Data were obtained from the FOS. The inclusion criteria were patients of both sexes with a confirmed diagnosis of FD, who were on, or were candidates for, ERT with agalsidase alfa. The exclusion criteria were patients who were treated with an ERT other than agalsidase alfa or were actively enrolled in blinded trials so not eligible for enrolling in the FOS. FOS was performed in accordance with the recommendations of Declaration of Helsinki. Where necessary, participating centres followed their respective national and/or local regulations regarding Ethics Committee/Institution Review Board requirements.
An internet-based application was used for entering data into the FOS database. Data were entered into the database by the responsible FOS physician or designee. The patients self-reported the survey instruments including the FPHPQ, and responses were entered into the database through the internet application. The cut-off date for this analysis was April, 2010.
Patients were followed as long as the physician found it appropriate. Only data collected during normal routine examinations were requested. At screening/baseline visit, age and gender, comorbidities, concomitant medication, and FOS - Mainz Severity Score Index (FOS MSSI)[10, 11] were collected. FPHPQ and other patient reported outcome (PRO) instruments were administered at subsequent visits. Given that FOS is an international, multi-centre, open survey, the PRO questionnaires were available to participating patients in their respective languages for most of the countries. Because FD is a rare disease, it was necessary to pool data of multiple languages to be able to include patients with a wide range of disease severity and to ensure a sufficient sample size for data analysis. Taking into consideration the language difference, efforts have been made to harmonize these questionnaires so that answers can be interpreted in international analyses. In addition, the translations were reviewed by bi-lingual clinical experts with particular scrutiny of wording appropriate for children. These patient-completed questionnaires are briefly described below.
Fabry-specific Paediatric Health and Pain Questionnaire (FPHPQ)
The FPHPQ assesses disease specific symptoms such as sweating, pain, dizziness and tiredness, heat and cold intolerance, swollen eyelids, gastrointestinal symptoms, feeling thirsty, difficulty hearing, ringing or buzzing noise in the ears, and ability and enjoyment to participate in sports. The FPHPQ measures the frequency of each specific symptom using a 5-point Likert scale (always, often, sometimes, seldom, and never), plus one item that measures the pain intensity using a 0-10 numeric rating scale. In addition, two items require numeric responses about the number of times experiencing onset of pain and the number of school days missed. Finally, there are two items with ‘yes’ or ‘no’ response asking about difficulty in hearing and other problems not mentioned. There are three separate age-specific versions for children aged 4-7, 8-12 and 13-18 years with questions that are phrased age appropriately. The parents of children aged 4-7 complete the FPHPQ; children aged 8 and above complete the questionnaire themselves. Translations of the FPHPQ are available in Spanish, Swedish, English, Norwegian, French, Dutch, Italian, and German.
FOS - Mainz Severity Score Index (FOS MSSI)
The MSSI is a clinical scoring system developed to assess the severity of signs and symptoms of FD and to monitor the progress of individual patients during ERT. The FOS MSSI is an adaptation of the MSSI scoring system for use in FOS and has been shown to be a useful and valid tool to evaluate disease severity and progression in adult patients with FD[10, 11].
The Brief Pain Inventory (BPI)
The BPI questionnaire is a validated tool developed to capture the “sensory” dimension of pain on a 0–10 numerical rating scale. It includes four pain items to capture the variability of pain over time and seven items assessing the “reactive” dimension of pain (interference with daily function)[12, 13]. Initially developed to assess pain related to cancer, the BPI has shown to be an appropriate measure for pain caused by a wide range of clinical conditions including FD[14, 15].
The KINDL questionnaire is a generic validated questionnaire developed for use in children and adolescents to evaluate the impact of health conditions on everyday living and quality of life (QOL)[16, 17]. The KINDL questionnaire comprises 24 items using a 5-point Likert scale (never, seldom, sometimes, often, all the time) and includes six subscales depicting physical well-being, emotional well-being, self-esteem, family, friends, and everyday functioning (school or nursery school/kindergarten).
The EuroQol valuation instrument, or EQ-5D, is an international standardized instrument designed to measure health and to allow the elicitation of patient and/or general population preference values for a wide range of standardized health states.
The EQ-5D self-report questionnaire defines health in terms of five dimensions: mobility, self-care, usual activities, pain or discomfort, and anxiety or depression. It also includes the respondents’ perception of their overall health on the visual analogue scale (EQ VAS), where 0 and 100 denote the worst and the best health states, respectively. The EQ-5D has been used in previous FD studies[20, 21].
Statistical analysis: item analysis, reliability, validity, and responsiveness
A battery of psychometric analyses, described below, was performed to a) identify poorly performing items and b) to assess the measurement properties of the instrument once these items were excluded. All analyses were conducted in several iterative steps with the use of Stata/MP Ver. 11.0, with the exception of the factor analysis which was performed using Mplus Ver. 6.0.
First, the distributional characteristics of each of the items (e.g., % missing, % at floor and ceiling, skewness and kurtosis) were reviewed to identify poorly performing items. Items were flagged for potential exclusion if the distribution of responses was highly skewed or kurtotic or they showed a floor or ceiling effect (minimum/maximum response > 67% of patients). Exploratory factor analysis (EFA) was also used as part of the initial item analysis to help identifying overlapping or redundant items. Items with factor loadings <0.34 (i.e., less than 10% of variance explained) on their primary subscale were flagged for possible deletion. Yes/no items were excluded as these violate the distributional assumptions of EFA. The final decision of item inclusion or exclusion was based on all of the psychometric analyses, and input from the clinical and instrument development experts.
After the exclusion of items from above, a follow-up EFA was conducted to investigate the dimensionality of the items retained. The purpose of the follow-up EFA was to identify items that cluster into conceptually related subscales as well as items that lacked or had weak relationships with the other items and did not belong to a specific subscale. Three types of rotation were consistently used to help interpret simple structure (quartimax and varimax, which are orthogonal rotations; and oblique, which allows correlations between subscales). A scoring algorithm was developed following the final selection of the items to be included in the instrument.
Next, the psychometric properties of the FPHPQ subscales that emerged from the follow-up EFA were examined. The psychometric analyses included descriptive statistics of the subscales and their internal consistency reliability, test-retest reliability, validity and responsiveness to change. The internal consistency of the FPHPQ subscales was assessed using Cronbach’s formula for coefficient alpha, and whenever Cronbach’s alpha exceeded 0.70 it was considered having good internal consistency reliability. The analysis for test-retest reliability of the FPHPQ involved calculating the intraclass correlation coefficient (ICC) based on data of those who reported no change on the patient global impression of change (PGIC) from first to second visit. An ICC of >0.70 among stable subjects is considered acceptable to demonstrate test-retest reliability.
Validity refers to the extent to which the instrument measures what it is intended to measure. To demonstrate construct validity, the score of an instrument should be correlated with scores of other validated instruments that measure similar concepts. The construct validity of the FPHPQ subscales was evaluated through correlation with the scores of the criterion measures BPI, KINDL, and EQ-5D. Rank order correlations were calculated and were expected to be ≥ 0.30 and statistically significant. Known-group validity is the extent that the average scores of the FPHPQ are significantly different among groups of FD patients with different level of disease severity. Analysis of variance (ANOVA) was used to examine the FPHPQ scores by the FOS MSSI grades.
Responsiveness (or ability to detect change) refers to the extent that the score of an instrument reflects the changes in the patient’s condition. The change of the score should be in sync with the direction of the condition: improving, worsening or unchanged. Patients with improving condition should have significantly better scores than the patients with worsening or unchanged condition. The magnitude of the score change can also be used to guide score interpretation. Change measured as standardized effect sizes with 0.20 denotes a small and not clinically relevant change; 0.50 denotes a moderate and clinically relevant change; and 0.80 or above denotes a large and clearly relevant change. A preliminary assessment of responsiveness of the FPHPQ was made by examining the amount of change in FPHPQ subscale scores between responders versus non-responders with responder defined as EQ-5D score increased 0.1 point from first to second visits, and separately with responder defined as BPI average pain and worst score rating decreased 20%, also from first to second visits.