The aim of this study is to improve outcomes in people with bipolar disorder (BD) by comparing two new online interventions specifically designed to improve quality of life amongst people who have had multiple (10 or more) episodes of BD.

People who have had significant experience with bipolar disorder (defined here as 10 or more episodes) may not benefit from existing psychosocial interventions targeting symptoms and relapse, and may be better served by interventions targeting quality of life (QoL). Our international team of researchers, clinicians and consumers has developed two different online interventions, both of which there is reason to believe will be useful. Both interventions are brief, with 4 weeks of new online content released weekly, plus one additional week of application. This 5-week 'active phase' is supported by email contact with a personal online coach. The remainder of the 6 months of participant involvement in the trial includes continued access to the website (without coaching support) and follow-up assessments. Both arms are equivalent in using cutting-edge internet technologies and design features to help people engage with the therapeutic content and generalise it into their real lives. The websites have been developed following best-practice principles of persuasive system design, and rely heavily on consumer videos, social engagement through discussion boards, personalised feedback, and intuitive content structure to maximise engagement.

Australia's NHMRC has funded a 4-year project (2016-2019) to develop and compare the effectiveness of the two websites in terms of a range of outcomes, primarily QoL. The randomized controlled trial (RCT) will definitively assess the QoL benefits of two websites for late stage Bipolar Disorder. The RCT has been designed to optimise various aims: minimise risk of bias to support definitive scientific findings (internal validity), support ready dissemination should outcomes be positive (external validity, end-user involvement), and to optimally manage the risks inherent in the population being studied. We expect to find definitive evidence of the comparative QoL benefits of the two interventions, and insights about secondary outcomes including self-rated state anxiety, self-rated depression, and clinician-rated depression. A number of clinical and functional secondary outcomes will also be explored, as will hypothesised mediators and baseline moderators of QoL outcomes. Economic analysis based on cost-consequence analysis, and a range of process evaluations will also be conducted.

A total of 300 participants will be block randomised to provide power to identify a small-moderate treatment effect on QoL. Participants will be blinded as to the experimental intervention. The study uses a single-site (internet-based) design, with advertising occurring primarily online, but also through traditional methods via clinical networks of the researchers in Australia, United Kingdom (UK), Canada and the US. Major assessment time points are baseline, post-treatment (primary endpoint), 3 months post-baseline and 6 months post-baseline. Participants will be remunerated for assessments, which include both online questionnaires and a (blinded) semi-structured clinical interview by phone.

A multi-layered risk-management approach has been developed based on our experience with online interventions for bipolar disorder and psychosis. First and foremost, we explain to participants that their participation does not replace usual care, and no emergency assistance is available through the website (a link to the international site unsuicide is provided). This devolving of responsibility to the participant is reinforced by the inclusion criterion of being under the care of a medical practitioner and having access to local emergency services. Second, both intervention sites contain general information about the potential risks (e.g., generating distress) of the interventions, as well as specific alerts to the potential challenges of particular exercises. Third, a comprehensive 'red flag decision tree' has been developed to guide the team's response to any risk issues arising (see Table 2). Finally, any adverse events arising will be reviewed weekly in the trial executive committee.

To ensure ready translation, minimally restrictive inclusion and exclusion criteria will be set.

Inclusion Criteria:

• diagnosis of BD from a mental health professional
• diagnosis of BD (BD I, BD II or Other Specified Bipolar and Related Disorder) confirmed by semi-structured interview using DSM-5 (Diagnostic and statistical manual of mental disorders-5) criteria, excluding criteria that mania/hypomania require abnormalities of activity/energy.
• must have experienced 10 or more episodes of mania, hypomania or depression
• must be under the care of and able to provide phone/mail contact details for a nominated medical practitioner
• must have local access to emergency services
• must have sufficient understanding of written and spoken English
• must have ready daily access to the internet and adequate internet literacy
• aged between 18 - 65 years

Exclusion criteria:

• currently experiencing an episode of depression or hypo/mania
• currently psychotic or actively suicidal

Outcomes Assessors will be blinded as to treatment allocation. To maintain blinding, the Assessor will not be involved in intervention delivery and participants will be instructed not to discuss treatment with their interviewer. When an interview leads to unblinding, the assessor will be replaced.

Participants will be blinded as to the primary hypothesis of which website will have superior benefits for QoL, but will of course be aware of the intervention they receive.