Schizophrenia spectrum disorders impact functioning, reduce quality of life and increase the risk of physical illness and premature mortality. Nutritional intervention studies aimed at decreasing body weight have demonstrated efficacy in improving metabolic outcomes; however, few studies have explored the impact of interventions designed to modify diet on mental health outcomes.

To synthesize the existing experimental studies of adjunctive diet modification as an intervention in the treatment of psychotic disorders, analyze findings related to effectiveness and safety, highlight knowledge gaps and limitations, and set forward recommendations for future research studies.

An extensive a priori search strategy was developed and the databases Embase, Embase Classic, Ovid MEDLINE were searched. Screening and data extraction were completed in duplicate. Studies included in this analysis were experimental studies of an adjunctive dietary intervention (overall dietary pattern or education on dietary change) for treatment of schizophrenia spectrum disorders. No restrictions were placed on control groups or blinding. The studies were required to report a mental health outcome.

Twenty-five clinical trials were identified, along with two additional protocols and two meta-analyses. Nineteen of the clinical trials reported improvement in one or more mental health domain including psychosis symptoms, cognition, and quality of life. A high level of heterogeneity was found with respect to patient population, intervention, and study design. All of the studies included lifestyle or psychosocial components in addition to dietary modification. The nutrition advice provided to participants was poorly described overall and compliance was not assessed. The studies that showed benefit tended to have a smaller sample size and were less likely to be randomized but were more likely to use a group delivery intervention.

Further research assessing effectiveness and efficacy of clearly reported dietary interventions is warranted, especially those using rigorous methodology, modifying diet in isolation and assessing participant compliance.

Schizophrenia spectrum disorders (SSD) include schizophrenia, schizoaffective disorder, and other psychotic disorders. The etiology is largely unknown and likely multifactorial with genetic, biological, and social factors implicated. SSD symptoms include positive symptoms (including delusions and hallucinations), negative symptoms (including social withdrawal and apathy), depression, and cognitive symptoms. While many theories exist, these symptoms may be attributed to imbalances in neurotransmitters and their receptors, dysregulation of other neurochemicals such as brain derived neurotrophic factor (BDNF), immune dysregulation, neuro-inflammation, and mitochondrial and metabolic dysfunction[1-3]. Treatment of SSD includes a combination of pharmaceutical and psychosocial interventions including anti-psychotic medications.

Individuals with psychotic disorders are at significantly elevated risk of medical comorbidities, ultimately resulting in an estimated 8 to 20-year reduction in lifespan[4,5]. Several factors contribute to increased risk of mortality such as socioeconomic status, increased tobacco use, low quality diet, and metabolic complications caused by antipsychotic medication[6-10].

Guidelines and programs exist to support SSD patients in dietary choices to manage weight and physical illness; however, nutritional guidelines for the treatment of psychiatric symptoms do not exist. Nutritional psychiatry is a relatively recent field, studying the effects of nutrients on mental illness incidence and progression. An expanding body of evidence including observational[11], mechanistic, and intervention studies[12,13] demonstrates a role of dietary constituents and patterns in the prevention and treatment of mental illnesses, although research in SSD is relatively limited.

Health care providers often accurately report obstacles faced by patients with SSD in the implementation of dietary interventions. Obstacles include motivational and cognitive barriers, poverty, as well as psychotic symptoms[14]. However, dietary intervention trials designed to improve aspects of physical health have documented feasibility and acceptability among SSD patients. One diet intervention study demonstrated an increased vegetable intake and decreased intake of discretionary foods and calories in first episode psychosis patients[5]. Another study involving participants with severe mental illness demonstrated improvements in diet quality and cardiovascular disease risk factors following a Mediterranean diet intervention[15].

Although fewer in number, some intervention studies have assessed mental health symptoms as primary or secondary outcomes. The purpose of the present systematic review is to synthesize the existing experimental studies of dietary modification as adjunctive treatment for psychotic disorders. We sought to analyze findings related to mental health outcomes (efficacy) and safety, highlight knowledge gaps and limitations, and set forward recommendations for future research studies.

An extensive a priori search strategy was developed by an experienced medical information specialist through an iterative process, which included over 300 search terms. Supplementary material appendix 1 contains the full search strategy. The search was executed in Embase, Embase Classic, and Ovid MEDLINE on January 6, 2017 and updated on April 15, 2018 with the removal of a small fraction of search terms that had failed to yield useful results in the initial search (see appendix 1 for details). No limits were placed on language, publication status, or year of publication.

Eligible studies included experimental studies with an intervention that attempted to manipulate participants’ overall dietary pattern (e.g., “healthy diet”, diet counselling, nutrition education), not just an individual type of food, dietary constituent or isolated component (e.g., supplement studies, low carbohydrate diets, gluten-free diets). The studies included participants with psychotic disorders. No restrictions were placed on blinding or the presence of a control group. Studies were required to assess a mental health outcome; studies that measured only physical health outcomes were not eligible.

Title and abstract screening were completed, based on the above criteria, in duplicate using Abstrackr, an online open-source program that facilitates rapid screening decisions and concurrent tagging of results. Disagreement was resolved by consensus by the two principle investigators (Monique Aucoin and Laura LaChance). Studies that failed to meet inclusion criteria on full text review were excluded. Data was extracted in duplicate using a piloted template. The data extracted included: author name, year of publication, geographic location, presence of control group, use of randomization, use of blinding, study population, inclusion of individuals with additional psychiatric illnesses, sample size, intervention description, method of intervention delivery (group or individual), additional interventions provided simultaneously, mental health outcome reported, change in outcome, whether the mental health outcome was identified as the primary outcome, weight or other physical health outcome and adverse events. Because of the significant heterogeneity of the populations, interventions and outcomes, meta-analysis was not attempted. Studies were grouped based on whether or not a positive mental health outcome was reported, the primary outcome of interest. χ² test was used to assess for significance in the differences between these groups of studies based on study design and factors which may contribute to bias.

The search strategy was initially designed for the conduct of a larger scoping review on the topic of dietary factors and psychosis which has been previously published[16].

On January 20, 2020, a search update was completed for any additional studies that met criteria for the present systematic review. Because this update was completed exclusively for the present review, a modified search string using the following terms was executed: (Diet or Dietary or Nutrition or food) and (psychosis or psychotic or schizophrenia).

The initial search and updates identified 76889 results, which were reduced to 55330 after de-duplication. Using Abstrackr to facilitate screening allowed study authors to manually screen slightly less than 50% of results (26053), relying on the artificial intelligence feature of the program to screen the remaining articles. 1290 articles were assessed in full-text form with 822 meeting criteria for the scoping review. Of these, 27 manipulated the dietary pattern of participants and met criteria for inclusion in the present systematic review. The update of the search January 2020 yielded 356 results, of which two met criteria for inclusion (Figure 1).

Open in New Tab   Full Size Figure   Download Figure

Figure 1 PRISMA flow chart.

Description of studies

The majority of the studies were published between 2015 and 2019 (Figure 2).

Open in New Tab   Full Size Figure   Download Figure

Figure 2 Distribution of studies based on year of publication.

Overall, most of the studies were published in the United Kingdom and Europe followed by North America with the remaining published in Australia, Asia, and South America.

Interventions

The details of the interventions themselves were generally poorly described. Most of the interventions consisted of nutritional education that focused on healthy diet habits. Some programs reported more specific macronutrient recommendations while other emphasized general concepts like meal planning and goal setting. Three interventions involved following specific energy-restricted diets. An additional three interventions mentioned education about calories. The interventions were mainly group-based. Five trials implemented interventions that had both individual and group components and five trials consisted of individual interventions only. It was unclear if the interventions were individual or group-based in two of the studies. Nearly all studies failed to identify the qualifications or training of the individual(s) delivering the interventions.

In most of the studies, the participants continued to take their psychiatric medications. All studies supplemented the dietary interventions with exercise and/or psychosocial interventions such as motivational interviewing, smoking cessation, or peer support. The additional interventions that were used alongside diet are displayed in Figure 3.

Open in New Tab   Full Size Figure   Download Figure

Figure 3 Additional interventions used alongside dietary modification in 25 studies.

Outcomes

Of the 27 study protocols described, twelve were designed with mental health symptoms as the primary outcome. The remaining studies described the mental health outcomes as secondary or exploratory.

Outcome measures included quality of life, psychosis symptoms, cognitive symptoms, self-esteem, functioning, depression, mental healthcare service utilization and levels of the biomarker BDNF. Nineteen studies reported at least one positive mental health outcome (Figure 4, Table 1). While several studies reported changes in total psychosis symptoms, three reported improvements in negative symptoms and one in positive symptoms. The prospective intervention trial that assessed prevention of psychotic symptoms in children reported a decrease in positive symptoms at the later outcome assessment[34]; however, given that this study’s design was significantly different compared to the others in this review, these results may be considered separately from the other studies reviewed.

Open in New Tab   Full Size Figure   Download Figure

Figure 4 Direction of findings for specific outcomes. Psy: Psychotic symptoms; QoL: Quality of life; Funct: Functioning; Dep: Depression; Med/use: Medication use or mental health care utilization; Esteem: Self-esteem/self-efficacy/perceived health.

Studies that did not report a positive mental health outcome were more likely to include 200 or more participants, use a control group or randomization, or include participants with diagnoses other than schizophrenia. They were more likely to involve one-on-one intervention rather than group intervention. Compared with those showing benefit, they were less likely to cite mental health symptoms as a primary outcome and less likely to report positive findings in an outcome related to body weight. The differences between studies reporting a benefit to mental health outcomes and those that did not are reported in Table 2. The differences in sample size, group vs individual intervention format and, randomization achieved statistical significance.

	Number of studies reporting benefit (%)	Number of studies not reporting benefit (%)	P value
Studies with greater than 200 participants	4 of 19 (21)	5 of 6 (83)	0.00559
Studies that included patients with non-SSD diagnoses (e.g., Depression, Bipolar, substance use disorder)	8 of 19 (42)	4 of 6 (67)	0.29380
Studies with an individual intervention component	2 of 19 (11)	3 of 6 (50)	0.03509
Studies designed with primary outcome related to mental health	9 of 19 (47)	2 of 6 (33)	0.54599
Studies that used a control	10 of 19 (53)	5 of 6 (83)	0.18081
Studies that used randomization	3 of 19 (16)	4 of 6 (67)	0.01553
Studies with positive finding in an outcome related to body weight	7 of 10 (70)	1 of 4 (25)	0.12428

With respect to duration of the effect, one study reported that positive mental health and weight outcomes persisted at a six month follow up assessment[32], the remainder did not report on long term outcomes.

A meta-analysis of six lifestyle intervention studies, including 289 participants, evaluated changes in levels of peripheral serum and plasma BDNF in patients with SSD and reported some positive findings compared to a control group[42]. A meta-analysis of 25 lifestyle intervention studies, including 1518 SSD participants, reported improvement in depression symptoms[43].

In addition to assessing mental health outcomes, 18 studies assessed physical health outcomes such as body weight, blood pressure, serum cholesterol, waist circumference, oral glucose tolerance, serum cytokine levels, and glycosylated haemoglobin. Nine of 18 studies found improvements in at least one physical measure with body weight, BMI and waist circumference most likely to benefit. It should be noted, that this represents a subset of the research on this topic; studies assessing exclusively physical outcomes were excluded from this review.

This systematic review identified the clinical trials which have attempted to modify dietary patterns in patients with SSD and reported mental health outcomes in response to treatment. The results suggest that these interventions may have promise for improving mental health symptoms in SSD patients. However, there are significant limitations to these studies and opportunities for further research.

Twenty-one of 27 primary and secondary research studies reported significant improvement in terms of quality of life, symptom improvement, increased functioning, decrease mental healthcare utilization and/or improved cognition. Studies with positive findings were generally less likely to include other patient populations (non-SSD), were more likely to have a group component; however, they tended to include smaller sample sizes and were less likely to be randomized.

Adverse events were often not reported and if they were reported, they were not attributed to the dietary intervention itself. Two studies reported worsening mental health in response to treatment. One study[20] noted that participants had to increase their use of medication during the trial. The medication that was typically increased was quetiapine (Seroquel), which is also used as a sleep aid. The authors theorized that the increase in medication may be due to sleep issues rather than increasing psychiatric symptoms as the participants’ sleep did worsen in the trial, possibly attributed to the various stages of withdrawal from substance use. The other study[37] noted that Quality of Life decreased in the intervention group. They attributed this to significant differences in levels of depression at baseline (non-optimal randomization) and a possible regression towards the mean. Furthermore, although the decrease in quality of life was statistically significant, it may not have been clinically relevant as it was a decrease of 1.5 points, on a scale with a range of 12-84 points.

Many of the studies utilized a group setting for the intervention. There is significant research suggesting a role of peer support and motivation in behavioural change[44,45]. In this review, studies using group interventions were more likely to report positive findings. Group delivery of interventions has the added benefit of cost effectiveness[46].

A significant limitation of this body of research is the generally small sample size and low methodological quality; few studies used randomization, blinding, or control groups. However, the most significant limitation is that all studies used multi-modal interventions rather than diet alone. These included exercise, psychosocial interventions, and smoking cessation in addition to medication. As the first two have significant evidence for a role in promoting improved mental health in SSD, it creates difficulty in understanding the exact role that diet modification may have played in achieving the outcomes reported.

Additionally, many of the studies were designed to test interventions for weight loss or other physical conditions such as hypertension, dyslipidemia and glucose dysregulation. Mental health outcomes were secondary or exploratory in most of the studies. As a result, studies may have been statistically powered to detect changes in the primary physical outcomes and not the mental health outcomes.

A significant limitation of this body of evidence is in the reporting of the diet interventions tested. Nineteen studies described the intervention as “a healthy diet” or “nutrition education”. While some studies provided details about the goals of the intervention (such as increased vegetable intake, decreased saturated fat intake), the majority of studies did not. Scientific understanding of a “healthy diet” has changed over time. For example, the American Dietary Guidelines removed cholesterol as a nutrient of concern in 2015 due to lack of evidence that dietary cholesterol impacts serum cholesterol levels[47]. However, previous guidelines had advised patients on limiting exposure to this nutrient. While certain dietary factors such as vegetables and fiber have been consistently viewed as healthy, fats and carbohydrates have experienced shifting attitudes over the previous decades[48]. As a result, studies designed at different times may have recommended significantly different macronutrient ratios. In addition to the role of macro- and micro-nutrients, a body of evidence has identified inflammation, oxidative stress, glycemic balance, food sensitivities, and the gut microbiome as intermediate factors between diet and mental health[49]. These various factors make it difficult to interpret the studies included in this review. It is possible that unique interventions with differing impacts on these pathophysiologic mechanisms contributed to the differing outcomes between studies. Future research may benefit from application of reporting checklists designed to improve the reporting of interventions[50]. A recent systematic review of methodology reporting in lifestyle interventions among people with psychosis also found generally incomplete reporting and put forward a recommendation for use of the CONSORT statement in future publications[51].

While it has been demonstrated that diet patterns in this patient population can be changed in response to nutrition education and counseling, it is widely acknowledged that this is challenging[14]. The lack of compliance data in the studies using educational programs and diet counselling creates significant difficulty in interpreting this body of evidence. It is possible that variability of compliance with the dietary interventions across studies could in part explain differing outcomes reported. Four studies included the provision of some or all of the participants’ food intake and thus had a higher degree of control. These included the prospective prevention study in children and three studies which took place in in-patient settings. Although these studies involved modification to the available food options, there was still a degree of control exercised by participants with respect to the food they consumed. Of note, three of these four studies reported benefit.

Further research should include assessment of compliance. In designing further studies, researchers could consider subjective assessment tools such as the Mediterranean Diet score or Alternative Healthy Eating Index[52] or objective assessment approaches such as plasma vitamin C and serum carotenoids[53] or metabolomic analysis[54]. While speculative, the non-significant association found between studies showing benefit in both mental health and physical health outcomes could be related to participant compliance. Alternatively, the improvement in mental health symptoms may be related to changes in weight, as a bidirectional relationship between the two is known[55]. In the absence of information related to intervention uptake, the studies showing a lack of change to mental health outcomes cannot conclude that diet does not impact psychosis and other mental health outcomes but rather that the educational intervention provided was not able to elicit these changes.

The results of this body of evidence suggest that diet may be a modifiable target for the treatment of psychotic disorders alongside treatment as usual. Dietary interventions were associated with improvements in psychosis symptoms, quality of life, and functioning in some studies. Improved nutrition has also been shown to improve the physical health in a population at high risk of medical illness[5]. There are studies reporting no effect of dietary intervention. These inconsistencies may be attributable to heterogeneity in dietary interventions (including the multi-modal nature), unclear participant compliance with educational interventions and heterogeneous participant populations.

Nutritional interventions have a few clear benefits. They are relatively low in cost, low in risk, and have the potential to benefit the highly prevalent and costly physical comorbidities of schizophrenia spectrum disorders. When interpreted in combination with the more advanced research in other psychiatric populations, such as randomized controlled trial of adjunctive dietary interventions in major depressive disorder, there is justification for more research in this area.

Future studies should aim to examine adjunctive dietary modification without manipulation of other health behaviours, assess compliance using subjective or objective methods, include more homogeneous patient populations, identify mental health symptoms as primary outcomes and be powered accordingly. In addition, high quality experimental studies with a control group, blinding, and randomization are needed in this area of research alongside continuing research exploring effectiveness in real world scenarios. Lastly, in reporting on these studies, clearer description of the experimental dietary interventions would assist in interpreting results, duplication of findings, and guiding further research.