Introduction
Methods
Design
Study population
Development of the substitute question
Baseline measurement
Pain intensity
Follow-up
Analysis
Validity
Correlations and hypotheses
Responsiveness
AUC method
Hypothesis testing
Predictive power
Basic model
Sensitivity analysis
Results
Patient characteristics
Population | Total cohort (n = 356) | Cohort “Follow-up” (n = 250) | Not included in the predictive study (n = 106) |
p value |
---|---|---|---|---|
Gender (male) (%) | 166 (47%) | 116 (46%) | 50 (47%) | 0.894 |
Age Mean (SD) | 49.5 (13.1) | 50.2 (13.0) | 47.8 (13.1) | 0.118 |
SPADI score (0–100) Mean (SD) | 46.7 (21.3) | 47.5 (21.2) | 45.0 (21.7) | 0.310 |
Substitute question (0–10) Median (IQR) | 4 (2–6) | 4 (2–6) | 3.5 (1–6) | 0.549 |
Duration of complaints in weeks Median (IQR) | 12 (6–26) | 12 (6–26) | 12 (6–24) | 0.502 |
Use of medication (%) | 171 (49%) | 129 (52%) | 42 (40%) | 0.055 |
Pain intensity (NRS) (0–10) Median (IQR) | 6 (4–7) | 6 (4–7) | 5 (4–7) | 0.068 |
Validity
Convergent validity
Divergent validity
Known groups validity
Group | Median score substitute question |
p value |
---|---|---|
Pain (n = 356) | ||
High initial pain > 7 | 6 (4–7) | 0.000 |
Low initial pain < 7 | 3 (1–5) | |
Work absence (n = 318) | ||
Work absence due to shoulder pain | 6 (5–7) | 0.000 |
No work absence due to shoulder pain | 3 (1–5.25) |
Responsiveness
Prediction model
Predictors for recovery | Model 1 (n = 250) | Model 2 (n = 250) | Model 3 (n = 250) |
---|---|---|---|
OR (95% CI) | OR (95% CI) | OR (95% CI) | |
Age (younger) | 0.98 (0.96–1.00) | 0.98 (0.96–1.01) | 0.98 (0.96–1.00) |
Duration of complaints (in weeks) (shorter) | 0.99 (0.99–1.00) | 0.99 (0.99–1.00) | 0.99 (0.99–1.00) |
Pain using an NRS (lower levels of pain) | 0.92 (0.80–1.05) | 1.02 (0.87–1.21) | 0.97 (0.83–1.13) |
Disability using the total SPADI score (lower level of functional disability) | 0.98 (0.97–1.00) | ||
Disability using the substitute question (lower level of functional disability) | 0.92 (0.81–1.04) | ||
Performance of the model | |||
Correct overall percentage | 64.8% | 65.6% | 65.2% |
Nagelkerke R
2
| 0.090 | 0.114 | 0.098 |
AUC (95% CI) | 0.64 (0.57–0.72) | 0.66 (0.59–0.73) | 0.65 (0.58–0.72) |
Hosmer and Lemeshow | 0.757 | 0.875 | 0.553 |
−2 Log likelihood | 319.286 | 314.534 | 317.594 |
Sensitivity analysis
Predictors for recovery | Model 4 (n = 241) | Model 5 (n = 241) | Model 6 (n = 241) |
---|---|---|---|
OR (95% CI) | OR (95% CI) p value | OR (95% CI) p value | |
Having a job | 1.77 (0.87–3.62) | 1.80 (0.88–3.68) | 1.75 (0.85–3.57) |
Being depressed (not being depressed helps to recover) | 0.41 (0.20–0.85) | 0.42 (0.21–0.88) | 0.43 (0.21–0.89) |
Age (younger) | 0.99 (0.97–1.02) | 0.99 (0.96–1.02) | 0.99 (0.97–1.02) |
Duration of complaints (in weeks) (shorter) | 0.99 (0.99–1.00) | 0.99 (0.99–1.00) | 0.99 (0.99–1.00) |
Pain using an NRS (lower levels of pain) | 0.95 (0.83–1.09) | 1.06 (0.89–1.27) | 0.98 (0.83–1.14) |
Lower disability (SPADI total score) | 0.98 (0.97–1.00) | ||
Lower disability (substitute question) | 0.96 (0.84–1.09) | ||
Performance of the model | |||
Correct overall percentage | 63.9% | 66.0% | 66.8% |
Nagelkerke R
2
| 0.127 | 0.149 | 0.130 |
AUC (95% CI) | 0.67 (0.60–0.74) | 0.69 (0.62–0.75) | 0.68 (0.61–0.74) |
Hosmer and Lemeshow | 0.310 | 0.853 | 0.051 |
−2 Log likelihood | 301.001 | 296.753 | 300.547 |