Development and internal validation of a baseline prediction model for ≥10-point Oswestry Disability Index improvement by episode end among adults receiving outpatient physical therapy for low back pain: A retrospective cohort study.

Authors

Yazeed Temraz, Ali Sahari, Faisal Alotaibi, Shouq Almutairi, Ahmed Hamami, Amal Alsaad, Rahaf Bin Salamah, Fay Almehrej

Published in

Medicine. Volume 105. Issue 44. Pages e51283. Oct 30, 2026.

Abstract

Low back pain outcomes after outpatient physical therapy are heterogeneous. This retrospective single-center cohort prediction-model study aimed to develop and internally validate a parsimonious baseline model using Oswestry Disability Index (ODI) and STarT Back risk category to predict ≥10-point ODI improvement by routine episode end and assess whether additional baseline variables improve prediction. The hospital data team supplied 400 adult low back pain episodes with baseline and final ODI available. After eligibility exclusions, 391 episodes with complete ODI and STarT Back data comprised the primary cohort; the extended model used 356 complete cases. ≥10-point ODI improvement occurred from baseline to the final recorded physical therapy visit; episode duration was variable (median 31 days [interquartile range 25-63], range 10-95). Model Development and Internal Validation: The primary model included baseline ODI and STarT Back risk category. Internal validation used 2000 bootstrap resamples with optimism correction; nested bootstrap resampling quantified uncertainty around corrected performance. Overall, 236/391 (60.4%) achieved ≥10-point ODI improvement. Baseline ODI was associated with higher odds of improvement (aOR 1.046; 95% confidence interval [CI], 1.027-1.066), whereas high versus low STarT Back risk was associated with lower odds (aOR 0.291; 95% CI, 0.157-0.539). Apparent AUC was 0.666 (95% CI, 0.612-0.719) and Brier score 0.216. Optimism-corrected AUC was 0.657 (95% CI, 0.607-0.722), Brier score 0.220 (0.200-0.233), calibration slope 0.959 (0.925-0.990), and intercept 0.018 (-0.011 to 0.041). The calibration curve showed localized miscalibration. Baseline ODI and STarT Back risk provided modest internally validated prediction of episode-end ODI improvement. The baseline-ODI association may reflect mathematical coupling, regression to the mean, and differential opportunity to achieve the threshold. The model does not estimate causal treatment benefit and requires external validation and local calibration. This prediction-model study was not prospectively registered.

PMID:
43012830
Bibliographic data and abstract were imported from PubMed on 11 Nov 2026.

Read full publication at:
Please sign in to see all details.

Stats

1-terrible, 9-excellent. How would you rate this publication? Sign in in to submit your rating.

Post a comment

You need to be signed in to post comments. You can sign in here.

Comments

There are no comments yet.