A randomized controlled trial to compare the effects of time-restricted eating versus Mediterranean diet on symptoms and quality of life in bipolar disorder

Psychopharmacological and psychotherapeutic treatments for BD have well-documented effects on circadian function. Animal research provides considerable evidence that lithium and antidepressants influence biological mechanisms involved in regulating circadian rhythms [46–50]. Parallel work in humans suggests that changes in neuronal circadian rhythms predict lithium response [51] and that psychosocial treatments for mood disorders decrease REM density [52–54]. Pilot research has shown the benefits of pharmacological agents such as ramelteon and agomelatine, which have selective affinities for the melatonin receptors MT1 and MT2, as adjuncts for mood stabilizers in BD [55, 56]

Multiple behavioral treatments for BD, including interpersonal and social rhythm therapy (IPSRT; [57]), light therapy [58], and interventions designed to restrict activity during dark periods [59, 60], focus on improving sleep and circadian function. Treatments designed to address sleep and circadian rhythms in BD have not been panaceas. Dark therapy, which may require years to successfully implement [59], appeared to help inpatients only when applied early in manic episodes [60]. IPSRT attempts to change the regularity of many different activities, and perhaps due to this complexity, many IPSRT patients are not successful at changing overall rhythms [61], which could explain the mixed findings on its effectiveness relative to control conditions [62, 63]. CBT for insomnia appears helpful for those with BD but does not target broader circadian disruptions [64]. Bright light therapy has shown mixed findings in BD [58]. Accordingly, despite the promise of behavioral and biological approaches to address disrupted circadian rhythms in BD, further treatment development is needed [65].

There is reason to think that TRE would be particularly helpful for those with BD because those with remitted BD demonstrate more irregular eating-related rhythms than individuals without psychiatric or sleep disorders [66]. Although no studies have examined TRE as a BD-specific intervention, researchers have considered mood changes in TRE trials among medical populations without diagnosable mood disorders. In the sole published RCT, depression scores significantly decreased after TRE compared to a control condition [67]. Other work has shown benefits of TRE compared to a control condition for quality of life [68, 69]. Uncontrolled studies of TRE have shown no significant decrease in pre- to post-depression scores [70, 71], although low depressive symptom levels at baseline might help explain these inconsistent findings.

This is a hybrid trial designed to assess acceptability and efficacy. H1: Compared with the control intervention, TRE will lead to improvements in interview-based mania and depression scores immediately postintervention, which will manifest in downstream improvements in self-rated quality of life (QoL) at 6 months compared to baseline. H2: To extend findings to BD, in that TRE will be rated as highly acceptable, and patients will exhibit high adherence. H3: Compared with those of a control intervention, the effects of TRE will be more powerful earlier in the course of the disorder (< 5 years from onset) than later in the course of the disorder. H4: Compared with those of a control intervention, the effects of TRE on symptoms will be more sustained at 6, 9, and 15 months. The effects of TRE on QoL, compared to those of a control intervention, will be sustained for 6, 9, and 15 months. H5: Compared with a control intervention, TRE will lead to improvements in self-rated symptoms of mania and depression at the end of intervention and at 3.5, 6, 9, and 15 months after baseline compared to baseline symptom severity scores.

We will conduct intent-to-treat analyses but will conduct sensitivity analyses to consider effects among those who adhere to interventions.

Individuals interested in participating will be invited to visit the study website, where they can view the study details, view a consent form, and complete basic screening questions for medical and demographic variables (e.g., have been diagnosed with BD, aged 18–65 years, not experiencing medical conditions that would interfere with taking part in TRE; currently report eating 12 or more hours per day at least twice per week; speaking English; and living in an English-speaking country where we have adequate expertise). Individuals will be asked to complete the EDE-QS [77], and those above clinical thresholds (≥ 15) will be excluded, as will those who endorse a history of diagnosis or treatment of eating disorders. Scores above clinical thresholds on the DIAMOND Self Report Screener subscale for psychosis [78], the Alcohol Use Disorder Identification Test (AUDIT; [79]) and the Drug Use Disorder Identification Test (DUDIT; [80]) will be used to indicate who should have diagnostic interviews to assess those exclusion criteria. Questionnaires to assess sleep disturbance will include the Insomnia Severity Index (ISI; [81–83]; ≥ 8 reflective of subsyndromal insomnia), the MCTQ [84] for circadian disturbance (midpoint of sleep on free days < 4.16 h to define early chronotype and > 4.53 to define late chronotype), and irregular sleep or social activity rhythms, as indicated by a score ≥ 30 on the Brief Social Rhythm Scale (BSRS; [85]).

Those who pass the online screening questions will be provided a link to schedule an appointment for an online interview to cover further eligibility criteria. The screening interview will begin with written informed consent. Then, interviewers will assess diagnoses of BD I or II and whether the potential participant is enrolled in (or willing to arrange) medical care for BD following a stable medication regimen, as well as exclusion criteria of active suicidality (CSSR) and current mood episode (DIAMOND; [78]). Those with BD II per the DIAMOND will be permitted to engage in the trial if not taking mood-stabilizing medication, as long as their provider has supported this plan and continues to see them at least once every 3 months.

We will use the Structured Clinical Interview for Sleep Disorders-Revised (SCISD-R; [86]) to assess sleep (hypersomnolence disorder) and circadian sleep–wake disorders (irregular sleep–wake disorder and non-24-h sleep–wake type disorder). We will use the DIAMOND eating disorders modules to assess symptoms since the age of 18. For those who endorse at least one relevant item on the DIAMOND-self report psychosis scale or who surpass thresholds on the AUDIT or DUDIT, we will assess current diagnoses of psychosis, alcohol use disorder, or substance use disorder. Participants who do not have a medical provider for their BD will be provided with support and referral information and invited to recontact us once they have enrolled in treatment. Participants who are in an acute episode or who are actively suicidal (assessed with the C-SSRS, discussed below) will be provided with support around engaging their provider and invited to return for follow-up assessment after one month to reconsider eligibility. Once participants provide us with their permission to contact a medical practitioner, we will send a letter to their practitioner noting that the patient has enrolled in the trial, emphasizing that we are not taking over their care and that we will potentially contact the provider for crisis management.

Participants who meet the eligibility criteria will complete two weeks of daily logs to monitor baseline food intake and timing via a cellphone app. Enrollment in the intervention will be contingent on the adequacy of their dietary logging on the app during the baseline period. Participants will be considered to have adequate logging for a given day if they enter at least two logs at least five hours apart (they will receive information about how to add missed entries). Participants who report a much shorter time window than they did on the screening questionnaire will be reminded to capture each eating epoch, and queried if the interval logged is substantively shorter. Participants’ logging will be reviewed on the first and second days after baseline, and participants will be given the opportunity to troubleshoot with staff as needed. Thereafter, the logs will be monitored at least twice per week. Those who successfully complete logs will receive encouraging messages through the logging app. At the end of the first week, participants with no days of adequate logging will be excluded from the study (unless we identify a barrier that can be addressed). After 10 days, those who have completed less than 50% of their assigned logs will be reminded that study continuation will depend on successful logging and given an opportunity to receive support around barriers to logging; if warranted (e.g., the patient experienced technical or travel barriers), we will extend the baseline period for four additional days.

Participants who complete at least seven days of adequate dietary logs during baseline will be randomly assigned to an intervention. To allow for shopping and preparation, individuals will receive their intervention assignment 4 days before the start of the intervention, with instructions not to begin the diet until the intervention period begins.

Starting on day 5 after baseline, we will gather several daily measures sent by REDCap through text message. Participants will complete the Core Consensus Sleep Diary (Core CSD; [87]) to assess sleep quantity, timing and quality each morning for 10 days until the end of the baseline period. Participants will also complete EMA of 5 prompts per day for 8 days with brief probes to assess mood (detailed below). We will send the CSD and EMA probes on at least 2 free days for each individual (e.g., one Saturday and Sunday). After the 8 days of mood monitoring, participants will be asked to complete day-after food frequency diaries on the final two days of baseline (one free day and one workday) to capture the quantity consumed for each component of the Mediterranean Diet (e.g., fruit, vegetables, whole grains, olive oil).

After at least 3 days of adequate logging, participants will be sent links to schedule the baseline interview (longitudinal interval follow-up evaluation [LIFE]; [76, 88, 89]; Montgomery-Asberg Depression Rating Scale [MADRS]; [90]); Young Mania Rating Scale [YMRS]; [91]; and the short form of the Rey Auditory Verbal Learning Test [RAVLT]; [92, 93]) and will be asked to complete a preintervention baseline battery of online questionnaires for demographic information, symptoms, sleep, and functional information (Patient Mania Questionnaire [PMQ-9]; [94]), Patient Health Questionnaire [PHQ-9]; [95], and Brief Quality of Life in Bipolar Disorder (QoL.BD; [96]), PROMIS Sleep Disturbance and Sleep-Related Impairment [97], Sleep Household Environment and In-Bed Behaviors [98] World Health Organization Disability Assessment Schedule (WHODAS 2.0-Brief; [99]). The participants were also asked to complete measures of potential moderators (Household Food Security Survey [HFSSM]; [100]), Positive Urgency [101], Personality Traits (PID5BF + ; [102]), Risky Families [103], and for women in appropriate age ranges, the Menopause-Specific QoL Sleep [hot-flash] items [104, 105]).

Participants will be asked to complete daily food logs using brief text entries and optional photos with a cellphone app throughout the 8-week intervention period. At 6, 9, and 15 months after the end of baseline, participants will log their food (first intake and last intake) for 10 days in the app. Sleep diaries (core CSD) and EMA data will be collected for 8 days at the midpoint of the intervention. For one weekday and one weekend after the EMA ends, all participants will be asked to provide day-after food frequency diaries to be able to code Mediterranean diet adherence. Symptom severity and LIFE interview assessments will be conducted at the midpoint, at the end of the intervention, and at 6, 9, and 15 months after the end of baseline. Self-rated assessments of symptom severity, QoL, and sleep/circadian variables will be conducted at each of those assessments and at 16 weeks post baseline. The secondary outcome and eating disorder measures will be administered postintervention but not at follow-ups.

Before commencing the study, interviewers will be trained to standardize the administration and to attain high interrater reliability on all observer-rated measures (the DIAMOND, SCISD-R, C-SSRS, LIFE, MADRS and YMRS). Interrater reliability reviews will be conducted monthly on randomly selected recordings throughout the trial. If rater drift is identified, assessors will be retrained.

In addition to the risk management procedures outlined above, we will take several steps to minimize the potential for adverse events. All team members involved in clinical assessments, coaching, or otherwise in direct contact with participants will receive careful training concerning self-harm, suicidality, and elevated eating disorder, mania or depression symptom severity scores. A detailed manual will provide assessment probes, decision rules, referral resources, and contact information for use during emergency situations. Training will emphasize procedures for providing feedback in a clinically sensitive manner, and role plays will be conducted to cover legal and ethical guidelines, including situations in which we do and do not have the authority to share concerns with treatment providers and/or emergency departments. Team members will be trained about the specific situations in which we have permission to break confidentiality and contact their treatment provider or emergency services, as well as be trained in the C-SSRS Interview to assess current intent, plan, and means and history of attempts. Interviewers will be trained to contact a licensed clinical psychologist while on the phone/videoconference if any participant endorses significant suicidal ideation, and clear flow charts will be available that detail “significant suicidal ideation” (e.g., intent, plan, or means are endorsed; history of suicide attempt with current suicidal ideation).

Feedback and referrals for participants will be provided based on the judgment of urgency of the situation. In situations that do not necessitate emergency intervention, the researcher reviews the nature of the symptoms with the participant and provides support for sharing these concerns with their provider. We will encourage sharing information with providers if participants are experiencing symptoms that meet the diagnostic criteria of the DIAMOND or the LIFE for a hypomanic, manic, or depressive episode or if the MADRS score ≥ 19 [108]. More intensive interventions might involve collaborating to remove lethal means or conjointly contacting treatment providers with the participant to facilitate effective communication. As warranted, the researcher will follow-up with the participants to ensure that they were able to receive adequate support. At all times, we balance respect for autonomy with the need for safety. Participants who experience symptoms that would interfere with the demands of the protocol will be granted a break from following the food plan and logging, and interviewers will schedule a time to check in to consider restarting trial engagement.

Throughout the trial, participants will be encouraged to reach the PI with any concerns about the study. Adverse events will be assessed formally post-intervention and at each follow-up. We may also learn of problems as participants complete logs (e.g., food restriction) or interact with coaches.

All sources of information will be monitored in a timely manner to consider potential adverse events. Where clinical deterioration is noted as potentially emerging, researchers will attempt to assess whether emergent symptoms or concerns appear in any way connected to the intervention or study procedures. If EDE-QS symptoms are above the clinical threshold after enrollment, we will include a probe to assess intervention effects on those symptoms (e.g., do you believe that this intervention has triggered food or appearance symptoms for you?), and we will conduct clinical interviews to consider deterioration of symptoms. Participants will be withdrawn from the trial if their participation compromises clinical care as determined by the study’s executive committee. On a case-by-case basis, this might include instances of hospital admission, acute hypomania or mania, psychotic symptoms, active suicidal intent, or increased eating disorder symptoms, particularly where there is concern that deterioration is related to trial engagement.

Adverse events will be reviewed at monthly executive meetings and at least annual meetings of the independent trial committee (who will consider patterns of adverse events by condition). Adverse events will be promptly reported to the Committee for the Protection of Human Subjects within one week.

For all participants, we will take strong measures to protect confidentiality. Our privacy standards match those of the countries that we are recruiting in. For US participants, we have obtained an NIH Certificate of Confidentiality. All study personnel will be required to complete comprehensive training in confidentiality and privacy. When interviewing participants, we will offer meetings by secure, HIPAA-protected zoom calls; participants have the option to turn off their camera during those meetings.

Once a participant is enrolled in the study, a study ID will be assigned, and this ID will be used in place of identifying personal information such as a name on all assessment materials. The list linking the IDs and names will be destroyed after study completion. All data will be encrypted and kept on a password-protected computers that can be accessed only by study personnel. Consent forms, payment information and contact information will be filed separately from the research records. Data for this project will be gathering REDCap, a highly secure and robust web-based research data collection and management system. Features of REDCap that protect participants' privacy and data security include data storage on servers located in ITG's Advanced Computing Center providing locked physical security; use of both an OHSU firewall and a second ACC firewall; data transmissions that are encrypted with industry-standard SSL methods.

TRE involves restricting the window of eating to 10 h per day. To make that happen, people typically avoid eating in the first 1–2 h after they wake up and in the 2–4 h before they go to sleep. Participants will log the timing of their food intake during the baseline weeks. At the time of randomization, they will receive a graph of that data, along with a guided worksheet to help set the eating window that works best for them.

Those with an eating window greater than 14 h will be asked to restrict their eating to 12 h in the first week and then 10 h in the second week. When selecting the eating window, we will ask them to consider family meals and social commitments and will review logs from baseline to help identify those social engagements. We will also review past week sleep schedules, to plan for avoiding food 2–3 h before sleep. In setting the start time for eating in the morning, we will review any special energy demands, such as exercise, and the time periods that might work for eating given family and work commitments. Participants will be reminded not to restrict their caloric intake. As one example, people who habitually snack at night will be coached to consider ways to add calories to their regular mealtimes to avoid hunger. We will also proactively review potential barriers, such as family attitudes toward meal schedules and children’s meal timing. We also will plan ahead for upcoming holidays and vacations that may disrupt following the plan.

During the fasting period, only water, unsweetened herbal tea or black coffee, and prescription medication are permitted. During the feeding window, no restrictions are placed on the type or quantity of food consumed. We will instruct participants to follow their habitual diet within the specified consistent 10-h period and to aim to consume the same number of calories per day as they did during the baseline.

Blogs and orientation materials will provide support for keeping eating within the time window, including ways to share the food plan and goals with family and friends, ways to have easily prepared foods on-hand to avoid delays in eating dinner, and suggestions for portable breakfast ideas that would allow one to eat during a mid-morning break at work. During the third and fourth weeks of the TRE intervention, participants will receive two blogs to provide brief educational content on sleep and circadian rhythms and then two blogs to provide tips for improving sleep and circadian rhythms.

To monitor the timing of eating, participants will be asked to log each time they consume food or beverages (except water) using a cellphone app. Participant logs will be reviewed at least twice per week, and participants will be sent feedback if they appear to not log, to struggle with the time window, or to restrict the time window or caloric intake beyond the recommended goals.

The Mediterranean Diet is a plan for healthy eating based on how people eat and drink in the 16 countries that border the Mediterranean Sea. Individuals will be encouraged to consume vegetables (6 servings/day), fruits (2–4 servings/day), whole grains (daily), legumes (3–4 times per week), nuts (0.5 oz per day), and oily fish (2 servings/week). Participants will be encouraged to choose lean meats and other sources of protein over red meat and processed meats. Sweets (including desserts and refined cereals), alcohol, and wine or alcohol will be labeled extras, and participants will be encouraged to limit their consumption of extras. The Mediterranean diet resource kit provided at the time of randomization will include a list of key foods, a plan for setting feasible goals for change in the first week, an outline for planning grocery shopping, suggestions for following the Mediterranean diet on a budget, healthy snack options, and tips for finding recipes that meet cultural or food allergy guidelines. Interventions will be followed for eight weeks. To match the burden and self-monitoring of the TRE condition, participants in the Mediterranean diet condition will also be asked to log their food intake using the app.

DIAMOND Self Report Screener [78]. We will administer the DIAMOND Self Report Screener psychosis subscale, a well-validated index of psychotic symptoms.

Alcohol Use Disorder Identification Test (AUDIT; [69, 79, 112]). We will administer the AUDIT to screen for potential alcohol use concerns. For individuals who score above the threshold on the AUDIT (≥ 8), we will administer the DIAMOND alcohol use disorder module to further evaluate participant eligibility.

Drug Use Disorder Identification Test (DUDIT; [80, 113]). We will administer the DUDIT to screen for potential drug use concerns. For individuals who score above the threshold on the DUDIT (≥ 6 for men; ≥ for women and nonbinary individuals), we will administer the DIAMOND substance use disorder module to further evaluate participant eligibility.

Insomnia Severity Index (ISI;[81, 83]). The ISI is a 7-item, well-validated index of nighttime and daytime insomnia symptoms.

Short Eating Disorder Examination Questionnaire (EDE-QS;[77]). The EDE-Q is a self-report measure designed to assess a broad range of eating-related symptoms, including restraint, eating concern, shape concern and weight concern [114]. We used the 12-item short form of the EDE-Q. This version has shown strong internal consistency (Cronbach’s α = 0.913, high temporal stability (ICC = 0.93), and scores are strongly correlated with the original EDE-Q scores (r’s > 0.81). It has also shown sensitivity in differentiating people with and without eating disorders.

Munich Chronotype Questionnaire (MCTQ; [84]). The MCTQ is designed to assess chronotype, with a focus on the midpoint of sleep during free days.

Brief Social Rhythms Scale (BSRS;[115]). The BSRS assesses the irregularity of engagement in daily activities such as mealtimes and social interaction on workdays and free days.

Orientation Memory Concentration Test (OMC;[116]). The OMC is a well-validated, brief test of mental status.

Diagnostic Interview for Anxiety, Mood, and OCD and Related Neuropsychiatric Disorders (DIAMOND; [78]). The DIAMOND is a brief structured diagnostic interview for major psychiatric disorders that has shown strong validity and reliability in comparison with other structured diagnostic interviews [78]. We will use the DIAMOND to document bipolar I or bipolar II disorder diagnoses to assess lifetime or current episodes of major depressive disorder and to ensure that we are excluding those with eating disorders (anorexia, bulimia, binge eating disorder or) at any time since age 18. Those whose symptoms on the DIAMOND are consistent with other specified eating or feeding disorders will also be excluded (e.g., repetitive binging or purging that does not meet the frequency or duration criteria). For those with scores above the threshold on the AUDIT (≥ 8) or DUDIT (≥ 6 for men; ≥ 2 for women and nonbinary individuals), we will administer the alcohol use disorder and/or substance use disorder modules as appropriate to consider exclusion criteria. For those who endorse any symptoms of psychosis on the DIAMOND-self report, we will complete the psychosis module of the DIAMOND.

Columbia Suicide Severity Rating Scale (C-SSRS; [75]). The C-SSRS is a brief, commonly used scale designed to cover suicidal ideation and behavior. Moreover, the use of other measures of suicide risk has been well validated. We will assess suicidal ideation, intent, plan and means in the past 2 weeks and lifetime history of suicide attempt using the screening version of the C-SSRS.

Structured Clinical Interview for Sleep Disorders—Revised (SCISD-R; [86]). The SCISD-R is a semi-structured clinical interview used to assess common sleep disorders. We will use the SCISD-R to assess hypersomnolence disorder and circadian sleep–wake disorders.

Our primary measures of mood will be interview-based assessments. We will include self-reported measures of mood in addition, as these allow for more frequent and affordable assessments. We will enrich our understanding of the temporal dynamics of change in these processes with our daily measures of mood (EMA), which will be gathered at baseline and at the midpoint of intervention. For comparison with the findings of other studies, we will also gather the Clinical Global Impression (CGI) and Patient Global Impression (PGI) indices. Our primary outcome measures will be interview-based measures of mood symptoms: the MADRS, YMRS, and LIFE score.

Montgomery-Asberg Depression Rating Scale (MADRS;[117]). The MADRS is a brief, widely used, psychometrically sound measure designed to briefly assess the severity of depressive symptoms. It provides standardized probes and behavioral anchors and attains high interrater reliability across studies of unipolar and bipolar disorder.

Young Mania Rating Scale (YMRS;[91]). The YMRS is one of the most frequently used rating scales for assessing the severity of manic symptoms. Each of the 11 items covers the severity of a core symptom of mania. The ratings reflect participant responses to the interview, coupled with clinician observation.

Longitudinal Interval Follow-Up Evaluation (LIFE; [76, 118]). We will use the LIFE to gather variations in manic mood states across the period between assessments to potentially enrich our ability to “catch” small changes between follow-ups. The LIFE will include probes for rapidly cycling symptoms [115].

We will also include the Wellcome Trust recommended self-report measures for depression and mania, the Patient Health Questionnaire (PMQ-9 for mania and PHQ-9 for depression; [94, 95]. Both are brief scales that have been shown to have high internal consistency and to be sensitive to treatment effects.

We will supplement our gold standard symptom severity interviews with ecological momentary assessment (EMA) probes of emotion states. An EMA has the advantage of providing a potentially more sensitive indicator of early shifts in mood variability [119]. Participants will be sent 5 surveys per day on their smartphones. Participants will be randomly assigned to one of four schedules (adapted to their designated time zone). These schedules allow for three hours between the final survey of the day and their usual bedtime. Surveys will be quasi-randomized during those hours, spaced on average 135 min apart. Participants will have up to 80 min to respond to each survey and will receive one reminder if they do not respond within the first 40 min.

In our EMA, we will use the Mood Zoom probes, which have been validated in BD [120], with additional probes for sleepiness, hyperfocus, and feeling inspired using a parallel format. Our core index derived from dimension reduction of the emotion probes above is the mean variability in daily negative affect (mean square of successive difference). As a secondary test of proposed mechanisms, we will also assess whether TRE, as a circadian intervention, leads to decreased evening levels of arousal (using factor scores). As a secondary outcome, we predict that both indices will decline more from baseline to the intervention midpoint in the TRE condition than in the Mediterranean diet condition and that this level of decrease will predict greater change in symptom severity and quality of life.

We will use the Brief Quality of Life in Bipolar Disorder (QoL.BD) questionnaire as our index of quality of life [121]. QoL. BD was designed to capture a full spectrum of relevant domains for individuals with BD (physical, sleep, mood, cognition, leisure, social, spirituality, finance, household, self-esteem, independence, and identity) and to be sensitive and pragmatic for repeated assessments in the context of treatment. The measure was developed based on previously available measures, with consultation from consumers and international experts in BD. It showed strong internal consistency and correlations with objective measures of functioning [122]. This short version is based on factor-analytic methods, has been validated for online use, and has been shown to be sensitive to change in two RCTs [123, 124].

We will use acceptability probes previously validated in the ORBIT clinical trial for BD [109), rated on a scale from “Strongly Disagree” to “Strongly Agree”. We will supplement these with specific questions to assess whether individuals perceive that their mania, depression, sleep, or quality of life improve or worsen following the intervention. Our primary index of acceptability will be the percentage of individuals who endorse that they agree or strongly agree that they would recommend the food plan to a friend. Ratings of “agree” or “strongly agree” for each item will be considered acceptable. As a further check of potential negative outcomes, participants will be asked open-ended queries regarding problems associated with the intervention. For transparency, we will present a table with the full set of items for evaluating acceptability.

To determine whether TRE specifically shifts sleep and day/night rhythms (hypothesized mechanisms), we will examine changes in sleep (the gold standard CORE CSD measures of quality of sleep and reduced variability in sleep midpoint, along with PROMIS sleep disturbance and sleep-related impairment) and day/night rhythms (BSRS and the MCTQ). To assess whether the sleep and light guidance offered as part of the TRE intervention leads to improved sleep hygiene, we will gather the Sleep Household Environment and In-Bed Behaviors [98].

We chose to include participants taking a broad range of medications to enhance the generalizability of the sample. To examine how medication levels shape intervention effects on circadian and symptom outcomes, we will gather the types of psychotropic and sleep medicines, doses, and adherence levels at each self-report assessment to calculate summary scores for each receptor type. At baseline, we will gather data on self-rated personality symptoms using the Personality Inventory for DSM-5, the Brief Form Plus (PID5BF + ; [102]), the Positive Urgency Scale [143], food insecurity (HFSSM; [100, 144]), Menopause QoL sleep items [104, 105], and early adversity (the Risky Families Scale; [145]) as potential predictors of intervention outcome.

In addition to reviewing careless responses during screening, we will review each self-report bundle for careless responding. We will qualitatively review and consider excluding questionnaire bundles where a person has responded with the same response more than 14 times in a row or has a high Mahalanobis distance score.

Changes in mania, depression, and QoL from baseline to immediately postintervention are hypothesized to be significantly greater in the TRE intervention than in the Mediterranean diet intervention. We will test this using a multivariate repeated measures linear mixed model analysis with random effects. All intervention-related effects will be estimated using an intention-to-treat approach, in which we will examine final scores (where available) for all randomized participants. To support intent-to-treat analyses, we will attempt to gather final assessments for all participants who discontinue the study (except those who explicitly discontinued or were withdrawn on ethical grounds). For H2, we will provide descriptive information, and we will test the hypothesis that relative to the Mediterranean Diet, the TRE condition will be associated with decreased attrition and higher ratings of satisfaction with the intervention. Statistical analyses will be conducted in accordance with the International Conference on Harmonization E9 statistical principles.

To corroborate the robustness of the findings, we will conduct analyses restricted to those participants who completed the baseline and postintervention symptom assessments and who also showed at least 80% adherence to the intervention during the last month of the intervention (early weeks are expected to include growth in the ability to follow both food plans). For TRE, we will define the eating window as the time window during which 95% of the eating events (the 2.5% to 97.5% interval) occur, and then calculate the proportion of days in which the person followed the planned time window. For the Mediterranean Diet, adherence will be quantified as a mean score of 6 or higher (roughly the 90th percentile) on the Alternate Mediterranean Diet score for the two days assessed at the midpoint of intervention [146].

Intervention-related changes in secondary outcomes will be tested using repeated measures linear mixed model analyses to determine whether TRE leads to changes in sleep (core CSD, PROMIS Sleep Disturbance and Sleep-related Impairment), day/night rhythms (BSRS and the MCTQ), or emotion variability (as assessed using ecological momentary assessment) as compared to the Mediterranean Diet.