AbstractBackground: People with severe mental illness die up to 20 years earlier and are more likely to have co-morbid physical illness and poorer health outcomes than a comparative population without these conditions. International and National policy has emphasised the need to integrate healthcare delivery across the primary and secondary care settings to improve healthcare for this population and collaborative care is one approach to integrating services. Collaborative care is considered a 'complex intervention' because it includes a number of different elements. Research suggests that collaborative care can improve symptoms of depression when compared to usual care. The evidence for collaborative care for severe mental illness has, to date, not been reviewed systematically. Aim: The aim of this thesis is to establish whether collaborative care is effective and if it can be implemented within routine clinical practice to improve the health and healthcare of people with severe mental illness. MethodsA mixed methods approach and triangulated design was used to explore the current context in which services are provided (studies 1 and 2), review evidence to determine the effectiveness of collaborative care for severe mental illness (study 3) and identify the factors that could facilitate or hinder the implementation of collaborative care through application of Normalization Process Theory (study 4). These studies combined constitute the development stage of the Medical Research Council's complex intervention development framework. Results: An estimated 31% of people with severe mental illness are seen only in primary care. Yet primary care and cross-boundary continuity is poor for this group. Examining consultation rate patterns demonstrated that rates are increasing over time and at a faster rate since the introduction of the Quality and Outcomes Framework. These findings point to the pivotal role of primary care in the context of severe mental illness. One randomized controlled trial of collaborative care for veterans with bipolar disorder (306 participants) indicated effectiveness in terms of reduced psychiatric admissions and quality of life; however, risk of bias was high. Stakeholders 'buy-in' to the core aims of collaborative care but there are notable factors which could hinder implementation within a UK NHS context. Conclusion:The findings from this thesis demonstrate how a mixed- methods approach can be used to inform the development of a complex intervention based on rigorous methods designed to provide complementary data. Findings are synthesised in a series of recommendations to inform the development and testing of collaborative care for severe mental illness.