A huge THANK YOU to my supervisor @andganna.bsky.social and all other coauthors who have been fundamental to get all this work done!

Take-home message 2:

The robust associations with some of the socio-demographic factors considered suggest that socio-economically disadvantaged groups would benefit from targeted interventions to improve the dispensing and uptake of pharmacological treatments. (13/n)

Take-home message 1:

Our results suggest a very limited role of genetics, including established pharmacogenes, on persistence and adherence. (12/n)

Moreover, we tested whether the GWAS results from FinnGen could be used to build a polygenic score for statin adherence and found this was significantly associated with observed adherence. (11/n)

We finally replicated the results for statins in Estonian Biobank, which have similar drug purchase and genetic data, and observed consistent effects for health and socio-economic factors present in both studies. (10/n)

We also wondered how ploygenic scores for the traits above would predict persistence and adherence.

Effect sizes were overall smaller than those for the health and socio-demographic factors at the beginning, and could explain a lower fraction of variance in adherence. (9/n)

Interestingly, adherence to statins and BP medications were positively correlated with participating in one or more of the optional questionnaires of UK Biobank. (8/n)

We ran a GWAS of persistence and adherence to each medication. We didn’t find significant associations with either phenotype but could use the results to explore the genetic overlap between persistence, adherence and other health and behavioral traits. (7/n)

None of the pharmacogenes we tested showed an effect on adherence. (6/n)

We then turned to genetics, the first suspect being relevant and well established pharmacogenes. CYP2C9 intermediate met. (higher risk of fluvastatins side effects) associated with lower odds of persistence. Similar for tamoxifen and CYP2D6 poor met. (lower drug response). (5/n)

We first considered the effect 8 health and sociodemographic factors.

Non-Finnish/Swedish mother tongue (proxy for recent immigration) and receiving social assistance benefits (lower socio-economic status) showed the biggest effects on both phenotypes, across all meds. (4/n)

We looked at these phenotypes across 5 relevant med classes, including up to N=1 814 591.

We explored health and socio-demographic factors in the nation-wide Finregistry data, and genetics for the subset of ppl from FinnGen.

We replicated our results in Estonian Biobank. (3/n)

Based on data from all prescription drug purchases made in Finland from 1998 to 2020, we defined two drug taking behaviors: continuing the treatment for at least one year vs early discontinuing it (persistence), and following the prescribed therapeutic regimen (adherence) (2/n)