Association between HIV infection and type 2 diabetes mellitus: global incidence, prevalence, and risk factors – a systematic review and meta-analysis

Abstract

Introduction

People living with HIV (PLHIV) are at an increased risk of type 2 diabetes mellitus (T2DM) due to aging, lifestyle, and antiretroviral therapy-related factors. This systematic review and meta-analysis estimated the pooled incidence, prevalence, and risk factors for T2DM among PLHIV globally, focusing on older populations.

Methods

This systematic review and meta-analysis was conducted following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines. We conducted a comprehensive literature search in PubMed, Embase, Web of Science, and Scopus for English-language articles published by December 25, 2024. The I² index and Cochrane’s Q statistic were also used to assess study heterogeneity. A random effects model was used to compute the pooled measure of association with 95% confidence intervals. Moreover, the Egger test was used to assess publication bias.

Result

After screening, 112 studies were selected for analysis. These included 491,367 PLHIV, of whom 39,021 had diabetes. Among them, 5,363 PLHIV were over 50 years old with diabetes. One hundred and twelve of these studies have assessed the prevalence of diabetes among people with HIV, while ten studies have reported both the incidence and prevalence of diabetes in this group. This analysis demonstrated an association between HIV and increased odds of T2DM (OR: 1.61, 95% CI: 1.09–2.38). Additionally, there was a significant association between T2DM and being over 50 years (OR: 3.35, 95% CI: 2.52–4.44), having a family history of DM (OR: 2.30, 95% CI: 1.58–3.73), a body mass index (BMI) greater than 25 (OR:2.43, 95% CI: 1.58–3.73), and having hypertension (OR: 1.10, 95% CI: 1.39–3.19) among PLHIV. Furthermore, the prevalence of T2DM among PLHIV was found to be 9% (95% CI: 8%–10%), and it was 11% (95% CI: 9%-13%)among PLHIV aged 50 and older. Additionally, a higher prevalence of T2DM was observed in PLHIV who use NRTIs14% (95% CI: 2%-28%), those who have been on antiretroviral therapy for more than 10 years 15% (95% CI: 6%-26%), and those diagnosed with advanced HIV disease 8% (95% CI: 4%-14%).

Conclusion

This analysis highlights a significant association between HIV and increased T2DM risk, with age, family history of DM, high BMI, and hypertension as key factors. Comprehensive diabetes screening and preventative interventions, including lifestyle modifications such as personalized nutritional guidance, promotion of regular physical exercise, and consistent glucose monitoring, are crucial for improving outcomes in PLHIV.

Clinical trial

Not applicable.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12879-025-12222-5.

Introduction

Recent advancements in antiretroviral therapy (ART) have resulted in considerable decreases in HIV mortality and morbidity due to opportunistic infections and advanced HIV disease (AHD)-related conditions among people living with HIV (PLHIV) [1]. These advancements have increased the life expectancy of those living with HIV, and the global HIV population is aging and growing [1, 2]. ART has resulted in a considerable number of people living with HIV after the age of 50, particularly in high-income countries [3]. However, mortality rates among PLHIV are typically greater than in the general population [1]. PLHIV also still suffer from an increased burden of comorbidity [4, 5], which has an impact on their quality of life, presenting a serious health equity concern [6].

Individuals with HIV experience increasingly complex aging processes, which are greatly influenced by the long-term effects of the virus and ART [7]. Studies have shown that those living with HIV and receiving ART have a higher risk of developing non-communicable diseases (NCDs) such as cardiovascular disease, diabetes mellitus, and non-AHD-related cancers than people without HIV [4, 8, 9]. ART, while life-saving, can contribute to metabolic disorders, including insulin resistance and dyslipidemia [10]. Furthermore, ART regimens, particularly protease inhibitors, are associated with adverse changes in fat deposition, glucose metabolism, and lipid metabolic dysregulation [11], thereby accelerating the development of diabetes.

The increasing lifespan of PLHIV necessitates proactive strategies to mitigate the growing burden of age-related NCDs, particularly diabetes. While the relationship among HIV, ART, aging, and diabetes development is complex, understanding these interactions is essential for improving long-term health outcomes. Furthermore, diabetes mellitus has emerged as an increasing health issue in countries with high HIV prevalence, such as Sub-Saharan Africa. According to studies, the comorbidities of diabetes and HIV complicate clinical management because of overlapping metabolic diseases and the impact of antiretroviral medication on glucose regulation [12–14]. Therefore, identifying modifiable risk factors and developing targeted methods for preventing and managing these age-related complications is crucial. Given the various reports on the prevalence of diabetes in PLHIV, this systematic review and meta-analysis aimed to estimate the pooled incidence, prevalence, and factors associated with type 2 diabetes mellitus (T2DM) among PLHIV globally.

Methods

Study design

This systematic review and meta-analysis evaluated the effect of HIV on the incidence and prevalence of T2MD among PLHIV, with a focus on PLHIV over 50 years of age. The study’s findings were reported according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (Fig. 1).

Fig. 1.

A flow chart depicting the stages of retrieving articles and checking eligibility criteria for meta-analysis to assess the impact of HIV on incidence and prevalence of diabetes

Search strategy

An electronic search strategy was developed to identify studies that examined HIV and diabetes. A comprehensive literature search was conducted across the following databases: PubMed, Embase, Web of Science, and Scopus for English-language publications. This search used a combination of keywords and MeSH terms related to HIV and diabetes, including “HIV,” “Human Immunodeficiency Virus,” “diabetes mellitus,” “type 2 diabetes,” and their variations. No date restrictions were applied, and searches were conducted from database inception until December 25, 2024. Furthermore, the reference lists of included studies were manually examined to identify additional relevant publications. We also conducted a grey literature search to locate unpublished papers, conference abstracts, dissertations, and related reports that might not be available in standard bibliographic sources databases. We prioritized the first 300 items from Google Scholar that were directly relevant to our research question [15]. Additionally, we reached out to the corresponding authors of relevant abstracts and unpublished studies to request full-text articles or more data. This step aimed to reduce publication bias and offer a thorough overview of the available evidence. The grey literature identified was then reviewed to ensure it met our inclusion and exclusion criteria. Afterward, any qualifying grey literature with usable data for our review was chosen for final analysis. The detailed search strategy for the PubMed database is provided in Additional file 1.

Inclusion criteria

Quantitative studies

Studies that evaluated the effect of HIV on diabetes

English-language publications

Peer-reviewed journal articles

No restrictions on sample size or study setting

Exclusion criteria

Studies without specific data on measuring the effect of HIV on diabetes

Non-original studies (reviews, editorials, commentaries)

Non-peer-reviewed articles

Qualitative studies (The primary aim of this study was to estimate the incidence, prevalence, and risk factors for diabetes mellitus among people living with HIV. To do this, our methodology was designed to focus only on synthesizing quantitative data that could be statistically merged and analyzed.)

Study selection

After removing duplicate articles, the titles and abstracts were screened based on the predetermined inclusion and exclusion criteria. Two independent reviewers (HM and MN) conducted a thorough screening of the titles and abstracts of all identified studies. Subsequently, full-text articles were reviewed for eligibility. Studies lacking adequate data regarding the outcomes of interest were excluded. Any disagreements between the reviewers during both the screening and full-text review phases were resolved through discussion and consultation with a third author (ZA). The inter-rater agreement between the two reviewers was evaluated using Kappa statistics, obtaining a Kappa value of 0.81, indicating substantial agreement.

Data extraction

Two independent reviewers extracted data from all included studies using a standardized form. Data were gathered using a data extraction form that included the following information: first author’s name, publication year, country, sample size, study design, age mean/median, sex distribution, HIV status (HIV or AHD), ART duration, type of ART regimen, outcome (diabetes incidence and prevalence), and key findings. The data extraction form was piloted on a sample of studies and refined before use. Any discrepancies in data extraction were resolved through discussion between reviewers or consultation with a third reviewer when necessary.

Quality assessment

The quality of the included studies (cross-sectional, case-control, and cohort studies) was assessed using the Newcastle-Ottawa Scale. This scale is designed to assess potential biases in various aspects of a study, including study design, conduct, and data analysis [16]. The total score ranges from 0 to 9, where scores of 0–4 were considered poor quality, and scores of 5–9 were considered good quality.

Ethical considerations

The protocol and procedures of the present study were reviewed and approved by the Research Ethics Committee of Tehran University of Medical Sciences (Ethics approval number: IR.TUMS.IKHC.REC.1404.012).

Statistical analysis

To evaluate the heterogeneity of study outcomes, I² statistics were used to categorize heterogeneity into three levels: low heterogeneity (≤ 25%), moderate heterogeneity (between 25% and 75%), and high heterogeneity (≥ 75%) [17]. Subgroup analyses were performed based on variables such as sex, HIV status (HIV or AHD), ART use (Yes or No), ART duration (< 10 or >= 10 years), and ART regimens (Pls or Nucleoside analog reverse-transcriptase inhibitors (NRTIs) or non-nucleoside reverse-transcriptase inhibitors (NNRTIs)), type of study, continent, quality score, body mass index (BMI), hypertension, and family history of diabetes. We didn’t apply a consistent, pre-established case definition across all studies for diabetes and HIV, but rather adhered to the criteria given by the original researchers. Furthermore, we used the reported prevalence and incidence from each study, as defined and computed by their respective methodology. To study the heterogeneity of sources, meta-regression was conducted based on sex, HIV status (HIV or AHD), ART duration, and ART regimens, type of study, continent, quality score, BMI, hypertension, and family history of diabetes. The Egger test was used to examine publication bias [18, 19], with p < 0.05 indicating significant bias. A random effects model was used to compute the pooled measure of association (risk ratio or odds ratio). We estimated effect sizes mainly using Odds Ratios (ORs) for comparisons when available. Additionally, we extracted and meta-analyzed Incidence Rate Ratios (IRRs) when studies included person-time data. However, due to variability in reporting across studies, we prioritized ORs for comparing incidence and specific risk factors. To provide a robust estimation of the association between HIV and diabetes, this study extracted adjusted IRRs (per 1000 person-years) and adjusted odds ratios, as reported in the included studies. These adjusted measures were used to control for potential confounding factors. Given the current dataset and the number of available studies, performing a stratified meta-analysis based on homogenous adjustment variables is not feasible due to an insufficient number of studies providing comparable adjustment sets, which limits the number of comparable studies. However, we have meticulously detailed the specific confounders adjusted for in each study within the Appendix table. Data analysis was conducted using Stata version 14 (Stata Corp, College Station, TX, USA). Additionally, a sensitivity analysis was conducted to determine the robustness of the findings, wherein one study was excluded at each stage, and the results were compared to those of the overall analysis.

Result

In the primary search across different databases, 2,392 studies were identified. After title, abstract, and full-text screening, 112 studies were included for analysis. The included studies comprised 491,367 PLHIV, 39,021 of whom had diabetes. Among them, 5,363 PLHIV were over 50 years old with diabetes (Table 1). Due to the large number of studies included, Table 1 presents only those focusing on diabetes prevalence in PLHIV aged over 50 years [20–44]. Within the cohort aged 50 and above, the same distribution of study quality was observed, with half classified as high-quality and the other half as low-quality. Complete data from all studies are provided in the supplementary materials (Supplementary file).

Table 1.

Characteristics of included studies to estimate the incidence and prevalence of diabetes among older adults living with HIV

Author	Country	Type of study	Mean/ Median of age (Year)	Sample size	Diabetes N (Total)	Diabetes N (Older PLHIV)	Ascertainment of T2DM*	OR Age (CI)	Quality
Bam N, 2022 [20]	South Africa	Case-control	NA**	531	177	29	NA	NA	High
Bratu A, 2021 [95]	Canada	Cohort	43.2	2792	129	59	physician billing and drug dispensation datasets	NA	High
Buendia J, 2021 [21]	USA	Cross sectional	NA	989	153	100	formal medical chart diagnosis, insulin/oral hypoglycemic prescriptions, or recent fasting blood glucose ≥ 126 mg/dL	NA	Low
Capeau J, 2012 [22]	France	Cohort	NA	1046	111	30	fasting glycemia at least 7.0mmol/l or 2-h OGGT glycemia at least 11.1mmol/l, and/or if initiation with antidiabetic drugs (including metformin) during follow-up	3.63 (2.22–5.92)	High
Chang D, 2022 [23]	4 African countries	Cohort	NA	3099	49	18	fasting glucose ≥ 126 mg/dl or antidiabetic medication	2.87 (1.49–5.16)	High
Chimbetete C, 2017 [52]	Zimbabwe	Cohort	37	4110	57	16	Random blood sugar higher than 11.1 mmol/L or fasting blood sugar higher than 7.0 mmol/L	8.35 (5.1–14.3)	High
Gunter J, 2017 [25]	Belgium	Cohort	46.7	5787	344	229	the clinical diagnosis of the condition or the prescription of antidiabetic drugs or insulin.		Low
Han W, 2019 [26]	13 asian countries	Cohort	35	1927	127	16	fasting blood glucose ≥ 126 mg/dL, glycated hemoglobin ≥ 6.5%, a two-hour plasma glucose ≥ 200 mg/dL, or a random plasma glucose ≥ 200 mg/dL	4.19 (1.12–8.28)	High
Kenyara L, 2024 [27]	Kenya	Cross sectional	NA	306	15	11	hemoglobin A1c ( > = 6.5)	NA	Low
Ledergerber B, 2007 [28]	switzerland	Cohort	NA	6513	123	43	plasma glucose level cut-off values of 17.0 mmol/L (fasting) and 111.1 mmol/L (non-fasting)	NA	High
Lo Y, 2009 [29]	Taiwan	Case-control	34	824	50	12	fasting blood glucose ≥ 126 mg/dL	NA	High
Shankalala P, 2017 [96]	Zambia	Cross sectional		270	40	28	random blood sugar levels and fasting glucose		Low
Sogbanmu O, 2019 [97]	South Africa	Cross sectional		335	21	7	HbA1c ≥ 6.5%		Low
Steiniche D, 2016 [62]	Guinea-Bissau	Cross sectional	37	953	52	10	Fasting blood glucose ≥ 7.0 mmol/L (≥ 126 mg/dL) and using anti-diabetic treatment	2.48 (0.95–6.45)	High
Ye Y, 2020 [37]	USA	Cohort	NA	3975	354	243	hemoglobin A1c (≥ 42 mmol/mol), fasting plasma glucose (≥ 126 mg/dl), and random plasma glucose test (≥ 200 mg/dl)	NA	High
Umare J, 2023 [57]	Nigeria	Cross sectional	45	440	90	68			High
Obimakinde A, 2020 [31]	Nigeria	Cross sectional	NA	62	2	2	fasting blood glucose ≥ 126 mg/dL	NA	Low
Justice A, 2021 [98]		Cohort		9189	1996	1996	NA		Low
Rubtsova A, 2021 [33]	USA	Cohort	57	356	103	103	self-report	NA	High
Vance D, 2011 [36]	USA	Cross sectional	NA	1478	132	59	NA	NA	Low
Moore D, 2018 [99]	USA	Cross sectional	59	99	26	26	NA	NA	Low
Sheppard D, 2017 [34]	USA	Cross sectional	55	40	3	3	NA	NA	Low
Morgan E, 2012 [30]	USA	Cross sectional	NA	92	9	9	NA	NA	Low
Pyarali F, 2021 [100]	USA	Cross sectional	52	985	166	166	NA	NA	Low
Zanella I, 2022 [38]	Italy	Cross sectional	69	60	10	10	taking antidiabetic drugs or persistently glycemia > 120 mg/ dL	NA	Low
Okyere J, 2022 [32]	South Africa	Cross sectional	NA	514	43	43	NA	2.2 (0.94–5.28)	High
Rabe M, 2020 [43]	South Africa	Cohort	NA	191	15	15	NA	NA	Low
Murray M, 2021 [40]	USA	Cross sectional	NA	562	73	73	NA	NA	Low
Mugisha M, 2016 [39]	Uganda	Cross sectional	NA	244	4	4	self-report	NA	High
Puhr R, 2019 [42]	Australia	Cross sectional	NA	200	30	30	self-report	NA	High
Patel R, 2016 [41]	UK	Cross sectional	58	299	33	33	NA	NA	Low
Shah S, 2002 [44]	USA	Cross sectional	59	198	22	22	NA	NA	Low
Bratt G, 2021 [101]	Sweden	Cohort	57	505	43	43	fasting glucose values □7.0 mmol/L		High
Mohammad G, 2024 [65]	USA	Cohort	35	9115	578	193	NA	1.19 (0.52–2.72)	High
Samad F, 2017 [102]	Canada	Cohort	NA	703	123	93	random blood sugar ≥ 11.1 mmol/L, fasting blood sugar ≥ 7 mmol/L, HbA1C ≥ 6.5%, antidiabetic medication use during the follow-up	0.53 (0.23–1.22)	High

* Type 2 diabetes mellitus

** Not available

Heterogeneity test

The heterogeneity was very high (I² >90%) in most analyses.

Effect size based on OR and IR

This meta-analysis estimated overall effect sizes using both ORs and IRRs for the incidence of Type 2 Diabetes Mellitus. Based on the pooled IRR from an analysis of 11 studies involving 79,351 participants, it was shown that HIV infection significantly increases the risk of developing Type 2 Diabetes Mellitus [45–53]. The pooled IRR was 4.79 (95% CI: 1.89–12.15 per 1000 person-years). “In parallel, an overall OR was calculated from the analysis of eight studies involving 16,590 PLHIV [42, 54–59]. The overall OR for Type 2 Diabetes Mellitus among PLHIV was 1.61 (95% CI: 1.09–2.38).

Moreover, in the analysis of 36 studies comprising 37,415 PLHIV over 50 years old, the OR of diabetes was higher at 3.35 (95% CI: 2.52–4.44). On the other hand, an analysis of seven studies comparing current ART users with non-ART users showed no statistically significant link between ART use and a higher risk of diabetes 0.80 (95% CI: 0.38–1.68) (Fig. 2). In addition, 10 studies (n = 14,120 PLHIV) found an OR of 2.30 (95% CI: 1.58–3.73) for diabetes among individuals with a family history of diabetes [29, 56, 60–67]. Five studies (n = 3,189 PLHIV) reported an OR of 2.43 (95% CI: 1.58–3.73) for diabetes in individuals with BMI >25 [59, 63, 68–70]. Ten studies (n = 3,083 PLHIV) showed a significantly higher OR for diabetes of 2.10 (95% CI: 1.39–3.19) among individuals with hypertension [55, 63–65, 68, 71–75]. The OR for diabetes was also higher among male PLHIV 1.14 (95% CI: 0.99, 1.32), but it was not statistically significant (Fig. 3).

Fig. 2.

Incidence rates, odds ratios, and associated odds ratios for diabetes among people living with HIV

Fig. 3.

Odds ratios associated with diabetes mellitus among people living with HIV

Effect size based on prevalence

The pooled prevalence of T2DM among PLHIV was found to be 9% (95% CI: 8%–10%; n = 112 studies; p < 0.001) (Supplementary file). To reduce heterogeneity and achieve greater convergence, we performed subgroup analyses based on age, sex, ART regimes, ART duration, HIV status, continent, and study quality.

The prevalence of T2DM among PLHIV aged over 50 years was 11% (95% CI: 9%–12%; p < 0.001) (Fig. 4). Additionally, the highest prevalence was each observed among individuals on nucleoside reverse transcriptase inhibitors (NRTIs) at 14% (95% CI: 2%–28%; p < 0.001). Also, individuals on ART for ≥ 10 years experienced a higher prevalence of T2DM compared to those on ART for < 10 years (15% vs. 6%). Furthermore, the prevalence of T2DM among HIV-positive individuals without progression to AHD differed from that among individuals with a diagnosis of AHD (6% vs. 8%) (Table 2). Further subgroup analyses uncovered notable variations by study design. Case-control studies had a higher prevalence at 21% (95% CI: 2%–41%; p < 0.001). Geographic analysis indicated marked continental differences, with the highest prevalence of T2DM recorded in America 10% (95% CI: 9%–14%; p < 0.001) and the lowest in Africa 2%(95% CI: 1.9%–3%; p < 0.001). Furthermore, study quality assessment revealed that high-quality studies reported a lower prevalence of T2DM compared to lower-quality studies (7% VS 12%) (Table 2).

Fig. 4.

Prevalence of diabetes mellitus among people living with HIV

Table 2.

Subgroup analysis for the prevalence of diabetes among people living with HIV

Variable	Number of studies	Sample size	Prevalence (%)	95% CI (%)	I2 (%)	p-value*
ART regimes
Pls	14	47,383	11	6–17	> 90	< 0.001
NRTIs	14	97,587	14	2–28	> 90	< 0.001
NNRTIs	10	44,978	9	5–13	> 90	< 0.001
ART Duration
< 10	7	4947	6	5–8	0.02	< 0.001
>=10	2	1051	15	6–26	> 90	< 0.001
Disease Status
HIV	11	14,634	6	4–9	88.40	< 0.001
AHD	11	22,813	8	4–14	> 90	< 0.001
Type of study
Cohort	47	332,988	8	7–9	> 90	< 0.001
Case-Control	3	1557	21	2–41	0.00	< 0.001
Cross sectional	50	15,577	10	8–12	> 90	< 0.001
Continent
Africa	42	107,582	8	7–9	> 90	< 0.001
America	39	218,349	10	10–14	> 90	< 0.001
Asia	12	47,898	6	5–8	> 90	< 0.001
European	14	99,660	7	6–8	> 90	< 0.001
Ocean	2	16,862	3	1.9-3	0.00	< 0.001
Quality
High	70	389,689	7	6–8	> 90	< 0.001
Low	40	146,072	12	9–15	> 90	< 0.001

*P-value for heterogeneity test

Publication bias

Assessment of publication bias did not detect significant bias concerning the odds ratio outcomes (Egger: p-value = 0.46) and prevalence (Egger: p-value = 0.24), suggesting robustness in the results reported (Fig. 5).

Fig. 5.

Funnel plot of included studies related to diabetes mellitus among people living with HIV

Meta-regression

Meta-regression was conducted to identify sources of heterogeneity based on age, sex, HIV status (HIV or AHD), ART regimens, ART duration, continent, BMI, hypertension, and family history of diabetes (Supplementary file). The results showed that none of these variables had significant associations with the effect size in this study. In other words, these variables could not explain the heterogeneity observed in the results.

Sensitivity analysis

To assess the robustness of our findings, we performed a sensitivity analysis. This involved systematically removing each study from the meta-analysis and examining the impact on the overall results (Fig. 6).

Fig. 6.

Results of sensitivity analysis for the odds ratio for diabetes risk among people living with HIV aged over 50 years

Discussion

According to the study findings, the overall OR of diabetes among PLHIV was higher than in those without HIV (OR = 1.61), with a total prevalence of 9%. Additionally, among adults over 50 years old, the OR was 3.35, with a diabetes prevalence of 11%. Beyond traditional risk factors, our study identified significant relationships between diabetes prevalence and key HIV-specific features. Specifically, a longer duration of HIV infection (> 10 years) and a history of an AHD were associated with a higher prevalence of T2DM, indicating that the long-term effects of chronic viral infection and prior immune suppression are important factors in metabolic health. Furthermore, having a family history of diabetes, a BMI above 25, and hypertension were linked to diabetes among PLHIV.

The incidence and prevalence of diabetes reported in our study were consistent with findings from other studies. A meta-analysis reported that the odds of diabetes among PLHIV was 3.8 times greater (OR = 3.8) compared to the non-infected group [76]. Moreover, a systematic review in Africa found that T2DM prevalence among HIV-infected adults was 6.1% (95% CI 3.8% to 9.7%) [77]. This increased risk stems from persistent immune activation during HIV infection, which increases pro-inflammatory cytokines (e.g., TNF-α, IL-6) [78]. These cytokines disrupt insulin pathways in muscle and adipose tissue, contributing to insulin resistance and diabetes development [13]. Adopting lifestyle changes, including consistent aerobic activities like brisk walking and a nutritious diet abundant in fruits, vegetables, and whole grains, can reduce the risk of diabetes among PLHIV [79]. Additionally, routine monitoring of blood glucose and cardiovascular risk factors, along with incorporating diabetes management into HIV care, is essential [80, 81].

Among PLHIV aged 50 years and older, our study observed a significantly increased odds of diabetes (OR = 3.35) and a prevalence of 11%. These results are consistent with those reported in previous studies [82]. A systematic review and meta-analysis found that aging is a major risk factor for diabetes and prediabetes among HIV-infected adults on antiretroviral therapy [82]. The increased risk is likely attributable to a combination of HIV-specific factors and age-related processes. Individuals over 50 years old were treated with earlier antiretroviral therapy regimens that disrupted glucose metabolism and increased insulin resistance [83]. Furthermore, long-term HIV infection causes increased oxidative stress, which can disrupt insulin function and blood sugar regulation, raising the risk of developing type 2 diabetes [84]. Lifestyle modifications, emphasizing nutrition, weight control, and exercise, have proven effective in lowering diabetes risk among these individuals [80]. Furthermore, regular diabetes screening and evaluation of risk factors, along with careful selection and close monitoring of antiretroviral therapy, are essential, especially for individuals at increased risk of diabetes [79]. Focused screening and health education tailored to this group can reduce the incidence of diabetes, improve metabolic health, and ultimately enhance quality of life.

Our research demonstrates that potential risk factors such as a family history of diabetes (OR = 2.3), a BMI over 25 (OR = 2.43), and hypertension (OR = 2.10) are linked to an increased likelihood of developing diabetes among PLHIV. These findings are comparable to those of other studies, which reported higher odds of DM among PLHIV with obesity (OR = 9.63), family history of DM (OR = 10.65) [12], and hypertension (OR = 4.95) [63]. Elevated BMI, high blood pressure, and a family history of diabetes are recognized traditional risk factors that increase the likelihood of diabetes by further reducing insulin sensitivity and affecting pancreatic function [66]. People with these risk factors, especially those who have a family history of diabetes, should undergo more frequent and careful screening for diabetes, as they have a significantly higher risk, and early detection can help delay or prevent disease progression and complications.

Among PLHIV, the prevalence of diabetes was highest among those treated with NRTI-based ART regimens and those with an HIV duration exceeding ten years, each about 14%. Additionally, 7% of individuals with an AHD diagnosis also had diabetes. These findings are consistent with previous studies, which reported that an association exists between NRTI use and diabetes risk [85, 86]. Furthermore, these studies reported a prevalence of diabetes of 12.9% in PLHIV for more than ten years. The observed association between NRTI-based ART and increased diabetes risk is likely due to multiple factors. First, well-known biological factors, such as mitochondrial toxicity and metabolic disruption caused by older NRTIs, may alter insulin signaling and glucose metabolism, providing a convincing explanation [13]. However, it is critical to consider potential epidemiological confounders. Patients who require NRTI-based ART may have specific baseline characteristics or disease profiles that predispose them to diabetes [82]. For instance, these regimens may be more prevalent in specific clinical contexts or resource-limited settings where other diabetes risk factors are also more common [87–89]. Furthermore, lifestyle factors such as diet, physical activity, and smoking, which are independent risk factors for diabetes, may be more common in specific subgroups of PLHIV, thereby affecting these associations [90]. Additionally, improved healthcare access and more frequent diabetes screening among people on ART may result in a higher detection rate of diabetes, regardless of the ART regimen [90, 91]. The increased likelihood of diabetes with longer HIV duration and in individuals with AHD, irrespective of common risk factors like obesity [92], further underscores the complexity of metabolic health in this population. The association between prolonged NRTI use and higher diabetes risk, especially with older regimens, emphasizes the importance of considering metabolic profiles when choosing ART, particularly given potential confounding factors. To reduce the heightened diabetes risk in individuals with HIV, particularly linked to ART, recent studies recommend careful choice and ongoing monitoring of ART regimens, prioritizing NRTIs that have more favorable metabolic effects when feasible [93]. Additionally, regular diabetes screening and timely intervention are essential to identify and manage prediabetes and diabetes early [94].

Limitations

This study has several limitations. First, some data were obtained from cross-sectional studies, which limit the ability to determine causal or temporal relationships between HIV, antiretroviral therapy, and the development of diabetes. Second, there is a lack of published research from certain regions, especially low-income countries and specific continents, which may restrict the global applicability of the findings. Third, a significant limitation of this meta-analysis is the variation in diagnostic criteria for diabetes mellitus across the included studies. While some studies used fasting plasma glucose thresholds, others relied on HbA1c levels. Additionally, several studies depended on self-reported diagnosis or other methods. The heterogeneity in diagnostic criteria is a significant source of clinical and statistical heterogeneity and may impact the comparability of pooled estimates. Fourth, considering the meta-regression results, subgroup analyses should be regarded as preliminary ideas for future longitudinal studies rather than definitive findings. Finally, meta-regression did not find significant sources of the high heterogeneity observed. This may be due to unmeasured population-level factors (such as socioeconomic status and genetic predisposition), clinical heterogeneity (such as varying stages of HIV or different standards of care), or the variation in T2DM diagnostic criteria across studies.

Conclusion

The findings of this study indicate that PLHIV have a threefold higher risk of developing T2DM compared to HIV-negative individuals, with an even greater risk among those aged 50 years or older. Key risk factors, including family history of diabetes, high BMI, high hypertension, and the use of NRTI-containing ART regimens, significantly contribute to this increased risk. It is recommended that policymakers and stakeholders prioritize routine T2DM screening and support lifestyle interventions targeting body weight and blood pressure management.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

PLHIV

People living with HIV

T2DM

Type 2 diabetes mellitus

ART

Antiretroviral therapy

NCDs

Non-communicable diseases

PRISMA

Preferred reporting items for systematic reviews and meta-analyses

OR

Odds ratio

IR

Incidence rate

BMI

Body mass index

NRTIs

Nucleoside reverse transcriptase inhibitors

Author contributions

HM and MN contributed to the study design, article screening, searching for articles, data interpretation, and confirming the submitted version. ZV and SK contributed to the study design, data extraction, searching for articles, data interpretation, and confirming the submitted version. ZA contributed to the study design, article screening, quality assessment, data interpretation, and confirming the submitted version. SR contributed to the analysis, data interpretation, and confirmed the submitted version. LA contributed to the study design, data interpretation, and confirmed the submitted version.

Funding

There are no sources of funding for this study.

Data availability

All data are provided within the manuscript or supplementary information files.

Declarations

Ethics approval and consent to participate

The protocol and procedures of the present study were reviewed and approved by the research ethics committee of Tehran University of Medical Sciences (Ethics Code: IR.TUMS.IKHC.REC.1404.012).

Consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.