Table 1 presents the calculation results for the factor loadings of mobility and RF support. The factor loading for each mobility was positive and greater than 0.2, indicating a positive correlation between the degree of disability and depression. The factor loadings for the surveyed items of RF support were negative and greater than 0.2, implying a negative correlation between support level and depression. Table 2 presents the participant demographic variables and the descriptive statistics for SD time point and Time since SD. Of the 710 participants, 256 were men and 454 were women. The participants were divided into six groups (on the basis of the spouses’ age at death), four education levels (from primary school to university and above), and four ethnic groups. The years 1989, 1993, 1996, 1999, 2003, and 2007 were used as the points of tangency for the time of SD occurrence. Time since SD was divided into five levels, namely less than 3 months, 3–6 months, 6–12 months, 12–24 months, and more than 24 months.
Table 3 presents data before and after SD according to participant sex. No significant difference was observed in income satisfaction between men and women before and after SD. A decline in the proportions of both men (ORm = 4.38, p < 0.001) and women (ORm = 7.0, p < 0.05) who smoked was observed. However, neither men nor women exhibited a significant difference in alcohol consumption before and after their spouses’ deaths. In terms of self-rated health, men did not exhibit a significant difference before and after SD, whereas the proportion of women who believed that they had favorable health declined after SD (ORm = 1.40, p < 0.05). In Table 3, diff represents the result of subtracting the post-SD score from the pre-SD score. Men exhibited a significantly decreased degree of depression after SD (mean = 1.52 ± 6.09, p < 0.0001), whereas the women’s degree of depression increased significantly (mean = − 1.31 ± 7.49, p < 0.01). A mobility score diff of < 0 indicates improved mobility. Men exhibited reduced mobility after SD (mean = 0.33 ± 1.25, p < 0.0001), whereas the mobility of women improved (mean = − 0.52 ± 1.50, p < 0.0001). Neither men nor women exhibited a significant difference RF support scores after SD, indicating no significant change in support from relatives and friends after SD.
Table 4 presents the multinomial logistic regression analysis of the effect of SD on depression. Model 1 consists of the simple effect without any variable adjustment; the mobility score was adjusted in Model 2, self-rated health was adjusted in Model 3, and the RF support score was adjusted in Model 4. Under the Model 1 SD effect, the proportion of depression among the participants after SD was 1.7 times that of before SD (P < 0.0001), whereas the SD effects of Models 2, 3, and 4 resulted in a proportion of depression after SD 1.5–1.7 times that of before SD, and the increase was significant. The odds ratios (ORs) of the adjustment variables in Models 1–4 were all significant, indicating that worsened mobility (OR = 1.3, p < 0.01), declined self-rated health status (OR = 0.5, p < 0.01), and higher RF support (OR = 1.5, p < 0.01) led to a higher proportion of depression. After variable adjustment, the SD effect on men remained significant, and the ORs of the adjustment variables were nonsignificant. After the mobility score (Model 2) and favorable self-rated health status (Model 3) were adjusted, the OR of the SD effect on women became nonsignificant, whereas the ORs of the adjustment variables became statistically significant (OR = 1.3, p < 0.01; OR = 0.4, p < 0.01).
Table 5 indicates that the proportion of post-SD depression was 6.0 times, 0.9 times, 1.7 times, and 1.3 times that of pre-SD depression when Time since SD was less than 6 months (OR = 6.0, p < 0.0001), 6–12 months (OR = 0.9, p > 0.05), 12–24 months (OR = 1.7, p < 0.05), and ≥ 24 months (OR = 1.3, p > 0.05), respectively. The proportion of depression decreased rapidly after 6 months. In the sex-stratified analysis, only men had a significantly higher proportion of post-SD depression compared with pre-SD depression in the 12–24 months after SD (OR = 2.38, p < 0.05), whereas women exhibited a significantly higher proportion of post SD depression less than 6 months after SD (OR = 4.0, p < 0.01). The remaining variables yielded no significant changes.
In the analysis of post-SD living arrangements, Table 5 indicates that the proportion of depression among widowers and widows living alone and those living with family members was 2.3 times (OR = 2.3, p < 0.01) and 1.5 times that before SD (OR = 1.5, p < 0.01). A further sex stratification analysis revealed that only men who lived with their families after SD exhibited significantly more severe depression compared with before SD (OR = 2.3, p < 0.01).
In terms of the stratified analysis of spouse’s health status, Table 5 indicates that a significantly large proportion of participants whose spouses had a favorable health status exhibited post-SD depression (all participants; OR = 2.0, p < 0.01, men: OR = 2.3, p < 0.05, women: OR = 1.8, p < 0.05). However, no significant difference was observed in the proportion of widowers and widows with post-SD depression whose spouses were in poor health.
Table 1
Factor loading derived by reduced rank regression: using pre SD mobility and social support items
Mobility items | Factor loading | | Social support items | Factor loading |
Squatting | 0.43 | | Willingness of significant others to talk with you | –0.40 |
Raising the arms up | 0.35 | | Willingness of significant others to care for you | –0.54 |
Grasping with the fingers | 0.42 | | Satisfaction with care from significant others | –0.57 |
Carrying 11 kg weight | 0.37 | | Significant others’ complaints to you | –0.27 |
Walking 200–300 m | 0.44 | | Willingness of significant others to ask your opinion | –0.38 |
Climbing up 2–3 floors | 0.43 | | -- | |
SD
Spouse Death; Factor loadings with absolute value ≥ 0.2 are shown in bold; higher mobility scores mean worse mobility function; higher social support scores mean lower level of social support.
Table 2
Characteristics of study subjects
| Total subjects (N = 710) | Male (n = 256) | Female (n = 454) |
| N | % | N | % | N | % |
Age of SD 60- | 39 | 5.5% | 5 | 2.0% | 34 | 7.5% |
60–65 | 87 | 12.3% | 20 | 7.8% | 67 | 14.8% |
65–70 | 148 | 20.9% | 49 | 19.1% | 99 | 21.8% |
70–75 | 160 | 22.4% | 54 | 21.1% | 105 | 23.0% |
75–80 | 165 | 23.2% | 68 | 26.6% | 97 | 21.4% |
80+ | 112 | 15.8% | 60 | 23.4% | 52 | 11.5% |
Education illiterate | 320 | 45.1% | 52 | 20.3% | 268 | 59.0% |
Elementary | 295 | 41.6% | 146 | 57.0% | 149 | 32.8% |
Junior/senior high | 85 | 12.0% | 50 | 19.5% | 35 | 7.7% |
Above college | 10 | 1.4% | 8 | 3.1% | 2 | 0.4% |
Ethnicity Fukinese | 510 | 71.8% | 172 | 67.2% | 338 | 74.5% |
Hakka | 113 | 15.9% | 46 | 18.0% | 67 | 14.8% |
Mainlander | 75 | 10.6% | 38 | 14.8% | 37 | 8.2% |
Other | 12 | 1.7% | 0 | 0.0% | 12 | 2.6% |
SD between 1989–1993 | 120 | 16.9% | 43 | 16.8% | 77 | 17.0% |
1993–1996 | 110 | 15.5% | 54 | 21.1% | 56 | 12.3% |
1996–1999 | 138 | 19.4% | 38 | 14.8% | 100 | 22.0% |
1999–2003 | 196 | 27.6% | 70 | 27.3% | 126 | 27.8% |
2003–2007 | 146 | 20.6% | 51 | 19.9% | 95 | 20.9% |
Time since SD 0–3 months | 40 | 5.6% | 16 | 6.3% | 24 | 5.3% |
3–6 months | 47 | 6.6% | 18 | 7.0% | 29 | 6.4% |
6–12 months | 105 | 14.8% | 37 | 14.5% | 68 | 15.0% |
12–24 months | 215 | 30.3% | 83 | 32.4% | 132 | 29.1% |
>24 months | 303 | 42.7% | 102 | 39.8% | 201 | 44.3% |
Table 3
Characteristics of study subjects: pre and post spouse death (SD)
Categorical variables | | | Male Post SD | ORm | | | | Female Post SD | ORm |
| | + | – | | | | + | – |
Satisfied Income | +: Yes -: No | Pre SD | + | 69 | 45 | 0.80 (NS) | | Pre SD | + | 104 | 68 | 0.94 (NS) |
- | 56 | 86 | | - | 72 | 210 |
| | | | + | – | | | | | + | - | |
Smoking habit | +: Yes -: No | Pre SD | + | 98 | 35 | 4.38 (`) | | Pre SD | + | 15 | 7 | 7 (*) |
- | 8 | 115 | | - | 1 | 431 |
| | | | + | - | | | | | + | - | |
Alcohol drinking | +: Yes -: No | Pre SD | + | 58 | 31 | 1.11 (NS) | | Pre SD | + | 18 | 20 | 1.11 (NS) |
- | 28 | 139 | | - | 18 | 398 |
| | | | + | - | | | | | + | - | |
self-rated health | +: Yes -: No | Pre SD | + | 162 | 44 | 1.42 (NS) | | Pre SD | + | 207 | 85 | 1.40 (*) |
- | 31 | 19 | | - | 61 | 101 |
| | Mean | (sd) | P | | Mean | (sd) | | P |
CESD Depressive: higher score | | | | | | | | | |
diff | 1.52 | (6.09) | *** | | -1.31 | (7.49) | | ** |
Mobility Worse: Lower score | | | | | | | | | | | | |
diff | 0.33 | (1.25) | *** | | | -0.52 | (1.50) | | *** |
RF support Low support: Higher score | | | | | | | | | | | | |
diff | 0.11 | (1.16) | NS | | | 0.05 | (1.15) | | NS |
# p < 0.1,* p < 0.05, ** p < 0.01, *** p < 0.0001 |
Table 4
Effect of spouse death on depression: multiple conditional logistic regression
| Total | Male | Female |
| OR | P | OR | P | OR | P |
Model 1: | | | | | | |
SD Effect (Post SD vs. Pre SD) | 1.7 | *** | 2.6 | ** | 1.5 | * |
Model 2: | | | | | | |
SD Effect(Post SD vs. Pre SD) | 1.5 | ** | 2.3 | ** | 1.2 | NS |
Mobility (higher score means worse) | 1.3 | ** | 1.3 | NS | 1.3 | ** |
Model 3: | | | | | | |
SD Effect(Post SD vs. Pre SD) | 1.7 | ** | 2.6 | ** | 1.4 | # |
Self-rated health (good vs. not good) | 0.5 | ** | 0.8 | NS | 0.4 | ** |
Model 4: | | | | | | |
SD Effect(Post SD vs. Pre SD) | 1.7 | ** | 2.3 | ** | 1.5 | * |
Higher social support | 1.5 | ** | 1.5 | # | 1.4 | * |
# p < 0.1,* p < 0.05, ** p < 0.01, *** p < 0.0001 |
Table 5
Effect of spouse death on depression: stratification analysis
| | Total | Male | Female |
| | OR | P | OR | P | OR | P |
Stratified by time since SD | | | | | | | |
< 6 months | Post SD vs. Pre SD | 6.0 | *** | Infinite | -- | 4.00 | ** |
6–12 months | Post SD vs. Pre SD | 0.9 | NS | 1.00 | NS | 0.88 | NS |
12–24 months | Post SD vs. Pre SD | 1.7 | * | 2.38 | * | 1.41 | NS |
>=24 months | Post SD vs. Pre SD | 1.3 | NS | 1.89 | NS | 1.18 | NS |
Stratified by living arrangement after SD | | | | | | | |
Live alone after SD | Post SD vs. Pre SD | 2.3 | ** | 3.0 | # | 2.0 | # |
Live with family members after SD | Post SD vs. Pre SD | 1.5 | ** | 2.3 | ** | 1.3 | NS |
Stratified by spouse’s heath status before SD(N = 548) | | | | | | | |
Spouse’s health status: Good | Post SD vs. Pre SD | 2.0 | ** | 2.3 | * | 1.8 | * |
Spouse’s health status: Not good | Post SD vs. Pre SD | 1.3 | NS | 1.9 | NS | 1.2 | NS |
# p < 0.1,* p < 0.05, ** p < 0.01, *** p < 0.0001 |