The burden of sarcopenia in people living with HIV: a systematic review and meta-analysis across WHO regions Authors: Vaibhavkumar N Shrivastav, Mohit N Makwana, Deep P Shah, Zinal Naresh Surti Published: 2026 ABSTRACT AbstractBackgroundSarcopenia, characterized by progressive loss of skeletal muscle mass and function, represents an underrecognized comorbidity in people living with human immunodeficiency virus, and comprehensive prevalence data across WHO regions remain lacking.ObjectiveTo estimate the prevalence of sarcopenia across WHO regions among people living with human immunodeficiency virus (HIV) and to identify demographic, clinical, and methodological factors associated with variations in reported prevalence.MethodsWe searched PubMed, Embase, Scopus, and Web of Science for observational studies published from 2000 onwards that reported the prevalence of sarcopenia in people living with human immunodeficiency virus. Prevalence estimates were pooled using a random-effects meta-analysis. Subgroup analyses examined variations by geographic region, diagnostic criteria, demographic characteristics, and virus-related factors. Publication bias was assessed using Egger’s and Begg’s tests.ResultsAcross 34 studies involving 7,216 participants in 15 countries, the pooled prevalence of sarcopenia was 16% (95% confidence interval: 12–21%). Substantial heterogeneity (93.6%) was observed. Prevalence estimates varied substantially by diagnostic criteria, ranging from 8% (EWGSOP2 2019) to 28% (Baumgartner criteria), highlighting the importance of standardised diagnostic approaches in future research. Geographic variation was evident, with Romania reporting the highest prevalence (34%). Meta-regression identified body mass index status as a significant moderator of variability in prevalence, and age above 50 years was associated with higher sarcopenia prevalence [odds ratio: 1.97 (1.30–2.98)]. No publication bias was detected. Sensitivity analysis demonstrated robustness. Evidence certainty was moderate.ConclusionsSarcopenia affects approximately one in six people living with HIV across the represented WHO regions, with considerable variation across diagnostic approaches and populations. While diagnostic harmonisation is needed to enable consistent epidemiological surveillance, clinical management of sarcopenia in PLHIV must adopt a personalised, risk-stratified approach that accounts for the considerable heterogeneity across individuals in age, body composition, ART history, and socioeconomic context.RegistrationCRD 420,251,063,820.Clinical trial numberNot applicable.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12879-026-13472-7.Keywords: Sarcopenia, Human immunodeficiency virus, Prevalence, Meta-analysis, Global health Introduction Sarcopenia, derived from the Greek words “sarx” (flesh) and “penia” (loss), represents a progressive and generalised skeletal muscle disorder characterised by the accelerated loss of muscle mass, strength, and physical performance [1]. This syndrome extends beyond normal age-related muscle decline, manifesting as a clinically significant condition associated with adverse health outcomes, including falls, fractures, physical disability, and mortality. The European Working Group on Sarcopenia in Older People (EWGSOP) [2, 3] and the Asian Working Group for Sarcopenia (AWGS) [4, 5] have established standardized diagnostic frameworks that integrate measurements of muscle mass, muscle strength, and physical performance, enabling consistent identification and classification of this condition across diverse populations and clinical settings. The global epidemiological landscape of sarcopenia reveals it to be a condition of significant public health concern, with prevalence estimates ranging from 10% to 27% among adults aged 60 years and older; however, these figures vary considerably based on diagnostic criteria, assessment methodologies, and population characteristics [6]. People all around the world, in both rich and poor countries, have sarcopenia. But not everyone has access to monitoring equipment and the skills to diagnose it. The fact that the world’s population is getting older makes things worse. There will be 2.1 billion people aged 60 and over by 2050. This will considerably increase the number of people who are at risk [7]. The economic effects are huge because the expenses of treating sarcopenia are expected to be in the billions of dollars per year in industrialized countries alone. This is because more people are being hospitalized, it takes longer to recover, and there is a greater need for long-term care services [8]. People living with HIV (PLHIV) face a particularly elevated risk of developing sarcopenia, even in the era of effective antiretroviral therapy (ART). The intersection of HIV infection and sarcopenia creates a distinctive clinical scenario characterized by multiple contributory mechanisms [9]. Chronic immune activation persists despite virological suppression, generating sustained inflammatory states marked by elevated cytokines, including tumour necrosis factor-alpha and interleukin-6, which directly interfere with muscle protein synthesis and promote proteolysis [10]. HIV and several antiretroviral medicines can damage mitochondria, which makes it tougher for muscle cells to consume energy. People with HIV often have hormonal changes that make it extra tough for their muscles to expand. These alterations include hypogonadism and growth hormone deficiency [11]. Additionally, metabolic complications such as insulin resistance and lipodystrophy syndrome contribute to altered body composition patterns. The ageing HIV population faces compounded risks as traditional age-related sarcopenia synergises with HIV-specific pathophysiological processes, creating an accelerated trajectory of muscle loss that may manifest at younger chronological ages compared to HIV-negative populations [12]. Critically, this accelerated trajectory means that sarcopenia in PLHIV may emerge at substantially younger ages than in the general population, with evidence suggesting clinically significant muscle loss occurring from as early as the fourth and fifth decades of life in this group [12]. This has important implications for the age at which screening and preventive interventions should begin: a focus exclusively on those aged 65 years and above may miss a substantial proportion of PLHIV already experiencing sarcopenia-related functional decline. Understanding the full age spectrum of sarcopenia burden in PLHIV is therefore essential to informing when healthy ageing interventions should be initiated in this population. Despite growing recognition of sarcopenia as a significant comorbidity in HIV care, current understanding remains fragmented. Available evidence comes from studies using heterogeneous diagnostic criteria, variable assessment methods, and diverse population characteristics, making it difficult to establish a coherent global perspective. Geographic disparities in research coverage leave substantial knowledge gaps, particularly in low- and middle-income countries where most PLHIV reside. Furthermore, the relationships between sarcopenia prevalence and factors such as ART regimens, HIV disease duration, immunological status, and nutritional parameters require systematic characterization. The clinical implications extend beyond academic interest, as sarcopenia in PLHIV associates with reduced quality of life, increased frailty, accelerated functional decline, higher rates of non-AIDS-defining comorbidities, and elevated mortality risk [13]. The global health community has increasingly prioritized healthy aging and the prevention of age-related functional decline. The United Nations Sustainable Development Goals (SDGs), particularly SDG 3, which aims to ensure healthy lives and promote well-being for all at all ages, explicitly encompasses the goal of reducing premature mortality from non-communicable diseases and supporting healthy ageing [14]. The World Health Organisation’s Decade of Healthy Ageing (2021–2030) emphasizes the importance of maintaining functional ability across the life course, recognizing sarcopenia as a key determinant of functional capacity in older populations [7]. For PLHIV, these global health priorities hold particular relevance as advances in HIV treatment have transformed the infection into a manageable chronic condition, with life expectancies approaching those of the general population in well-resourced settings. However, achieving truly healthy ageing for PLHIV requires addressing not only virological suppression but also the constellation of age-related and HIV-associated comorbidities, including sarcopenia [15]. Despite the well-characterised mechanistic pathways driving sarcopenia in PLHIV, a critical translational gap remains: the actual population-level burden of sarcopenia in this group has not been reliably quantified. Knowing that chronic immune activation, mitochondrial dysfunction, and hormonal dysregulation contribute to muscle loss does not, in itself, tell clinicians how prevalent sarcopenia is in real-world HIV populations, how this prevalence varies across different geographic and socioeconomic contexts, or which subgroups carry the greatest burden and therefore warrant priority screening. Without this epidemiological evidence, it is not possible to make a case for systematic screening, to allocate healthcare resources appropriately, or to design targeted interventions. Furthermore, while the biological mechanisms of sarcopenia in PLHIV are increasingly understood, the extent to which demographic and clinical factors — such as age, BMI, ART duration, and diagnostic criteria — modify observed prevalence in practice remains poorly characterised across diverse real-world settings. This distinction between mechanistic understanding and epidemiological evidence is the gap this review addresses. This systematic review and meta-analysis, therefore, aims to quantify the global prevalence of sarcopenia among PLHIV, examine variation across geographic regions, diagnostic frameworks, and population subgroups, and identify factors associated with differences in observed prevalence — with the ultimate goal of informing screening protocols, resource allocation, and evidence-based clinical practice for this growing population. Clearly defined primary and secondary objectives guided this review. The primary outcome was to determine the pooled global prevalence of sarcopenia among PLHIV aged 18 years and above. In addition to this primary aim, several secondary outcomes were examined: the identification of risk factors associated with sarcopenia in this population; variations in prevalence according to the diagnostic criteria applied, including AWGS, EWGSOP, and Baumgartner frameworks; the assessment methods used for diagnosing sarcopenia across different clinical and geographic settings; and country-specific and regional prevalence patterns worldwide. Methods Protocol development and registration Context and protocol development of the review: This systematic review adhered to the Preferred Reporting Items for Systematic Review and Meta-Analysis guidelines [16] and included published research from the year 2000 onwards. This date was chosen because it coincides with the post-combination antiretroviral therapy era, during which the clinical and body composition profiles of PLHIV changed substantially, and because validated sarcopenia diagnostic frameworks had become operationalised in research settings by this period [2–5, 17]. This systematic review and meta-analysis were registered on the International Prospective Register of Systematic Reviews (CRD 420251063820). Context and protocol development of the review: This systematic review adhered to the Preferred Reporting Items for Systematic Review and Meta-Analysis guidelines [16] and included published research from the year 2000 onwards. This date was chosen because it coincides with the post-combination antiretroviral therapy era, during which the clinical and body composition profiles of PLHIV changed substantially, and because validated sarcopenia diagnostic frameworks had become operationalised in research settings by this period [2–5, 17]. This systematic review and meta-analysis were registered on the International Prospective Register of Systematic Reviews (CRD 420251063820). This prospective registration enhances the transparency and reproducibility of our methodology while minimising the risk of outcome reporting bias. The protocol specified in advance all eligibility criteria, search strategies, data extraction procedures, risk of bias assessment methods, and planned statistical analyses, thereby establishing a rigorous framework for the systematic review process. Eligibility criteria and PICOS framework The following PICOS algorithm principles guided the initial retrieval: The following PICOS algorithm principles guided the initial retrieval: P(population): Adults aged ≥ 18 years or older with confirmed HIV infection, encompassing both treatment-naïve individuals and those receiving antiretroviral therapy, both community-dwelling and institutionalised individuals; Studies conducted exclusively in pediatric populations (individuals younger than 18 years) were excluded to maintain focus on adult sarcopenia manifestations. The decision to include all adults aged ≥ 18 years, rather than restricting inclusion to older age groups (e.g., ≥ 50 or ≥ 65 years), was deliberate and scientifically justified for three reasons. First, HIV-associated sarcopenia is recognised to manifest at earlier ages than in the general population due to chronic immune activation, ART-related metabolic effects, and prolonged catabolic states, making a restriction to older age groups likely to underestimate the true burden. Second, including a broad age range enables characterisation of age-related variation in prevalence across the full adult HIV care continuum, which is necessary to identify at which age groups screening should be prioritised. Third, the majority of available primary studies in this field recruited mixed-age adult populations; an age restriction would have excluded the majority of eligible evidence, severely limiting the scope and generalisability of the review. To address age-related variation analytically, age composition of study populations was examined as a dedicated subgroup variable, with studies stratified into younger versus older cohorts. P(population): Adults aged ≥ 18 years or older with confirmed HIV infection, encompassing both treatment-naïve individuals and those receiving antiretroviral therapy, both community-dwelling and institutionalised individuals; Studies conducted exclusively in pediatric populations (individuals younger than 18 years) were excluded to maintain focus on adult sarcopenia manifestations. The decision to include all adults aged ≥ 18 years, rather than restricting inclusion to older age groups (e.g., ≥ 50 or ≥ 65 years), was deliberate and scientifically justified for three reasons. First, HIV-associated sarcopenia is recognised to manifest at earlier ages than in the general population due to chronic immune activation, ART-related metabolic effects, and prolonged catabolic states, making a restriction to older age groups likely to underestimate the true burden. Second, including a broad age range enables characterisation of age-related variation in prevalence across the full adult HIV care continuum, which is necessary to identify at which age groups screening should be prioritised. Third, the majority of available primary studies in this field recruited mixed-age adult populations; an age restriction would have excluded the majority of eligible evidence, severely limiting the scope and generalisability of the review. To address age-related variation analytically, age composition of study populations was examined as a dedicated subgroup variable, with studies stratified into younger versus older cohorts. I(intervention): Not applicable (prevalence study); I(intervention): Not applicable (prevalence study); C(comparison): no comparison; C(comparison): no comparison; O(outcome): pooled prevalence of sarcopenia in PLHIV globally determined using validated diagnostic criteria. Studies were required to employ established sarcopenia diagnostic frameworks, including the EWGSOP criteria (2010 or 2019 versions) [2, 3], AWGS criteria (2014 or 2019 versions) [4, 5], or the Baumgartner criteria [17] and Studies utilising non-validated, investigator-defined, or unclear diagnostic criteria were excluded to ensure the validity and comparability of prevalence estimates. Detailed diagnostic criteria definitions and cut-off values for each consensus guideline are presented in Supplementary Table 1. O(outcome): pooled prevalence of sarcopenia in PLHIV globally determined using validated diagnostic criteria. Studies were required to employ established sarcopenia diagnostic frameworks, including the EWGSOP criteria (2010 or 2019 versions) [2, 3], AWGS criteria (2014 or 2019 versions) [4, 5], or the Baumgartner criteria [17] and Studies utilising non-validated, investigator-defined, or unclear diagnostic criteria were excluded to ensure the validity and comparability of prevalence estimates. Detailed diagnostic criteria definitions and cut-off values for each consensus guideline are presented in Supplementary Table 1. It is important to note that these three diagnostic frameworks do not capture identical constructs: the Baumgartner criteria rely exclusively on muscle mass (appendicular skeletal muscle mass indexed to height²), representing a mass-centred definition; the EWGSOP criteria (both 2010 and 2019 versions) incorporate muscle mass alongside strength and physical performance, with EWGSOP2 2019 placing primacy on low muscle strength as the key diagnostic parameter; and the AWGS criteria similarly integrate mass, strength, and performance but apply ethnicity-specific cut-off values developed for Asian body composition profiles. Consequently, studies using these frameworks are not measuring precisely the same biological construct, and differences in reported prevalence across criteria reflect genuine conceptual divergence in how sarcopenia is defined, not merely methodological variation. Studies had to report either the number of sarcopenia cases and total sample size or sufficient data to enable calculation of prevalence estimates with confidence intervals. S(study): cross-sectional study/retrospective study/cohort studies (baseline data)/surveillance data, hospital-based studies/community-based surveys/randomized control trials (baseline data)/population-based surveys/time series analysis were included and case reports, case series with fewer than 20 participants, narrative reviews, systematic reviews and meta-analyses (though their reference lists were examined for eligible primary studies), editorials, commentaries, opinion pieces, conference abstracts without accompanying full-text publications were excluded. S(study): cross-sectional study/retrospective study/cohort studies (baseline data)/surveillance data, hospital-based studies/community-based surveys/randomized control trials (baseline data)/population-based surveys/time series analysis were included and case reports, case series with fewer than 20 participants, narrative reviews, systematic reviews and meta-analyses (though their reference lists were examined for eligible primary studies), editorials, commentaries, opinion pieces, conference abstracts without accompanying full-text publications were excluded. Information sources and search strategy A comprehensive systematic search was conducted across many electronic bibliographic databases to identify all potentially valuable literature. The primary databases searched included PubMed/MEDLINE, Embase, Scopus, the Web of Science Core Collection, and Google Scholar. A professional medical librarian assisted us in developing the search strategy. The study employed a combination of Medical Subject Headings (MeSH) terms and relevant text words about three principal concepts: HIV/AIDS, sarcopenia/muscle wasting, and prevalence/epidemiology. A pilot study assessed the comprehensive search methodology, which was enhanced over several testing cycles. It was subsequently verified to ensure it provided the most precise and sensitive outcomes. Since January 2000, no limits on language or region have been imposed on database access. Detailed search strategies for each database are presented in Supplementary Table 2. To supplement electronic database searching and minimize the risk of missing relevant studies, we implemented several additional search methods. First, we manually screened the reference lists of all included studies and relevant systematic reviews identified during the search process, using a backward citation tracking approach to identify additional eligible studies not captured through database searching. Second, we performed forward citation tracking of seminal papers in the field using citation indexing tools to identify more recent studies citing these foundational works. Third, we searched clinical trial registries, including ClinicalTrials.gov and the WHO International Clinical Trials Registry Platform, to identify completed studies with published results. Fourth, we contacted leading experts and researchers active in the fields of HIV medicine and sarcopenia research to inquire about unpublished or recently published studies that might meet our inclusion criteria. The search methodology and documentation followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) [16] guidelines to ensure transparency and reproducibility of the review process. Study selection process After all the systematic searches were done, all the records that were found were brought into the Rayyan software. Then, duplicate citations were located and removed using both automated and human methods. There were two phases of screening to choose the studies. Each step was performed by two reviewers working independently to minimise bias and errors. Two reviewers assessed each reference independently and compared it to the eligibility criteria that had already been established during the first round of screening titles and abstracts. After that, they sorted the studies into three groups: possibly eligible, definitely not eligible, or unsure. At this time, an intentionally open approach was utilized, and papers were preserved for full-text examination if it wasn’t evident from the title and abstract alone if they were eligible. Subsequently, full-text articles of all potentially eligible studies were obtained and underwent detailed independent review by the same pair of reviewers, who assessed each study against the complete eligibility criteria and documented specific reasons for exclusion. Any disagreements between reviewers at either screening stage were resolved through discussion and consensus, with unresolved discrepancies adjudicated by a third senior reviewer. The study selection process, including the number of records identified, duplicates removed, studies excluded at each stage with reasons, and the final number of included studies, was documented according to PRISMA guidelines. Data extraction and management A standardised, piloted data extraction sheet was developed to ensure systematic and consistent collection of relevant information from included studies. Data extraction was performed independently by two reviewers for each included study, with extracted data subsequently compared and any discrepancies resolved through discussion or consultation with a third reviewer when necessary. For each included study, we systematically extracted the following categories of information: study identification data (first author, publication year, country of conduct, World Health Organization region, World Bank income classification), study design characteristics (study type, sampling method, recruitment setting, data collection period, relationship to COVID-19 pandemic phases), participant characteristics (sample size, age distribution, sex distribution, HIV disease duration, CD4 count, viral load, antiretroviral therapy status and duration, body mass index categories), sarcopenia assessment methodology (diagnostic criteria employed, specific version of criteria, muscle mass assessment method, use of single versus multiple diagnostic tools), and outcome data (number of sarcopenia cases, total sample size, reported prevalence with confidence intervals, or raw data enabling prevalence calculation). Additional data were extracted regarding potential effect modifiers and sources of heterogeneity, including whether studies were conducted in countries with official national sarcopenia policies, the predominant sex composition of study samples, and quality assessment indicators. When studies reported multiple prevalence estimates using different diagnostic criteria or assessment methods, all estimates were extracted and treated as separate data points in subgroup analyses. When essential data were missing or unclear in published reports, we attempted to contact corresponding authors via email with specific requests for the required information. For studies lacking confidence intervals but reporting prevalence and sample sizes, confidence intervals were calculated using standard binomial methods. Studies were excluded only when essential data (overall prevalence, sample size, or diagnostic criteria) could not be obtained or calculated. Missing moderator variables were coded as “not reported” rather than excluded to maximize study retention. Risk of bias assessment The methodological quality and risk of bias of included studies were systematically evaluated using tools appropriate for observational prevalence studies. We employed the Joanna Briggs Institute Critical Appraisal Checklist for Studies Reporting Prevalence Data, a nine-item tool specifically designed for prevalence studies that evaluates appropriateness of the sampling frame and technique, adequacy of sample size, detailed description of study subjects and setting, validity and reliability of condition measurement, appropriateness of statistical analysis, and adequacy of response rate [18, 19]. Additionally, we implemented GRADE (Grading of Recommendations Assessment, Development and Evaluation) staging to assess the overall certainty of evidence across studies (Supplementary Table 3). Data synthesis and statistical analysis All quantitative analyses were performed using R statistical software version 4.5.1, incorporating the ‘meta’ and ‘metafor’ packages for meta-analytical computations. Meta-analytic pooling of prevalence estimates was performed using the random-effects model based on the Sidik Jonkman method, which accounts for both within-study and between-study variance [20]. This approach was selected in anticipation of substantial heterogeneity in prevalence estimates arising from differences in populations, diagnostic criteria, assessment methods, geographic settings, and study designs. Raw prevalence data (number of cases and sample size) were extracted from each study and transformed using the Freeman-Tukey double arcsine transformation to stabilise variance and normalise the distribution, particularly for prevalence values approaching the boundaries of zero or one. Pooled prevalence estimates with 95% confidence intervals were calculated and back-transformed for presentation. We used the I² statistic to find out how different the studies were from each other. This shows that a lot of the disparities in effect estimates are actual and not due to sampling errors. People often think of 25%, 50%, and 75% as low, moderate, and high levels of diversity, respectively. We also used a chi-square test to check for differences. This test usually doesn’t have a lot of statistical power, hence p-values less than 0.05 were used as proof of statistically significant heterogeneity. We used tau-squared (τ²) to find out how different the experiments were from one another. We built forest plots to show the prevalence estimates from each study adjacent to the overall estimate. This made it easier to see how the data were spread out and how similar they were from trial to trial. Subgroup and sensitivity analyses To explore sources of heterogeneity and examine factors potentially influencing sarcopenia prevalence, we conducted extensive preplanned subgroup analyses stratified by multiple variables. Geographic subgroups included WHO regions (European Region, Region of the Americas, South-East Asia Region, Western Pacific Region, African Region), individual countries, and World Bank income classifications (high-income, upper-middle-income, lower-middle-income). Diagnostic and methodological subgroups encompassed sarcopenia diagnostic criteria versions (EWGSOP1 2010, EWGSOP2 2019, AWGS 2014, AWGS 2019, Baumgartner), muscle mass assessment method (DXA versus non-DXA methods), and use of multiple diagnostic tools versus single-method approaches. Clinical and demographic subgroups included HIV disease duration (≤ 10 years versus > 10 years versus not reported), antiretroviral therapy status (ART-experienced versus ART-naïve versus mixed populations), ART duration (≤ 5 years versus > 5 years versus not reported), body mass index status (normal, overweight, mixed, not reported), age composition (younger versus older cohorts), and sex distribution (predominantly male versus predominantly female versus balanced). Study design subgroups examined temporal variations (COVID-19 pandemic period: pre-COVID, during COVID, post-COVID), study design type (cross-sectional versus cohort studies), sampling methodology (consecutive, random, convenience), sample size categories (< 101, 101–500, 501–1000 participants), and presence versus absence of national sarcopenia policies in study countries. For each subgroup analysis, pooled prevalence estimates were calculated separately for each subgroup category, and statistical tests for subgroup differences were performed using chi-square tests. Meta-regression analyses were conducted when sufficient studies were available (typically ≥ 10 studies) to examine continuous or ordinal moderator variables and to explore whether specific study-level characteristics could explain between-study heterogeneity in prevalence estimates. These meta-regression analyses were conducted as exploratory investigations rather than hypothesis-driven confirmatory tests. Given that the moderator variables (BODY MASS INDEX status, World Bank income classification, diagnostic comprehensiveness, ART duration, study design) were selected based on their clinical and methodological plausibility as potential sources of heterogeneity — rather than from pre-specified a priori hypotheses — their findings should be interpreted as hypothesis-generating. The moderators examined were pre-listed in the registered protocol to ensure transparency; however, given the limited number of included studies relative to the number of moderators tested, the risk of chance findings and overfitting cannot be excluded. Accordingly, all meta-regression results should be interpreted cautiously and regarded as exploratory signals that warrant confirmation in prospectively designed future studies. Sensitivity analyses were performed to assess the robustness of the findings, including excluding small studies (sample size < 50), excluding the leave-one-out method, and sequentially excluding individual studies to identify influential outliers. Assessment of publication bias Publication bias was assessed using multiple complementary approaches. Funnel plots of prevalence estimates against standard errors were visually inspected for asymmetry. Egger’s regression test and Begg’s rank correlation test were performed as statistical assessments, with p-values less than 0.05 considered significant. When publication bias was detected, trim-and-fill analysis estimated the number of missing studies and provided adjusted pooled estimates. The Doi plot and Luis Furuya-Kanamori index were generated as alternative assessments, with index values exceeding plus or minus two indicating major asymmetry [21]. Given that publication bias is less concerning for prevalence studies, these assessments were interpreted cautiously alongside other sources of asymmetry, such as heterogeneity and methodological quality differences. Results Literature search and identified results The systematic literature search across five databases (PubMed, Scopus, Embase, Web of Science, and Google Scholar) identified 1,142 records, of which 533 duplicates were removed, leaving 609 unique records for screening. Following title and abstract screening, 564 records were excluded, and 45 full-text articles were sought for retrieval. Six articles could not be retrieved, resulting in 39 full-text articles assessed for eligibility. Five studies were subsequently excluded: 3 did not use validated diagnostic criteria, and two did not report prevalence estimates. Ultimately, 34 studies encompassing 7,216 participants met the inclusion criteria and were included in the systematic review and meta-analysis, providing comprehensive evidence on sarcopenia prevalence among people living with HIV across diverse geographic regions and clinical settings (Fig. 1). PRISMA flowchart of the selection of studies for sarcopenia prevalence among HIV patients globally: a systematic review and meta-analysis Features of the included studies The 34 included studies were published from 2013 to 2025 and involved 7,216 participants from 15 countries across five WHO regions. The majority were cross-sectional studies (n = 33), with one cohort study included. Sample sizes ranged from 33 to 860 participants, with a median of 150. The majority of the studies were conducted in the Americas (n = 13), followed by Europe (n = 8), the Western Pacific (n = 7), South-East Asia (n = 5), and one from Africa. Most studies originated from upper-middle-income countries (n = 21), followed by high-income (n = 10) and lower-middle-income countries (n = 3). Twelve studies were conducted pre-COVID, 11 during the pandemic, and 11 post-pandemic. Different criteria were used to diagnose sarcopenia: EWGSOP (n = 19), AWGS (n = 11), and Baumgartner (n = 4). Muscle mass was assessed using DXA (n = 16) or non-DXA methods (n = 18). The majority of participants were on antiretroviral medicine (n = 27 studies with ART-experienced populations), primarily utilizing convenience sampling (n = 22) or sequential sampling (n = 11). The quality scores ranged from 1 to 8, with a median score of 6. This suggests that the overall quality of the methods was between modest and good (Table 1). Table 1Features of the articles selected as part of the meta-analysis and systematic reviewSr no.1st AuthorPublication YearCovid PeriodCountryWHO RegionWorld Bank ClassificationStudy DesignSampling MethodSample SizeHIV duration (in years)BMI StatusDiagnostic Criteria UsedMuscle Mass Assessment MethodART StatusART Duration (Years)Prevalence (%)95% CIQuality Score1Serrano-Villar et al. [22]2013Pre-CovidSpainEuropean RegionHigh-incomeCross-sectionalConsecutive sampling132> 10NRBaumgartnerDXAART experienced> 523.816.35–31.0652Wasserman [23]2013Pre-CovidUSARegion of the AmericasHigh-incomeCross-sectionalConsecutive sampling80> 10OverweightEWGSOPNon-DXAART experienced> 550.2–9.7853Pinto Neto LF et al. [24]2016Pre-CovidBrazilRegion of the AmericasUpper-middle-incomeCross-sectionalConsecutive sampling33NRMixedEWGSOPNon-DXAART experienced> 524.29.58–38.8144Deep Dutta et al. [25]2017Pre-CovidIndiaSouth-East AsiaLower-middle-incomeCross-sectionalConsecutive sampling220≤ 10NormalEWGSOPDXAART experiencedNR4033.52–46.4775Dutta D et al. [26]2018Pre-CovidIndiaSouth-East AsiaLower-middle-incomeCross-sectionalConsecutive sampling214NRNormalEWGSOPNon-DXAART experiencedNR21.9316.38–27.4786Aziz et al. [27]2018Pre-CovidMalaysiaWestern Pacific RegionUpper-middle-incomeCross-sectionalconvenience sampling315≤ 10MixedAWGSNon-DXAART experienced> 585-10.977Aziz et al. [27]2018Pre-CovidMalaysiaWestern Pacific RegionUpper-middle-incomeCohortconvenience sampling153≤ 10MixedAWGSNon-DXAART experienced> 5105.24–14.7578Hawkins KL, et al. [28]2018Pre-CovidUSARegion of the AmericasHigh-incomeCross-sectionalconvenience sampling199NROverweightAWGSDXAART experienced> 54134.16–47.8319Patricia Echeverría et al. [29]2018Pre-CovidSpainEuropean RegionHigh-incomeCross-sectionalconvenience sampling860≤ 10NormalBaumgartnerDXANR> 525.722.77–28.62610Paula Debroy et al. [30]2019Pre-CovidItalyEuropean RegionHigh-incomeCross-sectionalconvenience sampling169> 10NormalBaumgartnerDXAART experiencedNR27.821.04–34.55811Alves et al. [31]2020Pre-CovidBrazilRegion of the AmericasUpper-middle-incomeCross-sectionalconvenience sampling99≤ 10OverweightEWGSOPNon-DXAART experienced> 518.210.59–25.8612Almeida LL, et al. [32]2020Pre-CovidBrazilRegion of the AmericasUpper-middle-incomeCross-sectionalconvenience sampling83NRNREWGSOPDXAART experiencedNR125-18.99213Celia L. Gregson et al. [33]2021During CovidSouth AfricaRegion of AfricaUpper-middle-incomeCross-sectionalRandom sample805NROverweightEWGSOPDXAART experiencedNR11.69.38–13.81814Oliviera, et al. [34]2021During CovidBrazilRegion of the AmericasUpper-middle-incomeCross-sectionalconvenience sampling302NROverweightEWGSOPDXAART experiencedNR4.32.01–6.58515Oliviera, et al. [34]2021During CovidBrazilRegion of the AmericasUpper-middle-incomeCross-sectionalconvenience sampling302NROverweightEWGSOPDXAART experiencedNR10-2.12516Almeida et al. [35]2021During CovidBrazilRegion of the AmericasUpper-middle-incomeCross-sectionalconvenience sampling44≤ 10NormalEWGSOPNon-DXABoth≤ 52512.20–37.8517Houssein et al. [36]2022During CovidTurkeyEuropean RegionUpper-middle-incomeCross-sectionalConsecutive sampling100≤ 10OverweightEWGSOPNon-DXAART experienced> 5104.12–15.88518Konishi, Keiji et al. [37]2022During CovidJapanWestern Pacific RegionHigh-incomeCross-sectionalConsecutive sampling87> 10NormalAWGSNon-DXAART experienced> 5 years10.33.91–16.68519Aprianta Agus et al. [38]2022During CovidIndonesiaSouth-East AsiaLower-middle-incomeCross-sectionalConsecutive sampling70≤ 10NormalAWGSNon-DXAART experiencedNR32.921.89–43.90820EC Barbu et al. [39]2022During CovidRomaniaEuropean RegionHigh-incomeCross-sectionalconvenience sampling41≤ 10NormalEWGSOPDXAART experienced≤ 543.928.70-59.09621EC Barbu et al. [39]2022During CovidRomaniaEuropean RegionHigh-incomeCross-sectionalconvenience sampling42≤ 10NormalEWGSOPDXAART experienced> 523.810.92–36.67622Fion Wing Lam Luk et al. [40]2022During CovidChinaWestern Pacific RegionUpper-middle-incomeCross-sectionalconvenience sampling150> 10NRAWGSDXAART experiencedNR17.311.24–23.35823Fion Wing Lam Luk et al. [40]2022During CovidChinaWestern Pacific RegionUpper-middle-incomeCross-sectionalconvenience sampling150> 10NRAWGSDXAART experiencedNR27.320.17–34.42824Cabrera DM, et al. [41]2023Post_CovidPeruRegion of the AmericasUpper-middle-incomeCross-sectionalconvenience sampling104> 10OverweightEWGSOPDXANR> 515.388.44–22.31525Peng T, et al. [42]2024Post_CovidChinaWestern Pacific RegionUpper-middle-incomeCross-sectionalconvenience sampling377NRNormalAWGSNon-DXANRNR29.524.89–34.10626Yuenyongchaiwat et al. [43]2024Post_CovidThailandSouth-East AsiaUpper-middle-incomeCross-sectionalconvenience sampling114> 10NormalAWGSNon-DXAART experienced> 521.9314.33–29.52827Jäger et al. [44]2024Post_CovidGermanyEuropean RegionHigh-incomeCross-sectionalconvenience sampling379> 10OverweightEWGSOPDXANRNR2.10.65–3.54728Martins et al. [45]2024Post_CovidBrazilRegion of the AmericasUpper-middle-incomeCross-sectionalconvenience sampling218> 10NormalEWGSOPNon-DXAART experiencedNR8.74.95–12.44729Thet D, et al. [46]2024Post_CovidThailandSouth-East AsiaUpper-middle-incomeCross-sectionalConsecutive sampling177> 10MixedAWGSNon-DXAART experienced> 519.7713.90-25.64530Yide Xu et al. [47]2024Post_CovidChinaWestern Pacific RegionUpper-middle-incomeCross-sectionalconvenience sampling378NRNRAWGSNon-DXAART experienced≤ 57.14.51–9.68631S. De Vincentis et al. [48]2024Post_CovidItalyEuropean RegionHigh-incomeCross-sectionalConsecutive sampling307> 10OverweightBaumgartnerDXAART experienced> 534.929.56–40.23732Martins et al. [49]2025Post_CovidBrazilRegion of the AmericasUpper-middle-incomeCross-sectionalconvenience sampling218NROverweightEWGSOPNon-DXABothNR1.40-2.96733Martins et al. [49]2025Post_CovidBrazilRegion of the AmericasUpper-middle-incomeCross-sectionalconvenience sampling218NRNREWGSOPNon-DXABothNR11.16.9-15.27734Lara Cristina Vieira et al. [50]2025Post_CovidBrazilRegion of the AmericasUpper-middle-incomeCross-sectionalConsecutive sampling76NRNREWGSOPNon-DXAART experienced> 536.3825.56–47.198*Footnote: Same-cohort multiple estimates: Sr. no. 6 & 7 (Aziz et al., 2018) [27], 14 & 15 (Oliviera et al., 2021) [34], 20 & 21 (EC Barbu et al., 2022) [39], 22 & 23 (Fion Wing Lam Luk et al., 2022) [40], and 32 & 33 (Martins et al., 2025) [49] each contribute more than one estimate from the same cohort, reflecting different diagnostic criteria versions or participant subgroups. For the overall pooled estimate, only one estimate per cohort was included to avoid double-countingAbbreviations: ART = Antiretroviral Therapy; AWGS = Asian Working Group for Sarcopenia; BMI = Body Mass Index; CI = Confidence Interval; DXA = Dual-energy X-ray Absorptiometry; EWGSOP = European Working Group on Sarcopenia in Older People; NR = Not Reported; WHO = World Health Organisation Features of the articles selected as part of the meta-analysis and systematic review *Footnote: Same-cohort multiple estimates: Sr. no. 6 & 7 (Aziz et al., 2018) [27], 14 & 15 (Oliviera et al., 2021) [34], 20 & 21 (EC Barbu et al., 2022) [39], 22 & 23 (Fion Wing Lam Luk et al., 2022) [40], and 32 & 33 (Martins et al., 2025) [49] each contribute more than one estimate from the same cohort, reflecting different diagnostic criteria versions or participant subgroups. For the overall pooled estimate, only one estimate per cohort was included to avoid double-counting Abbreviations: ART = Antiretroviral Therapy; AWGS = Asian Working Group for Sarcopenia; BMI = Body Mass Index; CI = Confidence Interval; DXA = Dual-energy X-ray Absorptiometry; EWGSOP = European Working Group on Sarcopenia in Older People; NR = Not Reported; WHO = World Health Organisation Meta-analysis results A meta-analysis of 34 studies including 7,216 participants revealed that 16% (95% CI: 12–21%) of individuals with HIV exhibited sarcopenia. The random effects model was employed for this analysis. The prevalence estimates from each study varied considerably, ranging from 1% to 44%. This indicates substantial heterogeneity across studies (I² = 93.6%, τ² = 1.0286, p < 0.05). Studies with larger sample sizes, such as those by Patricia Echeverra et al. (2018) and Celia Gregson et al. (2021), showed narrower confidence intervals and greater statistical significance in the pooled analysis. The prediction interval (2–60%) indicates that the actual outcome in a novel study context may significantly differ. This indicates that the prevalence of sarcopenia may vary with factors such as population characteristics, diagnostic criteria, and geographical location (Fig. 2). Meta-analysis of forest plot Cumulative meta-analysis The cumulative meta-analysis illustrates the progression of sarcopenia prevalence estimates as studies were incrementally incorporated from 2013 to 2025. The first study reported a 23.5% prevalence rate, but this estimate changed substantially when more data were added in the first phase. The estimate remained unstable across the first 15 studies, with wide CI reflecting early uncertainty. The estimate of how widespread it was gradually stabilised as more research was added, going beyond the first 15–20. This was the same as the final pooled estimate of 16.0%, which is shown by the dashed red line. The prevalence estimate becomes more accurate and reliable as the 95% CI narrows over time. This pattern illustrates that the estimate of the worldwide sarcopenia burden in adults with HIV has been more accurate and dependable as more research from different groups and situations has been included. This is true even though there wasn’t much data at first, and it wasn’t consistent. The estimate stabilised after approximately 20 studies (Fig. 3). Cumulative meta-analysis Subgroup analysis by country Supplementary Fig. 1 presents the forest plot for country-level subgroup analysis; studies are ordered by WHO region, with pooled estimates shown per country and overall. The country-level subgroup analysis revealed marked variation in sarcopenia prevalence across five WHO regions. Brazil had the most studies in the Americas, with 10 (n = 1,593) revealing a pooled frequency of 13% (95% CI: 11–15%). The USA had a far higher prevalence of 37% (95% CI: 31–43%), Italy had 32% (95% CI: 29–34%), and Peru had 15% (95% CI: 9–22%). Considerable variation was also observed within Europe. Spain had 25% (95% CI: 23–28%), Romania had 35% (95% CI: 25–46%), and Turkey had 10% (95% CI: 3–16%). Germany, on the other hand, had the lowest global estimate at 2% (95% CI: 1–4%). In the Western Pacific area, China had a prevalence of 22% (95% CI: 19–25%), Malaysia had 9% (95% CI: 5–11%), Thailand had 21% (95% CI: 16–26%), and Japan had 10% (95% CI: 4–17%). Within the South-East Asia region, prevalence estimates were broadly similar across countries — India reported 32% (95% CI: 28–37%) and Indonesia 33% (95% CI: 22–45%) — though the wide CI, particularly for Indonesia (based on a single small study), reflect considerable uncertainty around these estimates. The sole African study (South Africa) reported a prevalence of 12% (95% CI: 9–14%). There was still a lot of variation overall, both within and between regions (I² = 93.6%, χ² = 305.08, p < 0.0001). Subgroup analysis by diagnosis tool Supplementary Fig. 2 presents three pooled forest plots — one per diagnostic framework (EWGSOP, AWGS, Baumgartner) — with a subgroup difference test shown at the bottom. Sarcopenia prevalence varied considerably depending on the diagnostic framework applied. Research using the European Working Group on Sarcopenia in Older People (EWGSOP) criteria (n = 20 studies, 3,578 participants) reported a pooled prevalence of 12% (95% CI: 7–19%) under a random-effects model, with significant heterogeneity (I² = 93.6%). The Asian Working Group for Sarcopenia (AWGS) criteria (n = 11 studies, 2,170 participants) showed a higher pooled prevalence of 18% (95% CI: 13–26%) in the random-effects model, reflecting significant heterogeneity (I² = 93.5%). Studies using the Baumgartner criterion (n = 4, 1,468 participants) reported a pooled prevalence of 28% (95% CI: 24–33%) with random-effects pooling, indicating somewhat lower, though still considerable, heterogeneity (I² = 71.8%). Subgroup difference testing confirmed significant variation across diagnostic frameworks (χ² = 70.64, p < 0.0001 for common effect; χ² = 13.76, p = 0.001 for random effects). These results indicate that the selection of diagnostic criteria significantly affects estimates of sarcopenia prevalence, as the Baumgartner criteria reveal higher prevalence rates in comparison to the EWGSOP and AWGS criteria. Subgroup analysis by diagnostic version Supplementary Fig. 3 stratifies forest plots by specific criteria version; readers should note that each version subgroup is displayed separately, with its own pooled diamond. Prevalence estimates varied substantially across specific diagnostic criteria versions. This was based on specific versions of diagnostic criteria. The Baumgartner criteria had the highest combined frequency of 28% (95% CI: 24–33%) in four studies with 1,468 people. with moderate heterogeneity (I² = 71.8%). The EWGSOP criteria showed different changes for each version. The EWGSOP1 2010 study found that 16% of the 1,422 subjects (from 10 different studies) had the condition (95% CI: 10–26%) (I² = 93%). On the other hand, the new EWGSOP2 2019 study (n = 9 studies, 2,156 people) found a much lower frequency of 8% (95% CI: 3–19%), with an equally high level of variability (I² = 92.8%). The Asian criterion showed more consistent values across different versions. AWGS 2014 (which had 4 studies and 826 people) found a prevalence of 16% (95% CI: 7–33%), but with high heterogeneity (I² = 96.5%). On the other hand, AWGS 2019 (with 7 studies and 1,353 people) found a 20% prevalence (95% CI: 13–29%) with less variation (I² = 91.2%). Statistical analysis showed that there were big differences between the criteria versions (χ² = 110.99, p < 0.0001). This shows that the choice of diagnostic frameworks and version updates, especially the move from EWGSOP1 to EWGSOP2 with its stricter criteria, has a big impact on how sarcopenia is found and how common it is in people with HIV. Subgroup analysis by WHO region Supplementary Fig. 4 groups studies by WHO region; each regional subgroup displays an independent pooled estimate alongside the overall pooled effect. According to the WHO region subgroup research, sarcopenia prevalence varied considerably by geographic region. The overall frequency was highest in South-East Asia, at 27% (95% CI: 20–35%, n = 5 studies, 795 participants); nonetheless, with notable heterogeneity (I² = 85.7%). In the European Region, the prevalence was 20% (95% CI: 10–35%, n = 8 studies, 2,030 persons), with considerable variability among studies (I² = 92.4%).The Western Pacific Region had a prevalence of 14% (95% CI: 9–22%, n = 7 studies, 1,610 individuals, I² = 93.9%), while the Region of the Americas had a lower prevalence of 11% (95% CI: 6–20%, n = 13 studies, 1,976 participants) with high heterogeneityThe results were somewhat different from each other (I² = 71.8%)(I² = 93.9%). The African region was represented by a single study from South Africa (n = 805), with a prevalence of 12% (95% CI: 9–14%). Statistical testing revealed significant differences among WHO regions (χ² = 22.68, p < 0.0001), suggesting that regional factors, including healthcare infrastructure, HIV management practices, population demographics, nutritional status, and socioeconomic conditions, substantially influence the burden of sarcopenia in individuals living with HIV globally. Subgroup analysis by world bank classification Supplementary Fig. 5 organises studies by World Bank income group — lower-middle, upper-middle, and high-income — with separate pooled estimates and heterogeneity statistics per group. The World Bank’s subgroup analysis on income classification demonstrated a clear inverse correlation between income level and the prevalence of sarcopenia among individuals with HIV. The highest combined prevalence rate of 31% (95% CI: 22–43%) was identified in lower-middle-income countries (n = 3 studies, 504 participants), exhibiting relatively low heterogeneity (I² = 87.6%). This demonstrates that locations with limited resources face significant challenges. The prevalence was 13% (95% CI: 9–18%) in upper-middle-income countries (21 studies, 4,416 participants); however, significant heterogeneity was observed (I² = 91.4%), indicating considerable variability among populations and healthcare systems within this group. The lowest combined rate, at 20% (95% CI: 11–33%), was observed in high-income countries (n = 10 studies, 2,296 participants), although significant heterogeneity was present among the studies (I² = 92.8%). Statistical analysis (χ² = 11.88, p = 0.0026) revealed significant differences among groups with different income levels. Subgroup analysis by muscle mass assessment Supplementary Fig. 6 compares DXA versus non-DXA assessment subgroups side by side, with statistical testing for subgroup differences displayed at the foot of the figure. The subgroup analysis of various muscle mass assessment methods indicated disparities in the frequency of sarcopenia contingent upon the diagnostic methodology utilized. Research employing Dual-energy X-ray Absorptiometry (DXA), regarded as the gold standard for body composition evaluation, comprised 16 studies with 4,245 participants, yielding a pooled prevalence of 17% (95% CI: 10–27%) and exhibiting significant heterogeneity (I² = 95.1%). Conversely, research employing non-DXA methodologies, including bioelectrical impedance analysis, anthropometry, and computed tomography, comprised 18 studies with 2,971 participants, revealing a comparable pooled prevalence of 14% (95% CI: 10–20%) with considerable variability (I² = 90.4%). Statistical testing revealed no significant difference between the two assessment methods (χ² = 0.34, p = 0.5586), indicating that although DXA offers more accurate readings, alternative non-DXA approaches can produce similar prevalence estimates when appropriately calibrated. Subgroup analysis by national guideline adoption Supplementary Fig. 7 separates studies from countries with and without official sarcopenia policies; the subgroup difference test is shown at the bottom of the plot. The subgroup analysis based on national sarcopenia policy adoption showed that prevalence estimates were very different depending on whether or not there were official sarcopenia guidelines. In countries lacking explicit sarcopenia policies (n = 23 studies, 4,744 participants), the aggregated prevalence was 17% (95% CI: 11–24%), exhibiting significant heterogeneity (I² = 92.9%). Conversely, nations with formal sarcopenia policies (n = 11 studies, 2,472 participants) exhibited a reduced pooled prevalence of 14% (95% CI: 8–23%), accompanied by significant heterogeneity (I² = 95.1%). Nonetheless, statistical analysis revealed no significant difference between the two groups (χ² = 0.28, p = 0.5965), indicating that the existence of official programs does not inherently correlate with a reduced prevalence of sarcopenia. Subgroup analysis by multiple diagnostic methods Supplementary Fig. 8 compares single-method versus multi-method diagnostic approaches; the pooled estimates and subgroup difference test are displayed at the bottom of each group. The subgroup analysis by the number of diagnostic tools employed revealed significant differences in sarcopenia prevalence estimates based on the comprehensiveness of the diagnostic approach. Studies using single-method assessment (n = 8 studies, 2,194 participants) reported a substantially higher pooled prevalence of 26% (95% CI: 17–36%), with very high heterogeneity (I² = 93%). In contrast, studies employing multiple diagnostic tools (> 2 methods) for comprehensive sarcopenia assessment (n = 26 studies, 5,022 participants) demonstrated a lower pooled prevalence of 13% (95% CI: 9–19%) with similarly high heterogeneity (I² = 93%). Statistical testing confirmed a significant difference between these diagnostic approaches (χ² = 6.13, p = 0.0133), indicating that the number of assessment methods substantially influences case identification. Subgroup analysis by COVID period The COVID-19 period subgroup analysis revealed temporal variation in the frequency of sarcopenia among individuals with HIV. Prior research conducted before the COVID-19 pandemic (n = 12 investigations, 2,557 individuals) identified the highest aggregated prevalence at 20% (95% CI: 14–28%), marked by significant variability (I² = 91.9%). During the COVID-19 pandemic (n = 11 trials, 2,093 persons), the overall prevalence fell to 15% (95% CI: 8–26%), indicating substantial variability (I² = 91.7%). Post-pandemic studies showed a further reduction to 13% (95% CI: 7–23%), and the degree of variation was very high (I² = 95.2%). However, statistical analysis revealed no significant change between both time periods (χ² = 1.67, p = 0.4334). This temporal trend likely reflects methodological and diagnostic shifts across study periods rather than a true pandemic effect on sarcopenia burden, changes in how diagnoses are made, or changes in the populations studied, not the true effects of the epidemic (Supplementary Fig. 9). Subgroup analysis by ART status The subgroup analysis based on ART status showed that the prevalence of sarcopenia varied depending on treatment exposure. Research including ART-experienced individuals (n = 27 trials, 5,016 participants) revealed a combined prevalence of 17% (95% CI: 12–23%), characterized by significant heterogeneity (I² = 93.5%). Studies that did not disclose ART status (n = 4 studies, 1,720 participants) had a reduced pooled prevalence of 14% (95% CI: 4–37%) with significant heterogeneity (I² = 95.6%). Research involving both ART-experienced and ART-naïve individuals (n = 3 studies, 450 participants) demonstrated the lowest pooled prevalence of 8% (95% CI: 1–34%), characterized by considerable heterogeneity (I² = 91%). Nonetheless, statistical analysis revealed no significant difference between ART status groups (χ² = 0.94, p = 0.6261), indicating that ART exposure alone does not significantly affect sarcopenia prevalence in this meta-analysis (Supplementary Fig. 10). Subgroup analysis by ART duration The subgroup analysis by ART duration showed that the prevalence of sarcopenia varied significantly depending on how long the treatment lasted. Research comprising adults with extended ART exposure (> 5 years, n = 16 trials, 2,878 participants) revealed a combined prevalence of 19% (95% CI: 14–25%), exhibiting significant heterogeneity (I² = 90%). Participants with shorter ART duration (≤ 5 years, n = 3 trials, 453 participants) had a higher pooled prevalence of 21% (95% CI: 7–50%), indicating significant heterogeneity (I² = 95.4%). trials that did not disclose ART duration (n = 15 trials, 3,875 participants) showed the lowest pooled prevalence of 12% (95% CI: 6–21%) with very significant heterogeneity (I² = 95.2%). Nonetheless, statistical analysis revealed no significant difference across ART duration groups (χ² = 2.28, p = 0.3201), indicating that treatment length alone does not substantially influence sarcopenia prevalence (Supplementary Fig. 11). Subgroup analysis by study design The subgroup analysis by study design showed that cross-sectional studies (n = 33 studies, 7,063 participants) reported a pooled sarcopenia prevalence of 16% (95% CI: 12–21%) with very high heterogeneity (I² = 93.7%), while the single cohort study (n = 1 study, 153 participants) demonstrated a prevalence of 10% (95% CI: 6–16%) (Supplementary Fig. 12). Subgroup analysis by sampling method The subgroup analysis based on sampling method indicated that the prevalence of sarcopenia differed according to the selection criteria of individuals. Consecutive sampling studies (n = 11 trials, 1,436 participants) revealed a combined prevalence of 22% (95% CI: 15–31%), exhibiting considerable heterogeneity (I² = 87.3%). Conversely, convenience sampling studies (n = 22 studies, 4,915 people) indicated a lower prevalence of 13% (95% CI: 9–20%) with significant heterogeneity (I² = 94.1%). The only study that used random sampling (n = 1 study, 805 participants) found a prevalence of 12% (95% CI: 9–14%) (Supplementary Fig. 13). Subgroup analysis by HIV duration The subgroup analysis, predicated on the length of HIV infection, revealed an unforeseen trend in the prevalence of sarcopenia. Participants with a shorter HIV duration (≤ 10 years, n = 10 trials, 1,944 participants) had a higher pooled prevalence of 21% (95% CI: 14–31%), accompanied by significant heterogeneity (I² = 91.6%). Conversely, individuals with a longer duration (> 10 years, n = 12 studies, 2,067 participants) had a lower incidence of 15% (95% CI: 10–23%) accompanied by substantial heterogeneity (I² = 91.5%). Research that did not report HIV duration (n = 12 studies, 3,205 participants) exhibited the lowest prevalence of 12% (95% CI: 6–22%), accompanied by substantial heterogeneity (I² = 95.7%). Statistical testing indicated no significant difference between groups (χ² = 2.73, p = 0.2555); however, the reduced prevalence in individuals with prolonged HIV may result from enhanced viral suppression and immunological reconstitution facilitated by long-term antiretroviral therapy. Conversely, those with shorter HIV duration may still be experiencing suboptimal viral control, persistent inflammation, and immune activation associated with muscle loss. These observations suggest that long-term virologic suppression may be associated with a lower sarcopenia burden, though causal inference is not possible from these cross-sectional data (Supplementary Fig. 14). Subgroup analysis by BODY MASS INDEX status The subgroup study predicated on BODY MASS INDEX status revealed a significant discrepancy in the prevalence patterns of sarcopenia. Individuals with a normal BODY MASS INDEX (n = 12 studies, 2,456 participants) demonstrated the greatest pooled prevalence at 25% (95% CI: 19–32%), marked by substantial heterogeneity (I² = 85.2%). Conversely, individuals categorized as overweight (n = 11 studies, 2,895 participants) exhibited a markedly lower prevalence of 8% (95% CI: 4–17%), characterized by extremely high heterogeneity (I² = 96.4%). Mixed BODY MASS INDEX populations (n = 4 studies, 678 participants) exhibited an intermediate prevalence of 14% (95% CI: 8–23%, I² = 84.1%), whereas studies that did not record BODY MASS INDEX (n = 7 studies, 1,187 participants) had a prevalence of 17% (95% CI: 11–26%, I² = 90.7%). Statistical testing demonstrated highly significant differences among BODY MASS INDEX categories (χ² = 10.50, p = 0.0148), uncovering the unexpected result that normal-weight individuals with HIV are at a considerably greater risk of sarcopenia than their overweight peers, possibly illustrating the “obesity paradox,” wherein excess adiposity may confer metabolic advantages (Supplementary Fig. 15). Meta-regression Meta-regression analysis revealed three transformative moderators that substantially elucidated the high heterogeneity found across studies. Body mass index status emerged as the strongest predictor (Adjusted R² = 16.2%). Both overweight and normal body mass index categories were associated with significantly lower sarcopenia prevalence relative to the reference category, but the magnitude of reduction was substantially greater for overweight individuals (Beta = -2.337, p < 0.001) than for those with normal BMI (Beta = -1.123, p < 0.001). This indicates that overweight PLHIV had considerably lower sarcopenia prevalence than their normal-weight counterparts — a finding consistent with, and corroborating, the subgroup analysis (overweight: 8%, 95% CI: 4–17%; normal body mass index: 25%, 95% CI: 19–32%). Both analyses therefore point in the same direction: normal body mass index is associated with higher observed sarcopenia prevalence than overweight status in PLHIV, a pattern consistent with the obesity paradox observed in chronic disease populations. The World Bank income classification was just as useful (Adjusted R² = 10.1%). It showed that upper-middle-income countries had a much higher burden (Beta = -1.942, p < 0.001) than high-income countries (Beta = -1.396, p < 0.001). However, lower-middle-income countries did not show any significant associations, which goes against what is usually thought about resource-limited settings. Diagnostic comprehensiveness was also very important (Adjusted R² = 9.9%), with multi-method assessments giving very different estimates (Beta = -1.872, p < 0.001) than single-method assessments (Beta = -1.066, p < 0.05). This shows how important diagnostic rigour is for finding prevalence. Interestingly, parameters that were thought to be key contributors, such as ART duration, ART status, COVID period, sampling method, and study design, did not account for heterogeneity (all Adjusted R² = 0.0–1.5%), indicating that these variables had little independent effect on variation between studies (Supplementary Table 4). Sensitivity analysis: leave-one-out method The leave-one-out sensitivity analysis showed that the pooled estimate of sarcopenia prevalence was very stable. When each study was systematically excluded and the meta-analysis recalibrated, the aggregated prevalence consistently remained closely aligned with the overall estimate of 16% (as denoted by the red horizontal line), with all individual omission analyses producing estimates within a confined range of approximately 15–17%. The overall guess was very close to each of the individual data points. This proves that there wasn’t one study that changed the overall results a lot. The meta-analytic results are correct even though there are differences in sample size, location, or methods because remarkable consistency was observed across all 34 studies. Even when studies were taken away one after the other, the results stayed the same. This showed that the 16% sarcopenia prevalence estimate among PLHIV is robust across the represented study populations, though generalisability to underrepresented regions — particularly sub-Saharan Africa and South Asia — remains limited G(Supplementary Fig. 16). Galbraith plot analysis The Galbraith plot showed that there was a lot of variation between trials, with a few significant outliers going outside the 95% confidence boundaries (red dashed lines). Patricia Echeverra et al. 2018 was an extreme outlier with a very high standardized effect size (Z-score ~ 31), which means it was the biggest difference from the pooled estimate. Peng T et al. 2024, S. De Vincentis et al. 2024, Deep Dutta et al. 2017, Celia Gregson et al. 2021, and Hawkins KL et al. 2018 were all outliers. They were all much beyond the confidence ranges. The picture shows how effect size (on the y-axis) and accuracy (on the x-axis) are related. The studies are more likely to be near the regression line if the bubbles are bigger. Studies with less accuracy (greater standard errors) showed a wider range of prevalence estimates. The colour gradient by year of publication illustrates that the status of outliers doesn’t alter in a clear way over time. This means that the differences aren’t because things have changed throughout time, but because people are grouped and treated differently. These outliers likely indicate genuine disparities within the community, variations in diagnostic procedures, or distinctive research attributes, rather than deficiencies in the employed methodologies. This shows how complicated it is to study sarcopenia in HIV communities around the world (Fig. 4). Galbraith plot Sensitivity analysis excluding extreme outliers After removing two extreme outliers—Patricia Echeverra et al. 2018 (Z-score = 31.21, > 99th percentile) and Peng T et al. 2024 (Z-score = 22.31, > 95th percentile)—the meta-analysis of the remaining 32 studies (n = 5,979 participants) showed that the pooled prevalence estimate was very stable. The random effects model yielded a prevalence of 17% (95% CI: 12–21%), which closely aligns with the prior analysis encompassing all studies (16%, 95% CI: 12–21%). Heterogeneity remained significantly elevated (I² = 95.5%, τ² = 0.0269, p < 0.0001), indicating that the removal of these outliers did not substantially reduce between-study variability. The pooled estimate’s robustness, despite the exclusion of the most extreme studies, confirms that the overall prevalence of sarcopenia is not significantly influenced by outlier data (Supplementary Fig. 17). Risk of bias Publication bias assessment Statistical analysis verified the lack of publication bias within the meta-analysis. Egger’s regression test indicated the absence of significant funnel plot asymmetry (p = 0.399, t = 0.8548), while Begg’s rank correlation test demonstrated no evidence of selective reporting (p = 0.2353, tau = 1.1867). Both studies demonstrated that research with different prevalence estimates possesses an equivalent likelihood of publication. This reassured us that the 16% aggregated prevalence estimate accurately reflects the global number of HIV-positive individuals with sarcopenia, rather than being affected by selective reporting or the influence of small studies (Table 2). Table 2Egger’s & Beg’s test results for publication biasTestValueEgger’s Test P-value, Test resultP- 0.399, t-value: 0.8548Begg Test p-value0.2353Begg Test statistic (tau)1.1867 Egger’s & Beg’s test results for publication bias The funnel plot using Freeman-Tukey double arcsine transformation demonstrated good symmetry, with studies distributed relatively evenly around the pooled estimate (central vertical line). The majority of studies fell within the 95% confidence limits (indicated by the funnel boundaries), with only a few smaller studies (higher standard errors) positioned outside the funnel. The symmetric distribution and concentration of studies near the apex supports the findings from Egger’s (p = 0.399) and Begg’s (p = 0.2353) tests, indicating minimal publication bias and confirming the reliability of the 16% pooled sarcopenia prevalence estimate (Supplementary Fig. 18). Doi plot and LFK index assessment The Doi plot demonstrated a relatively symmetric distribution of effect sizes around the central vertical line, with studies distributed fairly evenly on both sides. The Luis Furuya-Kanamori (LFK) index yielded a value of 1.803, which falls within the range of ± 1 to ± 2, indicating minor asymmetry. According to LFK index thresholds (|LFK| < 1 = no asymmetry; 1–2 = minor asymmetry; >2 = major asymmetry), this finding suggests minimal publication bias. The result aligns well with conventional Egger’s (p = 0.399) and Begg’s (p = 0.2353) tests, which also indicated no significant publication bias. The slight asymmetry detected may reflect the substantial heterogeneity (I² = 93.6%) arising from diverse populations, diagnostic criteria, and methodological approaches rather than selective publication practices. Overall, these findings strengthen confidence in the validity and representativeness of the 16% pooled sarcopenia prevalence estimate (Supplementary Fig. 19). Quality of evidence: GRADE assessment GRADE criteria rated the evidence regarding the prevalence of sarcopenia in people living with HIV as moderate quality. A transparent account of how each GRADE domain influenced this rating is provided below. Evidence from observational studies begins at low certainty by default. Three potential sources of downgrading were specifically evaluated. Inconsistency: substantial heterogeneity (I² = 93.6%) was present, reflecting genuine variability in sarcopenia prevalence across populations, diagnostic criteria, and geographic settings. This triggered a one-level downgrade for inconsistency. and risk of bias: the predominance of convenience and sequential sampling (rather than random population-based sampling) across included studies was acknowledged. However, this did not prompt an additional downgrade, as convenience sampling is near-universal in HIV clinical cohort research, is unlikely to systematically bias prevalence estimates in a single direction, and was partially mitigated by the inclusion of studies from diverse clinical settings across 15 countries and, study design: the predominantly cross-sectional nature of included studies (33/34) was noted; however, this limitation is already captured within the baseline observational design rating and did not independently trigger a further downgrade. No concerns were identified for indirectness, imprecision (the 95% CI of 12–21% was relatively narrow despite high heterogeneity), or publication bias (Egger’s p = 0.399). Accordingly, beginning at low certainty, applying one downgrade for inconsistency, and finding no additional reasons for further downgrading, the overall certainty was rated as moderate. This moderate-quality evidence supports a plausible estimate that approximately one in six PLHIV globally may have sarcopenia, while acknowledging that standardised diagnostic methodologies and longitudinal study designs in future research could further strengthen this evidence base (Table 3). Table 3Grade assessmentOutcomeNo. of StudiesStudy DesignRisk of BiaInconsistencyIndirectnessImprecisionPublication BiasEffect (95% CI)CertaintySarcopenia in HIV(overall prevalence)34 studies(7216 participants)Observational(cross-sectionaland cohort)Not seriousVery serious(I² = 93.6%)Not seriousNot serious(narrow CI:12–21%)Not serious (Egger’s p = 0.399)~ 16% (CI:12–21%)⬤⬤⬤◯ Moderate Grade assessment Sarcopenia in HIV (overall prevalence) 34 studies (7216 participants) Observational (cross-sectional and cohort) Very serious (I² = 93.6%) Not serious (narrow CI: 12–21%) ~ 16% (CI: 12–21%) Meta-analysis of risk factor Age above 50 years emerged as a potent risk factor, conferring 97% increased odds of sarcopenia (OR: 1.97, 95% CI: 1.30–2.98, p = 0.001) in people living with HIV, establishing a clear imperative for targeted screening and early intervention in middle-aged and older adults with HIV infection (Supplementary Table 5). Discussion Principal findings and comparison with previous evidence This systematic review and meta-analysis aggregated data from 34 studies, including 7,216 persons across 15 countries, indicating a Who regions pooled prevalence of sarcopenia of 16% (95% CI: 12–21%) among adults living with HIV. This conclusion signifies a substantially lower estimate in contrast to the preceding comprehensive analysis by Seyed Alinaghi et al. (2023), which indicated a pooled prevalence of 23% from 26 studies [13]. Various methodological factors may elucidate this discrepancy. Our review utilized more rigorous inclusion criteria, necessitating validated diagnostic frameworks (EWGSOP, AWGS, or Baumgartner criteria) and excluding studies that employed investigator-defined or ambiguous diagnostic methodologies, thereby improving comparability and diminishing heterogeneity arising from non-standardized evaluations. We also included nine more recent publications that came out between 2023 and 2025. These papers illustrate the advancement of diagnostic techniques and demographic characteristics in the current era of ART. Third, our meta-regression analysis revealed that the variant of the diagnostic criteria functioned as a major moderator. The growing adoption of the EWGSOP2 2019 criteria (8% prevalence) instead of the EWGSOP1 2010 criteria (16% prevalence) likely contributed to our lower pooled estimate. Although EWGSOP2 places greater emphasis on muscle strength and introduces a hierarchical diagnostic algorithm, it requires the confirmation of both low muscle strength and low muscle mass for a definitive sarcopenia diagnosis. This more specific case definition may reduce overclassification compared to earlier mass-centred approaches, thereby lowering confirmed prevalence estimates. Thus, the observed decline from EWGSOP1 to EWGSOP2 likely reflects increased diagnostic specificity rather than a true reduction in disease burden [2, 3]. Geographic and regional variations Our subgroup analyses revealed substantial geographic diversity in sarcopenia prevalence, requiring careful interpretation. South-East Asia had the highest overall prevalence among WHO areas, at 27% (95% CI: 20–35%). The Americas finished in third with 11% (95% CI: 6–20%), and Europe came in second with 20% (95% CI: 10–35%). Romania had the highest frequency at the country level (35%, 95% CI: 25–46%), followed by the USA at 37% (95% CI: 31–43%), and Germany had the lowest incidence globally at 2% (95% CI: 1–4%). These substantial differences likely arise from complex interactions among population genetics, dietary practices, healthcare infrastructure, ART regimen formulations, and, importantly, the diagnostic criteria predominantly employed in each region. European studies mostly used EWGSOP criteria in different ways, while Asian studies used AWGS criteria that were made particularly for Asian body types [4, 5]. The unusually low occurrence in affluent countries like Germany may be associated with earlier initiation of antiretroviral therapy, improved dietary support of antiretroviral medication (ART) earlier, improved dietary support, structured sarcopenia screening programs, or even different body composition profiles. Conversely, the elevated rates in Southeast Asia and certain Latin American countries may be associated with delayed HIV diagnosis, a lower CD4 nadir at the initiation of treatment, a higher prevalence of infectious illnesses, and limited access to comprehensive HIV care beyond viral suppression [13]. These findings underscore that sarcopenia in HIV is not merely a biological phenomenon but is profoundly affected by healthcare systems, socioeconomic factors, and regional diagnostic practices. Impact of diagnostic criteria and assessment methods A significant outcome from our meta-regression analysis was the identification of diagnostic criteria and assessment procedures as the primary factors influencing reported sarcopenia prevalence, elucidating considerable between-study variability. Research utilizing Baumgartner criteria indicated the highest prevalence (28%, 95% CI: 24–33%), followed by AWGS criteria (18%, 95% CI: 13–26%), whilst EWGSOP criteria produced the lowest estimate (12%, 95% CI: 7–19%). In a version-specific subgroup analysis, it was even more interesting to see that the EWGSOP2 2019 criteria had a much lower prevalence (8%, 95% CI: 3–19%) than the EWGSOP1 2010 criteria (16%, 95% CI: 10–26%). This is a 50% drop in the number of instances found, which is only due to changes in the diagnostic framework. The EWGSOP2 version set stronger muscle strength limits, made it easier to find likely sarcopenia by looking for weakness alone, and made muscle quantity cut-offs clearer. This made the diagnostic algorithm more specific but possibly less sensitive [3]. In the same way, the AWGS 2019 criteria made muscle mass cut-off points stricter than the AWGS 2014 criteria, which led to differences in Asian populations [5]. Our results are consistent with the existing sarcopenia literature, indicating that the choice of diagnostic criteria significantly affects prevalence estimates across all groups, not just in HIV contexts [17]. The meta-regression indicated that studies utilising multiple diagnostic instruments (integrating muscle mass, strength, and performance evaluations) exhibited a significantly lower prevalence (13%, 95% CI: 9–19%) in contrast to single-method approaches (26%, 95% CI: 17–36%), implying that thorough multi-domain assessment enhances diagnostic specificity. These findings have significant implications: international consensus guidelines for sarcopenia screening in HIV populations must prioritise diagnostic harmonisation to facilitate meaningful cross-study comparisons and effective epidemiological surveillance. It is important to contextualise these meta-regression findings appropriately. The three significant moderators identified — BODY MASS INDEX status (adjusted R² = 16.2%), World Bank income classification (adjusted R² = 10.1%), and diagnostic comprehensiveness (adjusted R² = 9.9%) — together explain only a modest proportion of the total between-study variance, leaving the majority of heterogeneity unexplained. These analyses were conducted as exploratory investigations rather than confirmatory hypothesis-driven tests; the moderators were selected based on clinical and methodological plausibility, and the relatively small number of included studies (n = 34) relative to the number of moderators examined increases the possibility of chance associations. Notably, variables anticipated a priori to be important — including ART duration, ART status, COVID period, sampling method, and study design — explained negligible variance (adjusted R² = 0.0–1.5%), which itself is an informative finding suggesting that structural and contextual factors outweigh treatment-related variables in driving prevalence variability. Readers should therefore interpret these moderator findings as hypothesis-generating signals that identify priority variables for future confirmatory research, rather than as definitive explanations of heterogeneity. Body composition phenotypes and the obesity paradox The most clinically compelling outcome from our meta-regression study was the identification of body mass index status as the most significant independent mediator of sarcopenia prevalence (adjusted R²=16.2%), indicating a paradoxical link. People with a normal body mass index had the highest rate of sarcopenia (25%, 95% CI: 19–32%), which was much higher than the rate in overweight people (8%, 95% CI: 4–17%). This difference of three times was due to the body composition group alone. This unexpected result illustrates the “obesity paradox” or “sarcopenic obesity” phenomenon, which is gaining recognition in chronic illness populations, including HIV [13]. It is important to note that the subgroup analysis and meta-regression findings are fully consistent with one another, despite appearing contradictory at first reading. In the meta-regression, both BMI categories carry negative Beta coefficients relative to the reference, meaning both reflect reductions in prevalence from the intercept; the larger negative coefficient for overweight (Beta = − 2.337) compared to normal BMI (Beta = − 1.123) indicates that overweight status is associated with a greater reduction in sarcopenia prevalence — precisely mirroring the subgroup finding that overweight PLHIV had the lowest observed prevalence (8%) compared to normal-weight PLHIV (25%). The apparent inconsistency arose from imprecise reporting of the meta-regression direction in the original text, which has now been corrected. The apparent inconsistency arose from imprecise reporting of the meta-regression direction in the original text, which has now been corrected. There are a few mechanical explanations that should be considered. First, people who are HIV-positive and have a normal weight may be a phenotypically distinct group that has ongoing catabolic processes, not enough calories, persistent inflammation, or metabolic disturbances that may be associated with lower weight maintenance and higher rates of muscle loss and, overweight and obese people living with HIV may have more metabolic reserve, may be associated with lower observed rates of inflammation-induced muscle proteolysis by making more substrates available and possibly having more protective adipokine profiles and, the protective effect of a greater body mass index may indicate intact anabolic ability or diminished HIV-related metabolic stress in individuals who can maintain their weight despite viral infection and, measurement artefacts cannot be disregarded—DXA-derived appendicular skeletal muscle mass indices may be inappropriately adjusted in overweight persons when reported in relation to height², potentially concealing absolute muscular deficiencies. This contradiction has significant clinical ramifications: Routine body mass index-based risk stratification may consistently underestimate sarcopenia prevalence in normal-weight PLHIV while neglecting sarcopenic obesity in overweight persons. A thorough body composition evaluation that goes beyond basic anthropometrics is essential for precise sarcopenia detection in HIV populations [2, 3]. Socioeconomic determinants and resource settings The World Bank income classification subgroup analysis demonstrated a significant inverse correlation between national income level and sarcopenia prevalence. Lower-middle-income countries displayed the highest burden (31%, 95% CI: 22–43%), whereas upper-middle-income (13%, 95% CI: 9–18%) and high-income nations (20%, 95% CI: 11–33%) showed lower prevalence rates. This gradient challenge simplistic ideas that link limited resources to universally bad health outcomes. Instead, it shows how complicated the links are between healthcare access, HIV management systems, demographic factors, and diagnostic methods [13]. There are a number of different ways to look at this. In lower-middle-income countries, individuals may not undergo HIV testing until late, a pattern associated with more severe immunosuppression and prolonged catabolic states at treatment initiation before the initiation of treatment. Second, not having enough access to complete nutritional support, treatment of opportunistic infections, and multidisciplinary HIV care beyond antiretroviral therapy may keep muscle catabolism going after the virus has gone away. Third, places with low resources have higher rates of tuberculosis, chronic diarrhoea, and not having enough food. These factors may be associated with long-term energy deficits and chronic inflammation, which have been associated with higher sarcopenia prevalence in observational studies. But, remarkably, countries with high incomes had a mid-range prevalence instead of the lowest. This could mean that there are sarcopenic obesity phenotypes in groups with a higher baseline BODY MASS INDEX, different ART regimen formulations, or more sensitive diagnostic methods that can find mild cases of sarcopenia. The relatively lower prevalence in upper-middle-income countries may indicate an epidemiological “sweet spot,” marked by substantially enhanced access to ART while obesity-related metabolic disorders are less prevalent than in higher-income settings. These findings underscore that the prevention of sarcopenia in those living with HIV requires holistic interventions addressing nutritional security, complete HIV management, and socioeconomic determinants, rather than depending exclusively on antiretroviral therapy. Clinical and policy implications A fundamental consideration underlying the variation in reported sarcopenia prevalence is that the three diagnostic frameworks employed across included studies — Baumgartner, EWGSOP, and AWGS — do not define or measure the same clinical construct. The Baumgartner criteria adopt a purely mass-centred approach, identifying sarcopenia solely based on low appendicular skeletal muscle mass indexed to height², without requiring any assessment of muscle strength or physical performance. In contrast, the EWGSOP1 2010 criteria introduced a multi-domain definition requiring low muscle mass plus either low strength or low performance, while the EWGSOP2 2019 revision repositioned low muscle strength as the primary diagnostic criterion, making mass a confirmatory rather than entry-level parameter — a conceptual shift that fundamentally alters who is classified as sarcopenic. The AWGS criteria follow a similarly multi-domain approach but apply cut-off values specifically derived from Asian populations, reflecting ethnospecific body composition norms. This progression from a purely quantitative (mass-only) to a predominantly functional (strength-first) definition across generations of criteria explains, at least in part, why the Baumgartner criteria consistently yield the highest prevalence estimates (28%, 95% CI: 24–33%) while EWGSOP2 2019 produces the lowest (8%, 95% CI: 3–19%) — not because these studies examined fundamentally different populations, but because they were measuring different dimensions of the same syndrome. The overall results of this meta-analysis have significant consequences for HIV clinical practice, public health policy, and research priorities. First, sarcopenia screening should be consistently incorporated into standard HIV care protocols, especially for those over 50 years of age, those with a normal BODY MASS INDEX (which was paradoxically associated with higher observed prevalence in this review), those with extended HIV duration, and those living in resource-limited environments. The current HIV treatment guidelines from WHO and major national programs mainly focus on viral suppression, CD4 monitoring, and screening for cardiovascular disease, diabetes, and cancers that are more likely to occur with HIV. They don’t often talk about sarcopenia in a systematic way [7]. Our data endorse the integration of straightforward sarcopenia screening instruments—such as SARC-F questionnaires, handgrip strength evaluations, or chair stand tests—into annual comprehensive HIV exams, with affirmative results necessitating referral for body composition analysis and multidisciplinary management. It is important to distinguish between two complementary but conceptually distinct goals. First, diagnostic standardisation — the adoption of a consistent validated framework such as EWGSOP2 2019 across research and clinical settings — is necessary to enable meaningful cross-study comparison and population-level surveillance. This is a methodological requirement and does not imply clinical uniformity. Second, and equally important, clinical management of sarcopenia in PLHIV must be fundamentally personalised. The heterogeneity demonstrated throughout this review — in BMI-related prevalence patterns, age-related risk gradients, geographic variation, and socioeconomic determinants — underscores that no single intervention pathway will be appropriate for all individuals. Risk stratification should incorporate age, body composition phenotype, ART regimen and duration, immunological status, nutritional state, comorbidity burden, and functional capacity. Interventions should be tailored accordingly, drawing on multimodal strategies including individualised exercise prescription, personalised nutritional support, and where appropriate, pharmacological management. This aligns with the broader global movement toward personalised medicine in chronic disease management, which recognises that population-level evidence must be translated into individual-level care plans rather than uniform protocols. Second, diagnostic harmonisation is an urgent priority. The significant impact of diagnostic criteria on prevalence estimates (12% with EWGSOP compared to 28% with Baumgartner) makes cross-study comparisons and epidemiological surveillance almost futile without consistent methodologies. International HIV and geriatric medicine associations should convene to define consensus diagnostic criteria specifically tailored for PLHIV, considering HIV-specific pathogenesis, body composition phenotypes, and resource availability across various settings. Until such a consensus emerges, we recommend EWGSOP2 2019 criteria as the preferred framework given their comprehensive multi-domain approach, evidence-based thresholds, and global validation, while recognising that region-specific criteria (AWGS for Asia) may be more appropriate in certain contexts [2–5]. Health systems must allocate resources for sarcopenia prevention and management within the HIV care infrastructure. This includes teaching healthcare workers how to spot sarcopenia, making sure that diagnostic tools like handheld dynamometers, bioimpedance analysers, or DXA are available, setting up ways for patients to get to physiotherapy and nutrition services, and adding sarcopenia care to patient-centred medical homes for HIV. The World Health Organisation’s Decade of Healthy Ageing (2021–2030) and UN Sustainable Development Goal 3 put a lot of emphasis on keeping older people functional [7, 14]—sarcopenia in PLHIV is a clear example of these priorities in action that needs special attention and resources. Limitations There are some limitations with this systematic review that need to be pointed out. First, the studies included had a lot of differences (I²=93.6%), and meta-regression only partly explains this. Based on the totality of our subgroup analyses and meta-regression findings, the most influential drivers of between-study variability appear to be: diagnostic criteria, which alone produced a nearly threefold difference in prevalence estimates (28% for Baumgartner versus 8% for EWGSOP2 2019); BODY MASS INDEX status and body composition phenotype, which meta-regression identified as the single most significant independent moderator (adjusted R²=16.2%), reflecting the heterogeneous metabolic profiles of PLHIV across settings; and geographic and socioeconomic context, which introduces compounded variability through differences in HIV care infrastructure, ART initiation timing, nutritional status, and population age structure. Age composition of study populations, particularly the proportion of participants older than 50 years, further contributed to variability, as older PLHIV demonstrated nearly twofold higher odds of sarcopenia (OR: 1.97). Residual unmeasured confounders — such as specific ART regimens, HIV RNA suppression rates, CD4 nadir, comorbidity burdens, and lifestyle factors — likely contribute to the remaining unexplained variability. Importantly, readers should exercise caution in interpreting the pooled prevalence of 16% (95% CI: 12–21%) as a single representative “global” value. Given the substantial residual heterogeneity that persists even within individual subgroups, this pooled estimate is best understood as a weighted central tendency across widely varying clinical and epidemiological contexts, rather than a universal prevalence applicable to all PLHIV populations. Subgroup-specific estimates may therefore be more clinically meaningful for local programme planning than the overall pooled value. Second, the predominance of cross-sectional research (33/34) hinders causal inference about risk variables and limits the understanding of sarcopenia trajectories across time. Longitudinal cohort studies that track changes in muscle mass along the HIV care continuum are essential. A critical limitation affecting the generalisability of these findings is the pronounced geographic imbalance among included studies. The Americas contributed 13 studies, Europe 8, and the Western Pacific 7, while the African region — which bears the greatest global HIV burden, with over two-thirds of all PLHIV residing in sub-Saharan Africa — was represented by only a single study. South Asia and Central Asia were similarly absent. This geographic skew means that the pooled prevalence estimate of 16% predominantly reflects the experience of PLHIV in high- and upper-middle-income countries with established HIV care infrastructure, and should not be interpreted as a truly global estimate. We therefore recommend that the findings be contextualised within the WHO regions represented, and that readers exercise caution when extrapolating to low-income settings, particularly sub-Saharan Africa, where HIV burden, nutritional status, ART access, and body composition profiles differ substantially from the populations captured in this review. Future research must prioritise population-based prevalence studies in underrepresented regions to enable genuinely representative global estimates. A majority of the studies employed convenience or sequential sampling rather than random sampling from the general population. This may indicate that the clinical populations with an elevated risk of sarcopenia were disproportionately represented compared to the community-dwelling individuals living with HIV. The diversity of diagnostic criteria remains an insurmountable limitation due to the lack of HIV-specific sarcopenia definitions, despite some alleviation through subgroup analysis. The inclusion of studies enrolling adults from 18 years of age onwards, rather than restricting to older age groups, introduces heterogeneity in age composition across included studies and limits the direct applicability of findings to older PLHIV specifically. This was a necessary trade-off given that HIV-associated sarcopenia manifests across the full adult age spectrum and that restricting inclusion to older participants would have excluded the majority of available evidence. Readers should therefore interpret the overall pooled estimate with awareness that it reflects a broad adult HIV population, and consult age-stratified subgroup estimates when applying findings to older PLHIV specifically. Directions for future research Future research should prioritise longitudinal cohort studies to clarify the incidence and progression of sarcopenia across the HIV care continuum, alongside mechanistic investigations exploring HIV-specific pathways such as chronic inflammation, mitochondrial dysfunction, and antiretroviral therapy–related metabolic effects [10–12]. Randomised controlled trials evaluating multimodal interventions—including exercise, nutritional strategies, and pharmacotherapy—are needed, particularly across different age groups and resource settings [15]. Additionally, validation of simplified screening tools for resource-limited environments, implementation science to integrate sarcopenia assessment into routine HIV care, and cost-effectiveness and patient-centred research will be essential to translate evidence into sustainable clinical and policy practice. Conclusion Sarcopenia affects approximately one in six people living with HIV across the represented WHO regions (16%, 95% CI: 12–21%), with substantial variation across diagnostic criteria, geographic regions, and socioeconomic contexts. Given the marked underrepresentation of sub-Saharan Africa and other low-income settings in the available evidence base. Normal-weight individuals and those over 50 years were associated with the highest observed prevalence. These findings demand systematic integration of muscle health assessment into routine HIV care, diagnostic harmonisation, and evidence-based interventions. As the global HIV population ages, sarcopenia prevention represents a critical priority for achieving healthy ageing and optimal quality of life. Electronic Supplementary Material Below is the link to the electronic supplementary material. Supplementary Material 1 Acknowledgements We thank Dr Harish, Resident doctor, AIIMS Delhi, for his help in the search strategy. Author contributions V.S. was responsible for the conception and design of the study; V.S., M.M. & D.S., and Z.S. performed data acquisition; V.S. performed data analyses; V.S. interpreted the results; V.S. authored the work, and all authors gave it significant revisions before approving the finished draft. Funding The authors received no specific funding for this work. Data availability The datasets generated and/or analyzed during the current study are not publicly available to protect the privacy of the study. Still, they are available from the corresponding author upon reasonable request. Declarations Ethical approval Not applicable. Consent to participate Not applicable. Consent for publication Not applicable. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Supplementary Materials Supplementary Material 1 Data Availability Statement The datasets generated and/or analyzed during the current study are not publicly available to protect the privacy of the study. Still, they are available from the corresponding author upon reasonable request.