Introduction

As one of the major geriatric syndromes, frailty is a multidimensional condition that includes physical, cognitive, psychological, and social components1. Social frailty, a key aspect of frailty, has an incidence rate of 15–45.7% among the older population in China2. It refers to an individual’s prolonged lack of one or more essential social resources that support health, primarily including deficiencies in social behaviors, social activities, and self-management3. Research has shown that social frailty increases the risk of sarcopenia, disability, and other adverse outcomes in older patients and is significantly associated with all-cause mortality4. Thus, addressing social frailty in the older population is crucial in preventing negative outcomes and promoting healthy aging.

As one of the global public health issues, AIDS has been transformed from a fatal disease into a controllable and treatable chronic condition with the emergence of antiretroviral therapy (ART), which significantly extends the life expectancy of patients5. According to the Joint United Nations Programme on HIV/AIDS (UNAIDS), it is estimated that by 2023, approximately 39 million people worldwide will be infected with HIV6, and the number of HIV/AIDS patients aged ≥ 50 will increase from 5.4 million in 2015 to 8.2 million in 2022. In China, older adults living with HIV/AIDS (≥ 50 years old) account for 36.8% of the total HIV/AIDS population7. WHO defines individuals aged ≥ 60 in developing countries as the “older” population8. However, in AIDS research, patients experience accelerated immune aging, thus leading to geriatric syndromes (such as frailty and multimorbidity) approximately a decade earlier than the general population9. In order to differentiate them from the sexually active population aged 15–49, individuals aged ≥ 50 years with HIV/AIDS are typically classified as older adults living with HIV/AIDS10,11,12. Due to stigma, prejudice, negative emotions, multiple complications, and the economic burden of long-term treatment, older adults living with HIV/AIDS face increased depletion of psychological and social resources. This leads to reduced social participation and heightened feelings of loneliness, thereby further exacerbating social issues11,12. Studies have shown that social frailty in HIV/AIDS patients represents a significant proportion of frailty, and the most prominent issues are the weakening of social relationships and reduced social support in terms of social frailty13. Therefore, under an aging population, social frailty among older adults living with HIV/AIDS is becoming an increasingly critical issue. Existing research predominantly focuses on the physiological frailty in older adults living with HIV/AIDS, while little attention is given to social frailty in this group14. Early identification of social frailty in older adults living with HIV/AIDS and an exploration of related risk factors are urgent in clinical practices. This is essential for clinical staff and community healthcare institutions to formulate targeted intervention measures and prevent adverse clinical outcomes.

Research has shown that social frailty in the older population is influenced by multiple factors, including individual characteristics and external environments. For instance, a study based on older individuals in Chinese communities2 found that marital status and self-reported health were significantly related to social frailty. A meta-analysis7 highlighted that depression, chronic diseases, and sleep issues are risk factors for social frailty in community-dwelling older adults. In older adults with heart failure, social frailty was found to be associated with age, education, marital status, monthly household income per capita, living arrangements, exercise, disease duration, cognitive impairment, and activities of daily living (ADL)15. However, whether these factors affect social frailty in older adults living with HIV/AIDS requires further investigation. The Health Ecology Model (HEM) posits that health is the outcome of the interaction of individual factors and external environments, emphasizing the multi-level impacts of the environment on individuals and the complexity and diversity of influencing factors16. Due to the complexity and diversity of these factors, multicollinearity may arise. The Least Absolute Shrinkage and Selection Operator (LASSO) regression helps control model complexity by introducing a penalty that shrinks the coefficients of insignificant or minimally significant variables to zero, thus selecting the most relevant variables, as well as avoiding overfitting and multicollinearity. Unlike traditional regression models, LASSO allows for the simultaneous processing of all independent variables, significantly enhancing the model’s stability17. Random Forest (RF) helps identify key variables by providing more accurate and efficient comparisons of variable importance and analyzing complex nonlinear relationships18. The “black box” nature of machine learning means that it is difficult to understand how sample features influence the final results, but Shapley Additive Explanations (SHAP) provide the advantage of explaining individual prediction outcomes19. Therefore, based on the five aspects of HEM (personal traits, behavioral characteristics, interpersonal networks, living and working conditions, and policy environment), this study uses LASSO regression and the SHAP algorithm in RF to identify the risk factors of social frailty in older adults living with HIV/AIDSand provides a basis for the development of targeted intervention strategies.

Methods

Research design and participants

From January to December 2024, convenience sampling was used to select 335 older adults living with HIV/AIDS who visited the outpatient department of a tertiary Grade A hospital in Zunyi, Guizhou Province, China, as study participants.

Inclusion criteria:

(1) Meeting the diagnostic criteria outlined in the China HIV/AIDS Diagnosis and Treatment Guidelines (2021 Edition)20;

(2) Aged ≥ 50 years9;

(3) Able to communicate normally and be willing to participate in the study.

Exclusion criteria:

(1) Patients with coexisting mental disorders or severe systemic diseases;

(2) Patients with sensory impairments (e.g., visual or auditory) who are unable to provide valid information.

Sample size calculation: The event per variable (EPV)21 was used for sample size calculation. It is generally recommended to have at least 10 observations per predictor variable to achieve good prediction performance. In this study, 26 predictor variables were included. Considering a 20% rate of invalid questionnaires, the sample size should be at least 312 participants. Ultimately, 335 participants were included in the study. The selection of study participants is illustrated in Fig. 1. This study was reviewed and approved by the Ethics Committee of the Zunyi Medical University Affiliated Hospital (Ethics No: KLLY-2023-714). All participants provided informed consent, and the clinical investigation was conducted in accordance with the principles of the Declaration of Helsinki.

Fig. 1
figure 1

Flowchart of participant selection.

Full size image

Variables

According to HEM16, independent variables were selected from five dimensions: personal traits, psychological behaviors, interpersonal networks, work and living conditions, and policy environment (the relationship between independent variables and HEM is shown in Fig. 2). The detailed content and value assignment of the independent variables are listed in Table 1. Some independent variables are defined as follows. (1) Smoking history: ever smoked or currently smoking; (2) Alcohol history: Alcohol consumption in the past year; (3) BMI: Weight (kg) / Height (m²). Underweight (< 18.5 kg/m²), Normal (18.5–25.0 kg/m²), Overweight or higher (≥ 25.0 kg/m²); (4) Exercise: Never (0), Sometimes (1–3 times/week), Frequently (> 3 times/week).

Fig. 2
figure 2

Relationship between independent variables and HEM.

Full size image
Table 1 Selected independent variables and assigned values based on HEM.
Full size table

Measurement

Questionnaire of general information

Based on a literature review and consultations with experts, the general information questionnaire is designed to include demographic data and information on clinical diseases. Demographic data includes: gender, age, education, ethnicity, marital status, per capita monthly household income, smoking history, alcohol history, and exercise. The information of clinical diseases includes: time of HIV diagnosed, complications, ART payment, and CD4+ T lymphocyte count.

Social frailty

The Help, Participation, Loneliness, Financial, Talk (HALFT) scale, developed by Ma et al. in 2018 based on the Chinese context22, was used to assess social frailty in older adults living with HIV/AIDS. The scale consists of five items: (1) Help: Were you able to assist friends or family in the past year? (2) Participation: Did you engage in social or recreational activities in the past year? (3) Loneliness: Have you felt lonely in the past week? (4) Financial: Has your income in the past year been sufficient to sustain you for the entire year? (5) Talk: Do you have someone to talk to every day? Each item is scored based on a positive/negative response, with 0 or 1 point assigned. The total score ranges from 0 to 5, where a score of 0 indicates no social frailty, 1–2 points indicates the early stages of social frailty, and ≥ 3 points indicates social frailty. This scale has been widely applied in both community and clinical settings in China. After validation in the community-dwelling older population in China, Cronbach’s α was 0.602. In the older population in China, the test-retest Cronbach’s α was 0.723.

Depression

The Geriatric Depression Scale-15 (GDS-15)25, developed by Yesavage et al. in 198223, was used to assess depression in older adults living with HIV/AIDS. This scale is a specific tool for screening depression in older adults and was revised in 198626. It consists of 15 items, with items 1, 5, 7, 11, and 13 scored as 1 point for a “no” response, and the remaining items scored as 1 point for a “yes” response. The total score ranges from 0 to 15, and a score of ≥ 8 indicates the presence of depression. The scale has been validated in the older Chinese population, with a Cronbach’s α of 0.79327.

Cognitive impairment

Cognitive function in older adults living with HIV/AIDS was assessed using the Beijing version of the Montreal Cognitive Assessment (MoCA), developed by Yu et al. in 201228. The assessment includes seven dimensions: visuospatial and executive function, naming, attention, language, abstraction, delayed recall, and orientation, with a total score of 30 points. For individuals with ≤ 12 years of education, 1 point is added to the total score for correction. A score of < 26 is considered indicative of cognitive impairment. After validation in the Chinese older population, Cronbach’s α was 0.830. The scale has been widely used in the HIV-infected population29.

Sleep quality

The Pittsburgh Sleep Quality Index (PSQI), developed by Buysse et al. in 198930, was used to assess the sleep quality of older adults living with HIV/AIDS over the past month. It was adapted into Chinese by Liu et al.31. The scale consists of seven dimensions with 19 items and five scores rated by cohabitants (these are not included in the total score). The scale uses a Likert 4-point scoring system, with a total score range of 0 to 21 points. A PSQI score > 7 indicates the presence of sleep disorders. The Cronbach’s α for the scale was 0.83, with a test-retest Cronbach’s α of 0.85. Its sensitivity was 89.6%, and its specificity was 86.5%. After validation in the Chinese population, Cronbach’s α was 0.84231, demonstrating good reliability and validity.

ADL

The Chinese version of the Barthel Index (BI) was used to assess the ADL of older adults living with HIV/AIDS. The BI, developed by Mahoney et al. in 196532, was translated into Chinese by Hou et al.33. It includes 10 items: feeding, grooming, bathing, dressing, bladder control, bowel control, toilet use, mobility (on level surfaces), stair climbing, and transfers (bed to chair and back). The total score ranges from 0 to 100, with higher scores indicating better ADL ability. The scale has shown excellent reliability in the Chinese population, with a Cronbach’s α of 0.916.

Social support

The social support rating scale (SSRS), developed by Xiao34, was used to assess the level of social support among older adults living with HIV/AIDS. This scale consists of three dimensions and a total of 10 items: objective support (3 items), subjective support (4 items), and support utilization (3 items). The total score ranges from 12 to 66, with scores of 0–22 indicating low social support, 23–44 indicating moderate social support, and 45–66 indicating satisfactory social support. The Cronbach’s α for the scale items range from 0.890 to 0.940, demonstrating good reliability and validity. Its effectiveness has been confirmed in older adults living with HIV/AIDS in China35.

Data collection

To ensure the quality of the questionnaire, investigators received professional training. After informed consent was obtained from the participants, the questionnaires were distributed for the patients to complete, avoiding the use of suggestive language. For patients who were illiterate or unable to fill out the forms, the investigator read each survey item aloud to participants and recorded their responses accordingly. A total of 350 questionnaires were distributed, and those with logical errors or incomplete responses were excluded. Ultimately, 335 valid questionnaires were collected, resulting in a valid response rate of 95.7%.

Statistical analysis

Data processing and analysis were conducted using SPSS 29.0 (Version 29.0; IBM Corp., Armonk, NY, 2022), Python 3.11 (Version 3.11; Python Software Foundation, Wilmington, DE, 2022), and R 4.3.1 (Version 4.3.1; R Core Team, 2023). Qualitative data were presented as frequencies and percentages, with intergroup comparisons performed using the χ² test. Quantitative data were expressed as medians and interquartile ranges [M(P25, P75)], with intergroup comparisons conducted using the rank-sum test. LASSO combined with RF was employed to identify factors influencing social frailty. LASSO was implemented using the “glmnet” package in R, with 10-fold cross-validation to mitigate model overfitting and multicollinearity. Based on the regression coefficients at lambda.min, variables with non-zero coefficients were selected from an initial set of 26 variables. The selected variables were then incorporated into the RF model constructed using Python 3.11. Feature importance analysis was conducted to assess the contribution of each variable, and SHAP dependence plots were generated based on SHAP values to further examine the relationships between features and their SHAP values. A P-value of < 0.05 was considered statistically significant.

Results

General information about the surveyed subjects

This study included 335 older adults living with HIV/AIDS, with ages ranging from 50 to 89 (65.69 ± 9.02) years. Other general information is presented in Table 2.

Table 2 Characteristics of social frailty in older adults living with HIV/AIDS.
Full size table

Univariate analysis of social frailty in older adults living with HIV/AIDS

Among the 335 older adults living with HIV/AIDS, 105 (31.34%) experienced social frailty. The univariate analysis based on the HEM revealed that the following factors had statistically significant differences in social frailty among older adults living with HIV/AIDS (P < 0.05): Personal traits: age, complications, BMI, CD4+ lymphocyte count, HIV viral load, TC, LDL-C; Interpersonal network: marital status and social support; Behavioral characteristics: depression, sleep disorders, cognitive impairment, exercise, smoking history, alcohol history, and ADL; Living and working conditions: education and per capita monthly household income; Policy environment: ART payment. These results are shown in Table 2.

Variable selection using LASSO

To identify the most relevant variables associated with social frailty in older adults living with HIV/AIDS, LASSO17 was used for selection, and 26 variables were analyzed, with λ and binomial deviance calculated from the test data to evaluate the predictive performance of the fitted model. In Fig. 3(a), the red dot represents the target variable corresponding to the λ value, and the two dashed lines represent two specific λ values: the former one minimizes binomial deviance, while the latter one indicates the maximum λ value within one standard error of the minimum binomial deviance. In Fig. 3(b), each color represents a different variable. As the regularization parameter λ increases, the number of independent variables in the model gradually decreases. After ten-fold cross-validation, lambda.min (λ = 0.0127) and lambda.1se (λ = 0.035) were identified as the two parameters. Based on the regression coefficients at lambda.min, 12 variables with non-zero coefficients were selected from the 26 variables. These variables include age, per capita monthly household income, exercise, smoking history, complications, ART payment, CD4+ T lymphocyte count, depression, sleep disorders, cognitive impairment, ADL, and social support.

Fig. 3
figure 3

Results of LASSO variable selection (a) Cross-validation plot; (b) LASSO solution path.

Full size image

Variable selection using RF

Importance ranking

The 12 variables selected by LASSO were incorporated into the RF model, with the occurrence of social frailty as the dependent variable. By calculating the SHAP values of each variable (Fig. 4b), the importance ranking shown in Fig. 4(a) was derived. The variables, ranked by their importance, are ADL, depression, cognitive impairment, exercise, social support, per capita household monthly income, sleep disorders, age, complications, CD4+ T lymphocyte count, smoking history, and ART payment.

Fig. 4
figure 4

Importance ranking of independent variables based on SHAP values. (a) SHAP values of independent variables; (b) Importance ranking of independent variables.

Full size image

SHAP dependence plot of social frailty risk in older adults living with HIV/AIDS

The occurrence of social frailty was set as the dependent variable, and the value assignment of independent variables was detailed in Table 1. The SHAP dependence plot shows that older adults living with HIV/AIDS with lower ADL, lack of exercise habits, cognitive impairment, depression, inadequate social support, older age, sleep disorders, low CD4+ T lymphocyte count, insufficient per capita family monthly income, complications, smoking, and participation in the free ART program (dual or triple therapy) have a higher risk of experiencing social frailty. The SHAP dependence plots for the different independent variables are shown in Fig. 5.

Fig. 5
figure 5

SHAP dependence plots of risk factors for social frailty in older adults living with HIV/AIDS.

Full size image

Discussion

This study aims to analyze the current status and risk factors of social frailty in older adults living with HIV/AIDS. The results indicate that the incidence of social frailty in older adults living with HIV/AIDS is 31.34%, which is higher than the findings of Qi et al.2 and Huang et al.36. This may be attributed to factors such as chronic inflammatory responses caused by HIV infection, prolonged immune activation, and mitochondrial damage to cells caused by ART treatment, all of which accelerate the impact on the physiological, psychological, and social adaptation of older adults living with HIV/AIDS37 and results in a higher incidence of social frailty. Furthermore, there is currently no unified evaluation standard or assessment tool for social frailty, leading to different measurement outcomes. This emphasizes the need for healthcare staff at designated medical institutions and community healthcare workers to focus on identifying social frailty in older adults living with HIV/AIDS. Future studies should establish standardized criteria and evaluation tools for social frailty to facilitate the early identification of high-risk populations and the timely implementation of targeted interventions to improve clinical outcomes.

Based on the importance ranking from the RF model and the SHAP dependency plots, it was found that older adults living with HIV/AIDS with lower ADL, lack of exercise habits, cognitive impairment, depression, limited social support, older age, sleep disorders, low CD4+ lymphocyte count, insufficient per capita household income, complications, smoking history, and participation in free ART program have a higher risk of social frailty. These findings indicate that the factors influencing social frailty in older adults living with HIV/AIDS encompass personal traits, behavioral characteristics, living and working conditions, interpersonal networks, and the policy environment.

As the core component of HEM, personal traits mainly include biological characteristics such as physiological conditions and genetic factors, which are the most direct factors influencing social frailty in older adults living with HIV/AIDS. This study identified that under the category of personal traits, age, CD4+ T lymphocyte count, and complications are risk factors for the development of social frailty in older adults living with HIV/AIDS. Among these, age is the most important predictor under personal traits, which is consistent with Ragusa et al.38. With the combined effects of aging and the decline in CD4+ T lymphocyte count, the immune system of older adults living with HIV/AIDS is progressively compromised, leading to dysfunction in stress response, metabolism, and muscle function, which severely impacts overall health. This increases the risk of opportunistic infections and malignancies, thus causing a continuous decline in the ability of self-management and ultimately resulting in social isolation, reduced social interaction, and a lack of social roles1,39. CD4+ T lymphocyte count is a key indicator for assessing the disease progression, predicting clinical outcomes, and evaluating the effectiveness of antiviral therapy in older adults living with HIV/AIDS40. During follow-up, regular monitoring of CD4+ T lymphocytes and social frailty should be conducted. Multidisciplinary collaboration should be employed to further optimize treatment strategies, promote immune function recovery, reduce the incidence of complications, and support healthy aging in this population.

In behavioral characteristics, low ADL, lack of exercise habits, sleep disorders, cognitive impairment, depression, and smoking are associated with a higher incidence of social frailty in older adults living with HIV/AIDS, which is consistent with findings from other studies23,41. Among these, ADL and exercise are significant predictive factors. This is due to the fact that older adults living with HIV/AIDS are prone to developing other chronic conditions that can cause damage to important organs such as the brain or heart, thus further leading to a decline in physical function and a reduction in ADL42. Additionally, fatigue caused by the side effects of ART drugs can result in daytime lethargy and drowsiness, which disrupts regular routines, reduces physical activity, and suppresses the motivation to exercise43. The combined effects of these factors lead to lower social participation and social barriers in older adults living with HIV/AIDS44. Moreover, stigma and internalized discrimination caused by HIV infection make older adults living with HIV/AIDS more susceptible to negative emotions such as depression, which diminishes their social capital (personal relationships, social networks, social trust, social reciprocity, and social participation)45,46. Research has shown that older adults living with HIV/AIDS experience significant cognitive impairments in areas such as executive function, processing speed, language, recall, and psychomotor abilities due to neuroinflammation and apoptosis pathways activated by HIV and ART medications, which disrupt calcium homeostasis and promote oxidative stress, depleting neurotrophic factors47. This leads to a decline in their ability to access social resources, understand, and communicate, making it even more difficult for them to engage in social activities and establish connections with the outside world, thereby impairing their alternative compensatory mechanisms for social needs48. However, studies have shown that exercise not only improves the physical function of older adults living with HIV/AIDS but also enhances their psychological and sleep conditions, promotes social interactions, and increases social activities and participation49. It is recommended that healthcare professionals encourage exercise among older adults living with HIV/AIDS and incorporate multidisciplinary exercise interventions as a targeted strategy to prevent social frailty in this group, thus motivating patients to engage in social networks.

For the category of interpersonal network, the results of this study indicate that older adults living with HIV/AIDS with low social support have a higher incidence of social frailty, which is consistent with the findings of Qin et al.23. As a buffer against stress, adequate social support can expand the social support network of older adults living with HIV/AIDS, increase their social engagement and participation, and ultimately enhance their quality of life and sense of well-being50. Moreover, strong social support can provide patients with essential resources and information, including medical knowledge, medication guidance, material assistance, emotional support, and financial security, thereby improving their abilities for self-management51. With the rapid growth of older adults living with HIV/AIDS, traditional family, community, and outpatient service models are increasingly unable to meet their rising healthcare needs. The use of the internet and new media technologies has broken the information cocoon of the older population, creating more opportunities for social interaction. Therefore, healthcare professionals should integrate diverse resources to establish a multi-level, broad-ranging, refined, and personalized strategy for managing social support network management, thereby improving the health and well-being of older adults living with HIV/AIDS while also driving innovation and development in medical services.

In this study, per capita household monthly income under living and working conditions was identified as one of the predictive factors for social frailty in older adults living with HIV/AIDS. A study about the burden due to HIV across the world indicates that a higher proportion of HIV/AIDS patients reside in low- and middle-income countries, with disease burden closely linked to geographical location, healthcare systems, education, and socioeconomic status52. Most of the older adults living with HIV/AIDS in this study come from economically and culturally underdeveloped regions of China. Given the high likelihood of combinations associated with this disease, the costs of treatment and long-term ART monitoring impose a substantial financial burden relative to household income53. Additionally, low health literacy can lead to poor medication adherence, difficulties in health management, and frequent hospitalizations, and it is positively correlated with income levels54. This may explain why healthcare expenditures for older adults living with HIV/AIDS often exceed their financial capacity, further exacerbating economic burdens, lowering socioeconomic status, and increasing the risk of social isolation and negative coping toward the disease52,55. Therefore, it is crucial to consider the economic conditions of this population and facilitate their access to essential medical resources and social services to effectively lower the risk of social frailty.

For the policy environment, this study shows that older adults living with HIV/AIDS who choose the free ART program have a higher risk of social frailty. Since the implementation of China’s “Four Frees and One Care” policy, the free ART drug regimens have been continuously updated and optimized, with the current national free programs primarily consisting of dual or triple therapy regimens56. More than half of the study participants come from low-income rural areas, where financial constraints lead most patients to choose the national free program. Additionally, due to chronic inflammation and immune activation caused by HIV infection, along with aging-related factors, older adults living with HIV/AIDS are at increased risk of non-AIDS-defining illnesses. In order to manage symptoms in patients with multimorbidity and address disease-related challenges, polypharmacy has become increasingly prevalent, with its incidence ranging from 15 to 94% among older adults living with HIV/AIDS57. Polypharmacy increases the medication burden, raising the risk of adverse drug reactions, poor medication adherence, falls, and frailty58,59, thereby exacerbating the risk of social frailty in this population. These findings suggest that healthcare providers should adopt a multidisciplinary approach and leverage telemedicine technologies for dynamic follow-up of patients in free medication programs. By offering online consultations, medication guidance, and health monitoring services, medical professionals can better support these patients. Moreover, integrating resources from health and social welfare departments is crucial to providing older adults living with HIV/AIDS with comprehensive services, including medical care, financial assistance, and psychosocial support.

Research strengths and limitations

This study, based on HEM, used LASSO regression and RF to clarify the severity of social frailty among older adults living with HIV/AIDS and its influencing factors. However, it has several limitations: (1) As a cross-sectional study conducted in a single HIV/AIDS-designated hospital in China, the sample size is relatively small and lacks representativeness, making it difficult to identify the trajectory of social frailty changes in this population. (2) The cross-sectional design only establishes associations between independent and dependent variables without demonstrating causality. (3) The study focuses solely on older adults HIV/AIDS outpatients under follow-up, thus limiting its applicability to hospitalized patients. Therefore, further research is needed to address these limitations and enhance the reliability and generalizability of the findings.

Conclusion

This study reveals a high incidence of social frailty among older adults living with HIV/AIDS, underscoring the need for healthcare professionals to promptly assess social frailty in clinical practice and establish a dynamic monitoring mechanism. Based on the ranking of influencing factors, priority should be given to interventions targeting ADL, exercise, cognitive impairment, depression, and social support. Establishing a social support network for older adults living with HIV/AIDS based on HEM and strengthening health behavior interventions can effectively prevent or mitigate social frailty, thereby promoting healthy aging. Future research should further investigate the effectiveness of multidimensional intervention models to provide stronger evidence for enhancing the comprehensive care system for older adults living with HIV/AIDS.

Supplementary materials

The following supporting information can be downloaded at: https://www.kdocs.cn/l/ctLhFox4lz9o.