Eating cognitions, emotions and behaviour under treatment with second generation antipsychotics: A systematic review and meta-analysis

Weight gain and metabolic disturbances are frequent in people treated with second generation antipsychotics (SGA). We aimed to investigate the effect of SGAs on eating behaviors, cognitions and emotions, as a possible contributor to this adverse effect. A systematic review and a meta-analysis were conducted following the Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) guidelines. Original articles measuring outcomes relating to eating cognitions, behaviours and emotions, during treatment with SGAs were included in this review. A total of 92 papers with 11,274 participants were included from three scientific databases (PubMed, Web of Science and PsycInfo). Results were synthesized descriptively except for the continuous data where meta-analyses were performed and for the binary data where odds ratios were calculated. Hunger was increased in participants treated with SGAs with an odds ratio for appetite increase of 1.51 (95% CI [1.04, 1.97]; z = 6.40; p < 0.001). Compared to controls, our results showed that craving for fat and carbohydrates are the highest among other craving subscales. There was a small increase in dietary disinhibition (SMD = 0.40) and restrained eating (SMD = 0.43) in participants treated with SGAs compared to controls and substantial heterogeneity across studies reporting these eating traits. There were few studies examining other eating-related outcomes such as food addiction, satiety, fullness, caloric intake and dietary quality and habits. Understanding the mechanisms associated with appetite and eating-related psychopathology changes in patients treated with antipsychotics is needed to reliably inform the development of effective preventative strategies.


Introduction
Antipsychotics are classified into first-generation antipsychotics (FGAs), also known as typical antipsychotics and second-generation antipsychotics (SGAs), the so called atypical antipsychotics.SGAs are the first line of treatment for patients with schizophrenia and are also frequently used as mood stabilizers (Cepaityte et al., 2021;Lieberman, 2004).Whereas FGAs are known to cause extrapyramidal side effects such as akathisia and parkinsonism, whereas SGAs have been reported to cause metabolic side effects including an increase in appetite, weight gain and obesity (Deng, 2013).
Clozapine is regarded as the first SGA, introduced in the 1970s (de Maio, 1972).Other SGAs such as olanzapine were introduced in the 1990s and labelled as "atypical antipsychotics" (Moore et al., 1992).Some SGAs such as clozapine and olanzapine were found to induce a significant amount of weight gain (Alonso-Pedrero et al., 2019;Himmerich et al., 2015).Antipsychotic-related weight gain is thought to be associated with their affinity to histamine H1, dopamine and serotonin receptors (Kim et al., 2007;Kroeze et al., 2003;Roerig et al., 2011).
SGAs differ regarding their effect on body weight.Clozapine and olanzapine are associated with the most weight gain compared to other SGAs (Dayabandara et al., 2017).Quetiapine and risperidone lead to moderate weight gain and amisulpiride, aripiprazole, asenapine, lurasidone and ziprasidone are reported to be weight-neutral in most patients (Alonso-Pedrero et al., 2019;Barton et al., 2020;Dayabandara et al., 2017;Himmerich et al., 2015;Pillay et al., 2018;Ribeiro et al., 2018;Rognoni et al., 2021;Zhao et al., 2016).Additionally, there are individual differences in weight change between different patients who take the same medication.For example, in a study published by Kinon et al. (2001), some patients lost weight, gained no weight, or gained more than 20 kg during 3 years of treatment with olanzapine.
Changes in eating behaviours occur in a wide range of psychiatric disorders (Milaneschi et al., 2017;Wen Chi et al., 2015).For example, patients with acute schizophrenia, depression or anorexia nervosa might show more restrictive eating behaviours during acute episodes and tend to lose weight (Garfinkel et al., 1983), whereas patients with atypical depression or dementia can experience hyperphagia (Hsiao et al., 2013).Dementia was also shown to be related to the development of pica (Wen Chi et al., 2015).When patients recover from acute psychiatric disorders, this recovery is often associated with a normalisation of their eating behaviour and a return to their usual body weight.This phenomenon had already been described in the pre-psychopharmacological era by Kraepelin (1904) and Kryspin-Exner (1947).
Many studies have examined weight gain and weight-related outcomes during treatment with antipsychotics whereas fewer studies have specifically examined eating-related outcomes including eating behaviours, cognitions, emotions and the regulation of appetite.Understanding the mechanism of weight gain may be of value in devising treatment methods to counteract this unwelcome side effect.These could be psychological treatments or pharmacological approaches addressing changes in appetite or food-related behaviour and emotions.

Materials and methods
This systematic review and meta-analysis followed the Preferred Reporting Items for Systematic reviews and Meta-Analysis (PRISMA) guidelines (Page et al., 2021).See S.1 for the PRISMA 2020 main checklist, S.2 for PRISMA 2020 abstracts checklist and Fig. 1 for PRISMA identification of studies flow chart.A systematic search for eligible publications was conducted between the database date of inception until November 1st, 2021, using three databases: Web of Science, PubMed and APA PsycInfo (Ovid).
Additional publications were identified through a manual hand search through reference lists of relevant papers.A protocol and search strategy were developed by H.H. and H.M., which was prospectively registered with the Open Science Framework (protocol accessible at: htt ps://osf.io/6esm8/).

Inclusion and exclusion criteria
Original articles of clinical studies published at any date were included in our review.Case reports, review articles, letters and animal studies were excluded.Only studies written in English were considered in our systematic review.Studies were included if they measured outcomes relating to eating behaviour, food intake, and the regulation of appetite in patients or healthy participants who were exposed to SGAs.The duration of exposure to the SGA was unrestricted.Studies in which the effect of SGAs on weight change was the only measured outcome were also excluded.Objective and subjective outcome measures were accepted in our review.Studies scoring less than 50% in the quality assessmentwhich indicates a high risk of bias -were excluded.For more details about the inclusion and exclusion criteria see S.4.

Study selection
All publications identified by the search were screened independently by two reviewers (H.M. and S.B.).Endnote and Rayyan, an online software for systematic reviews (Ouzzani et al., 2016) were used for the management of the search results.After the removal of duplications of the results, titles and abstracts were screened against the aforementioned eligibility criteria.Queries regarding eligibility were discussed with the wider research team (J.T., H.H. and J.K.) before a decision on inclusion/exclusion was reached.Authors were contacted to provide missing full-texts and/or information where needed.

Data extraction
Data extraction was performed by H.M. and reviewed by S.B. Extracted data included the titles, abstracts, authors and lead author name and contact details, origin and aim of the study, recruitment methods and setting, study design, duration of the study/intervention, funding and possible conflict of interest, inclusion and exclusion criteria, sample size and population characteristics, diagnosis and criteria of diagnosis, intervention/s, comparisons, drug dosage, outcomes and measures, and the main findings.All data were collated in a Microsoft Excel spreadsheet.

Quality appraisal
Two reviewers (H.M. and S.B.) conducted a quality appraisal using the Joanna Briggs Institute critical appraisal tool (Moola et al., 2020).This tool offers different checklists for different study designs: cross sectional, case control, cohort, diagnostic, quasi-experimental and randomized controlled trials (RCTs), facilitating the comparison of quality across study types.The quality appraisal focused on the methodological quality of publications, in order to examine the possibility of bias in the study design, methods and results.The available JBI appraisal tools do not offer versions designed for open-label trials.Therefore, we used the quasi-experiment version to appraise open-label trials.Overall, the quality assessment decision to include or exclude studies was based on the frequency of responses with "no" or "unclear" with a 50% or more indicating high risk of bias and thus the exclusion of a report.The detailed quality appraisal of candidate studies is shown in S.5.

Quantitative analysis
For continuous outcomes such as eating behaviour (restraint, disinhibition) and hunger, meta-analyses were performed where sufficient (two or more) studies were available.Odds ratios were calculated for binary outcomes such as appetite increase.Both the meta-analyses and odds ratios utilized random effects models, using the DerSimonian and Laird method (DerSimonian and Laird, 1986) which was used to calculate standardized mean of differences (SMD) or odds ratios.All quantitative analyses were conducted using the meta set and summarize commands in Stata 16 (StataCorp, 2019).
Where possible, effect sizes (Cohen's d) were reported for studies that were not included in the meta-analysis, which were calculated using means and standard deviations reported in the studies.Effect size values were considered small (d = 0.2), moderate (d = 0.5) or large (d = 0.8) (Cohen, 2013).

Sensitivity analyses
Sensitivity analyses and heterogeneity between studies were assessed using the Higgins I 2 function which was considered high when the I 2 was higher than 75%.The Egger's test was also used together with funnel plots to identify any potential publication bias.The Duval and Tweedie trim and fill method (Duval and Tweedie, 2000) was used to identify present or absent studies causing funnel plot asymmetry, and to adjust for these studies.

Qualitative data synthesis
All studies that met the criteria for investigating eating cognitions, emotions and behaviours, but were not included in the meta-analyses and odds ration calculations, were reviewed and descriptively reported in this review, including available data from interventional and observational studies.

Characteristics of included studies and participants
A total of 92 studies with a total of 11,274 participants were eligible for inclusion in this systematic review.Of those, 11 studies were included in the quantitative analyses.Articles were published from the following countries: United States (n = 35; 37.6%), Italy and Australia (n = 6; 6.4% each); Canada and India (n = 5; 5.3% each); Germany, Turkey and Iran (n = 4; 4.3% each); Japan (n = 3; 3.2%); Netherlands, Israel, Spain, France, Hungary, UK, Thailand and Denmark (n = 2; 2.1% each); Austria, Taiwan, South Korea, Switzerland and Brazil (n = 1; 1% each).For more details about the summary of included studies see Table 1.-Disinhibited eating behaviours between male and female (d: 0.02).
-Caloric intake 143% higher than baseline in the 40 nausea-free patients.
-Caloric intake 124% higher in the 3rd day after exposure to 1.5 mg/ d olanzapine in the 13 nausea-free patients.
-No difference in caloric intake between a dose of 1.5 mg/d or more.b Mean dose was substituted with range, maximum, initial, median or fixed doses when not reported.c Duration of intervention in interventional studies was substituted with length of follow up in observational studies reported in mean ± standard deviations or range.
Most of the studies reported outcomes of adults (n = 56) either alone or in combination with sample of children, adolescents, adults or senior adults.Others included only children (n = 35), adolescents (n = 37) or senior adults (n = 18).The mean age was 26 ± 15 years and ranged from 2 to 85 years, reported by 83 studies.For more details about the characteristics of participants of included studies see Table 1.

Intervention
Olanzapine was the most widely used SGA in 33 studies followed by risperidone in 28 studies.The less frequently used SGAs were aripiprazole (n = 11), clozapine (n = 7), quetiapine (n = 7), ziprasidone (n = 1), asenapine (n = 1), ileoperidone (n = 1) and/or other mixes of SGAs.Three studies included patients treated with SGAs in combination with other type of drugs such as mood stabilizers, benzodiazepine, antidepressants and FGAs (de Beaurepaire, 2021; Lundgren et al., 2006;Morell et al., 2019;Teasdale et al., 2018).Only data related to SGAs were outlined in this review.All drug related data including drug dosages can be found in more detail in Table 1.

Table 2
Summary of studies included in the meta-analyses and odds ratio.Shaded rows indicate the inclusion of children and adolescents.All studies included in the odds ratio calculation are randomized controlled trials.Studies included in the eating behaviour meta-analyses are cross-sectional studies.NR: Not reported.a Age range in years was reported when mean ± standard deviation was not available.b Mean dose of the treatment agent was reported.c Represents the duration of intervention in weeks.d Drug dose was not mentioned.
Fig. 2. Forest plot for the hunger subscale of the three-factor eating questionnaire.

Outcomes and measures
Most of the studies reported eating-and appetite-related outcomes using subjective assessment tools.These include data reported from patients, carers or measured by health care practitioners, visual analogue scales, dosage records and treatment emergent symptom scales and adverse drug reactions checklists (n = 55).Diagnostic tools for studies reporting eating disorder symptomatology and diagnoses (n = 7) included the M-composite international diagnostic interview, criteria for the Diagnostic and Statistical Manual of mental disorders-Fourth edition (DSM-IV), the night eating questionnaire and other screening questions and self-report instruments based on DSM-IV criteria (de Beaurepaire, 2021; Gebhardt et al., 2007;Kluge et al., 2007;Kurpad et al., 2010;Lundgren et al., 2006;Moore et al., 2013;Theisen et al., 2003).
Eating behaviour was measured using the three-factor eating questionnaire (TFEQ) and the Dutch Eating Behaviour Questionnaire in 7 studies (Bachmann et al., 2012;Blouin et al., 2008;Bobo et al., 2011;Khazaal et al., 2009;Mathews et al., 2012;Sentissi et al., 2009;Stip et al., 2012).Caloric intake and dietary composition was measured by Fig. 3. Odds ratio for appetite increase.Fig. 4. Weighted mean percentage of participants reporting an increase in appetite.Self-reported appetite increase seen in participants treated with different SGA types.The chart indicates the weighted mean percentages with the minimum and maximum percentages obtained from available data.n = number of studies included in the weighted mean percentages calculations.either monitoring and calculating food intake manually by investigators (n = 7) (Ballon et al., 2018;Blouin et al., 2008;Fountaine et al., 2010;Gothelf et al., 2002;Okamoto et al., 2019;Roerig et al., 2005;Teff et al., 2013) or by using standardized tools such as the Food Frequency Questionnaire, 24-h dietary recall or built-in software such as the Dankost Pro (n = 7) (Calarge et al., 2012;Calarge and Ziegler, 2013;Daurignac et al., 2015;Henderson et al., 2006;Jakobsen et al., 2018b;Kluge et al., 2007;Lindsay et al., 2006;Teff et al., 2015).For more details about outcomes and outcome measures used in eligible studies, refer to Table 1.

Characteristics of studies included in the meta-analyses
A total of three studies were eligible for inclusion in the metaanalysis to compare the three eating behaviour subscales of the TFEQ cross-sectionally after treatment (Blouin et al., 2008;Khazaal et al., 2009;Sentissi et al., 2009).These studies included adult participants (18-65 years old) diagnosed with first episode psychoses (non-schizophrenia), schizophrenia and schizoaffective disorders.Independent analyses were conducted for each of the eating behaviour subscales: disinhibition, restraint and hunger.
Four additional studies also used the TFEQ but were ineligible for inclusion in the meta-analyses and thus were described narratively (Bachmann et al., 2012;Bobo et al., 2011;Mathews et al., 2012;Stip et al., 2012).See supplementary material S.6 for a qualitative appraisal of these studies.
Results from the eight studies that were combined using odds ratios were all longitudinal, reporting binary outcomes pertaining to appetite increase (Aman et al., 2005;Black et al., 2014;Guardia et al., 2004;Litten et al., 2012;McCracken et al., 2002;Roerig et al., 2005;Snyder et al., 2002;Srivastava et al., 2012).Diagnoses of participants included in this analysis were autism disorder (n = 2), alcohol dependence disorder (n = 2), conduct and disruptive behaviour disorder (n = 1), bipolar disorder (n = 1), borderline personality disorder (n = 1) and one study included healthy participants.Three studies included children aged from 5 to 12 years old (Aman et al., 2005;McCracken et al., 2002;Snyder et al., 2002), and the remaining included adults with an age range from 18 to 60 years old (Black et al., 2014;Guardia et al., 2004;Litten et al., 2012;Roerig et al., 2005;Srivastava et al., 2012).For more details about the characteristics of studies and participants included in the meta-analyses and odds ratio see Table 2.

Quality assessment
Overall, the quality ratings for all the 92 studies were of acceptable quality and thus all were included in our review.Many studies have used validated outcome measures while others used newly developed or nonvalidated tools to measure outcomes of interest.Other limitations found in the eligible studies included missing raw data and effect size presentations, uncertainties in the diagnosis of the study sample, undetailed inclusion and exclusion criteria, small sample sizes, mixed types of SGA exposure and limited follow up periods due to the nature of some studies such as cross-sectional studies.For more details about the quality assessment of included studies, refer to S.5.H. appetite,food craving and addiction 3.4.1.1.Cross-sectional meta-analysis of TFEQ hunger subscale.Data from a total of three studies, using a total sample of 201 participants (133 treated with SGAs and 68 controls) were analysed to compare the hunger subscales of the TFEQ.Overall, results suggested that there was a moderately greater score in the hunger subscale of the TFEQ in participants given SGAs (SMD = 0.62, 95% CI ]-0.26, [1.51, p = 0.17) (see Fig. 2 for a forest plot).
Additionally, Fig. 4 depicts the weighted mean percentages of participants reporting appetite increase with SGA use, across all studies included in the systematic review and meta-analysis, per type of SGA.51 studies were included, of which the numbers of studies examining each SGA specific drug were: 2 for clozapine; 21 for risperidone; 16 for olanzapine; 8 for aripiprazole and 4 for quetiapine.All included studies reported appetite increase experienced with these SGAs.

Food craving.
The effect sizes between the SGA and control groups in food cravings, which were calculated from the data of (Abbas and Liddle, 2013), were slightly higher for the fat (d = 0.24) and carbohydrate (d = 0.18) craving.Minimal differences were seen for the general foods (d = 0.11), sweets (d = − 0.05), and fast-food fat (d = − 0.16) cravings (Abbas and Liddle, 2013). 3.4.1.4. Food addiction.Only one study reported food addiction as an outcome in patients treated with SGAs (Goluza et al., 2017), finding that approximately 27% of participants treated with quetiapine, olanzapine and clozapine met the diagnostic criteria for food addiction and 77% had at least 3 food addiction symptoms.
For more details about the qualitative appraisal of studies investigating hunger, appetite, food craving and addiction, see supplementary material S.6.

Dietary disinhibition, binge eating and loss of control over eating
3.4.2.1.Cross-sectional meta-analysis of TFEQ dietary disinhibition subscale.Meta-analysis was performed using data from a total of three studies and a total sample of 201 participants (133 treated with SGAs; 68 controls) to compare the disinhibited eating subscales of the TFEQ.Results of this analysis revealed a small increase in disinhibited eating in patients treated with SGAs (SMD = 0.40, 95% CI ]-0.17, 0.97 [, p = 0.169) compared to controls and an overall substantial heterogeneity (I 2 = 66.43%) (See Fig. 5 for a forest plot).For more details about the qualitative appraisal of studies investigating disinhibited eating, binge and loss of control over eating, see supplementary material S.6.

Dietary restraint
3.4.3.1.Meta-analysis of TFEQ dietary restraint subscale.Data from a total of three studies, using a total sample of 201 participants of which 133 treated with SGAs and 68 controls were analysed to compare the restrained eating subscales of the TFEQ.Overall, this meta-analysis showed a small increase in restrained eating in patients treated with SGAs (SMD = 0.43, 95% CI ]-0.07, [0.94, p = 0.091) compared to controls with substantial heterogeneity between studies (I 2 = 57.68%)(see Fig. 6 for a forest plot).
One study found an increase in fullness (Khazaal et al., 2009) while two others reported no effect of olanzapine, aripiprazole and risperidone on satiety and fullness in an experimental food laboratory environment (Roerig et al., 2005;Teff et al., 2015).Treuer et al. (2009), however, found that approximately 35% of participants treated with olanzapine needed a larger amount of food in order to feel full.
Caloric intake of participants treated with SGAs was increased in many eligible studies in our review (Cicala et al., 2020;Fountaine et al., 2010;Gothelf et al., 2002;Kluge et al., 2007;Okamoto et al., 2019;Roerig et al., 2005).However, others reported reductions in calories consumed with SGA treatment (Henderson et al., 2006;Jakobsen et al., 2018b).Caloric intake was higher than requirements in Lindsay et al. (2006) and Teasdale et al. (2018)'s sample.Interestingly, Teasdale et al. (2018) found that treatment with olanzapine relative to other SGA monotherapy resulted in an increase in caloric intake.See Table 1 for detailed summary about the results.
An increase in food intake after the introduction of olanzapine was noted in only one laboratory-based experimental meal study (Ballon et al., 2018).Calarge and Ziegler (2013) found a relationship between iron status and caloric intake in that the iron depleted group ate more with risperidone.
Assessing macronutrients consumption revealed that the intake of carbohydrates in three studies (Ballon et al., 2018;Calarge and Ziegler, 2013;Lindsay et al., 2006), protein in two (Ballon et al., 2018;Lindsay et al., 2006) and fat in three studies (Ballon et al., 2018;Henderson et al., 2006;Lindsay et al., 2006) was increased with SGAs.Conversely, the consumption of macronutrients did not change from baseline (Gothelf et al., 2002), was in line with the dietary composition of the general populations (Jakobsen et al., 2018b) or lower than controls (Henderson et al., 2006).
Poor diet quality was detected in schizophrenic patients when compared to controls, and this did not change after treatment with SGAs in two studies (Jakobsen et al., 2018a(Jakobsen et al., , 2018b)).For detailed qualitative appraisal of outcomes relating to dietary composition, quality and eating habits see S.6.

Discussion
The present study systematically assessed and meta-analysed outcomes relating to eating cognitions, behaviours and emotions in individuals treated with SGAs regardless of their underlying psychiatric Both quantitative and qualitative synthesis of a broad range of study designs revealed an increase in appetite and hunger in participants treated with SGAs.Individuals exposed to SGAs were approximately 1.51 times more likely to experience an increase in appetite relative to controls.Moreover, a small increase in restrained and disinhibited eating were found in studies using standardised instruments although it is noteworthy that the largest study included found no effect (Sentissi et al., 2009).A small sized increase in restrained dietary behaviours in the SGA-treated participants was manifest, even though deliberate attempts to restrict dietary intake for the purpose of controlling weight are not usually a characteristic of this patient population (Bellisle, 2009).
Outcomes relating to the drive to eat including food craving and addiction were also considered in this review as a component of appetite control.The effect size for craving was slightly higher for fat (d = 0.38) and carbohydrates (d = 0.44) cravings in the SGA treated compared with the control groups (Abbas and Liddle, 2013).
Notably, one study in our review found that 27% of participants treated with SGAs met diagnostic criteria for food addiction and 77% demonstrated at least 3 markers of addiction to processed and energydense food (Goluza et al., 2017).However, more research is necessary into the relationship between SGA use and food addiction in order to substantiate these findings.
Studies on eating behaviours, including the quantity and quality of diets consumed by patients revealed interesting yet limited findings in our review.The caloric intake assessment showed that olanzapine was the most common SGA to cause increased food consumption, followed by clozapine, risperidone and aripiprazole (Cicala et al., 2020;Kluge et al., 2007).The dietary patterns including dietary composition (food source) and habits revealed some evidence for an increased preference for sweets, sugary food/drinks, and energy dense savoury foods (Henderson et al., 2006;Jakobsen et al., 2018b;Kluge et al., 2007;Lappin et al., 2018;Morell et al., 2019;Platzer et al., 2021).

Strength and limitations
To the best of our knowledge, our article is the first systematic review and meta-analysis investigating eating cognitions, emotions and behaviours in individuals treated with SGAs.Selection bias was mitigated by involving two independent reviewers during the screening, data extraction and quality assessment procedures.
Given the high heterogeneity in study designs, the limited number of interventional studies investigating the parameters of interest, we were only able to infer cautious conclusions.The high level of heterogeneity in the reporting of outcomes limited the number of studies eligible for inclusion in a meta-analysis and limited further investigations looking at the effect of drug dosage and treatment duration on meta-analytic findings.Crucially, there were insufficient studies included in the meta-analyses to allow us to conduct a meta-regression extension to our results as it has been indicated that ten or more studies are required for a meta-regression to be appropriate (Schmid et al., 1998).
One of the limitations of our review is including only English articles mostly from the United States which might have caused missing some important findings from non-English publications.Limitations also included the heterogeneity of outcome measures used from subjective, objective and validated questionnaires for each parameter of interest.

Clinical implications
The findings from this study suggest that some SGAs may impact on appetite regulation leading to overeating with resultant weight gain and problems associated with obesity, such as cardiovascular diseases and metabolic disturbances (Davis et al., 2020;Khosravi, 2020;Sankaranarayanan et al., 2021).In addition to the social adversity associated with severe mental illness (Compton et al., 2020), which may contribute to poor dietary patterns, it is possible that there is an interaction between the increase in appetite and our current food environment, which includes easily available, highly palatable and ultra-processed foods that may contribute to the development of food addiction (Küçükerdönmez et al., 2019).These disturbed dietary patterns have been shown recently to be greatly correlated with the risk of developing life-threatening illnesses (Wang et al., 2022).
The use of SGAs may also impact on impulsivity which may contribute to the development of maladaptive eating behaviours such as emotional, restrained, or disinhibited eating, as an addition to the weight gain and metabolic disturbance side effects of the drugs (Moe et al., 2016).Disinhibited eating of fat-and carbohydrate-rich foods may produce rapid weight gain in a short duration of time (de Beaurepaire, 2021; Kobayashi and Takano, 2018).This was commonly noted in patients exposed to the SGAs olanzapine and clozapine (Gebhardt et al., 2007).While not all patients on SGAs develop these side effects, those who do may benefit from additional medical support that may include lifestyle modifications and pharmacological interventions.Further large-scale interventional studies investigating the disinhibited eating associated with the use of SGAs, especially the ones linked with high risk of alteration in metabolic function such as olanzapine and clozapine, are needed in order to develop practical advice in modifying the altered dietary patterns for this patient population.
Available options to prevent and overcome SGA-related metabolic side effects may include choosing an antipsychotic with lower risk of metabolic alterations, involving patients in psychological, lifestyle and behavioural interventions, switching to different antipsychotic or adding adjunctive therapies to antipsychotics such as amantadine or topiramate (Hasan et al., 2019;Holt et al., 2019;NICE, 2014).However, some of the above options might not always be feasible as certain drugs are needed to control the intractable symptoms caused by certain psychiatric disorders.For example, some patients on a clozapine regimen cannot be switched back to a lower affinity SGA especially when clozapine was prescribed as the last remaining treatment option after multiple attempts with other SGAs (Correll and Howes, 2021).
There may be a promising role of some pharmacological treatments such as the glucagon-like peptide receptor agonists, liraglutide and semaglutide for the purpose of controlling weight gain (Shi et al., 2022).The latter is believed to be a practice-changing invention for weight control due to its low risk-benefit ratio seen in recent clinical trials (Wilding et al., 2021;Rubino et al., 2021;Davies et al., 2021;Wadden et al., 2021).Evidence also exists regarding the weight control benefits of some antidiabetic drugs such as metformin (Maayan et al., 2010;Seifarth et al., 2013;Chukir et al., 2021).Reboxetine, bupropion and exenatide were specifically tested as an adjunctive treatment for some SGAs and has shown clinical benefits in lessening the severity of weight gain caused by SGAs (Poyurovsky et al., 2003(Poyurovsky et al., , 2007;;Weizman et al., 2021;Siskind et al., 2018).Although the addition of some adjunctive drugs to SGAs were tested, and some have shown promising effects to alleviate antipsychotic-induced metabolic disturbances, further investigations are yet warranted to identify the most suitable options (Larsen et al., 2017;Lee et al., 2021).Assessing and managing changes in eating behaviours, cognitions and emotions may be key to managing increased appetite and food craving in patients treated with SGAs.The management of increased appetite and weight gain may include psychotherapeutic, psychopharmacological and dietary approaches as well as physical exercise (Taylor et al., 2012;Blundell et al., 2015;Mann et al., 2007;Jackson et al., 2015).
It is worthwhile mentioning the possible therapeutic role of SGAs for patients with eating disorders, specifically, anorexia nervosa, where weight gain remains a challenge (Williams et al., 2021).A systematic review and meta-analysis has shown a positive effect on weight gain (Han et al., 2022).However, we found no studies that examined changes in eating behaviours in this patient population for this review.Therefore, further investigations are warranted.

Fig. 5 .
Fig. 5. Forest plot for the dietary disinhibition subscale of the three-factor eating questionnaire.

Fig. 6 .
Fig. 6.Forest plot for the restraint subscale of the three-factor eating questionnaire.

Table 1
Summary of included studies Interventional studies.
(continued on next page) H.Mutwalli et al.

Table 1
(continued ) (continued on next page) H. Mutwalli et al. (continued on next page) H. Mutwalli et al. (continued on next page) H. Mutwalli et al.

Table 1
(continued ) (continued on next page)

Table 1
(continued on next page) H.Mutwalli et al.

Table 1
H.Mutwalli et al.diagnosis.A total of 92 studies were included in this systematic review encompassing participants from all age groups.