2

Setting the Context

This chapter summarizes the presentations of Nancy Krieger and Paula Lantz, which outlined some key concepts and issues important to moving population health science forward.

EMBODIED HISTORY, STRUCTURED CHANCE, AND FLEXIBLE PHENOTYPE AS CONTRIBUTORS TO HEALTH¹

In her presentation, Nancy Krieger of Harvard University described how, despite repeated robust refutations, for over a century numerous scientists and scientific reports have attempted to make causes of disease add up to 100 percent, for example, X percent due to “genes” and (100 – X percent) due to “environment” (or “chance”). However, Krieger stated, it is well known that interactions between causes means that population attributable fractions (PAFs) necessarily must add up to more than 100 percent. Challenging deep-rooted beliefs that underlie persistent errone-

___________________

¹ This section is based on the presentation by Nancy Krieger, professor of social epidemiology, Department of Social and Behavioral Sciences, Harvard T.H. Chan School of Public Health, director, Harvard School of Public Health Interdisciplinary Concentration on Women, Gender, and Health, and the statements have not been endorsed or verified by the National Academies of Sciences, Engineering, and Medicine. Krieger’s presentation “Embodied history + structured chance + flexible phenotype = ∑ ‘causes’ always >100%” presented material that will be part of a future publication; thus, a speaker-prepared synopsis is provided in lieu of a more detailed summary of the speaker’s remarks, along with an extensive bibliography to which the speaker referred to in her remarks, which is found in Appendix C.

ous efforts to force causes of population health to add up to 100 percent, Krieger drew on the ecosocial theory of disease distribution, which takes into account both embodied history and structured chance when analyzing population attributable risk. Embodiment refers to how people literally embody, biologically, their societal and ecological conditions, thereby creating population patterns of health, disease, and health inequities. Structured chance helps clarify how and why population parameters and individual risk are necessarily linked. Observed socially structured patterns of health inequities cannot be explained by either chance or population genetic structure. These health inequities are, in principle, preventable. Krieger stated that recognizing that causes must necessarily add to more than 100 percent can aid in framing and motivating the many different pathways and levels for historically grounded, multilevel, cross-sectoral action to promote equity, improve population health, and rectify health inequities.

RESEARCH DESIGNS AND FRAMEWORKS FOR POPULATION HEALTH IMPROVEMENT

To inform the discussions about research agendas for population health improvement, Paula Lantz of the University of Michigan provided a brief background on research designs, highlighted some of the current debates in the field of population health science, and offered a framework for identifying population health research priorities and actions (highlights are presented in Box 2-1).

RESEARCH QUESTIONS, DESIGNS, AND METHODS

Research questions drive everything, Lantz began, and different disciplines think about research questions in different ways. She described three major types of research questions, and offered examples from the field of public health.

• Exploratory questions describe initial hypotheses on a new topic or idea, such as “How might more stable housing improve quality of life?”
• Descriptive questions develop a deeper understanding and define trends and patterns, such as “What are adolescents’ most trusted sources of health information? What are historical trends and patterns in prescription drug abuse?”
• Explanatory questions seek to establish causal relationships, as in “How does chronic social stress increase cardiovascular disease risk? Will a tax on sugar-sweetened beverages reduce obesity?”

Once the research questions are defined, research design is the architecture of the plan for answering those questions. Basic components of a research design are the following:

• Purpose—to explore, describe, or explain
• Topic
• Unit of analysis, such as individual or micro level, organizational or mezzo level, and population or macro level
• Time dimension—cross-sectional or longitudinal
• Comparisons over time or across groups

Strong research designs are essential for explanatory research. The randomized controlled trial is the gold standard experimental design for studying causal relationships, Lantz said. In population health and policy research, where randomization to different situations is generally not possible, natural experiments (e.g., time–series designs) and quasi-experimental designs are used. Explanatory research designs also include economic analyses of the cost–benefit or cost-effectiveness of different interventions.

Within the context of research design, research methods are the specific ways in which data will be obtained and analyzed to answer the stated research questions, Lantz said. Design and methods are both important, she emphasized. A well-designed trial that does not produce quality data is of no value, as is data collected from a poorly designed study. Data can be primary data, newly collected directly by researchers, or secondary data from surveys, administrative systems, the U.S. Census,

or other existing data sources. Data analysis to answer the research question includes statistical procedures and qualitative analyses, and can be highly specialized.

CURRENT DEBATES AND CHALLENGES

Debates about definitions in population health persist, Lantz said, starting with defining what is meant by the terms population, health, and population health. There are also debates surrounding the definitions of community, socioeconomic status or position, and race. Lantz noted that race has been defined differently over time and across cultures and societies. Another debate revolves around quantitative versus qualitative methods, although Lantz noted a growing recognition of the value of qualitative data and mixed methods. Concerns remain, however, about attempts to use qualitative research to make causal arguments.

Disciplinary differences also come into play. Population health requires that a wide range of disciplines work together and learn from each other, Lantz said. Different disciplines bring different theoretical perspectives, conceptual frameworks, and methods to population health science, to develop new understandings, theories, and methods. Lantz described a recent exchange on Twitter that highlighted critiques of social epidemiology, echoing those she had previously heard from economists working on housing, transportation, and other social determinants of health. Frequent critiques include that epidemiology has weak study designs, and that bad epidemiologic studies are widely disseminated in the popular press, often driven by public relations offices of journals and universities that may overstate the findings of studies. Exchanges such as the one she experienced on Twitter reveal some of the issues with media coverage and the translation of research findings to the public, Lantz said, but they may also reflect some of the disciplinary differences in research design and methods, and legitimate concerns about the evidence required to establish a causal relationship.

IDENTIFYING RESEARCH NEEDS VERSUS DISSEMINATION NEEDS

Setting a research agenda for population health involves understanding disciplinary strengths and differences, and embarking on interdisciplinary research that creates new approaches and insights. In particular, Lantz highlighted the need for new exploratory research, a better understanding of population health phenomena, and better evidence regarding which policies and interventions work and which do not. In addition to new research, Lantz emphasized the need for better dissemination and

use of the large volume of existing research. She noted the need for both improved translation and dissemination of research, and research about translational science (i.e., how to best translate and disseminate findings for action).

In preparation for the small group discussions of priorities for a population health research agenda, Lantz offered a framework for considering where more evidence is needed, and where the evidence already exists but better dissemination and more action are needed (see Table 2-1). She called on participants to consider research priorities relative to the current state of evidence and consensus around that evidence, as well as the current state of evidence-based action (e.g., implementation by decision makers and stakeholders) on the issue.

Lantz provided examples of where she would rank some of the current population health issues in her sample framework, acknowledging that there could be debate about each. She suggested, for example, that there is strong evidence on the public health effects of climate change and gun violence, but weak action, including policy interventions, as these are highly politicized issues. As another example, Lantz placed the Drug Abuse Resistance Education (D.A.R.E.) program in this box, as the evidence and consensus are strong that the program does not work, yet action responding to that evidence has been weak and the program persists in 75 percent of elementary schools. Using this approach, Lantz said that priority areas for population health research would be those for which there is not enough evidence and for which evidence-based action on policy and practice is lacking (i.e., the bottom-right cell in Table 2-1).

TABLE 2-1 A Framework for Identifying Priorities for Population Health Research

NOTE: D.A.R.E. = Drug Abuse Resistance Education; LARC = long-acting reversible contraception.

SOURCE: Lantz presentation, September 30, 2015.

This page intentionally left blank.