Disordered Eating among People with Schizophrenia Spectrum Disorders: A Systematic Review

Disordered eating, or abnormal eating behaviours that do not meet the criteria for an independent eating disorder, have been reported among people with schizophrenia. We aimed to systemati-cally review literature on disordered eating among people with schizophrenia spectrum disorder (SSD). Seven databases were systematically searched for studies that described the prevalence and correlates of disordered eating among patients with SSD from January 1984 to 15 February 2021. Qualitative analysis was performed using the National Institutes of Health scales. Of 5504 records identified, 31 studies involving 471,159 subjects were included in the systematic review. The ma-jority of studies (17) rated fair on qualitative analysis and included more men, and participants in their 30s and 40s, on antipsychotics. The commonest limitations include lack of sample size or power calculations, poor sample description, not using valid tools, or not adjusting for con-founders. The reported rates were 4.4% to 45% for binge eating, 16.1% to 64%, for food craving, 27% to 60.6% for food addiction, and 4% to 30% for night eating. Positive associations were re-ported for binge eating with antipsychotic use and female gender, between food craving and weight gain, between food addiction and increased dietary intake, and between disordered eating and female gender, mood and psychotic symptoms. Reported rates for disordered eating among people with SSD are higher than those in the general population. We will discuss the clinical, treatment and research implications of our findings.

Extant research has shown consistent associations between diabetes, obesity, and metabolic syndrome, on the one hand, and behaviors such as smoking, substance use, unhealthy dietary choices, and a sedentary lifestyle, on the other [7,14]. People with  1.
(animal* or rat or rats or mouse or mice or rodent* or monkey*).ti. 14.
limit 16 to english language

Study Selection
Search results were exported to Endnote bibliographic management software and duplicates removed. In accordance with pre-determined inclusion and exclusion criteria ( Table 2), records were screened on publication type. Conference abstracts, conference papers, dissertations, irrelevant letters, notes, comments, and book reviews were excluded. All remaining records were uploaded into Covidence systematic review software for further screening. Two reviewers (AS and DV) independently screened the remaining records on title and abstract, then full text based on the predetermined inclusion and exclusion criteria ( Table 2). Conflicts were resolved through discussion between both reviewers and, where necessary, a third reviewer.

Inclusion Criteria Exclusion Criteria
• Participants: patients with a diagnosis of schizophrenia, schizoaffective disorder, schizophreniform disorders and first episode schizophrenia or includes Schizophrenia spectrum disorders. We also included studies that were done in people with mixed diagnosis provided they reported findings for people with SSD. • Interest: We included studies that examined or described "loss of control over eating", binge eating, disordered eating, night eating/nocturnal eating, obesity, overweight, bulimia nervosa, or disordered eating • Comparison: to those without disordered eating or healthy controls. • case-reports or case-series and non-empirical designs; • reviews, book reviews, comments, conference proceedings, abstracts only publications, duplicate publications, and grey literature.

Quality Appraisal
AS and SM undertook the quality appraisal, using items from the National Institutes of Health (NIH) Study Quality Assessment Tools for Case-Control studies, and NIH Study Quality Assessment Tools for Observational Cohort and Cross-Sectional studies [34]. The following items were focused on: (1) type of study design; (2) clear research question with clear aims, objectives, or hypothesis; (3) sample size justification or power calculation; (4) clearly defined study population (inclusion and exclusion criteria and details of the study subjects and setting were discussed); (5) clearly defined and valid outcome measures (objective measures and instruments that are reliable and valid); (6) analyses use appropriate statistical techniques and account for all participants; (7) key confounders measured and adjusted for. Some of the items on NIH scales (e.g., exposure measured prior to the outcome; whether the timeframe was sufficient to expect an association between exposure and outcome; were different levels of exposure measured; and blinding) were less relevant to this systematic review and were therefore dropped. The chosen items allowed for a maximum of 9 points and studies were ranked as poor (1-3 points), moderate or fair (4-6 points), or good in quality (7-9 points) depending on the total points scored.

Data Extraction
A data-extraction checklist was created and AS, KJ, and VM extracted data on: study details (first author's name, year of publication, country studied in), study design, sample characteristics (age and gender distribution, sample diagnosis and diagnostic criteria studied, selection criteria, study setting; and, for case-control studies, the description of controls), outcome variables studied (definition, instruments used, diagnostic criteria if relevant), any confounders adjusted for, and information on missing variables. All papers were reviewed by at least two investigators.
Data were synthesized descriptively. Figure 1 summarized the search and selection process. Our final selection included 31 studies [17,26,[28][29][30] (Table 3). Taken together, these studies included 471,158 individuals (including a single study by Striegel-Moore et al. [58] that encompassed 466,590 men), of whom 2651 had a diagnosis of SSD. A number of studies included patients with SSD as well as other psychiatric disorders but did not describe the exact breakup of the included diagnostic categories [26,28,43,57,58], making it impossible to determine the exact numbers of patients with SSD.    (2), and delusional disorder (1), compared with 20 "non-schizophrenic" healthy men.

Results
Strengths include use of multiple validated tools, and adjustment (although limited) for confounding. Limitations include small sample size, no power calculation, limited observations to variable analysed ratio, and focus on underlying cognitive mechanisms rather than on disordered eating.
Mean age of 30.5 (7.9) years for cases and 29.5 (6.7) years for controls, and all participants were men. Patients treated with SGAs were significantly more likely to be obese and have higher cognitive restraint, disinhibition, and susceptibility to hunger triggered by internal and external cues. The higher disinhibition accounts for overconsumption of food in response to variety of stimuli, and loss of control. Cross-sectional study of 156 outpatients with chronic schizophrenia or schizoaffective disorder on antipsychotic monotherapy for less than 8 years.
The strengths include modest sample size, clear research aims, use of standardised questions/criteria for binge eating.
The limitations include poor sample description, use of single item to screen for night eating, and not adjusting for confounders.
The mean age was 41.7 years. There were 88 men and 68 women. Prevalence of BN was 0, syndromal BED was 4.4%, and sub-syndromal BED was 18.7%. Prevalence of night eating was 30%. BED spectrum and night eating was more prevalent in the clozapine/olanzapine group, and the differences were significant in women. BED spectrum disorders were more common in the first two years of treatment, but night eating was stable over time.

Fawzi and Fawzi, 2012 Egypt
Disordered eating attitudes were measured using eating attitudes test (EAT) 40 To test the hypothesis that disordered eating attitudes co-occur with schizophrenia in a higher frequency and that disordered eating comorbidity would be associated with more severe schizophrenia psychopathology.
A case-control study of 50 consecutive antipsychotic naïve patients with DSM IV schizophrenia recruited from outpatient clinic, compared to 50, age, and gender matched healthy controls with no current or lifetime diagnosis of psychiatric disorder.
The strengths included sample size calculation, well-described sample and selection criteria, use of valid instruments, and adjustment for confounders. The limitation includes the study design.
The mean age of cases was 29.4 (10.2) years and that of controls was 31.1 (10.8) years. 29/50 (58%) of cases and controls were men. Disordered eating attitudes as measured by EAT 40 were two and half times more in schizophrenia than in controls (30% versus 12%). Presence of disordered eating behaviour in patients is associated with higher total PANSS scores, female gender, lower leisure time physical activity, and higher tea/coffee use. To examine the prevalence of food addiction and to explore the associations between participant characteristics and food addiction diagnosis.
Cross-sectional study of final sample of 93 outpatients with diagnosis of schizophrenia from tertiary regional hospital in NSW, Australia.
The strengths include use of valid instrument for food addiction. The authors did not describe their sample, or the diagnostic criteria used, did not adjust for confounders, or provide information about common factors associated with disordered eating.
The authors did not provide mean age. Majority of the sample was between 56 and 65 years of age. There were more males (M:F was 61:32). Most patients were on quetiapine, followed by clozapine and olanzapine.
In total, 27% of the sample met criteria for food addiction and among those who did not meet the criteria, 77.4% endorsed ≥ 3 symptoms but did not report distress or impairment criteria. The most common food addiction symptoms were persistent desire or repeated attempts to cut down.   (3), and schizophreniform (1) disorder who were enrolled in a randomised double-blind controlled trial comparing clozapine and olanzapine. Participants were followed up for 6 weeks.
The strength includes the design. Limitations include poor sample description, small sample size and no power calculation, not providing adequate information on dropouts, and limited information on the scales used.
The mean age of the clozapine group was 36.7 (13) years and for the olanzapine group was 32.8 (8.3) years. There were 18 women and 12 men. The number of patients reporting food cravings and binge eating increased significantly over time. In total, 67% of males and 61% of females reported food cravings and 33% males and 28% females reported binge eating. There were no statistically significant differences between the two antipsychotics on tendency to report food cravings or binge eating. The olanzapine group had numerically more people reporting binge eating, food craving, and weight gain. To study the prevalence of food addiction among schizophrenia patients and to assess whether there is a difference between individuals with and without food addiction in terms of nutritional status and anthropometric assessments.
Cross-sectional study of 104 DSM V schizophrenia patients recruited from the local hospital.
Strengths included decent sample size, use of validated instrument, and clear research aims. Limitations included poor sample description, and no adjustment for confounders.
The average age was 39.4 (10.78) and majority were females (60.8%). In total, 67 patients were on monotherapy and 37 were on more than 2 antipsychotics. Food addiction was found in 60.6% of the sample and more in females (62.9% of females versus 57.1% of men), although this was not statistically significant. In total, 41.3% of those with food addiction were obese and 33.3% were overweight and patients who had food addiction had significantly higher dietary energy intake.  Cross-sectional study of 288 sequential inpatients with DSM IV TR schizophrenia and schizoaffective disorder to study premorbid ED symptoms. They compared premorbid ED in those on treatment versus not on treatment.
The strengths include the sample size, definitions and variables used, and the analyses performed.
There were more men than women (182:106) and the mean age of the group was 32.72 (9.32) years. Women were higher among both the groups (with and without premorbid ED).
In total, 27/288 met the criteria for ED, most commonly AN, followed by BN, and EDNOS. The group with premorbid ED was significantly more psychotic (gustatory hallucinations), and had disorganisation (unusual thought content) symptoms, both during medication free and treatment phases, and higher depression scores in the treatment phase. Cases with ED had significantly higher IQ scores.

Palmese et al. 2013 USA
Night eating questionnaire and night eating interview.
To examine the frequency and clinical correlates of night eating syndrome (NES) in a sample of obese patients with schizophrenia.
Cross-sectional study of 100 obese/overweight patients with DSM IV TR schizophrenia or schizoaffective disorder.
The strengths included decent sample size, use of validated tools, and adjustment for confounders. Limitations include the cross-sectional design and the less than adequate description of selection criteria.
The mean age was 46.5 (10) years and there were more females (61). The mean BMI was 38.2 (7.7) and majority were African Americans (49).
In total, 12% of the sample met full criteria and an additional 10% met partial criteria for NES based on the interview, while the rates with the questionnaire was 8% for full criteria and further 8% for partial criteria. A total of 32% reported little or no control over night eating and 40% reported strong urges to eat. Night eating was associated with insomnia and depression but not current psychotic symptoms, antipsychotic medication use, or substance use disorder.  The aim of the study was to gain insight into the effects of different categories of antipsychotic drugs on the food attitudes of schizophrenia patients.
Cross-sectional study of 153 patients with DSM IV schizophrenia recruited from inpatient and outpatient settings.
The strengths include the sample description, clear selection criteria, use of valid tools, and adjustment for confounders. Limitations include small numbers within individual groups making it hard to compare and use self-report measures.
The mean age was 33.1 (8.7) years and there were 94 men (61.4%). The sample included 33 patients who were antipsychotic naïve. The mean BMI was 25.6 (5.5); 23.5% were overweight and 22.9% were obese. In total, 19% had metabolic syndrome and 37.3% had high waist circumference. Women had higher TFEQ restraint and disinhibition scores than men. Patients on atypical antipsychotics were more sensitive to external eating cues and have a greater tendency towards disinhibition scores. Overweight and obese patients have a higher susceptibility to hunger and disinhibition. DEBQ external eating score negatively correlated to total PANSS score and DEBQ emotional and external eating factors were higher in women compared to men. External DEBQ factor was also higher in atypical antipsychotic group compared to those on conventional antipsychotics and antipsychotic naïve group.

Srebnik et al. 2003 USA
EAT-26 was used to study eating disorders. A cut-off score of ≥20 was used to indicate presence of eating disorder.
This pilot study aimed to describe the prevalence of eating disorder symptoms and the clinical and demographic predictors of those symptoms among adults with severe and persisting mental illness (SPMI) receiving community mental health services.
Cross-sectional study of 149 community mental health participants with SPMI, 38% of who had a diagnosis of schizophrenia spectrum diagnosis.
The strengths include clear aims, use of valid tools, and adjustment for confounders.
The limitations include poor sample description, small sample size for individual illnesses, and lack of clarity on whether there were any diagnosis specific predictors for eating disorders.
The mean age was 40.8 (9.7) years and 51% of the sample were women. In total, 13% of SSD patients had an EAT score of >20. The mean scores and proportions were lower than in bipolar disorder (33.3%) and depression (27.5%).
Purging was more common in SSD, but this was not statistically significant. Female gender and BMI were significant predictors for eating behaviour for the entire sample. SSD formed 47% of the sample. The mean age for this group was 29.6 (5.6) years, and the mean BMI was 26.7 (6), which were greater than that for depressive and substance use disorders. In total, 51.7% of the entire sample were men; the information for SSD is not known. SSD patients scored significantly lower on depression, and anxiety scores and EAT scores. Participants with disordered eating in the entire sample had greater anxiety and depressive scores than those without disordered eating.

Binge Eating
Eleven studies [26,28,36,38,40,[44][45][46]50,53,60] reported on binge eating. All of these studies reported patients who were on antipsychotics. Eight of these were cross-sectional, one case-control, and the other two reported secondary results from multi-site clinical trials. In summary, these studies indicate that people with SSDs on antipsychotic treatment had elevated rates of binge eating, with rates ranging from 4.4% [38] to 16% [40,53] for binge eating disorder (BED) and 8.9% [46] to 45% [50] for binge eating symptoms (not meeting the diagnostic threshold for BED). In general, cross-sectional [45,50,53], case-control studies [26,44] and studies with small sample size [40] reported higher prevalence rates. Aguiar-Bloemer et al. [36] and Bromel et al. [40] describe an increase in binge eating after the commencement of clozapine (through qualitative survey or prospective measures). One patient in the study of Bromel et al. described that their binge eating symptoms completely reverted after temporarily stopping clozapine but recommenced with re-initiation. In these studies, binge eating overall was associated with female gender [26,38,45] or weight gain [28,60].

Food Cravings and FOOD Addiction
Eleven studies [28,35,36,[40][41][42]46,49,50,54,57] provided information on food crav-ings and/or food addiction. All these studies were in people prescribed antipsychotic medication; four [36,42,49,50] were cross-sectional and one case-control [35]. The re-maining six used prospective data or were post hoc analyses of patients recruited to clinical trials. Prevalence rates for food cravings varied from 16.1% [36] to 64% [46] and was highest for sweets (chocolates, candies). Food addiction was reported in 27% [42] to 60.6% of samples [49]; Goluza et al. [47] noted that 77.4% of their patients' reported symptoms of food addiction but did not experience associated distress or impairment. Study design could not account for the variation in rates seen. One study [35] found that patients on typical antipsychotics had higher craving scores than those on olanzapine (although not statistically significant). Craving for fast-food was associated with weight gain [28,41], and food addiction was associated with increased dietary in-take [49]. Stauffer et al. [57] demonstrated that reduction in carbohydrate craving was predictive of weight loss in treatment trials. Food craving inventory (28,35,41) and Yale food addiction scale (42,49) were the most commonly used tools while some studies (36,40,50,57) did not use validated tools.

Night Eating
Night eating in SSD was assessed in four studies [38,[50][51][52] two of which [51,52] were of overweight or obese individuals. The prevalence of night eating ranged from 4% [50] to 30% [38]. It is possible that the high rate reported in the study of de Beaurepaire [38] reflects their employing only a single screening question for night eating. However, Lundgren et al. [51] reported a rate of 27.2% for evening hyperphagia and 50% of their sample reported snacking on waking up in the middle of the night. Palmese et al. [52] reported that their patients were not aware of night eating but many endorsed symptoms of insomnia and depression. The four studies used different tools to study night eating.

Discussion
We undertook a systematic review of disordered eating among people with SSDs. We found 31 studies encompassing 471,158 individuals, of whom 2651 had a reported diagnosis of an SSD. Two of these studies [43,58] included mixed diagnosis (including schizophrenia patients) among those with disordered eating.
Our review of studies, most of which were done in people receiving antipsychotics, found elevated rates of binge eating, night eating, food addiction, food craving, and disordered eating behaviors among people with SSD. Taken together, these studies report a prevalence of 4.4% to 16% for binge eating disorder (BED) and 8.9% to 45% for binge eating symptoms, 16.1% to 64% for food craving, 27% to 77.4% for symptoms of food addiction, 4% to 30% for night eating, and 10% to 41.5% for disordered eating behaviors. The rates for other DEBs were 2.5 to 4 times higher than that in healthy controls. Recent reviews [61,62] suggest lifetime prevalence rate for BED of 1.53% (95% CI 1-2.17) in the general population, and a point prevalence of 1.2-8% for BED and 3.8% to 8.4% for NES in people with type 2 diabetes mellitus. Pursey et al. [63] reported a weighted prevalence of nearly 20% for food addiction in obese adults. A simple comparison shows that the rates for disordered eating are higher in people with SSD; thus, there is a 2 times higher rate for BED, and up to threefold higher for night eating syndrome and food addiction among people with SSD. We also found that patients with SSD often do not fulfil the full diagnostic criteria for these conditions and in particular, do not report distress; hence they may have symptoms without the full diagnosis, and the rates for these are even higher. This lack of distress has been attributed to apathy and negative symptoms of schizophrenia, or the general lack of insight in this population [42], or because specific items such as "eating alone because of feeling embarrassed" did not apply to this population as they tended to live alone [38]. Previous studies in general population have shown that the incidence of DEBs is higher than diagnosable eating disorders, but despite their adverse impact on the individual's functioning [64], are often untreated or missed by healthcare professionals [65]. There is also evidence for an association between disordered eating behaviors and excess eating, weight gain, and association with gender, and symptoms or chronicity.
It is unclear whether disordered eating in SSD is a part of the psychopathology, or whether it is secondary to medications, or an independent comorbid entity. We examine evidence for all three possibilities, below.

Role of Medications
Antipsychotic use, especially second-generation antipsychotics (SGA), is associated with higher risk of weight gain, obesity and metabolic complications [66]. A number of factors including neurotransmitter mechanisms, affect the brain's satiety center, whilst an impact on insulin and gastrointestinal hormones and lifestyle issues have been put forth to explain this association [17]. In addition, disordered eating, such as binge eating and sweet food craving, have been reported in association with SGAs [27,46,67]. Studies that used a prospective design [40,46] demonstrated an increase in food cravings or binge eating over time for people on SGAs. Bromel et al. [40] reported a temporary cessation of binge eating in one patient when the antipsychotic was temporarily ceased. Using a case-control design, Blouin et al. [39], demonstrated a higher tendency for hunger and disinhibition scores for those on SGAs than healthy controls, and Sentissi et al. [55] reported increased reactivity to external food cues for people on SGAs compared to those on typical antipsychotics. These above factors increase the risk of excessive eating and loss of control over eating [44]. It is likely that antipsychotic-induced reduced serotonergic neurotransmission might also play a role in binge eating and food craving [46]; this includes evidence for reduced serotonin transporter binding among obese binge eating women in a SPECT study [68], a role of SSRIs in BED [69,70], and an association of serotonin transport gene polymorphism with BED [71]. Further, the effect of SGAs on the satiety center could account for why patients taking these agents continue to eat and do not feel full after a meal [51]. Schizophrenia patients with simultaneous high restriction and high disinhibition scores are more likely to be overweight and to be treated with atypical antipsychotics [44,55].

Illness Related
The fact that disordered eating behaviors were reported prior to the advent of antipsychotics [24,25,53] and in antipsychotic-naïve individuals [29] lends weight to the corollary that core psychopathology could play a role. This is further strengthened by findings that antipsychotic use did not impact binge eating patterns [50]. Historically, restrictive eating behaviors in schizophrenia were considered to be consequent upon delusions and hallucinations [26,51], whereas BED and BN were thought to be associated with negative, cognitive, and mood symptoms [50,53]. Indeed, disturbances in appetite, eating and weight are not confined to diagnoses classified under "eating and feeding disorders" in schemes such as the DSM (e.g., hyperphagia and deficient appetite noted in mood disorders). In their sample, Malaspina et al. [30] showed that patients who had a history of an ED had more severe psychotic and disorganization symptoms, whereas Treuer et al. [72] suggested that clinical improvement of psychotic symptoms seemed to coincide with increased food intake, while De Beaurepaire [38] reflect an association with improvement in psychotic symptoms. Indirect evidence suggests that perturbations in brain regions that moderate reward mechanisms and dopamine dysfunction in schizophrenia could play a role in food addiction [42,49]. This could also explain the sweet and fast-food craving seen in schizophrenia [73,74]. Alternate explanations for food cravings impairments in schizophrenia include dysregulation of serotonin neurotransmission [75] or altered dopamine/acetylcholine ratio in certain brain areas such as the nucleus accumbens [76].
Finally, the role of comorbid diabetes and disordered eating behaviors in schizophrenia needs to be considered and further studied. It is well accepted that rates of type 2 diabetes mellitus is 2-5 times higher in people with schizophrenia than in general population [77]. Although often linked to second generation antipsychotics, the first documented evidence for this association dates back to the late nineteenth century, long before antipsychotics and modern obesogenic diets [78].In a recent systematic review and meta-analysis exploring prevalence of diabetes in first episode psychosis population, Pillinger et al. [79] provide evidence for impaired glucose homeostasis from illness onset in schizophrenia. It is possible that diabetes and schizophrenia share common early developmental risk factors (e.g., preterm birth, gestational diabetes, maternal malnutrition, etc.) suggesting that stress and hypercortisolemia may contribute to the association between the two conditions.
Diabetes mellitus is associated with increased appetite and disordered eating (especially bulimia and binge eating disorder) and this, in turn, is associated with elevated HbA1C and other diabetic complications including retinopathy, neuropathy, lipid abnormalities, ketoacidosis and poor metabolic control [80]. Diabetes is also associated with impairment in satiety and hunger signals. Insulin crosses the blood-brain barrier and acts on hypothalamus to regulate appetite and food intake [81]. Further, anti-diabetic medications have been found to be effective in reducing binge-eating in non-diabetic population [82]. It is therefore possible that, at least in some individuals, antipsychotics (particularly second generation) triggers, increases, or unmasks the vulnerability for diabetes mellitus in people with schizophrenia and leads to changes in the appetite and weight control centers, predisposing the individual to disordered eating behaviors. The inherent symptoms of schizophrenia could further reinforce disordered eating. For example, negative symptoms and hypodopaminergia can induce food addiction, or binge eating, mood symptoms could be associated with comfort eating, and cognitive rigidity could be associated with reduced cognitive restraint.

Other Explanations
A third explanation for the association of schizophrenia with disordered eating, they are distinct entities that coexist and overlap by chance [29]. Disordered eating problems are often missed because clinicians focus on psychotic symptoms, and/or because the disordered eating is not severe or typical enough to meet the diagnostic threshold for an independent eating disorder label. Some commentators have conceptualized DEBs as a physiological compensatory mechanism in reducing psychotic symptoms; this is based on the premise that starvation induces psychosis [83]. DEBs often predate psychosis and share common risk factors including developmental trauma [38]. A shared genetic susceptibility for schizophrenia and anorexia nervosa has also been described [84].

Strengths and Limitations
The strengths of this review include the comprehensive search strategy and use of multiple authors for quality appraisal and data extraction. Limitations included limiting to only published literature (no grey literature) and papers in English. Whilst the impact of the latter may have been mitigated by the inclusion of multiple studies from non-English speaking backgrounds, adding grey literature can lower the impact of publication bias [85]. Our review also captured a different literature to the scoping review by Stogios et al. [32] and there were only ten studies in common [35,[39][40][41]44,46,47,54,55,60]. We believe therefore, that our findings complement and lend additional weight to their findings.
We did not undertake a meta-analysis as the studies were highly heterogeneous, differing in the types of disordered eating studied, with differing methodologies, definitions, and thresholds. The quality of included studies was largely fair to good and we tried to link the information from individual data to the methodology, outcome measures, and qualitative analysis to give weight to the conclusions drawn. We had amended the NIH scale as some of the items were less relevant to our review and were therefore dropped. We do not believe this could have affected our quality rating.
While we did not exclude studies that focused on anorexia nervosa and bulimia nervosa, this review focused on disordered eating behaviors. There are overlaps between the different eating disorders and there is a possibility of diagnostic crossover or migration between the different disorders [86,87]. However, we are not aware of any studies that have necessarily studied the stability of eating disorder or disordered eating in SSD. Our review included studies from 1985 to 2021; our knowledge and understanding of eating disorders have evolved over this time and this is reflected in how the classificatory systems have changed. For example, the criteria for binge eating in Hay and Hall's [43] study using DSM III R criteria was "at least two days a week for three months", which changed to "at least two days a week for 6 months" in the DSM-IV, and now is at least one day a week for three months" in the DSM-5. Similarly, night eating syndrome, although initially described by Stunkard in 1955 [88], only found a place under OSFED criteria in the DSM-5. These changes have an impact on what is considered and accepted as disordered eating behaviors.

Clinical Implications
It is likely that multiple factors play a role in the association between schizophrenia with disordered eating. Antipsychotics influence the appetite and satiety centers of the brain and lead to hyperphagia [39,89]. There could be a simultaneous preference for sweet and hyperpalatable foods, which contain high dietary levels of sugar and fat. This can, in turn, have "high addictive potential" and reinforce further disordered eating behaviors [90]. Intrinsic illness factors in schizophrenia such as abnormalities in the mesolimbic reward circuit, or impaired cognitive restraint and executive functioning could mediate further disordered eating patterns in response to increased appetite and cravings.
It is possible that the actual prevalence of disordered eating in people with SSDs is substantially higher than reported in the studies reviewed here. Several factors could account for lower reported rates, including absence of distress or impairment [42], confusing criteria in people with schizophrenia (e.g., teasing out the differences between loss of control and intense desire, or between food craving and increased hunger), as well as lack of clear definitions (e.g., for food addiction). Further, de Beaurepaire [38] reported patients who experienced binge eating as pleasurable, in contrast with others who eat excessively but eat slowly. Hence it is likely that different thresholds or criteria are needed when assessing disordered eating behaviors in people with SSD. Similarly, future studies exploring night eating in schizophrenia should differentiate between nocturnal eating and evening hyperphagia.
Future studies need to examine the role of smoking in obesity-related behaviors [91,92]. SSDs are associated with high rates of cigarette smoking and substance use [15]. This has been partly explained by a putative self-medication hypothesis as a mechanism to offset the "negative" symptoms or hypodopaminergia. It is plausible that DEBs such as food addiction is a behavioral variant of this "impulse-control" issue and reflective of underlying impairments in the reward system. The possible role of smoking on cognition in schizophrenia and the impact on disordered eating needs to be studied further. Similarly, there is a need to study the role of smoking cessation or reduction on eating behaviors in people with SSDs. Studies have shown an association with all phases (acute and chronic) and all symptom domains of schizophrenia (positive, negative, cognitive, and affective). There is therefore a need for further research into disordered eating and in particular whether this forms an integral part of the symptoms of schizophrenia as discussed by Bleuler and Kraeplin more than a hundred years ago [24,25] and if disordered eating should be included among diagnostic criteria for schizophrenia. It is particularly important to look for "disordered eating" rather than an "eating disorder" as patients with SSD may not fulfil all the criteria for eating disorders, while they could still experience symptoms or manifest features of an eating disorder. An association between choking risk in schizophrenia (and in particular with antipsychotic use) [93] and choking risk with eating rapidly [94] have been described. The role of disordered eating behaviors in choking risk in SSD needs to be studied further.
As the rates for disordered eating are higher among people with SSD, it is important to train clinicians to look for, assess and manage eating disorders at various stages of the illness, particularly modelled along the lines of those for substance-use disorders [56]. Patients with SSD should be offered healthy lifestyle and dietary advice to develop skills to respond to physiological urges to eat, encouraged to eat on a regular schedule, eating on a budget, focus on reducing sugary food, and improve exercise and physical activity [51,56].
Cognitive behavioral therapy has shown to be effective for binge eating in a randomized controlled study of patients with antipsychotic-induced weight gain [95].
Pharmacological strategies have a role in mitigating antipsychotic induced weight gain. For example, preliminary research shows that naltrexone can be useful in antipsychotic induced weight gain [96] and targets both the endorphin and dopamine systems and could play a role in food cravings [97]. Naltrexone can reduce the rewarding aspects of food and reduce food consumption and reduce appetite [98]. It is important to consider that at least some studies have described that people with SSD did not report "distress" or "lack of impairment"; the role, place, and type of intervention needs to be studied further to elucidate whether intervention is warranted and what the best form of intervention is.
Studies have shown an association between SSD and higher scores on cognitive restraint, disinhibition, and hunger [37,39,47,48,55]; higher cognitive restraint is associated with lower BMI [48] and with weight loss [99] and better weight loss maintenance in treatment programs [100]. Studies of eating disorders have shown an association between cognitive flexibility and eating disorders, particularly anorexia nervosa [101,102] and a role for cognitive remediation therapy (CRT) for these disorders [103,104]. Although CRT has been studied in schizophrenia [105], its role in addressing disordered eating in schizophrenia has not, to our knowledge, been examined. Future research should explore the role for CRT in the different disordered eating in SSD.

Conclusions
Our systematic review provides evidence for an increased rate of binge eating, night eating, food cravings and addiction, and disordered eating behaviors among SSD patients. While this could be related to the illness itself, most of the included studies were performed in patients receiving antipsychotic treatment; hence, it would be difficult to establish whether this is a primary issue or secondary to medications or symptoms. There is a need for studies that assess disordered eating at baseline in early psychosis and prospectively with the commencement of antipsychotics. Studies need to be powered to be able to study the role of medications and other illness characteristics and should be able to inform the role of disordered eating in people developing metabolic complications.