Reduction of pulmonary function indices: comparison of two hierarchical methods

Area of Science:

• Pulmonary Medicine
• Biostatistics
• Occupational Health

Background:

• Pulmonary function tests (PFTs) generate numerous indices, leading to data redundancy.
• Identifying the most significant PFT indices is crucial for efficient analysis and interpretation.
• Firemen are a relevant occupational group for studying respiratory health due to potential workplace exposures.

Purpose of the Study:

• To apply and compare two hierarchical methods for selecting significant pulmonary function indices.
• To reduce redundancy in data derived from pulmonary function tests.
• To identify indices that best distinguish between asymptomatic and symptomatic individuals within a cohort of firemen.

Main Methods:

• Two hierarchical data reduction methods were employed: stepwise linear discriminant analysis and stepwise multiple correlation.
• Stepwise linear discriminant analysis selected indices based on their ability to discriminate between symptomatic and asymptomatic subjects.
• Stepwise multiple correlation selected indices based on their linear relationships with other variables.

Main Results:

• The two applied methods yielded distinct sets of significant pulmonary function indices.
• The choice of method influenced the selection of indices for data reduction.
• Discriminant analysis highlighted indices differentiating health status, while correlation analysis focused on inter-index relationships.

Conclusions:

• The selection of significant pulmonary function indices depends on the specific goals of data reduction.
• Both discriminant and correlational approaches offer valuable, yet different, perspectives for analyzing PFT data.
• Choosing the appropriate method is essential for targeted interpretation of pulmonary function test results.