What are Population Health Determinants or Factors?
Health outcomes, however defined and measured, are produced by determinants or factors. They often are sorted into the five categories presented on the right in the following model – health care, individual behavior, social environment, physical environment, and genetics.
Health care determinants generally include access, cost, quantity, and quality of health care services. Individual behavior determinants include choices about lifestyle or habits (either spontaneously or through response to incentives) such as diet, exercise, and substance abuse.Social environment determinants include elements of the social environment such as education, income, occupation, class, social support. Physical environment determinants include elements of the natural and built environment such as air and water quality, lead exposure, and the design of neighborhoods. Genetic determinants include the genetic composition of individuals or populations.
The subcomponents of these determinants or factors can be measured in many different ways. The County Health Rankings includes many such measures in each category that are available at the county level. A series of articles commissioned by the MATCH project, to be published in the online journal Preventing Chronic Disease starting in June 2010, outline current thinking regarding conceptualizing and measuring each of these categories.
In the model above, each category is depicted as the same size, implying that they each contribute equally to health outcomes. Although useful for illustration, in reality those determinants will carry different weights (and hence would be different sizes). Differences exist depending on the population studied, and because cross-sectoral analysis is complicated by interactions between determinants and the latency over time of their effects. In the MATCH County Health Rankings, health care is weighted 20%, behaviors 30%, the social environment 40%, and the physical environment 10%. An explanation of the process used to assign these particular weights is available. However, determining the correct weights for each category and the policies and programs underpinning them remains a major challenge for population health research.It is important, too, to realize the presence of “reverse causality,” which is why there is a small arrow in the above model going from outcomes to determinants/factors. This reflects the fact that outcomes such as morbidity can produce a change in a determinant or risk factor. For example, childhood illness can be responsible for lower educational attainment. In this case, the definitions of outcomes and determinants are reversed; morbidity would be the determinant or factor and educational attainment the outcome. Separating out the different directions of causality is an important and difficult research challenge.
Is Lower Health Care Spending Good for Our Health?
I believe the most important scholarly publication in January, 2012 was Health Affairs’ “Growth in U.S. Health Spending Remained Slow in 2010; Health Share of Domestic Product Was Unchanged From 2009.” This annual report from the Office of the Actuary and the National Health Expenditure Accounts Team of the federal Centers for Medicare & Medicaid Services (CMS) always gets substantial coverage in the mainstream press and this year was no exception.
Why are health expenditure trends so important to improving population health? Because the amount we spend on health care in the United States is the elephant in the room regarding aligning resources appropriately to make us healthier and reduce disparities.Some believe that our ever-increasing health care spending is a sign of market success, to be celebrated like Apple iPad sales. But the fact is we spend far more than any other country and still have poorer outcomes, and many experts believe a quarter to a third of what we spend is ineffective or wasted. In our resource-limited world, increases in health expenditures prevent investment in other health promoting areas like education.
The article reported that health care spending reached $2.6 trillion in 2010, or $8,402 per person. Due to low annual increases of 3.8% in 2009 and 3.9% in 2010 (lower than any period in the last fifty years), the health care share of total spending stabilized at 17.9 percent of GDP. Authors attributed this to slower growth in use of hospitals, physicians, and drugs; from losses of private health insurance coverage; lower median household income; and uncertainty about the financial future. Importantly, the federal government’s share of total expenditures increased to 29% (up from 23% in 2007) while state and local government’s share fell to 16% (from 18% in 2007). Employer contributions fell to 21% from 25% in 2001, while consumer out-of-pocket spending increased by 1.8%. While in previous years’ growth in use of services (as opposed to population growth or price increases) has been a major factor in the overall increase, this year it contributed only 0.1% of the 3.9% increase. Governmental public health spending increased by 8.2% to $82.5 billion, but this still accounts for only 3.2% of overall spending.
So is this good or bad news? Most news coverage highlighted the welcome second year of expenditure slowing, but noted that it might be transitory as the recession ends. Any slowing is in the right direction, but 3.9% is still greater than the 1.5% increase in the Consumer Price Index in 2010, and most of the increase was from price increases — not from greater demand. While slower growth could lead to better insurance coverage over the long haul, it likely had no impact in a single year. Consumers with economic challenges are using less health care, but this is not a good thing if due to delaying or avoiding necessary acute and preventive care. There is certainly no way to tell if lower spending rates translated into more investment in public health and other health promoting education and social service categories; any such result would have to await a longer term trend of getting health care spending close to the general inflation rate for all goods and services.
So one year during an economic slowdown does not a population improvement trend make. But we need to know how we are doing, both nationally and also at local levels where such data is often not available. Hats off to CMS and Health Affairs for continuing to produce and promote this important annual report card.
Are We Individually Responsible for Our Health Behavior Choices?
The answer is yes…and well, not really. This answer is often framed yes or no: either we have free-will, “just say no” responsibility for our behaviors, or we have no responsibility because of the life situations we find ourselves in. Of course the truth is somewhere in between – but what is the balance between them? The answer has important policy implications, particularly with regard to informing investment priorities.
I’ve often noted (as in last week’s Child Health post) that currently available evidence is insufficient to effectively guide public and private policy makers’ investment decisions with regard to population health improvement. That said, we must recognize the cutting edge research being done in this area. For several decades, Paula Lantz and Jim House and colleagues at the University of Michigan have used a longitudinal database called the Americans’ Changing Lives Survey (ACLS) to explore answers to important questions using the most sophisticated social science methods available. The ACLS has tracked 3600 adults from 1986 to 2005, querying them 4 times over that period.
In their most recent paper (published in the May 2010 issue of Social Science and Medicine) they developed models to explore relationships among various factors (such as age, gender, ethnicity, education, income, smoking status, physical activity level, and many others) and risk of death. They were particularly interested in connections between income and behavior and wanted to determine if poor people have worse health because they have poor health behaviors – or if something else is going on.
Not surprisingly, the findings showed that people with lower incomes tended to have more unhealthy behaviors and people with higher incomes tended to have healthier behaviors. Regardless of income, smoking and low levels of physical activity were both associated with an increased risk of death. But more importantly for this blog, the findings revealed that unhealthy behaviors are far from the only reason low income populations are at increased risk of death. Independent of health-related behaviors and other control factors, the risk of death among those in the lowest income category was 76% greater than those in the highest income category (which was similar to the risk differential between smokers and nonsmokers). In addition, the independent effect of the lowest level of physical activity was nearly triple that of income and smoking.
The authors conclude with a call for health policy and clinical incentives to enhance income security, promote smoking cessation/prevention, and support physical activity. While these results do not tell us exactly what our population health investment strategy should be across behavioral and non-behavioral factors, they make clear that a balance is required. To improve population health, policy and programmatic resources are needed to help make the healthy choice the easy choice – not only with respect to health behaviors but for healthcare, employment, and education as well.
The concept of behavioral “choice” is often interpreted to mean individual choices independent of external (and often constraining) factors. Careful studies like this one provide evidence that keeps us from operating from such a simplistic perspective. To improve population health, we all need to make better health choices – but we also need to understand and remedy the economic and educational upstream factors which determine the extent to which this is possible.
Have You Heard of “Primordial Prevention”?
I never had, until I started thinking about a post on prevention and went to John Last’s Dictionary of Epidemiology to get started. His basic definition is “actions aimed at eradicating, eliminating, or minimizing the impact of disease or disability,” and includes primary, secondary and tertiary prevention. Last defines primary prevention as “protection of health by personal and communal efforts such as enhancing nutritional status, immunizing against communicable diseases, and eliminating environmental risks such as contaminated water supplies.”
I’ve given a lot of thought to the issue of prevention this spring, especially since writing about the National Prevention Council’s Strategy Framework. In that post, I applauded the Framework’s call for Healthy Environments (such as affordable housing, employment opportunities, efficient transportation, good schools, and effective policing) but cautioned against interest group sidelining of these “Cross-cutting Priorities” in favor of the seven behavior-related “Priority Areas” which are more closely related to the primary prevention definition shown above.I was pleased with the Council’s May 24 Webinar discussion, which indicated a new recommendation in the “Empowered People Strategic Direction” to include “improving income, education, and employment opportunities.” I believe critical upstream factors should be central to all models of health, from the Prevention Framework to the Triple Aim. In the CDC’s Community Guide to Preventive Services, the Social Environment is one of 18 topics with systematic reviews. Yet only three reviews currently exist in this critical area: early childhood, culturally competent healthcare, and housing. The brevity of the list speaks volumes to what we consider important and where we have invested our research dollars.
So I was delighted to read that John Last’s definition of prevention includes a relatively recent classification called “primordial prevention”:
Primordial prevention…aspires to establish and maintain conditions to minimize hazards to health…it consists of actions and measures that inhibit the emergence and establishment of environmental, economic, social and behavioral conditions, cultural patterns of living known to increase the risk of disease.
This is, of course, the basic message of Link and Phelan’s seminal article on social conditions as fundamental causes of disease. I am a strong proponent of precision with terminology, because understanding informs and drives action. I sometimes worry that the more classic definition of prevention is too rooted in the lifestyle modification efforts of the past 40 years so that equal attention is not given to the upstream social determinants — or that it leads to taking the comfortable position that since improving income and education is so difficult we leave it to others (such as letting the Treasury and Federal Reserve worry about unemployment for us).
I hope no one doubts that I strongly support the work of the National Prevention Council and consider its strategy work fundamental to improving population health. But I also hope that a limited understanding of its name doesn’t get in the way. The Council cannot fulfill its cross-sectoral purpose unless specific goals and objectives for educational quality, for job creation and economic development, and for controlling health care costs are given equal or higher priority as those for immunizations, food deserts, smoking, and physical activity.
The time has come for “primordial prevention” to take center stage, along with public and private policy and resource allocation to reflect improved understanding of the many factors that drive health.
The Link Between Income and Health
Listening to the national conference call for the release of the 2012 County Heath Rankings last week, I was struck by the number of questions regarding the socioeconomic factors in the model. It made me recall the early morning call from a small town reporter when we released the first Wisconsin County Rankings in 2003, saying “do you mean that county income levels might be as important as the number of uninsured or the smoking rates?”
This concept is better understood today than it was in 2003, but it is still hard to communicate. We accept the health-compromising effects of smoking and lack of health care access, but how education and income ‘get under the skin” to produce disease and death is less obvious. Yet a growing body of literature supports this finding, and is the reason why we give social factors a weighting of 40% in the County Health Rankings model.
This week we’re calling attention to a study I co-authored with one of our PhD candidates, Erika Cheng. The article, Disparities in premature mortality between high and low income U.S. counties, appears in the April 2012 issue of Preventing Chronic Disease.
The figure below shows the relationship between median household income and age-adjusted, all-cause mortality rates for all counties in the country.
As expected, people tend to live longer in high income counties (HIC), with the diagonal line indicating the overall relationship. However, this relationship is not as strong in HIC; an increase of $9,000 in median household income was associated with an 18% better average mortality rate among low income counties (LIC) but only 12% better in higher income counties.
Equally interesting is the very large variation in mortality rates among LIC. Some of the LIC mortality rates are comparable to those of HICs (in the 200/100,000 range), while other LIC rates are triple or quadruple this level.
Also, a more nuanced picture emerges when controlling for other factors that impact mortality. Several factors were associated with longer lifespans, including percentages of adults with a 4-year college degree and percent Hispanic in the county. Other factors were associated with shorter lifespans, including preventable hospital stays, percent black in the county, percent children living below federal poverty guidelines, and percent adult smokers.
Importantly, when these other factors were controlled for, statistically significant linkages were found between median household income and mortality in the HICs, but not the LICs. This result highlights the main point of this study – that there is an interplay, apparent at the county level, between patterns of health factors and income. For example, two variables — the percent of children living below the federal poverty guidelines and the percent of single-parent households — are more closely linked to mortality in LICs than in HICs. We suggest that perhaps county-level income may buffer the effects of some variables associated with poor health. This buffer may operate through coping resources; county-level income may reflect overall availability of material and social resources that enable affluent single parents’ access to child care, neighborhood support, social networks, or higher-quality health care. Of course, we caution that the nature of the study and the data only allow for associations, not causal relationships.
From a policy perspective, these findings remind us that “one size fits all” does not apply to improving population health in communities. Our data reveal complicated relationships among the many factors that determine how long (not to mention how well) we live. While finding feasible and effective solutions can be difficult, we believe there’s a lot that can be done — including helping communities move on down the road toward better health and continuing to think about and work toward operationalizing the concept of locally customized policy approaches.
Will the Jobs Bill Impact Population Health?
The last week has seen much attention devoted to the issue of jobs and unemployment in our country. Presidential elections often hinge on the state of the economy and the persistence of high unemployment rates will likely play a greater role in 2012 than in previous periods. While a growing literature shows that displaced workers (defined as individuals who lose their jobs as part of plant closings, mass layoffs, and other firm-level employment reductions) tend to experience significant long-term earnings losses as well as decreased job stability, lower employment rates, earlier retirement, lower personal spending, and decreased health insurance coverage.
In addition to these primarily economic effects, there are likely to be health effects as well. Unemployment (measured as the percent of the population age 16 and over that is unemployed but seeking work) is one of seven key measures in the County Health Rankings’ Social and Economic Factors, but is only briefly mentioned in Healthy People 2020’s new section on the Social Determinants of Health.
An extensive literature demonstrates the association of unemployment with an increased likelihood of morbidity and mortality. In a 2009 review, Bambra summarized:
There are clear relationships between unemployment and increased risk of poor mental health and parasuicide, higher rates of all cause and specific causes of mortality, self-reported health and limiting long-term illness and, in some studies, a higher prevalence of risky health behaviours, including problematic alcohol use and smoking. The negative health experiences of unemployment are not limited to the unemployed only but also extend to families and the wider community…..links between unemployment and poorer health have conventionally been explained through two inter-related concepts: the material consequences of unemployment (e.g., wage loss and resulting changes in access to essential goods and services) and the psychosocial effects of unemployment (e.g., stigma, isolation and loss of self-worth).
In the year after displacement, Sullivan and von Wachter (2009) found mortality rates among long-tenured employees were 50%–100% higher than would otherwise have been expected. This effect on mortality hazards declined sharply over time, but persisted. Even twenty years after displacement, the authors estimated a 10%–15% increase in annual death hazards. If such increases were sustained indefinitely, they would imply a loss in life expectancy of 1.0–1.5 years for a worker displaced at age forty. Similarly, Bartley and Ferrie (2010) found unemployment elevated the risk of premature death by 57% among men ages 44-54.
But what comes first – unemployment or poor health? I do not usually focus on issues of research methodology in this blog, but the unemployment and health relationship is a good place to discuss reverse causality, since it features prominently in this literature and is of critical policy importance. In the unemployment relationship, not only are the unemployed likely to be less healthy, but it is also intuitive to expect that less healthy individuals are more likely to be unemployed because they are unhealthy and, therefore, less productive employees.
But sorting out these conflicting causal pathways is difficult. Lundin and colleagues (2010) estimated that 49% of the association between unemployment and poor health was due to poor health resulting in unemployment. Of course, context plays a major role. In Germany, where citizens have access to generous unemployment benefits, long entitlement durations, and universal health insurance, health status seems to drive employment status rather than vice-versa (Schmitz, 2011).
While researchers and policymakers need to be aware of such concerns, the U.S. evidence such as that of Sullivan cited above which control for reverse causality certainly suggest that we should consider job loss and unemployment a significant determinant of population health. While impact on income will likely remain the dominant policy concern, we would do well to keep in mind the impact on health and related worker productivity as the costs and benefits of employment policy are debated in the coming months and years.
Peter Orszag Has it Half Right
The former Obama Office of Management and Budget Director Peter Orszag made a compelling case in A Health Plan for Colleges (New York Times September 19) that when Medicaid costs increase (as they do every year), states respond by making cuts in higher education expenditures. Orszag estimates that if higher education’s share of state budgets over the past 25 years had remained constant rather than being crowded out by rising Medicaid costs, allocations would be $30 billion more (about $2000 per student) than they are currently. His strong and novel argument for controlling health care costs in order to preserve more funds for higher education goes beyond the more typical goals of increasing health care affordability and improving international competitiveness.
But he should have taken the argument a step further. Better educated populations are healthier populations. A large body of evidence supports this claim. Compared to adults having some education beyond high school, premature death rates (e.g., deaths before age 75) are double among those having only a high school education and triple among those not completing high school. Research shows that those with more education also have fewer disabilities and better physical functioning.
What role could education play in biological processes that produce death or disability? There are probably two main pathways. The first operates directly, through better knowledge about the importance of health care as well as healthy behaviors and prevention. The field of health literacy has been rapidly developing. According to the Institute of Medicine, approximately 90 million American adults lack the literacy skills needed to effectively use the US health care system. Low health literacy rates have been estimated to cost the health care system more than $100 billion per year. Also, education likely enhances the ability to make difficult short term decisions (such as stopping smoking, eating better, routinely exercising) which affect health later in life.
A second pathway is more indirect. People with more education tend to have better employment options, including access to jobs that pay more and provide better healthcare coverage. Jobs that require more education also tend to be less stressful, in part because they afford employees a greater sense of autonomy and control than typical “blue collar” jobs. A growing field of research examines how stress influences disease and mortality through endocrine and immune system response.
So to complete the argument Orszag begins: lower health care costs improve education, which improves health, which lowers health care costs. The result is an even more compelling case for health care cost containment.