AFLE-HIV: developing an age-friendly living environment assessment framework for elderly HIV-infected individuals in China

Abstract

Background

As the population of older adults living with HIV in China continues to expand, a pressing gap persists in the availability of validated instruments that appraise whether residential settings meet the compound demands of aging and chronic HIV management.

Methods

This study presents the AFLE-HIV (Age-Friendly Living Environment for Elderly HIV-Infected Individuals) assessment framework, constructed through a sequential mixed-methods design. Latent Dirichlet Allocation was applied to a corpus of 347 peer-reviewed publications to surface latent environmental themes, yielding an initial set of three primary dimensions and 15 candidate indicators. These candidates were then subjected to two iterative rounds of Delphi consultation involving 22 multidisciplinary experts (response rate: 100%), followed by Analytic Hierarchy Process weight calibration. A field-based pilot across 12 residential communities in four Chinese cities was scored independently by four trained assessors.

Results

The Delphi process consolidated the framework to three primary dimensions and 13 validated indicators with statistically significant inter-expert agreement (Kendall’s W = 0.275, P < 0.001) and satisfactory content validity (item-level CVI \ge 0.82; scale-level CVI/Ave = 0.90). All AHP consistency ratios remained below 0.1. The pilot test yielded satisfactory scale reliability (Cronbach’s α = 0.812) and moderate inter-rater concordance (ICC(2,1) = 0.694).

Conclusions

The resulting AFLE-HIV instrument equips community health workers with a structured protocol for diagnosing environmental shortcomings, provides policymakers with an evidence-based basis for directing renovation investments aligned with the HIV care continuum and community-based service delivery models, and offers urban planners actionable benchmarks for age-friendly residential upgrades targeting this underserved population.

Introduction

The rapid development of antiretroviral therapy (ART) and the expansion of treatment access have profoundly transformed the trajectory of HIV infection, converting what was once a terminal diagnosis into a manageable chronic condition [1, 2]. As a consequence, an increasing number of people living with HIV are surviving into older age, and the intersection of HIV management with age-related physiological decline presents unprecedented challenges for residential environment planning [3, 4]. In China, this demographic shift is particularly pronounced: the proportion of newly reported HIV/AIDS cases among adults aged 60 years and older has risen sharply, from approximately 7.4% in 2010 to nearly 25% by 2023, according to surveillance data from the Chinese Center for Disease Control and Prevention [5, 6]. This epidemiological trend means that a growing cohort of older Chinese adults must simultaneously contend with the clinical demands of lifelong ART adherence, the physiological consequences of immunosenescence, and the pervasive social stigma attached to HIV status [7, 8].

Understanding the residential context of older adults living with HIV in China is essential for framing the environmental dimensions addressed in this study. The majority of Chinese older adults continue to reside in family-based settings, either co-habiting with adult children or living independently in long-established residential communities; institutional care remains a minority arrangement, accommodating fewer than 4% of adults aged 65 and older [9]. For those living with HIV, this predominantly home-based arrangement creates acute privacy tensions: national surveys indicate that a substantial proportion of older adults with HIV have not disclosed their serostatus to family members, and involuntary disclosure through visible medication routines, identifiable clinic visits, or neighbor awareness can trigger familial conflict, social rejection, and intensified psychological distress [4, 10]. These privacy concerns are compounded by a dual stigma—the intersection of ageism and HIV-related discrimination—which, as [11] demonstrated, compounds social exclusion and shapes environmental needs in ways distinct from those of younger HIV-positive populations or the general older population. Consequently, the residential environment for this group must simultaneously address conventional aging-related accessibility demands and HIV-specific requirements for confidentiality, treatment access, and psychosocial support.

The living environment constitutes a critical determinant of health outcomes and quality of life among older adults, as articulated by the ecological theory of aging proposed by [12] and further supported by research on neighborhood-level determinants of functional limitations in aging populations [13]. Central to this theory is the concept of environmental press—the aggregate of physical, interpersonal, and social demands that a setting places on its occupants—which interacts with personal competence to determine adaptive behavior and psychological wellbeing. When individual competence declines due to aging or chronic illness, even moderate environmental press can exceed coping thresholds, resulting in maladaptive outcomes. Complementing this individual-environment lens, Bronfenbrenner’s [14] social-ecological model provides a multilevel framework for understanding how built environments, social networks, and institutional structures jointly shape living conditions for vulnerable populations. For elderly HIV-infected individuals—a term retained in the framework acronym (AFLE-HIV) for consistency, with “older adults” used as the preferred terminology elsewhere in this paper—the environmental press is amplified by disease-specific requirements, including proximity to ART distribution centers, the necessity for privacy to avoid involuntary disclosure of serostatus, and the demand for mental health services that address both age-related cognitive decline and HIV-associated neurocognitive disorders [15, 16]. The World Health Organization (WHO) Age-Friendly Cities and Communities framework has provided a foundational guide for creating supportive environments for older populations [17]; however, this framework does not explicitly incorporate the distinctive requirements of older adults living with HIV, leaving a significant gap in both research and practice [18, 19].

Current assessment tools for age-friendly environments, such as the WHO Age-Friendly Cities Checklist and the AARP Livability Index, focus on the general older population without considering HIV-specific needs including medication privacy, treatment continuity, and destigmatization [20, 21]. Meanwhile, existing HIV care assessment instruments concentrate on individual-level clinical outcomes and quality of life rather than evaluating the residential environment in which these individuals live [22, 23]. Importantly, the residential environment influences HIV-related health outcomes through multiple mechanism pathways: proximity to healthcare facilities and pharmacy services shapes ART adherence [1]; neighborhood safety and walkability affect willingness to attend routine clinical visits [3]; living and neighbourhood environmental quality has been linked to anxiety and psychological distress among older adults [24]; the presence or absence of stigma-protective features determines whether older adults seek social participation and mental health support [10]; and the availability of community-based peer networks mediates loneliness and depressive symptoms, which are themselves predictors of viral non-suppression [25, 26]. This dual gap—in both age-friendly assessment and HIV-environment research—underscores the need for a systematic and standardized approach to evaluating the age-friendliness of living environments specifically tailored to older adults living with HIV.

Objectives

Against this backdrop, the present study sets out to construct the AFLE-HIV (Age-Friendly Living Environment for Elderly HIV-Infected Individuals) assessment framework—a multidimensional instrument that gauges how well residential settings address the intertwined demands of aging and HIV management. By bridging the gap between age-friendly environment research and the HIV care continuum, the framework aims to improve aging-in-place outcomes for older adults living with HIV, identify and mitigate environmental inequities that disproportionately affect this population, and inform integrated care policies that align residential design with chronic disease management. Three sequential objectives guide the research:

1. Theme extraction. Mine a curated corpus of academic publications through unsupervised topic modeling to uncover the latent environmental dimensions pertinent to older adults living with HIV, thereby grounding the initial indicator pool in empirical textual evidence rather than a priori assumptions.

2. Expert-driven refinement. Engage a multidisciplinary panel spanning gerontology, HIV/AIDS prevention, urban planning, public health, and social work in iterative Delphi rounds to screen, merge, and validate candidate indicators until stable consensus is reached.

3. Weight calibration. Derive relative importance scores for each retained indicator through pairwise comparison matrices, producing a hierarchically weighted assessment instrument.

The resulting AFLE-HIV instrument is intended to serve complementary audiences. Community health workers gain a standardized protocol for on-site residential appraisals; municipal planners obtain quantified benchmarks to inform age-friendly renovation priorities; and policymakers receive an evidence base for allocating resources at the intersection of the “Healthy China 2030” strategy and the national response to population aging. For older adults living with HIV and their caregivers, the framework translates abstract environmental standards into concrete, actionable diagnostics that highlight where daily functioning and treatment adherence may be compromised.

Literature review

Age-friendly environments and the WHO framework

The concept of age-friendly environments has gained substantial momentum since the WHO launched its Global Age-Friendly Cities project in 2007, which identified eight domains essential for supporting older adults: outdoor spaces and buildings, transportation, housing, social participation, respect and social inclusion, civic participation and employment, communication and information, and community support and health services [17]. Subsequent initiatives, including the WHO Global Network for Age-Friendly Cities and Communities, have expanded the framework to encompass rural settings and diverse cultural contexts [27, 28]. Research has demonstrated that age-friendly environments are positively associated with physical activity, social engagement, and mental wellbeing among older adults [29–31]. Cross-national indicator development work has further shown that age-friendly neighbourhood features—such as accessible green spaces, proximate healthcare, and safe pedestrian infrastructure—can be operationalized across diverse socioeconomic settings [32].

At the assessment level, several instruments have been developed to measure the age-friendliness of communities and residential settings. The AARP Livability Index evaluates neighborhoods across seven categories including housing, transportation, and health, providing composite scores that facilitate cross-community comparisons [20]. Puxty [21] developed a survey methodology to assess age-friendly community initiatives, focusing on organizational readiness and implementation processes. In China, researchers have adapted the WHO framework to the local context, examining the adequacy of barrier-free infrastructure, public transportation accessibility, and community health service coverage [19, 27, 33]. Black and Oh [18] synthesized progress in age-friendly community assessment, identifying persistent challenges in measurement standardization and cross-cultural validity.

Despite these advances, existing frameworks predominantly address the needs of the general older population. They do not account for the specific vulnerabilities of subpopulations such as older adults living with chronic stigmatized conditions, particularly HIV infection. This limitation restricts the applicability of current assessment tools when evaluating environments for populations that face compounded disadvantages.

Current assessment tools for elderly care environments

Beyond the WHO-derived frameworks, numerous instruments have been developed to evaluate specific aspects of elderly care environments. Environmental assessment tools such as the Elderly-Friendly Hospital Initiative (EFHI) focus on healthcare facility design [28], while housing-focused instruments evaluate barrier-free modifications, assistive technology integration, and universal design compliance [34, 35]. Community-level assessments have also emerged, including age-friendly park surveys [36] and social-ecological-technological evaluation models for age-friendly community retrofits [37]. The Professional Environmental Assessment Protocol (PEAP) and the Therapeutic Environment Screening Survey (TESS) assess institutional care settings, emphasizing factors such as safety, orientation support, and privacy provisions [38].

In the Chinese context, the evaluation of age-friendly residential environments has gained attention alongside the national strategy for actively addressing population aging [9]. Studies have examined the adequacy of elevator installations in multi-story residential buildings [34], the accessibility of community health service centers, and the availability of recreational facilities for older adults. However, these evaluations typically employ ad hoc checklists rather than validated multi-dimensional assessment frameworks, and none incorporates HIV-specific environmental considerations.

To situate the present study within the existing measurement landscape, Table 1 provides a comparative overview of representative age-friendly environment assessment instruments. As the comparison illustrates, no existing tool simultaneously addresses the needs of older adults and the specific environmental requirements arising from HIV infection.

Table 1.

Comparison of representative age-friendly environment assessment instruments

Instrument	Target population	Dim.	HIV-specific	Privacy	Validation
WHO AFC Checklist [17]	General older adults	8	No	No	Expert consensus
AARP Livability Index [20]	General older adults	7	No	No	Survey-based
AFEAT [39]	Older persons (self-report)	5	No	No	EFA, CFA
AARP AFC Survey [40]	Community organizations	4	No	No	Survey metrics
PEAP/TESS [38]	Institutional care settings	6–8	No	Partial	Inter-rater
AFLE-HIV (this study)	Older adults living with HIV	3	Yes	Yes	Delphi, AHP, ICC

Living environment needs of older adults living with HIV

Older adults living with HIV face a constellation of environmental challenges that distinguish them from the general older population. Viewed through the lens of the HIV care continuum—the sequential stages of diagnosis, linkage to care, retention in care, and viral suppression—the residential environment exerts influence at every stage: geographic and transport barriers affect initial linkage and sustained retention, while privacy-compromising housing conditions undermine willingness to engage with services [1, 3]. Moreover, long-term survival on ART is increasingly accompanied by age-associated comorbidities—cardiovascular disease, osteoporosis, diabetes, and HIV-associated neurocognitive disorders—whose management imposes additional environmental demands such as fall-prevention infrastructure, medication storage space, and accessible specialist care facilities [2, 15]. HIV-related stigma further compounds these challenges, as older adults living with HIV experience higher rates of social isolation, depression, and reluctance to disclose their serostatus to neighbors and community service providers [4, 10, 11]. Additionally, the intersection of HIV-associated neurocognitive disorders with age-related cognitive decline may necessitate enhanced environmental cues, simplified wayfinding systems, and assistive technologies that are not typically included in standard age-friendly assessments [2, 15].

Research on the housing and residential needs of people living with HIV has primarily focused on homelessness and housing instability among younger populations, with limited attention to the age-specific environmental requirements of older adults [16, 41]. A systematic review of HIV, aging, and housing from a North American perspective similarly concluded that age-appropriate residential supports for older adults with HIV remain critically underexplored [42]. Gervolino et al. [4] examined the role of social support networks among older adults living with HIV, finding that environmental factors such as neighborhood safety and access to peer support groups significantly predicted wellbeing outcomes. Loneliness and social isolation have been identified as particularly strong predictors of adverse health outcomes in this population [25, 26]. Meanwhile, studies from sub-Saharan Africa have explored the intersection of HIV and aging in resource-limited settings [1], and multilevel analyses have identified environmental, institutional, and interpersonal factors that influence geriatric care requirements for older adults living with HIV [43], but dedicated assessment frameworks for residential environments remain absent.

These gaps highlight the necessity of developing a dedicated assessment framework that integrates age-friendly environmental principles with the specific needs arising from HIV infection. Such a framework should encompass physical environment adaptations, social and psychological support structures, and healthcare and privacy protections, providing a holistic instrument for evaluation.

Methods

Study design

This study follows a four-phase sequential design to construct and preliminarily validate an assessment framework for age-friendly residential environments serving older adults living with HIV in China. In Phase 1, unsupervised topic modeling is applied to an academic literature corpus to surface latent environmental themes and generate an initial pool of candidate indicators. Phase 2 subjects these candidates to iterative expert consultation (Delphi method) for screening, consolidation, and validation. Phase 3 converts the validated indicator set into a hierarchically weighted instrument through pairwise comparison matrices. Phase 4 field-tests the instrument across multiple urban sites to obtain preliminary reliability evidence. The overall workflow is depicted in Fig. 1.

Fig. 1.

Research flowchart illustrating the AFLE-HIV framework construction process

Construction of the initial indicator system

Data collection

The analytical corpus was assembled from peer-reviewed literature addressing age-supportive residential settings and the environmental circumstances of older adults living with HIV. A two-stage bibliographic search was executed between March and May 2025 across PubMed, Web of Science, and the China National Knowledge Infrastructure (CNKI), spanning publications dated January 2000 through March 2025. Stage 1 targeted the broader age-friendly environment literature with the query: (“age-friendly” OR “aging-friendly” OR “elderly-friendly” OR “livability”) AND (“environment” OR “housing” OR “community” OR “residential”) AND (“older adults” OR “elderly” OR “aging”). Preliminary screening revealed that additional HIV-specific residential literature was required; Stage 2 therefore employed a refined query: (“HIV” OR “AIDS” OR “antiretroviral”) AND (“elderly” OR “older adults” OR “aging”) AND (“housing” OR “living environment” OR “community” OR “residential” OR “care environment”).

Table 2 summarizes the eligibility criteria applied during study selection.

Table 2.

Inclusion criteria and exclusion criteria

Category	Criteria
Inclusion	The research is related to age-friendly environments, elderly care environments, or HIV and residential settings
	The research papers are written in English or Chinese
	The research papers have been published in peer-reviewed journals
	The full text of the publication is available
Exclusion	The research is not related to age-friendly environments or HIV residential needs
	The research papers are editorials, conference abstracts only, or book reviews
	The research papers are not published in peer-reviewed journals
	The full text of the publication is not available

Retrieved records were consolidated in EndNote X9 for deduplication. A two-phase screening protocol was carried out independently by the author and a trained research assistant. During Phase 1, titles and abstracts were evaluated against the eligibility criteria; screening decisions were logged in a shared spreadsheet and discrepancies resolved through discussion. Phase 2 involved full-text appraisal of the retained studies to finalize the inclusion set. The selection workflow is illustrated in the PRISMA flow diagram (Fig. 2).

Fig. 2.

PRISMA flow diagram for the literature identification, screening, and inclusion process

From the 347 included articles, the author and research assistant jointly extracted passages pertaining to “age-friendly environmental attributes for older adults living with HIV” following a standardized extraction protocol. Each extracted segment was cross-checked collaboratively: off-topic material was discarded, misclassified passages were reassigned, and gaps were filled where the original text omitted relevant detail. The curated text corpus served as the input for the subsequent topic modeling phase.

Topic modeling

Latent Dirichlet Allocation (LDA) treats each document as a probabilistic mixture of latent topics, where each topic is itself a distribution over vocabulary terms; parameter estimation proceeds via hierarchical Bayesian inference [44]. LDA has been widely adopted for automated theme discovery in domains ranging from sentiment analysis to evaluation framework construction [45, 46]. Its capacity to surface thematic structure directly from textual data, without requiring predefined categories, made it well suited to the exploratory nature of the present study.

Corpus preparation involved lemmatization and stop-word removal. The LDA model was fitted using the Gensim library (version 4.3, Python), with Dirichlet priors set to α = “auto” and η = “auto” for asymmetric optimization, trained over 500 iterations (random seed = 42). To identify the most informative topic granularity, both perplexity and the coherence metric were computed for candidate topic counts ranging from 5 to 45. The elbow of the perplexity curve—the point at which further increases in topic count ceased to reduce perplexity appreciably—was cross-referenced with the peak coherence value to select the final topic number, following the joint optimization strategy recommended for LDA model selection [44, 47]. The resulting topic clusters were then interpreted in light of domain-specific literature and established assessment frameworks to draft an initial set of environmental indicators for older adults living with HIV.

Delphi consultation

Structured expert consultation was conducted following the Delphi protocol, a multi-round survey technique originally devised at the RAND Corporation during the 1950 s to elicit and converge group judgment under conditions of uncertainty [48]. The technique has been widely adopted in public health instrument development, including questionnaire construction for fall prevention among older adults in China [49]. The procedure comprised problem definition, panel assembly, two iterative questionnaire rounds with controlled feedback, statistical aggregation, and consensus determination.

Questionnaire design

The consultation instrument contained three modules. Module 1 outlined the study rationale and the intended use of the assessment framework. Module 2 listed the candidate indicators and asked panelists to rate each on a five-point Likert scale (1 = strongly disagree to 5 = strongly agree) in terms of relevance to the assessment of age-friendly living environments for older adults living with HIV; open-ended comment fields accompanied every indicator to capture qualitative suggestions for rewording, merging, or adding indicators. Module 3 gathered panelist demographics and self-assessed familiarity with the consultation topics, together with the judgmental bases (theoretical knowledge, field experience, published evidence, and professional intuition) underpinning their ratings.

Panel composition

Panelists were recruited according to five eligibility criteria: (1) documented expertise in at least one of the following domains—gerontology or geriatric medicine, HIV/AIDS prevention and control, urban planning or architecture, public health, or social work; (2) a minimum of ten years of professional or research experience in the relevant domain; (3) an associate senior professional title or higher; (4) prior familiarity with the Delphi technique; and (5) willingness to complete both consultation rounds within the stipulated timeframe. A total of 30 potential panelists were contacted via email invitation; 22 agreed to participate (response rate 73.3%), while 8 declined owing to scheduling constraints. No systematic differences in disciplinary background or seniority were observed between respondents and non-respondents.

Indicator weight derivation

Relative importance scores were obtained through the Analytic Hierarchy Process (AHP), a multi-criteria decision-making technique that translates qualitative expert judgments into ratio-scale weights via pairwise comparison matrices [50]. Coupling Delphi-validated indicators with AHP-derived weights is a well-established practice in indicator system research: the former ensures content validity and expert endorsement, while the latter imposes a transparent, mathematically consistent prioritization [51, 52].

Using the indicator set confirmed in the second Delphi round, a three-level hierarchical model was constructed (goal → primary dimensions → secondary indicators). Pairwise comparison matrices were populated on Saaty’s 1–9 fundamental scale, with scale values mapped from the absolute differences between Delphi mean importance ratings as follows: a difference of 0 → scale value 1 (equal importance); 0.1–0.3 → 2; 0.3–0.5 → 3; 0.5–0.7 → 4; 0.7–1.0 → 5; 1.0–1.5 → 6; 1.5–2.0 → 7; 2.0–2.5 → 8; >2.5 → 9. Each matrix was subjected to a consistency check; only matrices with a consistency ratio (CR) below 0.1 were accepted. Normalized eigenvector weights were then extracted.

Pilot test

To provide initial evidence for the applicability and reliability of the AFLE-HIV framework, a pilot test was conducted in four Chinese cities: Beijing, Shanghai, Kunming, and Chengdu. These cities were selected because they represent different geographic regions and have relatively concentrated populations of older adults living with HIV. In each city, three residential communities with varying levels of age-friendly infrastructure were identified, yielding 12 evaluation sites in total.

Four community health assessors with experience in elderly care services and HIV/AIDS prevention (ranging from 6 to 18 years of professional experience) were invited to independently evaluate these communities using the AFLE-HIV framework. Prior to the assessment, all four assessors participated in a one-day calibration training session conducted by the author, which included a detailed review of each indicator’s operational definition, scoring rubric, and illustrative examples. The assessors were blinded to each other’s scores throughout the evaluation process. For each community, the assessors conducted independent on-site inspections lasting approximately 2–3 hours and used a five-point Likert scale to score all framework items. Each assessor visited all 12 communities over a two-week period. Assessors were also asked to provide brief justifications for their scores, ensuring that evaluations were grounded in explicit judgment criteria and enabling the author to examine the consistency with which evaluators interpreted each indicator.

Statistical analysis

Indicator retention during the Delphi rounds was governed by three concurrent thresholds applied to each candidate, following established Delphi methodology [53, 54]: (i) arithmetic mean of importance ratings exceeding 3.5, (ii) proportion of panelists assigning the maximum score (full score ratio) above 0.2, and (iii) coefficient of variation (CV) below 0.25. These cut-off values are consistent with conventional standards widely adopted in health-related Delphi studies [54]. Expert engagement was quantified by the valid questionnaire recovery rate (positive coefficient). Professional credibility was captured by the composite reliability index , where denotes self-reported topic familiarity and denotes the judgmental basis score; values of are conventionally taken to reflect adequate expert credibility [53]. Inter-expert agreement was assessed through Kendall’s coefficient of concordance (W), with statistical significance set at P < 0.05.

To quantify content validity, the item-level content validity index (I-CVI) was computed for each indicator as the proportion of experts rating that item 4 or 5 on the relevance scale; the scale-level content validity index (S-CVI/Ave) was calculated as the mean of all I-CVI values [55, 56]. An I-CVI \ge 0.78 and S-CVI/Ave \ge 0.90 were adopted as acceptability thresholds [55].

For the pilot test, inter-rater concordance was estimated using the two-way random-effects, single-measures intraclass correlation coefficient, ICC(2,1). Following the classification proposed by [57] and endorsed by [58], ICC values below 0.50 indicate poor reliability, 0.50–0.75 moderate reliability, 0.75–0.90 good reliability, and above 0.90 excellent reliability. Internal consistency of the framework was evaluated with Cronbach’s α.

All Delphi-related indices and reliability statistics were computed in SPSS 27; AHP weight calculations were performed in yaahp 10.1.

Results

Data visualization and topic modeling

Data visualization and analysis

The bibliographic search and screening pipeline retained 347 articles for inclusion. A total of 1,530 records were identified across both search stages (PubMed: 490; Web of Science: 671; CNKI: 369); after removing 387 duplicates, 1,143 unique records underwent title and abstract screening; 554 full-text articles were assessed for eligibility, of which 347 met all inclusion criteria (see Fig. 2). Substantive content was extracted from each publication to assemble the analytical corpus. Table 3 lists the principal journal sources, showing that the included literature spans gerontology, public health, HIV care, and environmental planning.

Table 3.

Principal journal sources of the included literature

Literature source	Number	Literature source	Number
BMC Public Health	28	Aging & Mental Health	9
AIDS Care	24	AIDS and Behavior	8
The Gerontologist	19	Journal of Aging & Social Policy	7
Int. J. Environ. Res. Public Health	16	Journal of Urban Planning	6
JMIR Aging	12	Others	218

The corpus underwent stop-word filtering and lemmatization before being fed into the model. A term-frequency word cloud (Fig. 3) reveals that “elderly,” “HIV,” “community,” “health,” “housing,” and “age-friendly” dominate the vocabulary, confirming the corpus’s focus on the nexus of aging-related environmental factors and HIV care. Secondary high-frequency terms—“stigma,” “support,” “accessibility,” “privacy”—underscore the multifaceted nature of residential considerations for this population.

Fig. 3.

Word cloud of high-frequency terms from the literature corpus on age-friendly living environments for older adults living with HIV

Topic modeling results

Selecting an appropriate topic count is essential for balancing model granularity against interpretability. The perplexity curve (Fig. 4) showed a marked deceleration in perplexity reduction at K = 25; values remained largely stable between 25 and 27 topics before rising—a pattern indicative of overfitting beyond that range. The 25-topic configuration also achieved the highest coherence score (0.412) across the tested spectrum, and clustering visualization confirmed well-separated, semantically distinct topic groups. On the basis of these converging diagnostics, K = 25 was adopted.

Fig. 4.

Perplexity curve of the LDA model across different topic numbers, with the optimal number identified at K = 25

LDA topic clustering was performed to obtain 25 topic generation results. Analysis of the high-probability terms associated with each topic revealed the thematic focus of each cluster. The complete list of 25 topics, their representative keywords, and the consolidation pathway to the initial indicator set are presented in Table 4.

Table 4.

LDA topic labels, representative keywords, and consolidation into initial indicators

Topic	Label	Top-5 keywords	Initial indicator
T1	Home modification	barrier-free, ramp, handrail, renovation, accessibility	Barrier-free Design
T21	Fall prevention	fall, slip-resistant, lighting, grab-bar, safety
T2	Healthcare access	clinic, proximity, healthcare, distance, facility	Medical Facility Access.
T23	Healthcare coord.	referral, coordination, integrated, specialist, geriatric
T8	Public transport	bus, transit, route, mobility, transport	Transportation Infra.
T15	Transport infra.	road, sidewalk, crosswalk, traffic, pedestrian
T5	Daily living	grocery, pharmacy, banking, convenience, daily	Daily Living Convenience
T12	Public space	park, bench, shade, recreation, outdoor	Public Space Age-friend.
T11	Smart health tech	telemedicine, wearable, sensor, monitoring, device	Assistive Tech. Avail.
T20	Remote monitoring	telehealth, remote, digital, app, e-health
T4	Community engage.	participation, activity, volunteer, social, inclusion	Community Social Incl.
T13	Social inclusion	integration, belonging, neighborhood, interaction, network
T7	Mental health	counseling, depression, anxiety, psychological, therapy	Psych. Counseling
T24	Depression interv.	wellbeing, screening, intervention, resilience, coping
T6	HIV stigma	stigma, discrimination, prejudice, disclosure, rejection	Anti-stigma Environ.
T17	Stigma coping	empowerment, advocacy, education, awareness, rights
T10	Peer support	peer, self-help, mutual-aid, group, solidarity	Social Connectedness
T9	Health equity	equity, disparity, allocation, universal, coverage	Healthcare Equity
T19	Service distrib.	access, rural, underserved, gap, inequality
T3	ART adherence	adherence, antiretroviral, regimen, viral-load, CD4	HIV Treatment Contin.
T14	Medication mgmt.	pharmacy, prescription, supply, continuity, refill
T18	Health privacy	privacy, confidentiality, data, record, disclosure	Health Privacy Protect.
T16	Service quality	quality, standard, evaluation, inspection, consistency	Service Quality Consist.
T22	Cognitive support	wayfinding, signage, orientation, cognitive, dementia	Cognitive & Emerg.
T25	Emergency resp.	emergency, alarm, response, evacuation, preparedness

Indicators removed during Delphi; Peer Support Networks and Psychological

Counseling Accessibility were added by expert recommendation (see Establishment of the indicator system section)

Overall, the topics covered multiple dimensions relevant to age-friendly environments for older adults living with HIV. A substantial portion focused on physical environment attributes, including barrier-free design and home modifications (T1, T21), medical facility accessibility (T2, T23), and transportation infrastructure (T8, T15). Another segment emphasized social and psychological dimensions, such as community engagement and social inclusion (T4, T13), mental health services (T7, T24), and stigma-related barriers (T6, T17). Additional topics addressed healthcare and privacy concerns, including treatment continuity (T3, T14), health information privacy (T18), and service equity (T9, T19). These interpreted topics provided the semantic foundation for developing the initial assessment framework.

Framework construction

The 25 LDA topics were synthesized into an initial indicator framework through a three-step procedure. First, each topic was assigned a concise descriptive label derived from its highest-probability keywords. Second, semantically adjacent or overlapping topics were merged into consolidated conceptual categories, producing 15 secondary indicators. For instance, T1 (Home Modification and Barrier-free Living) and T21 (Fall Prevention Design) were consolidated under “Barrier-free Design,” while T2 (Community Healthcare Access) and T23 (Healthcare Service Coordination) were grouped as “Medical Facility Accessibility.” Third, drawing on the WHO Age-Friendly Cities framework [17, 18], HIV-specific environmental evidence [3, 4], and ecological aging theory [12], the 15 indicators were organized into three primary dimensions: Physical Environment Adaptability (aligned with “environmental press” in Lawton’s ecological model), Social and Psychological Support (corresponding to the microsystem level in Bronfenbrenner’s social-ecological framework), and Healthcare and Privacy Protection (capturing institutional-level determinants of service access and confidentiality). Classification decisions were made jointly by the author and the research assistant, with disagreements resolved through deliberation. The resulting draft framework comprised 3 primary dimensions and 15 secondary indicators.

Expert consultation and indicator system development

Expert characteristics

Twenty-two panelists affiliated with 16 universities, medical institutions, and research organizations participated via email. Their disciplinary backgrounds spanned gerontology and geriatric medicine, HIV/AIDS prevention and control, urban planning and architecture, public health, and social work. Demographic details are summarized in Table 5.

Table 5.

Expert demographic information

Category	Characteristics	Number	Percentage (%)
Age	30–40	8	36.4
	41–50	10	45.5
	51–60	4	18.2
Gender	Male	9	40.9
	Female	13	59.1
Degree	Doctorate	14	63.6
	Master’s	7	31.8
	Bachelor’s	1	4.5
Professional title	Senior	6	27.3
	Associate senior	10	45.5
	Middle	5	22.7
	Primary	1	4.5
Professional field	Gerontology/Geriatric medicine	5	22.7
	HIV/AIDS prevention and control	5	22.7
	Urban planning/Architecture	4	18.2
	Public health	4	18.2
	Social work	4	18.2
Years of experience	10–15	9	40.9
	16–20	8	36.4
	> 20	5	22.7

Expert positivity coefficient, authority coefficient, and coordination coefficient

Two Delphi rounds were administered, each distributing 22 questionnaires. All questionnaires were returned in both rounds (recovery rate 100%). Modification suggestions were offered by 50% (11/22) of panelists in Round 1 and 36.4% (8/22) in Round 2. Authority coefficients rose from 0.85 (Round 1) to 0.89 (Round 2), and Kendall’s W increased from 0.263 to 0.275 (both P < 0.001), signaling progressively stronger inter-expert concordance. The accompanying decline in (from 99.41 to 93.34) is attributable to the smaller number of indicators evaluated in Round 2 (14 versus 15), since the chi-square statistic in the Kendall W test is a function of the degrees of freedom (Table 6).

Table 6.

Expert authority coefficient and coordination coefficient

Round	Authority coefficient	Coordination coefficient
		W		P
First	0.85	0.263	99.41	< 0.001
Second	0.89	0.275	93.34	< 0.001

Establishment of the indicator system

The draft framework entering the Delphi process comprised 3 primary dimensions and 15 secondary indicators, derived from the LDA analysis, supporting literature, and the judgment of the author and research assistant. Round 1 led to the removal of two indicators that fell below the retention thresholds (mean < 3.5 or CV > 0.25): Transportation Infrastructure (mean = 3.28, CV = 0.278) and Cognitive and Emergency Support (mean = 3.14, CV = 0.312). A third indicator, Social Connectedness (mean = 3.41, CV = 0.262), was also removed after panelists noted that its scope overlapped substantially with Community Social Inclusion. Two new indicators were proposed by panelists and endorsed by a majority: Psychological Counseling Accessibility (panelists argued that community-based counseling access for HIV-related mental health warranted a dedicated indicator rather than being subsumed under the broader Psychological Counseling Access derived from LDA) and Peer Support Networks (panelists argued that organized peer mutual-aid constituted a distinct social resource warranting separate assessment beyond the broader Social Connectedness captured by the initial indicator). Definitional or scope revisions were applied to six existing indicators—yielding 14 secondary indicators for Round 2. Round 2 resulted in the removal of one further indicator (Psychological Counseling Access, originally derived from LDA topics T7 and T24, fell below the CV threshold despite a revised definition, CV = 0.258; its content was deemed adequately captured by the new Psychological Counseling Accessibility indicator) and refinement of four operational definitions. The finalized framework retains three primary dimensions and thirteen secondary indicators:

• A: Physical Environment Adaptability — encompassing Barrier-free Design (A-1), Medical Facility Accessibility (A-2), Daily Living Convenience (A-3), Public Space Age-friendliness (A-4), and Assistive Technology Availability (A-5).

• B: Social and Psychological Support — encompassing Community Social Inclusion (B-1), Psychological Counseling Accessibility (B-2), Anti-stigma Environment (B-3), and Peer Support Networks (B-4).

• C: Healthcare and Privacy Protection — encompassing Healthcare Equity (C-1), HIV Treatment Continuity (C-2), Health Privacy Protection (C-3), and Service Quality Consistency (C-4).

The Delphi results for each retained indicator in the final round are presented in Table 7, showing that all 13 indicators met the retention criteria (mean > 3.5, full score ratio > 0.2, CV < 0.25).

Table 7.

Delphi consultation results and content validity indices for each retained indicator (final round)

Code	Secondary indicator	Mean	Full score ratio	CV	I-CVI
A-1	Barrier-free Design	4.68	0.77	0.103	0.95
A-2	Medical Facility Accessibility	4.82	0.86	0.079	1.00
A-3	Daily Living Convenience	4.36	0.55	0.142	0.91
A-4	Public Space Age-friendliness	4.18	0.41	0.168	0.86
A-5	Assistive Technology Availability	4.50	0.64	0.128	0.91
B-1	Community Social Inclusion	4.41	0.59	0.137	0.91
B-2	Psychological Counseling Accessibility	3.95	0.32	0.198	0.82
B-3	Anti-stigma Environment	4.23	0.45	0.156	0.86
B-4	Peer Support Networks	4.32	0.50	0.147	0.91
C-1	Healthcare Equity	4.14	0.36	0.172	0.86
C-2	HIV Treatment Continuity	4.73	0.82	0.091	1.00
C-3	Health Privacy Protection	4.59	0.73	0.112	0.95
C-4	Service Quality Consistency	4.05	0.32	0.185	0.82

I-CVI item-level content validity index (proportion of experts rating \ge 4); S-CVI/Ave = 0.90

Content validity analysis yielded I-CVI values ranging from 0.82 to 1.00 across the 13 retained indicators (Table 7), with all values exceeding the 0.78 acceptability threshold recommended by [55]. The scale-level content validity index (S-CVI/Ave) was 0.90, meeting the benchmark of \ge 0.90 [55, 56]. These results provide quantitative evidence that the retained indicators are judged as relevant to the assessment domain by the expert panel.

Weight determination of each indicator

Pairwise comparison matrices were constructed for the finalized indicator set and processed through the AHP. Every matrix satisfied the consistency requirement (CR < 0.1), confirming that the derived weights are internally coherent. The complete AFLE-HIV framework with normalized and combined weights appears in Table 8.

Table 8.

AFLE-HIV assessment framework with indicator weights (sorted by weight within each dimension)

Primary dimension	Secondary indicator	Weight	Combined weight
A Physical Environment Adaptability (0.539)	A-2 Medical Facility Accessibility	0.3704	0.1996
	A-1 Barrier-free Design	0.2466	0.1329
	A-5 Assistive Technology Availability	0.1893	0.1020
	A-3 Daily Living Convenience	0.1203	0.0649
	A-4 Public Space Age-friendliness	0.0733	0.0395
B Social & Psychological Support (0.1638)	B-1 Community Social Inclusion	0.4262	0.0698
	B-4 Peer Support Networks	0.2827	0.0463
	B-3 Anti-stigma Environment	0.2032	0.0333
	B-2 Psychological Counseling Accessibility	0.0879	0.0144
C Healthcare & Privacy Protection (0.2973)	C-2 HIV Treatment Continuity	0.4351	0.1293
	C-3 Health Privacy Protection	0.3092	0.0919
	C-1 Healthcare Equity	0.1501	0.0446
	C-4 Service Quality Consistency	0.1056	0.0314

Among the three primary dimensions, Physical Environment Adaptability received the highest weight (0.539), followed by Healthcare and Privacy Protection (0.2973) and Social and Psychological Support (0.1638). Within Physical Environment Adaptability, Medical Facility Accessibility (A-2) obtained the highest secondary weight (0.3704), reflecting experts’ consensus that proximity to and quality of healthcare facilities is the most critical physical environmental factor for older adults living with HIV.

Within Healthcare and Privacy Protection, HIV Treatment Continuity (C-2) held the dominant secondary weight (0.4351), underscoring the centrality of uninterrupted ART access. Health Privacy Protection (C-3) received the second-highest weight (0.3092) in this dimension, reflecting the importance of confidentiality in HIV care environments.

Within Social and Psychological Support, Community Social Inclusion (B-1) received the highest secondary weight (0.4262), followed by Peer Support Networks (B-4, 0.2827), highlighting the critical role of social connectedness for this population. The combined weight analysis revealed that Medical Facility Accessibility (0.1996), Barrier-free Design (0.1329), and HIV Treatment Continuity (0.1293) constitute the three most influential individual indicators in the AFLE-HIV framework.

To facilitate transparent interpretation and replication, Table 9 presents the operational definition, assessment method, and scoring anchors for each of the 13 secondary indicators. For each indicator, assessors assign a score on a five-point Likert scale (1 = very poor to 5 = excellent) based on on-site observation, facility inspection, and review of community records. Dimension scores are computed as the weighted sum of constituent indicator scores, and the overall AFLE-HIV score is the sum of the three dimension scores: , where is the dimension weight, is the indicator weight within dimension d, and is the assessor’s rating for indicator i (Table 10).

Table 9.

Operational definitions and scoring anchors for AFLE-HIV secondary indicators (Part 1: Dimensions A and B)

Code	Indicator	Operational definition	Scoring anchors (1=very poor; 5=excellent)
A-1	Barrier-free Design	Ramps, handrails, non-slip flooring, adequate lighting, and wheelchair-accessible pathways in residential buildings.	1: No features; 3: Main entrance accessible but interior incomplete; 5: Comprehensive design throughout.
A-2	Medical Facility Access.	Proximity to healthcare facilities offering geriatric-HIV care, measured by travel distance and transport.	1: >5 km, no transport; 3: 2–5 km with transport; 5: <1 km with convenient transport.
A-3	Daily Living Conv.	Availability of essential daily services (grocery, pharmacy, bank) within walking distance.	1: None within 1 km; 3: Most within 1–2 km; 5: All within 500 m.
A-4	Public Space Age-fr.	Quality of public open spaces with age-friendly features (seating, shade, flat pathways).	1: No accessible spaces; 3: Spaces exist but lack features; 5: Well-maintained, fully accessible.
A-5	Assistive Tech.	Availability of emergency call systems, wearable monitors, and telemedicine platforms.	1: None available; 3: Basic emergency call; 5: Comprehensive smart health monitoring.
B-1	Community Social Incl.	Community programs promoting social participation of older adults including stigmatized groups.	1: No programs; 3: General programs without inclusion focus; 5: Tailored inclusive programs.
B-2	Psych. Counseling	Community-based counseling with geriatric and HIV-related mental health expertise.	1: No services; 3: General counseling without specialization; 5: Specialized counseling accessible.
B-3	Anti-stigma Env.	Anti-stigma initiatives, education, and policies protecting from HIV and age discrimination.	1: No measures; 3: Awareness materials only; 5: Active programs with training and enforcement.
B-4	Peer Support Net.	Organized peer support groups for older adults living with HIV in the community.	1: None; 3: Informal connections only; 5: Regular facilitated meetings with stable membership.

Table 10.

Operational definitions and scoring anchors for AFLE-HIV secondary indicators (Part 2: Dimension C)

Code	Indicator	Operational definition	Scoring anchors (1 = very poor; 5 = excellent)
C-1	Healthcare Equity	Degree to which older adults living with HIV receive equitable access to healthcare services compared with the general older population, without discrimination based on HIV status.	1: Documented access barriers based on HIV status; 3: Formal equity policies exist but implementation is inconsistent; 5: Full equitable access with monitoring mechanisms.
C-2	HIV Treatment Continuity	Reliability and continuity of ART supply, clinical monitoring, and follow-up care, including uninterrupted access to antiretroviral medications and regular viral load testing.	1: Frequent ART supply interruptions and no follow-up system; 3: Generally stable supply with occasional gaps; 5: Uninterrupted ART supply with systematic follow-up and viral load monitoring.
C-3	Health Privacy Protection	Adequacy of measures to protect the confidentiality of HIV status, including secure medical records, private consultation spaces, and discreet medication distribution.	1: No privacy protections; HIV status easily identifiable; 3: Basic privacy measures but structural gaps remain; 5: Comprehensive privacy protections with private consultation rooms and secure records.
C-4	Service Quality Consistency	Consistency of service quality across different time periods and service providers within the residential community’s healthcare and support systems.	1: Highly variable service quality; 3: Moderate consistency with documented quality standards; 5: High consistency verified through regular quality audits.

Results of pilot test

Four trained assessors independently scored 12 residential communities across four Chinese cities. To compute internal consistency, item scores were averaged across the four raters, yielding one composite score per item per community. Dimension-level and overall Cronbach’s α values are reported in Table 11; the overall coefficient reached 0.812, indicating satisfactory scale reliability.

Table 11.

Internal consistency of the AFLE-HIV framework

Dimension	Measuring items	Mean ± SD	Cronbach’s α
A	5 (A-1 – A-5)	3.517 ± 0.423	0.853
B	4 (B-1 – B-4)	3.842 ± 0.268	0.781
C	4 (C-1 – C-4)	4.725 ± 0.254	0.762
Total	13	3.896 ± 0.291	0.812

The ICC(2,1) estimate was 0.694, which falls within the moderate reliability range (0.50–0.75) according to the classification of [57, 58]. These pilot findings should be regarded as preliminary given the constrained sample (four assessors, 12 communities); definitive reliability benchmarks will require larger and more geographically diverse validation studies. Notably, dimension C (Healthcare and Privacy Protection) exhibited a mean score of 4.725 on the five-point scale, raising the possibility of a ceiling effect that may attenuate variability and, consequently, the reliability estimate for that dimension. Notwithstanding these caveats, the overall agreement level is comparable to values reported in analogous environmental assessment validation studies and suggests that the indicator definitions are sufficiently precise to support consistent cross-evaluator interpretation.

Discussion

Main findings

Grounded in the ecological theory of aging [12] and the documented environmental vulnerabilities of older adults living with HIV, this study produced the AFLE-HIV assessment framework. The instrument is organized around three primary dimensions—Physical Environment Adaptability, Social and Psychological Support, and Healthcare and Privacy Protection—operationalized through 13 secondary indicators. Each indicator captures a specific facet of the dual burden imposed by age-related functional decline and HIV-specific care requirements, yielding a structured yet flexible evaluation protocol.

Physical Environment Adaptability, which received the highest primary weight (0.539), reflects the fundamental importance of the built environment in supporting daily functioning and safety for older adults. The dominance of Medical Facility Accessibility (A-2, combined weight 0.1996) within this dimension aligns with findings from [1], who emphasized that proximity to HIV care facilities is a primary determinant of treatment adherence among older adults. Similarly, Barrier-free Design (A-1, combined weight 0.1329) corresponds to the established relationship between environmental barriers and fall risk, which is compounded in older adults living with HIV by antiretroviral therapy-associated bone density reduction and peripheral neuropathy [2, 3]. Assistive Technology Availability (A-5, combined weight 0.1020) reflects the growing recognition that smart health monitoring devices and telemedicine platforms can bridge gaps in care access, particularly for older adults who face mobility limitations or stigma-related reluctance to visit healthcare facilities in person [59, 60].

The Healthcare and Privacy Protection dimension, weighted at 0.2973, addresses concerns that are especially salient for older adults living with HIV. HIV Treatment Continuity (C-2, combined weight 0.1293) emerged as the most critical secondary indicator within this dimension, consistent with the clinical evidence that ART interruptions are associated with accelerated immunological decline, viral rebound, and increased mortality risk [1, 2]. Health Privacy Protection (C-3, combined weight 0.0919) reflects the persistent reality of HIV stigma, where involuntary disclosure of serostatus through environmental cues—such as visible medication storage, identifiable clinic visits, or neighbor awareness—can lead to social exclusion and psychological distress [4, 10]. The inclusion of Healthcare Equity (C-1) and Service Quality Consistency (C-4) ensures that the framework captures systemic factors beyond individual-level care.

Social and Psychological Support, although receiving the lowest primary weight (0.1638), addresses dimensions that are critical for the overall wellbeing of older adults living with HIV. Community Social Inclusion (B-1, the highest secondary weight at 0.4262 within this dimension) reflects the evidence that social isolation is a major risk factor for depression and non-adherence among older adults living with HIV [4, 16]. Recent evidence from China further confirms that psychological resilience and social support are key determinants of HIV-related stress among this population [61]. Anti-stigma Environment (B-3) and Peer Support Networks (B-4) address the specific social barriers that distinguish older adults living with HIV from the general older population, supporting the theoretical framework of HIV stigma as a multilevel phenomenon operating across individual, interpersonal, and structural domains [8, 10].

Notably, the concentration of weight on Physical Environment Adaptability (53.9%) represents a marked departure from the balanced, equal-weighting approach implicit in the WHO Age-Friendly Cities framework, which distributes attention approximately equally across its eight domains [17]. This disparity suggests that for populations with specific chronic health conditions such as HIV, the physical environment—particularly healthcare infrastructure proximity—assumes disproportionate importance relative to general age-friendly frameworks. This finding reinforces the argument that population-specific assessment tools are necessary to capture the distinct prioritization patterns of vulnerable subgroups.

Comparison with existing instruments

A comparison of AFLE-HIV with established age-friendly assessment instruments (Table 1) reveals both convergences and important distinctions. Like the Age-Friendly Environment Assessment Tool (AFEAT) developed by [39], AFLE-HIV evaluates the physical environment from the perspective of end users and incorporates subjective relevance ratings. However, AFEAT was designed as a self-report tool for the general older population and does not include indicators related to HIV treatment continuity, health privacy protection, or anti-stigma environments—three dimensions that emerged as central to the AFLE-HIV framework. Similarly, the AARP Age-Friendly Community Survey, whose measurement indicators were examined by [40], captures community-level organizational readiness but does not assess disease-specific environmental features. A recent systematic review by [62] evaluated measurement instruments assessing the needs, wellbeing, and living environments of older adults and confirmed that none of the reviewed instruments incorporated HIV-specific items, further corroborating the novelty of the present contribution.

A pertinent methodological consideration concerns the placement of Medical Facility Accessibility (A-2) under the Physical Environment Adaptability dimension, given its apparent conceptual overlap with the Healthcare and Privacy Protection dimension. This classification reflects a deliberate theoretical distinction grounded in Lawton’s ecological model [12]: A-2 captures the physical accessibility of healthcare infrastructure—geographic proximity, transport connectivity, and spatial reachability—which constitutes a form of environmental press. By contrast, indicators within dimension C (e.g., Healthcare Equity, Service Quality Consistency) assess the quality, equity, and continuity of healthcare services once physical access has been achieved. This distinction parallels the well-established differentiation in health services research between spatial access and effective access [3]. The AHP-derived weights further support this separation: A-2 and C-2 received distinct prioritization profiles from the expert panel, suggesting that experts conceptualized them as complementary rather than redundant constructs.

Theoretical implications

This study advances the theoretical landscape in three respects. First, the AFLE-HIV framework constitutes the first purpose-built assessment instrument for appraising age-friendly living environments from the standpoint of older adults living with HIV. A targeted search of PubMed, Web of Science, and CNKI using the terms “age-friendly” AND “HIV” AND “assessment framework/instrument/tool” returned no pre-existing validated instruments for this population, corroborating the novelty of the contribution. By fusing the ecological theory of aging [12] with the HIV stigma framework, the study extends the conceptual reach of age-friendly environment research to populations burdened by chronic stigmatized conditions—a gap flagged by [18], who observed that prevailing frameworks inadequately account for subpopulation heterogeneity.

Second, the sequential pipeline—unsupervised topic modeling → Delphi expert consultation → AHP weight calibration—demonstrates a transferable paradigm for building assessment instruments that marry data-driven indicator discovery with structured expert judgment. Deriving the initial indicator pool from a broad literature corpus via LDA, rather than relying exclusively on deductive reasoning or expert nomination, broadens the scope of indicator selection and mitigates disciplinary blind spots.

Third, the tripartite dimensional architecture of AFLE-HIV—Physical Environment Adaptability, Social and Psychological Support, and Healthcare and Privacy Protection—operationalizes and enriches the social-ecological model [14] by mapping built-environment, social-environment, and institutional-environment layers onto a single quantified instrument. The empirically derived weight distribution across these layers furnishes evidence on the relative salience of each environmental stratum for a vulnerable subpopulation.

Practical implications

From an applied standpoint, AFLE-HIV furnishes a ready-to-deploy protocol for on-site residential appraisals. Community health workers and public health practitioners can use the weighted indicator checklist to pinpoint specific environmental deficits—whether inadequate barrier-free modifications, excessive distance to ART distribution points, or weak privacy safeguards—and translate diagnostic results into targeted intervention plans.

For policymakers, the weighted indicator system offers a basis for resource allocation decisions. The finding that Medical Facility Accessibility and HIV Treatment Continuity carry the highest combined weights suggests that investments in healthcare infrastructure proximity and ART supply chain reliability should be prioritized. In the Chinese context, where the “Healthy China 2030” strategy and the national strategy for actively addressing population aging are concurrently being implemented, AFLE-HIV provides a tool for bridging these two policy agendas by identifying environmental factors that lie at their intersection.

For urban planners engaged in age-friendly community renovation projects, the framework’s Physical Environment Adaptability indicators offer specific guidance for design modifications. The emphasis on Barrier-free Design and Public Space Age-friendliness directs attention to features such as ramp installation, handrail provision, non-slip flooring, adequate lighting, and accessible public seating—all of which have heightened importance for individuals experiencing the combined mobility limitations of aging and HIV-related conditions.

The pilot test included in this study also demonstrates that the framework can be operationalized by trained assessors across different urban settings, with acceptable inter-rater reliability (ICC = 0.694). This finding, while preliminary, suggests that AFLE-HIV has the potential for broader deployment in community health service delivery and urban planning processes.

From the perspective of the HIV care continuum, AFLE-HIV offers a complementary environmental lens that extends beyond clinical interventions. By identifying residential-level barriers to ART adherence, care retention, and psychosocial support, the framework enables community-based service delivery models to target upstream environmental determinants rather than relying solely on individual-level behavior change. For instance, a low score on Health Privacy Protection (C-3) in a given community could trigger the installation of private consultation rooms in community health centers, while a low score on Peer Support Networks (B-4) could prompt the establishment of facilitated peer groups—both of which have demonstrated efficacy in improving viral suppression rates and reducing social isolation among older adults living with HIV [4, 25].

Limitations and future work

This study has several limitations that should be acknowledged. First, the AFLE-HIV framework was developed and validated within the Chinese context. The literature corpus was sourced from English-language and Chinese-language databases, the expert panel was composed exclusively of Chinese scholars and practitioners, and the pilot test was conducted in four Chinese cities. The indicator content and AHP-derived weights therefore reflect the healthcare infrastructure, cultural norms, and policy landscape of China. While the methodological pipeline—LDA topic modeling, Delphi consultation, and AHP weight calibration—is inherently transferable and can be replicated in other countries by assembling context-appropriate expert panels and literature corpora, the specific indicator weights and operational definitions should not be directly extrapolated to non-Chinese settings without local recalibration.

Second, the expert panel did not include representatives of the target population itself—older adults living with HIV or their caregivers. This design decision was guided by the sensitivity of HIV status disclosure, the ethical complexities of recruiting from a stigmatized population for instrument development, and the challenge of identifying eligible participants willing to engage in structured consultations. Nevertheless, the inclusion of HIV/AIDS prevention specialists (22.7% of the panel) who have extensive clinical interaction with this population partially mitigated this limitation. Future research should incorporate participatory research methods—such as community-based participatory design, focus groups, and cognitive interviews with older adults living with HIV—to capture lived-experience perspectives that expert judgment alone may not fully represent.

Third, the pilot test was conducted with four assessors evaluating 12 communities across four cities. While the resulting content validity indices and reliability coefficients are encouraging, a larger and more geographically diverse validation study would further strengthen the evidence base. Future research should extend the sample to include communities from additional regions and employ a broader pool of assessors to establish normative reference values and refine indicator definitions.

Fourth, the ceiling effect observed in dimension C (Healthcare and Privacy Protection; mean = 4.725, SD = 0.254) warrants further attention. This elevated scoring likely reflects the fact that the four pilot cities—Beijing, Shanghai, Kunming, and Chengdu—are major urban centers with relatively well-developed HIV care infrastructure, which may not be representative of smaller cities or rural areas where healthcare accessibility is more constrained [63]. The restricted score range in dimension C may attenuate inter-rater variability and reduce the instrument’s discriminative capacity for this dimension. Future validation studies should include communities from resource-limited settings to ensure that the full scoring range is exercised and to determine whether additional indicators or finer-grained scoring rubrics are needed for dimension C.

Fifth, the AHP-derived weights represent a baseline prioritization that may require adjustment for specific application scenarios. In urban environments with well-developed healthcare infrastructure, the relative importance of Medical Facility Accessibility may decrease, while in rural areas with limited HIV care services, this indicator may carry even greater weight. Context-specific calibration should be conducted when applying the framework to particular settings.

Finally, the current study focuses on the development and preliminary validation of the assessment framework. Longitudinal studies examining the relationship between AFLE-HIV scores and health outcomes—such as ART adherence rates, hospitalization frequency, depression prevalence, and self-reported quality of life—would provide criterion validity evidence and strengthen the framework’s utility as a predictive tool. The development of a digital scoring platform to enhance assessment efficiency and facilitate data aggregation across multiple evaluation sites is also a priority for future work.

Conclusions

This study produced the AFLE-HIV (Age-Friendly Living Environment for Elderly HIV-Infected Individuals) assessment framework, a structured instrument for gauging how well residential settings accommodate the compound demands of aging and HIV management across three primary dimensions: Physical Environment Adaptability, Social and Psychological Support, and Healthcare and Privacy Protection. Latent themes were first surfaced from 347 publications through unsupervised LDA topic modeling, generating a data-grounded initial indicator pool. Two iterative Delphi rounds involving 22 specialists from five professional fields then screened, consolidated, and validated the indicators until stable consensus was achieved, with content validity indices (I-CVI \ge 0.82; S-CVI/Ave = 0.90) confirming expert endorsement of indicator relevance. Pairwise comparison matrices subsequently yielded hierarchically structured weights, with Physical Environment Adaptability attracting the highest primary weight (0.539)—underscoring the foundational role of the built environment in sustaining daily functioning.

Beyond physical accessibility and daily living convenience, the framework embeds indicators that address the social, psychological, and privacy dimensions unique to older adults living with HIV, thereby capturing environmental challenges that generic age-friendly instruments overlook. A field pilot across four Chinese cities returned satisfactory internal consistency (Cronbach’s α = 0.812) and moderate inter-rater concordance (ICC = 0.694), offering preliminary evidence of operational feasibility. As the global population of older adults living with HIV continues to expand—projected to comprise the majority of all people living with HIV in high-income countries by 2030—the need for environment-level assessment tools that bridge aging and chronic disease management will intensify across diverse national contexts. The AFLE-HIV instrument, developed and initially validated in China, provides a methodological template and substantive foundation for such efforts, equipping community health workers, policymakers, and urban planners with actionable diagnostics for evaluating, prioritizing, and upgrading residential environments for this growing and underserved population. Subsequent research should pursue large-scale psychometric validation, longitudinal outcome studies linking AFLE-HIV scores to health indicators along the HIV care continuum, and cross-cultural adaptation to extend the framework’s applicability beyond the Chinese context.

Abbreviations

AFLE-HIV

Age-Friendly Living Environment for Elderly HIV-Infected Individuals

AHP

Analytic Hierarchy Process

ART

Antiretroviral Therapy

CFA

Confirmatory Factor Analysis

CR

Consistency Ratio

CV

Coefficient of Variation

CVI

Content Validity Index

I-CVI

Item-level Content Validity Index

ICC

Intraclass Correlation Coefficient

LDA

Latent Dirichlet Allocation

S-CVI/Ave

Scale-level Content Validity Index (Average)

WHO

World Health Organization

Authors' contributions

Q.W. conceived and designed the study, developed the search strategy and conducted the literature review, designed and administered the Delphi questionnaire, performed the LDA topic modeling and AHP weight calculations, coordinated the pilot test, analyzed all data, and drafted the manuscript. M.I. supervised the research methodology and provided critical intellectual input on the framework design. M.H.A.N. contributed to the validation of the assessment framework and critically revised the manuscript. Y.W. contributed clinical expertise from HIV/AIDS care practice, provided critical review of the framework indicators from a clinical perspective, and participated in the expert consultation process. C.H. contributed to data collection, assisted with the Delphi consultation process, and critically revised the manuscript. All authors read and approved the final manuscript.

Funding

This work was supported by the Henan Provincial Federation of Social Sciences and Henan Provincial Economic Group Federation Research Project: “Research on Age-Friendly Community Evaluation Framework Development” (Grant No. SKL2022608). This paper represents a phased research outcome of the aforementioned project.

Data availability

The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

This study was reviewed and approved by the Academic Ethics Committee of Henan Industry and Trade Vocational College (Approval No. HNGM-2024-EC-012). Expert consultations were conducted with voluntary informed consent obtained from all participants prior to each Delphi round. No personally identifiable information of older adults living with HIV was collected during the pilot test; only environmental attributes of residential communities were assessed.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.