Placebo response on the weight change and its influencing factors in overweight or obesity with or without diabetes: a meta-analysis of 182 anti-obesity medications trials

Abstract

Background

To further determine the placebo response and the associated factors in trials of anti-obesity medications (AOMs).

Methods

PubMed, EMBASE, the Cochrane Center Register of Controlled Trials, and Clinicaltrial.gov website were searched from the inception to December 2025 for placebo-controlled trials of AOMs conducted in patients with overweight or obesity with or without diabetes.

Results

Overall, 182 trials were included. Placebo treatment in AOMs trials generally resulted in a weight loss of 1.13 kg (95% CI, −1.29 to −0.97 kg, P < 0.01). Participants receiving injectable placebo experienced a greater weight reduction than those receiving oral placebo (ES = −1.46 kg, 95% CI, −1.74 to −1.19 kg, P < 0.01 vs. ES = −0.82 kg, 95% CI, −1.05 to −0.59 kg, P < 0.01; P for subgroup comparison <0.001). Patients without diabetes showed a greater weight loss of placebo response than those with type 2 diabetes (ES = −1.92 kg, 95% CI, −2.22 to −1.62 kg, P < 0.01 vs. ES = −0.72 kg, 95% CI, −0.92 to −0.52 kg, P < 0.01; P for subgroup comparison <0.001), whereas individuals with type 1 diabetes had a significant weight gain (ES = 0.76 kg, 95% CI, 0.46 to 1.05 kg, P < 0.01). Meta-regression analyses showed younger age, lower male percentage, higher baseline body mass index and higher baseline body weight were significantly associated with greater weight reduction in placebo response.

Conclusion

Our meta-analysis indicated a significant placebo response in trials of AOMs, which was associated with age, sex, baseline BMI and baseline body weight in patients with overweight or obesity with or without diabetes.

Plain language summary

This study looked at how often a “placebo” (a pill or injection with no medicine) causes weight loss in clinical trials for obesity treatments. Researchers analyzed 182 past studies. They found that people taking a placebo still lost some weight—about 1.13 kg on average. Interestingly, “fake” injections led to more weight loss than “fake” pills. People without diabetes lost more weight than those with type 2 diabetes, while people with type 1 diabetes tended to gain weight. Younger people, those with a higher starting weight, and more women also lost more weight on placebo. These results show that simply taking placebo in a treatment trial can influence weight. Therefore, future obesity studies should carefully account for the placebo response to ensure new drugs are truly effective.

Guo and Yang et al. examine placebo response and associated factors in placebo-controlled trials of anti-obesity medications. Their meta-analysis indicated a significant placebo response, which was associated with age, sex, baseline BMI, and baseline body weight in patients with overweight or obesity with or without diabetes.

Introduction

Recognized as one of the greatest public health problems globally, obesity has reached pandemic status in the last 50 years¹, with the prevalence nearly tripling since 1975². New data presented in Obesity Atlas 2022 showed that, by 2030, it is predicted that 1 in 5 women and 1 in 7 men will be living with obesity, equating to over 1 billion people globally³. Furthermore, obesity is also the largest contributor of the metabolic disease burden and almost 5.02 million people worldwide died of obesity in 2019 according to the Global Burden of Disease Study⁴.

First-line treatment for obesity is lifestyle intervention, followed by pharmacotherapies⁵. Anti-obesity medications were recommended as one of treatments for obesity in combination with lifestyle interventions. Numerous randomized controlled trials (RCTs) of anti-obesity medications have been conducted to test the efficacy of pharmacological agents, with placebo used as a control. However, weight changes due to placebo response have been found in these trials. The placebo response is described as a series of positive changes that occur after taking placebo, including the natural history of the disease (e.g., spontaneous remissions), regression to the mean, biases from patients and clinicians, ambiguity in symptom detection, potential hidden effects of unidentified cointerventions, and the expectations of patients and clinicians regarding therapeutic outcomes⁶. The assessment of medication efficacy in clinical trials may be hampered by the large placebo response, which makes it more difficult to observe the difference in efficacies between the tested medication and the placebo. Therefore, it is important to have a comprehensive understanding of the placebo response and the associated factors to better interpret the published trials objectively and guide the design of future trials scientifically.

Involving 70 trials and 20,555 patients, a meta-analysis estimated the placebo response rate in obesity pharmacological trials, reporting that the proportion of patients who experienced ≥5%, ≥10%, and ≥15% weight loss was 20.4%, 8.3%, and 6.2%, respectively, and the proportion increased gradually with increasing duration of study⁷. However, considering the numerous trials of hypoglycemic medications with weight-loss effects, the number of trials included in this meta-analysis was limited. Given the rising amplitude of the placebo response and the growing number of placebo-controlled trials of anti-obesity medications in recent years, it is essential to further investigate the placebo response. In addition, the potential factors associated with the placebo response of anti-obesity medications have not been adequately explored so far. Therefore, for further clinical trial design and decision-making, we performed an updated systematic review and meta-analysis for RCTs of anti-obesity medications to appraise the placebo response and the associated factors in patients with overweight or obesity with or without diabetes.

Methods

Study design

This systematic review and meta-analysis were conducted according to the guidelines of Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) protocol⁸. The protocol of this meta-analysis was registered in International Prospective Register of Systematic Reviews (PROSPERO) as CRD42023432174.

Data sources and searches

According to the recommendations from the Cochrane Handbook for Systematic Reviews for meta-analysis⁹, pairs of investigators (YG, WY, ZL, and SB) conducted systematic searches of PubMed, EMBASE, the Cochrane Center Register of Controlled Trials, and clinicaltrials.gov for RCTs of anti-obesity medications conducted in patients with overweight or obesity with or without diabetes. The date of the search was from the inception to December 2025. The search strategy used both Medical Subject Headings (MeSH) and free text terms, including obesity, overweight, body weight change, weight loss, weight reduc*, weight decreas*, weight control*, placebo-controlled, and RCTs. A professional librarian has reviewed the search strategy according to the PRESS guidelines¹⁰. The complete search strategy can be found in Table S1. Any discrepancies were resolved by a senior reviewer (XC).

Study selection and data extraction

The inclusion criteria for eligible studies were as follows: (1) placebo-controlled, randomized clinical trials; (2) trials of anti-obesity medications conducted in patients with overweight or obesity with or without diabetes; (3) trials reporting weight changes from baseline to the end of study in placebo treatment arm. Observational studies, reviews, meta-analyses, or cross-over design studies were excluded. Studies conducted with an active agent control or published in non-English language were excluded as well.

Two investigators (YG and WY) independently reviewed the titles, abstracts, and full-texts of the research articles. They excluded any duplicate or ineligible items, assessed the risk of bias of the remaining studies using the Cochrane risk-of-bias tool¹¹, and extracted data from eligible studies with a standardized form: publication data (first author, published year), study design, treatment arms, study duration, baseline characteristics (mean age, sex ratio, baseline body mass index, baseline body weight) and weight changes from baseline to the end of study. If the above data were not found in both articles and supplementary files, the data from clinicaltrials.gov website would be extracted. Another three investigators (GL, SB, and ZL) checked the results of extractions, with any disagreements or discrepancies resolved by a senior reviewer (XC).

Assessment of risk of bias

The risk-of-bias of the included trials was independently assessed by two reviewers with the Cochrane risk-of-bias tool, which has seven domains: adequate randomization sequence generation, adequate allocation concealment, blinding of participants and caregivers, blinding of outcome assessors and adjudicators, free of frequent missing outcome data, free of selective outcome reporting, and free of other bias¹¹. For each domain, we rated definitely yes (low risk of bias), probably yes, probably no, and definitely no (high risk of bias) based on the description in the articles.

Data synthesis and analysis

The meta-analysis was performed by computing the pooled effect size (ES) and 95% confidence intervals (CIs) of weight changes to evaluate the placebo response in placebo treatment group. We used mean changes from baseline and standard deviations (SDs) extracted from published data when reported. In case only baseline and post-intervention values were available, the means and SDs of change were estimated by using the method described in the Cochrane Handbook Version 6.3¹². When SDs were missing, we estimated them from standard errors (SEs) or confidence intervals. The heterogeneity between included studies was evaluated by Higgins I² statistics and the tau-squared test. For estimated I² values of 0–25%, 25–50%, 50–75%, and 75–100%, heterogeneity was defined as low, moderate, substantial, or considerable, respectively. The random-effect model was used in this meta-analysis. Publication bias was assessed via the funnel plot and evaluated by Egger’s test. A sensitivity analysis was conducted to reduce the heterogeneity in our analyses. Subgroup analyses by baseline characteristics and other related factors, including baseline body mass index (BMI), baseline body weight, treatment design, administration route, indication, study duration, publication year, and comparator agent type, were also performed. To calculate the associations between different variables and the placebo response on body weight, we used meta-regression analysis for continuous variables: age, male percentage, baseline BMI, baseline body weight, and study duration, and used ANOVA tendency analysis for discontinuous variable: publication year.

Pool ES analyses, subgroup comparisons, and meta-regression analyses were conducted

by STATA software, version 17.0 (STATA, College Station, TX, USA). ANOVA tendency analysis was conducted by SPSS software (SPSS 24.0, Armonk, NY: IBM Corp). Egger’s test and funnel plot were conducted by STATA software, version 17.0 (STATA, College Station, TX, USA). Statistical significance was considered at P < 0.05.

Results

Characteristics and quality assessment of included studies

Overall, 182 trials were included in this meta-analysis (Fig. 1). Among these trials, 69 trials were conducted in patients with overweight or obesity without diabetes mellitus (DM), 107 trials were conducted in patients with overweight or obesity with type 2 diabetes mellitus (T2DM), 5 trials were conducted in patients with overweight or obesity with type 1 diabetes mellitus (T1DM) and 1 trial was conducted in patients with hypothalamic obesity with or without T2DM. Baseline characteristics of included studies were systematically summarized in Table S2.

Fig. 1. PRISMA flow diagram of included studies.

Records were identified via database searching. After removal of duplicates, records were screened and excluded according to title/abstract and full-text review. A total of 182 studies were included in the final analyses.

In all, 30 trials were with high risks of bias in frequent missing outcome data, 3 trials were with high risks of bias in allocation concealment, and no trials were with high risks of bias in other domains. Outcomes in detail were organized methodically in Table S3. The publication bias was evaluated by Egger’s test (β = −0.63, P = 0.479) (Table S4) and the funnel plot (Fig. S1), which implied that there was some publication bias in our analysis.

Effects of placebo response on body weight

It was found that in patients with overweight or obesity with or without diabetes, receiving placebo in anti-obesity medications trials generally resulted in a considerable reduction in body weight (ES = −1.13 kg, 95% CI, −1.29 to −0.97 kg, P < 0.01).

With or without diabetes

The placebo response of trials conducted in patients with overweight or obesity without diabetes showed a significantly greater weight reduction than those with T2DM (ES = −1.92 kg, 95% CI, −2.22 to −1.62 kg, P < 0.01 vs. ES = −0.72 kg, 95% CI, −0.92 to −0.52 kg, P < 0.01; P for subgroup comparison <0.001), while a significant weight increase was observed in patients with T1DM (ES = 0.76 kg, 95% CI, 0.46 to 1.05 kg, P < 0.01). Details were shown in Fig. 2.

Fig. 2. Summary of placebo response on body weight.

Pooled ES and 95% CIs were calculated using a random-effects model; between-study heterogeneity was evaluated by Higgins I² statistic and tau-squared test. Subgroup stratifications were performed by treatment design, administration route, indication, baseline BMI, baseline body weight, study duration, publication year, and comparator agent type. All statistical tests were two-sided, and no adjustment for multiple comparisons was applied. ES, effect size; T1DM, type 1 diabetes mellitus; T2DM, type 2 diabetes mellitus. * Deleting 1990-1999 subgroup. # Compared with oral subgroup, P < 0.05. $ Compared with obesity subgroup, P < 0.05. P value < 0.05 denotes statistical significance.

Oral or injectable route

As for administration route, significant weight reduction of placebo response was observed in both oral route and injectable one. Furthermore, participants receiving injectable placebo (ES = −1.46 kg, 95% CI, −1.74 to −1.19 kg, P < 0.01) experienced a greater weight reduction than those receiving oral placebo (ES = −0.82 kg, 95% CI, −1.05 to −0.59 kg, P < 0.01; P for subgroup comparison <0.001). However, patients with intranasal placebo tended to experience a weight gain (ES = 0.53 kg, 95% CI, −0.38 to 1.44 kg, P = 0.26).

Monotherapy or add-on therapy

In terms of monotherapy and add-on therapy, significant weight reduction of placebo response was found in both monotherapy (ES = −1.17 kg, 95% CI, −1.37 to −0.96 kg, P < 0.01) and add-on therapy (ES = −1.04 kg, 95% CI, −1.35 to −0.73 kg, P < 0.01) subgroups without group difference (P = 0.68).

Age and sex

As for age, the weight loss of placebo response gradually decreased as the patients got older (Fig. 3a). Meta-regression analysis consistently showed that younger age was significantly associated with more weight reduction in placebo response (β = 0.04, 95% CI, 0.02 to 0.06, P < 0.01).

Fig. 3. Meta-regression analysis for factors associated with placebo response.

a Association between age and placebo response (β = 0.04, 95% CI, 0.02 to 0.06, P < 0.01), b Association between male percentage and placebo response (β = 0.01, 95% CI, 0.02 × 10⁻² to 0.02, P = 0.044). c Association between baseline BMI and placebo response (β = −0.14, 95% CI, −0.21 to −0.08, P < 0.01). d Association between baseline body weight and placebo response (β = −0.06, 95% CI, −0.07 to −0.04, P < 0.01). e Association between study duration and placebo response (β = −0.4 × 10⁻², 95% CI, −0.9 × 10⁻² to 0.02 × 10⁻², P = 0.057). Meta-regression was used for age, male percentage, baseline BMI, baseline body weight, and study duration to assess their associations with placebo response. All statistical tests were two-sided, and no adjustment for multiple comparisons was performed.

In terms of sex, the weight loss of placebo response was negatively correlated with the proportion of men (Fig. 3b). As meta-regression analysis further shown, lower male percentage was associated considerably with greater weight reduction in placebo response (β = 0.01, 95% CI, 0.02 × 10⁻² to 0.02, P = 0.044).

Baseline BMI

As baseline BMI decreased, the magnitude of weight loss in placebo response diminished gradually (Fig. 2). Meta-regression analysis found that greater baseline BMI was significantly associated with more weight loss in placebo response (β = −0.14, 95% CI, −0.21 to −0.08, P < 0.01) (Fig. 3c).

Baseline body weight

For participants whose baseline body weight was above 70 kg, with the increase of baseline body weight, an increasing tendency in the magnitude of weight loss in placebo response was found (Fig. 2). Meta-regression analysis also indicated that higher baseline body weight was significantly associated with greater weight reduction in placebo response (β = −0.06, 95% CI, −0.07 to −0.04, P < 0.01) (Fig. 3d).

Study duration

Various degrees of weight loss in placebo response were shown in all strata of study duration (Fig. 2). No significant association between weight change of placebo response and study duration was found (β = −0.4 × 10⁻², 95% CI, −0.9 × 10⁻² to 0.02 × 10⁻², P = 0.066) (Fig. 3e).

Publication year

Before 2019, the weight loss of placebo response got smaller with publication year getting later. Until the last 4 years, there was a slight increase in the amplitude of weight loss of placebo response, but little changed overall (Fig. 2). ANOVA tendency analysis indicated that publication year was significantly associated with the weight change in placebo response (Linear term = 22.03, P = 0.008). However, when stratified by the published year, we found that only one trial prior to 1999, which probably increased the bias of our analysis. Therefore, sensitivity analysis was conducted by excluding the trial published before 1999 and then we found that the correlation between the publication year and the weight change in placebo response turned to be insignificant (Table 1).

Table 1.

Meta-regression analysis and ANOVA tendency analysis of placebo response with associated factors

Variables	Weight change
Multivariate meta-regression analysis
	β	95% CI	P value
Age (year)	0.04	0.02, 0.06	<0.01
Male percentage (%)	0.01	0.02×10−2, 0.02	0.044
Study duration (week)	−0.4 × 10−2	−0.9 × 10−2, 0.02 × 10−2	0.066
Baseline BMI (kg/m2)	−0.14	−0.21, −0.08	<0.01
Baseline body weight (kg)	−0.06	−0.07, −0.04	<0.01
ANOVA tendency analysis
	Linear term	F value	P value
Publication year	22.03	8.26	0.008
Publication yeara	4.32	1.65	0.314

Meta-regression was used for continuous variables (age, male percentage, baseline BMI, baseline body weight, study duration) to assess their associations with placebo response; ANOVA tendency analysis was applied for the discontinuous variable (publication year). All statistical tests were two-sided, and no adjustment for multiple comparisons was performed.

BMI body mass index.

^aDeleting 1990–1999.

P value < 0.05 denotes statistical significance.

Discussion

Involving 182 placebo-controlled trials of AOMs, this meta-analysis described that the overall placebo response on body weight was a reduction of 1.13 kg, which was associated with age, male percentage, baseline BMI, and baseline body weight. Realizing the placebo response in placebo-controlled AOMs trials may help us to interpret the existing published trials more objectively and to judge the true clinical values of AOMs. Additionally, understanding the associated factors of placebo response may help us to control these factors for further improving the design and the accuracy of future trials.

Weight loss in individuals with obesity could improve cardiometabolic parameters such as blood glucose, blood pressure, and lipid profiles, enhance insulin sensitivity, and reduce hepatic fat accumulation. These effects collectively contribute to a marked reduction in the risk of major adverse cardiovascular events (MACE) and a slower progression of chronic kidney disease (CKD). In clinical trials of weight-loss medications, a 1.13 kg weight reduction observed in the placebo group suggested that even patients receiving placebo may experience modest weight loss, which could potentially benefit cardiorenal and metabolic risk profiles. However, this significant placebo-associated weight reduction may also confound the assessment of the true efficacy of the active drug. Therefore, interpretation of trial results should carefully account for the placebo response.

The underlying mechanisms of placebo response are still complex and are not fully understood, which are commonly attributed to natural history of diseases, fluctuation of symptoms, and statistical regression to the mean⁶. In addition, placebo response has proved to be a psychobiological phenomenon attributable to the overall therapeutic context, where individual patient, clinician factors, and the interaction between the patient, clinician, and treatment environment are included^13,14. For obesity, lifestyle modification is the acknowledged first-line treatment¹⁵, including improved dietary patterns, increased physical activity, or enhanced health education and awareness. In most of the trials included in this study, participants received lifestyle interventions in addition to either the active drug or placebo, which might also be responsible for the weight reduction observed in patients with placebo.

According to this meta-analysis, it was found that the injectable placebo resulted in a greater weight reduction than the oral one. Two articles estimating the placebo response in trials of hypoglycemic medications reported similar findings^16,17. Although mechanisms underlying has not been fully understood, it was speculated that the greater placebo response might result from the stronger psychological feedback by injection administration¹⁶. Patients injected with placebo received a stronger sensory shock and greater psychological suggestion, which probably raised their expectation and adherence to the anti-obesity treatment.

We observed a greater weight loss in placebo response for individuals with overweight or obesity only, when compared with those simultaneously with T2DM. Similarly, another study found that the placebo response rate was higher in patients with overweight or obesity without DM than those with DM⁷. Possible reasons for this might be first speculated as that, compared with those without DM, the basic physical and health conditions of patients with overweight or obesity with DM might be worse, since those with DM were more likely to develop cardiovascular or renal diseases, which probably limited the intensity of physical activity and affected the weight loss of placebo response. Furthermore, according to our meta-analysis, patients with overweight or obesity with DM were older than those without DM (P < 0.01). And the age difference between these two populations may lead to different cognitive levels of the disease. It was supposed that patients with DM might have to receive more complex management strategies, which may affect their adherence to anti-obesity interventions. Additionally, many patients with DM included also received hypoglycemic pharmacotherapies, such as insulin, sulfonylurea, and thiazolidinedione, which resulted in body weight gain as well, thus interfered the placebo response.

Whether age is a significant associated factor for placebo response has always been a concern for many researchers. A meta-analysis across medicine reported that age seemed to have no impact on the placebo response in RCTs, as they revealed that most analyses (54/74) included failed to find an association between placebo response and age¹⁸. However, the findings of Lin et al.¹¹ and Chin et al.^7,16, which determined that age was significantly associated with placebo response in diabetes or obesity, are contrary to the conclusion of the meta-analysis across medicine. According to our study, younger age was associated with greater weight loss of placebo response. The pathophysiological and psychological differences between the young and old may account for it. Younger patients probably have more expectations for weight loss and are more likely to adhere to routine management. On the contrary, older patients may have experienced more failure during treatment and have less confidence and enthusiasm for beneficial outcomes of AOMs treatment.

The relationship between sex and placebo response remains a subject of ongoing debate in the literature. A systematic review examining sex differences in placebo effects noted that, due to insufficient data prior to 2010, it was unable to draw definitive conclusions regarding the influence of sex¹⁹. For instance, in studies of placebo analgesia, while one report suggested that males may demonstrate a stronger placebo response than females²⁰, others conducted in models such as tooth extraction²¹ or transcutaneous electrical nerve stimulation²² found no significant sex-based differences. More recently, a 2020 study by Lin et al. indicated that a lower proportion of males was associated with a greater placebo-induced reduction in HbA1c in type 2 diabetes, supporting a potential role of sex in modulating placebo response¹⁶. Our findings align with this observation, as we also recorded greater weight loss in patient groups with a lower male percentage. In contrast, Chin et al. reported no such association in their analysis, observing no significant sex-related trends in either the proportion of patients achieving weight loss thresholds or in absolute weight reduction values⁷. Thus, sex remains an inconsistent and contested predictor of placebo response, and more researches are needed to further prove their relationship.

In T2DM, baseline HbA1c and baseline BMI, as indicators of disease severity, were proved to be positively associated with the HbA1c reduction with placebo treatment¹⁶. However, after being adjusted for other baseline factors, such relationships turned to disappear, which indicated this correlation may be due to the regression to the mean. In Crohn’s Disease, a negative correlation was shown between disease severity and placebo response rate, and the concept that spontaneous remission is more difficult to obtain in patients with more severe conditions may explain it²³. In our study, we found that higher baseline BMI and higher baseline body weight were associated with greater weight reduction of placebo response in overweight or obesity, which is consistent with Chin’s findings⁷. Regarding the distinct relationships between disease severity and placebo response shown in different diseases, we speculate that it may be attributed to the different pathophysiological etiologies. As for overweight or obesity, patients with higher baseline BMI or higher baseline body weight may gain more benefits from non-pharmacological treatments. More researches need to be conducted to find more evidence to further explain the association between disease severity and placebo response.

Our study showed that study duration was not associated with placebo response, which had been observed in the previous studies as well, such as in migraine prevention²⁴ and ulcerative colitis²⁵. But another study reported that there was a positive correlation between study duration and placebo response in obesity pharmacological trials. They speculated that the possible reason was due to more intensive and regular follow-up in the longer study duration, which could lead to more benefits when receiving the same pharmaceutical treatment⁷. Conversely, several trials of antipsychotic medications reached the opposite conclusion that shorter study duration resulted in a greater placebo response^26–28. Consequently, whether study duration is related to placebo response requires more evidence.

Publication year has been a controversial factor for placebo response. In this analysis, we found that earlier publication year was significantly associated with greater weight reduction of placebo response. Considering that only 1 trial in 1990–1999 was included, we conducted a sensitivity analysis by excluding the extreme study and found this association turn to disappear. Additionally, an earlier meta-analysis reported that there was no correlation between publication year and placebo response in obesity pharmacological trials⁷. However, other studies focusing on neuropathic pain²⁹, depression³⁰, and migraine prevention²⁴ found that the placebo response was shown to be increasing over time. The discrepancy in the selection criteria for trials and the analysis method used to validate the correlation might be the probable explanation for the different conclusions observed in the above studies²⁶.

There are still some limitations in our study. Firstly, we were unable to evaluate the true placebo effect because most trials included did not set a no-treatment arm. Hence, part of the placebo response observed in our study may be attributed to the natural history of the disease, the spontaneous fluctuation of symptoms, and the regression to the mean. In addition, many trials set a program of diet, activity, and behavioral guidance as concurrent background weight loss treatments, which have been confirmed to associate with metabolic improvements. However, we were unable to quantify the magnitude of weight loss resulting from these lifestyle interventions. Besides, we also tried to investigative whether trials with lifestyle interventions would observe a greater weight reduction than those without them. But it was difficult to find a definitive answer due to the poor reporting of these background treatments in most studies. Hopefully, it will be enriched in further investigations. Additionally, the included trials differed in terms of inclusion criteria, baseline characteristics, and experimental design, which might lead to heterogeneity among them. To control the heterogeneity, we conducted subgroup analyses and meta-regression analyses. Moreover, only trials reporting weight changes from baseline to the end of study in placebo group were included, while trials reporting placebo-adjusted weight changes were excluded due to the failure to calculate exact weight change values in placebo group based on the incomplete data reported. And for trials without reporting SD, SE, or 95% CI of weight change either in articles (including Supplements) or on the Clinicaltrial.gov website, we also removed them from our study. The exclusion of these trials may have resulted in the absence of some data available for analysis in our study. To deal with this issue, we have contacted the authors to get as much information as possible. Besides, we failed to conduct subgroup analysis stratified by ethnicity, because most included trials covered multiple races. Additionally, according to a review of novel noninvasive treatments to the obesity, SGLT2i (sodium-glucose cotransporter 2 inhibitor) was recognized as one of the potential anti-obesity strategies³¹. Therefore, we also included 44 trials of SGLT2i, which would enable us to obtain more comprehensive data, but also interfere with the interpretation of placebo response in trials of true anti-obesity medications to some extent. Moreover, the ecological bias of meta-regression analyses might be inevitable, as the analyses were conducted based on the trial level data. Due to the lack of individual data reported in the original studies, we were unable to evaluate the placebo response of anti-obesity trials at individual level. Therefore, the results of meta-regression analyses need to be explained seriously. Hopefully it will be enriched in further investigations to validate our findings. Besides, the publication bias might also influence the accuracy and reliability of the results, therefore, we should take cautions to explain the final results. Finally, although we have demonstrated the placebo response in AOM trials, we were unable to definite the components involved in this due to the complex mechanisms of placebo response. Further research will be encouraged to explore this issue.

Conclusions

Our meta-analysis systematically described the placebo response in RCTs of anti-obesity medications, which was evaluated as a significant weight loss of 1.13 kg in patients with overweight or obesity with or without diabetes. Younger age, lower male percentage, higher baseline BMI and higher baseline body weight were potentially associated with greater placebo response in patients receiving placebo treatment. Results from this study may urge the researchers to take more considerations when designing and conducting placebo-controlled clinical trials in future AOMs trials for obesity.

Author contributions

L.J. and X.C. conceptualized this study and designed the systematic review protocol; Y.G., W.Y., Z.L., S.B., and G.L. performed the study selection and data extraction; W.Y., Y.G., H.W., R.J., and X.Y. performed the statistical analyses; Y.G., W.Y., C.L., and X.C. prepared the outlines and wrote the manuscript. All authors contributed to the critical revision of manuscript drafts.

Peer review

Peer review information

Communications Medicine thanks Hojin Oh and the other, anonymous, reviewer for their contribution to the peer review of this work.

Funding

This work was supported by National Natural Science Foundation of China (No.81970698) and Beijing Natural Science Foundation (No.7202216). The funding agencies had no roles in the study design, data collection, or analysis, decision to publish, or preparation of the manuscript.

Data availability

All data relevant to the study are included in the article or uploaded as appendix information. No more additional data are available.

Competing interests

Linong Ji has received fees for lecture presentations and for consulting from AstraZeneca, Merck, Metabasis, MSD, Novartis, Eli Lilly, Roche, Sanofi-Aventis, and Takeda. All other authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s43856-026-01659-7.