Participation and social functioning in patients with fibromyalgia: development and testing of a new questionnaire

To determine the specific participation restrictions FMS patients experience it is worthwhile questioning patients about how they personally experience problems employing qualitative methodology (cf. [5, 20]). In a pre study we examined eight focus groups (N = 38), each containing three to six FMS patients [21]. The pre study was conducted in September and October 2010. The interviews lasted between 60 and 90 minutes and were audio-recorded in digital format. In line with Anatchkova & Bjorner [22], we had two data-collection phases: in the first, we used a half-structured interview guideline, asking the patients to freely describe the participation restrictions they experience from their FMS on a daily basis. The key questions were: ‘In which areas of your daily life does your illness hamper you?’ and ‘How important to you are those areas in your daily life?’ The second phase contained activating elements [23]. Using colored index cards, we illustrated eight areas of participation (covering chapters 6 to 9 in the ICF) with examples, and the patients were to select three areas in which their illness hampered them the most. We thereby tested the relevance of the areas of participation we had theorized that the patients would experience. Once the patients had selected the index cards, the results were discussed in a focus group.

The group discussions were recorded and transcribed, and the contents analyzed by two coders using Atlas.ti software (version 6.2). A coding system was developed in several steps. The final version contained 10 supercategories (i.e., “social activities/recreation/leisure”) and a total of 105 subcategories (e.g., “going to a concert”). The patients experienced especially strong impairment in the following areas of daily life: contact with close family and friends, grandchildren, colleagues, partners, during leisure activities, at work, and when doing housework (for more detail, see [21]). 69 items were generated from the subcategories while we made sure that the contents of the most frequently-coded subcategories were also adequately represented in the items. Of the 69 items (=FPQ version 1), 11 address work and job-training and are only to be answered when the subject had answered “yes” to the question “Are you employed or in a job-training program (i.e., school, apprenticeship, higher education)?”.

The FPQ version 1 was presented to 10 FMS patients in a cognitive interview [24] in which thinking-aloud and verbal-probing techniques were used. The interviews were conducted in May 2011. To reduce the patients’ burden, the FPQ was subdivided into two approximately equally long versions A and B. Five patients were given Version A and five Version B. Their remarks were used to revise and if necessary, omit items. After the cognitive interviews, 47 items remained (FPQ version 2). Of those 47 items, 9 address employment and training. The FPQ contains questions (such as “To what extent did you have difficulty engaging in your hobbies during the last four weeks because of your illness?”) for which there were six possible response categories (no difficulty, mild difficulty, moderate difficulty, severe difficulty, impossible, not applicable).

A summary of the entire study design – from instrument development to psychometric testing – is presented in Figure 1.

To test the FPQ psychometrically, N = 256 FMS patients undergoing inpatient rehabilitation in Germany were surveyed at the beginning of rehabilitation (t0), 3 months (t1) and 6 months (t2) after the end of rehabilitation. The recruitment phase was between August 2011 and September 2012. An interdisciplinary treatment program was carried out during rehabilitation that comprised coordinated psychological, medical, pharmacologic, educational, and physiotherapeutic components. The study’s inclusion criterion was a definitive diagnosis of “fibromyalgia syndrome” (ICD-10; M 79.7). FMS was diagnosed according to the American College of Rheumatology (ACR) guidelines. The study was approved by the ethics committee of the University of Freiburg (approval number 172/10). The percentage of patients that did not fill out the questionnaire was 35.4%. The most important reason for non-inclusion was refusal to participate (70.9%) followed by language difficulties (11.9%). Table 1 provides information on the study patients. The return rate at t1 was 83.6%, and 72.3% at t2. The dropout analysis revealed that the participants at t2 did not differ significantly from dropouts regarding education, gender, and chronification, as well as the FPQ and FIQ values. Only age made a significant difference. The dropout patients were somewhat older (53.4 versus 50.7 years, p = 0.019). We referred to t0 data in all the analyses, adding t1 and t2 data only to determine the responsiveness.

In addition to the FPQ version 2, we administered three other instruments to test construct validity and responsiveness, and to compare the FPQ’s contents with other instruments. The FIQ’s physical scale (German version FIQ-G [25]) contains 10 items and approaches the construct participation and social functioning in that it addresses key daily tasks (i.e., going shopping) and because one item refers to visiting friends and relatives. The Pain Disability Index (PDI [26]; German version: [27]) is a generic measure of pain-related disability that can be used for pain in all body regions. The short instrument contains 7 items addressing participation restrictions in various fields of life (i.e. relaxation, social activities, work) and has been administered often in FMS patients [28, 29]).

The PROMIS® item banks for satisfaction with participation are generic instruments that capture participation and social functioning (differentiating between social roles (SR) and discretionary social activities (DSA)). The two item banks are components of the PROMIS® social health domain framework [30, 31]. Our research group received permission from the PROMIS network to translate the item banks regarding satisfaction with participation into German, and the results of the psychometric assessment in patients with chronic back pain will be published soon [32]. Results of the final 13-item (SR-G) and 10-item (DSA-G) German scales regarding unidimensionality were satisfactory. Both scales are reliable and show good item response model fit and distribution characteristics. Both scales are sensitive to clinical changes, and we observed initial proof of construct validity [32].

We eliminated eight items because of too many missing values. These were generally questions that addressed a participation area that was irrelevant to some individuals (i.e., gardening, activities with grandchildren, dancing, etc.). The subsequent exploratory factor analysis revealed a two-factor solution for the non-job-related items that can be interpreted as: Factor 1: Participation in social life (FPQ-S), Factor 2: Participation in daily life (FPQ-D). The job-related items showed a one-factor solution (participation in work-life FPQ-W).

After the exploratory factor analysis, 32 items remained. To ensure sufficient unidimensionality, three items in the FPQ-S scale were eliminated. No items needed to be eliminated from the other two scales. To ensure the fit to the Rasch model, one item in the FPQ-S scale and one item in the FPQ-W scale had to be eliminated. The final version of the FPQ questionnaire thus contained 27 items (FPQ-S: 11 items, FPQ-D: 11 items, FPQ-W: 5 items). The scale values lie between 0 and 100, with the higher values corresponding to better participation and social functioning.

Table 2 illustrates all FPQ’s psychometric properties. The instrument displays good distribution properties, all the scales are unidimensional and there is only minor local dependence. The FPQ-S and FPQ-D scales fit to the Rasch model very well, the FPQ-W scale’s fit is satisfactory (one item’s outfit value is at 1.46 somewhat too high, that of another at 0.50 too low). The fit to the Rasch model guarantees some methodological advantages [37, 41]: The FPQ person parameters are interval scale measures, person parameters can (as in adaptive testing) also be derived from a portion of the scale items, and the person fit can be analyzed so that inattentive responders or cognitively-challenged patients can be identified. The reliability values are good to very good (with Cronbach’s Alpha ranging from 0.85 and 0.94); none of the items reveals differential item functioning. The corrected item-scale correlations are generally above 0.60 (see Table 3).

Table 3 shows descriptive results on the scale and item level. The scale values illustrate considerable participation restrictions in social interactions and activities in the FPQ-S area as well as in the professional area (FPQ-W) (mean values ranging from 35 to 40). In the FPQ-D scale (daily tasks and responsibilities) we noted considerably fewer participation restrictions (mean value approx. 60). FMS patients have particular difficulty socializing, i.e. going to parties or accepting invitations, going on vacation, and generally with leisure activities and attending cultural events. At work they seem to have trouble coordinating multiple duties at the same time and maintaining the necessary working tempo.

Table 4 illustrates that we proved our hypothesis. The FPQ correlates closely with the four other scales that also capture facets of participation and social functioning. The FPQ-S and FPQ-D scales correlate higher than 0.60 with the other instruments, while the job-related FPQ-W scale correlates at between 0.44 and 0.59. This is most probably because job-related issues are not addressed at all in the other scales and if they are, they are captured only through a few items.

To investigate any differences between the FPQ and other instruments in terms of responsiveness, Table 5 shows the effect sizes 3 and 6 months after rehabilitation, categorized by subgroups. One notes that moderate effects were achieved over the short term (3 months after rehabilitation) that diminished 6 months after rehabilitation. The highest responsiveness was revealed by the FPQ scales (especially the FPQ-W) and the PDI, followed by the SR-G scale. SRM values were lower in the FIQ-G and DSA-G scales.

Our subgroup analyses revealed that the FPQ shows effects in all the subgroups we considered. The effects are less pronounced 6 months after rehabilitation in older persons and in those whose illness is of longer duration. As the other instruments tend to reflect such results also, we cannot claim to have demonstrated group-specific FPQ responsiveness, but rather the influence of general risk factors on the success of interdisciplinary treatment programs in FMS patients.

To determine whether the FPQ covers relevant areas of participation, and the extent to which these differ from those addressed in the other instruments, we linked items in all the instruments to the ICF categories (Table 6). We chose specific and less specific ICF categories according to how specific or general the content was in a given item. We observed that no other instrument contains items from so many different ICF chapters of activity and participation (7 of 9 chapters) as do the three FPQ scales. The PDI measures at a similar breadth (5 of 9 chapters). However, the FIQ (physical scale), SR-G, and DSA-G each contains only items from three ICF chapters. In contrast to the other instruments, the FPQ focuses strongly on the general tasks and demands areas (i.e., “managing your general daily routine (job, housework, private life)”) as well as on community, social and civic life - areas that no other instrument addresses at such a high rate (30% of items). The FPQ is furthermore the only instrument containing a specific scale for participation restrictions addressing work and employment (see [5]). Although the SR-G scale does cover this area with several items, the items do not specify whether the patient is describing housework or paid employment (thus explaining the double allocation in Table 6).