Digging deeper into the HRSA data for Cambria and Somerset Counties I thought I would look at predictors of mortality in the area. The graph above shows a strong positive correlation between preventable hospital stays and all cause mortality. This means that zip-codes with higher rates of preventable hospital stays have higher mortality rates.
This relationship accounts for 74.6% of the variability. If the relationship were perfect accounting for 100% of the variability it would look like the graph below with the zip-code labels taken out.
Preventable Hospital Stay Definition
The CDC defines preventable hospital stays as “admissions to a hospital for certain acute illnesses (e.g., dehydration) or worsening chronic conditions (e.g., diabetes) that might not have required hospitalization had these conditions been managed successfully by primary care providers in outpatient settings.” These admissions are illnesses that can be nipped in the bud. They are costly for hospitals and can lead to fatal chronic illnesses.
The zip-codes 15469 (Normalville, PA) and 15464 (Mill Run, Pa) had the highest mortality rates at 950 deaths/100,000. They were followed closely by the zip-code 15459 (Markleysburg, PA). These zip-codes are all found in Fayette County near the Somerset County line. However, they did not have the highest rates of preventable hospital stays.
Forty zip-codes tied for the highest preventable hospital stay rate at 3,361 hospital stays per 100,000 enrolled in Medicare. These zip codes include all of the Johnstown zip codes which are all of the high need zip-codes. There are racial disparities for preventable hospital stays in Cambria County as seen below.
| Group | 2017 | 2018 | 2019 | 2020 | 2021 |
| Overall | 5,919 | 5,762 | 5,518 | 3,251 | 3,429 |
| White | 5,773 | 5,779 | 5,505 | 3,239 | 3,328 |
| African American | 6,653 | 6,154 | 7,326 | 4,036 | 9,936 |
One should be cautious about concluding a cause and effect relationship exists between preventable hospital stays and mortality. There is the issue of directionality. For example does A cause B or does B cause A? Also, there could be another mediating variable C. Other correlations between mortality and other variables have been found. I will detail them in subsequent posts.