Statistical modelling needed to find the effects from a community-based elderly safety promotion program.

Area of Science:

• Public Health
• Gerontology
• Epidemiology

Background:

• Community-based health promotion evaluations often use multiple control areas and time-series analyses.
• Significant yearly and area-specific fluctuations can obscure true intervention effects.

Purpose of the Study:

• To evaluate the effectiveness of a community-based elderly safety promotion program.
• To assess the impact of the program on hip fracture incidence using a quasi-experimental time-series design.

Main Methods:

• A quasi-experimental time-series analysis was conducted with one intervention and six control communities.
• Data spanned 12 years (6 pre-intervention, 6 post-intervention) for individuals aged 65+.
• Hip fracture incidence was analyzed using multivariate methods accounting for intra-annual and intra-panel variance.

Main Results:

• Conventional time-series analyses showed no clear patterns or conclusive results.
• Multivariate analysis revealed underlying trends, indicating 14 fewer hip fractures in the intervention community than expected.
• The results were sensitive to extreme values, highlighting the need for robust statistical approaches.

Conclusions:

• Effect evaluations of community-based health promotion in small communities can be unreliable without appropriate statistical modeling.
• Advanced statistical techniques are crucial for accurately interpreting time-series data and determining intervention efficacy.