Page 83 - The efficacy and effectiveness of psychological treatments for eating disorders - Elske van den Berg
P. 83

  Chapter 4 83
 (patient and repeated measures). To control for possible confounding, the baseline score of the dependent outcome variable plus the variables from the baseline analysis were added with a p < .05. Cohen’s d were used to examine effect sizes.
Clinical outcome analyses were performed on a completers dataset (complete measures available at baseline and end-of-treatment) and on a pooled imputed data- set. For missing data, multiple imputation, with 50 imputations for each missing obser- vation, was used under a missing-at-random assumption; no differences were found between patient groups with and without complete EDE-Q global scores. Analyses were performed first on the imputed datasets separately, and then the outcomes of the 50 imputations were combined using Rubin’s rules (Rubin, 1987), using SPSS.
All cost-effectiveness analyses were run on intention-to-treat basis, using a pooled multiple imputed dataset. Unless otherwise indicated, cost-effectiveness findings are based on these imputed data.
Cost-effectiveness calculations
The cost-effectiveness analyses were performed from the direct treatment costs, health care provider perspective. For each patient, treatment costs (in euros) were established by multiplying standard Dutch cost prices, index year 2014 (Zorginstituut Nederland, 2016) by the amount of time spent on outpatient contacts (both contacts directly with patients and contacts concerning patients), by the number of days in day-care and / or number of hospitalization days. The time horizon of this study is from start to end-of-treatment; since this horizon was little over a year for both cohorts, no discounting for future costs / effects was applied. Differences in costs and effects between both cohorts were calculated as difference in cumulative direct costs. A total of 2,500 nonparametric bootstrapped samples was extracted from the 50 imputed data sets, with the number of patients per sample equal to the patient numbers in the original dataset.
Two different incremental cost-effectiveness ratios (ICERs) were calculated for each bootstrapped sample as ICER = (CostsCBT-E – CostsTAU) / (EffectsCBT-E – EffectsTAU) where effect was either EDE-Q global score < 2.77 or BMI ≥ 18.5. ICERs were calcu- lated separately for all patients and for the subgroup of outpatients only, to explore the influence of hospitalization costs. The ICERs were plotted on cost-effectiveness planes and used for further calculations. TAU as comparator intervention is posi- tioned in the origin of the cost-effectiveness plane with the horizontal axis indicating differences in effect and the vertical axis differences in costs. Figures 2 is divided into four specific quadrants along the horizontal and vertical axis with ICERs in the upper right quadrant indicating CBT-E generating better health gain at additional costs,




























































































   81   82   83   84   85