The epidemiology of stroke in the pre-hospital setting in Perth, Western Australia: factors influencing ambulance usage, accuracy of diagnosis and characterisation of stroke mimics

[Truncated abstract] Aim To explore the epidemiology of stroke in patients who presented to Metropolitan Perth Emergency Departments (EDs) during the five year study period from July 2001 to July 2007 through a determination of the sensitivity of prehospital stroke diagnosis, characterisation of conditions which mimic stroke and evaluation of a prehospital stroke screen to improve diagnostic accuracy. Setting This thesis was set in the Metropolitan Perth region, which is located in the southwest of Western Australia (WA), and is home to 72% of the state's total population. The adult population of Perth at the midpoint of this study was 1.06 million people. Method Administrative data sourced through the Western Australian Data Linkage Unit were used to create two cohorts. The first cohort included all presentation to EDs for stroke and the second included all transports to hospital where patients were given a provisional diagnosis of stroke by the attending paramedic. Descriptive analyses were conducted using demographic and clinical factors to compare patients transported to hospital by ambulance with those transported by other means. Logistic regression modelling was used to quantify the influence of variables predictive of ambulance transport to hospital while controlling for potential confounders. Logistic regression modelling was also used to identify predictors of death within 30 days of stroke. The diagnostic accuracy of the clinical assessment carried out by ambulance paramedics was assessed by comparing the ambulance diagnosis of stroke with the final ED diagnosis. Cases correctly diagnosed as stroke to determine the temporal and clinical characteristics of stroke patients transported to hospital by ambulance and cases misdiagnosed were examined to characterise the types of conditions that mimic stroke.