Social Determinants of Health in the BaseHealth Predictive Model
Dr. Nick van Terheyden, CMO, BaseHealth
More than ever before, the social determinants of health (SDOH) are getting some well-deserved attention when it comes to their role in impacting an individual’s relative health status. Here at BaseHealth, seeing as we’re in the business of predicting health status, we’ve focused significant time and effort on understanding how SDOH can be incorporated into our model. As such, we’ve developed a fairly nuanced view when it comes to the utility of SDOH in predicting health status. Our approach to identifying ‘qualified’ predictors of health status are based on the following criteria:
- Are they proven to be directly causal?
- Are they quantifiable at an individual level?
- Are they intervenable?
For example, we know high blood pressure or hypertension is a qualified predictor of coronary artery disease (CAD) because (1) there have been multiple studies that demonstrate its direct causal link to CAD, (2) it has been quantified at the individual level, and (3) we can intervene to reduce the risk with diet, exercise and a variety of medications. Unfortunately, to date, SDOH, at least as they are defined by the World Health Organization (WHO), struggle to stand up to our criteria for establishing predictors of health status.
Most social determinants are not causal, but they are associative. For example, high income usually gives better access to medical care, clean water, safe housing, and is at times even linked to being a primary caregiver. BaseHealth’s Causality model focuses on the most evidence-based causal risk factors, which means adding social determinants directly to the model would overestimate or in some cases underestimate the accurate health status prediction.
Quantifiable at The Individual level
Though social determinants can be easily observed and reported in populations, it is much harder to assign effects from these determinants to specific individuals. There are not many methods of measuring social determinants at a standardized individual level, since they are all baskets of various underlying characteristics and behaviors within a population. Data about some determinants (such as being a care giver to an incapacitated patient) are actually very hard to collect because they require information about relationships that are hardly ever monitored or even reported.
Interventions for most social determinants can be applied at a political or systemic level, but it is extremely difficult to address the same social determinants at an individual level. While certain risk factors that SDOH are associated with can be addressed at the individual level, like access to clean water, there is no way to comprehensively and methodically intervene on all SDOH at an individual level. There may be some opportunity to apply some analysis of existing data where such SDOH factors were changed albeit inappropriately. For example, the Flint Michigan water crisis dating back to 2014 that exposed 100,000 residents to high levels of lead that has since been reversed may provide some view under the covers of SDOH factors in this population.
Having said this, we believe that there are certain well-defined narrow contexts where we could use some well-proven social determinants to improve our understanding of the mapping between the SDOH, the causality model and the actual causative risk factors within a population. BaseHealth has built an AI empowered associative model which can make the association between the social determinants, casual risk factor, and actual health conditions.
For example, if information is available about whether a person lives close to a major highway or not, we could use that, along with other risk factors we have about that person, to sharpen our associative model with respect to lung problems. With that information, we would likely generate a better estimate of risk factor impact and improve the ranking of members who need relevant interventions. However, the interventions would still be oriented around the known causal risk factors — as opposed to the associated SDOH — since the only intervention opportunity available for an individual who lives near a busy highway is to have the member move, which is hardly practical.
In short, finding an effective role for SDOH in any causal model is challenging, but we do see them being used to refine the cohort predictions in our associative model. As discussed earlier, adding social determinants to the model would likely contribute to clarified liability measurements, instead of generating new member-level interventions. Of course, as we’ve discovered in the course of our work, social determinants of health, like access to a vehicle or being a primary caregiver, do impact the effectiveness of the interventions we recommend. That being the case, we’re investigating how to incorporate that data into our model when calculating intervention effectiveness or deciding on target cohorts.