Hypothesis / aims of study
The ICIQ-B is a patient-completed questionnaire for evaluating symptoms of anal incontinence and its impact on quality of life (QoL) . The ICIQ-B consists of 21 items, with 17 arranged in three scored domains: bowel pattern, bowel control and the impact on quality of life. Four further unscored items are included in the questionnaire to evaluate important issues from both a clinical and a patient perspective. The questionnaire is designed for general practitioners and clinicians in both primary and secondary care institutions, and its purpose is to screen for incontinence, to obtain a brief yet comprehensive summary of the level, impact and perceived cause of symptoms of incontinence and to facilitate better patient-clinician discussions. It is also an ideal research tool.
The ICIQ-B is part of the International Consultation on Incontinence (ICI) suite of validated continence questionnaires (www.iciq.net). To be able to perform comparisons between different countries, the goal of the current study was to translate the ICIQ-B questionnaire to Norwegian, and validate the translation using a sample of patients with symptomatic anal incontinence.
Study design, materials and methods
Translation and cultural adaptation: The ICIQ-B was initially translated from English into Norwegian using a bi-lingual Norwegian/English translator, followed by a back translation into English conducted by another bi-lingual Norwegian/English translator. The English back translation were then sent to, and evaluated by, the ICIQ study group, which provided useful comments regarding possible ambiguities or other flaws. The translated instrument was then pilot tested for comprehensibility, readability and equivalence through cognitive interviews with nine men and women with anal incontinence. In addition, opinions from six bi-lingual or monolingual multidisciplinary clinical experts were obtained (colorectal surgeons, nurses and physiotherapists). Minor discrepancies were identified and amended, resulting in a comprehensible Norwegian version of the ICIQ-B ready for validation.
A total of 208 Norwegian women and men with anal incontinence completed the questionnaires at baseline, with 50 completing them again after one to six weeks (re-test). The respondents were outpatients of University Hospitals from three different regions of Norway.
Reliability: Cronbach’s alpha (α) was calculated as a measurement of internal consistency. Test-retest reliability was evaluated using intra-class correlation coefficients (ICC) to measure stability over time. Additionally, change scores such as Standard Error of Measurement and Smallest Detectable Change were reported.
Content validity: Floor and ceiling effect and the percentage of missing items were calculated for baseline data. Floor and ceiling effect are considered problematic if more than 15% of participants achieve the highest or lowest possible score. Based on Cosmin recommendations <3% missing data at item level is acceptable and >15% is unacceptable.
Construct validity: The ICIQ-B factor structure was tested through two steps. First, a confirmatory factor analysis (CFA) was performed using a structural equation modeling (SEM). Thereafter, an exploratory factor analysis (EFA) was performed, using a principal component analysis with Varimax rotation.
We had a 58.8% response rate at base line and a 45.7% response rate in the re-test. At base line, most of the respondents were women (87.3%). The age range was from 18-89 years with a mean of 59.2 years (SD= 15.0). The participants had a mean bowel pattern score of 7.68 (SD 3.12) (scale from 1-21), mean bowel control score of 17.29 (SD 5.21) (scale from 0-28) and mean impact on QoL 17.82 (SD 6.46) (scale score from 0-26). A higher score indicates worse symptoms or impact than a low score in all three domains.
Reliability: Crohnbach’s alpha (α) ranged from 0.59 to 0.85 and intraclass correlation coefficients (ICC) ranged from 0.91 to 0.95. The instrument measurement error (SEM) was estimated to 1.89 - 4.40 points, and the smallest detectable change (SDC90) was 4.49 - 10.45 points (Table 1). SEM is an expression of the average measurement error, while SDC is the uncertainty related to this average. For instance, the Standard error of measurement is 1.89 for bowel pattern, but to be 90% certain that a real change beyond the measurement error has occurred; the patient score has to change by 4.49 points.
Content validity: Small adjustments to the questionnaire were made based on the results from the cognitive interviews. Overall percentage of missing data at baseline was 3.3%, ranging from 0.5% to 11.1% (Restriction of sexual activities) for single items. No floor or ceiling effects were found in the total score distributions of bowel pattern, bowel control or impact on QoL. Similarly, no floor or ceiling effect was observed in any of the single items.
Construct validity: The original 3-factor solution from the English version of ICIQ-B was tested in the Norwegian version by means of CFA. The goodness-of-fit statistics did not show an acceptable fit and suggested that the model should be rejected. The next step, EFA, suggested a 5-factor solution of the 17 items, with 2-6 items in each factor (Table 2).
Interpretation of results
The testing of ICIQ-B demonstrates good reliability in terms of internal consistency and stability of the instrument. Two of three domains had acceptable and good internal consistency in terms of Crohnbach’s alpha, bowel control and impact on QoL respectively. However, for the bowel pattern domain (α= 0.59), the result is borderline to questionable (>0.6). Stability over time was excellent for all domains with ICC ranging from 0.91 to 0.95.
The instrument shows an acceptable level of missing data (mean 3.3%). The absence of floor and ceiling effect demonstrates that the instrument is able to produce a good distribution of responses to an item and that there is no clustering / skewness of scores at the upper or lower level on the scale. The original English version of ICIQ-B suggests a 3-factor solution based on 17 of the instrument’s items. However, the analysis of the Norwegian version suggest a 5-factor solution with an almost completely different factor structure than the original. One of the five factors from the EFA is almost identical to the impact on QoL domain, except that it also loads on the medication item. The poor fit of constructs between the original and translated versions may be due to a small Norwegian sample (n=161 after cases with missing items were excluded). However, another explanation may be that the original English version of ICIQ-B lacks a good factor structure; A review of Quality of life measures in FI concludes that the original English ICIQ-B version did not show sufficient construct validity, since the factor analysis did not consistently show factor loadings of >0.60.