Table 1 reports descriptive characteristics of listed variables in this study. Among 2003 respondents, female (N = 1183, 59.06%) and those over 65 years old (N = 1129, 56.45%) accounted for the majority. Additionally, more than half of the respondents were Non-Hispanic white (N = 1300, 65.07%), been married or coupled (N = 1256, 62.80%), and had high school education or above (N = 1682, 83.98%). Although 62.1% (N = 1242) of the elderly had ADL (Activities of daily living, ADL) difficulty(s), the proportion for a self-reported poor or fair health status was only 38.92% (N = 778). In line with the high proportion of self-fulfilling affective profile (N = 1277, 72.11%), older adults in this study reported averagely higher positive (Mean = 12.69, SD = 4.23, range 0–20) versus negative scores (Mean = 3.33, SD = 3.07, range 0–20) in emotional responses. Regarding vulnerability items, unable to do chores (N = 502, 25.10%) or pay the bills (N = 419, 20.9%) by themselves, had to help others with the chores (N = 428, 21.4%), and had delayed dental care (N = 434, 21.7%) were the most frequent ones. The mean score of positive aging attitudes was 20.09 (SD = 5.42, range 5–30), and other details are shown in Table 1.
Latent Class Analysis reveals that older adults in the current study developed three distinct patterns of vulnerability when confronted with the COVID-19 threat. Accordingly, older adults distributed unfairly across vulnerability types, as the proportion in three classes was 19%, 9%, and 72%, respectively. Details could be seen in Table 2
Table2. Comparing Models with Different Latent Classes: Fit indices (No. of Obs = 2003).
No. of Groups
|
loglikelihood
|
AIC
|
BIC
|
ssaBIC
|
Entropy
|
LMR
|
BLRT
|
Proportion in class
|
|
|
|
|
|
|
|
2LL
|
P
|
2LL
|
P
|
1
|
2
|
3
|
4
|
1
|
-9828.10
|
19690.21
|
19785.45
|
19731.44
|
|
|
|
|
|
|
|
|
|
2
|
-9315.25
|
18700.49
|
18896.57
|
18785.38
|
0.80
|
1018.27
|
0.00
|
1025.72
|
0.00
|
0.24
|
0.76
|
|
|
3
|
-9223.31
|
18552.62
|
18849.55
|
18681.16
|
0.80
|
182.54
|
0.00
|
183.87
|
0.00
|
0.19
|
0.09
|
0.72
|
|
4
|
-9167.00
|
18476.00
|
18873.77
|
18648.20
|
0.69
|
111.80
|
0.28
|
112.61
|
0.00
|
0.06
|
0.59
|
0.19
|
0.15
|
Notes.
ssaBIC: Sample-Size Adjusted BIC (n* = (n + 2) / 24); LMR: LO-MENDELL-RUBIN Test; BLRT: Bootstrapped Likelihood Ratio Test.
Selection Criteria: Model selection starts from one latent class and should stop if a) the AIC/BIC/ssaBIC begins to grow with another new group added; or b) the p value for LMR/BLRT turned nonsignificant with another new group added (p>0.05). Model with an Entropy of 0.8 or over is acceptable.
|
Figure 1 presents the detected vulnerability types based on the estimated probability of respondents from each latent class answered yes to vulnerability items. The solid black line refers to older adults who demonstrated slight vulnerabilities during the COVID-19 pandemic. Apart from several experiences commonly witnessed by populations, such as income deduction and spending growth, older adults in this group are prone to experience a certain degree of inconveniences regarding chores and bills. About 72% of older adults in this study had slight vulnerabilities in sustaining daily routine and could be comparatively healthier and more capable than two other types. Besides, about one in five of the older adults had a vulnerability in healthcare use. Comparatively, they had intense demands in surgery, doctor visits, dental care et al., but failed to accommodate due to lockdowns in COVID-19. Noteworthy, another 9% of older adults were comprehensively vulnerable when confronted with the COVID-19 threat. Besides elevated risks in life inconveniences and pandemic exposure, persons in this group are prone to experience healthcare shortages and economic hardships. A 20–50% probability of experiencing delayed survey, doctor visits, and dental care, along with income deduction, spending growth, food shortage, and financial due miss, went to older adults with dual vulnerabilities. Details could be seen in Fig. 1.
Table 3 further demonstrates each vulnerability type's sociodemographic and health-related characteristics and identifies the differences between them. As noted in Table 3, there was no significant difference in affective profiles between individuals from different vulnerability types. Comparatively, the female was prone to have either healthcare use vulnerability or dual vulnerabilities other than slight vulnerability, whereas older adults aged over 65 were more likely to experience slight vulnerability. The proportion of Non-Hispanic white presented significant differences between vulnerability types, which was highest in healthcare use vulnerability (72.47%) and lowest in the dual vulnerability type (39.13%). Besides, older adults with less education, had ADL difficulties, and been uncoupled were most likely to have dual vulnerabilities. Details are presented in Table 3.
Table 3. Distribution of Level-2 Affective Profiles and Level-1 Characteristics across the Identified Latent Vulnerability Groups.
Characteristics
|
Class Description
|
Overall chi-square/F Test
|
|
Slight Vulnerability
|
Healthcare Use Vulnerability
|
Dual Vulnerability
|
Chi-square/F
|
P value
|
N (%)
|
1439(71.84)
|
426(21.27)
|
138(6.89)
|
|
|
Level-2 Affective Profile
|
|
|
|
1.25
|
0.29
|
Self-fulfilling %
|
73.49
|
72.70
|
54.78
|
|
|
High affective %
|
1.25
|
1.05
|
3.48
|
|
|
Low affective %
|
23.69
|
23.88
|
31.30
|
|
|
Self-destructive %
|
1.57
|
2.36
|
10.43
|
|
|
Level-1 Features
|
|
|
|
|
|
Female %
|
57.31
|
63.62
|
64.49
|
6.33**
|
0.01 a, b
|
Age > 65 %
|
60.17
|
49.77
|
38.41
|
20.34***
|
<0.001 a, b
|
Non-Hispanic white %
|
65.37
|
72.47
|
39.13
|
14.96***
|
<0.001 a, b, c
|
Uncoupled %
|
36.14
|
35.92
|
52.17
|
4.05*
|
0.02 b, c
|
Less than high school %
|
17.10
|
10.56
|
21.74
|
10.81***
|
<0.001 a, c
|
Relatively poor %
|
49.68
|
47.17
|
48.15
|
1.48
|
0.23
|
Eligible for Medicaid %
|
10.70
|
7.04
|
22.46
|
3.96
|
0.05
|
With ADL Difficulties %
|
60.79
|
60.80
|
79.71
|
9.85***
|
<0.001 b, c
|
Poor health status %
|
7.87
|
7.06
|
6.52
|
0.85
|
0.49
|
Notes.
Significant difference by class (P<0.05 with multinominal regression detecting partial relationship between group membership and each variable after controlling for covariates): a: Slight VUL vs Healthcare use VUL; b: Slight VUL vs Dual VUL; c: Healthcare use VUL vs Dual VUL.
P<0.05 *, P<0.01 **, P<0.001 ***.
|
In Table 4, the relationships between vulnerability type, aging attitudes, and positive emotional responses during the COVID-19 pandemic are presented. As affective profile was hypothesized with an anchoring effect on emotional responses, this study used the affective profile as a Level-2 identity. Socioeconomic and health-related variables entered on the first step, and aging attitudes were entered on the second step to predict positive emotional responses. In the second model, positive aging attitudes presented a significant association with positive emotional responses towards the pandemic (B = 0.263, SE = 0.017), also explained an additional 6.27% of the variability in positive emotions over and above socioeconomic and health-related factors. However, the third model indicated that no significant differences in positive emotional responses existed between vulnerability types. With aging attitudes and vulnerability types controlled, had some college or higher education (B = 0.661, SE = 0.254) and lived with better economic situations (B = 0.605, SE = 0.192 for mid-level; B = 0.617, SE = 0.211 for rich) was associated with more positive emotional responses. On the contrary, persons who had ADL difficulties (B=-0.436, SE = 0.183) and perceived themselves as good or better health status were less likely to have positive emotions than their counterparts. See more details in Table 4.
Table4. Relationship between Aging Attitudes, Vulnerability Group and Positive Emotional Responses among Older Adults (Level-2 Identity: Affective Profile)
Variables
|
|
Model 1
|
Model 2
|
Model 3
|
Fixed-effects
|
|
B
|
S.E.
|
B
|
S.E.
|
B
|
S.E.
|
Constant
|
|
12.614
|
1.077
|
6.538
|
0.988
|
6.393
|
0.998
|
Gender (ref: Male)
|
Female
|
0.414*
|
0.176
|
0.254
|
0.167
|
0.250
|
0.167
|
Marital Status (ref: Married)
|
Partnered
|
-0.002
|
0.397
|
0.051
|
0.375
|
0.076
|
0.375
|
|
Uncoupled
|
-0.462*
|
0.192
|
-0.338
|
0.182
|
-0.356
|
0.182
|
Age (ref: <=65)
|
>65
|
-0.121
|
0.176
|
0.078
|
0.167
|
0.119
|
0.168
|
Education (ref: < High School)
|
High school diplomacy
|
0.442
|
0.258
|
0.438
|
0.244
|
0.425
|
0.244
|
|
Some college or higher
|
0.683*
|
0.268
|
0.687**
|
0.254
|
0.661**
|
0.254
|
Race (ref: Non-Hispanic White)
|
Hispanic
|
0.104
|
0.278
|
-0.204
|
0.264
|
-0.246
|
0.265
|
|
Non-Hispanic Black
|
0.754***
|
0.234
|
0.246
|
0.225
|
0.224
|
0.226
|
|
Others
|
0.706
|
0.411
|
0.292
|
0.391
|
0.282
|
0.391
|
ADL Difficulties (ref: None)
|
One or more
|
-0.746***
|
0.191
|
-0.420*
|
0.183
|
-0.436*
|
0.183
|
Self-rate Health (ref: Poor)
|
Fair
|
-1.006**
|
0.332
|
-0.551
|
0.316
|
-0.546
|
0.316
|
|
Good
|
-2.085***
|
0.342
|
-1.178***
|
0.329
|
-1.176***
|
0.329
|
|
Very good
|
-2.691***
|
0.384
|
-1.098**
|
0.377
|
-1.097**
|
0.377
|
|
Excellent
|
-3.682***
|
0.510
|
-1.284**
|
0.506
|
-1.365**
|
0.508
|
Medicaid Eligibility (ref: No)
|
Eligible
|
-0.172
|
0.297
|
-0.080
|
0.284
|
-0.082
|
0.284
|
Household Wealth (ref: Poor)
|
Medium level
|
0.608**
|
0.202
|
0.611**
|
0.192
|
0.605**
|
0.192
|
|
Relatively rich
|
0.622**
|
0.222
|
0.616**
|
0.211
|
0.617**
|
0.211
|
Aging Attitudes (AA)
|
|
|
|
0.263***
|
0.017
|
0.266***
|
0.017
|
VUL Group (ref: Slight VUL)
|
Healthcare use VUL
|
|
|
|
|
0.190
|
0.191
|
|
Dual VUL
|
|
|
|
|
0.591
|
0.328
|
Variance components
|
|
|
|
|
|
|
|
var (_cons)
|
|
1.936
|
0.712
|
1.579
|
0.586
|
1.607
|
0.597
|
var (Residual)
|
|
3.379
|
0.057
|
3.167
|
0.054
|
3.165
|
0.054
|
Obs
|
|
1752
|
|
1709
|
|
1709
|
|
Chi-square
|
|
244.35***
|
|
504.69***
|
|
509.58***
|
|
Ho: OLS versus HLM
|
|
<0.001
|
|
<0.001
|
|
<0.001
|
|
Notes: AA: Aging Attitudes, VUL: Vulnerability; * P<0.05; ** P<0.01; *** P<0.001. |
Table 5 focuses on negative emotional responses and explores their relationship with aging attitudes and vulnerability types. A set of hierarchical linear regressions was conducted with socioeconomic and health-related variables entered in the first step, and aging attitudes entered the second step to predict negative emotional responses. Positive aging attitudes explained 5.16% of the variance in negative emotions and presented a significant relationship with these responses (B=-0.194, SE = 0.014). The vulnerability type entered in the third model, revealing that older adults who had healthcare use vulnerability (B = 0.476, SE = 0.159) and dual vulnerabilities (B = 1.186, SE = 0.274) were prone to develop more passive emotions than their slightly vulnerable counterparts. Moreover, there was a significant moderation effect of aging attitudes and vulnerability type on negative emotional responses. Positive aging attitudes are critical for persons being dual vulnerable, as negative emotions reduced faster in older adults with dual vulnerabilities than the slight vulnerable counterparts when aging attitudes improved (B=-0.118, SE = 0.045). With aging attitudes and vulnerability type controlled, female (B = 0.542, SE = 0.140) was more likely to have negative emotions than male counterparts, whereas having Medicaid (B=-0.530, SE = 0.253) and being uncoupled (B=-0.336, SE = 0.152) associated with fewer negative emotional responses.
Table5. Relationship between Aging Attitudes, Vulnerability Type and Negative Emotional Responses among Older Adults (Level-2 Identity: Affective Profile)
Variables
|
|
Model 1
|
Model 2
|
Model 3
|
Model 4
|
Fixed-effects
|
|
B
|
S.E.
|
B
|
S.E.
|
B
|
S.E.
|
B
|
S.E.
|
Constant
|
|
2.930
|
0.871
|
7.762
|
0.951
|
8.513
|
0.988
|
8.204
|
0.989
|
Gender (ref: Male)
|
Female
|
0.493***
|
0.146
|
0.564***
|
0.140
|
0.553***
|
0.140
|
0.542***
|
0.140
|
Marital Status (ref: Married)
|
Partnered
|
-0.231
|
0.329
|
-0.281
|
0.314
|
-0.274
|
0.314
|
-0.268
|
0.313
|
|
Uncoupled
|
-0.283
|
0.158
|
-0.380*
|
0.152
|
-0.371*
|
0.152
|
-0.366*
|
0.152
|
Age (ref: <=65)
|
>65
|
-0.616***
|
0.147
|
-0.749***
|
0.141
|
-0.735***
|
0.141
|
-0.727
|
0.141
|
Education (ref: < High School)
|
High school diplomacy
|
-0.198
|
0.214
|
-0.162
|
0.205
|
-0.154
|
0.205
|
-0.114
|
0.205
|
|
Some college or higher
|
0.166
|
0.222
|
0.191
|
0.213
|
0.191
|
0.213
|
0.225
|
0.213
|
Race (ref: Non-Hispanic White)
|
Hispanic
|
-0.006
|
0.231
|
0.323
|
0.223
|
0.332
|
0.223
|
0.374
|
0.224
|
|
Non-Hispanic Black
|
-0.110
|
0.193
|
0.309
|
0.188
|
0.324
|
0.189
|
0.336
|
0.189
|
|
Others
|
-0.805*
|
0.340
|
-0.458
|
0.328
|
-0.433
|
0.327
|
-0.439
|
0.326
|
ADL Difficulties (ref: None)
|
One or more
|
0.302
|
0.158
|
-0.002
|
0.153
|
0.001
|
0.153
|
0.009
|
0.153
|
Self-rate Health (ref: Poor)
|
Fair
|
0.560*
|
0.272
|
0.196
|
0.262
|
0.196
|
0.262
|
0.197
|
0.262
|
|
Good
|
0.740**
|
0.280
|
0.109
|
0.273
|
0.105
|
0.273
|
0.128
|
0.273
|
|
Very good
|
1.488***
|
0.316
|
0.380
|
0.315
|
0.375
|
0.314
|
0.402
|
0.314
|
|
Excellent
|
2.347***
|
0.420
|
0.763
|
0.422
|
0.789
|
0.422
|
0.706
|
0.422
|
Medicaid Eligibility (ref: No)
|
Eligible
|
-0.433
|
0.244
|
-0.530*
|
0.236
|
-0.525*
|
0.236
|
-0.530*
|
0.235
|
Household Wealth (ref: Poor)
|
Medium level
|
-0.017
|
0.167
|
-0.038
|
0.161
|
-0.045
|
0.161
|
-0.055
|
0.161
|
|
Relatively rich
|
0.058
|
0.183
|
0.003
|
0.176
|
-0.005
|
0.176
|
-0.003
|
0.175
|
Aging Attitudes (AA)
|
|
|
|
-0.194***
|
0.014
|
-0.194***
|
0.014
|
-0.183***
|
0.016
|
VUL Group (ref: Slight VUL)
|
Healthcare use VUL
|
|
|
|
|
0.746***
|
0.159
|
0.948
|
0.620
|
|
Dual VUL
|
|
|
|
|
1.186***
|
0.274
|
3.245***
|
0.841
|
Inter (ref: Slight VUl *AA)
|
Healthcare VUL * AA
|
|
|
|
|
|
|
-0.010
|
0.030
|
|
Dual VUL * AA
|
|
|
|
|
|
|
-0.118**
|
0.045
|
Variance components
|
|
|
|
|
|
|
|
|
|
var (_cons)
|
|
1.454
|
0.715
|
1.528
|
0.726
|
1.702
|
0.737
|
1.673
|
0.628
|
var (Residual)
|
|
2.754
|
0.047
|
2.612
|
0.045
|
2.609
|
0.045
|
2.604
|
0.045
|
Obs
|
|
1709
|
|
1668
|
|
1668
|
|
1668
|
|
Chi-square
|
|
105.29***
|
|
299.05***
|
|
374.92***
|
|
383.17***
|
|
Ho: OLS versus HLM
|
|
<0.001
|
|
<0.001
|
|
<0.001
|
|
<0.001
|
|
Notes: AA: Aging Attitudes, VUL: Vulnerability; * P<0.05; ** P<0.01; *** P<0.001. |