Abstract
Objective To compare physical function scales of the Multidimensional Health Assessment Questionnaire (MDHAQ) with that of the Health Assessment Questionnaire–Disability Index (HAQ-DI) in patients with psoriatic arthritis (PsA), and to examine whether either questionnaire is less prone to “floor effects.”
Methods Data were collected prospectively from 2018 to 2019 across 3 UK hospitals. All patients completed physical function scales within the MDHAQ and HAQ-DI in a single clinic visit. Agreement was assessed using medians and the Bland-Altman method. Intraclass correlation coefficients (ICCs) were used to assess test-retest reliability.
Results Two hundred ten patients completed the clinic visit; 1 withdrew consent. Thus, 209 were analyzed. Sixty percent were male, with mean age of 51.7 years and median disease duration of 7 years. In clinic, median MDHAQ and HAQ-DI including/excluding aids scores were 0.30, 0.50, and 0.50 respectively. Although the median score for HAQ-DI was higher than for MDHAQ, the difference between the 2 scores was mostly within 1.96 SDs from the mean, suggesting good agreement. The ICCs demonstrated excellent test-retest reliability for both the MDHAQ and HAQ-DI. Similar numbers of patients scored 0 on the MDHAQ and HAQ-DI including/excluding aids (48, 47, and 49, respectively). Using a score of ≤ 0.5 as a cutoff for minor functional impairment, 23 patients had a MDHAQ ≤ 0.5 when their HAQ-DI including aids was > 0.5. Conversely, 4 patients had a MDHAQ > 0.5 when the HAQ-DI including aids was ≤ 0.5.
Conclusion Both the MDHAQ and HAQ-DI appear to be similar in detecting floor effects in patients with PsA.
- activities of daily living
- Health Assessment Questionnaire
- psoriatic arthritis
- quality of life
- self-assessment
Psoriatic arthritis (PsA) is a chronic inflammatory arthritis that can limit daily activities, with detrimental effects on patients’ quality of life.1,2 Assessment of function is important in both clinical practice and in trial settings to enable evaluation of disease activity and treatment effect. The Health Assessment Questionnaire–Disability Index (HAQ-DI) is a well-validated patient self-report questionnaire for assessing physical function in rheumatic diseases.3 However, it is lengthy to complete, and scoring can be complex. Scores can also be artifactually elevated when aids are used, despite improving patient function, and the scoring method may lead to different activities being compared from visit to visit.4 Moreover, some activities included in the HAQ-DI may not be relevant to certain patients, such as “cutting meat” in patients who are vegetarians.
To improve clinical utility and relevance of the HAQ-DI, other versions have been developed, including the Multidimensional Health Assessment Questionnaire (MDHAQ).4,5,6,7 Compared to the HAQ-DI, the MDHAQ has a shorter 10-item physical function scale. However, to allow a holistic assessment, it also includes additional questions on pain, sleep, mood, fatigue, joint symptoms, and patients’ other medical history, all of which are not part of the HAQ-DI. The physical function scale within the MDHAQ has been compared against the HAQ-DI previously in other rheumatic diseases, and more recently, in PsA.4,5,8 Prior research has shown that it may be less susceptible to “floor effects,” whereby patients report normal scores of “0” despite experiencing functional impairment.5,8 Detection of floor effects is becoming more important, as treatment strategies increasingly shift toward achieving minimal disease activity.9 Our study aimed to assess the agreement between the physical function scales of the MDHAQ and HAQ-DI in patients with PsA, and evaluate whether either questionnaire was less susceptible to floor effects.
METHODS
Study design. We conducted a cross-sectional questionnaire study comparing the physical function scales of the MDHAQ to the HAQ-DI in patients aged ≥ 18 years with definite PsA (according to the ClASsification of Psoriatic Arthritis [CASPAR] criteria10 or previous diagnosis by a rheumatologist). Patients were recruited from 3 UK hospital trusts (Oxford University Hospitals, Leeds Teaching Hospitals, and Bradford Teaching Hospitals) from December 20, 2018, to August 22, 2019.
All patients completed the MDHAQ and HAQ-DI in a single visit within usual care. The HAQ-DI consists of 20 activities of daily living grouped into 8 categories, and the difficulty in carrying out each activity is ranked using a semiquantitative 0–3 scale. The MDHAQ physical function scale consists of 8 activities of daily living chosen from the HAQ-DI (1 per category). It also contains 2 additional questions pertaining to more challenging activities aimed at identifying more minor, but still relevant, functional impairment, and is also scored from 0 to 3.3,4 The order that both questionnaires were completed was alternated to exclude the effects of participant fatigue. Both questionnaires were compared with the 12-item PsA Impact of Disease questionnaire (PsAID-12),11 which is scored from 0 to 10. This has a validated patient acceptable symptom state (PsAID-12 score ≤ 4) to stratify high- and low-impact disease. Data were also collected on patient demographics, PsA subtype, disease duration, concurrent fibromyalgia (FM), self-reported disease activity, and current therapy.
Patients were given an identical pack with a prepaid self-addressed envelope and the instruction to complete the questionnaires 1 week later. This ceased when returned questionnaire numbers were sufficient to evaluate test-retest reliability. The 1-week timepoint was chosen as it was assumed that most patients’ disease activity state would not have changed significantly. This was clarified with an additional question on disease activity at 1 week.
Sample size calculation. There is no gold standard for measuring physical function in PsA. Based on statistical advice, we calculated our sample size to detect noninferiority between the physical function scales of the MDHAQ and HAQ-DI. Using data from a prior pilot study of 51 patients with PsA, we calculated that for a powered study to detect noninferiority between the 2 scales with a margin of 0.125 at a 2-sided 0.025 significance level with > 90% power, 210 participants were needed,12 given that 30% of these cases had a HAQ score of 0.
Statistical analysis. Median scores and IQR were calculated for all questionnaires. Agreement between the HAQs were assessed using the Bland-Altman method.13 The t test was used to evaluate whether the order of questionnaire presentation affected the scores. Spearman rank was used to assess the correlation between HAQ and PsAID-12 scores. Intraclass correlation coefficients (ICC; 2-way mixed model absolute agreement) were used to assess test-retest reliability, with ICCs > 0.75 considered to demonstrate concordance.14 The proportion of patients who scored 0 was calculated for the HAQs, to allow assessment of floor effects. All analyses were performed using R (version 3.6.1; R Foundation for Statistical Computing).
Ethical considerations. This study was approved by the London-Surrey Research Ethics Committee (reference 18/LO/2057). All patients gave written informed consent.
RESULTS
Patients and questionnaire scores. Two hundred ten patients completed the initial clinic visit; 1 withdrew consent. Thus, data from 209 patients were analyzed. Sixty-two of the 107 patients given an identical pack to complete at 1 week returned the questionnaires.
Table 1 details baseline characteristics of the cohort and median questionnaire scores, with score distributions in clinic shown in Figure 1. Although the median HAQ-DI score is consistently higher than the MDHAQ, the difference mostly lies within 1.96 SDs of the mean, suggesting good agreement (Figure 1). Patients with PsAID-12 scores ≤ 4 had HAQ scores clustered around the lower half of the 0–3 range, whereas those with PsAID-12 scores > 4 had an even distribution of HAQ scores. Of the 11 patients with concurrent FM, 3 had a PsAID-12 score ≤ 4, 6 had a PsAID-12 score > 4, and 2 had incomplete PsAID-12 scores.
Clinical characteristics and questionnaire scores in clinic and at home 1 week later.
Comparison of the HAQs. (A–C) Histograms of HAQ scores. (D–G) Bland-Altman plots. (H,I) Scatter plots by impact (high-impact: PsAID-12 score > 4; low-impact: PsAID-12 score ≤ 4). MDHAQ: Multidimensional Health Assessment Questionnaire; HAQDI: Health Assessment Questionnaire–Disability Index; PsAID-12: 12-item Psoriatic Arthritis Impact of Disease questionnaire.
Using Spearman rank, we found statistically significant correlations between clinic PsAID-12 scores and clinic MDHAQ, HAQ-DI including aids, and HAQ-DI excluding aids scores (Spearman ρ 0.75, 0.72, 0.71, respectively, all P < 0.001).
The order in which HAQ questionnaires were completed did not affect the score (MDHAQ first vs second, P = 0.72; HAQ-DI including/excluding aids first vs second, P = 0.86 and P = 0.92, respectively).
Floor effects. When considering individual question scores (Table 2), questions i and j of the MDHAQ (walk 3 km, and participate in sport, respectively) were least susceptible to floor effects, with the lowest proportion of patients who scored 0. They also had an increased proportion of patients with higher scores, as with questions 5b (take a bath), 6a (reach and get down 5-lb object from overhead), and 8c (household chores) of the HAQ-DI.
Summary of scores for individual questions within the MDHAQ and HAQ-DI in clinic.
However, when considering total scores, similar numbers of patients scored 0 in MDHAQ and HAQ-DI including/excluding aids (48, 47, and 49 respectively). Moreover, similar numbers of patients scored 0 in the HAQs when their PsAID-12 score was > 4 (2, 3, and 3, respectively). There was also no clear difference between the numbers of patients scoring 0 in the HAQs when analyzing by patient-reported remission (remission: 15, 13, 15 vs nonremission: 31, 32, 32, respectively) and presence/absence of FM (yes: 1, 0, 0 vs no: 47, 37, 39, respectively).
Using ≤ 0.5 as a cutoff for minor functional impairment, 23 patients had a MDHAQ ≤ 0.5 when their HAQ-DI including aids was > 0.5. This reduced to 17 when HAQ-DI excluding aids was > 0.5. Conversely, 4 patients had an MDHAQ > 0.5 when the HAQ-DI including aids was ≤ 0.5. This increased to 5 when HAQ-DI excluding aids was ≤ 0.5. For patients with PsAID-12 scores > 4, there were 24 patients who had a MDHAQ ≤ 0.5, compared to 20 and 16 with HAQ-DI excluding/including aids ≤ 0.5, respectively.
Test-retest reliability. Clinic and home questionnaire scores were similar, although consistently slightly numerically higher, at home (Table 1). The ICCs (95% CI) for MDHAQ, HAQ-DI including/excluding aids, and PsAID-12 were 0.97 (0.95–0.98), 0.98 (0.97–0.99), 0.98 (0.96–0.99), and 0.96 (0.93–0.97), suggesting excellent test-retest reliability.
DISCUSSION
Our study found good agreement between the physical function scales of the HAQs in patients with PsA, with both demonstrating excellent test-retest reliability. This corroborates results from previous studies.4,5
Previous studies have suggested that the physical function scale of the MDHAQ may be less susceptible to floor effects compared to the HAQ-DI. In a US study of 144 patients with rheumatic diseases, 23 scored 0 on the HAQ-DI, whereas 14 scored 0 on the MDHAQ.5 In another US study of 140 female patients with rheumatoid arthritis (RA) or systemic lupus erythematosus (SLE),15 scores of 0 were seen in > 49% of patients with RA and > 63% of patients with SLE for items shared between the HAQ-DI and MDHAQ. In contrast, for MDHAQ-specific physical function items, scores of 0 were seen in < 29% of patients with RA and < 41% of patients with SLE. More recently, in a study of 274 patients with PsA, 26.6% scored 0 on the HAQ-DI compared to 17.9% on the MDHAQ.8
Our results showed that when considering individual questions, the MDHAQ-specific questions did seem to be less susceptible to floor effects, with 38.3% patients reporting normal scores of 0 compared to 64.6% for questions shared with the HAQ-DI. However, when considering total scores, a similar number of patients had normal scores of 0 in both questionnaires including the subgroups with high-impact disease (PsAID-12 score > 4), FM, and patient-reported remission, suggesting overall similar performance in detecting floor effects.
In our cohort, HAQ-DI scores were consistently higher than MDHAQ scores, even with aids excluded. A greater proportion of patients had MDHAQ scores ≤ 0.5, suggesting minor functional impairment when their HAQ-DI scores were > 0.5, than the other way around. For those with high-impact disease, more patients also had MDHAQ scores ≤ 0.5 compared to the HAQ-DI. Although the 2 questionnaires are not directly interchangeable, there is a heavy overlap of questions, with both aimed at assessing physical function. Patients with PsA are generally younger with a higher baseline function compared to other rheumatic diseases, and they may also have more variable manifestations of functional impairment due to disease heterogeneity.16,17 Therefore, the more detailed HAQ-DI questionnaire with its increased breadth of questions may help to detect slightly more minor functional impairment compared to the MDHAQ. However, it is important to balance the value of potential information gleaned from a lengthier questionnaire against the time and burden to patients from filling it out, especially when a more succinct questionnaire is available that provides similar information.
For low-impact disease (PsAID-12 score ≤ 4), the HAQ scores clustered around the lower half of the 0–3 range, whereas for high-impact disease the distribution is more even. This suggests that low functional impairment generally occurs with low-impact disease. However, high-impact disease does not necessarily equate to high functional impairment, as other factors including skin symptoms and psychological impact contribute.
Strengths of our study include recruitment of patients from 3 separate centers and comparing the HAQs in an unselected group of patients with PsA within routine clinical practice. Limitations include the different administration settings of the questionnaires to assess test-retest reliability, and the 1-week interval meant some patients felt their disease activity state had changed. We were unable to analyze ceiling effects in our dataset, as only 1 patient had a HAQ-DI score of 3. Moreover, we did not include the psychological and clinical components of the HAQs within our study, which would be important to assess in the future.
In conclusion, we show that although the MDHAQ-specific physical function questions are less susceptible to floor effects individually, when considering total scores, both HAQ questionnaires perform similarly in detecting floor effects in patients with PsA. The breadth of questions within the HAQ-DI may allow it to detect slightly more minor functional impairment compared to the MDHAQ, but this needs to be carefully balanced against the increased patient burden.
Footnotes
WY is a National Institute for Health Research (NIHR) Academic Clinical Fellow. LCC is an NIHR Clinician Scientist and Senior Clinical Research Fellow funded by a NIHR Clinician Scientist award. The research was supported by the NIHR Oxford Biomedical Research Centre. The views expressed are those of the author(s) and not necessarily those of the NHS, the NIHR, or the Department of Health. We acknowledge the support of the NIHR Clinical Research Network (NIHR CRN).
The authors declare no conflicts of interest relevant to this article.
- Accepted for publication April 26, 2021.
- Copyright © 2021 by the Journal of Rheumatology
