Irritable Bowel Syndrome (IBS) affects large numbers of the population (prevalence estimates ranging from 12% to 30%) [1–7] and accounts for a significant proportion of the workload of both primary care practitioners and gastroenterologists [8]. Treatment options are wide ranging and include symptomatic prescribing, dietary management and a range of psychotherapeutic and complementary therapies. The breadth of the problem and the range of treatment options make this a potential area for much research activity.

The diversity of symptom types and profiles in IBS means that there is no agreement on appropriate determination of efficacy in trials. Researchers have therefore used a range of techniques and measures, which are largely arbitrary and unvalidated.

Three large systematic reviews completed in the UK, Europe and the USA [9–11] demonstrated large variety in the primary outcome measures adopted by trialists, with many trials reporting individual symptoms, such as pain or urgency or more arbitrary outcomes such as percentage of unformed stools. A review of randomised controlled trials of pharmacological treatments [12], noted that symptom data were most commonly collected by daily diary (56% of trials), or during consultation. Only 4/70 included trials reported use of a symptom questionnaire and only 2 of these reported validation of the chosen measure [13, 14].

Selection of a suitable outcome measure is further complicated by the fluctuating nature of IBS. It has been suggested [15], that due to the variance of symptoms over time, clinical trials in this area should have a minimum duration of 2–3 months. The increase in psychotherapeutic and complementary therapies such as hypnotherapy for the management of IBS, would suggest that follow-up should be even longer, as such therapies may prove to be effective (and cost-effective) over extended periods by dealing with root causes or the learning of self-management techniques. The use of daily diary data collection strategies over long periods may be unfeasible or may increase drop-out rates leading to a biased sample.

The ROME I, II and III trial committees have consistently agreed that the main assessment of outcome in IBS trials should be patient centred [16–18] and over a decade ago Tally et al [17], strongly proposed the need to develop sensitive outcome measures with multiple domains as a priority in this area. A recent move towards global assessment, led to the adoption of 'adequate relief' as a primary endpoint in some trials [19–21] and a literature review of outcome measures identified a dichotomous 'adequate relief' question as the measure of first choice in assessing global symptomatology [22]. The ROME III committee now indicate that 'adequate' or 'satisfactory' relief are the current standards for primary outcome assessment in treatment trials in IBS [18] and this was supported by a recent review of primary endpoints in IBS trials [23]. It is however generally indicated that integrative symptom questionnaires are also acceptable endpoints subject to appropriate validation [18].

There is the risk that exclusive use of dichotomous or 'adequate relief' endpoints in a disease with such differing symptom profiles may result in the failure to identify therapies of benefit to sub-groups of patients or effective on specific symptoms. The GSRS-IBS symptom score [24], based on Rome I criteria has been shown to discriminate by severity and frequency of disease symptoms. However this tool has not been widely used in practice and its generalisability across all age groups has yet to be demonstrated. A recent visual analogue scale for IBS (VAS-IBS) has been developed [25], however this tool was validated on a sample of women only and requires further testing in clinical practice to establish responsiveness.

Therefore the continuing need for a symptom measure suitable for use in a range of clinical and management trials in the area of IBS remains. The principle characteristics of this tool were determined to be 1) patient reported, 2) suitable for repeated testing over extended time periods, 3) acceptable to patients for ongoing completion, and 4) containing multiple dimensions covering relevant symptoms. This paper therefore describes the development and validation of a tool which meets these requirements, is suitable for self-completion and postal return and fulfils the recommendations of the ROME working group regarding validation of a primary outcome measure [16]. Identification of a well validated and increasingly adopted IBS specific quality of life measure [26, 27], limited the development of this new tool to a symptom only score which would be used to complement this.

The aim of this study was to develop a symptom score suitable for use in a range of therapeutic trials in IBS and establish its acceptability, reliability, validity and sensitivity to a change in health.

Unlike other symptom scores validated using referral-based populations, the validation of this questionnaire was carried out on a large representative community based cohort. Our primary aim was to create a patient reported score for repeated use in trials – in this setting patient interpretation of symptoms is not an issue and could be beneficial, allowing patients to self-define symptoms within tight parameters. Translation to a clinical setting will require further testing especially in light of known differences in the way patients and doctors define key symptoms such as constipation [38].

The high response rate and the high completion rates of all questions demonstrate the questionnaire's acceptability and suitability for self-completion. The possibility of a response bias due to 15% of responders not working because of health problems may exist. However the prevalence of a range of conditions such as diabetes (5%), hypertension (31%) and cancer (4%) are similar to UK figures and thus our responders reflect the general morbidity of a community population. Patients with other pre-existing GI conditions were excluded from the study.

Three underlying dimensions were identified within the questionnaire with Cronbach's alpha coefficients of between 0.7 and 0.9. These are consistent with recommendations of internal reliability [34, 39, 40] and provide evidence that the items within each dimension are measuring the same characteristic.

The stability of the responses was demonstrated by the high intra-class correlation coefficients of between 0.75 and 0.81. Consequently we can be confident that changes measured by this tool will be due to the effect of treatment and not due to measurement errors of the tool.

Increases in pain, diarrhoea and constipation scores were associated with decreasing quality of life confirming construct validity thus providing evidence that they measure what they purport to. As anticipated, the Qol scores correlated more highly with the pain scale [26].

The decrease in all scores, and corresponding small to medium effect sizes following an improvement in symptoms, indicate the tool's sensitivity to detect a change in health. The small effect sizes found for those with worse symptoms are most probably due to a ceiling or regression to the mean effect. In these cases, baseline scores, with the exception of constipation, were from subjects with greater symptoms than an average IBS sufferer and therefore scores were unlikely to increase. These changes were observed after a short duration of time (7–10 days); however differences at 3 months have also been shown. A randomised controlled trial of gut-directed hypnotherapy demonstrated an improvement in the Birmingham IBS pain, diarrhoea and overall symptom scores at 3 months compared to usual management [41]. An ongoing randomised controlled trial of a dietary product will provide further estimates of responsiveness over time to supplement these results.

The questionnaire performed well for both genders with similar results for reliability, validity and reproducibility being found for men and women. However, gender specific effect sizes could not be reliably estimated due to the small sample of males reporting a change in symptoms.

Whilst the instrument meets validation criteria, it is acknowledged that it may not query the full and diverse range of symptoms experienced by IBS sufferers. The inclusion of additional questions, or the opportunity for patient nomination of personal outcomes (as in MYMOP [42]), may further enhance the tool.

This IBS specific symptom questionnaire has been appropriately developed and tested. It is patient-centred, suitable for self-completion and is acceptable. The score includes multiple dimensions that cover representative and relevant questions. The dimension and summary scores provide reliable and valid outcome measures of IBS symptoms. High levels of reproducibility indicate that the score is suitable for use in therapeutic trials where it is important to establish that measured changes reflect a real change in status, not just variability in measurement. The increase in effect size, with improvement in symptoms confirms the instrument's ability to measure a clinically meaningful change in health. It is particularly important that a symptom questionnaire is responsive to a change in health if it is to be used as an outcome in a clinical trial.

The severity of symptoms experienced by populations recruited to treatment trials should be reported if generalisability is to be established. The lack of an acceptable and validated disease specific symptom score has hampered the comparison of the efficacy of different management strategies or attempts to synthesise the results of existing studies in the area of IBS. We recommend that future studies of IBS utilise the Birmingham IBS symptom score, together with the previously well validated IBS-QoL questionnaire, to facilitate the production of high quality patient centred research and to enable future comparisons between published studies and meta-analysis of data.