While 800,000 people die by suicide every year worldwide, it is estimated that self-harm occurs 10 to 20 times more frequently. However, in Hong Kong there has been no comprehensive epidemiological information relating to self-harm produced since 2004. Prevention measures must be guided by local knowledge, and the findings of previous studies mainly conducted in the West may not be generalizable to Hong Kong. This study aims to investigate the epidemiological features of self-harm in order to understand its pattern and characteristics. Four topics were analyzed: 1) incidence rates and patients’ demographic profile; 2) non-fatal repetition; 3) suicide following self-harm; and 4) temporal variations. Since there exists no citywide surveillance and monitoring system for self-harm patients in Hong Kong, self-harm data are not collected in a timely or systematic manner. Emergency attendance and inpatient admissions data of public hospitals managed by the Hospital Authority were therefore used to trace relevant medical records. It should be noted that, as hospital administrative data are not collected for research purpose, there have been some concerns about the feasibility of using the data for a study of self-harm. Thus this study also examined the features of the best available hospital data, finding some evidence that these records offer a representative cohort of self-harming patients for research purpose. Between 2002 and 2011, it was found that the average annual incidence rate of self-harm in Hong Kong was 116.0 per 100,000 people. This estimate is a lot lower than those in the West, but is at the upper end of findings in Asian countries. The risk of non-fatal repetition was 6.9% within the year following the index episode of self-harm. Using a survival model specifically for recurrent events data, 8.9 cases of repeated self-harm were expected per 100 patients. The corresponding risk of suicide was much lower at 0.6%. A higher risk of repeated self-harm was observed during the first three months, while the high-risk period of suicide lasted for five months. A diminishing seasonality in self-harm was found, from a bi-seasonal pattern in 2002-2006 to a one-cyclic pattern in 2007-2011, with a peak from May to July and anadir in December. A strong holiday variation was detected around Lunar New Year, but only among males. The temporal variation in Hong Kong due to holidays revealed a different pattern from the West. The findings provide the epidemiological pattern of self-harm in Hong Kong, which can assist clinical assessment and suicide prevention strategy, and also offer directions for future research. The latter include the extraordinary increase in self-poisoning among males in 2003 and 2007; the cost burden on the healthcare and social system; the association between immediate risk of repetition and hospital stay; frequent repeaters of self-harm; and explanations of reported temporal patterns. The results also suggest the importance of implementing a surveillance system to better collect information on self-harming patients, not only for research and evaluation study but also to facilitate management and service provision. / published_or_final_version / Social Work and Social Administration / Doctoral / Doctor of Philosophy

published_or_final_version / abstract / Community Medicine / Doctoral / Doctor of Philosophy

published_or_final_version / Social Work and Social Administration / Master / Master of Social Sciences