Background: An estimated 87% of torture survivors experience chronic pain such as brachial plexopathy from upper extremity suspension or lumbosacral plexus injury from leg hyperextension. However, a vast majority of pain is undetected by evaluators due to a lack of diagnostic tools and confounding psychiatric illness. This diagnostic gap results in exclusive psychological treatment rather than multimodal therapies, substantially limiting rehabilitation. We hypothesized that the United Nations Istanbul Protocol (UNIP) would have a sensitivity of approximately 15% for pain detection, and that the use of a validated pain screen would improve its sensitivity by at least 29%, as compared to the reference standard (pain specialist evaluation).
Methods and findings: This prospective blind-comparison-to-gold-standard study of survivors of torture, as defined by the World Medical Association, took place at Weill Cornell Medicine between February 1, 2017, and June 21, 2019. 11 women and 9 men, for a total of 20 participants, were included in the analysis. Five participants received 2 UNIP evaluations, for a total of 25 unique evaluations included in the analysis. Participants were representative of a global population, with home countries in Africa, Central America, South Asia, the Caribbean, and the Middle East. Methods of torture experienced were homogeneous, following the predictable pattern of systematic torture. Participants first received the standard evaluation protocol for torture survivors (UNIP) by a trained evaluator, and subsequently received a validated pain screen (Brief Pain Inventory-Short Form [BPISF]) followed by a noninvasive examination by a pain specialist physician (reference standard). The primary outcome was the diagnostic and treatment capability of the standard protocol (index test) versus the validated pain screen (BPISF), as compared to the reference standard. Trained evaluators performing the initial assessment with the UNIP (index test) were blinded to the study, and the pain specialist physician (reference standard) was blinded to the outcome of the initial UNIP evaluation and the BPISF; data from the initial UNIP assessment were not gathered by the principal investigator until all other study procedures were completed. Providers using only the UNIP captured pain in a maximum of 16% of evaluations, as compared to 85% of participants being diagnosed with pain by the reference standard. When employed, the validated pain screen had a sensitivity of 100% (95% CI 72%-100%) and a negative predictive value of 100%, as compared to a sensitivity of 24% (95% CI 8%-50%) and a negative predictive value of 19% (95% CI 5%-46%) for the index test. The difference in the sensitivity of the UNIP as compared to the BPISF was significant, with p < 0.001. No adverse events owing to participation in the study were reported by participants. Limitations of the study include small sample size, its single-site nature, and the exclusion of individuals who did not speak 1 of the 5 study languages.
Conclusions: These data indicate that a validated pain screen can supplement the current global standard assessment of torture survivors, the UNIP, to increase the accuracy of pain diagnosis.
Trial registration: ClinicalTrials.gov NCT03018782.