Cultural Adaptation, Validity, Reliability and Responsiveness of the Swedish Version of the Effective Musculoskeletal Consumer Scale (EC-17)

Abstract

Background

Endorsed by the Outcome MEasures in Rheumatoid Arthritis in Clinical Trials (OMERACT) group, The Effective Consumer Scale (EC-17) was developed in English for patients with musculoskeletal diseases to assess the skills and perceptions important for participating in and managing health care. The objective of this study was culturally to adapt the questionnaire into Swedish and to study its psychometric properties.

Methods

After translation of the questionnaire into Swedish, two different groups of outpatients from two specialist rheumatology departments participated in the study. Face validity was assessed, together with internal consistency, test–retest and responsiveness of the questionnaire. Construct validity was assessed using the Arthritis Self-Efficacy Scale (ASES), and responsiveness to a five day educational intervention was analysed using the standardized response mean (SRM).

Results

Analyses were based on 124 patients with inflammatory rheumatic diseases, of whom 50 attended the intervention. Data quality met the requirements, with missing values <5%, and floor and ceiling effects <15%. Item total correlations were all >0.4, ranging from 0.49 to 0.88. Cronbach's alpha was 0.93 and 0.95 for the two groups. The test–retest correlation (ICC_2.1) was 0.94, and there was a significant improvement as a result of the intervention, with an SRM of 0.43. However, the questionnaire had a higher construct validity with the ASES subscale ‘other symptoms’ than hypothesized a priori (r_s 0.75).

Conclusion

The Swedish version of the EC-17 met the required recommendations for face validity, internal consistency, test–retest reliability and responsiveness. Its construct validity needs to be further established, and the questionnaire needs further testing in different groups of patients and in different interventions. Copyright © 2012 John Wiley & Sons, Ltd.

Introduction

Over the past decade, there has been increasing support for the importance of self-management in arthritis care. Patient education and self-management are often considered as first-line non-pharmacological treatments in the assessment of patients with arthritis (American College of Rheumatology Subcommittee on Osteoarthritis, 2000, 2002). Several educational forms have been conducted to foster self-management and a focus on behavioural change by increasing patients' self-efficacy and skills to cope with arthritis in everyday life; most often, pain, symptoms and function have been the target for education (Lorig et al., 1989, 2006, 2008). Research supports patient benefits of knowledge, behaviours and health outcomes from self-management education programmes (Chodosh et al., 2005; Warsi et al., 2004), but the effects do not tend to last long (Riemsma et al., 2003).

In further developing outcome measures for self-management knowledge, patients and researchers were brought together at the Outcome MEasures in Rheumatoid Arthritis in Clinical Trials (OMERACT) 7 meeting, to discuss the characteristics that make people effective at managing musculoskeletal diseases. This workshop provided a list of the potential characteristics of an effective health consumer (Tugwell et al., 2005), and contributed to the development of a tool to measure health consumer effectiveness. Consequently, the Effective Consumer Scale (EC-17) was developed by the OMERACT initiative to assess the skills and perceptions of patients which are important for them to participate in and manage their health care (Kirwan et al., 2009; Kristjansson et al., 2007). After testing for psychometric properties, the number of items in the questionnaire was reduced from 48 to 17 (see below).

The questionnaire has been found to measure different skills and attributes compared with the Arthritis Self-Efficacy Scale (ASES) and the Patient Activation Measure (PAM) (Hamnes et al., 2010; Kirwan et al., 2009; Kristjansson et al., 2007). It was also found to have acceptable responsiveness after an educational intervention (Santesso et al., 2009). The EC-17 has been translated and culturally adapted into Norwegian with acceptable validity, reliability and responsiveness (Hamnes et al., 2010). The aim of the present study was to adapt the EC-17 culturally into Swedish and to assess the validity, reliability and responsiveness of the questionnaire.

Materials and methods

The translation and validation process included several steps to ensure validity, reliability and responsiveness of the questionnaire based on published recommendations (Beaton et al., 2000; Terwee et al., 2007).

The EC-17

The EC-17 consists of 17 questions, covering five domains: use of health information, clarifying personal priorities, communicating with others, negotiating roles and taking control, and deciding and taking action. Each item is scored on a Likert scale ranging from 0 to 4 (from never to always), and a total sum is calculated and transformed into a 0 to 100 scale (worst to best score) (Santesso et al., 2009).

Cultural adaptation and translation process

Four experienced health professionals and one patient representative independently translated the questionnaire. Consensus was then reached among the translators and a back translation was performed by a native English-speaking health professional and professional translator. The two English versions were compared, and after a second consensus in the group, the Swedish version of the questionnaire was pre-tested in 15 patients with inflammatory arthritis who were attending a specialist clinic. The patients were asked to complete the questionnaire and to comment and provide feedback on the questions included and on interpretability. No changes were suggested, and the patients gave the questionnaire face validity and supported its content validity.

Subjects and intervention

Two different groups of patients with inflammatory rheumatic diseases participated in the validation study of the translated questionnaire. The first group of patients included (GI) came from an ongoing randomized controlled trial (RCT) studying nurse versus physician care in patients with a stable disease activity score who were treated with biologic therapy at a specialist rheumatology clinic. GI patients filled out the EC-17 twice, one to two weeks apart, either before entering or after finishing participation in the RCT, to establish the test–retest reliability of the questionnaire. The second group (GII) consisted of consecutively enrolled patients who were participating in a five-day arthritis self-management education course at a rheumatology specialist clinic. GII patients filled out the ED-17 at admission and at discharge, to test the responsiveness of the questionnaire.

Self-management intervention

The education course was provided over five consecutive days for six hours per day. A multidisciplinary team (nurse, physician, physiotherapist, occupational therapist and social worker) provided the education for groups of four people, who were educated theoretically and practically. The theoretical education was aimed at informing them about the natural history of the disease, pharmacologic/non-pharmacologic treatment, the benefit of healthy diet and physical activity, how to manage living with a chronic disease and the health care system. In practical terms, the course participants learned how to exercise in order to maintain a good physical capacity (mobility, muscle strength and aerobic capacity), as well as to how to use devices and orthoses to prevent joint damage, in order to manage work life and the full range of other activities of daily living.

Psychometric properties and statistical analysis

Data from all subjects were analyzed for floor and ceiling effects and missing data in the items. During the development of the EC-17 scale, items with ≥5% missing response data were reconsidered, so we also used this as our limit (Kristjansson et al., 2007). A summary score cannot be calculated if more than one item in a subscale is left blank or if more than three items in the questionnaire are missing (Santesso et al., 2009). Floor and ceiling effects should be present in <15% of the responders (Terwee et al., 2007). Both groups also contributed baseline data for the analyses of internal consistency, determining reliability obtained from a single application. Internal consistency was assessed by Cronbach's alpha, and an alpha between ≥0.80–0.95 is generally regarded as acceptable (Streiner and Norman, 1995). An item-total correlation of 0.4 or higher has been recommended if sample sizes are small (Streiner and Norman, 1995).

Test–retest assessment was based on data from patients participating in the RCT study only (GI). Test–retest reliability was assessed by using the intraclass correlation coefficient (two-way random effects model, ICC2.1), for which a coefficient of 0.85 has been suggested as an acceptable limit for individual comparisons, while for groups a coefficient of 0.70–0.75 is acceptable (Rosner, 1995; Terwee et al., 2007).

Patients participating in the self-management intervention (GII) filled out the EC-17 and the ASES and its subscales of pain and other symptoms (Lorig et al., 1989) at admission and at discharge for construct validity and responsiveness purposes. No other questionnaires of this type were available in Swedish at the time. The construct validity of the EC-17 questionnaire was assessed by Spearman's rank correlation and hypothesized a priori (Terwee et al., 2007). We hypothesized medium correlations (r = 0.3–0.6) between the EC-17 and ASES subscales (Bland and Altman, 1999). Change as a result of the intervention was assessed by the Wilcoxon signed rank test; to test for responsiveness, the SRM (calculated as mean score change divided by the standard deviation of the mean score change) was calculated. An SRM is considered low if it is <0.5, moderate at 0.5–0.8 and large at >0.8 (Beaton et al., 1997).

Data from some variables were not normally distributed, and as the data were ordinal (Likert scale), non-parametric analyses were used, although, to allow comparison with earlier validity studies on the EC-17, the mean (SD) is also presented. SPSS version 18.0 (IBM SPSS Statistics, version 18.0, New York, United States) was used for analyses.

Ethics

The ethics committee approved the study No. 2007/10. All subjects gave informed consent to participate in the study.

Results

The GI group included 74 patients, of whom 29 (39%) were male; the mean age was 58 (SD 12) years, and the disease duration ranged from 0 to 39 years. The GII group consisted of 50 patients, mean age 54 (SD 13) years, and a disease duration ranging from 0–26 years (see Table 1).

Table 1. Patient descriptive

	Group IN = 74	Group IIN = 50
Female/male, n (%)	45/29 (61/39)	38/12 (76/24)
Age, years
mean (SD)	57.5 (11.6)	54.2 (12.8)
median (range)	59.5 (54.0)	56.0 (52.0)
Disease duration, years
mean (SD)	15.4 (9.3)	3.0 (5.6)
median (range)	14.0 (39.0)	1.0 (26.0)
Diagnoses, n (%)
Rheumatoid arthritis	53 (72)	39 (78)
Spondyloarthritis	18 (24)	9 (18)
Polyarthritis	2(3)	2 (4)
Inflammatory systemic diseases	1(1)

The GI group had a longer disease duration (p < 0.001) compared with the GII group (see Table 1); it also had a higher (i.e. ‘better’) EC-17 mean total score (p = 0.004) and a consistently higher mean score in all items (p ≤ 0.05) except for items 2, 8, 9 and 14, for which no statistical differences were found (Table 2). The mean item score was the lowest for both groups for item 13, ‘I feel a sense of control over my disease’ (GI: 3.15, SD 0.63 and GII: 2.56, SD 1.03) and the highest for item 9, ‘I have built an open and trusting relationship, based on mutual respect, with my health care providers’ (GI: 3.61, SD 0.52 and GII: 3.38, SD 0.73) at baseline.

Table 2. EC-17 item and raw score†, mean (SD), frequency and item total correlation for group I (GI, n = 74) and II (GII, n = 50)

Item	G	Mean (SD)	Never %	Rarely %	Some-times %	Usually %	Always %	Item-total corr‡
† 0–4 scale (0 = never and 4 = always). ‡ Item–total correlation.
1. I know who can help me judge the quality of the information I receive about my disease	I.	3.47 (0.56)	0	0	2.7	47.3	50.0	0.58
	II.	2.88 (1.1)	4	8	18	36	34	0.76
2. I understand the information I receive about my disease	I.	3.36 (0.49)	0	0	0	63.5	36.5	0.64
	II.	3.30 (0.58)	0	0	6	58	36	0.60
3. I know how to adapt general health information to my own situation	I.	3.34 (0.50)	0	0	1.4	63.5	35.1	0.65
	II.	2.92 (0.88)	2	4	18	52	24	0.73
4. I can be clear about what is important in my life when I make decisions about my disease	I.	3.39 (0.54)	0	0	2.7	55.4	41.9	0.52
	II.	3.08 (0.78)	0	4	14	52	30	0.76
5. I can weigh the pros and cons of a decision about my disease	I.	3.34 (0.56)	0	0	4.1	58.1	37.8	0.67
	II.	3.00 (0.67)	0	4	10	68	18	0.74
6. I can set realistic goals about the management of my disease	I.	3.19 (0.77)	0	2.7	6.7	59.5	31.1	0.60
	II.	2.90 (0.84)	0	8	16	54	22	0.70
7. I can express my concerns well to health care providers	I.	3.38 (0.72)	1.4	0	5.4	45.9	47.3	0.72
	II.	3.14 (0.7)	0	0	18	50	32	0.61
8. I know how to ask good questions about my health and my disease	I.	3.16 (0.78)	1.4	0	14.9	48.6	35.1	0.72
	II.	3.12 (0.63)	0	0	14	60	26	0.50
9. I have built an open and trusting relationship, based on mutual respect, with my health care providers	I.	3.61 (0.52)	0	0	1.4	36.5	62.1	0.59
	II.	3.38 (0.73)	2	0	4	46	48	0.72
10. I am able to play the role I want to in my health care team	I.	3.57 (0.50)	0	0	0	43.2	56.8	0.59
	II.	3.28 (0.78)	0	4	8	44	44	0.75
11. I know who to work with to meet my health needs	I.	3.38 (0.57)	0	0	4.1	54.1	41.8	0.66
	II.	2.94 (0.99)	4	2	22	40	32	0.83
12. I can be assertive to get what I need to meet my health needs	I.	3.28 (0.73)	0	2.7	8.1	47.3	41.9	0.65
	II.	3.04 (0.75)	0	0	26	44	30	0.46
13. I feel a sense of control over my disease	I.	3.15 (0.63)	0	1.4	9.5	62.1	27.0	0.49
	II.	2.56 (1.03)	4	12	24	44	16	0.70
14. I feel confident in making decisions about my health	I.	3.34 (0.67)	0	1.4	6.8	48.6	43.2	0.65
	II.	3.10 (0.74)	0	4	10	58	28	0.67
15. I can negotiate with others about what we need to do to manage my disease	I.	3.26 (0.83)	0	5.4	8.1	41.9	44.6	0.74
	II.	2.88 (0.98)	2	10	12	50	26	0.72
16. I can negotiate with the health care system about what to do to manage my disease	I.	3.47 (0.53)	0	0	1.4	50	48.6	0.69
	II.	2.98 (0.96)	4	2	16	48	30	0.88
17. I can organize my life to act on decisions about how to manage my disease	I.	3.28 (0.61)	0	0	8.1	55.4	36.5	0.70
	II.	2.90 (0.81)	2	0	26	50	22	0.85
Total score converted to 0–100 scale (worst to best)	I.	83.8 (10.6)
	II.	75.6 (15.5)

No single item had >5% missing data. Item 1 had 2.7% missing data in GI, while item 11 had 4% missing data in GII at discharge. In total, the highest proportion of missing data was present in GI at retest (1.4%), but it was possible to summarize a total score for all but two subjects at the retest (Table 3).

Table 3. Missing data, floor and ceiling effects of the EC-17 questionnaire

EC-17	Outpatients group I (test–retest)		Outpatients group II (responsiveness)
	Test N = 74	Retest N = 74	Admission N = 50	Discharge N = 49
† Total score was only possible to calculate for n = 72. ‡ Based on 17 items × n patients in the group.
Total score†
mean (SD)	83.8 (10.6)	83.2 (10.8)	75.6 (15.5)	79.2 (11.7)
min - max	60–100	56–100	25–100	56–100
median, range	82.4 (39.7)	80.9 (44.1)	77.9 (75)	77.9 (44.0)
Missing data‡
item, n (%)	5/1258 (0.4)	18/1258 (1.4)	3/850 (0.4)	0
Floor; ceiling
total scale, n (%)	0 ; 7 (9%)	0 ; 8 (11%)	0 ; 2 (4%)	0 ; 3 (6%)

Floor and ceiling analyses met the requirements (present in less than 15% of the respondents), with ceiling effects varying between 4% and 11% but demonstrating no floor effects at any time (Table 3).

Construct validity with the ASES subscale pain and other symptoms was a priori hypothesized to be in the range of 0.3–0.6. This was true for the ASES pain subscale (r_s = 0.59) but the correlation for the ASES subscale other symptoms and the EC-17 was higher than hypothesized (r_s = 0.75); (p = 0.001 for both subscales) (Table 4).

Table 4. Responsiveness and construct validity for the EC-17 and ASES after the five-day self-management intervention, n = 49

Scale	Baseline score		Change score		SRM§	Spearman r with EC-17
	mean (SD)	median (range)	mean (SD)	median (range)
† EC-17 scale 0–100 (worst to best). ‡ SES scales, 10–100 (worst to best). § Standardized response mean (mean score change divided by the standard deviation of the mean score change).
EC-17†	75.5 (15.6)	77.9 (75)	3,7 (8.6)	2.9 (54.4)	0.43	1.00
ASES pain‡	57.3 (22.1)	56.0 (80)	9.9 (19.7)	10.0 (96)	0.51	0.59
ASES other symptoms‡	61.4 (19.5)	66.3 (85)	5.6 (15.2)	3.3 (61.7)	0.37	0.75

The internal consistency of the EC-17, assessed by Cronbach's alpha, was 0.93 for GI and 0.95 for GII. In the GI group, item-total correlation were all above 0.4 and ranged from 0.49 for item 13, ‘I feel a sense of control over my disease’, to 0.74 for item 15, ‘I can negotiate with others about what we need to do to manage my disease’. In the GII group, item 12, ‘I can be assertive to get what I need to meet my health needs’, had the lowest (0.46), and item 16, ‘I can negotiate with the health care system about what to do to manage my disease’, the highest (0.88) total-item correlation (Table 2).

The ICC_2.1 of the EC-17 Swedish version was 0.94. The time between test and retest varied between three and 16 days for 64 of the 74 subjects; questionnaires with shorter (n = 1) or longer (n = 7) time frames between the test and the retest date were omitted from the analysis. Two subjects left too many items blank, making a sum score impossible to calculate at the retest. Summary scores for the 64 subjects included in the retest analysis had a median of 80.9 and ranged from 55.9 to 100 (mean 83.5; SD 10.2).

Change as a result of the intervention was possible to analyse, based on 49 subjects, and we found a difference between the admission and discharge EC-17 scores, as well as in ASES pain and other symptoms scores(p = 0.005 [EC-17]; p = 0.001[ASES pain]; p = 0.02 ]ASES other symptoms]). EC-17 had an SRM of 0.43, displaying a small change as a result of the five-day intervention. For the ASES pain and other symptom subscales, the SRM was 0.51 and 0.37, respectively (Table 4).

Discussion

The EC-17 questionnaire was developed and endorsed by the OMERACT for patients with musculoskeletal diseases, to assess the skills and perceptions which are important for participating in and managing health care. In the present study we culturally adapted the questionnaire into Swedish, and found that it met the requirements for the psychometric properties tested.

The EC-17 is valid and reliable in both the English and the Norwegian language, and our findings have large similarities with earlier results. The Swedish EC-17 questionnaire had few missing items, no problems with floor and ceiling effects, and a Cronbach's alpha in line with earlier validation studies (Hamnes et al., 2010; Kristjansson et al., 2007; Santesso et al., 2009). Item correlations were similar to those of the Norwegian sample, where they ranged from 0.59 to 0.82, and the ICC of 0.94 was in line with Norwegian findings, which had an ICC of 0.90 (Hamnes et al., 2010).

The Swedish questionnaire also showed an acceptable responsiveness in accordance with the Norwegian study. The interventions provided in Sweden and Norway consisted of a five-day educational self-management programme, and they yielded similar responsiveness, with a small SRM of 0.43 and of 0.48, respectively (Hamnes et al., 2010). In the validation study of the English version of EC-17, the SRM was 0.34 after a six-week arthritis self-management programme intervention, while the ASES subscales pain and other symptoms had a higher SRM (0.65 and 0.54, respectively) (Santesso et al., 2009). In the Swedish sample, the ASES had an SRM of 0.51 (pain) and 0.37 (other symptoms), similarly to the EC-17. Differences in responsiveness between the English and the two Scandinavian versions of the questionnaire might be as a result of differences between the translated questionnaires, the educational interventions or to the different mix of diagnoses in the samples.

We did find some other differences between the countries, concerning the psychometric properties of the EC-17 questionnaire. According to Terwee et al. (2007), in testing for construct validity, the hypothesis has to be set a priori, and we hypothesized correlations between the ASES subscales and the EC-17 to be in the range r_{s =} 0.3–0.6. This was true for the ASES pain scale and the EC-17 (r_s = 0.59), but we found a higher correlation between the ASES subscale other symptoms and the EC-17 (r_s = 0.75). In the English validation study, correlations between the corresponding ASES subscales and EC-17 were lower, at 0.25 and 0.27 (Santesso et al., 2009). The ASES subscale other symptoms comprises six items concerning handling and coping with fatigue, pain, depressive symptoms and the impact of disease on daily living. Santesso et al. (2009) found some overlap of items between the EC-17, the ASES and the Patient Activation Measure but concluded that most items in the EC-17 were not covered by the two other questionnaires. According to the present findings, it seems that reporting high self-efficacy in managing arthritis symptoms implies that the person also trusts him/herself to manage individual health care fairly well. The reason for the differences between the studies concerning both construct validity and responsiveness might be down to the translation of the two questionnaires or to a different focus in each of the interventions. The five-day intervention in Sweden focuses on managing arthritis problems, as well as on understanding the health care system. However, this should be further explored, to understand fully the potential overlaps between the questionnaires.

In the present study, we found a consistently higher mean value for both groups in all items and in the total score of the EC-17 compared with the other studies. We also found a difference between the two Swedish samples. The GI group, which included patients with stable inflammatory rheumatic diseases with a long disease duration, had consistently higher scores compared with the GII group, which had a shorter disease duration and included patients eligible for a self-management intervention. One possible explanation may be that patients in the GI group have dealt with the health care system for a long time and have well established contacts with the clinic, resulting in differences in the scores. Nevertheless, the GII group had higher scores compared with the samples in the two other validation studies. One possible explanation is that GII was a more homogeneous group, consisting of 78% with a diagnosis of rheumatoid arthritis (RA), compared with 17% (Hamnes et al., 2010) and 30–36% (Santesso et al., 2009), in the other studies. In the present study, we did not include subjects with osteoarthritis or fibromyalgia, which represented the largest groups in the two other studies, and hypothetically this might have affected the mean scores or the responsiveness of the questionnaire. Further studies are needed to explore whether patients with inflammatory rheumatic diseases treated in specialist care facilities report higher skills in managing their health care in general, compared with patients with fibromyalgia or osteoarthritis, or with patients treated in primary care. The level of education was not correlated with EC-17 scores in the Norwegian study (Hamnes et al., 2010), which was supported by post hoc analysis of the present data (r_s = 0.03).

In the present study, the main limitations concern the relatively small sample size. According to Terwee et al. (2007), groups of 50 subjects are the smallest recommended number for the tests performed. In our responsiveness test, we had 49 subjects participating, but the change was statistically significant, at a p-value of 0.005, so we suggest that this sample size was appropriate.

Moreover, our subjects had inflammatory rheumatic diseases, including mostly RA, but the EC-17 questionnaire was developed for all musculoskeletal diseases, so further validation studies should include a wider range of diseases. In the present study, all subjects were involved in specialist care, while many patients with musculoskeletal diseases utilize primary care, and the questionnaire should be validated for people using different levels of care. Worldwide, there are plans to use the EC-17 in studies of health equity in vulnerable populations (Rader et al., 2011).

The importance of a questionnaire such as the EC-17 has been supported by patients attending the OMERACT meetings, and a proposed online decision aid inclusive of the EC-17 categories is under development (Santesso et al., 2009). Cultural adaptation of questionnaires makes comparisons and collaboration among countries possible; the lack of studies using the same outcome measures has been put forward by several authors (Hagel et al., 2011; O'Donnell et al., 2010). During the implementation of the present study, a new revised English version was proposed, to improve clarity and understanding (available at http://www.cgh.uottawa.ca/eng/effective_consumers.html).

The new improved version of the EC-17 affects a few of the Swedish questions. The order, but not the meaning, of the first three questions has been changed, and questions 10, 13, 14 and 16 might also need some correction in Swedish, in order to be comparable with the new English version. A new translation including face and content validity will be undertaken in Sweden, in order to align the Swedish version with the new English version.

In conclusion, the Swedish version of the EC-17 met the required recommendations for face validity, internal consistency and test–retest reliability. The questionnaire showed responsiveness to a five-day educational intervention in self-management. Its construct validity needs to be further tested, as the overlap with the ASES symptom subscale was higher than hypothesized. However, psychometric testing of a questionnaire is an ongoing task, and the questionnaire needs further testing in different groups of patients and in different interventions.

Acknowledgements

This work was supported by grants from the Swedish Rheumatism Association.