Acceptability and feasibility of two interventions in the MooDFOOD Trial: a food-related depression prevention randomised controlled trial in overweight adults with subsyndromal symptoms of depression

Objectives We report on the acceptability, feasibility, dose-response relationship and adherence of two nutritional strategies to improve mood (multinutrient supplements; food-related behavioural activation (F-BA)) studied in a randomised controlled depression prevention trial (the Multi‐country cOllaborative project on the rOle of Diet, Food‐related behaviour, and Obesity in the prevention of Depression (MooDFOOD) Trial). We also assessed baseline determinants of adherence and assessed whether better adherence resulted in lower depressive symptoms. Design Randomised controlled trial with a 2×2 factorial design conducted between 2015 and 2017. Setting Germany, the Netherlands, UK and Spain. Participants Community sample of 1025 overweight adults with elevated depressive symptoms without a current episode of major depressive disorder. Main eligibility criteria included age (18–75 years), being overweight or obese, and having at least mild depressive symptoms, shown by a Patient Health Questionnaire Score of ≥5. A total of 76% of the sample was retained at the 12-month follow-up. Interventions Daily nutritional supplements versus pill placebo or an F-BA therapy, delivered in individual and group sessions versus no behavioural intervention over a 1-year period. Primary and secondary outcome measures Primary outcome: self-reported acceptability of the interventions. Secondary outcomes: adherence and self-reported depressive symptoms. Results Most participants reported that the F-BA was acceptable (83.61%), feasible to do (65.91%) and would recommend it to a friend (84.57%). Individual F-BA sessions (88.10%) were significantly more often rated as positive than group F-BA sessions (70.17%) and supplements (28.59%). There were statistically significant reductions in depressive symptoms for those who both adhered to the F-BA intervention and had a history of depression (B=−0.08, SE=0.03, p=0.012) versus those who had no history of depression. Supplement intake had no effect on depressive symptoms irrespective of adherence. Conclusions F-BA may have scope for development as a depression prevention intervention and public health strategy but further refinement and testing are needed. Trial registration number NCT02529423.

I, the Submitting Author has the right to grant and does grant on behalf of all authors of the Work (as defined in the below author licence), an exclusive licence and/or a non-exclusive licence for contributions from authors who are: i) UK Crown employees; ii) where BMJ has agreed a CC-BY licence shall apply, and/or iii) in accordance with the terms applicable for US Federal Government officers or employees acting as part of their official duties; on a worldwide, perpetual, irrevocable, royalty-free basis to BMJ Publishing Group Ltd ("BMJ") its licensees and where the relevant Journal is co-owned by BMJ to the co-owners of the Journal, to publish the Work in this journal and any other BMJ products and to exploit all rights, as set out in our licence.
The Submitting Author accepts and understands that any supply made under these terms is made by BMJ to the Submitting Author unless you are acting as an employee on behalf of your employer or a postgraduate student of an affiliated institution which is paying any applicable article publishing charge ("APC") for Open Access articles. Where the Submitting Author wishes to make the Work available on an Open Access basis (and intends to pay the relevant APC), the terms of reuse of such Open Access shall be governed by a Creative Commons licence -details of these licences and which Creative Commons licence will apply to this Work are set out in our licence referred to above.
Other than as permitted in any relevant BMJ Author's Self Archiving Policies, I confirm this Work has not been accepted for publication elsewhere, is not being considered for publication elsewhere and does not duplicate material already published. I confirm all authors consent to publication of this Work and authorise the granting of this licence. depressive symptoms for those who both adhered to the F-BA intervention and had a history of depression (B = -0.08, SE = .03, p = 0.012) versus those who had no history of depression. Supplement intake had no effect on depressive symptoms irrespective of adherence.
Conclusions: F-BA may have scope for development as a depression prevention intervention and public health strategy but further refinement and testing are needed.

Patient and Public Involvement
The study was supported by a volunteer public advisory group in the UK which provided input to the programme of research at the planning stage. This volunteer group primarily advised us on the design of the informational material to support the trial and the consent to participate process and forms. A patient/lay person representative (Germany) served as a member of our Trial Steering Committee for the duration of the trial.  Depression is currently the leading global health problem 1 with associated annual costs estimated at $1 trillion 2 . Similarly, obesity has become a global pandemic 3 , which as well as being associated with a range of physical health conditions such as diabetes 4 , may also increase the risk for depression 5 . Given the high prevalence of major depressive episodes (MDE) and major depressive disorder (MDD) 6,7 , depression prevention interventions have now become a global priority 8, 9 . One emerging and promising approach to depression prevention falls under the 'nutritional psychology' framework, which suggests that change in diet, nutrition and food-related behaviour may prevent depression 10, 11 . Recent meta-analyses have provided robust observational evidence of the inverse association between diet and depression, showing that higher adherence to a Mediterranean diet is linked to a lower risk of depression 1213 . The data from these meta-analyses included cross-sectional and longitudinal studies: prospective and in particular randomised controlled prevention trials are necessary to test the causal direction of the relationship between diet and depression.
One recent attempt to conduct a large scale randomised controlled trial of diet and nutrition to prevent depression is the MooDFOOD trial 14 . Participants in this 2x2 factorial designed trial were randomised to daily nutritional supplementation versus pill placebo, and to a nutritionally based psychological therapy; a multi-modal food-related behavioural activation intervention (F-BA) versus no behavioural intervention. The F-BA mode of delivery was comprised of a mix of individual and group sessions 14 .
The MooDFOOD trial found that whilst neither intervention had significant effects on incidence of MDE over 12-months, 14   depression at 12-month follow up for those with higher initial severity and prevented MDE for those with higher levels of treatment adherence 14 , as well as reducing unhealthy food consumption and increasing healthy food consumption 15 . There were no beneficial effects of daily nutrient-supplementation on depressive symptoms, and there was even evidence that outcomes were worse for those taking supplements versus placebo.
However, examining intervention outcomes alone does not provide relevant information to evaluate the real-world implementation and dissemination of an intervention, such as whether and in what form patients will use the intervention; how much patients find the intervention helpful and engaging, how adherent patients are with the intervention (i.e., number of treatment sessions attended, or percentage of pills taken) and whether the dose of therapy received influences its outcomes. These are important questions to determine whether the therapy can be implemented beyond the trial, whether it would be used, and what influences its effects. The Medical Research Council (MRC) guidance on delivering complex interventions makes it clear that assessing the feasibility of delivering a complex intervention is an essential but often overlooked step and that interventions are hampered by a range of problems including those around acceptability, adherence to intervention and delivery 16 .
To address this critical issue, we report on the two interventions in more depth, in particular exploring their adherence and acceptability as well as testing for any doseresponse relationships and assessing potential for adherence (both unmoderated and moderated by baseline characteristics) to influence depressive symptoms over time.
We note that in the main trial outcome analysis 17 there was not only no evidence to suggest that the supplements had beneficial effects but also evidence to suggest that they may cause harm. With this in mind, we concentrated on the following hypotheses: 1. Acceptability. Do participants value and like the preventive nutritional intervention? This is critical, as without acceptability, the intervention is unlikely to be taken up and used by the target population in the future. Because the trial was advertised openly as an investigation into two interventions with equal potential to help with depression, we hypothesised that both would be equally acceptable to participants. We hypothesised that individual sessions would be favoured more than group sessions because of their  previously there were no significant beneficial effects on depressive symptoms found for supplements 14 , but also report the effects for supplements here as a comparison analysis. 4. Are there baseline determinants of higher adherence and are there significant moderators of the effect of adherence on outcome? That is, are there different populations or subgroups that may be most likely to benefit from the intervention?
Given previous findings implicating either 'main effects' of the MooDFOOD intervention on dietary behaviour (e.g. 15 ) or interaction effects between intervention, adherence and pre-existing depression and symptom levels 14 we hypothesized a priori that adherence to intervention has a stronger effect on reducing depressive symptoms for those individuals with i) a history of clinical depression, ii) higher baseline depressive symptoms or, iii) lower baseline diet quality scores.

Participants
The present study used the data from the MooDFOOD randomised controlled prevention trial 14

Food-related Behavioural Activation Therapy (F-BA)
The F-BA consisted of a protocol-based intervention incorporating standard approaches of BA proven effective in depression treatment 19 including selfmonitoring, functional analysis, and activity scheduling, with a focus on changing dietary habits to improve mood. F-BA was provided over 21 sessions (15 individual, 6 group) for one year. A range of potential strategies were employed including foodrelated strategies and food group targets to aim for (see Table 3 for details).

Intervention with pills
Patients received either multi-nutrient supplements (1412mg of eicosapentaenoic and docosahexaenoic omega-3 fatty acids (ratio 3:1), 30μg selenium, 400μg folic acid, and 20μg vitamin D3 coupled with 100mg calcium) or placebos provided in two pills per day, to be taken daily for one year. Further details on both interventions have been reported elsewhere 14 20 .

Measures
Bespoke trial acceptability questions were completed at the final 12-month assessment by participants who completed the trial. Participants were asked standardised questions about their experiences and opinions on the trial supplements/placebo. For example, 'How satisfied were you with the supplements?' Similarly, questions relating to the F-BA included 'How helpful was the coaching in improving your mood?' These questions are itemised in Table 1. In addition, participants of the F-BA intervention were asked which mood-related strategies and food targets they had tried and the extent to which they believed they were successful. Dietary intake was assessed with the 250-item GA2LEN food frequency questionnaire 21 . Intake of the eleven MooDFOOD diet quality food groups were summed to obtain a MooDFOOD diet quality score (range: 0 indicating poor adherence, 77 indicating optimal adherence).

Depressive symptoms -Patient Health Questionnaire-9
The PHQ-9 22 is a well-validated nine-item self-report instrument that assesses current symptoms of depression over the last two weeks based on DSM-IV criteria for major depression, which is used extensively in primary care and clinical trials and is proven to be as good at detecting depression as clinician-administered instruments 23 .

Major Depressive Disorder
At baseline, a previous history of MDD was measured using a brief standardised diagnostic interview (duration of 15-30 min): the Major Depression module from the Mini International Neuropsychiatric Interview (MINI) 24 .

Measuring adherence to intervention
Adherence to the F-BA intervention was defined as having attended at least eight out of the 21 sessions. The adherence criterion for the pills was 70% of pills taken over the course of the trial. The latter calculation was based on weights of provided versus returned supplement containers as well as self-reported supplement use 14 .
We also assessed the proportion of F-BA participants who reported attempting either mood-related strategies or food-related strategies and food targets as promoted in the F-BA intervention, as well as the degree to which these attempts were successful.
Better adherence to the F-BA would be evidenced by higher proportions of participants attempting strategies and reporting them as successful. 1. Acceptability. We calculated the percentage of self-report responses to acceptability questions for both pills and F-BA psychological therapy. We compared differences between groups (supplement vs F-BA, individual vs group sessions) using chi-square tests of association.
2. Adherence. We reported the levels of treatment adherence for pills and F-BA and explored different influences on adherence by baseline participant characteristics (gender, age, past history of depression, depressive symptoms, diet quality and BMI) using chi square and t-tests. 3. Dose effects of adherence. We used the total number of F-BA sessions, individual and group separately, as well as the binary variable for overall good F-BA adherence, to predict the longitudinal change in depressive symptoms from baseline (T0) to trial follow up at 12 months (T12). The dose effects of adherence for pills were assessed using the binary pill adherence variable.
These relationships were tested using linear, quadratic and cubic polynomials in linear regression models. T12 depressive symptom level was the outcome, adjusted for T0 depressive symptoms, history of depression, sex and trial site. 4. Moderators of the effect of adherence on outcome. We tested interactions between good F-BA adherence / pill adherence and the three potential moderators outlined in hypothesis 4 (history of depression, baseline depressive symptoms and diet quality score). This was achieved by using separate linear regression models with T12 depressive symptoms as the outcome adjusted for T0 depressive symptoms and including the variables adherence (binary), the three potential moderators and the interaction terms between adherence and the potential moderators. Only significant interactions were followed up. We decomposed any significant interactions by assessing the influence of adherence at different levels of a significant moderator. For the binary moderator (a history of depression) we tested the effect of adherence on outcome for those with a history of depression versus those without. For continuous variables (baseline depressive symptoms and diet quality), we planned to test the effect of adherence at levels of the moderator, equal to and above the median versus below the median average.
The models were run in the Mplus programme using full information maximum likelihood (FIML) estimation to account for missing data, which is a robust unbiased and efficient technique outperforming traditional approaches to missing data 25, 26 .

Food-related behavioural activation
A total of 76% of the sample was retained at the 12-month follow up. The F-BA acceptability questionnaire data were available for a maximum of 60.7% of the F-BA group (311 from a possible 512 individuals randomised). The remainder of participants had either dropped out of the trial and were not available to complete an end of trial questionnaire pack or choose not to answer questions on the F-BA. The results showed that the F-BA was well received by participants, with 68.2% saying F-BA had either met their needs completely or a lot, 18.7% said somewhat and 13.2% said slightly or not at all. Most participants (83.6%) said they were either very satisfied or somewhat satisfied with the F-BA intervention, overall. The majority reported that 'it was easy to do' (65.9%), 'worth their time doing it' (84.9%) and 'most would recommend the F-BA to a friend in similar need' (84.6%). The majority of participants were also satisfied with both the length (82.3%) and frequency (68.5%) of F-BA sessions. Most participants also reported that the burden of homework was "just right" (67.9%) with 27.7% preferring less and 4.5% preferring more.

F-BA strategies, behaviours and targets
The majority of behaviour strategies had been attempted by most F-BA participants (70% to 86% had tried each strategy). The top three most tried strategies were limit your snacks (86.8%), pay attention to habitual snacking (84.6%) and avoid emotional eating (81.7%). See Table 3 Table 3).

Multi-nutrient supplement and placebo pills
Participants reported that it was easy to comply with the daily pill intervention (supplement pill or placebo) (75.4%), although in contrast to the F-BA results, most said that they would not recommend them to a friend in similar need (55.6% No, 44.4% Yes). Participants most often reported a 'neutral' response to how satisfied they were with the supplements (62.3%), with 28.6% to some extent satisfied and 9.1% to some extent dissatisfied. Similarly, the majority were 'neutral' when asked how helpful the supplements were in improving mood (67.5%), where 22.9% said they were to some extent helpful and 9.6% said they were to some extent unhelpful. Participants were significantly less positive about the supplements than the individual F-BA sessions on all equivalent questions including on whether they would refer a friend to the intervention (χ 2 = 146.40, p < .001), their overall satisfaction with the intervention (χ 2 = 385.36, p < .001) and how helpful the intervention was in improving mood (χ 2 = 333.60, p < .001).

Baseline moderators of the effect of adherence on outcome
There was a significant interaction between good F-BA adherence and a history of depression (B = -3.12, SE = 1.07, p = 0.003) such that adhering to F-BA led to a reduction in depressive symptoms only for those with a history of depression (B = - Because we found evidence for the hypothesised significant interaction between good F-BA adherence and a history of depression, we explored the linear dose-response relationship between the combined number of F-BA sessions and a reduction of depressive symptoms, in those with a history of depression. There was a significant association between total number of F-BA sessions and depressive symptoms for those with a history of depression, such that for every additional F-BA session attended, there was an associated average reduction in depressive symptoms (B = -0.08, SE = .03, p = 0.012).

Adherence to supplement and placebo pills
For pill compliance (supplement or placebo), 513 participants (78.8%) were categorised as adherent (defined a priori as taking ≥70% of the supplements during the 12 months).

Discussion
This depression prevention study was administered largely as intended (71% attending 8 or more F-BA sessions; 79% taking 70% or more of the study pills) and we found a reasonable overall retention rate (76%) over the course of the programme (12 months). The results suggest that while both MooDFOOD interventions (F-BA and daily nutritional supplementation) were reasonably acceptable to study participants, they reported being significantly more satisfied with the F-BA than the supplements.
Participants reported that F-BA was significantly more helpful in improving mood and were significantly more likely to refer a friend in similar need. There was evidence of reasonable adherence in both interventions and a large majority of those reporting on F-BA activities said both that they had attempted the mood and food-related strategies contained within it and that these attempts had been to some extent successful.
These findings are ostensibly to some extent contrary to our findings published elsewhere on the effect of the F-BA on food intake using a self-reported food frequency questionnaire 15 . In that analysis, the F-BA changed the eating behaviour for a number of food groups (e.g. increasing fruit and vegetable consumption, whole grains and fish; decreasing the consumption of sugary snacks) but for not for others (e.g. reducing meat, alcoholic beverages, high fat dairy and soft drinks). It should be noted that in the present study, the questionnaire used to assess the F-BA was tapping into beliefs about the successfulness of the strategies used while the food frequency questionnaire, albeit self-report, is effectively a count of actual food consumed over a given period. Future studies should include both subjective reports on interventions and more objective measures indicating behavioural change.  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  60   F  o  r  p  e  e  r  r  e  v  i  e  w  o  n  l  y   19 The overall evidence from the present analysis suggests that the F-BA would be acceptable to the public/patients in a way that the supplements would not. This is in keeping with the main trial results 14 which showed a placebo effect for the supplement trial arm for the secondary outcomes, indicating possible harmful effects, whereas the F-BA was effective in reducing symptoms of depression for those with higher severity at baseline and prevented incidence of depression at high levels of adherence to therapy. In the present analysis, a history of depression moderated the effect of adherence on depressive symptoms. Taken as a whole, the present findings suggest that the F-BA has scope for further testing and development. It is however, too early to make recommendations for it to be rolled out as a public health prevention strategy in overweight adults with subsyndromal symptoms of depression. It is also important to carry out health economics evaluations for such an intervention. Most sessions of the F-BA are a 1:1 face-to-face intervention which involves trained psychologist time and may therefore be more costly to run than other interventions (e.g. internet-based interventions).
Participants reported preferring F-BA significantly more than daily supplement pills, which is consistent with analogous research on patient therapy preference. For example, a meta-analysis has shown a three-fold patient preference for psychological over pharmacological treatments 27 . Similarly, there was a clear preference for individual over group F-BA therapy which is also consistent with previous research on mental health treatment preference in primary care 28 . It is important to take into consideration patient preference as it can predict good outcomes, as shown in a recent study that found that depressed patients who showed a stronger preference for a  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  60   F  o  r  p  e  e  r  r  e  v  i  e  w  o  n  l  y 20 treatment (person-centred counselling versus low intensity CBT) fared better in that treatment compared with the alternative 29 .
It is also important to note that meta-analysis has shown smaller therapeutic effects for group versus individual modality in psychotherapy for depression 30 and this should be factored in to the future development of prevention interventions. Individual F-BA sessions appear to be the best way to ensure such interventions are practically feasible.
An important moderating factor on the effect of good F-BA adherence on depressive symptoms in the present study was having a past history of depression which suggests that future prevention approaches may be more effective when they are selected (targeted at those with prior history or elevated risk) rather than universal (aimed at the general population), which is consistent with prior findings 31 . We also found some evidence to suggest that F-BA session attendance was linearly related to the reduction in symptoms over time, only for those with a history of depression, partially replicating previous findings 30 . The a priori use of a threshold for good F-BA adherence (8/21 sessions) in our moderation analyses showed that, for those with a history of depression, attending eight or more F-BA sessions predicted significant reductions in depressive symptoms. Future research using a similar intervention should therefore consider achieving eight sessions of intervention but continue to test where the best threshold lies. Future work should also continue to test the alternative hypothesis that change in depressive symptoms may affect adherence to intervention levels.
Similarly, in a meta-analysis on the treatment effect of Mindfulness-Based Cognitive Therapy in randomised clinical trials, the risk of relapse in depression was found to be larger in patients with higher initial depressive symptoms 32 . Although the sample in  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  60   F  o  r  p  e  e  r  r  e  v  i  e  w  o  n  l  y   21 the MooDFOOD trial was ostensibly a high-risk sample with elevated depressive symptoms and individuals being overweight or obese, the relatively low incidence of depression observed over 12-months (~10%) 14 suggests the inherent risk was not high. Future research should focus on higher-risk populations in order to improve the feasibility of interventions. For example, participants could be recruited with higher levels of depressive symptoms at baseline and a history of recurrent depression, which would likely confer a higher risk for subsequent depression.
Early work on the dose-response relationship has suggested that as few as eight therapy sessions are sufficient for 50% of depressed patients to reach clinically significant change 33 although more recently there has been a general consensus that somewhere between 13-18 sessions are required 34 . Less is known on the optimal number of sessions for depression prevention, however meta-analytic reviews of intervention studies where the number of sessions ranged from 1-15, found a 21-22% reduction in incidence of depression 31,35 . No dose-response relationship was found in either of the above meta-analyses. An alternative explanation for the results is that those participants who attended no or few F-BA sessions differed in some essential way to those engaged more with the intervention. For example, initial depressive symptom levels may have been much higher for those individuals not engaging.
However, we found no significant difference between those adhering to F-BA and those not adhering to F-BA on any baseline characteristics including a history of depression, lending support to the interpretation that elements in the F-BA caused the reduction of symptoms. Similarly, it is entirely possible and is clinically intuitive that early changes in depressive symptoms in the intervention may predict adherence.
However, in the post-hoc analysis we found no evidence for this association.

Limitations
The present study has a number of limitations. First, although the F-BA intervention was shown to be reasonably acceptable and feasible to deliver, the mediating mechanisms underlying reductions in depressive symptoms in this population are not yet known. These should be teased out as it is possible that the responsible mechanisms are psychological, food-related or both and it is necessary to correctly identify the responsible mechanisms in order to ensure that the intervention is feasible, in the sense that it needs to be maximally effective. Second, it is not clear whether the F-BA approach to intervention is relevant for community samples with a full range of BMI values or only for overweight and obese individuals. This has important feasibility and policy implications and would influence the nature of prevention intervention potentially changing it from targeted or indicated to universal. A major limitation with the self-reported assessment of the F-BA was that this came only at the end of the trial. That is, we did not have intermediate evaluations. Although we were successful in reaching some participants at the end of the trial who had left the trial at an earlier point, it is entirely possible that a proportion of those who had dropped out of the trial and were unreachable at the end point held more negative views than those participants who responded. More broadly, the F-BA approach to depression prevention requires development and refinement through research to answer key questions on modality of delivery (e.g. individual supported by web-based tools and mobile technology) and patient group (is there scope to roll out to different groups including chronic depression to prevent relapse, prevention in adolescents and young people?) before it can be considered for a larger-scale public health intervention. In conclusion, the present study showed that the F-BA depression prevention strategy was preferred by study participants over daily use of supplements, was well-received and largely considered acceptable. It has potential for good reach, was administered as intended, including a reasonable retention rate, and therefore has potential as a depression prevention intervention. Further work is needed before making clear recommendations for depression prevention in this and other groups.     F-BA χ 2 = 39.13 (4)** Note. F-BA Chi squared statistics refer to the difference between individual and group F-BA overall ratings. Tx Chi squared statistics refer to the difference between individual F-BA and the supplement group overall ratings. Pills Chi squared statistics refer to the difference between pill and placebo groups. * P < 0.05 ** P < 0.001. Results were unchanged when using Fisher's exact test for small cell sizes.

Abstract
Objectives: We report on the acceptability, feasibility, dose-response relationship and adherence of two nutritional strategies to improve mood (multi-nutrient supplements; food-related behavioural activation) studied in a randomised controlled depression prevention trial (the MooDFOOD trial). We also assessed baseline determinants of adherence and assessed whether better adherence resulted in lower depressive symptoms.  (70.17%) and supplements (28.59%). There were statistically significant reductions in depressive symptoms for those who both adhered to the F-BA intervention and had a history of depression (B = -0.08, SE = .03, p = 0.012) versus those who had no history of depression. Supplement intake had no effect on depressive symptoms irrespective of adherence.

Design
Conclusions: F-BA may have scope for development as a depression prevention intervention and public health strategy but further refinement and testing are needed.

Strengths and limitations of this study
 A large sample from a multi-country European prospective randomised controlled depression prevention trial was used.  which as well as being associated with a range of physical health conditions such as diabetes 4 , may also increase the risk for depression 5 . Given the high prevalence of major depressive episodes (MDE) and major depressive disorder (MDD) 6, 7 , depression prevention interventions have now become a global priority 8,9 .
One emerging and promising approach to depression prevention falls under the 'nutritional psychology' framework, which suggests that change in diet, nutrition and food-related behaviour may prevent depression 10, 11 . Recent meta-analyses have provided robust observational evidence of the inverse association between diet and depression, showing that higher adherence to a Mediterranean diet is linked to a lower risk of depression 12,13 . The data from these meta-analyses included cross-sectional and longitudinal studies: prospective and in particular randomised controlled prevention trials are necessary to test the causal direction of the relationship between diet and depression. Several randomised controlled trials (RCT) have been conducted demonstrating a positive effect of dietary intervention strategies either in individuals with major depression or with elevated symptoms of depression [14][15][16] , however, there is still a limited number of trials examining the effect of nutritional strategies on the prevention of depression.
One recent attempt to conduct a large scale RCT of diet and nutrition to prevent depression is the MooDFOOD trial 17 . Participants in this 2x2 factorial designed trial were randomised to daily nutritional supplementation versus pill placebo, and to a nutritionally based psychological therapy; a multi-modal food-related behavioural The MooDFOOD trial found that whilst neither intervention had significant effects on incidence of MDE over 12-months, 17 F-BA, relative to no F-BA, reduced symptoms of depression at 12-month follow up for those with higher initial severity and prevented MDE for those with higher levels of treatment adherence 17 , as well as reducing unhealthy food consumption and increasing healthy food consumption 18 . There were no beneficial effects of daily nutrient-supplementation on depressive symptoms, and there was even evidence that outcomes were worse for those taking supplements versus placebo.  Assessing whether community samples and those at risk of common mental health problems such as depression are likely to adhere to nutrition-based interventions to reduce symptoms or increase wellbeing is an important next step, and while work on this issue has begun [14][15][16][17] , more in depth analysis is needed.
To address this critical issue, we report on the two interventions in more depth, in particular exploring their adherence and acceptability as well as testing for any doseresponse relationships and assessing potential for adherence (both unmoderated and moderated by baseline characteristics) to influence depressive symptoms over time.
We note that in the main trial outcome analysis 17 there was not only no evidence to suggest that the supplements had beneficial effects but also evidence to suggest that they may cause harm. With this in mind, we concentrated on the following hypotheses: 1. Acceptability. Do participants value and like the preventive nutritional intervention?
This is critical, as without acceptability, the intervention is unlikely to be taken up and used by the target population in the future. Because the trial was advertised openly as an investigation into two interventions with equal potential to help with depression, we hypothesised that both would be equally acceptable to participants. We hypothesised that individual sessions would be favoured more than group sessions because of their

Participants
The present study used the data from the MooDFOOD randomised controlled The trial has received approval from the local regional ethics committees and was conducted according to Helsinki good practice guidelines 20 .

Patient and Public Involvement
The study was supported by a volunteer public advisory group in the UK which provided input to the programme of research at the planning stage. This volunteer group primarily advised us on the design of the informational material to support the trial and the consent to participate process and forms. A patient/lay person representative (Germany) served as a member of our Trial Steering Committee for the duration of the trial.

Food-related Behavioural Activation Therapy (F-BA)
The F-BA consisted of a protocol-based intervention incorporating standard approaches of BA proven effective in depression treatment 21   participants of the F-BA intervention were asked which mood-related strategies and food targets they had tried and the extent to which they believed they were successful.

Diet quality
Dietary intake was assessed with the 250-item GA2LEN food frequency questionnaire 23 . Intake of the eleven MooDFOOD diet quality food groups were summed to obtain a MooDFOOD diet quality score (range: 0 indicating poor adherence, 77 indicating optimal adherence).

Depressive symptoms -Patient Health Questionnaire-9
The PHQ-9 24 is a well-validated nine-item self-report instrument that assesses current symptoms of depression over the last two weeks based on DSM-IV criteria for major depression, which is used extensively in primary care and clinical trials and is proven to be as good at detecting depression as clinician-administered instruments 25 .

Major Depressive Disorder
At baseline, a previous history of MDD was measured using a brief standardised diagnostic interview (duration of 15-30 min): the Major Depression module from the Mini International Neuropsychiatric Interview (MINI) 26 .

Measuring adherence to intervention
Adherence to the F-BA intervention was defined as having attended at least eight out of the 21 sessions. The adherence criterion for the pills was 70% of pills taken over the course of the trial. The latter calculation was based on weights of provided versus returned supplement containers as well as self-reported supplement use 17 .
We also assessed the proportion of F-BA participants who reported attempting either mood-related strategies or food-related strategies and food targets as promoted in the F-BA intervention, as well as the degree to which these attempts were successful.
Better adherence to the F-BA would be evidenced by higher proportions of participants attempting strategies and reporting them as successful.

Statistical approach
1. Acceptability. We calculated the percentage of self-report responses to acceptability questions for both pills and F-BA psychological therapy. We compared differences between groups (supplement vs F-BA, individual vs group sessions) using chi-square tests of association.
2. Adherence. We reported the levels of treatment adherence for pills and F-BA and explored differences influences in adherence by baseline participant characteristics (gender, age, past history of depression, depressive symptoms, diet quality and BMI) by adherence using chi square and t-tests.
3. Dose effects of adherence. We used the total number of F-BA sessions, individual and group separately, as well as the binary variable for overall good F-BA adherence, to predict the longitudinal change in depressive symptoms from baseline (T0) to trial follow up at 12 months (T12). The dose effects of adherence for pills were assessed using the binary pill adherence variable. The models were run in the Mplus programme using full information maximum likelihood (FIML) estimation to account for missing data, which is a robust unbiased Consistent with views expressed elsewhere 29 , our approach here is to simply report, describe and discuss any tests of significance carried out in order to maintain a balance between type I and type II error rates, rather than apply any corrections to pvalues. We encourage the reader to evaluate the size of effect and regard all findings as tentative until further corroboration 30 .

Food-related behavioural activation
A total of 76% of the sample was retained at the 12-month follow up. The F-BA acceptability questionnaire data were available for a maximum of 60.7% of the F-BA   Table 3).

Multi-nutrient supplement and placebo pills
Participants reported that it was easy to comply with the daily pill intervention (supplement pill or placebo) (75.4%), although in contrast to the F-BA results, most said that they would not recommend them to a friend in similar need (55.6% No, 44.4%

Adherence to supplement and placebo pills
For pill compliance (supplement or placebo), 513 participants (78.8%) were categorised as adherent (defined a priori as taking ≥70% of the supplements during the 12 months).

Discussion
The results suggest that while both MooDFOOD interventions (F-BA and daily nutritional supplementation) were reasonably acceptable to study participants, they reported being significantly more satisfied with the F-BA than the supplements.
Participants reported that F-BA was significantly more helpful in improving mood and were significantly more likely to refer a friend in similar need. A large majority of those reporting on F-BA activities said both that they had attempted the mood and foodrelated strategies contained within it and that these attempts had been to some extent successful. These findings are ostensibly to some extent contrary to our findings published elsewhere on the effect of the F-BA on food intake using a self-reported food frequency questionnaire 18 . In that analysis, the F-BA changed the eating behaviour for a number of food groups (e.g. increasing fruit and vegetable consumption, whole grains and fish; decreasing the consumption of sugary snacks) but for not for others (e.g. reducing meat, alcoholic beverages, high fat dairy and soft The overall evidence from the present analysis suggests that the F-BA would be acceptable to the public/patients in a way that the supplements would not. This is in keeping with the main trial results 17 which showed a placebo effect for the supplement trial arm for the secondary outcomes, indicating possible harmful effects, whereas the F-BA was effective in reducing symptoms of depression for those with higher severity at baseline and prevented incidence of depression at high levels of adherence to therapy. An important moderating factor on the effect of good F-BA adherence on depressive symptoms in the present study was having a past history of depression which suggests that future prevention approaches may be more effective when they adopt a more selective approach (e.g. targeted at those with prior history or elevated risk of major depression), which is consistent with prior findings 31 . Taken as a whole, the present findings suggest that the F-BA has scope for further testing and development. It is however, too early to make recommendations for it to be rolled out as a public health prevention strategy in overweight adults with subsyndromal symptoms of depression. It is also important to carry out health economics evaluations for such an intervention. Most sessions of the F-BA are a 1:1 face-to-face intervention which involves trained psychologist time and may therefore be more costly to run than other interventions (e.g. internet-based interventions). It is also important to note that meta-analysis has shown smaller therapeutic effects for group versus individual modality in psychotherapy for depression 35 and this should be factored in to the future development of prevention interventions. Individual F-BA sessions appear to be the best way to ensure such interventions are effective in realword settings.
We also found some evidence to suggest that F-BA session attendance was linearly related to the reduction in symptoms over time, only for those with a history of depression, partially replicating previous findings 35 . The a priori use of a threshold for good F-BA adherence (8/21 sessions) in our moderation analyses showed that, for those with a history of depression, attending eight or more F-BA sessions predicted significant reductions in depressive symptoms. Future research using a similar intervention should therefore consider achieving eight sessions of intervention but continue to test where the best threshold lies. Future work should also continue to test the alternative hypothesis that change in depressive symptoms make affect adherence to intervention levels.

Similarly, in a meta-analysis on the treatment effect of Mindfulness-Based Cognitive
Therapy in randomised clinical trials, the risk of relapse in depression was found to be larger in patients with higher initial depressive symptoms 36 . Although the sample in the MooDFOOD trial was ostensibly a high-risk sample with elevated depressive symptoms and individuals being overweight or obese, the relatively low incidence of depression observed over 12-months (~10%) 17 suggests the inherent risk was not high. Future research should focus on higher-risk populations in order to improve the feasibility of interventions. For example, participants could be recruited with higher levels of depressive symptoms at baseline and a history of recurrent depression, which would likely confer a higher risk for subsequent depression.
Early work on the dose-response relationship has suggested that as few as eight therapy sessions are sufficient for 50% of depressed patients to reach clinically significant change 37 although more recently there has been a general consensus that somewhere between 13-18 sessions are required 38 . Less is known on the optimal number of sessions for depression prevention, however meta-analytic reviews of intervention studies where the number of sessions ranged from 1-15, found a 21-22% reduction in incidence of depression 31,39 . No dose-response relationship was found in either of the above meta-analyses. An alternative explanation for the results is that those participants who attended no or few F-BA sessions differed in some essential way to those engaged more with the intervention. For example, initial depressive symptom levels may have been much higher for those individuals not engaging.
However, we found no significant difference between the F-BA sessions attendance groups on any baseline characteristics including a history of depression, lending support to the interpretation that elements in the F-BA had caused the reduction of symptoms. Similarly, it is entirely possible and is intuitive, clinically, that early changes A recent RCT using a brief dietary intervention (dietician-delivered intervention via a 13-minute online video) reported a significant reduction in depressive symptoms for young people with elevated symptoms 16 . The results of this study are informative in that they suggest that nutritional interventions (brief ones) are feasible and in young people with elevated depressive symptoms. Therefore, it is possible that a modified
Conclusions: F-BA may have scope for development as a depression prevention intervention and public health strategy but further refinement and testing are needed.

Strengths and limitations of this study
 A large sample from a multi-country European prospective randomised controlled depression prevention trial was used.   1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  60   F  o  r  p  e  e  r  r  e  v  i  e  w  o  n  l  y  Introduction Depression is currently the leading global health problem 1 with associated annual costs estimated at $1 trillion 2 . Similarly, obesity has become a global pandemic 3 , which as well as being associated with a range of physical health conditions such as diabetes 4 , may also increase the risk for depression 5 . Given the high prevalence of major depressive episodes (MDE) and major depressive disorder (MDD) 6, 7 , depression prevention interventions have now become a global priority 8,9 .
One emerging and promising approach to depression prevention falls under the 'nutritional psychology' framework, which suggests that change in diet, nutrition and food-related behaviour may prevent depression 10, 11 . Recent meta-analyses have provided robust observational evidence of the inverse association between diet and depression, showing that higher adherence to a Mediterranean diet is linked to a lower risk of depression 12,13 . The data from these meta-analyses included cross-sectional and longitudinal studies: prospective and in particular randomised controlled prevention trials are necessary to test the causal direction of the relationship between diet and depression. Several randomised controlled trials (RCT) have been conducted demonstrating a positive effect of dietary intervention strategies either in individuals with major depression or with elevated symptoms of depression [14][15][16] , however, there is still a limited number of trials examining the effect of nutritional strategies on the prevention of depression.
One recent attempt to conduct a large scale RCT of diet and nutrition to prevent depression is the MooDFOOD trial 17 . Participants in this 2x2 factorial designed trial were randomised to daily nutritional supplementation versus pill placebo, and to a nutritionally based psychological therapy; a multi-modal food-related behavioural  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  The MooDFOOD trial found that whilst neither intervention had significant effects on incidence of MDE over 12-months, 17 F-BA, relative to no F-BA, reduced symptoms of depression at 12-month follow up for those with higher initial severity and prevented MDE for those with higher levels of treatment adherence 17 , as well as reducing unhealthy food consumption and increasing healthy food consumption 18 . There were no beneficial effects of daily nutrient-supplementation on depressive symptoms, and there was even evidence that outcomes were worse for those taking supplements versus placebo.   1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  60   F  o  r  p  e  e  r  r  e  v  i  e  w  o  n  l  y   6 Assessing whether community samples and those at risk of common mental health problems such as depression are likely to adhere to nutrition-based interventions to reduce symptoms or increase wellbeing is an important next step, and while work on this issue has begun [14][15][16][17] , more in depth analysis is needed.
To address this critical issue, we report on the two interventions in more depth, in particular exploring their adherence and acceptability as well as testing for any doseresponse relationships and assessing potential for adherence (both unmoderated and moderated by baseline characteristics) to influence depressive symptoms over time.
We note that in the main trial outcome analysis 17 there was not only no evidence to suggest that the supplements had beneficial effects but also evidence to suggest that they may cause harm. With this in mind, we concentrated on the following hypotheses: 1. Acceptability. Do participants value and like the preventive nutritional intervention?

Patient and Public Involvement
The study was supported by a volunteer public advisory group in the UK which provided input to the programme of research at the planning stage. This volunteer group primarily advised us on the design of the informational material to support the trial and the consent to participate process and forms. A patient/lay person representative (Germany) served as a member of our Trial Steering Committee for the duration of the trial.  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59

Measures
Measures were collected at baseline (T0), three months (T3), six months (T6) and at  Table 1. In addition, participants of the F-BA intervention were asked which mood-related strategies and food targets they had tried and the extent to which they believed they were successful.

Diet quality
Dietary intake was assessed with the 250-item GA2LEN food frequency questionnaire 23 . Intake of the eleven MooDFOOD diet quality food groups were summed to obtain a MooDFOOD diet quality score (range: 0 indicating poor adherence, 77 indicating optimal adherence).

Depressive symptoms -Patient Health Questionnaire-9
The PHQ-9 24 is a well-validated nine-item self-report instrument that assesses current symptoms of depression over the last two weeks based on DSM-IV criteria for major depression, which is used extensively in primary care and clinical trials and is proven to be as good at detecting depression as clinician-administered instruments 25 .  26 .

Measuring adherence to intervention
Adherence to the F-BA intervention was defined as having attended at least eight out of the 21 sessions. The adherence criterion for the pills was 70% of pills taken over the course of the trial. The latter calculation was based on weights of provided versus returned supplement containers as well as self-reported supplement use 17 .
We also assessed the proportion of F-BA participants who reported attempting either mood-related strategies or food-related strategies and food targets as promoted in the F-BA intervention, as well as the degree to which these attempts were successful.
Better adherence to the F-BA would be evidenced by higher proportions of participants attempting strategies and reporting them as successful.

Statistical approach
1. Acceptability. We calculated the percentage of self-report responses to acceptability questions for both pills and F-BA psychological therapy. We compared differences between groups (supplement vs F-BA, individual vs group sessions) using chi-square tests of association.
2. Adherence. We reported the levels of treatment adherence for pills and F-BA and explored differences influences in adherence by baseline participant characteristics (gender, age, past history of depression, depressive symptoms, diet quality and BMI) by adherence using chi square and t-tests. The models were run in the Mplus programme using full information maximum likelihood (FIML) estimation to account for missing data, which is a robust unbiased and efficient technique outperforming traditional approaches to missing data 27, 28 .
Consistent with views expressed elsewhere 29 , our approach here is to simply report, describe and discuss any tests of significance carried out in order to maintain a balance between type I and type II error rates, rather than apply any corrections to pvalues. We encourage the reader to evaluate the size of effect and regard all findings as tentative until further corroboration 30 .

Food-related behavioural activation
A total of 76% of the sample was retained at the 12-month follow up. The F-BA acceptability questionnaire data were available for a maximum of 60.7% of the F-BA group (311 from a possible 512 individuals randomised).  Table 2 for details.

F-BA strategies, behaviours and targets
The majority of behaviour strategies had been attempted by most F-BA participants (70% to 86% had tried each strategy). The top three most tried strategies were limit your snacks (86.8%), pay attention to habitual snacking (84.6%) and avoid emotional eating (81.7%). See Table 3 Table 3).

Multi-nutrient supplement and placebo pills
Participants reported that it was easy to comply with the daily pill intervention

Baseline differences by adherence groups
There was no evidence that good F-BA adherence (

Discussion
The results suggest that while both MooDFOOD interventions (F-BA and daily nutritional supplementation) were reasonably acceptable to study participants, they reported being significantly more satisfied with the F-BA than the supplements.
Participants reported that F-BA was significantly more helpful in improving mood and were significantly more likely to refer a friend in similar need. A large majority of those reporting on F-BA activities said both that they had attempted the mood and foodrelated strategies contained within it and that these attempts had been to some extent successful. These findings are ostensibly to some extent contrary to our findings published elsewhere on the effect of the F-BA on food intake using a self-reported food frequency questionnaire 18 . In that analysis, the F-BA changed the eating behaviour for a number of food groups (e.g. increasing fruit and vegetable consumption, whole grains and fish; decreasing the consumption of sugary snacks) but for not for others (e.g. reducing meat, alcoholic beverages, high fat dairy and soft drinks). It should be noted that in the present study, the questionnaire used to assess the F-BA was tapping into beliefs about the successfulness of the strategies used while the food frequency questionnaire, albeit self-report, is effectively a count of  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  The overall evidence from the present analysis suggests that the F-BA would be acceptable to the public/patients in a way that the supplements would not. This is in keeping with the main trial results 17 which showed a placebo effect for the supplement trial arm for the secondary outcomes, indicating possible harmful effects, whereas the F-BA was effective in reducing symptoms of depression for those with higher severity at baseline and prevented incidence of depression at high levels of adherence to therapy. An important moderating factor on the effect of good F-BA adherence on depressive symptoms in the present study was having a past history of depression which suggests that future prevention approaches may be more effective when they adopt a more selective approach (e.g. targeted at those with prior history or elevated risk of major depression), which is consistent with prior findings 31 . Taken as a whole, the present findings suggest that the F-BA has scope for further testing and development. It is however, too early to make recommendations for it to be rolled out as a public health prevention strategy in overweight adults with subsyndromal symptoms of depression. It is also important to carry out health economics evaluations for such an intervention. Most sessions of the F-BA are a 1:1 face-to-face intervention which involves trained psychologist time and may therefore be more costly to run than other interventions (e.g. internet-based interventions).
The fact that participants reported preferring F-BA significantly more than daily supplement pills is also consistent with analogous research on patient therapy preference. For example, a meta-analysis has shown a three-fold patient preference for psychological over pharmacological treatments 32 . Similarly, there was a clear  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  It is also important to note that meta-analysis has shown smaller therapeutic effects for group versus individual modality in psychotherapy for depression 35 and this should be factored in to the future development of prevention interventions. Individual F-BA sessions appear to be the best way to ensure such interventions are effective in realword settings.
We also found some evidence to suggest that F-BA session attendance was linearly related to the reduction in symptoms over time, only for those with a history of depression, partially replicating previous findings 35 . The a priori use of a threshold for good F-BA adherence (8/21 sessions) in our moderation analyses showed that, for those with a history of depression, attending eight or more F-BA sessions predicted significant reductions in depressive symptoms. Future research using a similar intervention should therefore consider achieving eight sessions of intervention but continue to test where the best threshold lies. Future work should also continue to test the alternative hypothesis that change in depressive symptoms make affect adherence to intervention levels.
Similarly, in a meta-analysis on the treatment effect of Mindfulness-Based Cognitive Therapy in randomised clinical trials, the risk of relapse in depression was found to be larger in patients with higher initial depressive symptoms 36 . Although the sample in  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  Early work on the dose-response relationship has suggested that as few as eight therapy sessions are sufficient for 50% of depressed patients to reach clinically significant change 37 although more recently there has been a general consensus that somewhere between 13-18 sessions are required 38 . Less is known on the optimal number of sessions for depression prevention, however meta-analytic reviews of intervention studies where the number of sessions ranged from 1-15, found a 21-22% reduction in incidence of depression 31,39 . No dose-response relationship was found in either of the above meta-analyses. An alternative explanation for the results is that those participants who attended no or few F-BA sessions differed in some essential way to those engaged more with the intervention. For example, initial depressive symptom levels may have been much higher for those individuals not engaging.