Purchase this article with an account.
Hla M. Htoon, Zachary Ang, Ai T. Aw, Baskaran Mani, Ecosse L. Lamoureux, III, Tin Aung; Optimal Fractional Factorial Designs In The Construction Of Choice Sets For Developing A Glaucoma Utility Index In Singapore. Invest. Ophthalmol. Vis. Sci. 2012;53(14):1423.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
The development of discrete choice experiments (DCE) choice sets requires the application of design theory based evaluation of design efficiency. The study aims to provide optimal fractional factorial designs in the construction of choice sets in developing a glaucoma utility index in Singapore.
Focus group/interviews were initially conducted to (a) validate 6 a-priori dimensions: central & near vision, lighting & glare, mobility, activities of daily living, eye discomfort; (b) understand the effects of glaucoma and its treatment; and (c) further explore other potential dimensions in glaucoma patients in Singapore. Then, design theory based fold-over techniques were used to create varying choice sets for each health state. R package DoE.base and an orthogonal array (OA) online library were the source for these choice sets. Choice sets were then inspected for level balance, orthogonality and for the existence of trivial choice sets. Burgess online software was used to check the efficiency of the design matrix with an optimality criterion determinant C. Determinant C from the information matrix was used to estimate efficiency.
Focus group/ interviews were conducted on 39 glaucoma patients. 67% were Chinese speakers. The mean (±SD) age was 67.5±9.9 years and 28.2% were female. 17.9%, 43.6% and 38.5% of our sample had mild, moderate and severe glaucoma, respectively. The 6 a-priori dimensions, with an additional psychosocial dimension, were validated. After creating choice sets, we identified 2 different question sets based on the 7 dimensions that provided excellent design efficiency. The first question set consisting of 18 choice sets, 7 dimensions, 3 levels of difficulty (none, some, severe) provided 100% efficiency, orthogonality and level balance without presenting trivial choice sets. The second one had 32 choice sets, 7 dimensions, 4 levels of difficulty (none, some, quite some, severe) with 94.49% efficiency.
Our study showed that, based on 7 dimensions, there were two possible optimally designed sets of questions (consisting of 32 and 18 choice sets) which will contribute to the development and validation of the first glaucoma utility index in Singapore.
This PDF is available to Subscribers Only