Introduction

Besides death, discomfort, drug side effects, and financial costs, disability has been identified as one of the primary dimensions in the assessment of outcome in rheumatoid arthritis (RA). Several instruments have been developed to measure functional disability, ranging from physical assessment by trained assessors to self-administered questionnaires [1]. The Stanford health assessment questionnaire (HAQ) is one of the instruments available for measuring the health status and physical function of RA patients [2]. It is used widely and has been tested extensively for reliability, validity, and responsiveness [3]. It is self-administered, short, simple, easy to score and has been extensively used in clinical trials. However, in order to determine functional status of patients with RA from different ethnic backgrounds, this self-administered questionnaire must be adapted and validated in different cultural contexts. The HAQ has been successfully translated into different languages [4–19]. It has been translated two times to Arabic classic language. Our aim was to adapt the HAQ to Moroccan culture and life style and validate this dialectal Arabic version to suit the needs of Moroccan patients with RA.

Materials and methods

Patients

We led a prospective study from July 2004 to September 2005. One hundred consecutive dialectal Moroccan-speaking patients with RA attending the rheumatology clinic at El Ayachi hospital were recruited into the study. All patients fulfilled the American Rheumatism Association (ACR) 1987 revised criteria for RA [20]. They have been seen either in consultation or during a hospitalization. They all gave informed consent. Each patient benefited from two clinical visits with an interval of between 2 and 10 days. No adjunction or modification in treatment has been authorized. The questionnaire was self administered on each visit. For analphabetic patients, the questionnaire was read by another person without any modification of the content. The HAQ explores eight domains of activity with two to three items for every domain. Every item is quoted from 0 to 3 according to the level of difficulty encountered by the use of an instrument or the help of a third person. To measure disease activity and severity, clinical parameters, laboratory tests and radiographic evaluation were performed at the first visit (patient self-assessment of pain using visual analog scale going from 0 to 100; morning stiffness; disease duration; number of tender joints; number of swollen joints; erythrocyte sedimentation rate (ESR); rheumatoid factor and radiological Steinbrocker score).

Translation and cultural adaptation of HAQ

The translation of HAQ into dialect Arabic followed proposed guidelines by Guillemin and colleagues [21, 22]. The initial translation from the original language to the target language was done by two groups of translators. Each group contained two bilingual translators. Their mother tongue was dialect Arabic. The first group was not informed of the concepts being quantified and had no medical qualifications. This group was less influenced by the academic goal and offered a translation that reflected the language used by the population. The second group included one rheumatologist. It was aware of the concepts being examined. To ensure accuracy, the forward translation was back-translated into English by two other groups of translators with English culture totally blinded to the original version. The expert committee contained translators, back-translators, a sociologist, a teacher in linguistics, and two rheumatologists. Its role was to consolidate all the translated and back translated versions of the questionnaire, review the discrepancies, and develop the prefinal version of the questionnaire for field testing. A few questionable items were discussed and resolved. To adapt the HAQ to Moroccan culture and life style, the following modifications were made: to the instruction “cut your meat”, was added the words “with the knife”, because Moroccan people eat with their fingers. “Lift a full cup or glass to your mouth” was substituted by “lift a full glass to your mouth” because the cup is not frequently used, and “or” in the item means that we can use one of them without any modification in the meaning. In the item “open a new milk carton,” new was deleted because “milk carton” in our context is opened once. The question “Reach and get down a 5-pound object (such as a bag of sugar) from just above your head”, 5 pound has been replaced by 2 kg, measure appropriate to our culture. The difference between the two measures was considered by the expert committee as non-significant. Figure 1 presents the dialectal Arabic HAQ (HAQAd).

Fig. 1
figure 1

Final version of HAQAd

Full size image

Psychometric evaluation of HAQAd

Acceptability

The acceptability factor was tested by studying the percentage of refusals, missing items, and complete questionnaires. The acceptability questionnaire, which comprises the percentage of disturbing items, items that were hard to understand or confusing, and the willingness to fill out the questionnaire a second time, has been also tested.

Reliability

Internal consistency refers to the degree of correlation between items. Cronbach’s coefficient is the most widely used method for assessing internal consistency. Cronbach’s alpha was calculated in each dimension of the instrument to assess the internal consistency reliability. A high alpha coefficient (≥0.70) suggests that the items within a dimension measure the same construct and support the construct validity [23]. Internal consistency was calculated from the results of the first administration.

Test–retest reliability refers to the stability of a score derived from several administrations of a measure by the same person. Here, patients in the stable phase (between visits one and two) and in whom the treatment remained unchanged were assessed. Test–retest reliability was tested using intraclass correlation coefficients (ICC) for the global score because the format of answer is continuous. The ICC estimates the correlation between two measures concering the same subject. Its value is comprised between +1 (perfect reproducibility) and 0 (hopeless reproducibility). A value >0.80 is considered satisfactory [24]. The reproducibility for each domain has been appreciated using the coefficient kappa. This one measures the degree of agreement between the two responses. A coefficient kappa >0.8 indicates a good concordance [25].

Construct validity

Convergent validity

The convergent validity was assessed by correlating the HAQAd with disease activity and severity markers including pain rated on a visual analog scale; morning stiffness; disease duration; number of tender joints; number of swollen joints; erythrocyte sedimentation rate (ESR); rheumatoid factor and radiological Steinbrocker score. This correlation was evaluated by the Spearman correlation coefficient (rs). It is considered to be excellent if rs > 0.91, good if 0.71 < rs < 0.90, moderate if 0.51 < rs < 0.70, low if 0.31 < rs < 0.50, and minim or absent if rs < 0.3 [26].

Item internal convergency represents the correlation between different domains. The domain which measures similar dimensions produces high correlations. Rates >0.60 correspond to a high correlation, moderate between 0.30 and 0.60, and low correlation below 0.30[27].

Structural validity was assessed by the principal component analysis. We compared the correlations of the revised items with the first principal component and the correlations of the unchanged items with this principal component.

Statistical methods

Data were analysed using SPSS for Windows version 13 (SPSS, Chicago, IL, USA). Test results are reported as significant if P < 0.05. Quantitative variables were described by using mean ± standard deviation, and ranges. Psychometric properties were studied as cited previously.

Results

Study population

A total of 100 patients (85 women and 15 men) with RA were recruited in the psychometric study. They came from different regions, urban and rural. Their demographic and medical characteristics are listed in Table 1.

Table 1 Demographic and medical characteristics of our patients
Full size table

Acceptability

No refusal or missing item has been noted, and the HAQAd has been willingly fulfilled a second time by patients.

Reliability

Cronbach’s alpha showed strong reliability with a standardized α = 0.978 among the 20 items, and the Cronbach’s alpha of domains ranged from 0.960 to 0.988. The domains with the highest correlation to the total instrument were “Grip” and “Activities” (α = 0.96), and domains with the lowest correlation were “Dressing and grooming” and “Walking” (α = 0.988) (Table 2). None of the 20 items significantly increased or decreased the standardized alpha if it was eliminated from the scale.

Table 2 Test–retest reliability and internal consistency of HAQAd
Full size table

Table 2 displays mean and standard deviation of test–retest results in each of the eight domains of HAQAd. The test–retest reliability yielded intraclass correlation coefficients of 0.987 for HAQAd score. The kappa test ranged from 0.70 to 0.92 for all domains.

Construct validity

As can be seen in the Table 3, statistically significant correlation was observed between the HAQAd and pain rated on a visual analog scale, morning stiffness, disease duration, number of tender joints, number of swollen joints, erythrocyte sedimentation rate (ESR), rheumatoid factor and radiological Steinbrocker score (rs between 0.26 and 0.82). This correlation was the strongest with the painful joint index (rs = 0.82) and the lowest but still statistically significant with rheumatoid factor (rs = 0.26).

Table 3 Correlation between HAQAd and disease activity and severity parameters
Full size table

The correlation matrix between the domains yielded correlation coefficients of 0.62 (Walking and Eating) to 0.86 (Reach and Hygiene) (Table 4).

Table 4 Correlation matrix for HAQAd
Full size table

The principal component analysis resulted in one factor explaining 71.5% of the total variance of the scale. All 20 items had similar correlation coefficients (range 0.747–0.897) with this principal component (Table 5). The correlations between each of the four revised items (items: 5, 6, 7, 13) and the first principal component were similar in magnitude to those between the 16 unchanged items and this component.

Table 5 First component in principal component analysis
Full size table

Discussion

The results from this study support the hypothesis that the HAQAd is a valid and reliable instrument to measure functional disability in Moroccan-speaking patients with RA. This study suited a major need of Moroccan people. Indeed, an important problem in the applicability of functional disability questionnaire is the variation in lifestyle and culture among countries. Two Arabic versions of HAQ exist and have been developed by Kuwaitis and Egyptians [12, 18]. However, both have been done in classical Arabic. Our translation was done from English to the dialect Arabic and not to the classical Arabic and therefore it could be used by the illiterate patients. On the other hand, the Egyptian version was validated on a cohort of 184 patients. They have different origins: 78 Egyptians, 42 patients from Saudi Arabia, 20 Sudanese, 14 Syrians, 12 from Bahrain, 10 Kuwaitis and only 8 Moroccans [18]. Therefore, this version is not adapted to Moroccan culture and cannot be used by our patients. We obtained a translation of the Functional Disability Index of the Stanford health assessment questionnaire appropriate to Moroccan social and culture conditions, without modifying the structure of the instrument. It was necessary to adapt four items. No adjunction or deletion of any item had been noted.

The characteristics of patients included for psychometric evaluation of HAQAd rejoin the literature data [28]. The feminine predominance and the young age were noted among our patients. All patients came from different regions of the country. This plurality in the origin allows taking account of the linguistic and cultural differences that exists in our country. Our patients can be considered as representative of Moroccan population.

During the translation, we chose the simplest terms and the frequently used ones. Moroccan dialect is akin to those of the other countries of the Maghreb (Algeria, Tunisia, Mauritania, etc.), the HAQAd could be used therefore in these countries.

The internal consistency was high and indicates that the components of the scale measure the same construct. Cronbach’s alpha was 0.978 for the total score and >0.70 for the domains of HAQAd, which are in the acceptable ranges for internal consistency. These results are comparable to those of other versions. Indeed, the Cronbach’s alpha test range from 0.86 [13] to 0.979 [18].

Therefore, HAQAd permits to value the patient’s state at different moments, and give similar results in comparable situations. Otherwise, although variations from items exist, all items of the HAQAd measure the same phenomenon, i.e., the functional disability. Test–retest reliability showed a strong reliability with high values for kappa and ICC = 0.987. These results are similar to those presented in other adaptations [4–19]. Indeed, the intraclass correlation coefficients reported in the literature ranged from 0.829 [18] to 0.99 [14], and our result matches previous results.

These results underline the strong reliability and the precision of the HAQAd questionnaire.

The validity of content is not discussed here in so far as it has been established at the time of the development of the instrument. Particular attention has been given to the translation using linguistic and conceptual equivalences in order to respect the principles and concepts having permitted its construction.

The HAQAd was found to have correlation with disease activity and severity markers. This correlation was statistically significant with disease activity and severity parameters. The highest correlation was observed for the number of tender joints (rs = 0.82), and the lowest but still statistically significant for rheumatoid factor (rs = 0.26). These results are comparable to those of previous studies [12, 13, 17, 18]. However, the Korean version had not found significant correlation with ESR [14].

A convergent validity exists between the HAQAd instrument and the different parameters of assessment of RA. Therefore, this adapted questionnaire constitutes an element of assessment of RA and a means for the follow-up of its evolution.

Spearman correlation coefficients between all domains comprised between 0.62 and 0.86. Less good results have been presented in the two Arabian versions [12, 18]. El Meidany et al. reported a rate of rs between 0.57 and 0.83 [18]. In the Kuwaitis version rs ranged from 0.43 to 0.79 [12].

The principal component analysis resulted in one factor explaining 71.5% of the total variance of the scale. These results are similar to those of other versions [8, 11, 14, 17, 18]. The four revised items were appropriate expressions of physical function since the correlations between the revised items and the first principal component are similar in magnitude to those between the unchanged items and this principal component.

Conclusion

The Moroccan HAQAd is a reliable and valid instrument that can be self-administered to evaluate functional disability in Moroccan-speaking patients with RA. Its property to explore the functional disability in all its diversity and its consistency is conserved. Several practical applications of this instrument can be considered: assessment by the patients of their own disability; regular follow-up of arthritis patients by the physician; assessment of the effect of the treatment; and principal or secondary judgment criteria of clinical trials. It will allow the medical team to evaluate pertinently the functional prognosis, and therefore the quality of life of RA Moroccan patients.