Table 2.
Selected baseline covariates measured in 2019 and loneliness assessed in 2022 stratified by quartiles of estimated conditional average treatment effects (CATEs) for the association between frequent virtual communicationa and loneliness among older adults in Japanb.
| Baseline characteristic | Quartile 1 (n=1207) | Quartile 2 (n=1200) | Quartile 3 (n=1200) | Quartile 4 (n=1200) |
| Estimated CATEs, mean (SD) | −0.21 (0.03) | −0.13 (0.02) | −0.07 (0.02) | −0.02 (0.03) |
| Loneliness in 2022 (3-9)c, mean (SD) | 4.5 (1.6) | 4.3 (1.5) | 4.1 (1.4) | 3.9 (1.3) |
| Age (y), mean (SD) | 73.2 (5.9) | 74.1 (6.1) | 74.0 (6.0) | 73.4 (5.8) |
| Gender (women), n (%) | 51 (4) | 376 (31) | 960 (80) | 1161 (97) |
| Marital status (married), n (%) | 989 (82) | 918 (76) | 830 (69) | 870 (72) |
| Educational attainment (≥13 y), n (%) | 515 (43) | 308 (26) | 298 (25) | 239 (20) |
| Annual household income (¥10,000; US $1= ¥164 as of July 27, 2026), mean (SD) | 240 (154) | 227 (143) | 233 (135) | 269 (150) |
| Alcohol consumption (yes), n (%) | 266 (22) | 423 (35) | 737 (61) | 828 (69) |
| Smoking status (yes), n (%) | 164 (14) | 200 (17) | 77 (6) | 11 (1) |
| Number of natural teeth (≥20), n (%) | 594 (49) | 563 (47) | 615 (51) | 772 (64) |
| Frequency of going out (once a week or more), n (%) | 1167 (97) | 1163 (97) | 1172 (98) | 1177 (98) |
| Civic participationd (0-5), mean (SD) | 0.8 (1.1) | 1.0 (1.2) | 1.2 (1.3) | 1.3 (1.3) |
| Reciprocityd (0-3), mean (SD) | 2.5 (0.8) | 2.7 (0.6) | 2.7 (0.5) | 2.9 (0.3) |
| Social cohesiond (0-3), mean (SD) | 2.0 (1.1) | 2.1 (1.1) | 2.1 (1.1) | 2.1 (1.0) |
| Living alone (yes), n (%) | 128 (11) | 139 (12) | 167 (14) | 137 (11) |
| Self-rated health (very good), n (%) | 162 (13) | 221 (18) | 193 (16) | 152 (13) |
| GDS-15e (0-15), mean (SD) | 3.2 (3.3) | 3.1 (3.2) | 2.8 (2.8) | 2.7 (2.5) |
| Frequency of meeting friends (≥4 times weekly), n (%) | 164 (14) | 172 (14) | 229 (19) | 160 (13) |
| Internet use in 2018-2019 (a few times a month or less), n (%) | 636 (53) | 688 (57) | 701 (58) | 610 (51) |
aFrequent virtual communication was defined using a frequency item for internet or email use and a communication purpose item for communication with family, friends, or acquaintances at least 2 to 3 times per week in the preceding year. The communication purpose item included examples such as email, Line, Zoom, and video calls but did not separately identify each platform or modality. Nonusers or infrequent users were defined as participants who did not meet this frequent virtual communication definition.
bQuartile 1 to quartile 4 represent quartiles of the estimated CATEs. Quartile 1 had the most negative estimated CATEs, whereas quartile 4 had estimated CATEs closest to the null. CATE denotes the conditional average treatment effect estimand; estimated CATEs denote the individual-level forest predictions used for stratification. Because this study was an observational association analysis, the estimated CATEs are interpreted as exploratory heterogeneity in adjusted associations, not as definitive causal effects.
cLoneliness was assessed in 2022 using the 3-item UCLA Loneliness Scale (range 3-9), with higher scores indicating greater loneliness.
dThe 3 key dimensions of social capital.
eGDS-15: 15-item Geriatric Depression Scale.