P119. A prospective, psychometric validation of NIH PROMIS physical function, pain interference and upper extremity CAT in cervical spine patients: successes and key limitations

• Published in: The spine journal 2019-09, Vol.19 (9), p.S213-S213
• Main Authors: Iyer, Sravisht, Koltsov, Jayme, Steinhaus, Michael E., Ross, Thomas, Young, Kelsey, Stein, Dan, Yang, Jingyan, Lafage, Virginie, Albert, Todd J., Kim, Han Jo
• Format: Article
• Language: English
• ISSN: 1529-9430; 1878-1632
• DOI: 10.1016/j.spinee.2019.05.544

Current patient-reported outcomes in cervical pathology have substantial limitations. Patient Reported Outcomes Measurement Information System (PROMIS) Computer Adaptive Tests (CATs) offer the potential for improved psychometric properties with reduced questionnaire burden. To validate PROMIS CATs against existing patient-reported outcomes in patients undergoing surgery of the cervical spine. Prospective cohort study. Adult patients undergoing cervical spine surgery, excluding those undergoing surgery for instability due to trauma. Validation of the instruments was completed with a Rasch Model as well as measurement of coverage, efficiency, test-retest reliability, responsiveness, and convergent validity. The following instruments were tested: SF-36, NDI, VAS arm/neck, and PROMIS Pain Interference (PI), Physical Function (PF) and Upper extremity (UE). Adult patients undergoing cervical spine surgery at a single institution between 2016-2018 were prospectively enrolled. Patients completed questionnaires (SF-36, NDI, VAS arm/neck, and PROMIS PI, PF, and UE) preoperatively and at 6 months postoperatively. Demographic data, diagnosis, and procedural data were recorded. Of 198 eligible patients, 164 had 6-month follow-up data. The most common diagnoses were radiculopathy (37.6%), myeloradiculopathy (30.5%), and myelopathy (28.4%). All studied instruments had acceptable fit to a Rasch model. PROMIS CATs demonstrated improved average time to completion for PI (39s), PF (47s), and UE (54s), compared to NDI (117s) and SF-36 PCS (175s). Responsiveness for PROMIS CATs was similar to NDI and SF-36. Test-retest reliability was lower for PI (ICC:0.68), PF (0.70), and UE (0.59), compared to NDI (0.86) and PCS (0.85). For convergent validity, PI was strongly correlated to NDI and PF to SF-36 PCS. There were no significant floor or ceiling effects for the PROMIS domains, although UE had postoperative clustering (n=18) at a high score (56.4) and PI had postoperative clustering (n=27) at a low score (38.7). PROMIS CATs demonstrate several advantages, including improved efficiency and comparable responsiveness and convergent validity with legacy instruments in cervical spine patients. Nevertheless, CATs had lower test-retest reliability and had significant clustering at higher levels of function for the PI and UE CATs. These limitations must be considered before broad adoption of CATs in cervical spine patients. This abstract does not discuss or include any applicable devices