Heterogeneity in exercise interventions and metabolic outcomes

Patients with severe mental illnesses, especially those with psychotic disorders such as schizophrenia, face an elevated risk of developing metabolic syndrome due to intertwined biological, inflammatory, and iatrogenic factors [Reference Polcwiartek, O’Gallagher, Friedman, Correll, Solmi and Jensen7, Reference Correll, Detraux, De Lepeleire and De Hert17, Reference Schmitt, Maurus, Rossner, Röh, Lembeck and von Wilmsdorff18, Reference DE Hert, Schreurs, Vancampfort and VAN Winkel53]. Antipsychotics promote central obesity and insulin resistance through weight gain and adipocyte dysfunction [Reference Correll, Detraux, De Lepeleire and De Hert17, Reference Schmitt, Maurus, Rossner, Röh, Lembeck and von Wilmsdorff18], while psychotic disorders related inflammation independently worsens cardiometabolic risk [Reference Henderson, Vincenzi, Andrea, Ulloa and Copeland9, Reference Lambert, Copeland, Sampson and Duffy10, Reference Pillinger, McCutcheon, Vano, Mizuno, Arumuham and Hindley20].

Despite the well-established link between metabolic syndrome and cardiovascular disease [Reference Mottillo, Filion, Genest, Joseph, Pilote and Poirier12], the included trials assessed its components inconsistently (e.g., prioritizing isolated markers such as waist circumference or triglycerides over syndromic criteria). This methodological inconsistency underscores the need for standardized protocols in future studies [Reference Zhou, Guan, Xiu and Wu42–Reference Masa-Font, Fernández-San-Martín, Martín López, Alba Muñoz, Oller Canet and Martín Royo51].

Methodological heterogeneity in defining and measuring metabolic syndrome components significantly hampered cross-study comparability. Key examples include waist circumference measurement, protocols between the WHO-standardized midpoint (iliac crest to lower rib margin), umbilical positioning, and inconsistent subject posture or respiratory phase documentation. Despite universal reporting of triglyceride and HDL, lipid assessments were similarly compromised by variable fasting durations (8–12 h), sample processing methods, and assay techniques. This reflects nonadherence to the established diagnostic criteria (e.g., IDF and NCEP-ATP III). Future trials should implement standardized protocols (WHO STEPPS for anthropometry and CDC-NHLBI guidelines for lipids) with explicit methodological reporting. The development of harmonized COS for SMI metabolic research would enable meaningful evidence synthesis.

Notably, multimodal interventions combining aerobic and resistance training demonstrated modest benefits [Reference Scheewe, Backx, Takken, Jörg, van Strater and Kroes46, Reference Jae, Kil, So, Lee, Song and Kim47], supporting that varied exercise regimens enhance metabolic flexibility [Reference Heissel, Heinen, Brokmeier, Skarabis, Kangas and Vancampfort27, Reference Yang, Yuan, Zhang, Fu, Wang and Wang29, Reference Vancampfort, Stubbs, Mitchell, De Hert, Wampers and Ward54]. On the other hand, trials using only aerobic or low-intensity workouts reported no benefits, suggesting that both exercise type and intensity play a critical role [Reference Fernández-Abascal, Suárez-Pinilla, Cobo-Corrales, Crespo-Facorro and Suárez-Pinilla48, Reference Wu, Wang, Bai, Huang and Lee50, Reference Masa-Font, Fernández-San-Martín, Martín López, Alba Muñoz, Oller Canet and Martín Royo51]. While high-intensity interval training (HIIT) has potential advantages, it is difficult to implement in this population owing to practical challenges, including motivation and tolerance issues. A key limitation of this study is the paucity of trials comparing exercise modalities. We could not determine the optimal exercise type for metabolic improvement with only two HIIT studies, one yoga trial, and two combined training regimens. This heterogeneity prevents robust recommendations regarding specific protocols (e.g., HIIT vs. resistance training) and underscores the need for head-to-head comparative trials.

We aggregated data across schizophrenia, schizoaffective disorder, and schizophreniform disorder owing to their shared cardiometabolic risk profiles and universal exposure to second-generation antipsychotics (SGAs), which frequently induce weight gain, dyslipidemia, and glucose intolerance. While the illness course and symptomatology show some heterogeneity, this grouping aligns with established psychiatric research paradigms and reflects clinical practice realities. Future studies with larger sample sizes may facilitate diagnostic subgroup analysis.

Blood pressure and exercise intensity

Exercise interventions were not associated with significant reductions in systolic or diastolic blood pressure compared to nonexercise controls, with moderate to high heterogeneity across studies. These findings are consistent with prior evidence suggesting that exercise alone, without adjunctive dietary or pharmacological strategies, may be insufficient to treat hypertension in individuals with psychotic disorders [Reference Polcwiartek, O’Gallagher, Friedman, Correll, Solmi and Jensen7]. Notably, antipsychotic medications have been associated with hypertension, especially second-generation medications such as clozapine and olanzapine, through mechanisms such as weight gain, sodium retention, and sympathetic activation [Reference Correll, Detraux, De Lepeleire and De Hert17, Reference Schmitt, Maurus, Rossner, Röh, Lembeck and von Wilmsdorff18, Reference Newcomer55]. While Bredin et al. [Reference Bredin, Warburton and Lang56] reported blood pressure improvements after 12 weeks of aerobic or resistance training, their cohort lacked severe psychiatric comorbidities, potentially inflating generalizability. Exercise intensity is a major moderator of training outcomes; studies involving high-intensity interval training (HIIT; 70–95% HR peak) or vigorous aerobic protocols (50–90% HR_max) [Reference Heggelund, Nilsberg, Hoff, Morken and Helgerud43, Reference Romain, Fankam, Karelis, Letendre, Mikolajczak and Stip44] demonstrated greater metabolic adaptations than moderate-intensity regimens [Reference Fernández-Abascal, Suárez-Pinilla, Cobo-Corrales, Crespo-Facorro and Suárez-Pinilla48, Reference Brobakken, Nygård, Güzey, Morken, Reitan and Heggelund49], in line with the dose–response relationships observed in nonpsychiatric populations [Reference Hwang, Yoo, Kim, Hwang, Handberg and Petersen57–Reference Ramos, Dalleck, Tjonna, Beetham and Coombes59]. However, HIIT may not be feasible in psychosis, where motivational deficits, sedation with antipsychotics, and cognitive impairments pose significant barriers. Large trials stratifying patients according to illness severity and medication regimens are needed to clarify optimal intensity thresholds.

Lipid profiles and medication interactions

Despite the role of aerobic exercise in improving lipid metabolism [Reference Vancampfort, Rosenbaum, Schuch, Ward, Richards and Mugisha33], our meta-analysis did not find any significant effects on triglyceride or HDL cholesterol levels. This discrepancy likely reflects pronounced metabolic dysregulation in the SMI. Antipsychotics directly exacerbate lipid abnormalities via hepatic lipogenesis (endogenous hepatic synthesis of fatty acids from carbohydrates) and adipocyte dysfunction (impaired function of fat-storing cells) [Reference McIntyre, Soczynska, Konarski and Kennedy14, Reference Buhagiar and Jabbar60]. For instance, clozapine and olanzapine elevate triglyceride levels by up to 40% and reduce HDL levels by 15–20% in psychotic disorders [Reference Correll, Detraux, De Lepeleire and De Hert17, Reference Schmitt, Maurus, Rossner, Röh, Lembeck and von Wilmsdorff18, Reference Newcomer55], which may overwhelm the modest lipid-modulating capacity of exercise. A pilot study demonstrating HDL reduction at week 12 of training [Reference Bredin, Warburton and Lang56] further highlights the complexity of the lipid response in medicated populations. Although exercise may mitigate antipsychotic-induced hypertriglyceridemia without compromising psychiatric stability [Reference Kim, Lang, Warburton, Barr, Smith and Thornton61], the mechanisms involved remain unclear. The proposed pathways include enhanced lipoprotein lipase activity and β oxidation [Reference Mann, Beedie and Jimenez62], though inflammation and oxidative stress, which are hallmarks of psychosis, may blunt these adaptations. Future studies must incorporate direct biomarkers (e.g., apolipoprotein B and LDL particle size) and control for medication dose, duration, and polypharmacy to disentangle these interactions.

Standardization extends beyond measurement techniques to encompass the diagnostic criteria. Only 3 of 10 included studies applied established metabolic syndrome definitions (e.g., IDF or NCEP-ATP III), while others reported isolated components [Reference Romain, Fankam, Karelis, Letendre, Mikolajczak and Stip44, Reference Scheewe, Backx, Takken, Jörg, van Strater and Kroes46, Reference Fernández-Abascal, Suárez-Pinilla, Cobo-Corrales, Crespo-Facorro and Suárez-Pinilla48]. These risks misclassify outcomes and dilute actual intervention effects. Future work should mandate adherence to internationally recognized criteria and report individual components along with composite syndrome status.

Glucose metabolism and adherence challenges

Data from all studies in this meta-analysis assessed glucose levels after exercise interventions, but limited improvements were observed despite exercise is established to enhance insulin sensitivity [Reference Correll, Solmi, Veronese, Bortolato, Rosson and Santonastaso3, Reference Lambert, Copeland, Sampson and Duffy10, Reference Henderson, Cagliero, Copeland, Borba, Evins and Hayden63]. Previous studies using an intervention based on aerobic training alone [Reference Fernández-Abascal, Suárez-Pinilla, Cobo-Corrales, Crespo-Facorro and Suárez-Pinilla48, Reference Wu, Wang, Bai, Huang and Lee50] did not improve hyperglycemia, perhaps because of suboptimal intensity, insufficient duration, or poor adherence. Of note, the only adherence report – a critical determinant of exercise efficacy – reflected data at point, FFD 30 months and FFD 40 months in five trials with an adherence range of 60–75% [Reference Romain, Fankam, Karelis, Letendre, Mikolajczak and Stip44, Reference Fernández-Abascal, Suárez-Pinilla, Cobo-Corrales, Crespo-Facorro and Suárez-Pinilla48, Reference Brobakken, Nygård, Güzey, Morken, Reitan and Heggelund49]. However, one trial reported a 50% dropout rate in the intervention arm, which was primarily attributed to motivational deficits and program duration [Reference Romain, Fankam, Karelis, Letendre, Mikolajczak and Stip44].

Attrition of this type is consistent with larger evidence associating poor adherence in SMI populations with psychosocial obstacles (e.g., cognitive impairment and social isolation) and adverse consequences of medications (e.g., sedation and fatigue) [Reference Korman, Stanton, Vecchio, Chapman, Parker and Martland22]. The lack of standardized adherence metrics across studies – half omitted attendance thresholds or compliance criteria – further complicates the interpretation. These findings underscore the necessity of integrating behavioral strategies (e.g., motivational interviewing and peer support) to sustain engagement in this population.

The lack of glucose-lowering effects may also indicate profound metabolic dysregulation inherent to psychotic disorders. Antipsychotics, including olanzapine and clozapine, modulate insulin signaling via direct β-cell toxicity and adipose tissue inflammation [Reference Mendelson64, Reference Nielsen, Skadhede and Correll65], which could cancel out exercise-induced benefits. Furthermore, exercise prescriptions in the included trials rarely accounted for medication specific metabolic risks or individualized glycemic targets. Future interventions should prioritize stratified designs that control for antipsychotic class, dose, and illness chronicity while incorporating continuous glucose monitoring to capture dynamic responses.

Antipsychotic medications: A critical confounder

All participants were prescribed antipsychotics, which are known to promote weight gain, dyslipidemia, and insulin resistance [Reference Mendelson64, Reference Fleischhacker, Siu, Bodén, Pappadopulos, Karayal and Kahn66, Reference Allison, Mentore, Heo, Chandler, Cappelleri and Infante67]. One trial exclusively included patients treated with clozapine [Reference Pillinger, McCutcheon, Vano, Mizuno, Arumuham and Hindley20, Reference Wu, Wang, Bai, Huang and Lee50], a subgroup with a greater metabolic risk [Reference Mendelson64, Reference Nielsen, Skadhede and Correll65]. This timing is further challenged by the progressive metabolic burden in chronic (vs. first-episode) psychosis [Reference Nielsen, Skadhede and Correll65, Reference Fleischhacker, Siu, Bodén, Pappadopulos, Karayal and Kahn66, Reference Kessler, Sisson and Short68, Reference Izquierdo, De Souto Barreto, Arai, Bischoff-Ferrari, Cadore and Cesari69].

Although structured exercise interventions offer multiple health benefits, this meta-analysis found no evidence that exercise improves metabolic syndrome in individuals with SMI receiving antipsychotic treatment. Therefore, although physical activity should remain a safe and broadly beneficial component, it appears insufficient to overcome the metabolic burden associated with antipsychotics. Pharmacodynamic interactions, such as how exercise alters drug metabolism or receptor sensitivity, are understudied, but should be urgently investigated [Reference Izquierdo, De Souto Barreto, Arai, Bischoff-Ferrari, Cadore and Cesari69].

In all trials selected for inclusion, participants were administered antipsychotics [Reference Zhou, Guan, Xiu and Wu42–Reference Masa-Font, Fernández-San-Martín, Martín López, Alba Muñoz, Oller Canet and Martín Royo51], known independent agents of metabolic dysregulation via mechanisms of weight gain, hypertriglyceridemia, and insulin resistance [Reference Correll, Detraux, De Lepeleire and De Hert17, Reference Schmitt, Maurus, Rossner, Röh, Lembeck and von Wilmsdorff18, Reference Mendelson64]. Reversible with dose reduction, second-generation antipsychotics (SGAs) such as clozapine, olanzapine, risperidone, and quetiapine worsen cardiovascular risk through direct actions on lipid metabolism (elevating triglycerides, reducing HDL) and glucose homeostasis [Reference Henderson, Cagliero, Copeland, Borba, Evins and Hayden63, Reference Samaras, Correll, Mitchell and De Hert70, Reference Yood, DeLorenze, Quesenberry, Oliveria, Tsai and Willey71]. Clozapine, in particular, demonstrates the most severe metabolic liabilities, with studies reporting a fourfold increased risk of type 2 diabetes and 20–25% weight gain within the first year of treatment [Reference Mendelson64, Reference Nielsen, Skadhede and Correll65].

Critically, none of the trials examined exercise–psychotropic medication interactions, despite the possibility that physical activity may modulate drug pharmacokinetics (e.g., hepatic CYP450 enzyme activity) or counteract receptor-mediated metabolic harm (e.g., 5-HT2C antagonism-induced hyperphagia) [Reference Haddad and Sharma72, Reference Chaouloff73]. This limitation greatly hinders our ability to separate the independence of exercise as a therapeutic modality from its potential ameliorative effect as a treatment adjunct (e.g., metabolic salvage therapy) in drug-treated populations.

Chronicity and the case for early intervention

The higher proportion of cases classified as chronic psychosis in our analysis [Reference Zhou, Guan, Xiu and Wu42–Reference Masa-Font, Fernández-San-Martín, Martín López, Alba Muñoz, Oller Canet and Martín Royo51] is consistent with the progressive nature of metabolic syndrome in SMI, whereby long exposure to antipsychotic medications and illness-related determinants (e.g., chronic inflammation, sedentariness) contribute to increased risk over time [Reference Nielsen, Skadhede and Correll65, Reference Fleischhacker, Siu, Bodén, Pappadopulos, Karayal and Kahn66, Reference Kessler, Sisson and Short68]. In contrast, patients with first-episode disease have a lower metabolic burden of morbidity, leaving a crucial time window for early intervention [Reference Kessler, Sisson and Short68, Reference Fitzgerald, Sahm, Byrne, O’Connell, Ensor and Ní Dhubhlaing74]. Structured exercise may be protective when significant weight gain and insulin resistance begin, and exercise prevention studies initiated upon illness onset may help counter eating-induced weight gain and insulin resistance before irreversible pathophysiological derangements occur [Reference Fitzgerald, Sahm, Byrne, O’Connell, Ensor and Ní Dhubhlaing74]. However, pragmatic barriers, including motivational deficits, medication induced fatigue, and limited access to tailored programs, hinder adherence in this population [Reference Firth, Solmi, Wootton, Vancampfort, Schuch and Hoare75]. Adding exercise-coordinated specialty care models, nutritional counseling, and peer support may enhance engagement and sustainability. Our findings must be interpreted in the light of several limitations. The small, heterogeneous sample size of trials and inconsistent outcome report constrain generalizability. Although statistical tests did not detect publication bias, this remains plausible, given the niche focus of metabolic interventions in psychosis [Reference Zhou, Guan, Xiu and Wu42–Reference Masa-Font, Fernández-San-Martín, Martín López, Alba Muñoz, Oller Canet and Martín Royo51]. Critically, intervention heterogeneity, marked by disparate exercise modalities, intensities, and durations, precluded dose–response insights, while the pervasive metabolic effects of second-generation antipsychotics (SGAs) likely obscure exercise-specific benefits.

Furthermore, adherence variability, compounded by poor reporting and high attrition rates, introduces potential bias. To address these gaps, future studies should standardize metabolic syndrome criteria (e.g., incorporating direct biomarkers such as insulin sensitivity), stratified analyses by antipsychotic class, dose, and illness chronicity, and integrate evidence-based behavioral strategies to enhance adherence. Such refinements are essential for isolating the therapeutic potential of exercise within the complex biopsychosocial landscape of severe mental illnesses.

Future studies should explicitly evaluate the interactions between specific antipsychotic agents and exercise-induced metabolic responses. This includes stratifying participants by antipsychotic type (e.g., clozapine vs. aripiprazole) and dose and exploring potential mechanisms such as altered receptor binding, hepatic metabolism (e.g., CYP450 modulation), and downstream effects on lipid and glucose regulation. A deeper understanding of these interactions will enable more tailored and clinically relevant exercise prescriptions for this population.