Jared C. Tishelman, Dennis Vasquez-Montes, David S. Jevotovsky, Nicholas Stekas, Michael J. Moses, Raj J. Karia, Thomas Errico, Aaron J. Buckland and Themistocles S. Protopsaltis
The Patient-Reported Outcomes Measurement Information System (PROMIS) has become increasingly popular due to computer adaptive testing methodology. This study aims to validate the association between PROMIS and legacy outcome metrics and compare PROMIS to legacy metrics in terms of ceiling and floor effects and questionnaire burden.
A retrospective review of an outcomes database was performed at a single institution from December 2016 to April 2017. Inclusion criteria were age > 18 years and a chief complaint of back pain or neck pain. The PROMIS computer adaptive testing Pain Interference, Physical Function (PF), and Pain Intensity domains; Oswestry Disability Index (ODI); Neck Disability Index (NDI); and visual analog scale (VAS) back, VAS leg, VAS neck, and VAS arm were completed in random order. PROMIS was compared to legacy metrics in terms of the average number of questions needed to complete each questionnaire and the score distributions in the lower and higher bounds of scores.
A total of 494 patients with back pain and 130 patients with neck pain were included. For back pain, ODI showed a strong correlation with PROMIS-PF (R = −0.749, p < 0.001), Pain Intensity (R = 0.709, p < 0.001), and Pain Interference (R = 0.790, p < 0.001) domains. Additionally, the PROMIS Pain Intensity domain correlated to both VAS back and neck pain (R = 0.642, p < 0.001 for both). PROMIS-PF took significantly fewer questions to complete compared to the ODI (4.123 vs 9.906, p < 0.001). When assessing for instrument sensitivity, neither survey presented a significant ceiling and floor effect in the back pain population (ODI: 0.40% and 2.63%; PROMIS-PF: 0.60% and 1.41%). In the neck pain cohort, NDI showed a strong correlation with PROMIS-PF (R = 0.771, p < 0.001). Additionally, PROMIS Pain Intensity correlated to VAS neck (R = 0.642, p < 0.001). The mean number of questions required to complete the questionnaire was much lower for PROMIS-PF compared to NDI (4.417 vs 10, p < 0.001). There were no significant differences found in terms of ceiling and floor effects for neck complaints (NDI: 2.3% and 6.92%; PROMIS-PF: 0.00% and 5.38%) or back complaints (ODI: 0.40% and 2.63%; PROMIS-PF: 1.41% and 0.60%).
PROMIS correlates strongly with traditional disability measures in patients with back pain and neck pain. For both back and neck pain, the PROMIS-PF required patients to answer significantly fewer questions to achieve similar granularity. There were no significant differences in ceiling and floor effects for NDI or ODI when compared with the PROMIS-PF instrument.
Ian McCarthy, Michael O'Brien, Christopher Ames, Chessie Robinson, Thomas Errico, David W. Polly Jr. and Richard Hostin
Incremental cost-effectiveness analysis is critical to the efficient allocation of health care resources; however, the incremental cost-effectiveness ratio (ICER) of surgical versus nonsurgical treatment for adult spinal deformity (ASD) has eluded the literature, due in part to inherent empirical difficulties when comparing surgical and nonsurgical patients. Using observed preoperative health-related quality of life (HRQOL) for patients who later underwent surgery, this study builds a statistical model to predict hypothetical quality-adjusted life years (QALYs) without surgical treatment. The analysis compares predicted QALYs to observed postoperative QALYs and forms the resulting ICER.
This was a single-center (Baylor Scoliosis Center) retrospective analysis of consecutive patients undergoing primary surgery for ASD. Total costs (expressed in 2010 dollars) incurred by the hospital for each episode of surgical care were collected from administrative data and QALYs were calculated from the 6-dimensional Short-Form Health Survey, each discounted at 3.5% per year. Regression analysis was used to predict hypothetical QALYs without surgery based on preoperative longitudinal data for 124 crossover surgical patients with similar diagnoses, baseline HRQOL, age, and sex compared with the surgical cohort. Results were projected through 10-year follow-up, and the cost-effectiveness acceptability curve (CEAC) was estimated using nonparametric bootstrap methods.
Three-year follow-up was available for 120 (66%) of 181 eligible patients, who were predominantly female (89%) with average age of 50. With discounting, total costs averaged $125,407, including readmissions, with average QALYs of 1.93 at 3-year follow-up. Average QALYs without surgery were predicted to be 1.6 after 3 years. At 3- and 5-year follow-up, the ICER was $375,000 and $198,000, respectively. Projecting through 10-year follow-up, the ICER was $80,000. The 10-year CEAC revealed a 40% probability that the ICER was $80,000 or less, a 90% probability that the ICER was $90,000 or less, and a 100% probability that the ICER was less than $100,000.
Based on the WHO's suggested upper threshold for cost-effectiveness (3 times per capita GDP, or $140,000 in 2010 dollars), the analysis reveals that surgical treatment for ASD is cost-effective after a 10-year period based on predicted deterioration in HRQOL without surgery. The ICER well exceeds the WHO threshold at earlier follow-up intervals, highlighting the importance of the durability of surgical treatment in assessing the value of surgical intervention. Due to the study's methodology, the results are dependent on the predicted deterioration in HRQOL without surgery. As such, the results may not extend to patients whose HRQOL would remain steady without surgery. Future research should therefore pursue a direct comparison of QALYs for surgical and nonsurgical patients to better understand the cost-effectiveness of surgery for the average ASD patient.