METHODOLOGICAL CHALLENGES IN THE EVALUATION OF PRESCRIPTION SUPPORT SOFTWARE: LESSONS FROM PMCF ALERT VALIDATION
Author(s)
Clara Bouvard, MSc1, Tiphaine Porte, MSc1, Emilie Chaslin, BSc2, Marion Dansette, MSc2, Audrey Lajoinie, PharmD, PhD1.
1RCTs, Lyon, France, 2Orisha Socialecare (Teranga Software), Paris, France.
1RCTs, Lyon, France, 2Orisha Socialecare (Teranga Software), Paris, France.
OBJECTIVES: To illustrate how the current narrative format of Summary of Product Characteristics (SmPC) complicates alert validation and calls for more standardized, interoperable regulatory documents, based on the use of NETSoins DM, a prescription support software (PSS) used in nursing homes.
METHODS: A randomized sample of 30 centers and 100 patients was analyzed. Alerts encompassed precautions for use, pharmacovigilance, interactions, dosing, and incompatibility. Experts reported missing alerts. Compared to SmPCs by two pharmacists, discordant alerts were assessed for patient impact.
RESULTS: Centers were predominantly EHPAD (87%, median bed capacity: 73 (47;81). Patients were female (73%), with a median age of 89 years (80;93). NETSoins DM generated 19 (±12) alerts per prescription. Among 345 alerts, mainly concerned dosing (28%) and pharmacovigilance (24%). Overall, 200 alerts were SmPC-discordant, with discordance rates of 98% for pharmacovigilance, 87% for interactions and 84% for dosing. Discordances reflected SmPC limitations: missing clinical qualifiers, non-operational wording, and fragmented information. Over 80% of pharmacovigilance alerts arose from duration rules absent from SmPCs, particularly for chronic treatments. Nearly 90% of dosing alerts reflected SmPCs blending recommendations, examples, and edge cases without clear structure. Almost 33% of contraindication alerts corresponded to precautions, overestimating severity. Around 20% of interaction alerts overestimated clinical risk due to absent decision thresholds. Same problematics were observed in alerts deemed missing by experts.
CONCLUSIONS: These findings underscore the impact of structural SmPC limitations on PSS alert validation. Observed discrepancies reflect not erroneous alerts, but an interoperability gap between drug documentations underpinning PSS development, which would benefit from machine-readable, semantically structured databases. SmPC usability was assessed retrospectively, beyond the scope of the primary PMCF study, which may not fully capture the true extent of limitations. While clinical expert review remains a reliable reference, combining it with drug documentation analysis strengthens the regulatory robustness of PMCF-derived Clinical Evaluation Report evidence.
METHODS: A randomized sample of 30 centers and 100 patients was analyzed. Alerts encompassed precautions for use, pharmacovigilance, interactions, dosing, and incompatibility. Experts reported missing alerts. Compared to SmPCs by two pharmacists, discordant alerts were assessed for patient impact.
RESULTS: Centers were predominantly EHPAD (87%, median bed capacity: 73 (47;81). Patients were female (73%), with a median age of 89 years (80;93). NETSoins DM generated 19 (±12) alerts per prescription. Among 345 alerts, mainly concerned dosing (28%) and pharmacovigilance (24%). Overall, 200 alerts were SmPC-discordant, with discordance rates of 98% for pharmacovigilance, 87% for interactions and 84% for dosing. Discordances reflected SmPC limitations: missing clinical qualifiers, non-operational wording, and fragmented information. Over 80% of pharmacovigilance alerts arose from duration rules absent from SmPCs, particularly for chronic treatments. Nearly 90% of dosing alerts reflected SmPCs blending recommendations, examples, and edge cases without clear structure. Almost 33% of contraindication alerts corresponded to precautions, overestimating severity. Around 20% of interaction alerts overestimated clinical risk due to absent decision thresholds. Same problematics were observed in alerts deemed missing by experts.
CONCLUSIONS: These findings underscore the impact of structural SmPC limitations on PSS alert validation. Observed discrepancies reflect not erroneous alerts, but an interoperability gap between drug documentations underpinning PSS development, which would benefit from machine-readable, semantically structured databases. SmPC usability was assessed retrospectively, beyond the scope of the primary PMCF study, which may not fully capture the true extent of limitations. While clinical expert review remains a reliable reference, combining it with drug documentation analysis strengthens the regulatory robustness of PMCF-derived Clinical Evaluation Report evidence.
Conference/Value in Health Info
2026-11, ISPOR Europe 2026, Vienna, Austria
Value in Health, Volume 29, Issue 12S
Code
MT11
Topic
Health Service Delivery & Process of Care, Medical Technologies
Topic Subcategory
Digital Health
Disease
Geriatrics