Our randomized controlled trial of a multifaceted evidence-based intervention for improving the inpatient management of fever, hyperglycemia, and swallowing dysfunction in the first three-days following stroke improved outcomes at 90 days by 15%. We designed a quantitative process evaluation to further explain and illuminate this finding.
Blinded retrospective medical record audits were undertaken for patients from 19 stroke units prior to and following the implementation of three multidisciplinary evidence-based protocols (supported by team-building workshops, and site-based education and support) for the management of fever (temperature ≥37·5°C), hyperglycemia (glucose >11 mmol/l), and swallowing dysfunction in intervention stroke units.
Data from 1804 patients (718 preintervention; 1086 postintervention) showed that significantly more patients admitted to hospitals allocated to the intervention group received care according to the fever (n = 186 of 603, 31% vs. n = 74 of 483, 15%, P < 0·001), hyperglycemia (n = 22 of 603, 3·7% vs. n = 3 of 483, 0·6%, P = 0·01), and swallowing dysfunction protocols (n = 241 of 603, 40% vs. n = 19 of 483, 4·0%, P ≤ 0·001). Significantly more patients in these intervention stroke units received four-hourly temperature monitoring (n = 222 of 603, 37% vs. n = 90 of 483, 19%, P < 0·001) and six-hourly glucose monitoring (194 of 603, 32% vs. 46 of 483, 9·5%, P < 0·001) within 72 hours of admission to a stroke unit, and a swallowing screen (242 of 522, 46% vs. 24 of 350, 6·8%, P ≤ 0·0001) within the first 24 hours of admission to hospital. There was no difference between the groups in the treatment of patients with fever with paracetamol (22 of 105, 21% vs. 38 of 131, 29%, P = 0·78) or their hyperglycemia with insulin (40 of 100, 40% vs. 17 of 57, 30%, P = 0·49).
Our intervention resulted in better protocol adherence in intervention stroke units, which explains our main trial findings of improved patient 90-day outcomes. Although monitoring practices significantly improved, there was no difference between the groups in the treatment of fever and hyperglycemia following acute stroke. A significant link between improved treatment practices and improved outcomes would have explained further the success of our intervention, and we are still unable to explain definitively the large improvements in death and dependency found in the main trial results. One potential explanation is that improved monitoring may have led to better overall surveillance of deteriorating patients and faster initiation of treatments not measured as part of the main trial.