Chapter 6 124 with Engelberg: substituting sample median values for responses of “don’t know” or “no response”; and imputing a value of 0 for “doctor didn’t do this”. Engelberg et al. chose for the imputation of a 0 for “doctor didn’t do this” based on the assumption that all of the items identified important aspects of end-of-life communication, and therefore the failure to complete or address an item warranted a low score.25 However, our patient population was not (necessarily) at the end-of-life and therefore we deemed it inappropriate for two of the items (“Talking about how long you have to live” and “ Talking about what dying might be like”) to impute a 0 for “doctor didn’t do this”. Instead, we treated it as “no response”. We summarized scores in accordance with Engelberg to a median score for general communication (QOC-gen) and median score for care decision communication (QOC-CD).25 To adjust for potential confounders, we assessed the following baseline characteristics: age, gender, educational level, health perception and presence of a family member. To take into account the possibility of other factors that improve care decision discussions over time, we also explored the effect of time. Secondary outcomes are: 1) how many patients recall a care decision discussion, 2) who initiated this discussion, 3) how many patients recall to have received the leaflet, 4) their evaluation of this leaflet, 5) whether they had a previous care decision discussion and with whom, and 6) what they perceive to be an appropriate moment for care decision discussions. Originally, we planned to evaluate the physicians preparedness for care decision discussion as well, using a questionnaire inspired by the one used by Smith et al. to evaluate residents approaches to advance care planning. 27 We intended to compare these before the implementation and 3 months after implementation. However, because of the extremely low response rate before implementation (<20%), possible due to extensive workload due to the COVID-19 pandemic, we dropped this analysis. 2.5 Statistics We compared baseline characteristics between the two groups with Chi-square test for nominal data and Mann Whitney U for ordinal data. Primary outcome (care decision communication, QOC-CD) was first compared using univariate
RkJQdWJsaXNoZXIy MTk4NDMw