Friday, August 26, 2011

Actual causes of death in the U.S.: not what you think

Any standard public health or medical school prevention text includes (or ought to include) some version of the figure below, which illustrates that the leading causes of death in the U.S. at the turn of the century (heart disease, cancer, stroke) were actually surrogates for what have come to be known as the actual causes of death: unhealthy behaviors such as tobacco use, poor diet, and physical inactivity.

The most effective preventive services that primary care clinicians provide, then, are not screening tests but counseling interventions that aim to change one or more of these behaviors for the better. Community-level initiatives such as tobacco-free restaurants and campuses, pedestrian-friendly cities, and increasing access to nutritious food sources play a critical role in changing health-related behaviors, too.

Unfortunately, the impact of behavioral or "lifestyle" approaches to prevention is likely to be limited by two factors: 1) even intensive interventions produce very modest benefits; and 2) behaviors don't exist in a vacuum, but are largely shaped by economic and social circumstances. Family medicine professor and former U.S. Preventive Services Task Force member Steven Woolf has published a number of studies showing that the risk of death is strongly associated with levels of college education and income; his research team at Virginia Commonwealth University worked with the Robert Wood Johnson Foundation to develop an interactive County Health Calculator that illustrates how many premature deaths could be avoided by eliminating educational and income disparities.

Researchers from Columbia University went a step further by publishing "Estimated Deaths Attributable to Social Factors in the United States" in this month's issue of the American Journal of Public Health. Using estimates derived from the literature on social determinants of health and year 2000 mortality data, they found that the "actual" causes of death looked like this:

1) Low education: 245,000
2) Racial segregation: 176,000
3) Low social support: 162,000
4) Individual-level poverty: 133,000
5) Income inequality: 119,000
6) Area-level poverty: 39,000

Clearly, we know a great deal more about successful strategies for fighting clinical and behavioral causes of death than we do about social causes, some of which often appear intractable. But I could not agree more with the authors' conclusion that "these findings argue for a broader public health conceptualization of the causes of mortality and an expansive policy approach that considers how social factors can be addressed to improve the health of populations." The point being: poverty, discrimination, and low education aren't just social or political issues best left to non-clinicians - they're health issues, too.


  1. The graphic you've posted stunned me. The sheer gap of the poor diet to alcohol is really big. Makes you think the only thing we can really do is to change our lifestyle. AND Influence others to do so. But that's easier said than done. :(

  2. Having worked in the Canadian public health system, the social determinants of health are an important component of their health policy efforts. The literature is consistent in showing strong associations between factors such as income and education and health outcomes. However, what remains truly a mystery is how these "upstream" factors exert their influence on health outcomes (research has not been able to elucidate the causality path). Until we know much more into this area, we cannot ascertain causality with any confidence. More importantly, there exists little research into the efficiency of social interventions into health outcomes: that is, it is not clear how cost-effective social interventions are on health outcomes and given difficult economic times, it is difficult to justify some of the proposed actions. On the other hand, causality has been established with a high degree of certainty between the lifestyle factors and the corresponding health outcomes (hence the graph that is shown). What needs to happen for lifestyle interventions to be effective is then to take into account the environment (social, physical and economic) when helping patients. In addition, few providers are properly trained in lifestyle interventions (ie. motivational interviewing, exercise prescriptions, nutrition counseling and smoking cessation counseling for example) as these are not topics usually taught well in professional schools. Moreover, reimbursement is not adequate for these types of interventions hence decreasing their effectiveness (the 5 minute consult and a prescription for a pill is reimbursed more adequately). Until these issues are addressed, lifestyle interventions will not be effective in improving health outcomes. Moreover, more research is needed to address how the famous social determinants of health exert their influence on health outcomes.

  3. Some would argue that in the last couple of years obesity may have passed tobacco use as the top bar on your comparative graph. Either way you are right on track with the "actual" causes of death. Keep up the good work.

  4. While I agree with your conclusion that the "lifestyle" approach to prevention is a vital and under-appreciated aspect of health care, I think that calling behavioral factors the "actual" causes of death is misleading. It seems to promote the idea that dying is the fault of the deceased by confusing an archaic meaning of the word "actual" (i.e., having to do with action) with the modern meaning of "actual" (i.e., real or genuine). You're absolutely right to highlight the importance of changing one's lifestyle for the better (and to propose that clinicians should play a larger role in helping people with this). I would just point out that Alzheimer's disease, and instances of heart disease not related to lifestyle, are nowhere to be found among the "actual" causes of death.