Skip to main content

Table 3 Candidate prognostic models and key model statistics

From: Predicting the outcome of conservative treatment with physiotherapy in adults with shoulder pain associated with partial-thickness rotator cuff tears – a prognostic model development study

No

Candidate model

N* factors

Main characteristic

AICC

∆AICC

SEE

R2ADJ§

1

Age + sex + physical demands + disability (WORC) + pain + history of shoulder pain + symptom duration + smoking + pain catastrophizing (PCS)

(+ diabetes removed)

9

Full model (all factors)

891

11

313

0.12

2

Smoking + pain catastrophizing (PCS) (+ diabetes removed)

2

Potential for modification (could be modified (addressed) by some action (e.g. treatment)

880

0

314

0.11

3

Age + sex

2

Factors that cannot be modified

889

9

336

−0.02

4

Age + sex + physical demands + pain + history of shoulder pain + symptom duration + smoking (+ diabetes removed)

7

Type of assessment: “no questionnaires”

899

19

344

−0.06

5

Disability (WORC) + pain catastrophizing (PCS)

2

Type of assessment: “questionnaires”

880

0

314

0.11

6

Smoking (+ diabetes removed)

(1)

Type of factor: “bio(logical) factors”

Excluded from analysis due to removal of diabetes

7

History of shoulder pain + symptom duration

2

Background (patient history)

889

9

336

−0.02

8

Pain + history of shoulder pain + symptom duration

3

Further models: pain-related factors (excluding pain catastrophizing)

889

9

335

−0.01

9

Pain + pain catastrophizing (PCS)

2

Further models: pain and attitude towards pain

882

2

318

0.09

  1. *Denotes the number of factors in each model as analyzed (i.e. after removal of diabetes). An ∆AICC value of 0 denotes the model(s) with the lowest AICC value(s), representing the “best” model(s) within the set of candidate models; Model initially included diabetes, which was excluded from the analyses due to its low prevalence in the sample; §Negative R2 values are generally interpreted as “0”