10 – Derivation of a short form of the dentine hypersensitivity questionnaire*

The original Dentine Hypersensitivity Experience Questionnaire (DHEQ) has 48 items, 34 of which comprise the scale that measures the subjective impacts of dentine hypersensitivity (DH). If a questionnaire is extensive, it is helpful to develop a short form to minimize participant burden by reducing the time to complete it. Doing so will also limit the cost of data collection. A short form of a questionnaire will facilitate its use in broader segments of the population and in special groups such as the elderly. It may also have applicability in clinical settings and in larger population samples. Further, a shorter version may reduce missing data. Naturally, such a short form should preserve the properties of the original when possible. This chapter reports the work to derive and evaluate short forms of the DHEQ. Two methods were used: the item-impact and the stepwise regression methods. The short forms were then evaluated cross-sectionally and in two randomized controlled trials, in part by assessing their performance compared with the original DHEQ. The two techniques produced quite similar short forms, so the choice between them required careful consideration. Aim: To derive and evaluate a short form of the DHEQ. Methods: Data from three previous studies of DH (n=353) were pooled and randomly divided in half. Ten-item and 15-item short forms were derived using the first half of the data using the item-impact and regression methods. The four short forms were evaluated using the second half of the data. Results: The 10-item and 15-item versions of the regression short form detected impacts in 37% and 61% of participants, respectively, compared with 68% and 93% using the item-impact method. All short forms had internal consistency (Cronbach’s α>0.84) and test–retest reliability (intraclass correlation coefficient, ICC >0.89). All correlated with the long form (all r>0.93, P<0.001) and with the impact of the effect on the mouth on everyday life (all r≥0.73, P<0.001). None of the short forms detected a treatment effect in two trials, although all four showed a tendency to detect an effect in a trial when the long form had also done so. Conclusions: The 15-item short form derived from the item-impact method performed better than other short forms and appeared to be sufficiently robust for use by individual patients.