Clin Rheumatol DOI 10.1007/s10067-014-2515-4
ORIGINAL ARTICLE
Reliability, validity, and responsiveness of the Persian version of the rheumatoid and arthritis outcome score (RAOS) in patients with rheumatoid arthritis Hossein Negahban & Fatemeh Masoudpur & Elham Rajaei & Mohammad Ali Nazarinia & Masood Mazaheri & Mahyar Salavati
Received: 12 August 2013 / Revised: 26 December 2013 / Accepted: 24 January 2014 # Clinical Rheumatology 2014
Abstract The aims of this study were to cross-culturally translate the original rheumatoid and arthritis outcome score (RAOS) into Persian and evaluate its reliability, validity, and responsiveness in a group of patients with rheumatoid arthritis (RA). The questionnaire was translated through a standard forward-backward translation. A sample of 103 patients was asked to complete the Persian RAOS, the Short Form-36 (SF36), and the arthritis impact measurement scale-short form (AIMS2-SF). To determine test-retest reliability, the Persian RAOS was readministered to a sample of 50 patients, 3–6 days H. Negahban (*) : F. Masoudpur Musculoskeletal Rehabilitation Research Center, Ahvaz Jundishapur University of Medical Sciences, Ahvaz, Iran e-mail: [email protected] F. Masoudpur e-mail: [email protected] E. Rajaei Department of Rheumatology, Ahvaz Jundishapur University of Medical Sciences, Ahvaz, Iran e-mail: [email protected] M. A. Nazarinia Shiraz Geriatric Research Center, Shiraz University of Medical Sciences, Shiraz, Iran e-mail: [email protected] M. Mazaheri MOVE Research Institute Amsterdam, Faculty of Human Movement Sciences, VU University Amsterdam, Amsterdam, The Netherlands M. Mazaheri Musculoskeletal Research Center, Isfahan University of Medical Sciences, Isfahan, Iran e-mail: [email protected] M. Salavati Department of Physical Therapy, University of Social Welfare and Rehabilitation Sciences, Tehran, Iran e-mail: [email protected]
after the first visit. To evaluate responsiveness, 50 patients completed the Persian RAOS at baseline and at the end of a pharmacological intervention. Test-retest reliability and internal consistency were assessed using intraclass correlation coefficient (ICC) and Cronbach’s alpha, respectively. Construct validity was assessed by comparing the results of the RAOS with the Persian SF-36 and AIMS2-SF using Spearman's correlation coefficient. Responsiveness was assessed by the calculation of effect size (ES) and standardized response means (SRM). The acceptable level of ICC> 0.70 and Cronbach’s alpha>0.70 were obtained for the most RAOS subscales. As expected, moderate to strong correlations were observed between subscales of the RAOS and the SF-36/AIMS2-SF intended to measure similar constructs. The ES range of 0.18 to 0.51 and the SRM range of 0.25 to 0.91 were obtained for the RAOS subscales. In conclusion, the Persian RAOS is a reliable, valid, and responsive outcome measure for patients with RA suffering from arthritis in the lower limb joints. Keywords Outcome . Persian version . Psychometric property . Rheumatoid arthritis
Introduction Rheumatoid arthritis (RA) is a chronic inflammatory disease that affects women more than men with a ratio of 3:1 [1–3]. The peak onset age of the disease lies between the third and fifth decades [3]. RA leads to joint pain and swelling, morning stiffness, decreased range of motion, and in most cases, impaired functional activities (disability) and reduced quality of life [1, 2]. Laboratory and radiologic tests and/or other hightechnology instruments used for assessing RA may not
Clin Rheumatol
sufficiently address the major problems of pain, functional limitations, and disability [4]. Therefore, in the past decade parallel to advancement in treatment strategies, various instruments of outcome measurements have been developed to assess pain, disability, and health-related quality of life in RA patients. These instruments include the Short Form-36 (SF-36), the health assessment questionnaire (HAQ), and the arthritis impact measurement scale-short form (AIMS2-SF) [4]. The RA disease especially affects the weight-bearing joints in the lower limb [5]. Therefore, there is a need to concentrate on validation of an outcome measure that is specifically developed to evaluate the effects of hip, knee, and ankle RA on functional limitation and disability [5]. The rheumatoid and arthritis outcome score (RAOS) is an adaptation of the knee injury and osteoarthritis outcome score (KOOS), developed to assess pain, functional limitation, and quality of life in patients with chronic inflammatory diseases of the joints in the lower extremity [2]. To use an outcome measure in a given country, it is necessary to cross-culturally translate the instrument and to evaluate its psychometric properties [6]. To date, the original RAOS has been translated and validated into the Swedish [2], English [2], Turkish [1], and French [5] languages. However, no study has yet translated and validated the RAOS into Persian. Therefore, the aims of this study were to cross-culturally translate the original RAOS into Persian and to evaluate its reliability, validity, and responsiveness in a group of patients with RA.
Materials and methods Translation process Permission for translation was obtained from the developer (Prof. E. Roos). The cross-cultural translation from the original English to Persian was performed based on the international guidelines provided in the literature [6, 7]. First, forward translation of this questionnaire was produced by two native Persian translators who independently translated the original English version into Persian. Translators were asked to focus on conceptual rather than on literal translation to preserve the meaning of each item [8]. A single forward translation was obtained after a consensus meeting with both translators and the principle investigator. The final forward translation of the RAOS was translated back into English by a native American-English translator who was blinded to the original version. Finally, a multidisciplinary committee consisting of two physiotherapists, two rheumatologists, and one outcome researcher reviewed the backward translation against the original version to check for the discrepancies. They also checked the comprehension of the translation and verified cross-cultural equivalence of the final
Persian version. In the final step, a pilot study on 20 patients was conducted to check whether the patients fully understand all questions and response choices. Patients A consecutive sample of 103 patients with RA were recruited from Foolad rheumatology clinic and rheumatology department of Golestan Hospital in Ahvaz and also Motahhari rheumatology clinic and rheumatology department of Hafez Hospital in Shiraz between June 2012 and June 2013. Patients were diagnosed with RA based on the American College of Rheumatology (ACR) 2010 criteria [9]. Patients were included if they had (1) medium to high disease activity as measured by a rheumatologist based on disease activity score-28≥3.2, (2) at least one symptomatic joint arthritis of the lower limbs [5], and (3) ability to speak Persian. Patients were excluded if they had (1) rheumatic diseases other than RA [5], (2) recent fracture or joint surgery [5, 8], (3) intraarticular injection in the joints of the lower limbs [5], and (4) cognitive impairment [8]. In the first visit, patients were asked to complete the Persian versions of the RAOS, SF-36, and AIMS2-SF. To determine the test-retest reliability, the Persian RAOS was readministered to a sample of 50 patients whom health status were not changed during the first and second visit. An interval of 3–6 days between the test and retest sessions was chosen to minimize the influence of learning and change in health status. To ensure that the health status of the patients remained stable between test-retest sessions, they were asked to answer whether their symptoms were better, same, or worse since filling out the initial questionnaire. Only patients with the answer “same” were included in the retest session. As the respondents were expected to answer all questions, the researchers checked all items after completing the questionnaire. In case of a missing item, the participants were asked to respond to the unanswered questions. Also, to evaluate responsiveness, 50 patients completed the Persian versions of the RAOS at baseline and after 2 month of pharmacological intervention including diseasemodifying anti-rheumatic drugs (DMARDs): hydroxychloroquine sulfate, methotrexate, ARAVA, and prednisolone. This subgroup of patients should not participate in any other forms of treatment such as physical therapy and intraarticular injection. Due to the beneficial effect of combined DMARDs therapy compared with monotherapy [10, 11], all patients received more than one DMARD. All patients were requested to sign a written informed consent form approved by the Ethics Committee at Ahvaz Jundishapur University of Medical Sciences, Ahvaz, Iran. Instruments The RAOS is a self-reported instrument and consists of 42 items of five subscales: pain (9 items), symptom (7 items),
Clin Rheumatol
activities of daily living (ADL) (17 items), sports and recreations (sport/rec) (5 items), and quality of life (QoL) (4 items). All items were scored from zero to four, and each of the five subscale scores was calculated as the sum of the items included. Raw scores were then transformed to a zero (indicating extreme problem) to 100 (indicating no problem) scale [2]. The SF-36 is a 36-item generic, self-reported measure which quantifies general health and wellness [12]. It consists of eight subscales including physical functioning (PF), rolephysical (RP), bodily pain (BP), general health (GH), vitality (VT), social functioning (SF), role-emotional (RE), and mental health (MH). Scores for each subscale range from 0 (poor health status) to 100 (good health status). From these eight subscales, two summary measures namely physical health summary measure (PHSM) (including PF, RP, BP, and GH) and mental health summary measure (MHSM) (including VT, SF, RE, and MH) can be computed [13]. The Persian version of SF-36 has been validated and can be used in Iran [14]. The AIMS2-SF consists of 26 items which covers five subscales including Physical, Symptom, Affect, Social interaction, and Role. The scores of each subscale ranges from 0 to 10 for which higher scores represent poorer health status [8].
Data analysis Data were analyzed using the statistical package for the social sciences version 16.0 (SPSS Inc., Chicago, IL, USA). The test-retest reliability was assessed using the two-way, randomeffect model of intraclass correlation coefficient (ICC) with absolute agreement definition and 95 % confidence interval (95 % CI) [15, 16]. An ICC equal to or greater than 0.70 was considered acceptable for test-retest reliability [15]. The standard error of measurement (SEM) was calculated from the root mean square error term of the repeated measures analysis of variance table and aimed to estimate measurement precision associated with repeated measurements [17]. The SEM is useful for computing the smallest detectable change (SDC) which is the smallest change in the health status that could be considered clinically important between pre-post interventions [18]. The SDC was defined as the 95 % CI of SEM (±1.96 SEM) [19]. Internal consistency was assessed for the baseline administration by calculation of Cronbach’s alpha coefficient [15]. A Cronbach’s alpha level equal to or greater than 0.70 was considered satisfactory for items homogeneity of the Persian RAOS [15]. The property of items homogeneity was further assessed thorough the correlation between each item and its corresponding subscale as a whole excluding that item (itemsubscale correlation after correction for overlap). Spearman's correlation coefficient equal to or greater than 0.40 was considered acceptable [15].
To determine whether the Persian RAOS actually consists of five subscales, a forced five-factor solution with principal component analysis (PCA) was performed on the first administration of the RAOS. However, it is necessary to note that an absolute minimum of five patients per item (i.e., 42×5=210 patients) is necessary to conduct factor analysis on the sample [20], while the sample size of the current study consisted of 103 patients. A forced five-factor solution of the Persian RAOS was performed using the PCA with varimax rotation. Only factor loadings above 40 % were considered as indicative of item loading. Construct validity was assessed by comparing the results of the RAOS with validated Persian version of the SF-36 and AIMS2-SF for the first administration using Spearman's correlation coefficient (rs). Correlation coefficients greater than 0.60, 0.60–0.30, and less than 0.30 were considered as strong, moderate, and weak, respectively [1, 15]. It was hypothesized a priori that there is moderate to strong correlations between (1) the RAOS Pain and SF-36 BP subscales, (2) the RAOS ADL and SF-36 PF subscales, and (3) the RAOS sport/rec and SF-36 PF subscales. Finally, it was hypothesized that the correlations between the RAOS subscales and the SF-36 subscales of PHSM (PF, RP, BP, GH) should be higher than between the RAOS subscales and the SF-36 subscales of MHSM (VT, SF, RE, MH). For the AIMS2-SF, we hypothesized strong correlation between the RAOS and AIMS2-SF physical, symptom, and role subscales. Also, weak correlation was expected between the RAOS and AIMS2-SF affect and social interaction subscales since they do not measure similar constructs. Content validity was tested by assessing the ceiling and floor effects of the RAOS on the first administration. Achievement of highest or lowest scores by more than 15 % of the patients is considered as a cutoff point for a poor content validity [21]. Large percentage of ceiling and floor effects limits the interpretability of change scores [22]. Responsiveness is defined as the capacity of an instrument to detect clinically meaningful changes over time [22, 23]. Changes due to the pharmacological intervention were assessed by paired t test. There is a lack of consensus on how to evaluate the responsiveness in the literature [23, 24]. To increase the comparability of our study, the authors of this study reported the results of effect size (ES) and standardized response mean (SRM) as these measures have been used in other validation studies of the RAOS. The ES is defined as the mean change score of pre-post intervention divided by the standard deviation (SD) of the baseline score [24–26]. The SRM is defined as the mean change score of pre-post intervention divided by the SD of the change score [24–26]. The ES and SRM of 0.20 to 0.50, 0.50 to 0.80, and greater than 0.80 were considered as small, medium, and large responsiveness, respectively [24, 25].
Clin Rheumatol
Results Table 1 shows the demographic and clinical characteristics of the patients. Due to cultural and social differences, some minor changes were made to the Persian version. Item 13 of the ADL subscale regarding the degree of difficulty for “get in/out of bath” was changed to “take a bath” because tubs are not used frequently in Iran. The examples of “praying in sitting position” and “use of European toilet instead of Iranian toilet” were added to the second item of QoL for better understanding of the question. All patients felt the Persian RAOS as clear and easy to understand. The results of test-retest reliability and internal consistency were shown in Table 2. The acceptable level of ICC>0.70 and Cronbach’s alpha>0.70 (except QoL that was in the marginal level of 0.67) were obtained for the Persian RAOS. The ICC and Cronbach’s alpha for the five subscales ranged from 0.85
Table 1 Demographic and clinical characteristics of participants completing the RAOS (n=103) n (%) unless stated Demographic data Age (year), mean (SD) Height (m), mean (SD) Weight (kg), mean (SD) Sex Male Female Years of education 6–8 9–12
– 47 (10.62) 1.61 (0.08) 73.64 (14.44) – 12 (11.7) 91 (88.3) – 42 (40.8) 43 (41.7)
>12 Marital status Single Couple Clinical data Disease duration (month), mean (SD) Side of involved leg Right Left Both Affected joint Hip Knee Foot Hip/knee Knee/foot Triple joints
18 (17.4) – 5 (4.9) 98 (95.1) – 35.08 (55.53) – 13 (12.6) 7 (6.8) 83 (80.6) – 3 (2.9) 31 (30.1) 5 (4.9) 7 (6.8) 38 (36.9) 19 (18.4)
SD standard deviation
to 0.93 and the 0.67 to 0.88, respectively. The results of corrected item-total correlations were shown in Table 3. Spearman's correlation≥0.40 was obtained for almost all items of the RAOS with the exception of one item of the pain subscale, two items of the symptom, two items of the ADL, one item of the sport/rec, and one item of the QoL subscale. The results of the factor analysis showed that the forced five-factor solution of the Persian RAOS constituted almost 52 % of the variance with 30.52.78, 6.77, 5.86, 4.45, and 4.01 % of variance explained by each factor, respectively. The eigenvalue of 1 was exceeded by all components as eigenvalues of 12.82, 2.84, 2.46, 1.87, and 1.68 were seen for the five factors. Factor 1 corresponded to most of the RAOS items. Factor 2 corresponded to one item of symptom and two items of ADL. Three items of the symptom had overlap loading on factors 1, 2, and 3. One item of symptom and one item of ADL had overlap loading on factors 1 and 4. Factor 5 corresponded to two items of pain and one item of ADL. All items of sport/rec and QoL had loading on factor 1. As expected, strong positive correlation was observed between subscales of the RAOS and the SF-36 intended to measure similar constructs (Table 4): RAOS pain vs. SF-36 BP (rs =0.64); RAOS ADL vs. SF-36 PF (rs =0.65); and RAOS sport/rec vs. SF-36 PF (rs =0.64). Also, with the exception of moderate correlation between the RAOS ADL and SF-36 MH (rs =0.32), weak positive correlation was obtained between subscales of the RAOS and the SF-36 RE and MH (rs range=0.07 to 0.26). Finally, as hypothesized, there were moderate to strong positive correlations between the RAOS and the SF-36 PHSM (rs range=0.43 to 0.61) while low to moderate positive correlations were detected between the RAOS and the SF-36 MHSM (rs range=0.27 to 0.48). For the AIMS2-SF, moderate to strong negative correlations were seen between the RAOS and AIMS2-SF physical, symptom, and role subscales (rs range=−0.36 to −0.66). Also, as expected weak negative correlations were found between the RAOS and AIMS2-SF affect and social interaction subscales (rs range=−0.01 to −0.20). There were no ceiling and floor effects for the Persian RAOS and AIMS2-SF. However, a ceiling effect of 51.5 % and a floor effect of 33 % were obtained for the RP and RE subscales of the SF-36, respectively. The mean (SD) of days from baseline (preintervention) to follow-up (postintervention) measurement was 59.5 (2.6). A significant improvement was found for all subscales of the Persian RAOS (p
ORIGINAL ARTICLE
Reliability, validity, and responsiveness of the Persian version of the rheumatoid and arthritis outcome score (RAOS) in patients with rheumatoid arthritis Hossein Negahban & Fatemeh Masoudpur & Elham Rajaei & Mohammad Ali Nazarinia & Masood Mazaheri & Mahyar Salavati
Received: 12 August 2013 / Revised: 26 December 2013 / Accepted: 24 January 2014 # Clinical Rheumatology 2014
Abstract The aims of this study were to cross-culturally translate the original rheumatoid and arthritis outcome score (RAOS) into Persian and evaluate its reliability, validity, and responsiveness in a group of patients with rheumatoid arthritis (RA). The questionnaire was translated through a standard forward-backward translation. A sample of 103 patients was asked to complete the Persian RAOS, the Short Form-36 (SF36), and the arthritis impact measurement scale-short form (AIMS2-SF). To determine test-retest reliability, the Persian RAOS was readministered to a sample of 50 patients, 3–6 days H. Negahban (*) : F. Masoudpur Musculoskeletal Rehabilitation Research Center, Ahvaz Jundishapur University of Medical Sciences, Ahvaz, Iran e-mail: [email protected] F. Masoudpur e-mail: [email protected] E. Rajaei Department of Rheumatology, Ahvaz Jundishapur University of Medical Sciences, Ahvaz, Iran e-mail: [email protected] M. A. Nazarinia Shiraz Geriatric Research Center, Shiraz University of Medical Sciences, Shiraz, Iran e-mail: [email protected] M. Mazaheri MOVE Research Institute Amsterdam, Faculty of Human Movement Sciences, VU University Amsterdam, Amsterdam, The Netherlands M. Mazaheri Musculoskeletal Research Center, Isfahan University of Medical Sciences, Isfahan, Iran e-mail: [email protected] M. Salavati Department of Physical Therapy, University of Social Welfare and Rehabilitation Sciences, Tehran, Iran e-mail: [email protected]
after the first visit. To evaluate responsiveness, 50 patients completed the Persian RAOS at baseline and at the end of a pharmacological intervention. Test-retest reliability and internal consistency were assessed using intraclass correlation coefficient (ICC) and Cronbach’s alpha, respectively. Construct validity was assessed by comparing the results of the RAOS with the Persian SF-36 and AIMS2-SF using Spearman's correlation coefficient. Responsiveness was assessed by the calculation of effect size (ES) and standardized response means (SRM). The acceptable level of ICC> 0.70 and Cronbach’s alpha>0.70 were obtained for the most RAOS subscales. As expected, moderate to strong correlations were observed between subscales of the RAOS and the SF-36/AIMS2-SF intended to measure similar constructs. The ES range of 0.18 to 0.51 and the SRM range of 0.25 to 0.91 were obtained for the RAOS subscales. In conclusion, the Persian RAOS is a reliable, valid, and responsive outcome measure for patients with RA suffering from arthritis in the lower limb joints. Keywords Outcome . Persian version . Psychometric property . Rheumatoid arthritis
Introduction Rheumatoid arthritis (RA) is a chronic inflammatory disease that affects women more than men with a ratio of 3:1 [1–3]. The peak onset age of the disease lies between the third and fifth decades [3]. RA leads to joint pain and swelling, morning stiffness, decreased range of motion, and in most cases, impaired functional activities (disability) and reduced quality of life [1, 2]. Laboratory and radiologic tests and/or other hightechnology instruments used for assessing RA may not
Clin Rheumatol
sufficiently address the major problems of pain, functional limitations, and disability [4]. Therefore, in the past decade parallel to advancement in treatment strategies, various instruments of outcome measurements have been developed to assess pain, disability, and health-related quality of life in RA patients. These instruments include the Short Form-36 (SF-36), the health assessment questionnaire (HAQ), and the arthritis impact measurement scale-short form (AIMS2-SF) [4]. The RA disease especially affects the weight-bearing joints in the lower limb [5]. Therefore, there is a need to concentrate on validation of an outcome measure that is specifically developed to evaluate the effects of hip, knee, and ankle RA on functional limitation and disability [5]. The rheumatoid and arthritis outcome score (RAOS) is an adaptation of the knee injury and osteoarthritis outcome score (KOOS), developed to assess pain, functional limitation, and quality of life in patients with chronic inflammatory diseases of the joints in the lower extremity [2]. To use an outcome measure in a given country, it is necessary to cross-culturally translate the instrument and to evaluate its psychometric properties [6]. To date, the original RAOS has been translated and validated into the Swedish [2], English [2], Turkish [1], and French [5] languages. However, no study has yet translated and validated the RAOS into Persian. Therefore, the aims of this study were to cross-culturally translate the original RAOS into Persian and to evaluate its reliability, validity, and responsiveness in a group of patients with RA.
Materials and methods Translation process Permission for translation was obtained from the developer (Prof. E. Roos). The cross-cultural translation from the original English to Persian was performed based on the international guidelines provided in the literature [6, 7]. First, forward translation of this questionnaire was produced by two native Persian translators who independently translated the original English version into Persian. Translators were asked to focus on conceptual rather than on literal translation to preserve the meaning of each item [8]. A single forward translation was obtained after a consensus meeting with both translators and the principle investigator. The final forward translation of the RAOS was translated back into English by a native American-English translator who was blinded to the original version. Finally, a multidisciplinary committee consisting of two physiotherapists, two rheumatologists, and one outcome researcher reviewed the backward translation against the original version to check for the discrepancies. They also checked the comprehension of the translation and verified cross-cultural equivalence of the final
Persian version. In the final step, a pilot study on 20 patients was conducted to check whether the patients fully understand all questions and response choices. Patients A consecutive sample of 103 patients with RA were recruited from Foolad rheumatology clinic and rheumatology department of Golestan Hospital in Ahvaz and also Motahhari rheumatology clinic and rheumatology department of Hafez Hospital in Shiraz between June 2012 and June 2013. Patients were diagnosed with RA based on the American College of Rheumatology (ACR) 2010 criteria [9]. Patients were included if they had (1) medium to high disease activity as measured by a rheumatologist based on disease activity score-28≥3.2, (2) at least one symptomatic joint arthritis of the lower limbs [5], and (3) ability to speak Persian. Patients were excluded if they had (1) rheumatic diseases other than RA [5], (2) recent fracture or joint surgery [5, 8], (3) intraarticular injection in the joints of the lower limbs [5], and (4) cognitive impairment [8]. In the first visit, patients were asked to complete the Persian versions of the RAOS, SF-36, and AIMS2-SF. To determine the test-retest reliability, the Persian RAOS was readministered to a sample of 50 patients whom health status were not changed during the first and second visit. An interval of 3–6 days between the test and retest sessions was chosen to minimize the influence of learning and change in health status. To ensure that the health status of the patients remained stable between test-retest sessions, they were asked to answer whether their symptoms were better, same, or worse since filling out the initial questionnaire. Only patients with the answer “same” were included in the retest session. As the respondents were expected to answer all questions, the researchers checked all items after completing the questionnaire. In case of a missing item, the participants were asked to respond to the unanswered questions. Also, to evaluate responsiveness, 50 patients completed the Persian versions of the RAOS at baseline and after 2 month of pharmacological intervention including diseasemodifying anti-rheumatic drugs (DMARDs): hydroxychloroquine sulfate, methotrexate, ARAVA, and prednisolone. This subgroup of patients should not participate in any other forms of treatment such as physical therapy and intraarticular injection. Due to the beneficial effect of combined DMARDs therapy compared with monotherapy [10, 11], all patients received more than one DMARD. All patients were requested to sign a written informed consent form approved by the Ethics Committee at Ahvaz Jundishapur University of Medical Sciences, Ahvaz, Iran. Instruments The RAOS is a self-reported instrument and consists of 42 items of five subscales: pain (9 items), symptom (7 items),
Clin Rheumatol
activities of daily living (ADL) (17 items), sports and recreations (sport/rec) (5 items), and quality of life (QoL) (4 items). All items were scored from zero to four, and each of the five subscale scores was calculated as the sum of the items included. Raw scores were then transformed to a zero (indicating extreme problem) to 100 (indicating no problem) scale [2]. The SF-36 is a 36-item generic, self-reported measure which quantifies general health and wellness [12]. It consists of eight subscales including physical functioning (PF), rolephysical (RP), bodily pain (BP), general health (GH), vitality (VT), social functioning (SF), role-emotional (RE), and mental health (MH). Scores for each subscale range from 0 (poor health status) to 100 (good health status). From these eight subscales, two summary measures namely physical health summary measure (PHSM) (including PF, RP, BP, and GH) and mental health summary measure (MHSM) (including VT, SF, RE, and MH) can be computed [13]. The Persian version of SF-36 has been validated and can be used in Iran [14]. The AIMS2-SF consists of 26 items which covers five subscales including Physical, Symptom, Affect, Social interaction, and Role. The scores of each subscale ranges from 0 to 10 for which higher scores represent poorer health status [8].
Data analysis Data were analyzed using the statistical package for the social sciences version 16.0 (SPSS Inc., Chicago, IL, USA). The test-retest reliability was assessed using the two-way, randomeffect model of intraclass correlation coefficient (ICC) with absolute agreement definition and 95 % confidence interval (95 % CI) [15, 16]. An ICC equal to or greater than 0.70 was considered acceptable for test-retest reliability [15]. The standard error of measurement (SEM) was calculated from the root mean square error term of the repeated measures analysis of variance table and aimed to estimate measurement precision associated with repeated measurements [17]. The SEM is useful for computing the smallest detectable change (SDC) which is the smallest change in the health status that could be considered clinically important between pre-post interventions [18]. The SDC was defined as the 95 % CI of SEM (±1.96 SEM) [19]. Internal consistency was assessed for the baseline administration by calculation of Cronbach’s alpha coefficient [15]. A Cronbach’s alpha level equal to or greater than 0.70 was considered satisfactory for items homogeneity of the Persian RAOS [15]. The property of items homogeneity was further assessed thorough the correlation between each item and its corresponding subscale as a whole excluding that item (itemsubscale correlation after correction for overlap). Spearman's correlation coefficient equal to or greater than 0.40 was considered acceptable [15].
To determine whether the Persian RAOS actually consists of five subscales, a forced five-factor solution with principal component analysis (PCA) was performed on the first administration of the RAOS. However, it is necessary to note that an absolute minimum of five patients per item (i.e., 42×5=210 patients) is necessary to conduct factor analysis on the sample [20], while the sample size of the current study consisted of 103 patients. A forced five-factor solution of the Persian RAOS was performed using the PCA with varimax rotation. Only factor loadings above 40 % were considered as indicative of item loading. Construct validity was assessed by comparing the results of the RAOS with validated Persian version of the SF-36 and AIMS2-SF for the first administration using Spearman's correlation coefficient (rs). Correlation coefficients greater than 0.60, 0.60–0.30, and less than 0.30 were considered as strong, moderate, and weak, respectively [1, 15]. It was hypothesized a priori that there is moderate to strong correlations between (1) the RAOS Pain and SF-36 BP subscales, (2) the RAOS ADL and SF-36 PF subscales, and (3) the RAOS sport/rec and SF-36 PF subscales. Finally, it was hypothesized that the correlations between the RAOS subscales and the SF-36 subscales of PHSM (PF, RP, BP, GH) should be higher than between the RAOS subscales and the SF-36 subscales of MHSM (VT, SF, RE, MH). For the AIMS2-SF, we hypothesized strong correlation between the RAOS and AIMS2-SF physical, symptom, and role subscales. Also, weak correlation was expected between the RAOS and AIMS2-SF affect and social interaction subscales since they do not measure similar constructs. Content validity was tested by assessing the ceiling and floor effects of the RAOS on the first administration. Achievement of highest or lowest scores by more than 15 % of the patients is considered as a cutoff point for a poor content validity [21]. Large percentage of ceiling and floor effects limits the interpretability of change scores [22]. Responsiveness is defined as the capacity of an instrument to detect clinically meaningful changes over time [22, 23]. Changes due to the pharmacological intervention were assessed by paired t test. There is a lack of consensus on how to evaluate the responsiveness in the literature [23, 24]. To increase the comparability of our study, the authors of this study reported the results of effect size (ES) and standardized response mean (SRM) as these measures have been used in other validation studies of the RAOS. The ES is defined as the mean change score of pre-post intervention divided by the standard deviation (SD) of the baseline score [24–26]. The SRM is defined as the mean change score of pre-post intervention divided by the SD of the change score [24–26]. The ES and SRM of 0.20 to 0.50, 0.50 to 0.80, and greater than 0.80 were considered as small, medium, and large responsiveness, respectively [24, 25].
Clin Rheumatol
Results Table 1 shows the demographic and clinical characteristics of the patients. Due to cultural and social differences, some minor changes were made to the Persian version. Item 13 of the ADL subscale regarding the degree of difficulty for “get in/out of bath” was changed to “take a bath” because tubs are not used frequently in Iran. The examples of “praying in sitting position” and “use of European toilet instead of Iranian toilet” were added to the second item of QoL for better understanding of the question. All patients felt the Persian RAOS as clear and easy to understand. The results of test-retest reliability and internal consistency were shown in Table 2. The acceptable level of ICC>0.70 and Cronbach’s alpha>0.70 (except QoL that was in the marginal level of 0.67) were obtained for the Persian RAOS. The ICC and Cronbach’s alpha for the five subscales ranged from 0.85
Table 1 Demographic and clinical characteristics of participants completing the RAOS (n=103) n (%) unless stated Demographic data Age (year), mean (SD) Height (m), mean (SD) Weight (kg), mean (SD) Sex Male Female Years of education 6–8 9–12
– 47 (10.62) 1.61 (0.08) 73.64 (14.44) – 12 (11.7) 91 (88.3) – 42 (40.8) 43 (41.7)
>12 Marital status Single Couple Clinical data Disease duration (month), mean (SD) Side of involved leg Right Left Both Affected joint Hip Knee Foot Hip/knee Knee/foot Triple joints
18 (17.4) – 5 (4.9) 98 (95.1) – 35.08 (55.53) – 13 (12.6) 7 (6.8) 83 (80.6) – 3 (2.9) 31 (30.1) 5 (4.9) 7 (6.8) 38 (36.9) 19 (18.4)
SD standard deviation
to 0.93 and the 0.67 to 0.88, respectively. The results of corrected item-total correlations were shown in Table 3. Spearman's correlation≥0.40 was obtained for almost all items of the RAOS with the exception of one item of the pain subscale, two items of the symptom, two items of the ADL, one item of the sport/rec, and one item of the QoL subscale. The results of the factor analysis showed that the forced five-factor solution of the Persian RAOS constituted almost 52 % of the variance with 30.52.78, 6.77, 5.86, 4.45, and 4.01 % of variance explained by each factor, respectively. The eigenvalue of 1 was exceeded by all components as eigenvalues of 12.82, 2.84, 2.46, 1.87, and 1.68 were seen for the five factors. Factor 1 corresponded to most of the RAOS items. Factor 2 corresponded to one item of symptom and two items of ADL. Three items of the symptom had overlap loading on factors 1, 2, and 3. One item of symptom and one item of ADL had overlap loading on factors 1 and 4. Factor 5 corresponded to two items of pain and one item of ADL. All items of sport/rec and QoL had loading on factor 1. As expected, strong positive correlation was observed between subscales of the RAOS and the SF-36 intended to measure similar constructs (Table 4): RAOS pain vs. SF-36 BP (rs =0.64); RAOS ADL vs. SF-36 PF (rs =0.65); and RAOS sport/rec vs. SF-36 PF (rs =0.64). Also, with the exception of moderate correlation between the RAOS ADL and SF-36 MH (rs =0.32), weak positive correlation was obtained between subscales of the RAOS and the SF-36 RE and MH (rs range=0.07 to 0.26). Finally, as hypothesized, there were moderate to strong positive correlations between the RAOS and the SF-36 PHSM (rs range=0.43 to 0.61) while low to moderate positive correlations were detected between the RAOS and the SF-36 MHSM (rs range=0.27 to 0.48). For the AIMS2-SF, moderate to strong negative correlations were seen between the RAOS and AIMS2-SF physical, symptom, and role subscales (rs range=−0.36 to −0.66). Also, as expected weak negative correlations were found between the RAOS and AIMS2-SF affect and social interaction subscales (rs range=−0.01 to −0.20). There were no ceiling and floor effects for the Persian RAOS and AIMS2-SF. However, a ceiling effect of 51.5 % and a floor effect of 33 % were obtained for the RP and RE subscales of the SF-36, respectively. The mean (SD) of days from baseline (preintervention) to follow-up (postintervention) measurement was 59.5 (2.6). A significant improvement was found for all subscales of the Persian RAOS (p
