Aims and objectives: To assess agreement between data retrieved from interviews with nurses and data from electronic patient records (EPR) about hospitalised patients' symptoms, clinical signs and treatment during the last three days of life.
Background: Patient records have been used to map symptom prevalence in dying hospitalised patients. However, deficiencies have been found regarding nursing documentation. To our knowledge, this is the first study to assess the agreement between nurse interviews and patient electronic records during the last three days of life in a hospital.
Design: This retrospective study was undertaken in a Norwegian hospital.
Method: We used the resident assessment instrument for palliative care to interview nurses on 112 dying patients, and we independently extracted data from EPR. The agreement between the two data sets was computed with the kappa coefficient. Sensitivity and specificity were calculated. Interview data were used as a reference.
Results: The agreement between the two data sets ranged from poor to good and was highest among symptom variables, including pain, dyspnoea, nausea and the clinical sign falls. In contrast, several clinical variables ranged from poor to fair levels of agreement. The majority of the treatment variables ranged from moderate to good levels of agreement.
Conclusions: Data from the EPR on symptoms (e.g., pain, dyspnoea and nausea) and treatment variables appeared to be reliable and trustworthy, but the data related to fatigue, dry mouth, bloating and sleep interfering with normal functioning should be interpreted carefully.
Relevance to clinical practice: This study contributed to knowledge of agreement between data from nurse interviews and electronic records on symptoms, clinical signs and treatment of dying patients in last three days of life.
© 2011 Blackwell Publishing Ltd.