Exploration of subgroups and predictors of health-promoting lifestyle among older adults in the community: a latent profile analysis

Jianyi Bao, Shasha Li, Yuwei Lu, Guojing Guo, Yuecong Wang, Shufang Liao, Yue Li, Yingxue Xi, Xiaofang Song, Xinyu YangFrontiers in Public Health 2026 · DOI

TL;DR

Researchers used Latent Profile Analysis (LPA) to move beyond "one-size-fits-all" health metrics for the elderly, identifying three distinct subgroups of community-dwelling older adults based on their health-promoting lifestyles. The study reveals that factors like smartphone usage, friend networks, and socioeconomic status are critical predictors of which "health profile" an individual falls into, suggesting that future public health interventions must be tailored to specific subgroup deficits rather than general populations.

What Problem Does This Paper Address?

As global populations age, "healthy aging"—maintaining functional ability in later life—has become a public health priority. While we know that health-promoting lifestyles (HPL) reduce depression and chronic disease, most research treats HPL as a monolithic score. This "variable-centered" approach masks the reality that two people with the same total score might have vastly different needs; one might struggle with nutrition while another lacks social support.

The authors address this gap by using a "person-centered" approach to identify unobserved heterogeneity. They aim to categorize older adults into specific profiles based on six dimensions: self-actualization, health responsibility, physical exercise, nutrition, interpersonal support, and stress coping.

Methods

The study utilized a cross-sectional survey design, collecting data from 503 community-dwelling Chinese older adults (aged ≥60) in Huzhou, Zhejiang.

The core analytical technique was Latent Profile Analysis (LPA). Unlike standard clustering, LPA is a model-based approach that identifies subgroups (latent profiles) within a population based on their response patterns across continuous variables (in this case, the 36 items of the HPLP-CE scale).

To identify the best-fitting model, the researchers compared 1-profile through 5-profile solutions using several statistical indices:

  • Information Criteria: AIC, BIC, and aBIC (lower is better).
  • Classification Quality: Entropy (values ≥ 0.80 indicate high accuracy).
  • Likelihood Ratio Tests: LMRT and BLRT to determine if adding another profile significantly improves the model.

Once profiles were established, multivariate logistic regression was used to identify which sociodemographic and psychosocial factors predicted membership in each group.

Datasets

The researchers generated a primary dataset through face-to-face interviews. Key instruments included:

  • HPLP-CE (Health-Promoting Lifestyle Profile-Chinese Older Adults): A 36-item scale covering six health dimensions.
  • LSNS-6 (Revised Lubben Social Network Scale): Specifically the friend network subscale.
  • Health Risk Perception Questionnaire: Based on the health belief model (perceived susceptibility and severity).

Key Insights and Ideas

  • The "Three-Profile" Reality: The elderly population isn't a gradient of "healthy to unhealthy." Instead, they fall into three distinct archetypes:
  • Low HPL Group (32.80%): Low scores across all dimensions, likely facing "cumulative disadvantages" like poor literacy and socioeconomic constraints.
  • Moderate HPL-Low Health Responsibility Group (44.33%): Generally healthy but specifically weak in proactive health monitoring and vigorous exercise.
  • High HPL-Low Stress Management Group (22.86%): High overall engagement but surprisingly weak at coping with "unreasonable demands" and life stressors.
  • The Digital Dividend: Smartphone usage was a massive predictor. Users were significantly more likely to belong to the moderate or high HPL groups, likely due to easier access to health information and digital social integration.
  • Friends vs. Family: While family support is often emphasized in Chinese culture, this study highlights that friend networks were a potent predictor of higher HPL profiles, suggesting that peer-led community activities may be more effective for lifestyle changes than family-based ones.
  • Gendered Responsibility: Older women were more likely to fall into the "Low Health Responsibility" group. The authors hypothesize that heavy family obligations (childcare and housework) may cause women to deprioritize their own health monitoring.

Key Results

  • Profile Distribution: The largest group was the "Moderate-Low Responsibility" group (44.33%), followed by the "Low HPL" group (32.80%).
  • Predictive Power: The logistic regression model was robust (Pseudo-R² = 0.571).
  • Income & Health: Monthly income >3,000 RMB was a strong predictor for the High HPL group (OR = 7.365 compared to the Low HPL group).
  • Smartphone Impact: Smartphone users had over 17 times the odds (OR = 17.148) of being in the High HPL group compared to non-users.
  • Social Participation: Participation in geriatric activities increased the odds of being in the High HPL group by over 8 times (OR = 8.356).

Limitations and Future Work

The authors acknowledge several constraints:

  • Regional Bias: The study was conducted in Huzhou, a relatively developed area in East China. Results may not generalize to less developed or rural regions.
  • Cross-Sectional Nature: The data captures a snapshot in time; it cannot prove that getting a smartphone causes a healthier lifestyle, only that they are correlated.
  • Self-Reporting: The reliance on interviews may introduce social desirability or recall bias.

Future Work: The authors suggest moving toward longitudinal studies to track how these profiles change over time and incorporating objective physical measures (like frailty or mobility tests) to supplement self-reported lifestyle data.

Summary updated

Abstract

This study classified health-promoting lifestyles among community-residing older adults and identified predictors for each, helping to develop tailored health intervention programs.
BibTeX
@inproceedings{bao2026exploration,
  author = {Jianyi Bao and Shasha Li and Yuwei Lu and Guojing Guo and Yuecong Wang and Shufang Liao and Yue Li and Yingxue Xi and Xiaofang Song and Xinyu Yang},
  title = {{Exploration of subgroups and predictors of health-promoting lifestyle among older adults in the community: a latent profile analysis}},
  booktitle = {Frontiers in Public Health},
  year = {2026},
  doi = {10.3389/fpubh.2026.1583243},
}
Figure