Eable size without losing essential information through an orthogonal design (when

However, the problem of multiple testing has to be borne in mind: with more than 200 statistical tests to be performed some significant title= jir.2014.0026 results will be CX-5461 manufacturer produced "by chance", they title= srep39151 are however artefacts (approx. Health Economics Review 2013, 3:30 http://www.healtheconomicsreview.com/content/3/1/Page 4 ofTable 1 Mean importance of therapy characteristics assessed by physicians taking the view of their patients and by the patients themselvesMean importance Physicians1 Benefits 85 84 83 77 76 76 73 62 60 Side effects 92 87 87 79 57 56 Mode of administration 90 86 86 83 81 77 76 69 68 47Patients2 89 94 90 81 82 78 85 88Therapy characteristic Drug improves physical state (e.g. better mobility, pain relief) Drug has very high efficacy (reduction of viral load) Drug promises maximum prolongation of life expectancy Drug improves emotional and mental state (e.g. less thoughts about disease) Drug allows for improved mobility (e.g. longer journeys possible) Drug improves social contact opportunities (e.g. visits possible) L.Eable size without losing essential information through an orthogonal design (when making certain assumptions about interaction effects). The SPEED software package was used to select the optimal subset of scenarios [23], maximum dissimilarity between therapy alternatives was achieved by generating the alternative B as exact mirror image of alternative A (using the fold-over technique) [12]. In this study we conducted the DCE technique with eight pair decisions, each with six characteristics. Respondents had to choose eight times between treatment A or B. The presented treatment pairs, as well as the characteristics were the same as in the previous patient study with verbal adaptions to physicians. Linguistically, the questions were adapted in the questionnaire for physicians from a self-assessment: "What would you choose ...?" into a judgment: "How do you think your patients would rate" and "What would your patient choose?".Statistical analysisThe Discrete Choice Experiment is a choice based method, and a variant of the conjoint analysis that was made possible through the theoretical work of Lancaster [16] and McFadden [17]. In the Discrete Choice Experiment different therapies are presented pair-wise and the subjects have to decide for one of the options [18]. In a first step all characteristics that are relevant for each target group have to be identified [19]. The treatment alternatives are presented to the subject and the subject has to decide for one of the presentedFor statistical data analysis we used analysis of variance, regression analysis, factor analysis, and random effect logit models for the DCE. All statistical analyses were done using SPSS and STATA. A p-value title= jir.2014.0026 results will be produced "by chance", they title= srep39151 are however artefacts (approx. 10 of such results must be expected at a 95 confidence level). A possible increase of significance level (e.g. Bonferroni correction) may minimize these errors but increases the risk of ignoring really existing relationships.