Skip to main content

Table 5 Overall performance of full logistic multivariate regression models (n = 246)

From: Prediction of healthcare utilization following an episode of physical therapy for musculoskeletal pain

Dependent variable

Any care

Opioids

Injection

Surgery

Diagnostic tests or imaging

Emergency room

Complete Case

 Block 1

Demographic, Clinical and Comorbidity

.258**

.274**

.292**

.226*

.250*

.400**

 Block 2

OSPRO-YF (10 items)

OSPRO-ROS (10 items)

.267

.294

.293

.234

.253

.404

 Block 3

OSPRO-YF (+ 7 items)

OSPRO-ROS (+ 13 items)

.275

.296

.315

.259

.271

.457

 Block 4

4-week change

(Pain, Disability, OSPRO-YF)

.337*

.424**

.353

.426**

.340*

.560*

Inverse Probability of Attrition Weighted

 Block 1

Demographic, Clinical and Comorbidity

.306**

.294**

.313**

.236

.304**

.430**

 Block 2

OSPRO-YF (10 items)

OSPRO-ROS (10 items)

.314

.317

.317

.250

.305

.435

 Block 3

OSPRO-YF (+ 7 items)

OSPRO-ROS (+ 13 items)

.321

.321

.334

.284

.321

.473

 Block 4

4-week change

(Pain, Disability, OSPRO-YF)

.382*

.448**

.373

.464**

.389*

.611*

  1. All values are variance explained (pseudo-R2); Health care utilization (dependent) variables refer to utilization during the follow-up period, whereas independent variables refer to measurements taken before the follow-up period
  2. *p < 0.05, **p < 0.01