Table 2. Results of the multilevel logistic regression analysis of recurrent fall experiences.
| Recurrent fall experience | Model 1 | Model 2 | Model 3 (PM10) | Model 3 (PM2.5) | ||
|---|---|---|---|---|---|---|
| Individual level | ||||||
| Age (65 yr or over) | 1.131 (1.062–1.204) | 1.130 (1.061–1.203) | 1.130 (1.061–1.203) | 1.130 (1.061–1.203) | ||
| Sex (female) | 1.394 (1.315–1.478) | 1.393 (1.314–1.477) | 1.393 (1.314–1.477) | 1.393 (1.314–1.477) | ||
| Marital status (ref. unmarried) | ||||||
| Married | 0.376 (0.357–0.397) | 0.377 (0.357–0.397) | 0.377 (0.357–0.397) | 0.377 (0.357–0.397) | ||
| Other (divorced, etc.) | 0.461 (0.429–0.495) | 0.462 (0.430–0.496) | 0.462 (0.430–0.496) | 0.462 (0.430–0.496) | ||
| Education (9 yr or over) | 0.709 (0.666–0.754) | 0.708 (0.665–0.753) | 0.707 (0.665–0.752) | 0.707 (0.665–0.753) | ||
| Household income (ref. first quartile) | ||||||
| Second quartile | 0.778 (0.733–0.826) | 0.778 (0.733–0.825) | 0.777 (0.732–0.825) | 0.777 (0.732–0.825) | ||
| Third quartile | 0.756 (0.711–0.805) | 0.756 (0.710–0.804) | 0.754 (0.709–0.803) | 0.755 (0.709–0.803) | ||
| Fourth quartile | 0.720 (0.674–0.770) | 0.719 (0.673–0.769) | 0.718 (0.672–0.767) | 0.718 (0.672–0.768) | ||
| Alcohol consumption (ref. never drinker) | ||||||
| Former drinker | 1.186 (1.108–1.270) | 1.186 (1.108–1.269) | 1.185 (1.108–1.269) | 1.185 (1.108–1.269) | ||
| Occasional drinker | 1.182 (1.099–1.270) | 1.181 (1.099–1.270) | 1.181 (1.099–1.269) | 1.181 (1.099–1.269) | ||
| Current drinker | 1.237 (1.162–1.317) | 1.237 (1.162–1.316) | 1.236 (1.162–1.316) | 1.236 (1.162–1.316) | ||
| Smoking status (ref. never smoker) | ||||||
| Former smoker | 1.199 (1.118–1.285) | 1.199 (1.119–1.285) | 1.199 (1.118–1.285) | 1.199 (1.118–1.285) | ||
| Current smoker | 1.381 (1.293–1.475) | 1.381 (1.293–1.475) | 1.381 (1.293–1.475) | 1.381 (1.293–1.475) | ||
| Physical activity (yes) | 1.005 (0.963–1.048) | 1.003 (0.962–1.046) | 1.003 (0.962–1.046) | 1.003 (0.962–1.046) | ||
| BMI categories (ref. normal) | ||||||
| Underweight | 1.158 (1.060–1.265) | 1.157 (1.059–1.264) | 1.157 (1.059–1.264) | 1.157 (1.059–1.264) | ||
| Overweight | 1.125 (1.075–1.177) | 1.125 (1.075–1.177) | 1.125 (1.075–1.177) | 1.125 (1.075–1.177) | ||
| Obesity | 1.361 (1.257–1.475) | 1.361 (1.256–1.475) | 1.361 (1.256–1.474) | 1.361 (1.256–1.474) | ||
| Depressive symptom (yes) | 3.483 (3.247–3.736) | 3.479 (3.243–3.731) | 3.476 (3.241–3.728) | 3.475 (3.240–3.727) | ||
| Diagnosis of diabetes (yes) | 1.221 (1.149–1.299) | 1.222 (1.149–1.299) | 1.221 (1.149–1.299) | 1.221 (1.149–1.299) | ||
| Diagnosis of hypertension (yes) | 0.973 (0.925–1.024) | 0.973 (0.925–1.024) | 0.973 (0.924–1.024) | 0.973 (0.924–1.024) | ||
| Residence area (urban) | 0.997 (0.931–1.069) | 1.027 (0.955–1.105) | 1.035 (0.962–1.113) | 1.030 (0.957–1.108) | ||
| Regional level | ||||||
| PM10 (per 10 µg/m3) | 1.131 (1.026–1.247) | |||||
| PM2.5 (per 10 µg/m3) | 1.202 (1.051–1.374) | |||||
| Urban forest area ratio (ref. first quartile)a | ||||||
| Second quartile | 1.278 (1.096–1.491) | 1.277 (1.097–1.487) | 1.278 (1.098–1.488) | |||
| Third quartile | 1.094 (0.937–1.278) | 1.094 (0.938–1.275) | 1.097 (0.942–1.279) | |||
| Fourth quartile | 0.950 (0.804–1.123) | 0.981 (0.830–1.160) | 0.985 (0.833–1.163) | |||
| Urban forest of life zone per capita (ref. first quartile)a | ||||||
| Second quartile | 1.084 (0.924–1.273) | 1.064 (0.907–1.247) | 1.063 (0.907–1.246) | |||
| Third quartile | 1.052 (0.885–1.252) | 1.059 (0.892–1.257) | 1.043 (0.879–1.237) | |||
| Fourth quartile | 1.022 (0.854–1.223) | 1.060 (0.885–1.268) | 1.034 (0.866–1.234) | |||
| Population density (per person/km2) | 1.000 (1.000–1.000) | 1.000 (1.000–1.000) | 1.000 (1.000–1.000) | |||
| ICC | 4.67% | 4.25% | 4.14% | 4.11% | ||
| PCVb | 6.97% | 9.40% | 2.82% | 3.33% | ||
Values are presented as beta coefficients (95% confidence interval). In the null model, the ICC was 5.00%
PM = particulate matter, BMI = body mass index, ICC = intraclass correlation coefficient, PCV = proportional change in variance.
aUrban forest area ratio and urban forest of life zone per capita were acquired for 237 regions and categorized into quartiles; bThe proportional change in the area level variance between the empty model and model 1; model 1 and model 2; and, model 2 and model 3.