Dying in hospital - A retrospective medical record analysis on care in the dying phase in intensive care units and general wards.

Abstract

Background Care in the dying phase is often suboptimal in hospitals outside specialised palliative care. Studies of the implementation of recommendations for care in the dying phase are rare. Medical records can provide information in this regard. Methods Retrospective medical record analysis of n=400 patients who died on six intensive care units and four general wards of two German medical centres. To evaluate the care in the dying phase, we descriptively analysed 37 variables reflecting recommendations of the German Palliative Care Guideline. To identify factors associated with important aspects of care, seven of these variables were determined as appropriate for multiple logistic regression. Results In 299/400 (74.8%) patients, recognition of the dying phase was documented. Patients with ability to communicate were informed about the impending death in 46/121 (38.0%) and informal caregivers in 282/390 (72.3%). Patients' wishes regarding care were documented in 33/122 (27.1%). Monitoring was rarely stopped before death, with significantly lower percentages on intensive care units: e.g. vital signs in 30% (14.9% in intensive care units vs. 52.9% in general wards; confidence interval [28.4; 47.6]). Validated symptom assessment tools were used in 272/400 (68.0%), mainly for pain (66.3%). Logistic regression analysis identified setting (intensive care units vs. general ward), recognition of the dying phase, patient age, communication ability, and malignant neoplasm as factors significantly associated with aspects of care. Conclusion The dying phase was not consistently documented, many presumably non-beneficial interventions continued until death, systematic symptom assessment beyond pain was lacking, and communication with patients and informal caregivers was poorly documented. Findings suggest a need for setting-specific strengthening of healthcare professionals in these settings.