Screening real-world data for evidence of unsuspected drug benefits: An application of the sequence symmetry analysis

The aim of this is to test the feasibility of identifying unsuspected, previously unknown drug-outcome associations, that is, collateral drug benefits (CDBs), through a systematic screening analysis of real-world health-care databases. Ultimately, such screening could lead to drug repurposing.

We analysed data from the Danish National Prescription Registry and the Danish Patient Registry, covering 1996-2022. The study employed the sequence symmetry analysis (SSA), an exposure-anchored self-controlled design that compares the number of clinical outcomes in symmetrical windows before and after the exposure drug initiation. To verify the directionality and robustness of these associations, we incorporated the case-crossover (CCO) design, another self-controlled design. The obtained associations were ranked according to the hypothetical number of averted outcomes, if a causal effect could be assumed.

The analysis included 1.3 billion prescriptions and 260 million diagnosis records, resulting in 27 820 976 drug-diagnosis combinations and 7 920 323 drug-drug combinations. Preventive associations in both the SSA and CCO were found in 7795 drug-diagnosis and 5088 drug-drug combinations. A manual review of the highest ranked 100 associations resulted in 11 drug-diagnosis and 2 drug-drug associations as potential unknown CDBs. Notable findings included selective serotonin reuptake inhibitors linked to a lower risk of certain cardiovascular outcomes, anticholinesterases associated with fewer delirium diagnoses and progestogens associated with a reduced risk of obesity.

The study confirms that hypothesis-free screening is feasible and that combining sequence symmetry analysis and case-crossover designs can identify potential collateral drug benefits. Further validation studies are required to confirm these findings and explore their clinical implications.