Association between patient-reported symptoms and nurses' clinical impressions in cancer patients admitted to an acute palliative care unit

Background: Clinicians typically rely on their own or the nurses' clinical impression (NI) of symptoms rather than patient self-reports. It is unclear whether these means of assessment yield similar results. Objective: To prospectively compare patient-reported symptoms on a modified Edmonton Symptom Assessment System (ESAS) with NI scores. Methods: Consecutive patients with advanced cancer admitted to our acute palliative care unit between April and July 2010 were studied. We collected the results of the ESAS on the day of admission (D1) to the unit and 2 days later (D3). We also collected the NI of each patient's physical and psychological distress on D1 and D3. Results: One hundred eighteen patients completed the ESAS on D1 and 116 on D3. On D1 there was no significant association between NI score and ESAS assessment except for dyspnea, which was weakly correlated with NI score for physical distress (r=0.22, p=0.02). The median ESAS physical and psychosocial scores were 31 and 12 in patients with NI of low or no physical distress, versus 34 (p=0.07) and 15 (p=0.18) in patients with NI of moderate or severe distress, respectively. On D3, we found a significant association between ESAS and NI only for pain (r=0.32, p<0.001) and anxiety (r=0.30, p=0.001). Sensitivity and specificity of the NIs for ESAS scores were low on both days. Conclusion: The clinical impression of highly trained palliative care nurses showed poor association with patient-reported symptom intensity. Validated symptom assessment tools are needed for bedside clinical assessment.