Article Text

Original Article
Validation of the Inflammatory Bowel Disease Disability Index in a population-based cohort
  1. Corinne Gower-Rousseau1,2,
  2. Hélène Sarter1,2,
  3. Guillaume Savoye3,
  4. Noémie Tavernier4,
  5. Mathurin Fumery5,
  6. William J Sandborn6,
  7. Brian G Feagan7,
  8. Alain Duhamel8,
  9. Nathalie Guillon-Dellac1,2,
  10. Jean-Frédéric Colombel9,
  11. Laurent Peyrin-Biroulet10
  12. and the International Programme to Develop New Indexes for Crohn's Disease (IPNIC) group
    1. 1Public Health, Epidemiology and Economic Health, Registre Epimad, Maison Régionale de la Recherche Clinique, Centre Hospitalier Universitaire Régional, Lille Cedex, France
    2. 2Lille Inflammation Research International Center LIRIC—UMR 995 Inserm/Université Lille 2/CHRU de Lille; Equipe «IBD and environmental factors: epidemiology and functional analyses, Lille University, France
    3. 3Gastroenterology Unit, Epimad Registry, Hôpital Charles Nicolle, Rouen University Hospital, Rouen, France
    4. 4Gastroenterology Unit, Hôpital Huriez, Lille University Hospital, France
    5. 5Gastroenterology Unit, Epimad Registry, CHU Amiens Sud, Avenue Laennec- Salouel, Amiens University Hospital, France
    6. 6Division of Gastroenterology, University of California San Diego, La Jolla, California, USA
    7. 7University of Western Ontario, London, Ontario, Canada
    8. 8Univ. Lille, CHU Lille, EA 2694—Santé publique : épidémiologie et qualité des soins, Lille, France
    9. 9Icahn School of Medicine at Mount Sinai, New York, New York, USA
    10. 10Gastroenterology Department and Inserm U954, Nancy University Hospital, Université de Lorraine, Nancy, France
    1. Correspondence to Dr Corinne Gower-Rousseau, Service de Santé Publique, Epidémiologie, Economie de Santé et Prévention, Registre Epimad, Maison Régionale de la Recherche Clinique, Centre Hospitalier Universitaire Régional, CS 70001, Lille Cedex 59037, France; corinne.gower{at}chru-lille.fr

    Abstract

    Background IBDs are chronic destructive disorders that negatively affect the functional status of patients. Recently, the Inflammatory Bowel Disease Disability Index (IBD-DI) was developed according to standard WHO processes. The aims of the current study were to validate the IBD-DI in an independent patient cohort, to develop an index-specific scoring system and to describe the disability status of a well-defined population-based cohort of French patients with IBD.

    Methods From February 2012 to March 2014, the IBD-DI questionnaire was administered to a random sample of adult patients with an established diagnosis of IBD issued from a French population-based registry. The IBD-DI consists of 28 items that evaluate the four domains of body functions, activity participation, body structures and environmental factors. Validation included item reduction and data structure, construct validity, internal consistency, interobserver and intraobserver reliability evaluations.

    Results 150 patients with Crohn's disease (CD) and 50 patients with UC completed the IBD-DI validation phase. The intraclass correlation coefficient for interobserver reliability was 0.91 and 0.54 for intraobserver reliability. Cronbach's α of internal consistency was 0.86. IBD-DI scores varied from 0 to 100 with a mean of 35.3 (Q1=19.6; Q3=51.8). IBD-DI scores were highly correlated with Inflammatory Bowel Disease Questionnaire (−0.82; p<0.001) and SF-36 (–0.61; p<0.05) scores. Female gender (p<0.001), clinical disease activity (p<0.0001) and disease duration (p=0.02) were associated with higher IBD-DI scores.

    Conclusions The IBD-DI has been validated for use in clinical trials and epidemiological studies. The IBD-DI showed high internal consistency, interobserver reliability and construct validity, and a moderate intraobserver reliability. It comprises 14 questions and ranges from 0 to 100. The mean IBD-DI score was 35.3 and was associated with gender, clinical disease activity and disease duration. Further research is needed to confirm the structural validity and to assess the responsiveness of IBD-DI.

    Trial registration number 2011-A00877-34

    • QUALITY OF LIFE
    • INFLAMMATORY BOWEL DISEASE

    Statistics from Altmetric.com

    Significance of this study

    What is already known on this subject?

    • IBDs are disabling conditions that negatively affect physical, psychological, familial and social dimensions of life.

    • The concept of quantifying disability has become established for the evaluation of many other chronic diseases.

    • However, the majority of IBD studies have focused on health-related quality of life (HRQL) assessments in distinction to measurement of disability.

    • Nevertheless, HRQL is subjective and existing instruments such as the Inflammatory Bowel Disease Questionnaire (IBDQ) have not been accepted as valid by the US Food and Drug Administration.

    What are the new findings?

    • We developed the Inflammatory Bowel Disease Disability Index (IBD-DI) according to the WHO classification.

    • The quality of data generated by the IBD-DI was excellent with a maximum 1.5% of missing data per item.

    • The quantitative validation found a good reliability, a good correlation between IBD-DI and Physical SF-36 and IBDQ, a very good inter-interviewer reliability and a moderate intra-interviewer reliability.

    • IBD-DI score ranged from 0 to 91 (possible maximum score of 100) and the mean of IBD-DI was 35.3 (±SD 20.5), and associated risk factors for high IBD-DI scores were female gender, long disease duration and high clinical disease activity.

    How might it impact on clinical practice in the foreseeable future?

    • We described for the first time the disability status in a population-based cohort of patients with IBD using a thorough methodology including all steps (item reduction and data structure, construct validity, internal consistency, interobserver and intraobserver reliability) of quantitative validation.

    • A validated version of the IBD-DI is now available that can be used as an outcome measure in clinical trials and prospective epidemiological studies.

    Introduction

    The chronic IBDs are disabling conditions that negatively affect physical, psychological, familial and social dimensions of life.1 Over the past few decades, the concept of quantifying disability has become established for the evaluation of many other chronic diseases. However, the majority of IBD studies have focused on health-related quality of life (HRQL) assessments in distinction to measurement of disability.2 ,3 Nevertheless, HRQL is subjective and existing instruments such as the Inflammatory Bowel Disease Questionnaire (IBDQ) have not been accepted as valid by the US Food and Drug Administration (FDA).4 Both HRQL and disability assessments are patients-reported outcomes (PROs) that provide important insights regarding the impact of disease activity and severity on patients’ well-being.5 ,6 However, none of the existing IBD PROs were developed according to FDA guidance for PRO development, which is a critical limitation to their use as outcome measures in controlled studies of new therapies.

    Disability indexes have been developed and validated in other chronic progressive diseases, such as the Health Assessment Questionnaire (HAQ) for rheumatoid arthritis and the expanded disability status scale for multiple sclerosis (MS).7 ,8 These instruments have been used in disease-modification trials,9 ,10 which assess the long-term effectiveness of different treatment algorithms. Given that UC and Crohn's disease (CD) are chronic disabling diseases, an international collaboration was initiated, which led to the development according to the WHO classification of the first Inflammatory Bowel Disease Disability Index (IBD-DI), namely the International Classification of Functioning, Disability and Health (ICF).11 As stated by Joseph Sellin, “this ICF framework seeks to examine the multifactor understanding of functioning and disability taking into account a bio-psycho-social perspective”.12 Leong et al13 showed that the IBD-DI may be a valid tool measuring disability in both CD and UC and correlates with workforce participation. However, a robust scoring system and validation in an independent cohort of patients with IBD at the population level were lacking.

    Accordingly, our aims were to validate the IBD-DI in an independent cohort of patients with IBD; to develop a valid scoring system for the IBD-DI to be used in clinical trials and cohort studies and to evaluate using this instrument the disability status of a well-defined population-based cohort of French patients with IBD.

    Patients and methods

    Preliminary steps

    A preliminary step before the administration of the questionnaire was backward-forward translation in French as the instrument was developed in English: this step included forward translations by two expert translators (French native) followed by backward translations by two expert translators (English native) with synthesis of the translators versions after each step by an expert committee as recommended by Beaton et al.14 The IBD-DI was then tested by one gastroenterologist and one IBD specialist nurse in 15 consecutive adult patients with IBD (10 with Crohn's disease and 5 with UC) in March 2011. Good acceptability and lack of ambiguity of the questionnaire was demonstrated. All characteristics of these 15 patients and main results of this step development are detailed in online supplementary table 1.

    Study population

    From 1 February 2012 to 31 March 2014, the questionnaire was administered to a random sample of 200 adult patients with an established diagnosis of IBD for at least 3 months. Included patients were selected from a cohort issued from a population-based registry (EPIMAD). This cohort concerns all adult patients with an IBD diagnosis between 1988 and 2004 and diagnosed before 17. EPIMAD registry is a prospective population-based study that records all incident cases of IBD in northern France. This study population base consists of 5 790 526 persons, which represents 9.3% of the French population.15–18 All patients diagnosed with IBD from 1 January 1988 to 31 December 2004 are recorded from all adult (n=250) and paediatric (n=12) gastroenterologists practising in the private and public sectors. Only residents of the studied areas at the time of diagnosis were included. Each gastroenterologist reported all patients consulting for the first time with clinical symptoms compatible with IBD. The main data collected included age, gender, year of diagnosis, interval between onset of symptoms and diagnosis, and clinical, radiological, endoscopic and histological findings at the time of diagnosis. The final diagnosis of IBD was established by two expert gastroenterologists and was recorded as definitive, probable or possible according to previously published criteria. Only definitive and probable cases were eligible for inclusion.16

    The following disease characteristics were prospectively collected for every patient: disease duration; smoking status (current, former or never); phenotype (disease location and behaviour in CD) according to the Montreal classification;19 extraintestinal manifestations; medical and surgical treatments; and clinical disease activity index assessed by the Harvey Bradshaw Index (HBI) for CD20 and the partial Mayo score for UC.21

    Questionnaire administration

    Patients with IBD have been informed of study's aim by letter from their gastroenterologist (GE). An IBD specialist nurse specifically trained in this questionnaire contacted them by phone to explain the study objectives. Patients who accepted to participate in the study went to the clinical investigation centre of the EPIMAD registry to give their written consent.

    The validation phase consisted of administering this questionnaire during a face-to-face interview by the IBD specialist nurse.

    The published questionnaire contains 28 items. The first 18 items cover different aspects of disability (overall health, sleep and energy, affect, body image, pain, regulating defecation, looking after one's health, interpersonal activities, work and education covering the three domains of body functions, body structures and activities and participation). Eight other items are complementary to these disability items and provide information on how patients environment interact with his/her disease and are considered as facilitators or barriers as recommended by the WHO classification.3 ,11 These eight supplementary items are related to the weight of patients’ environment concerning medications, family, food and health professionals having a positive or negative impact on their disability. The latter two relate to access to social security and access to healthcare system. It is noteworthy that these two questions on healthcare systems were not relevant in France (98% answered that they have a good healthcare). Hence, quantitative validation concerned only the 18 items. To note, before quantitative validation, questions 12 and 13 regarding work and education were merged into one question as question 12 concerned ‘work and household activities’ and question 13 dealt with ‘school and studying activities’ and concerned two different study populations. Hence, the questionnaire used for quantitative validation in the present study was composed of 17 questions.

    In total, 12 out of 17 items were scored on a five-point Likert scale (no limitation, slight limitation, moderate limitation, severe limitation, extreme limitation), 2 out of 17 items required a dichotomous answer (‘yes’ or ‘no’) (questions on arthritis/arthralgia, weight loss), 2 were continuous variables (body mass index (BMI) and number of liquid or very soft stools during last week) and 1 (blood in stool) was divided into three categories (‘none’, ‘little’, ‘a lot’) (online figure 1).11

    Statistical analysis

    Sample

    The sample size was determined to enable the introduction of 17 questions into one principal component analysis (PCA). There is no consensus on the number of patients necessary to run a PCA and a rule of thumb consists of calculating the sample size based on a minimum of subjects by variable. Based on a subject to item ratio of 10:1, 170 subjects are necessary to run a PCA.22 This number was rounded to 200 patients to assess quantitative validation. Included patients were selected randomly from a cohort issued from our population-based registry, including 77% of patients with CD and 23% of patients with UC. Random lists of patients were extracted from this cohort and stratified according to disease type (CD or UC) to have a sample representative of UC in this population of patients with IBD (150 CD and 50 UC). The first 30 included patients answered the questionnaire three times to assess reliability. They answered the questionnaire during a first visit at study inclusion and during a second visit that took place 2–4 weeks later. The second interview was conducted twice, once by the original interviewer to assess intraobserver reliability (test–retest method) and again by a second interviewer to evaluate interobserver reliability. According to Giraudeau and Mary,23 assuming a true intraclass correlation coefficient (ICC) of 0.75 and two measures by patient, there is 94% chance to obtain a lower bound of the two-sided CI for ICC of 0.6. Next 30 included patients with IBD completed IBDQ and SF-36 questionnaires (V.2 Standard, US V.2.0) to assess construct validity.24 ,25

    Item reduction and data structure

    Descriptive statistics were performed for each item. The number of missing data was examined to check that the item was well accepted. In case of high missing answers (>20%), reasons for such non-response had to be explored and item deleted from the questionnaire. Distribution of items was examined to detect items not sufficiently informative. For items based on a Likert scale, the percentage of responses on extreme points was thus examined to detect floor or ceiling effects (>50%). For binary items, the percentage in each response level was examined.

    Identification of the dimensions of the instrument was performed by means of an exploratory PCA as factor structure was never explored after qualitative validation and as this study is the first attempt to validate this index. The principle is that items that correlate highly with each other are clustered in one factor, while items within one factor preferably show a low correlation with items belonging to other factors.26 The number of components was determined by examination of the scree plot of eigenvalues with Horn's procedure (parallel analysis with 20 random simulations).27 A substantial loading of an item on one principal component was considered >0.4.22 Items with factor loading <0.4 were considered for deletion.

    Score construction

    The sum of item values from all selected items was computed to calculate the overall score of the IBD-DI, with a higher score indicating a greater level of disability. All items were rescaled to range from 0 to 4. Binary items were coded as ‘0’ for ‘no’ and ‘4’ for ‘yes’ (weight loss, arthritis, blood in stools). BMI was scored as ‘0’ for a BMI >18.5 and ‘4’ for a BMI <18.5. Number of liquid stools was categorised as follows: 0 were grouped together and other values were categorised into quartiles. In case of <20% of missing items for an individual, the total score was then rescaled from 0 to 100 using the following formula: score×100/(p×4), where p represents the number of answered items. In case of >20% of missing items for an individual, the score is not calculated and the total score is thus missing for that individual.

    The COnsensus-based Standards for the selection of health Measurement Instruments (COSMIN) taxonomy of psychometric properties and definitions for health-related outcomes was used to investigate the following measurement properties of IBD-DI score: reliability and construct validity.26 ,28

    Reliability

    Cronbach's α coefficient and corrected item-total correlations were used to assess internal consistency. Internal consistency refers to the degree of interrelatedness among the items. The corrected item-total correlation is the correlation of a given item in one dimension with the remaining item of that dimension. Cronbach's α with one-at-a-time deletion procedure was used to assess the impact of items on internal consistency. Cronbach's α is expected to exceed 0.7.20 Corrected item-total correlations should exceed 0.4.

    Reliability refers to the degree to which participants can be distinguished from each other, despite measurement error.29 Intraobserver and interobserver reliabilities were assessed by ICC. Intraobserver reliability was based on a one-way random analysis of variance using restricted maximum likelihood method. Interobserver reliability was assessed with the so-called ICC for ‘agreement’, and calculation was based on a two-way random analysis of variance (random patient and observer) using restricted maximum likelihood method. Two-sided 95% CIs for ICC were calculated by bootstrap method with 2000 replicates obtained with replacement. According to Hernaez,29 an ICC <0.40 is considered as ‘poor’, an ICC between 0.40 and 0.59 as ‘fair’, an ICC between 0.60 and 0.75 as ‘good’ and an ICC >0.75 as ‘excellent’. Detailed variance components are presented in online supplementary table as recommended by Hernaez.29 This table allows us to calculate the SE of measurement that relates to how close the scores on repeated measurements are.

    Construct validity

    Construct validity was assessed by the ‘hypothesis testing’ method.26 Construct validity was defined by COSMIN as the degree to which the scores of a measurement instrument are consistent with hypothesis (eg, relationships with scores of other instruments) based on the assumption that the instrument validly measures the construct to be measured.27 We hypothesised that IBD-DI negatively correlates with IBDQ and Physical SF-36. The Pearson coefficients of correlation between IBD-DI and IBDQ and between IBD-DI and Physical SF-36 were computed and are supposed to be >0.4.

    Disability level and associated factors in a population-based cohort

    The IBD-DI score was described for the 200 adult patients with IBD using mean and SD, median, IQR and range. Factors associated with disability level were identified by means of bivariate analyses of variance.

    Statistical analyses were performed using SAS V.9.3 and R 3.0.2 (‘psy’ packages). Statistical significance was considered as p<0.05.

    Results

    Baseline characteristics of the 200 included patients

    A total of 474 patients were contacted to reach a final number of 200 included patients (response rate 42%) without any significant differences between respondents and non-respondents concerning age, gender, IBD type, disease duration, disease location and behaviour at diagnosis.

    Sociodemographic and clinical details of the 200 included patients with IBD are shown in online supplementary table 2. In total, 150 patients had CD (75%) and 50 UC (25%). Eighty-eight (44%) patients were males (47%), with a median age of 26 years (IQR 22–31) at study entry. Median disease duration was 12 years (IQR 3–17). In CD, 92 patients (61%) had inactive disease, 54 (36%) moderate disease and 4 had severe disease according to HBI. In UC, 13 patients (26%) had inactive disease and 37 (74%) mild or moderate disease according to the Montreal classification. None of the patients with UC had severe disease at inclusion (online supplementary table 2). Concomitant medications at the time of interview were 5-aminosalicylic acid (n=42; 21%), systemic steroids (n=5; 2.5%), immunosuppressants (azathioprine or methotrexate; n=46 (23%)) and anti-tumour necrosis factor therapy (n=82; 42%). In CD, 71 patients (47%) had at least one intestinal resection; and in UC, 8 patients (16%) had colectomy, including 4 patients with subtotal colectomy and ileorectal anastomosis and 4 with coloprotectomy with ileoanal anastomosis. Four patients (2%) had a current stoma. At the time of interview, 61 patients (32%) were active smokers.

    Item reduction and data structure

    Item analysis

    The proportion of missing answers per item was low with a maximum of 1.5%, showing good quality of data. Also, 9 out of 17 (53%) items had no missing responses. Ninety-three per cent (n=186) of respondents had no missing answers. A maximum of one missing item (0.5%) was observed in 14 (7%) patients. No ceiling or floor effects (>50%) were observed for questions in five-point Likert scales.

    Factor structure

    All 17 items were entered into a PCA. The scree plot of eigenvalues (figure 1) revealed a predominant eigenvalue. This was confirmed by parallel analysis22 that showed that only one eigenvalue is clearly above what could be expected by chance. Hence, the scale can be considered unidimensional, meaning that items measure a single latent trait or construct. First component explained 33% of total variance. Four items (BMI, weight loss, blood in stool, arthritis/arthralgia) had low correlation with the principal component (<0.4) (table 1). Following these results, BMI, weight loss and blood in stool were removed from the questionnaire. Arthritis was kept in the score as it is known to be a major disabling manifestation for patients with IBD based on expert opinion. We checked that the three removed factors did not correlate with the second component of PCA.

    Table 1

    Factor loadings of principal component analysis computed on the 17 items through the analysis of IBD-DI questionnaires in a population-based IBD study (n=200)

    Figure 1

    Scree plot of eigenvalues from principal component analysis on 17 items through the analysis of Inflammatory Bowel Disease Disability Index questionnaires in a population-based IBD study (n=200). A scree plot displays the eigenvalues associated with each factor in descending order. Point line represents the plot of observed eigenvalues obtained from the 200 patients with IBD included in this study. Plain lines represent plots of eigenvalues obtained from 20 random samples (parallel analysis). Only factors with eigenvalues that are greater than simulated values are interpreted. Here, only the first eigenvalue is clearly above simulated values.

    Score calculation and disability status in a population-based registry using the IBD-DI

    The IBD-DI score was calculated as the sum of the remaining 14 items and rescaled to range from 0 to 100. Score was calculated only when at least 80% of items were answered. In our study, this was true for all included patients. The validated version of the questionnaire as well as the scoring system is depicted in online supplementary figure 2. In the overall IBD population, the mean score of the IBD disability index was 35.3 (±SD 20.5) and ranged from 0 to 91. Distribution of the IBD-DI is presented in figure 2 and online supplementary table S3. Based on IQR of the IBD-DI that was evaluated at the population level, this index can be categorised as follows: 0–20 (no disability), 20–35 (mild disability), 35–50 (moderate disability) and 50–100 (severe disability).

    Figure 2

    Distribution of Inflammatory Bowel Disease Disability Index (IBD-DI) score in a population-based IBD study (n=200).

    Reliability

    The Cronbach's α coefficient calculated on the remaining 14 items was 0.86. One-at-a-time deletion of items indicated that each item contributed to IBD-DI score and confirmed the unidimensionality of the set of 14 items (Cronbach's α with deleted items are between 0.84 and 0.87). All inter-item correlations where under 0.7, demonstrating the absence of redundancy between items. Corrected item-total correlations were all over 0.4 except for arthritis (0.33). Interobserver reliability was very good with an ICC of 0.91 (0.76 to 0.93). Intraobserver reliability was moderate with an ICC of 0.54 (0.10 to 0.69). No patient experienced a significant disease flare leading to a therapeutic intervention between the two evaluations. Variance components are reported in online supplementary table S4, and a complementary analysis of intraobserver reliability is presented in online supplementary table S5.

    Construct validity (hypothesis testing)

    The IBD-DI score correlated with the physical and mental component subscales of the SF-36 (−0.61 and −0.71, respectively) and total IBDQ score (−0.82) (Pearson's correlation coefficients).

    Environmental factors

    Facilitators and barriers showed good quality of data with low item non-response (a maximum of 1% of missing answer per item). Facilitators and barriers scores were calculated as the sum of the four facilitators and barriers items respectively and rescaled to go from 0 to 10. In our population-based study, facilitators varied from 0 to 10 with a mean of 4.8 (±SD 2.5) and barriers varied from 0 to 10 with a mean of 2.5 (±SD 2.3).

    Factors associated with disability level using the IBD-DI

    IBD type did not influence the IBD-DI (p=0.12) with a mean score of 33.9±19.5 in CD compared with 39.2±23.1 in UC (table 2). In the overall IBD population, IBD-DI was associated with gender (p<0.001), disease activity (p<0.0001) and disease duration (p=0.02). IBD-DI was higher for female gender (score of 39.7±20.2 compared with 29.6±19.6 for men) and for active disease (43.5±19.7 vs 27.7±18.4 for inactive disease). Interestingly, 16% (n=17) of patients with inactive disease (n=105) had severe disability (>50) and 14% (n=13) with active disease (n=95) had no disability (<20). IBD-DI was also higher in patients with more recent onset of disease (<8 years) and those with longer disease duration (≥17 years). In both disease, female gender (p=0.006 for CD and p=0.048 for UC) and disease activity (p<0.0001 for CD and p=0.039 for UC) remained significantly associated with greater IBD-DI scores. No relation between age and IBD-DI has been found.

    Table 2

    Analysis of sociodemographic and clinical factors associated with IBD-DI score in a population-based IBD study (n=200)

    Discussion

    IBD is a chronic progressive, destructive condition, resulting in loss of function.1–3 Therefore, a need exists to develop, as in other chronic diseases,30 ,31 a valid instrument evaluation of disability. Other validated indexes have been proposed: one evaluating work disability,32 another one using a different methodology to assess disability33 and a proposal for a scoring system of the ICF questionnaire.13

    We developed the IBD disability index according to the WHO classification,3 ,11 which mandates a comprehensive process of patient concept elicitation interviews, expert interviews, item generation, content validity, patient cognitive interviews and a quantitative study.34 However, a formal validation study conducted in a unique population was lacking, hence the current study.

    Of the 200 included patients with IBD, more than half of the patients had inactive disease, which is consistent with previous epidemiological assessments in the literature. The quality of data generated by the IBD-DI was excellent with a maximum 1.5% of missing data per item. The quantitative validation found an internal consistency of 0.86 showing a good reliability of the scale, a good correlation between IBD-DI and Physical SF-36 and IBDQ, and a very good inter-interviewer reliability with an ICC of 0.91. Intra-interviewer reliability was shown to be only moderate with an ICC of 0.54, but changes in disability of patients with IBD over time may potentially explain this result. The median time between two interviews was 3.8 weeks (2.7 to 4.6). As demonstrated in clinical trials, clinical disease activity may rapidly change over time. As clinical disease activity influences the IBD-DI score in this validation study, this may have contributed to the moderate intrarater reliability of the IBD-DI.

    In the whole IBD population, IBD-DI score ranged from 0 to 91 (possible maximum score of 100) and the mean of IBD-DI was 35.3 (±SD 20.5). Interestingly, no important difference in scores was observed between patients with CD and those with UC. This finding is surprising given the perception in the clinical community that the former disease is associated with a higher risk of progression to serious complications. In our population-based study, in the whole incident recorded cases, UC represents one-quarter of all IBD cases.15–17

    It is noteworthy that there was no difference between the 200 included patients and the 274 non-participating patients, thus confirming that our sample is representative of our population-based IBD cohort. Nevertheless, the low number of patients with UC in the studied cohort could result in a lack of statistical power to highlight potential differences between UC and CD. Important risk factors for high IBD-DI scores were female gender, short (<8 years) or long disease duration (≥17 years), and high clinical disease activity. Interestingly, 16% of patients had severe disability as defined by the fourth quartile of IBD-DI while in clinical remission, thus indicating that there may be some disconnect between clinical disease activity and disability in some patients with IBD. Further research is necessary on biological, environmental and psychosocial determinants of disability.

    These findings are consistent with those of other studies. In a population-based cohort (n=3666) of patients with inflammatory polyarthritis, women had worse HAQ scores than men.35 Similarly, in a European MS cohort I of 1992 patients, a multivariate multiple regression analysis found that female gender was significantly associated with a lower Physical SF-36 index and higher disability level.36

    The strengths of our study are the methods used to validate the IBD-DI and the first description of the disability status in a population-based cohort of patients with IBD. We used a thorough methodology including all steps (item reduction and data structure, reliability and construct validity) of quantitative validation. We performed this validation in an independent cohort from the cohort used to develop the IBD-DI.11 Indeed, the recruitment of patients in our study from general population provided the full spectrum of disability necessary to validate the IBD-DI. Moreover, the distribution of the IBD-DI in a population-based IBD cohort best reflects the real-world setting.

    However, our study had some limitations. First, we did not evaluate responsiveness to change. This is a critical property of an instrument for use in either randomised controlled trials of therapy as an outcome measure or in longitudinal cohort studies to evaluate disease burden in a population over time. We plan to conduct a rigorous evaluation of responsiveness. We will use the same cohort and will question patients again during the year 2016. Patients will be their own control, and we will study the change of IBD-DI score and also the associated factors with these changes. Second, we observed a moderate intrarater reliability with an ICC of 0.54 given that this was 1 month apart. This aspect further requires further investigations. Third, the study was conducted exclusively in France where the model of healthcare delivery and social supports differs from those present in many other jurisdictions. It is noteworthy that the two questions on healthcare systems were not relevant in France (98% answered that they have a good healthcare). Regarding ‘Social Security Disability Insurance benefits’, in France, IBD are considered as a long-term disease among 29 long-term diseases and all patients with IBD are totally taken in charge by the French healthcare system. In countries other than France, these questions might be considered as a facilitator or barrier. The effects of these societal factors on disability require further assessment, which will require comparative international studies. The potential influence of environmental factors on disability status will require further investigation. Finally, IBD-DI is validated in its French version and will need to be translated for use in other languages following guidelines of translation process (forward/backward translation) and assessing cross-cultural validity.

    In conclusion, a validated version of the IBD-DI is now available that can be used as an outcome measure in clinical trials and prospective epidemiological studies. Indeed, the resulting IBD-DI, containing 14 items, showed highly satisfactory internal consistency, interobserver reliability, construct validity and a moderate intraobserver reliability. Nevertheless, further research is needed to confirm the structural validity and to assess the responsiveness of IBD-DI. There is also a need to assess its validity in specific populations such as elderly patients and to assess the impact of drug intervention in clinical trials.

    Acknowledgments

    The authors wish to thank the interviewing practitioners who collected data: I. Rousseau, A Pétillon, B. Turck, V. Kail, C. Cunisse, S. Auzou, M. Leconte, C. Le Gallo, D. Rime. The authors thank all patients and all gastroenterologists who participated in this study and the European Charity Fondation Digestscience. The authors would also like to thank all the people who were actively involved in the qualitative study.

    References

    View Abstract

    Supplementary materials

    • Supplementary Data

      This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

    Footnotes

    • Additional material is published online only. To view please visit the journal online (http://dx.doi.org/10.1136/gutjnl-2015-310151).

    • This work was presented in part at the European Crohn's and Colitis Organization (ECCO) meeting held in Barcelona in 2015, the Digestive Disease Week (DDW) meeting held in Washington, DC in 2015.

    • Collaborators The IPNIC group: Walter Reinisch and Herbert Tilg (Austria); Michael Kamm (Australia); Geert D'Haens, Edouard Louis and Geert Van Assche (Belgium); Brian Feagan and E Jan Irvine (Canada); Pierre Michetti (Switzerland); Toshifumi Hibi (Japan); Jürgen Schölmerich and Stefan Schreiber (Germany); Pia Munkholm (Denmark); Julian Panes and Cieza Alarcos (Spain); Jean-Frédéric Colombel, Jacques Cosnes, Marc Lémann, Maïté Lewin, Jean-Yves Mary, Benjamin Pariente and Laurent peyrin-Biroulet (France); Simon Travis (UK); Yehuda Chowers (Israel); Silvio Danese and Maurizio Vecchi (Italy); Daan W. Hommes (The Netherlands); Tom Oresland (Norway); Joel Fletcher, Edward V Loftus Jr., Wlliam J. Sandborn and Bruce E. Sands (USA).

    • Contributors CG-R: concept and study design, acquisition and interpretation of data; and drafting the manuscript. HS: data management, interpretation of data and statistical analysis. GS, NT and MF: interpretation of data; drafting and critical revision of the manuscript. WJS and BGF: initiation of the study; interpretation of data; drafting and critical revision of the manuscript. AD: supervisor of data management and statistical analyses. NaG-D: concept and study design; study logistics. J-FC: concept and study design; interpretation of data; drafting and critical revision of the manuscript. LP-B: concept and study design; acquisition of data, interpretation of data; drafting and critical revision of the manuscript.

    • Funding EPIMAD is organised under an agreement between the Institut National de la Santé et de la Recherche Médicale (INSERM) and the Institut de Veille Sanitaire (InVS) and also received financial support from the François Aupetit Association, Lille, Amiens and Rouen University Hospitals. This study has been funded by «le Programme Hospitalier de Recherche Clinique Inter Regional» 2011 (Promoteur : CHRU Lille, Financement: Ministère de la Santé, France). AbbVie Company participated in developing the content for the IPNIC meetings, but was not involved in the development or review of this article with the authors. Editing services, including for the slides at the IPNIC meetings, were provided by Margaux-Orange, and financial support for these services was provided by AbbVie. The Health French Ministry grant registration number is 2011-A00877-34.

    • Competing interests CG-R has served as speaker for Abbvie France, Ferring International, Janssen International and MSD France. GS has served as speaker for MSD France, Ferring France, Abbvie France and Vifor France. MF has served as speaker for Abbvie France, Ferring France and MSD France. WJS has received consulting fees from Abbott, ActoGeniX NV, AGI Therapeutics, AlbaTherapeutics Corp, Albireo, Alfa Wasserman, Amgen, AM–Pharma BV, Anaphore, Astellas, Athersys, Atlantic Healthcare Ltd, Aptalis, BioBalance Corp, Boeh-ringer-Ingelheim, Bristol-Myers Squibb, Celgene, CelekPharmaceuticals, Cellerix SL, Cerimon Pharmaceuticals, ChemoCentryx, CoMentis, Cosmo Technologies, Coro-nado Biosciences, Cytokine Pharmasciences, Eagle Phar-maceuticals, EnGene, Eli Lilly, Enteromedics, ExagenDiagnostics, Ferring Pharmaceuticals, Flexio Thera-peutics, Funxional Therapeutics Ltd, Genzyme Corp, Gilead Sciences, Given Imaging, GlaxoSmithKline, HumanGenome Sciences, Ironwood Pharmaceuticals, KaloBiosPharmaceuticals, Lexicon Pharmaceuticals, Lycera Corp, Meda Pharmaceuticals, Merck Research Laboratories,Merck Seron, Millenium Pharmaceuticals, Nisshin Kyorin Pharmaceuticals, Novo Nordisk, NPS Pharmaceuticals, Optimer Pharmaceuticals, Orexigen Therapeutics, PDL Biopharma, Pfizer, Procter and Gamble, Prometheus Laboratories, ProtAb Ltd, Purgenesis Technologies, Relypsa, Roche, Salient Pharmaceuticals, Salix Pharmaceuticals, Santarus, Schering Plough, Shire Pharmaceu-ticals, Sigmoid Pharma Ltd, Sirtris Pharmaceuticals, SLAPharma UK Ltd, Targacept, Teva Pharmaceuticals, Therakos, Tillotts Pharma AG, TxCell SA, UCB Pharma, ViametPharmaceuticals, Vascular Biogenics Ltd, Warner ChilcottUK Ltd, and Wyeth; WJS received research grants from Abbott, Bristol–Myers Squibb, Genentech, GlaxoSmithKline, Janssen, Milennium Pharmaceuticals, Novartis, Pfizer, Procter and Gamble, Shire Pharmaceuticals, and UCB Pharma; WJS received payments for lectures/speakers bureaux from Abbott, Bristol-Myers Squibb, and Janssen; and holds stock/stock optionsin Enteromedics. BGF has served as consultant for Abbott/AbbVie, Actogenix, Akros, Albireo Pharma, Amgen, Astra Zeneca, Avaxia Biologics, Avir Pharma, Axcan, Baxter Healthcare Corp., Biogen Idec, Boehringer-Ingelheim, Bristol-Myers Squibb, Calypso Biotech, Celgene, Elan/Biogen, EnGene, Ferring Pharma, Roche/Genentech, GiCare Pharma, Gilead, Given Imaging, GSK, Ironwood Pharma, Janssen Biotech (Centocor), JnJ/Janssen, Kyowa Kakko Kirin, Lexicon, Lilly, Lycera BioTech, Merck, Millennium, Nektar, Novonordisk, Pfizer, Prometheus Therapeutics and Diagnostics, Protagonist, Receptos, Salix Pharma, Serono, Shire, Sigmoid Pharma, Synergy Pharma, Takeda, Teva Pharma, TiGenix, Tillotts, UCB Pharma, Vertex Pharma, VHsquared, Warner-Chilcott, Wyeth, Zealand, Zyngenia. BGF received Research and Grant from Abbott/AbbVie, Amgen, Astra Zeneca, Bristol-Myers Squibb (BMS), Janssen Biotech (Centocor), JnJ/Janssen, Roche/Genentech, Millennium, Pfizer, Receptos, Santarus, Sanofi, Tillotts, UCB Pharma. BGF is speaker's bureau for Abbott/AbbVie, JnJ/Janssen, Takeda, Warner-Chilcott, UCB Pharma. BGF is Member or/and Scientific Advisory Board for Abbott/AbbVie, Amgen, Astra Zeneca, Avaxia Biologics, Bristol-Myers Squibb, Celgene, Centocor, Elan/Biogen, Ferring, JnJ/Janssen, Merck, Novartis, Novonordisk, Pfizer, Prometheus Laboratories, Protagonist, Salix Pharma, Takeda, Teva, TiGenix, Tillotts Pharma AG, UCB Pharma. Brian G Feagan is Board of Directors for Officer—Robarts Clinical Trials. JFC has served as consultant or advisory board member for Abbvie, ABScience, Amgen, Bristol Meyers Squibb, Celltrion, Danone, Ferring, Genentech, Giuliani SPA, Given Imaging, Janssen, Immune Pharmaceuticals, Medimmune, Merck & Co., Millenium Pharmaceuticals, Neovacs, Nutrition Science Partners, Pfizer, Prometheus Laboratories, Protagonist, Receptos, Sanofi, Schering Plough Corporation, Second Genome, Shire, Takeda, Teva Pharmaceuticals, Tigenix, UCB Pharma, Vertex, Dr August Wolff GmbH & Co. J-FC has served as speaker for Abbvie, Ferring, Janssen, Merck & Co., Nutrition Science Partners, Takeda. LP-B has served as consultant for Merck, Abbvie, Janssen, Genentech, Mitsubishi, Ferring, Norgine, Tillots, Vifor, Therakos, Pharmacosmos, Pilège, BMS, UCB-pharma, Hospira, Celltrion, Takeda, Biogaran, Boerhinger-Ingelheim, Lilly, Pfizer, HAC-Pharma, Index Pharmaceuticals. LP-B has served as speaker Lecture fees from Merck, Abbvie, Takeda, Janssen, Ferring, Norgine, Tillots, Vifor, Therakos, Mitsubishi, HAC-pharma.

    • Patient consent Obtained.

    • Ethics approval The study protocol has been approved by the ANSM with the number 2011-A00877-34 and by Nord-Ouest IV IRB with the number 11/53.

    • Provenance and peer review Not commissioned; externally peer reviewed.

    • Data sharing statement The data belong to the Lille University Hospital (EPIMAD Registry). The scoring system belongs to Lille University Hospital. The questionnaire belongs to the IPNIC group.

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.