Addressing ‘difficulty in dining’ among older adults: optimizing community senior dining halls from external and internal built environments

Global population aging, particularly acute in China, raises challenges in meeting older adults’ daily needs. Community-based senior dining halls (CSDHs) provide affordable, nutritious meals and social opportunities but many face financial strain due to low dining volume. Prior work suggests food quality, income, and built environment influence dining behaviors, yet direct, objective evidence linking external and internal built environment characteristics to CSDH usage is limited. This study asks how built environment factors around and within CSDHs relate to dining volume, positing two hypotheses: (1) comfortable and convenient external environments (e.g., density, transit, street/greenery design, surrounding POI diversity) increase older adults’ likelihood of traveling to CSDHs; (2) attractive, safe, and comfortable internal environments with social spaces (e.g., dine-in seating, amenities) encourage use. Using objective dining records and built-environment measures, the study seeks to inform planning for community-based senior care.

Evidence links built environments to older adults’ travel and destination use via the 3Ds framework (Density, Diversity, Design). Density (e.g., building density, FAR) correlates with activity and walking, though findings are mixed for FAR around parks. Design elements—street density/connectivity, transit supply (bus, subway), and greenery—affect walking and active travel among older adults. Diversity of points of interest can relate to active travel but results vary by context and perception. Internally, restaurants and canteens serve as social venues for older adults; features such as dine-in seating and supportive interior elements (barrier-free handrails, water access, seating) can enhance comfort and frequency of visits. Much prior research relies on self-reported data subject to bias, underscoring the need for objective usage measures. Few studies have directly examined CSDH usage in relation to built environments, creating a gap this study addresses.

Study area: Yuexiu District, Guangzhou, China—dense urban core with highest aging level in the city (28.41% aged 65+). In 2023, 50 CSDHs operated in Yuexiu, providing adequate sample and environmental variation. Data on dining volume: Official dining volumes for Q2 2023 (to avoid Q1 COVID-19 effects) obtained from the Yuexiu District Civil Affairs Bureau, including dine-in, take-out, and deliveries. External built environment factors: Using GIS, CSDH locations were geocoded and Euclidean buffers of 400 m, 600 m, and 800 m created (typical older adult walking distances). Variables (from OSM and remote sensing/Gaode): Density—floor area ratio (FAR); Design—street density, number of bus stops, distance to nearest subway station, NDVI; Diversity—catering POI diversity (Shannon index). FAR was preferred over building coverage ratio to capture 3D density. Internal built environment factors: On-site audit recorded facilities (availability of kitchen, dine-in seating, health/eldercare center), architectural design (number of entrances, building floor, accessible entrances), and interior design (barrier-free handrails, number of tables, number of seats [≤10, 11–20, 21–30, >30], wash basins, water dispensers, clocks, TVs, air conditioning, menu, bulletin board, spice racks, sterilization lamps). Covariates: Older population density (from the 7th national census) and a proxy for food taste (vote counts from a district competition; 566,353 total votes distributed among CSDHs). Prices were uniform (15 Yuan/serving). Statistical analysis: Multiple linear regression models estimated separately for each buffer (400/600/800 m). Preprocessing: variables constant across sites (air conditioning, menu, bulletin board) removed; multicollinearity assessed via VIF (threshold >4), removing spice rack, sterilization lamp, NDVI, and number of tables. Dining volume was log-transformed; independent variables standardized. Models report coefficients, p-values, VIFs, adjusted R², and AIC. Additional stratified summaries examined thresholds for FAR (≤5.00, 5.01–6.00, 6.01–7.00, ≥7.00) and distance to subway (≤300 m, 301–500 m, ≥500 m), and seat groups (≤10, 11–20, 21–30, ≥30).

• Model fit: Adjusted R² ≈ 0.478 (400 m), 0.476 (600 m), 0.494 (800 m). External environment:
• FAR: Negative association with dining volume; significant at 400 m and 600 m (coef −0.271, p<0.05), marginal at 800 m (p=0.101). Dining volume rises notably when FAR <6.00.
• Distance to nearest subway station: Negative association (closer is better); marginal at 400 m (p=0.080), significant at 600 m (p=0.046) and 800 m (p=0.029). Sites ≥500 m from a subway had ~500 fewer meals than those within 500 m.
• Street density, number of bus stops, and catering POI diversity were not significant (street density marginal at 800 m, p≈0.085). Internal environment:
• Dine-in seating: Positive association; significant at 400 m and 600 m (coef ≈0.392, p<0.05), marginal at 800 m (p≈0.060).
• Building floor: Negative (lower floors associated with higher volume); marginal significance at 600 m (p≈0.080) and 800 m (p≈0.087).
• Number of seats: Relative to ≤10 seats, both 21–30 and >30 categories showed significantly higher dining volumes across buffers (e.g., 400 m: 21–30 coef 0.317, p=0.027; >30 coef 0.319, p=0.025). Average dining volume exceeded 4000 meals for >30 seats group.
• Water dispenser: Positive; significant at 600 m (p=0.028) and 800 m (p=0.016), marginal at 400 m (p=0.057).
• Barrier-free handrail marginal at 400 m (p=0.083); other interior items and facility variables (kitchen, healthcare center) not significant. Covariates:
• Older population density and food taste votes were generally not significant (food taste marginal at 400 m, p≈0.071).

Findings support both hypotheses. Externally, lower FAR areas and proximity to subway stations are associated with higher dining volumes, likely reflecting older residential districts with greater older adult presence, more comfortable walking environments, and potentially better cognitive/social engagement contexts. A practical threshold emerged: FAR below ~6 and within 500 m of a subway station are favorable. Other external factors (street density, bus stops, POI diversity) showed limited or no associations. Internally, dine-in options, being on lower floors, more seating (≥21 seats, especially >30), and water dispensers positively relate to usage, reinforcing the importance of comfortable, accessible spaces that enable social interaction. Features like on-site kitchens, health centers, or additional entrances did not show measurable associations with dining volume in this sample. Overall, objective usage data combined with GIS and audit measures indicates specific architectural and amenity choices that can drive CSDH patronage and, by extension, financial sustainability and social benefits.

Analyzing 50 CSDHs in Yuexiu District, Guangzhou, the study demonstrates that specific external (lower FAR, closer to subway stations) and internal (dine-in seating, lower floors, higher seating capacity, water dispensers) built environment attributes are linked to higher dining volumes. These insights can guide site selection (prefer FAR <6, within 500 m of subway) and interior programming (provide dine-in seating and ≥20–30 seats, ground/low floors, water dispensers) to enhance patronage, social interaction, and sustainability of CSDHs. Future work should incorporate individual-level factors, ambient environmental qualities, service quality/marketing variables, and subjective perceptions to further validate and refine these relationships.

• Individual-level confounders (socioeconomic status, social needs, health conditions) were not included due to data limitations.
• Ambient environmental factors (lighting, acoustics, noise) were not measured owing to scope, time, and equipment constraints.
• Service quality (cleanliness, staff friendliness) and marketing/advertising effects were not assessed.
• Study did not include subjective perceptions; future surveys could validate causal pathways.
• Generalizability is limited to similar high-density urban contexts and requires replication in other cities and time periods.