Supporting someone with bipolar disorder: what the latest research says

A 2017 analysis pooled 12 studies of 5,837 people with bipolar disorder in ordinary care. In the first year, 44% had a new episode of high or low mood. For those who stayed well through that year, the risk in the second year was 19% [1]. In the US STEP-BD study, 48.5% of 858 people who'd recovered had a new episode within two years, and episodes of depression were more than twice as common as highs. Remaining mood symptoms were linked with earlier recurrence [2]. Yet in an Italian study that followed 266 people for four years, about a third had no new episode at all [3]. If they're "better" but still flat, irritable or sleeping badly, encourage them to discuss these changes with their GP or mental health team.

Many people have their own pattern of early changes before an episode, often called a relapse signature: sleeping less, talking faster or pulling away from friends. In a UK trial, 69 people had seven to 12 sessions learning to spot their own early signs and rehearsing an action plan they carried on a card. Over 18 months, fewer had a manic relapse. It took 65 weeks for a quarter of the intervention group to have one, against 17 weeks with usual care. The study found no clear benefit for depressive relapse [4]. It was one small, older trial, but much relapse planning today builds on it, and relatives often notice the changes first.

In a Cardiff study of 3,140 people with bipolar disorder, 20% reported that losing sleep had triggered mania or hypomania; about 12% of those who had experienced depression reported it as a trigger for a low [5]. These were people's recollections, not an experiment. A 2025 study monitored 164 people with a wearable ring for over a year and found changes in sleep and activity around the onset of hypomanic symptoms [6]. It is early research, not a reliable home warning system. Interpersonal and social rhythm therapy, which works on relationships and regular routines, helped people stay well for longer in a 2005 trial of 175 adults with bipolar I disorder [7]. A later trial of 100 people aged 15 to 36 found no clear advantage over specialist supportive care [8]. If sleep changes have preceded episodes before, include them in the plan agreed with their clinician.

In a trial of 120 people who were well at the start, those offered 21 weekly group sessions on the illness, early warning signs and routine had fewer than half as many episodes on average over five years as people in an unstructured group. Ninety-nine of the 120 participants completed that follow-up [9]. The larger UK PARADES trial was less clear-cut. Of 304 people, 58% of those offered a course had a new episode within about two years, against 65% in peer support groups, a gap that could be chance [10]. A 2026 BMJ review also identified group psychoeducation, added to usual treatment, as an effective maintenance option for adults [11]. A lift there each week can make it easier to keep going.

Family-focused therapy brings the person and relatives together to learn about bipolar and practise talking problems through. In a US trial of 101 people, all receiving medication, 35% of those offered 21 sessions over nine months relapsed within two years, against 54% with brief crisis support [12]. A 2020 analysis of 39 trials also favoured learning in a family or group over one to one, though the estimate was very uncertain [13]. A 2026 review of seven studies, with 528 people, found relapse, especially into depression, was more likely in families with a lot of criticism or over-involvement. The authors call this preliminary [14]. These studies cannot establish that families cause relapse. Family sessions offer support and ways to handle communication and problems, and NICE says they should be offered to people who live with, or are close to, their family [15]. If they agree, ask the team what's available.

In a European study of 1,922 people treated for a high, those who kept using cannabis had more new episodes over two years. Outcomes for people who stopped were similar to those who had never used it; this does not prove the risks were identical [16]. In a Norwegian study of 112 people in their first treatment for bipolar, carrying on drinking harmfully went with about three and a half times the odds of a new episode within a year, mostly depression. The study found no clear difference between those who stopped harmful drinking and those who had never done so. It also found no clear independent link between cannabis-use patterns and relapse [17]. Both show links rather than proof, and the Norwegian study was small. Asking how they sleep and feel the day after tends to land better than a lecture, and their team or a local drug and alcohol service can help them cut down.

Bipolar UK's free Mood Tracker records mood, sleep, medication and feelings. The charity says entries are stored on the device and can be exported to share. Its privacy notice also describes optional cloud backups and usage analytics, so check the app's settings and current privacy information [18]. A different app, LiveWell, was tested with coaching alongside usual care in 205 US adults with bipolar I disorder. It did not clearly reduce relapse overall. A promising result in participants who were well without symptoms at the start needs confirmation [19]. This trial does not establish whether Bipolar UK's app prevents relapse. Sharing is their choice; agree what would be useful and for how long:

> "Would you be OK sending me your sleep each week for a month? You can stop whenever you like."

A 2019 review pooled five trials of crisis planning for adults with psychotic illness or bipolar disorder. The risk of compulsory admission over twelve to twenty-four months was about 25% lower relative to usual care; this is not a 25-percentage-point fall, and the review found no clear reduction in total admissions [20]. NICE recommends making a risk-management plan with the person and their carer if possible [15]. Bipolar UK explains how an advance choice document can record wishes about future care [21]. Ask their team how these wishes will be recorded, shared and considered. Our page on planning ahead covers what to put in a plan.

Medication changes are for them and their prescriber to agree. A 1996 observational study of 161 people stopping lithium found that faster stopping was linked with earlier recurrence [22]. It studied lithium specifically, and its short taper schedules should not be used as a stopping guide today. NICE advises gradual stopping, a plan for early signs of relapse and monitoring for two years after long-term medication has stopped [15]. The pace depends on the medicine and the person's circumstances.

Bipolar UK runs groups in person and online, including a group for family and friends. Check the current listing for dates and registration [23]. Its peer support line lets UK residents over 18 who are affected by bipolar book a call or email a question; a UK phone number is needed for calls [24]. The PARADES trial did not establish that a course was better than its similarly intensive peer-support programme overall [10]. That does not show that every community group has the same effect.

Next check due: 3 November 2026. Details to check: Group dates and booking routes.

The 2025 LuBi trial compared morning with midday bright light in people with bipolar depression who were taking mood stabilisers. Thirty-four people were analysed. Both groups received active light after a placebo lead-in, so there was no continuing dummy-light group to test the antidepressant effect. Seven participants developed raised-mood symptoms, including two with hypomania; these improved within three days of reducing the light dose [25]. This was a small, closely monitored study. Use bright light only with the care team's agreement and a plan for monitoring mood.

Sleepio is an online CBT course for insomnia. NICE recommends it as an option in primary care for people who would otherwise receive sleep-habit advice or sleeping tablets [26]. That recommendation is not specific evidence for bipolar disorder. Part of the programme reduces time in bed, and the provider advises people with bipolar disorder to consult their doctor before starting [27].

Evidence for ketogenic diets in bipolar disorder is still limited. In an Edinburgh pilot, 20 of 26 people who started the diet completed it. They were well at the start, and standard depression and mania scores did not significantly improve. The reported link between daily ketones, mood and energy came from just 14 participants [28]. With no control group, this cannot show that keto treats bipolar depression. A report from the same study recorded a serious episode of ketoacidosis in someone also taking a diabetes medicine [29]. ENERGISE-BD, funded in 2026, is comparing a ketogenic diet with NHS healthy-eating guidance [30]. Its register aims for 206 adults and, as checked on 5 October 2026, lists recruitment as open and study completion in July 2030; results are not yet available [31]. Anyone considering the diet should discuss it with their prescriber and a dietitian first.

Next check due: 3 April 2027. Details to check: ENERGISE-BD recruitment, completion date and results.