The International Society for Pharmacoeconomics and Outcomes Research (ISPOR) recently held its annual conference in Boston, MA. The conference highlights the global accomplishments in health economics and outcomes research (HEOR) and ensures excellence and rigor in HEOR to guide evidence generation for healthcare decision making.
I had a profound learning experience by attending this year's conference. In these sessions I gained technical skills that could be applied to the many healthcare evidence generation activities supported by RTI, and I share a few below.
Causal models strengthen evidence generation
Some sessions introduced the audience to more rigorous methodologies for evidence generation, such as methods of applying causal models to assess the impact of an exposure on an outcome, particularly when conducting observational studies. Causal inference modeling helps to provide evidence on cause-effect relationships, which is more concrete and accurate in lieu of associative relationships. By applying a causal inference model, such as marginal structural modeling, we can provide stronger evidence to our clients around how exposure to treatment or intervention X causes outcome Y.
Real-world data can advance health equity
The topic of health equity was addressed at this year's conference by highlighting ways to incorporate equity into HEOR methodologies. With clinical trials still considered the “gold standard" of evidence generation in HEOR research, the lack of representation in clinical trials is a real public health problem that needs to be addressed if we are to ever reach a more equitable future in healthcare.
A session on “Representation vs Representativeness" discussed how real-world data (RWD) can support underrepresentation in clinical trials. Evidence generated from observational studies, such as exploring impacts of treatments and/or interventions and outcomes among historically underserved populations can help to inform the design of an appropriate randomized clinical trial (RCT).
Additionally, RWD can complement RCT design by informing what geographical locations should be targeted when determining clinical trial site selection or where large clusters of underrepresentation in clinical trials exist. This synergy can assist in successful interventions in these areas. It can also close gaps to increase participation and attrition of historically underserved populations in RCTs.
Countering biases in predictive models
One session of particular interest focused on addressing biases in predictive models and algorithms. Many healthcare organizations in the private sector rely on risk segmented algorithms and risk stratified predictive analytics to identify populations who are at the highest risk of adverse health outcomes. The objective is to be more proactive and to prevent unwanted and costly outcomes among those individuals. The root cause of inherent biases when building predictive models and algorithms is usually underrepresentation of historically underserved populations in data and lacking comprehensive social needs data among served populations.
Inherent biases can also be a result of the methods and measures used to build and validate the predictive model or algorithm. These inherent biases can lead to predictive algorithms that misclassify patients as having low risk for an adverse event when they are truly high risk. Making clinical decisions based on models with inherent biases can lead to a widening of health disparities where there are gaps among those who are being identified as needing additional resources.
Improvements to cost-effective methodologies
Lastly, this year's conference also highlighted the need of considering more advanced modeling methodologies to address health equity when conducting cost-effective analysis (CEA) for costly digital health solutions. A method defined as distributional cost-effective analysis (DCEA) can be instrumental in CEAs by stratifying and quantifying the health benefits and burdens of a solution by socioeconomic status, age, geography, race, or other relevant social variables. This approach helps to provide evidence around the equity benefit of a health solution as well as its efficiency impact.
My summary just barely scratched the surface of the many learnings associated with ISPOR 2023. Reflecting on my own personal experience, I walked away with multiple methods that can be incorporated into my daily work. These newly identified skills and approaches can be applied to enhance the rigor of the methods we employ to generate evidence around health interventions. Additionally, the need for health equity incorporation into HEOR is more profound now than ever and the expertise and tools housed at RTI, like RTI Rarity™, can provide a great benefit to the discipline of HEOR. Contact us to learn more.