GLS Regression of Factors Influencing Functional Health Among Older Adults in China Based on Panel Data

Technical Summary: GLS Regression of Factors Influencing Functional Health Among Older Adults in China Based on Panel Data

Problem Statement
With the accelerating trend of population aging, understanding the determinants of functional health among older adults is critical for maintaining health and longevity. While Activities of Daily Living (ADL) and Instrumental Activities of Daily Living (IADL) are established metrics for evaluating physical function and self-care capacity, existing research predominantly treats these metrics as predictors for other endpoints (e.g., mortality or cognitive decline). There is a scarcity of studies that model ADL and IADL as outcome variables to elucidate the specific risk factors driving their progression. This study addresses this gap by defining ADL and IADL as core metrics of functional health and investigating the factors influencing their decline among Chinese adults aged 65 and above.

Methodology
The study utilizes a longitudinal panel dataset derived from the Chinese Longitudinal Healthy Longevity Survey (CLHLS), a nationally representative cohort covering 23 provinces in mainland China.

• Data Source: Four waves of survey data (2008, 2011, 2014, and 2018) were extracted. After cleaning and excluding missing values, the final balanced panel consisted of 20,502 individuals aged 65+.
• Dependent Variables: Functional health was operationalized via two indices:
  • ADL: Composed of six items (bathing, dressing, mobility, toileting, feeding, continence). Scores were aggregated and normalized (0–12 scale), where higher values indicate greater functional impairment.
  • IADL: Composed of eight items (social visits, shopping, meal prep, laundry, walking 1km, lifting 5kg, sit-to-stand, public transport). Scores were aggregated and normalized (0–24 scale), with higher values indicating greater impairment.
• Independent Variables: Eight predictors were analyzed: chronological age, Body Mass Index (BMI), years of education, and five neurocognitive domains (general ability, reaction time, attention, memory, and language/comprehension/motor coordination).
• Statistical Approach: Generalized Least Squares (GLS) regression was employed to estimate the associations. The Hausman test was conducted to determine the appropriate model specification (Fixed Effects vs. Random Effects). For both ADL and IADL models, the Hausman test indicated a statistically significant difference ($P < 0.0001$) between estimators, leading to the selection of the Random Effects (RE) model as the optimal specification.

Key Results
The GLS regression analysis identified distinct categories of factors influencing functional health, with results varying slightly between ADL and IADL models:

1. Protective Factors (Negative Coefficients): Higher scores in specific neurocognitive domains were associated with a reduction in functional impairment (i.e., better functional health).

   • ADL Model: General ability, attention, memory, and language/comprehension/coordination were statistically significant ($P < 0.05$). Reaction capacity was not significant.
   • IADL Model: General ability, reaction capacity, attention, memory, and language/comprehension/coordination were all statistically significant ($P < 0.05$).
   • Interpretation: These cognitive domains act as positive regulators, inhibiting or delaying the decline of functional autonomy.

2. Risk Factors (Positive Coefficients): Higher values in demographic and physiological variables were associated with an increase in functional impairment.

   • Common to Both Models: Age and Years of Education showed a positive quantitative relationship with functional decline ($P < 0.05$).
   • ADL Specific: BMI was a significant predictor of decline ($P = 0.002$).
   • IADL Specific: BMI was not statistically significant ($P = 0.204$).
   • Interpretation: Advancing age, higher BMI (for ADL), and longer years of education are associated with accelerated functional decline.

Key Contributions

• Methodological Application: The study applies a GLS random-effects framework to panel data, moving beyond categorical data analysis to quantify the specific associations between continuous functional health metrics and a broad set of predictors.
• Outcome Modeling: It shifts the analytical focus from using ADL/IADL as predictors of mortality to modeling them as outcomes, thereby identifying specific drivers of functional deterioration.
• Nuanced Findings on Education: The study highlights a counterintuitive finding where higher educational attainment correlates with faster functional decline, contrasting with literature that often views education as protective. The author suggests this may be linked to sedentary lifestyles associated with intellectually demanding occupations.
• Cognitive-Functional Link: The analysis confirms a strong, positive moderating role of cognitive function (across multiple domains) in preserving functional independence.

Significance and Claims
The paper claims that its findings provide a quantitative basis for understanding geriatric health dynamics in China. By categorizing factors into those that positively regulate (cognitive abilities) and those that negatively regulate (age, BMI, education) functional health, the study aims to inform targeted active aging policies.

The author concludes that public health interventions should prioritize the early detection of ADL and IADL decline. Specifically, they advocate for the promotion of moderate- to high-frequency physical activities within community settings for older adults. The stated goal of these interventions is to improve overall health levels and self-care capacities, thereby mitigating functional deterioration. The study does not propose new experimental protocols but rather calls for the application of these insights to existing community health strategies.