The Social Determinants of Prescriptions

People have more prescription medications as they get older. This is not a new insight, but rather an intuitive one that has been proven with data. An interesting question is whether there are other attributes - social determinants of health - that influence prescriptions. The simple answer is yes, and with the appending of key attributes to anonymized claims data we can understand which attributes are impacting population health.

Using claims data specifically focused on prescription utilization as a marker of health and healthcare engagement, we looked at the impact of six attributes to explain differences in behavior. These attributes are age, gender, race/ethnicity, income, education, and geography.

The correlation between age and prescription utilization is easily identified as seen in the chart above. Also, the role of the patient’s biologic gender shows a well pronounced difference between males and females. The separation in behavior begins around age 17 and maintains its widest gap between the ages of 22 and 32. Even as the gap narrows slightly after 33, the clear separation continues.

A lot has been written about the impact of geography on health. At a micro level, the existence of food deserts has been discussed for over a decade, and they have been linked to other social determinates of health such as race, ethnicity, and income. At a macro level, the data suggest there are factors beyond access to healthy food options which are influencing prescribing habits.

In the US, differences at the state level indicate multiple factors influencing the role of geography as seen in the map above. These factors range from cultural in terms of diet (The Southern diet) to socioeconomic issues related to the above-mentioned food deserts.

Three significant socioeconomic factors that influence health and are present within various levels of geography are race/ethnicity, income, and education. Ethnicity data, when accounting for population size, provides evidence of greater health needs and disease prevalence which can easily be observed as Hispanic and Black patients have a significantly greater use of prescription therapies across all ages.

To explore the impact of education and income using ethnicity as a comparison, an NRx/Patient ratio is used in order to normalize the data. With this metric, clear differences in healthcare demand can be observed between segments. As education and income are highly correlated variables, income is almost always a double-click on education.

Education and income demonstrate more uniform and consistent behavioral characteristics than ethnicity. Similar patterns were observed in prior MedFuse analysis of telehealth utilization. This is due to the ability of members within an ethnic or race cohort to transcend education and economic factors. The fact that some ethnic/racial groups display characteristics similar to socioeconomic segments is due to their population distribution more than ethnicity/race alone.

Emphasizing the impact of socioeconomic factors does not dismiss the role of race and ethnicity in disease prevalence (racial and ethnic genetics impact numerous diseases such as diabetes and heart disease), nor does it discount ethnic and racial cultural factors. As previously mentioned, a person’s geographic location can impact their health, in part due to cultural elements such as diet and feelings toward healthcare. What the totality of this data highlights is the importance of understanding the drivers of health and health behaviors. Beyond understanding, marketers and other stakeholders who want to communicate and connect with patients must segment and adapt both the message and their delivery if they want to be most effective in impacting population health.