OBJECTIVETo analyze the correlation between population characteristics and work ability of employees with a multilevel model, to investigate the important influencing factors for work ability, and to provide a basis for improvement in work ability.

METHODSWork ability index (WAI)was applied to measure the work ability of 1686 subjects from different companies (n=6). MLwi N2.0 software was applied for two-level variance component model fitting.

RESULTSThe WAI of employees showed differences between various companies (χ2=3.378 6, P=0.0660); working years was negatively correlated with WAI (χ2=38.229 2, P=0.0001), and the WAI of the employees with 20 or more working years was 1.63 lower than that of the employees with less than 20 working years; the work ability of manual workers was lower than that of mental-manual workers (χ2=8.2726, P=0.0040), and the work ability showed no significant difference between mental workers and mental-manual workers (χ2=2.086 0, P=0.148 7).

CONCLUSIONFrom the perspective of probability, the multilevel model analysis reveals the differences in work ability of employees between different companies, and suggests that company, work type, and working years are the important influencing factors for work ability of employees. These factors should be improved and adjusted to protect or enhance the work ability of employees.