235: Identifying Clinically Meaningful Chronic Pain Using Routine Primary Care Records: C-PICTURE

Topic: Clinical Science
Subtopic: Non-specific to any diagnosis, Assessment and Measurement of Pain/Nociception, Epidemiology, Digital Health Technologies

Background and Aims: Chronic pain is common, yet there is currently no reliable way to identify people with chronic pain in the general population, for example through electronic health records (EHRs). This limits population surveillance and service planning [1, 2]. Analgesic prescribing is often used as a surrogate marker but is likely to underestimate prevalence of chronic pain [3]. Existing approaches either show limited sensitivity/specificity or have not been validated [2, 4]. This study evaluated an algorithm (designed using a combination of diagnostic and prescribing codes) for use in primary care. We aimed to identify people with clinically meaningful, high-impact chronic pain.

Methods: The algorithm was co-designed by healthcare professionals, people with chronic pain, and health data scientists. A mixed-methods evaluation was conducted using Scottish primary care data from six diverse practices. Algorithm classifications were compared with: (1) medical records, including free-text review; (2) patient survey responses using validated measures of pain severity, pain interference, and expressed need; and (3) qualitative insights from patients and clinicians. Diagnostic accuracy, agreement, and receiver operating characteristic analyses were performed. Qualitative interviews explored coding practices, pain recognition, healthcare use, and lived experience. Additional analyses assessed alignment with high pain burden and unmet need.

Results: The algorithm demonstrated high sensitivity (~92%) and negative predictive value (~99%) compared with medical records (n=1,200), with few false negatives. Chronic pain was documented in 31% of medical records, compared with 57% self-reported chronic pain in the survey (n=889). Agreement with survey definitions was modest (κappa≈0.25–0.30), reflecting differences between documented and self-reported pain. However, the algorithm identified 75-77% of people reporting high pain interference and expressed need, with a greater number of algorithm indicators among those with higher pain burden (p<0.001). Qualitative findings explained discrepancies through inconsistent coding and pain normalisation, supporting identification of persistent, high-impact pain.

Conclusions: The algorithm performs well as a high-sensitivity tool for identifying clinically meaningful chronic pain in primary care. Apparent over-identification largely reflects documentation practices and differences between lived pain experience and coded records rather than algorithm error. Importantly, the algorithm preferentially captures individuals whose pain interferes with daily life and is associated with unmet need. These findings support the use of routine data algorithms for chronic pain research, surveillance and service planning, while highlighting areas for future refinement. This approach could be adapted in other healthcare systems with electronic health records to improve chronic pain identification and monitoring.

References: 1. Stubhaug A, Hansen JL, Hallberg S, Gustavsson A, Eggen AE, Nielsen CS. The costs of chronic pain—Long‐term estimates. European Journal of Pain. 2024
2. Lacasse A, Cauvier Charest E, Dault R, Cloutier A-M, Choiniere M, Blais L, et al. Validity of algorithms for identification of individuals suffering from chronic noncancer pain in administrative databases: a systematic review. Pain Medicine. 2020;21(9):1825-39.
3. Torrance, N., Mansoor, R., Wang, H., Gilbert, S., Macfarlane, G. J., Serpell, M., Baldacchino, A., Hales, T. G., Donnan, P., Wyper, G., Smith, B. H., & Colvin, L. (2018). Association of opioid prescribing practices with chronic pain and benzodiazepine co-prescription: a primary care data linkage study. British Journal of Anaesthesia, 120(6), 1345–1355.
4. Hafezparast N, Bragan Turner E, Dunbar-Rees R, Vusirikala A, Vodden A, de La Morinière V, et al. Identifying populations with chronic pain in primary care: developing an algorithm and logic rules applied to coded primary care diagnostic and medication data. BMC Primary Care. 2023;24(1):184.

Presenting Author: Ian-Ju Liang
Submitting Author: IAN-JU LIANG

Senior Author: Lesley Colvin

IASP 2026 Global Year theme of Neuropathic Pain: Yes

• Cassie Higgins - University of Dundee
• Nouf Abutheraa - University of Dundee
• Paul Cameron - NHS Fife
• Frank Sullivan - University of St Andrews
• Gary Macfarlane - University of Aberdeen
• Kathryn Martin - University of Aberdeen
• Suzanne Grant - University of Dundee
• Peter Donnan - University of Dundee
• Hussein Patwa - Patient Partner
• Cherry Holland - Patient Partner
• Neil Robson - Patient Partner
• Sue Cole - Patient Partner
• Blair Smith - University of Dundee