The design of this audit was a retrospective case-note review of two groups of consecutive patients with stroke admitted to a district general hospital serving an urban population of 135,000 people. About 150 patients with stroke are admitted annually to the general medical service. Neurology and neuro-surgical services are provided at a regional centre about 25 minutes drive from Hutt Hospital. Inpatient rehabilitation is provided within an age-unrelated rehabilitation service on the hospital campus.
The stroke pathway consisted of a paper document inserted into each stroke admission, either in the emergency department or medical unit. It contained general advice about acute stroke management and then detailed daily activities for each of the first 5 days of the acute admission. The daily activities were listed under medical, nursing, therapist and discharge planning headings. Each item was to be ticked, initialled and dated as achieved by relevant staff on a daily basis. The content of the pathway was determined by literature review, consensus statements and regular meetings of a stroke pathway team over a 3-year period.
Cases were identified from International Codes for Diagnosis (ICD-10) discharge codes for the 6-month periods of June through December 2003 (pre-pathway group) and June through December 2004 (post-pathway group). In addition, after the pathway was introduced a logbook was kept that prospectively recorded patients with stroke admitted to the medical unit; additional cases were identified from this source. The case-notes were reviewed to confirm that the primary reason for admission was stroke according to the World Health Organisation definition .
The following data were abstracted from the case-notes of confirmed stroke patients: length of stay on medical and rehabilitation services, demographics, neurological impairments necessary to classify the stroke according to the Oxfordshire Community Stroke Project (OCSP) , Charlson Comorbidity Index , Barthel Index  at admission and discharge, estimated Modified Rankin Scale  prior to admission and at discharge, discharge disposition, presence and management of specific issues (fever, hyperglycemia, hypertension, prophylaxis for thromboembolism, aspirin given within 48 hours of admission), use of investigations (computed tomography, magnetic resonance imaging, echocardiography, carotid doppler ultrasound, fasting lipid profile, acute phase marker [erythrocyte sedimentation rate or C-reactive protein]), and use of secondary prevention treatment at discharge (blood pressure lowering treatment, smoking cessation programme, optimisation of diabetic control, cholesterol lowering treatment, anti-platelet treatment, anticoagulation for atrial fibrillation).
The Barthel Index (BI) is a 10-item scale of independence in mobility and self-care activities, with scores ranging from 0 (complete dependence) to 20 (complete independence). Scores of 0 to 10 typically indicate severe functional disability. Although the BI was not routinely administered at admission and discharge during either period, it was possible to retrospectively score the items by reference to nursing, occupational therapy and physiotherapy notes within the case-record. The Modified Rankin Scale (MRS) is a simple 6-level categorisation of functional independence: 0 indicates no symptoms, 1 indicates symptoms but no disability, 2 indicates slight disability but independent, 3 indicates moderate disability but can walk independently, 4 indicates moderately severe disability, 5 indicates severe disability. In-hospital functional status at admission was measured with the BI rather than the MRS since the BI is more informative and the MRS is more frequently employed as a community-based outcome, rather than an adequately scaled functional index for hospital inpatients .
A good outcome at discharge from hospital was defined in two ways: discharge to home, nearly independent survival (MRS<4) and independent survival (MRS<3). A MRS<4 has not previously been used to define a good outcome in acute stroke trials, but was chosen here to capture modest improvements in patients severely affected by stroke.
Non-parametric statistical tests were used to evaluate differences between the two cohorts (Mann-Whitney U for continuous or ordinal data, Chi-square for categorical data). To adjust for differences in case-mix , the influence of being admitted in the post-pathway group upon outcome was assessed using logistic regression models incorporating age, gender, comorbidity index, admission BI, whether transferred to rehabilitation and OCSP classification. SPSS version11.5 was used for all analyses.