Aneurysmal subarachnoid haemorrhages : a retrospective cohort study

INTRODUCTION Aneurysmal subarachnoid haemorrhage (aSAH) is a severe subtype of stroke that accounts for 5 to 10% of strokes. With a mortality rate of close to 50% and a propensity to affect younger individuals, aSAHs are associated with a significant loss of productive lives. This highlights the urgent need to increase the understanding of the epidemiology of aSAH and in turn improve clinical care and functional outcomes. The motivation for this PhD came from being a neurosurgical nurse wanting to understand the geographical and socioeconomic differences in incidence, management and outcome for aSAH patients. Geography has been recognised as a determinant of health however few aSAH studies include people outside metropolitan neurosurgical units (specialty hospital wards caring for neurosurgical patients). This fails to capture early deaths and the impact of geographic, sociodemographic and system-level factors that contribute to incidence, time to treatment and poor outcomes post-aSAH. AIMS This study had four main aims. 1. To explore the validity of coding of aSAH in administrative data. 2. To define aSAH incidence within the Australian population, with consideration of differences by rurality and socioeconomic status. 3. To define time to treatment post aSAH, examining the associations with socioeconomic status, rural place of residence and inter-hospital transfers. 4. To understand early (<24-hour) and late (up to 12-month) survival post aSAH, with a focus on rurality and socioeconomic status. Methods and materials A retrospective, population-based cohort study of aSAH cases in Tasmania, Australia from 2010 to 2014 was created. Cases were identified through searches of multiple overlapping sources, including all regional and direct admissions to Tasmanian hospitals. Data linkage with Ambulance Tasmania and The Registry of Births, Deaths and Marriages ensured that all cases were identified including early fatal cases and cases that were not admitted to the neurosurgical unit. Major exposure variables across chapters were socioeconomic status and rurality, which were defined using spatial data from the Australian Bureau of Statistics. Social, demographic, health and clinical data were extracted from medical records. For aim 1 (Chapter 5), sequential combinations of International Classification of Disease (ICD-10-AM) search algorithms were used. Sensitivity, specificity and positive predictive values (PPV) using the three ICD-10-AM code algorithms to predict confirmed aSAH were calculated. For aim 2 (Chapter 6), Age standardised rates (ASR) of aSAH were calculated using the 2001 Australian population. Poisson regression was used to calculate differences in incidence rate ratios by age, sex, area-level socioeconomic status and geographical location. For aim 3 (Chapter 7), the cohort was used to examine time intervals from ictus and the onset of symptoms to treatment intervals including essential imaging, inter-hospital transfer and intervention times. Linear regression was used to examine the associations between transfer status, place of residence and socioeconomic status and log transformed times to treatment. For aim 4 (Chapter 8), a conceptual framework of rural vulnerability was used to explore interrelationships between demographics, sociodemographic, individual risk factors and survival. Unadjusted differences in survival were examined using Kaplan- Meier survival curves and compared using the log rank test, with cox proportional hazards models applied to identify independent predictors of survival. RESULTS There were 414 possible events identified with n=237 being confirmed as aSAH with n= 185 of these confirmed admissions and n= 52 data linked out of hospital deaths (defined as deaths that occurred out of hospital and without a hospital admission). Chapter 5: Medical record review within the ICD-10-AM codes of I60.0 to I60.9 resulted in a sensitivity and specificity (95% Confidence interval (CI)) of 0.90 (8.86- 0.94) and 0.23 (0.16-0.13) and a PPV of 65.1%. The population-based sensitivity and specificity were 0.74 (0.68-0.79) and 0.25 (0.18-0.33) with a PPV of 67.4%. When analysing the aneurysm morphology, the sensitivity was 0.52 (0.45-0.58) and the specificity was 0.91 (0.85-0.95) with a PPV of 90.5%. Chapter 6: The ASR for aSAH was 9.99 (95% CI 8.69-11.29) per 100,000. There was a significant association between incidence and area-level socioeconomic disadvantage with the rate of aSAH in geographical areas identified as disadvantaged 1.40 times higher than that in areas identified as advantaged (95% CI 1.11-1.82, P= .012). When fitted as a continuous variable, socioeconomic disadvantage measured by Socioeconomic Index for Areas (SEIFA) was associated with a 5.8% lower incidence of aSAH per unit increase in SEIFA (95% CI .91-.98; P= .01). Chapter 7: The median inter quartile range (IQR) time from ictus to intervention was 13.78 (6.48-20-63) hours. Socioeconomic disadvantage was associated with a 1.52-fold increase in time from the onset of symptoms to hospital (p=‚Äöv¢¬ß0.05) and a 1.76-fold increase in time to neurosurgical admission time (P= <0.05). Residing in an outer regional area was associated with a 2.27-fold increase in time from ictus to neurosurgical admission (P= <0.05). Inter-hospital transfer was associated with a 6.26- fold increase in time from ictus to neurosurgical admission (P= <0.05). Chapter 8: The 12-month mortality of the cohort was 52.3% with 54.0% of these deaths occurring in the first 24-hours post ictus, and with this rate similar across demographic, socioeconomic and geographical categories. In an univariable analysis of 12-month survival, outcome was not influenced by socioeconomic status or geographical location. However, the potentially modifiable risk factors of hypertension and hypercholesterolemia (Hazard Ratio (HR) 1.70, 95% CI 0.99-2.91) were significantly associated with 12-month survival in multivariable analyses (hypertension, HR 1.81, 95% CI 1.08-3.03 and hypercholesterolemia, HR 1.71, 95% CI 1.00-2.94). CONCLUSION This study increases knowledge of the epidemiology of aSAH. Improving on previous studies through the use of data linkage this study provided a contemporary estimate of incidence within the Australian population including its relationship to socioeconomic status. The search strategy demonstrated that applying a broad administrative database search algorithm to aSAH results in a number of inaccuracies. The study also provided novel estimates of time to treatment and its relationship not only to individual factors related to rural vulnerability but also health system factors, including regional admissions and inter-hospital transfers. This highlights the need to review the timeliness and organisation of inter-hospital transfers to overcome the potential vulnerability for individuals from rural and disadvantaged areas. Survival to 12-months was not related to socioeconomic status or geographical location but to modifiable risk factors. These findings may inform efforts to improve early identification, treatment and outcome of individuals with aSAH.