Session Information
Session Type: ACR Poster Session A
Session Time: 9:00AM-11:00AM
Background/Purpose: The improved Health Assessment Questionnaire (iHAQ) is a novel version of the HAQ (ACR-recommended metric of physical function for RA) based on Item Response Theory. However, there is limited evidence examining its performance in OA. Our purpose was to evaluate the cross-sectional and longitudinal construct validity of the iHAQ in knee OA.
Methods: In a randomized trial comparing Tai Chi with physical therapy among adults with symptomatic knee OA (ACR criteria), we pooled the similar treatment effect that was found and performed a cross-sectional and longitudinal analysis. Participants completed the iHAQ, Patient Global Assessment, Short Form-36, and WOMAC at baseline and after 12 weeks. We examined two iHAQ scoring methods: method A used all 20 items, method B used 16 (Range: 0-100; 100=more disability). Applying the Consensus-based Standards for the Selection of Health Measurement Instruments (COSMIN) standard to assess cross-sectional construct validity, we tested 6 a priori hypotheses with the expected Spearman’s correlations between iHAQ scores and legacy scores at 12 weeks. After a comparison of instruments, we formulated each hypothesis with the direction, strength, and rationale for expected correlation. Longitudinal construct validity was investigated similarly but used correlations in score changes from baseline to 12 weeks. Cross-sectional or longitudinal construct validity was based on number of confirmed hypotheses: High, 5-6 of 6 (>80%); Moderate, 3-4 of 6 (80-35%); or Poor, 0-2 of 6 (<35%).
Results: In 159 participants (71% female, 57% white, mean age 61 years), the 12-week median (Interquartile Range) iHAQ scores were respectively 12.5 (3.1, 25) and 9.4 (1.6, 23) for A and B methods. The change scores from baseline were -4.7 (-15.6, 3.1) for A; and -3.1 (-12.5, 3.1) for B. All associations were in the anticipated direction (Table 1). Affirming high cross-sectional construct validity, all 6 respective hypotheses (100%) were confirmed for both A and B methods (Table 2). Indicating high longitudinal construct validity, 5 of 6 respective hypotheses (83%) were confirmed (Table 3).
Conclusion: iHAQ is able to measure its intended construct and detect changes over time among those with knee OA, which supports its consideration as a novel, recommended outcome instrument in OA.
|
Table 1. Spearman’s Correlations between iHAQ and Legacy Measures |
||
|
Legacy Measure |
iHAQ Method A†, rho |
iHAQ Method B†, rho |
| SF-Physical Function* (Score range: 0-100) |
-0.6604 |
-0.6729
|
| Change SF-Physical Function | -0.4757 | -0.4868 |
| SF-Mental Health
(Score range: 0-100) |
-0.3812 | -0.3795 |
| Change SF-Mental Health | -0.3042 | -03278 |
| WOMAC function
(Score range: 0-1700) |
0.5404 | 0.5487 |
| Change WOMAC function | 0.4972 | 0.5212 |
| SF-Role Physical (Score range: 0-100mm) | -0.4629 | -0.4830 |
| Change SF-Role Physical | -0.3734 | -0.3864 |
| Patient Global (Score range: 0-10cm) | 0.4946 | 0.5363 |
| Change Patient Global | 0.3055 | 0.3299 |
|
*Cross-sectional scores were taken at 12-week follow-up. †Change iHAQ scores were correlated with change legacy measures. Change scores reflected change from baseline to 12 weeks. Note: Consensus guidelines from COSMIN do not recommend reporting p values of correlation coefficients for the purposes of instrument validation. |
||
|
Table 2. Hypotheses for Cross-Sectional Construct Validity of the Improved HAQ |
|||||
|
Hypotheses |
Rationale |
Correlation Result(s) |
Confirmed? |
||
| 1. There will be a strong* (≥ 0.5) negative correlation between iHAQ and Short Form (SF)-Physical Function after intervention |
iHAQ and SF-Physical Function attempt to measure similar constructs |
Expectೠ0.50 |
Method A: -0.66 Method B: -0.67 |
Yes |
|
|
2. The positive correlation between iHAQ with SF-Physical Function will be at least 0.1 higher than the correlation between iHAQ with SF-Mental Health |
iHAQ and SF-Physical Function attempt to measure similar constructs, but iHAQ and SFMH attempt to measure unrelated constructs | Expect ≥ 0.10 |
Method A: 0.28 Method B: 0.29 |
Yes | |
|
3. There will be at least a mod-strong (≥ 0.4) positive correlation between iHAQ and WOMAC Function after intervention. |
iHAQ and WOMAC Function attempt to measure similar constructs, but the WOMAC function is lower limb-specific | Expect≥ 0.40 |
Method A: 0.54 Method B: 0.55 |
Yes | |
|
4. There will be at least a moderate (≥ 0.3) positive correlation between iHAQ and WOMAC Pain after intervention.
|
iHAQ and WOMAC Pain attempt to measure loosely-related constructs | Expect≥ 0.30 |
Method A: 0.49 Method B: 0.51 |
Yes | |
|
5. There will be at least a moderate (≥ 0.3) negative correlation between iHAQ and SF-Role Physical after intervention. |
iHAQ and SF-Role Physical attempt to measure loosely-related constructs | Expect≥ 0.30 |
Method A: -0.46 Method B: -0.48 |
Yes | |
|
6. There will be at least a weak-moderate (≥ 0.2) positive correlation between iHAQ and Patient Global Assessment after intervention. |
iHAQ and Patient Global Assessment attempt to measure partially related constructs, | Expect≥ 0.20 |
Method A: 0.49 Method B: 0.54 |
Yes | |
|
Cross-Sectional Construct Validity Rating: High† |
Hypotheses Confirmed: 6 of 6; 100% |
||||
| *Correlation strength: r ≥ 0.5, Strong; 0.5 > r ≥ 0.4, Moderate-Strong; 0.4 > r ≥ 0.3, Moderate; 0.3 > r ≥ 0.2, Weak-Moderate; 0.2 > r ≥ 0.1, Weak; and 0.1 > r, Negligible. †Overall construct validity was assigned based on percent of total hypotheses confirmed: High, (5-6 of 6 (≥75%); Moderate, 3-4 of 6 (50%≤ x <75%); or Poor, 0-2 of 6 (<50%). .‡For clarity, ”Expect” values are expressed as absolute values.
|
|||||
|
Table 3. Hypotheses for Longitudinal Construct Validity of the Improved HAQ |
|||||
|
Hypotheses |
Rationale |
Correlation Result(s) |
Confirmed? |
||
| 1. There will be a strong* (≥ 0.5) negative correlation between change of iHAQ and change of Short Form (SF)-Physical Function |
iHAQ and SF-Physical Function attempt to measure similar constructs |
Expect.‡ ≥ 0.50 |
Method A: -0.48 Method B: -0.49
|
No |
|
|
2. The positive correlation of change on the iHAQ with that of the SF-Physical Function will be at least 0.1 higher than the correlation of change on the iHAQ with the SF-Mental Health |
iHAQ and SF-Physical Function attempt to measure similar constructs, but iHAQ and SF-Mental Health attempt to measure unrelated constructs | Expect ≥ 0.10 |
Method A: 0.17 Method B: 0.16 |
Yes | |
|
3. There will be at least a moderate-strong (≥ 0.4) positive correlation between change of iHAQ and change of WOMAC Function |
iHAQ and WOMAC-Function attempt to measure similar constructs, but the WOMAC function is lower limb-specific | Expect≥ 0.40 |
Method A: 0.497 Method B: 0.52 |
Yes | |
|
4. There will be at least a moderate (≥ 0.3) positive correlation between change of iHAQ and change of WOMAC Pain |
iHAQ and WOMAC Pain attempt to measure loosely-related constructs | Expect≥ 0.30 |
Method A: 0.43 Method B: 0.44 |
Yes | |
|
5. There will be at least a moderate (≥ 0.3) negative correlation between change of iHAQ and of SF Role Physical |
iHAQ and SF-Role Physical attempt to measure loosely-related constructs | Expect≥ 0.30 |
Method A: -0.37 Method B: -0.39 |
Yes | |
|
6. There will be at least a weak-moderate (≥ 0.2) positive correlation between change of iHAQ and change of Patient Global Assessment |
iHAQ and Patient Global Assessment attempt to measure partially related constructs | Expect≥ 0.20 |
Method A: 0.31 Method B: 0.33 |
Yes | |
|
Longitudinal Construct Validity Rating: High† |
Hypotheses Confirmed: 6 of 6; 100% |
||||
| *Correlation strength: r ≥ 0.5, Strong; 0.5 > r ≥ 0.4, Moderate-Strong; 0.4 > r ≥ 0.3, Moderate; 0.3 > r ≥ 0.2, Weak-Moderate; 0.2 > r ≥ 0.1, Weak; and 0.1 > r, Negligible. †Overall construct validity was assigned based on percent of total hypotheses confirmed: High, (5-6 of 6 (≥75%); Moderate, 3-4 of 6 (50 %≤ x <75%); or Poor, 0-2 of 6 (<50%). .‡For clarity, ”Expect” values are expressed as absolute values.
|
|||||
To cite this abstract in AMA style:
Chung M, Lee AC, Wong JB, Han X, Wang C. Cross-Sectional and Longitudinal Construct Validity for the Improved Health Assessment Questionnaire Among Adults with Knee Osteoarthritis [abstract]. Arthritis Rheumatol. 2016; 68 (suppl 10). https://acrabstracts.org/abstract/cross-sectional-and-longitudinal-construct-validity-for-the-improved-health-assessment-questionnaire-among-adults-with-knee-osteoarthritis/. Accessed .« Back to 2016 ACR/ARHP Annual Meeting
ACR Meeting Abstracts - https://acrabstracts.org/abstract/cross-sectional-and-longitudinal-construct-validity-for-the-improved-health-assessment-questionnaire-among-adults-with-knee-osteoarthritis/