Every year since 2010, the University of Wisconsin Population Health Institute and the Robert Wood Johnson Foundation have produced the County Health Rankings—a “population health checkup” for the nation’s over 3,000 counties. This is a fascinating resource that measures both health outcomes and health factors.
The study pulls data from a number of sources in an attempt to measure counties’ well being. You can find a description of their complete methodology here. Using a number of measures they ranked counties within each state. Utah’s rankings are shown in figure 1 for both health outcomes (measuring length and quality of life) and health factors (measuring behaviors, clinical care, social and economic factors and the physical environment). The researchers also have a report specific to Utah worth your time.
The study includes some suggestions to improve a community’s health (again, I recommend you visit the website). But as I looked at the data, I wondered what the correlation was between certain health outcomes and health factors. Particularly, I looked at the percent of adults within each county that reported fair or poor health and compared it with a host of health factors including county obesity rates, smoking rates, excessive drinking rates, and percent physically inactive. Additionally, I included other potential health factors not included in the study such as county average age, population, and percent of the population living below the poverty rate.
I used a scatter plot and the correlation coefficient to determine the correlation between each of the above mentioned health factor to the percent of adults within the county that reported fair or poor health. I also compared the correlation coefficient with its R-squared value that measures the confidence one can have in the correlation. My work suggests that not every health factor is created equally.
Before we consider the results of my study, figure 2 shows the data I used.
Now let’s consider in turn the correlation between each factor and the health outcome we’re measuring. We’ll start with the factor with the least correlation and work our way to the highest correlation.
Excessive Drinking Rate (chart 1)
While one would intuitively expect a county with a higher excessive drinking rate to result in a higher level of fair or poor health, the opposite is actually the case in Utah. In fact, the county with the highest percentage of excessive drinking (Summit County) is also the county with the lowest percent of its citizens self-reporting fair or poor health. Chart 1 shows the scatter plot for excessive drinking rate. The correlation suggests a 4 percent increase in health outcome for a 10 percent increase in excessive drinking. Don’t put too much stock in this, though, as the R-squared value of 0.177 suggest very little confidence in this correlation.
Average Age and Population (charts 2 & 3)
I’ve lumped average age and population together as factors as neither seems to impact self reported fair or poor health. Not only that, but their R-squared values suggest even less confidence in their results than that of excessive drinking rates. According to the results, there should be a 0.3 percent increase in fair or poor health for every 10 percent increase of average age (see chart 2) and a 1 percent decrease in fair or poor health for every 10 percent increase in population (chart 3).
Obesity and Physically Inactive Rates (charts 4 & 5)
Obesity and physical activity show more correlation to well being, but perhaps not as much of a correlation as one would think (certainly not as much a correlation as I suspected). For every 10 percent increase in obesity rates among citizens of a county, there is a three percent increase in self-reported fair or poor health citizens. The R-squared value is a shaky 0.095 (chart 4). And for every 10 percent increase in physically inactive citizens, there is a four percent increase in self-reported fair or poor health citizens. The R-squared value of this correlation is 0.176 (chart 5).
Percent below Poverty Level and Smoking Rates (charts 6 & 7)
Finally, we are getting somewhere. Both the measure of percent of population below the poverty level and a county’s smoking rates show a strong correlation to self-reported physical well being. For every 10 percent increase in the percentage of citizens living below the poverty level there is a corresponding nine percent increase in self-reported fair or poor health (chart 6). Smoking rate numbers are equally correlated. For every 10 percent increase in smoking rates in a county, there is a nine percent increase in self-reported fair or poor health (chart 7). And both health factors have a robust R-squared value of 0.765. Interestingly, while both health factors similarly correlate to the measured outcome, they don’t correlate to each other as much as they do to the outcome.
From a quick study of the data, it appears we can readily identify two health factors that greatly impact health outcomes in county smoking rates and percent of the population living below poverty levels. The challenge county officials face is how to craft policy that effectively addresses those two factors.
If you have a subject you’d like UAC to explore in a future Counties by the Numbers article, please email Arie Van De Graaff at email@example.com