Effects of total diet replacement programs on mental well-being: A systematic review with meta-analyses

Abstract

This systematic review with meta-analyses assessed the effects of total diet replacement (TDR) programs on mental well-being in clinical trial participants with a body mass index greater than or equal to 25 kg/m² . TDR programs involve replacing all dietary requirements with nutritionally replete formula foods and are generally administered to induce rapid weight loss. To date, it is largely unclear what effects TDR programs may have on mental well-being, particularly in the long-term. To address this, we screened 25,976 references across six databases and extracted 35 publications. These 35 publications provided sufficient data to evaluate the effects of TDR programs on depression, anxiety, stress, positive affect, negative affect, vitality, role-emotional, social functioning, mental health, mental composite summary score, self-esteem, and general psychological health in 24 meta-analyses. Due to the lack of research comparing TDR programs to comparator groups, 22 of our 24 meta-analyses explored change in these mental well-being sub-domains over time in TDR programs without comparators. Specifically, we assessed the change from pre-diet (before the TDR program) to either post-diet (up to and including two months after the TDR program); and/or follow-up (more than two months after the TDR program). For depression and anxiety, we were also able to assess the change from pre-diet to mid-diet (which fell within two weeks of the diet half-way point). The remaining two meta-analyses assessed the difference in depression scores between a TDR group and a food-based comparator group from pre-diet to post-diet and from pre-diet to follow-up. Across all meta-analyses, our results found no marked adverse effects of TDR programs on any mental well-being sub-domain. In fact, clear improvements were observed for depression, anxiety, stress, vitality, role-emotional, and social functioning at post-diet. Interestingly, the improvements for depression, vitality and role-emotional were maintained at follow-up. All improvements were observed in meta-analyses without comparators. While the two comparator-based meta-analyses showed no difference between TDR programs and food-based diets in depression symptoms, there was low statistical power. For all meta-analyses containing three or more independent samples, we constructed prediction intervals to determine the range within which the mean of the true effects may fall for future populations. While these prediction intervals varied between sub-domains, we found that mean depression scores are only likely to increase (i.e., depression will worsen) in less than 3% of future TDR interventions which meet our inclusion/exclusion criteria. Taken together, we concluded that for adults with a body mass index greater than or equal to 25 kg/m² , TDR programs are unlikely to lead to marked adverse effects on mental well-being. These findings do not support the exclusion of participants from trials or interventions involving TDR programs based on concerns that these programs may adversely affect mental well-being. In fact, by excluding these participants, they may be prevented from improving their metabolic health and mental well-being.

Keywords: diet; mental well-being; meta-analysis; obesity.

FIGURE 1

Domains, sub‐domains and measurement scales for all extracted data. Note: BDI, Beck Depression Inventory; HADS‐A, Hospital Anxiety Depression Scale (anxiety); HADS‐D, Hospital Anxiety Depression Scale (depression); STAI‐S, State Trait Anxiety Inventory (state); HAM‐A, Hamilton Anxiety Rating Scale; HAM‐D, Hamilton Depression Rating Scale; IDS‐SR, Inventory of Depression Symptomatology (self‐report); NHP, Nottingham Health Profile; PANAS, Positive and Negative Affect Schedule; PSS, Perceived Stress Scale; RAND‐36, RAND 36‐Item Health Survey; RSES, Rosenberg Self‐Esteem Scale; SF‐12, Short Form Health Survey with 12 Items; SF‐36, Short Form Health Survey with 36 Items; SCL‐90‐R, Symptom Checklist‐90‐Revised

FIGURE 2

Combined totals for literature searching, screening and extraction across six databases. Note that some publications were included in at least one meta‐analysis as well as a qualitative analysis for another outcome, thus the totals do not add up to 35 publications. [Correction added on 1 September 2022, after first online publication: In the preceding sentence, the missing word ”as“ was inserted before "a qualitative".]

FIGURE 3

Risk of bias within publications. D1: Quality of Reporting; D2: External Validity; D3: Internal Validity; D4: Selection Bias