The Use of Real-World Evidence in Health Technology Assessment Decisions in Sweden and the UK: A Natural Language Processing (NLP) Approach

OBJECTIVES: The objective of this study was to apply NLP techniques to analyse the utilization of Real-World Evidence (RWE) in HTA decisions in Sweden (TLV) and the UK (NICE).

METHODS: A comprehensive analysis was conducted using data from national HTA assessment bodies, IQVIA HTA Accelerator, and PubMed publications, employing NLP techniques such as sentiment analysis and topic modelling to extract insights and identify related clusters of terms and concepts.

RESULTS: The PubMed search yielded a total of 3,213 results of relevant publications, and analysis of the IQVIA HTA Accelerator data disclosed the inclusion of RWE in 666 HTA submissions. The PubMed review revealed that cohort studies and design from patient registries were the most accepted source of RWE and highlighted a growing integration of RWE in HTA processes. However, prevailing negative sentiments surrounding RWE utilization indicated significant acceptance concerns.

Since 2018, there has been a significant increase in the inclusion of RWE in HTA appraisals, with NICE accounting for 459 instances and TLV for 171 instances, highlighting the growing recognition of its value in informing HTA decision-making. TLV has successfully accepted RWE for supporting effectiveness (42.86%) and treatment patterns (28.57%), while NICE primarily utilized RWE for effectiveness (27.17%). The sentiment analysis reveals that NICE shows a favourable acceptance of RWE with a higher count of positive sentiments (326) compared to negative sentiments (289), while TLV exhibits a more cautious approach with a higher count of negative sentiments (149) compared to positive sentiments (103).

CONCLUSIONS: The growing integration of RWE in HTA, coupled with divergent acceptance patterns between TLV and NICE, highlights the importance of comprehensively addressing challenges and leveraging RWE as a valuable tool in evidence synthesis for HTA decision-making.