• 검색 결과가 없습니다.

C. Data Collection and Analysis Procedure

Ⅳ. RESULTS

A. Demographic Status of Subjects Surveyed

The number of subjects who responded to the survey for this study was 600 individuals in the beginning, among which, the samples who were removed by the Listwise deletion method because the response rate was low or the response was odd was 124. Therefore, the final sample for the analysis was 476 individuals, and the results of frequency analysis conducted for identifying such sociodemographic status appeared like in

<Table 2> below.

Among the total 476 individuals, there were 154 males (32.4%) and 322 females (67.6%). In the relationship with dementia patient, spouse occupied 183 individuals (38.5%), which was the highest, daughter 127 (26.7%), daughter-in-law 82 (17.2%), and son 72 (15.1%). In average monthly income level, 1.5 ~ 3.5 million won occupied 185 individuals (39.0%), which was the highest. In academic ability, high school graduate occupied 170 individuals (35.7%) and college graduate 161 (33.8%). The care period for dementia patient was 4.3 years on average, 26.5 days on monthly average, and 14.1 hours on daily average. And care burden showed 40.5 point on average in the range from point 0 to point 88, and depression showed 14.1 point on average in the range from point 0 to point 59.

<Table 2> Characteristics of Caregiver

Categories mean ± SD / N (%)

Age (y) 57.0 ± 13.0

Gender

Male 154 (32.4)

Female 322 (67.6) Relationship to patient Spouse (M) 76 (16.0) Spouse (F) 107 (22.5) Daughter-in-law 82 (17.2) Son 72 (15.1) Daughter 127 (26.7) Others (Son-in-law ) 12 (2.5) Monthly income(won) <150 153 (32.2)

150-350 185 (39.0)

≥350 136 (28.7) Education level No formal education 11 (2.3)

Elementary school 46 (9.7) Middle school 61 (12.8) High school 170 (35.7) College or university 161 (33.8) Graduate school 27 (5.5) Duration of caring (y) 4.3 ± 4.6 Caring day per month (d) 26.5 ± 8.2 Caring time per day (h) 14.1 ± 8.4

ZBI 40.5 ± 20.2

BDI 14.1 ± 10.0

* Characteristics of the categorical variables described as N (%), and characteristics of continuous variables described as Mean (SD)

And the results of descriptive statistic analysis conducted to look at the status of dementia patient state appeared as shown in the following

<Table 3>.

<Table 3> Dementia Patient state Categories mean ± SD

S-SDQ 21.3 ± 7.4 NPI 39.1 ± 36.1 S-ADL 5.9 ± 6.5 S-IADL 28.5 ± 13.1

Cognitive function (S-SDQ) was at 21.3 point on average in the range from point 0 min. to point 30 max. Problem Behavior (NPI) appeared at 39.1 point on average in the range from point 1 min. to 180 point max.

And Activities of Daily Living-Basic(S-ADL) appeared at 5.9 point on average in the range from point 0 min. to 24 point max. and Activities of Daily Living-Instrumental (S-IADL) appeared at 28.5 point on average in the range from point 0 min. to 45 point max.

B. Confirmatory Factor Analysis

The measurement tools used in this study are depression, patient state, care activity, care burden, and social activity. Dependent variable depression consisted of 21 question items, and independent variable patient state consisted of a total of 54 question items: largely 15 question items

for cognitive function, 12 for problem behavior, 12 for Activities of Daily Living-Basic, and 15 for Activities of Daily Living-Instrumental.

Intervening variable caregiving activity consisted of a total of two items:

largely 1 item for time and 1 item for number of days. And intervening variable care burden consisted of 22 question items and moderating variable social activity consisted of 7 question items.

To identify the configuration system and theoretical validity of this measurement tool, we conducted a measurement model analysis as one of the confirmatory factor analyses. But, we used Full Information Maximum Likelihood (FIML) for parametric estimation, since some missing values were allowed. Also the number of appropriate samples for structural model analysis can be calculated by the following formula.

Structure equation modeling’s appropriate number of samples = 1.5 × p(p+1)

*p=number of measurement variables

*Source: Bae(2011), 174.

When we substituted 8, number of measurement variables for the structural equation modeling presented in this study into this formula, the number of appropriate samples was drawn as 108.

But according to the Monte Carlo simulation, minimum number of samples for the structural equation modeling should be 200 or higher (Bae, 2011), and it was reported that if the sample size is more than 400, ML reacts sensitively and so the model fit gets bad (Bae, 2011). Given that the number of samples for the analysis in this study is more than 400, we

presented NFI(Normed Fit Index), IFI(Incremental Fit Index), and CFI(Comparative Fit Index) that did not react sensitively to the number of samples.

According to the analysis, the goodness of model fit appeared like in

<Table 4> below.

<Table 4 > Measurement model’s goodness of model fit

Index of fit NFI* IFI* CFI*

Standard >.90 >.90 >.90

Goodness of fit 0.905 0.920 0.918

* NFI(Normed fit index): is an index representing how much the proposed model has improved compared to the basic model. If it is generally 0.90 or higher, it is regarded as acceptable.

* IFI(Incremental fit index): has a value from 0 to 1. If it is 0.9 or higher, it is judged as a good fit.

* CFI(Comparative fit index): was developed from the perspective of indicating parameter and distribution of parent population to complement NFI. has a value from 0 to 1. If it is 0.9 or higher, it is judged as a good fit.

In the goodness of model fit of the measurement model, it appeared that NFI, IFI, and CFI all were 0.9 or higher and so the model was judged to be appropriate.

And the results of parametric estimation to identify the factor loading of the measurement model appeared like in <Table 5> and <Figure 5> below.

<Table 5> Measurement model’s factor loading

      B S.E. C.R. P β SMC

Patient state → S-SDQ 1       0.749 .514

Patient state → NPI 4.770 0.475 10.036 *** 0.516 .308 Patient state → S-ADL 0.845 0.069 12.336 *** 0.624 .338 Patient state → S-IADL 1.914 0.127 15.071 *** 0.823 .332

Care activity → Time 1       0.576 .678

Care activity → Day 1.011 0.172 5.891 *** 0.581 .390

Care burden → ZBI 1       0.555 .266

Caregiver

depression → BDI 1       0.717 .561

***p<.001

<Figure 5> Result of measurement model analysis

As a result of checking the factor loading, it appeared that the factor loading values of all measurement variables composed of patient state, care activity, care burden, and depression were at least 0.516 or higher, which obtained validity. And the explanation power of each measurement variable, SMC(Squared Multiple Correlations) was at least .266, which appeared 26.6% or higher.

And the correlation between latent variables appeared like in <Table 6>

below.

<Table 6> Correlation of latent variables

      r

Patient state ↔ Care activity 0.504**

Patient state ↔ Care burden 0.658**

Patient state ↔ Caregiver depression 0.394**

Care activity ↔ Care burden 0.567**

Care activity ↔ Caregiver depression 0.252**

Care burden ↔ Caregiver depression 0.876**

**p<.01

Patient state had a positive (+) correlation with all care activity, care burden, and depression. Care activity had a significant positive (+) correlation with care burden and depression. Care burden had a significant positive (+) correlation with depression.

C. Structural Model Analysis

This study conducted a structural equation modeling analysis to verify the research hypotheses presented and the analyzed goodness of model fit was shown like in <Table 7> below.

In the goodness of model fit, NFI, IFI, and CFI all appeared 0.9 or higher, and so the model was judged to be appropriate.

The results of parametric estimation for verifying the research hypotheses appeared like in <Table 8> below.

<Table 7> Goodness of model fit of the structural model

Index of fit NFI* IFI* CFI*

Standard >.90 >.90 >.90

Goodness of fit 0.905 0.920 0.918

* NFI(Normed fit index): is an index representing how much the proposed model has improved compared to the basic model. If it is generally 0.90 or higher, it is regarded as acceptable.

* IFI(Incremental fit index): has a value from 0 to 1. If it is 0.9 or higher, it is judged as a good fit.

* CFI(Comparative fit index): was developed from the perspective of indicating parameter and distribution of parent population to complement NFI. has a value from 0 to 1. If it is 0.9 or higher, it is judged as a good fit.

<Table 8> Result of parametric estimation in structural model

      B S.E. C.R. P β

Patient

state → Care

activity 5.172 .859 6.022 *** .504

Patient

state → Care

burden 1.121 .170 6.580 *** .499

Patient

state → Caregiver

depression .069 .021 3.291 *** .316

Care

activity → Care

burden .231 .109 2.123 .034* .250

Care

activity → Caregiver

depression .028 .043 1.781 .076 .192 Care

burden → Caregiver

depression .500 .061 8.132 *** .518

*p<.05 ***p<.001

As a result of analysis, the impact of patient state on care activity was β=.504***, which appeared significant in the positive (+) direction. And the impact of patient state on care burden was β=.499***, which appeared significant in the positive (+) direction. Also the impact of patient state on depression was β=.250, which appeared significant in the positive (+) direction. In other words, it can be known that if the patient state is worse, the caregiver activity and burden get higher and the depression becomes higher. Therefore, the Research Hypotheses H1-2, H1-3 and H1-1 presented in this study were adopted.

And the impact of care activity on care burden was β=.316***, which appeared significant in the positive (+) direction. In other words, it can be

known that the more the care activity, the higher the care burden.

Therefore, the Research Hypothesis H1-4 presented in this study was adopted.

In addition, the impact of care burden on depression was β=.518***, which appeared significant in the positive (+) direction. In other words, it can be known that the more the care burden, the higher the depression.

Therefore, the Research Hypothesis H1-6 presented in this study was adopted.

But the impact of care activity on depression was β=.192, which appeared not significant. Therefore, the Research Hypothesis H1-5 presented in this study was rejected.

The results so far can be represented like in <Figure 6> below.

SMC: Care activity(.254), Care burden(.507), Caregiver depression(.890)

<Figure 6> Result of direct effect

And the results drawn by verifying the mediating effect hypotheses presented in this study with Sobel test are like in <Table 9> below.

<Table 9> Result of the mediating effect analysis

Pathway z p Result

Patient state

→ Care activity

→ Caregiver depression

.647 .517 Rejected

Patient state

→ Care burden

→ Caregiver depression

5.138 .000*** Adopted

Care activity

→ Care burden

→ Caregiver depression

2.052 .040* Adopted

**p<.01 ***p<.001

As a result of analysis, the mediating effect of care burden between patient state and depression was z=5.138***, which had a significant mediating effect. And the mediating effect of care burden between care activity and depression was z=2.052*, which had a significant mediating effect. But the mediating effect of care activity between patient state and depression was z=0.647, which was not significant in mediating effect.

Therefore, the Research Hypotheses H2-2 and H2-3 presented in this study were adopted and H2-1 was rejected.

As a result of verifying the moderating effect of social activity in structural routes of dementia elderly patient state, caregiver activity,

burden, and depression, the goodness of model fit appeared like in <Table 10> below.

<Table 10> Goodness of fit of the moderating effect model

Index of fit NFI* IFI* CFI*

Standard >.90 >.90 >.90

Goodness of fit .887 .916 .913

* NFI(Normed fit index): is an index representing how much the proposed model has improved compared to the basic model. If it is generally 0.90 or higher, it is regarded as acceptable.

* IFI(Incremental fit index): has a value from 0 to 1. If it is 0.9 or higher, it is judged as a good fit.

* CFI(Comparative fit index): was developed from the perspective of indicating parameter and distribution of parent population to complement NFI. has a value from 0 to 1. If it is 0.9 or higher, it is judged as a good fit.

As IFI and CFI appeared 0.9 or higher in the goodness of model fit, the model was judged to be appropriate.

D. Analysis of Moderating Effect

1. Social Support’s Moderating Effect

The results from the parametric comparative analysis in multiple group analysis to verify the moderating effect of social support presented in this study appeared like in <Table 11> below. At this time, to make social support moderating variable, we recoded the group of less than median

value as ‘low group’ and the group of more than median value as ‘high

low 3.922 1.106 3.545 *** .439

1.302

high 6.083 1.240 4.907 *** .559

Patient

low .188 .177 1.061 .289 .163

.062

high .174 .088 1.977 .048* .286

Care

activity → Care burden

low .069 .033 2.119 .034* .258

1.546

high .068 .030 2.297 .022* .354

Care

activity → Caregiver depression

low .055 .022 2.507 .012* .425

1.268

high .014 .015 .891 .373 .144

Care

As a result of analysis, it appeared that the moderating effect of social support was not significant at C.R<±1.96 in all routes between patient state, care activity, care burden, and depression. Therefore, the Research Hypothesis H3-1, presented in this study was rejected. The analytical

results can be shown by group of social support like in <Figure 7> below.

<Figure 7> Result of the moderating effect of social support

2. Social Activity’s Moderating Effect

The results from the parametric comparative analysis in multiple group analysis to verify the moderating effect of social activity presented in this study are shown like in <Table 12> below. At this time, to make social activity moderating variable, we recoded the group of less than median value as ‘low group’ and the group of more than median value as ‘high group.’

low 2.523 1.029 2.453 0.014* 0.390

2.481*

high 6.776 1.371 4.941 *** 0.487

Patient

state → Care burden

low 1.680 0.253 6.650 *** 0.635

2.855*

high 0.727 0.219 3.326 *** 0.393

Patient

state → Caregiver depression

low 0.257 0.189 1.363 0.173 0.224

0.369 high 0.250 0.120 2.089 0.037* 0.435

Care

activity → Care burden

low 0.067 0.034 1.971 0.048* 0.278

0.034 high 0.051 0.22 2.299 0.022* 0.387

Care

activity → Caregiver depression

low 0.023 0.019 1.205 0.228 0.128

0.061 high 0.024 0.014 1.783 0.075 0.134

Care

burden → Caregiver depression

low 0.435 0.077 5.663 *** 0.830

0.724 high 0.523 0.093 5.605 *** 0.984

*p<.05 **p<.01 ***p<.001

As a result of analysis, in the impact of patient state on care activity, social activity’s moderating effect appeared (C.R=2.481*). In this route, the group of high social activity was β=.487***, which was significantly higher than the group of low social activity (β=.390*). In other words, the impact of patient state on care activity was found to be greater than the group of high social activity.

And in the impact of patient state on care burden, social activity’s moderating effect appeared (C.R=2.855*). In this route, the group of low social activity was β=.635***, which was significantly higher than the group of high social activity (β=.393***). In other words, the impact of patient state on care burden was greater in the group of low social activity.

Therefore, the Research Hypothesis H3-2, presented in this study was partially adopted. The analytical results can be shown by group of social activity like in <Figure 8> below.

<Figure 8> Result of the moderating effect of social activity

Ⅴ. DISCUSSION A. Summary and Discussions

This study was conducted to identify the mediating effect of care activity and burden and the moderating effect of social support and social activity in the relationship between dementia patient state and caregiver depression. Hypotheses were tested with structural equation modeling and the tested results were compared with the findings from the precedent studies as follows:

1. Considerations on Direct Effect

The precedent studies have reported about the relationship between dementia patient state and depression that care situations like dementia elderly’s Activities of Daily Living (ADL) and Problem Behavior (BPSD) and family caregiver’s coping style and social support had an impact on family caregiver depression (Ma and Kim, 1995; Kim, 2002) and that the higher the dependence in dementia elderly’s activities of daily living and the worse the problem behavior, the higher the family caregiver’s depression or stress (Brodaty and Luscombe 1998; Meshefedjian et al., 1998). As can be seen in Jang(2013) or Hong(2016) who reported a similar relationship between dementia patient state and depression, caregiver depression is influenced by dementia patient state. Such a relationship was also drawn from this study.

As reported in many precedent studies on the relationship between burden and depression in dementia patient caregivers (Ryu, 2017, Hong, 2017, Park, 2015), the variables that had the biggest impact on depression were subjective health condition, care period, occupation, self-efficacy, quality of marital relationship before caring, and care burden prior to social support (Park, 2015), and the correlation between dementia patient caregiver’s care burden and depression is high, but both cannot be said as the same concept, and there are different opinions about the directivity, and unlike the care burden that individuals feel because they care for the dementia patients, depression is regarded as a psychological construct that is influenced by care situations including care burden and other different states and care burden is seen as a predictor of depression (Ryu, 2017).

Also, considering Pearlin(1990) and Tennstedt et al.(1989) arguing that care burden had an impact on depression, the same results could also be found in this study.

As can be seen in studies (Kwon et al., 2002) that patient state causes care activity and studies (Song et al. 2013; Bae et al, 2006) that patient state had an impact on care burden, such care activity or burden is associated with the symptoms that dementia patients show. The Ministry of Health and Welfare(2009) reported that among dementia patient’s declined cognitive function, deteriorated activities of daily living, and problem behavior, patient’s problem behavior had the biggest impact on dementia caregiver’s burden, and restlessness and aggressive behavior which are shown in the group of behavioral and psychological symptoms aggravates the relationship with caregiver and the aggravated relationship in turn causes more severe behavioral problems from the patients who are

cared, which is a vicious cycle (Rachel et al., 2005). In fact, dementia patient’s unregulated behavioral and psychological symptom like aggression is the cause why nursing facilities refuse them to admit, which aggravates caregiver burden. Song et al.(2013) stated that dementia severity is judged as declined cognitive disorder, behavioral and psychological symptoms, and activities of daily living, and depending on dementia severity, main caregiver’s time to take care of patients increases and care burden also increases. Compared with the findings from such precedent studies, this study is in the same context with such studies.

Precedent studies present various causes of care burden that dementia patient caregivers feel, but among which, care activity is emphasized. Song et al.(2013) maintained that in case of dementia severity, its severity increases main caregiver’s time to take care of patient and increases care burden, too. Lee et al.(2008) claimed that if the effort and time to take care of dementia elderly increases due to dementia elderly’s problem behavior and dependence, caregivers feel considerable care burden in physical, psychological, and economic aspects. Multiple precedent studies on care burden stated that care period has an impact on care burden and especially mentioned the relationship between role constriction and care period, as one of the care burden factors that dementia elderly caregivers feel strongly. Due to role constriction that threatens the existing ordinary lives, caregiver loses his self-esteem, which leads to anger at the patient and the burden from role constriction aggravates as the time to take care increases and care period cannot be separated in understanding the care burden. The more the care period, the care burden increases, and if one should take care of a patient all day long, care burden is great, and care

burden lessens if sharing the burden rather than if taking care of a patient alone, which shows a close relationship between care period and care burden (Kwon et al. 2015). When compared with the findings from this study based on the reports from such precedent studies, the explanations about the relationship between dementia patient caregiver’s care activity and care burden are in the same context.

However, Park et al.(2005) claimed that as caregivers have to be with dementia elderly all day long, they are socially isolated and their physical health is aggravated and at the same time, care burden increases and depression gets worse. Compared with Kim(2000) or Bae et al.(2018) who reported that caregiver’s activity has an impact on depression, this study shows a different context in that insignificant results were drawn in this study.

2. Considerations on Mediating Effect

As a result of analyzing the mediating effect of care activity and care burden that this study presented as core research hypotheses based on different research findings presented in precedent studies, it was found that the mediating effect of care activity between patient state and depression was not significant, but care burden showed a significant mediating effect in the relationship between dementia patient state and caregiver depression and the relationship between care activity and depression, which showed differentiation from the precedent studies. The moderating effect of social support did not appear in all routes between patient state, care activity, care burden, and depression, but the moderating effect of social activity

관련 문서