Wednesday, November 18, 2009

Interpreting Public Health “Facts”

As I discussed in my last blog, interpretation of medical facts is often crucial for individuals who need to decide on their own approach to care, such as in the decision whether to obtain screening mammography at age 40 or 50.  Equally important, especially with the current public debate about health care reform, is the ability to interpret public health “facts”.  A recent article in the Journal of Public Health entitled “Analysis of 23 million US hospitalizations: uninsured children have higher all-cause in-hospital mortality” illustrates how the straightforward often is not as straightforward as it seems. 

Causation and Association

The article starts out by stating in the introduction that the goal of the study is to “characterize the impact of insurance status on inpatient mortality and costs of care”.  It then, in the discussion at the end states, “we estimate that 16,787 deaths might have been prevented over this time period assuming lack of insurance was the driving factor”.  It actually gives no data to support the premise that lack of insurance was the driving factor.  The study does show that lack of insurance is associated with an increased death rate but it is important to remember that just because two things are associated does not mean that one causes the other.

For example, is it possible that people without insurance are of a lower socio-economic strata and may have more social problems?  Are there more single parents in that group?  Are there more children whose parents, due to social issues, are battling drug abuse and alcoholism and then the children are not receiving adequate care on that basis?  There are many factors that are associated with lack of insurance that suggest that the increased mortality is just as likely related to those other factors as it is to the lack of insurance.  One thing is clear and that is that this study does not tell us anything about the “impact of insurance status on inpatient mortality” and instead only tells us of the association between lack of insurance and inpatient mortality.  No causation is shown. 

Other Factors Not Addressed

In every public health study, it is often the other factors that are not addressed that are most important.  In this study, for example, the largest number of deaths above what you saw in children with insurance, occurred in the  newborn period.  The lesson I would learn from this is that there is an association between poor prenatal care and death rate for newborns as any woman who wants prenatal care can get it via public health clinics as well as urban health centers.  Women who receive prenatal care at those centers also tend to obtain Medicaid coverage or other coverage thus the newborns would also be expected to have some coverage at the time of birth.  That factor was not mentioned in the study.

Be Careful of Hidden Agendas and Biases

There is no such thing as a study totally without bias and studies can be useful even if bias is present.  However to really use a study, be sure to look at what was studied and what was stated on the basis of what was studied.  Often the topics studied have little to do with the conclusions that are drawn.  In this political environment, that is especially true.  So read the studies and be careful of the headlines.