Exploring and reconciling discordance between documented and preferred resuscitation preferences for hospitalized patients: a quality improvement study.

Purpose A discordance, predominantly towards overtreatment, exists between patients' expressed preferences for life-sustaining interventions and those documented at hospital admission. This quality improvement study sought to assess this discordance at our institution. Secondary objectives were to explore if internal medicine (IM) teams could identify patients who might benefit from further conversations and if the discordance can be reconciled in real-time. Methods Two registered nurses were incorporated into IM teams at a tertiary hospital to conduct resuscitation preference conversations with inpatients either specifically referred to them (group I, n = 165) or randomly selected (group II, n = 164) from 1 August 2016 to 31 August 2018. Resuscitation preferences were documented and communicated to teams prompting revised resuscitation orders where appropriate. Multivariable logistic regression was used to determine potential risk factors for discordance. Results Three hundred and twenty-nine patients were evaluated with a mean (standard deviation) age of 80 (12) and Charlson Comorbidity Index Score of 6.8 (2.6). Discordance was identified in 63/165 (38%) and 27/164 (16%) patients in groups I and II respectively. 42/194 patients (21%) did not want cardiopulmonary resuscitation (CPR) and 15/36 (41%) did not prefer intensive care unit (ICU) admission, despite these having been indicated in their initial preferences. 93% (84/90) of patients with discordance preferred de-escalation of care. Discordance was reconciled in 77% (69/90) of patients. Conclusion Hospitalized patients may have preferences documented for CPR and ICU interventions contrary to their preferences. Trained nurses can identify inpatients who would benefit from further in-depth resuscitation preference conversations. Once identified, discordance can be reconciled during the index admission.