Methods

Following an SCP, experts in AS from different geographical areas integrated evidence from preceding studies into ICF Core Sets during a 3 day consensus conference (online supplementary figure).

Expert rheumatologists were selected from the full member list of the Assessment SpondyloArthritis international Society (ASAS), assuring representation of all world regions. ASAS members have an established record in the field of AS as membership requires a dedicated interest in spondylitis and a minimum of publications in the field.14 Non-rheumatologist expert health professionals were recommended by ASAS members. In line with the standardised approach for all ICF consensus conferences, it was intended to establish three working groups, each consisting of five to seven participants. This number assures representation of the different health professionals and geographical areas and enhances the possibility of contribution of each individual in the discussions.

Preceding studies included qualitative studies to identify which aspects are typical and/or relevant for functioning in AS according to health professionals (Delphi exercise), patients (focus interviews) and outcome research (literature review).11 The concepts revealed in these steps were linked to the closest ICF category. In addition, empirical data collection with the ICF checklist among patients provided quantitative information on the relevance of the category for patients.12 Overall, 374 ICF categories at the second, third and fourth ICF level were identified in the preceding studies with 93 (24%) categories on Body functions, 58 (15%) on Body structures, 146 (38%) on Activities and participation and 77 (20%) on Environmental factors.

After the training in the ICF, participants were asked to read the compiled categories from the preceding studies. For each second level category, they were asked to give an individual vote (yes/no) as to whether this category was relevant and/or typical for AS and should be included in the C-ICF-CS. Subsequently, the SCP was started, which involved alternating group and plenary discussions. The Appendix (online supplementary figure) explains in detail the steps and the agreement rules for inclusion and exclusion of the categories. In part I categories from the preceding studies for the C-ICF-CS and in part II categories for the B-ICF-CS were selected. For the C-ICF-CS categories were selected from the list of categories obtained from the preceding studies. The first cycle was intended to decide on inclusion or exclusion of the second level categories and the second cycle to decide on inclusion or exclusion of the more specific third or fourth level categories. Selection of categories for the B-ICF-CS was limited to categories included in the C-ICF-CS and involved individual ranking followed by plenary agreement on the number of categories for each component to be included. For the C-ICF-CS, participants were instructed to choose as few categories as possible but as many as necessary for global assessment. For the B-ICF-CS participants had to consider the minimum number of categories important for clinical trials.

Results

Nineteen experts (13 rheumatologists, three physiotherapists, one specialist nurse, one occupational therapist and one psychologist) from 12 different countries (12 from Europe, four from North America, one from Asia and one from Mexico and South America each) attended the consensus conference that was held in September 2007 in Nottwill, Switzerland. Each working group comprised rheumatologists, one physiotherapist and one other health professional. Each working group was chaired by a person from a different world region (IEvdHB, Europe; WM, North-America and FH, Asia). The plenary discussions were facilitated by the coordinator (JB).

Comprehensive ICF Core Set

Tables 1–4 show the ICF categories included in the C-ICF-CS. The 80 categories comprised two categories at the first level (chapters), 54 categories at the second level, 12 at the third level and 12 at the fourth level. The 56 higher level categories (first and second level) were made up of 12 (21%) categories from the component Body functions, seven (13%) from Body structures, 24 (43%) from Activities and participation and 13 (23%) from Environmental factors. When combining the point of decision on inclusion in the table with decision process described in the Appendix, it is possible to have an approximate impression of the level of agreement for each specific category.

Within the Body functions, the majority of identified categories belong to chapter 7 Neuromusculoskeletal and movement-related functions and chapter 2 Sensory functions and pain. In the plenary discussions it was noted that b6601 Functions related to pregnancy was identified as relevant in the Delphi by 84%. In a final vote this category was not selected, favouring the opinion of those who felt frequency of impairments in pregnancy were insufficiently examined to decide whether this is typical. Similarly, several experts reminded the group that patients were more susceptible to infections or allergies (b435 Immunological system functions). In the final vote this category was not selected, favouring the argument that this susceptibility is probably subjective, and not supported by research evidence. Further, the category b1801 Body image was discussed. Some argued that patients might suffer emotionally from their altered posture. The final vote favoured those stating that the body image of patients with AS is realistic and not distorted and that the emotional aspects of this “true” change are reflected in b152 Emotional functions.

Of the seven categories at the second level of the component Body structures, five belonged to chapter 7 Structures related to movement. Most categories were further specified into more specific categories. The plenary discussion concentrated first on the relevance of enthesitis, osteoporosis and influence of muscle composition. This resulted in the inclusion of the category s770 Additional musculoskeletal structures related to movement with the three specifications s7700 Bones, s7702 Muscles and s7703 Extra-articular ligaments to cover these aspects of the disease. Further the point was raised that “postural change” is a typical structural impairment with relevant impact on functioning but not represented by a specific ICF category. It was decided to add this higher level structural impairment (structural impairment not covered; s-nc) to the component Body structures.

Of the 24 categories of the component Activities and participation most belong to chapter 4 Mobility (seven categories) and chapter 5 Self-care (five categories). In the plenary discussions there was split opinion on the inclusion of d240 Handling stress. Some stated that AS does not impair the possibility to handle stress. The plenary vote favoured the argument that there is increasing evidence that acute and chronic inflammation alters the stress response and therefore the disease itself poses an additional challenge to handling stress.

The component Environmental factors was represented by 13 categories at the first or second level of the ICF hierarchy. Most belong to chapter 1 Products and technologies (six categories) and chapter 5 Services, systems and policies (five categories). In the plenary discussion it was specified that the category e580 Health services, systems and policies covers the concept “delay in diagnosis”, which had been raised in the Delphi among experts as well as during focus interviews with patients as an important barrier. Further, it was decided to include chapter 3 Support and relationships and chapter 4 Attitudes at the chapter level without specification into lower-level categories (specifying the group of people whose attitudes and support are relevant such as family, friends, health professionals, people in position of authority, etc) because experts found it difficult to decide whose attitudes and support are most relevant for patients. Finally, experts discussed category e225 Climate. Despite some evidence that working in cold and humid environment hampered worker participation, it was concluded that this was an atypical/uncommon situation and that for overall functioning or health, climate was not a typical (objective) facilitator or barrier.

The asterisks in tables 1–4 denote the categories that would not have been selected based on the individual votes, when accepting a threshold of 75% agreement. This threshold was chosen because this was the threshold for consensus of a category for the first working group vote. Nine categories were included in the C-ICF-CS that would not have been selected based on the individual votes only.

Brief ICF Core Set

Of the 56 second level categories from the C-ICF-CS, 19 (35%) were selected for the B-ICF-CS, of which six (32%) were from the component Body functions, four (21%) from Body structures, seven (37%) from Activities and participation and two (11%) from Environmental factors (table 5). No category was specified at the lower level.

Although rheumatologists favoured the inclusion of Seeing functions and Structures of the eye as typical for AS, most other health professionals argued against inclusion since they felt attacks are usually temporary without major impairments and that treatment-resistant uveitis is rare. In the final vote Seeing functions were not selected for the Brief Core Set. Using the same argument, the category Structures of the respiratory cage was discussed and not selected.

Discussion

An SCP integrating expert knowledge on AS and evidence from preceding studies, has led to the definition of the ICF Core Sets for AS. The core sets aim to represent what is typical and relevant for functioning and health in AS. The C-ICF-CS is meant for multidisciplinary assessment and outcome research and the B-ICF-CS for clinical studies. By using the universally accepted language of the ICF and a standardised approach, comparisons across disease will now become possible.13 Since ICF Core Sets will have unique but also shared categories across diseases, this will offer opportunities to specify or generalise intervention programmes or research across disease.

In view of the clinical heterogeneity of AS, experts first agreed that spinal and extraspinal articular disease as well as uveitis were unique and typical manifestations of AS that should be considered when defining the core sets. Other AS-related comorbidities such as clinical inflammatory bowel disease and psoriasis were considered to add to the burden of disease and it was preferred to assess these in separate core sets.

Altogether, 80 categories were selected for the C-ICF-CS and 19 for the B-ICF-CS. Consistent with the consequences of the inflammatory process in the spine, peripheral joints and enthesis, the majority of the categories relate to pain, movement and mobility. In addition, impairments in energy and drive, sleep and emotional functions as well as restrictions in a large number of activities and life situations (participation) were considered important for the core set. Consistent with the bio-psycho-social model of disease, 17% of all selected categories belonged to the component environmental factors. Experts felt strongly about the role of drugs and of support and relationships as reflected by the inclusion of this category and chapter respectively in the Brief Core Set.

The organisers of the conference took great care to invite experts in AS from different professional backgrounds and different countries. Although not an easy task, it proved useful as became clear during the plenary discussions of categories for which no consensus was achieved in the working group votes. For example, physiotherapists had a strong input to reasons for including b740 Muscle endurance functions and d920 Recreation and leisure, the psychologist contributed arguments when discussing b152 Emotional functions, b1801 Body image and d240 Handling stress, the rheumatology nurse added arguments for inclusion of d910 Community life.

Experts enjoyed the SCP and felt they had achieved an important step in defining functioning in AS. One of the main advantages that were mentioned is that it will help in the study of the complexity of functioning in AS (see article on page 108).15 Notwithstanding they also had concerns. The ICF classification and language is not always easy to understand. Also, when selecting categories for the B-ICF-CS, experts encountered difficulties, partly because perspectives of different health professionals diverged, and partly, because it was difficult to judge which categories (within and across components) would actually provide redundant opposed to distinct information about the patients’ abilities. It is clear that the core sets are first versions that will need extensive validation (fig 1) not only with regard to “truth” for the entire spectrum of patients and for all perspectives and geographical regions, but also with regard to reliability and feasibility. In rheumatoid arthritis test–retest repeatability of the C-ICF-CS was moderate with agreement for all categories of the core set in 57% and 59% of patients for inter- and intrarater (2 week interval) reliability, respectively. Repeatability improved for some categories after collapsing the five-level qualifier system into three levels.16 Probably, also the wording of the categories and unfamiliarity of professionals and patients with ICF interviews contribute to the moderate reliability.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Illustration of the process to develop and further validate the International Classification of Functioning and Health (ICF) Core Set for ankylosing spondylitis (AS).

Sensitivity to change of the C-ICF-CS 6 months after the start of disease-modifying antirheumatic drugs in rheumatoid arthritis was moderate and associations with changes in disease activity and other patient reported outcomes were weak.17 It should be noted also that changes in established outcomes in this trial were low.

One of the future directions for the ICF, therefore, is to measure categories by items (and scores) from existing questionnaires. Applying item response theory (IRT), it was possible to constitute a psychometrically sound interval scale when integrating items from several existing instruments that deal with a similar ICF category; in this case example the category b130 Energy and drive functions.18 IRT will also have a role in further validation of the ICF Core Sets to confirm that the categories belong to the same dimension “functioning and health” and identify whether certain categories are redundant. An exploratory study applying IRT to interviews of 111 patients with AS using a preliminary ICF checklist for AS (after exclusion of environmental factors), showed that four categories did not fit the unidimensionality. Category d850 Remunerative employment had a negative misfit, indicating the ability to work does not share the unidimensionality of “functioning” with the remaining categories. The categories (d560 Drinking; d6200 Shopping and d830 Higher education had a negative misfit, which indicates they are too highly correlated or are redundant with other categories. These results need to be confirmed in large patient samples.5 The large number of ICF categories included in the C-ICF-CS, with an assessment time between 30 and 45 min, might be a potential drawback. However, it should be realised that functioning is complex and we should not try to reduce the necessary categories with the risk of ignoring an important building block. Further, computer-assisted testing will help in the future to efficiently measure aspects of outcome in individual patients using only fragments of large computerised item pools.18

In conclusion, the C-ICF-CS for comprehensive classification and the B-ICF-CS for clinical studies in AS are now available. The core sets aim to represent the new reference to define functioning in AS and facilitate clinicians’ and researchers’ efforts to incorporate a patient-oriented, multilevel and comprehensive view on functioning with AS.