (2015). Stories from the Shadows: Reflections of a Street Doctor. Boston, MA: Boston Health Care for the Homeless Program.
13 Baggett, S., Chang, Y., Porneala, B.C., Bharel, M., Singer, D.E., and Rigotti, N.A. (2015). Disparities in cancer incidence, stage, and mortality at Boston Health Care for the Homeless Program. American Journal of Preventive Medicine, 49(5):694-702.
14 Pew Research Center. (2012). The Rise of Asian Americans. Available: http://www.pewsocialtrends.org/files/2013/04/Asian-Americans-new-full-report-04-2013.pdf [March 2018].
TABLE 2-1 Breast Cancer Incidence Rates,* California, 1988-2004
|Race/Ethnicity||Combined (U.S. + Foreign born)||U.S. born||Foreign born||Rate Ratio (95% CI) (U.S./Foreign)|
NOTES: *Rates, per 100,000, adjusted to the U.S. 2000 standard. CI = cancer incidence.
**N-H white = non-Hispanic white.
***Asian = Chinese, Japanese, Filipino, Korean, Vietnamese, and South Asian.
SOURCE: From workshop presentation by Scarlett Gomez. Adapted from Gomez et al. (2010).
limited English proficiency ranges anywhere from 2 to 52 percent depending on the AANHPI group.15 The report also showed the high variability in rates of poverty for AANHPI, which Gomez said was important because poverty is a critical social determinant of health that can greatly impact health outcomes.
Gomez presented information from a case study she and her colleagues conducted on breast cancer in Asian American women.16 While rates of breast cancer among Asian Americans on the whole are considerably lower than rates among the non-Hispanic white population, the study found that once the Asian ethnic groups were disaggregated, the rates ranged greatly, from a low of 46 per 100,000 among Korean Americans to 102 and 100 per 100,000 among Japanese and Filipino Americans. When they broke the groups down further and looked at place of birth (U.S. born versus foreign born), they found almost two-fold differences among Chinese Americans—from 122 per 100,000 for U.S. born to 66 per 100,000 for foreign born (see Table 2-1).
Gomez and her colleagues’ study showed other differences in cancer rates based on immigration patterns and length of residence in the United States. They have since measured breast cancer mortality rates among dif-
15 From “A Community of Contrasts. Asian Americans, Native Hawaiians, and Pacific Islanders in California.” Asian American Center for Advancing Justice. 2013. Available: https://www.advancingjustice-alc.org/wp-content/uploads/2013/03/Communities_of_Contrast_California_2013-1.pdf [March 2018].
16 Gomez, S.L., Quach, T., Horn-Ross, P.L., et al. (2010). Hidden breast cancer disparities in Asian Women: Disaggregating incidence rates by ethnicity and migrant status. American Journal of Public Health, 100(Suppl. 1):S125-S131.
ferent groups, looking at trends based on tumor type and looking at stage of diagnosis as an indicator of early detection.
Several of her colleagues were curious why an increasing number of Asian American women were presenting with breast cancer at a much younger age. Using Surveillance, Epidemiology, and End Results Program data as the basis for more information, three separate studies have shown higher incidence rates in younger Asian American women (age 50 or younger) relative to non-Hispanic white women in the same age range. Gomez and colleagues (2010) showed that this pattern was much more pronounced among U.S.-born Asian American women. However, Gomez said, the increasing incidence trends among young Asian women also have been seen in Asian countries, which is likely attributable to changing breast cancer risk factors in these countries.
In Gomez and colleagues’ most recent paper (2017),17 they used California Cancer Registry information to address the question of breast cancer subtypes. The 2017 study showed higher prevalence and higher incidence of HER2/neu subtype tumors among some Asian American ethnic groups. Compared to non-Hispanic white women, Filipino and older Vietnamese women in particular had higher incidence rates of several HER2 subtypes of breast cancer.
Gomez said although her studies focused on breast cancer, diverging heterogeneous patterns of cancer among Asian Americans hold true for many cancer sites. Likewise, diverging cancer trends can also be observed for Native Hawaiian and Pacific Islander groups. She pointed to a lack of representation of Asian American subgroups in the current NCI-funded epidemiology cohorts, and acknowledged that small numbers and inconsistent capturing of ethnicity data play a part. She emphasized the need to consider the intersectional effects of multiple social determinants of health—not just ethnicity and nativity, but also language, socioeconomic status, and other factors. Gomez noted the potential in leveraging designs such as migrant studies, longitudinal studies, and intergenerational studies to gain insights into cancer and other diseases.
FIELDING STUDIES IN UNDERREPRESENTED POPULATIONS
Lisa Signorello began her presentation by highlighting research she and her colleagues conducted through the Southern Community Cohort Study
17 Gomez, S.L., Von Behren, J., McKinley, M., Clarke, C.A., Shariff-Marco, S., Cheng, I., Reynolds, P., and Glaser, S.L. (2017). Breast cancer in Asian Americans in California, 1988-2013: Increasing incidence trends and recent data on breast cancer subtypes. Breast Cancer Research and Treatment, 164(1):139-147.
(SCCS), which officially started in 2001.18 The SCCS was a large study that followed 85,000 people in the southeastern United States in a prospective cohort, highly motivated by a known high cancer burden in that region. The cohort comprised targeted groups that were underrepresented in cancer research and/or had issues accessing proper care, including African Americans, rural Americans, and Americans living in poverty. She said these groups experience high geographical dispersion and major competing priorities, such as housing, food, and employment. The groups also experience major health disparities, which may be due in part to lack of seeking care because of a long-standing distrust of medical and research communities. While the study focused on minorities in the southeastern United States, she said that the issues translate to millions of people throughout the country.
Signorello said that at the time the cohort was conducted, she and her colleagues had not considered what they were doing to be “small population research” in the way that Shobha Srinivasan and her NCI colleagues have framed it, especially considering the size of the study. The populations were hard to reach, she acknowledged, and required increased effort and the expenditure of greater resources to capture them. She noted commonalities, however, between the hard-to-reach and the small data or small population paradigms.
For both small and hard-to-reach populations, Signorello said the goal is to fill important gaps in knowledge. Both scenarios involve dealing with populations who have been untapped for research and may be unfamiliar with or have trepidations about biomedical research. Conversely, if a population is very small and specific, she said, it may have already been tapped several times and could be fatigued or disenchanted by the research process, as is the case for some Native American communities. These circumstances require enhanced efforts in the primary stages of study design: accessing the population, recruitment, data collection, and retention/follow-up.
For the SCCS, Signorello and her colleagues solicited the help of local community health centers (CHCs) to enroll cohort participants in person. The CHCs are federally certified, have served these communities over time, and care about their health statuses. They chose this recruitment method over mail and telephone recruitment strategies, which had yielded less-than-optimal results in early testing. For more than 7 years and with the assistance of 71 CHCs, they recruited 73,000 participants throughout the Mississippi Delta, Appalachia, and the rural South; 65 percent were African American, and 61 percent had a total household income of less than $15,000 per year.
According to Signorello, the success of the cohort enrollment stemmed
18 Signorello said her views do not necessarily reflect official views of the U.S. government.
from the work her group did prior to the study, which involved community engagement, trust building, and developing cultural competencies. The CHCs are led by a board of health service users in their communities and are part of a larger national network of centers. Signorello and her colleagues took time to understand decision-making structures and priorities at the centers to build trust and confidence. They spent a lot of time “on their turf and on their terms,” engaging and absorbing information from the CHC staff and their patients. Buy-in was especially important since the longitudinal observation study would not yield immediate or direct benefits to the participants. She said they found a great will on the part of community members to address the high burden of disease in their communities for the future.
Signorello explained the study required increased staff capacity for complicated and comprehensive research activities. Instead of bringing in research staff, they hired from the local community and trained new hires and internal staff on handling specimens and following quality control guidelines. To do this, SCCS staff had to account for differences in infrastructure, patient flow, and other characteristics between the CHCs and tailor their approaches accordingly. While the research was a priority, Signorello said the main goal was to continue providing quality and timely health care at the CHCs. To help with this, the SCCS made aggregate data available to CHCs at least once annually; the CHCs could use the data and draw on the academic partnerships to inform and enhance their work. They could also incorporate the data into federal, state, and local grant requests.
Signorello discussed challenges with collecting data for unique populations that may have unique types and rates of exposure. It is important to understand the context of the population. Data collection tools and methodology may need tailoring—for example, with new biomarkers or geospatial indexes. A survey validated in one population may not work for another. Language and literacy (both general and health-related) issues often come into play, as well as residential instability such as homelessness. She said the latter issue also comes into play when attempting to follow up with cohort participants, which is essential for longitudinal studies. She and her colleagues collected as much data as possible during a first interview. They are hopeful that technological advances will assist future efforts with participant retention.
Signorello emphasized the importance of including underrepresented groups in research to improve science, public health, and health equity. It may be unchartered territory that requires careful planning and additional resources; as such, the goal should be to develop methods that are effective in reaching the target populations but also are cost efficient.
Mandi Pratt-Chapman (George Washington Cancer Center) commented that she appreciated Koh’s discussion about data on the LGBT population. She noted a need for systems change, better research, and medical curricula on the topic, as well as heightened attention to diversity and cultural competency for researchers, clinicians, and the public health workforce.
Kellan Baker (Johns Hopkins Bloomberg School of Public Health) asked Koh about the discussions in the health community around scaling back the focus on subpopulations for Healthy People 2030. Koh responded that emerging data on some subpopulations have clarified needs and stimulated programs to address them and that more subpopulations with specific needs have yet to be identified. He expressed his hope that Healthy People 2030 would continue to emphasize the importance of such work as the country grows increasingly diverse.
Anna Napoles (National Institute on Minority Health and Health Disparities) asked Signorello if the workforce trained to manage the cohort data collection had been or would be trained to implement the interventions yielded by the data collected. Signorello said SCCS wrote grants for several CHCs to start programs including smoking cessation, weight reduction, and other intervention-based programming. SCCS staff also formally trained CHC workers to become patient navigators.
Robert Croyle (NCI) commented that small populations can prove difficult to research because they do not fit the standard paradigm that encourages researchers to fully understand a problem before they intervene. He said iterative research models are a better choice for small population research. Instead of surveillance occurring before and/or apart from interventions and modifications, Croyle suggested that the observations need to generate hypotheses for survey mechanisms, which can then inform intervention strategies.
James Allen (University of Minnesota Medical School) underscored the concept of establishing trust in studying hard-to-reach populations. Noting that trust is created and strengthened over time and often builds an expectation of reciprocity, Allen asked about funding mechanisms to help develop these relationships. He suggested the group think about how to do this more effectively. Croyle responded that the National Institutes of Health is addressing this issue in the All of Us research cohort.19 Canada and other countries are already combining cohort and interventional research. He noted some cohort leaders are hesitant to “spoil” the cohort or undermine the research by introducing early interventions, but other countries have
shown the negative effects can be combatted by more frequent serial data collection.
Patrick Sullivan (Emory University) asked about questions on sexual identity in the NHIS, YRBSS, and American Community Survey. He asked if this would turn into a partisan issue affected by the political environment and, if so, the best strategies to promote the value of these added measures. Koh said that these issues and other diversity concerns should center on data and science and not politics. All members of society (particularly those desiring increased representation) can raise their voices to policymakers and other leaders to align forces and advocate for these changes.
Amy Kilbourne (University of Michigan) said a robust area of implementation science is growing, thanks in part to efforts within NCI. She pointed to many opportunities for investigators working in epidemiologic surveys and for trained community health workers to serve as catalysts to deploy and implement best practices in routine care more rapidly.
This page intentionally left blank.