Article Text
Abstract
Introduction Stool form is frequently used as a proxy measure of stool consistency, which is a central component in the description of normal or altered bowel habit. The Bristol Stool Form Scale (BSFS) is in widespread use in clinical practice and research and yet has undergone incomplete assessment of validity and reliability. The aim of this study was to determine concurrent validity and intra-rater reliability of the BSFS.
Method Healthy adults provided a single real stool that was independently categorised using the BSFS by both themselves and by an expert in gastroenterological research, following which stool water was measured using lyophilisation. Participants also used the BSFS to categorise 26 stool models, 7 representing the BSFS types (reference types) and 6 intermediate types, all covertly provided in duplicate. Concurrent validity (comparison with gold standard) was determined by measuring (i) agreement between participants and experts categorisation; (ii) comparison with stool water content; and (iii) the accuracy of categorisation of reference stool models. Intra-rater reliability was measured by the accuracy of categorisation of covert duplicate stool models.
Results 86 healthy adults categorised the stool models, of which 59 provided a real stool. Stool water content was significantly different between stools categorised to different stool types (P < 0.05) and water content correlated with BSFS type (rs = 0.320, p < 0.05). Participants and experts agreed on 36% of real stool categorisations, representing only fair agreement (κ = 0.25). Stools categorised as Type 2 and 5 by participants (both at the border of normal consistency) were actually assigned a wide range of types by the experts. Participants correctly categorised reference stool models on 81% of occasions, representing substantial accuracy. However, the Type 2, 3 and 6 reference models were categorised correctly on less than 75% of occasions. Familiarity with the scale improved accuracy of model categorisation (p < 0.05). Overall substantial intra-rater reliability was demonstrated (κ = 0.72), although low exact agreement was seen with the Type 2 and 3 models.
Conclusion Overall, the BSFS has good concurrent validity and intra-rater reliability. However, categorisation of real stools proves difficult for some. In both categorisation of real stools and stool models issues arise around the borders of stool types representing normal and abnormal bowel habit (Types 2 and 3 and Types 5 and 6) and this should be taken into consideration when using the scale. Familiarity with the scale and education in its use may improve accuracy of stool categorisation. This study should be repeated in patients with abnormal bowel habit to determine its utility in clinical practice.
Disclosure of interest None Declared.