One hundred eighty-seven dyads informal caregivers were part of this study, together with their relative with dementia in included in “Quality of life and functional ability of older adults living with dementia: outcomes of a Portuguese sample” (manuscript submitted to Ageing & Society, June 2021). All the sociodemographic data of included caregivers and PwD is presented in Table 1 and Table 2, respectively, of “Quality of life and functional ability of older adults living with dementia: outcomes of a Portuguese sample” (manuscript submitted to Ageing & Society, June 2021).
Briefly, the informal caregivers were family members of a PwD (32 wife’s, 29 husbands, 70 daughters, 38 sons, 6 sisters and 12 were other family members) and the mean age was 61 years old. Of the caregivers, 92.5% were living with their relative with dementia and seventy-five caregivers had a professional occupation (36.4% were in working in full-time work and 3.7% in part-time). The PwD were mostly female (74.3%), the mean age was 82 years old, with dementia diagnosis from 6.6 years (ranged from 1 and 20 years). The PwD scores obtained in “Quality of life and functional ability of older adults living with dementia: outcomes of a Portuguese sample”(manuscript submitted to Ageing & Society, June 2021), were in mean -0.005±0.31 (-0.536-0.766), indicating that PwD had complications in multiple domains of EQ-5D-3L, 36.9±1.4 (varying from 5 to 90) for EQ-VAS, 19.5±4.0 (15.5-23.5) for QoL-AD score, 15.8±4.8 (11.0 to 20.6) to BDRS score and 16.0±5.5 (10.5 to 21.5) for MMSE.
3.2 Quality of life
The frequency distributions for the EQ-5D-3L dimensions, as well as the means of EQ-5D-3L/EQ-VAS for the overall sample and by age group are presented in Table 1. Regarding the EQ-5D-3L index, the mean was 0.375±0.28 (-1.11-1.00), indicating that the informal caregivers had complications usually in two or three domains of EQ-5D-3L. According to L. Ferreira et al., 2014, the EQ-5D-3L tariffs for the Portuguese population varied between -0.536 (worst health state) and 1 (best health state). In our study, the minimum obtained for EQ-5D-3L index was the value of -1.112, revealing a lower minimal value (revealing worst states of health) and a lower mean value, in comparison to the Portuguese population norms (0.375±0.28 vs 0.758 (SD not shown), respectively) (L. N. Ferreira et al., 2014). Despites the low QoL of these caregivers, there were some cases (six cases, corresponding to 3.2%) reaching a maximum of 1 (the best health status) in the EQ-5D-3L index (“no problems” in all domains). Thus, one of the main limitations of the EQ-5D-3L regarding the ceiling effect, seen in this sample, in 3.2% of the cases and 2.7% in the EQ-VAS (value of 100). These six caregivers were a wife (85 years old), two daughters (23 and 43 years old), two sons (52 and 55 years old) and a grandchild (42 years old), the EQ-VAS varied from 55 to 90 and the time since PwD diagnosis was mostly less than 4 years, with exception to two cases where it reached 7 and 16 years. The EQ-VAS mean considering all cases was 60.7±19.7 (varying from 10 to 100), indicating that there were cases of very bad QoL related to extreme problems in multiple domains (probably informal caregivers with multiple health conditions, or caring for PwD in advanced stages), and some mild cases (caregivers of PwD in their initial phases, younger caregivers, or caregivers that dedicated less hours in caring tasks). As well as in the EQ-5D-3L, the mean value found for EQ-VAS was lower when comparing to Portuguese population norms (60.7±19.7 vs 74.9 (SD not shown), respectively) (L. N. Ferreira et al., 2014).
In our study, there were statistical differences regarding both EQ-5D-3L and EQ-VAS between age groups. In detail, for the EQ-5D-3L index, there were differences between the group of caregivers aged 70 or more years and the group aged 30-49 years old (p=0.002), and between 50-59 years old and 30-49 years old (p=0.01), with the EQ-5D-3L mean value being higher caregivers aged 30-49 years old, so, younger caregivers. Additionally, the older group (aged 70 or more years old), differed from 30-49 (p=0.001) and 50-59 years old groups (p=0.02) for the EQ-VAS score, presenting the lowest value. Also, there were differences by caregivers’ gender, with EQ-5D-3L being most of the times better in men’s comparatively to woman’s (with exception to 18-29 years old group), with this difference being statistically significant for the caregivers aged 50-59 years old (p<0.01).
Regarding all the cases (Table 1), the dimensions such as mobility and ability to do self-care tasks reveal mostly “no problems” (96.8% and 93.6%, respectively). In the dimension of usual activities, informal caregivers had mostly “some problems” (41.7%) and “no problems” in 35.3%. The dimensions that gathered a higher number of cases with “extreme problems” were pain/discomfort and anxiety/depression, (42.8% and 59.4% respectively), while 34.8% and 29.4% had “some problems” and 22.5% and 11.2% had “no problems” regarding these dimensions. In detail, it is possible to see in Table 1 that despites the group age, caregiver’s mobility was almost not compromised. Some of the caregivers aged 70 or more years presented some or extreme problems regarding self-care tasks, while the difficulties in usual activities (i.e., work, housework, family activities) started sooner, in the group aged 30-49 and aggravated in older ages. Finally, in the last two domains of EQ-5D-3L (pain/discomfort and anxiety/depression), the younger caregivers (18-29 years old), present almost “no problems”, while the remain groups revealed mostly “extreme problems”, mainly in the groups of 50-69 and more than 70 years old, and in anxiety/depression dimension (Table 1).
The comparisons between the results obtained for QoL in our study and the overall Portuguese population (L. N. Ferreira et al., 2014), showed that, despites the age group and the gender, the informal caregivers had lower QoL than the non-caregivers, at same age: 18-29 years old (0.59±0.36 vs 0.86 (SD not shown)), 30-49 years old (0.51±0.23 vs 0.82 (SD not shown)), 50-69 years old (0.35±0.27 vs 0.69 (SD not shown)), 70 or more years old (0.29±0.29 vs 0.60 (SD not shown)). The same occurred for the EQ-VAS score: 18-29 years old (70±19 vs 84.3 (SD not shown)), 30-49 years old (68±20 vs 79.8 (SD not shown)), 50-69 years old (62±19 vs 68.5 (SD not shown)), 70 or more years old (53±19 vs 62.1 (SD not shown)).
< Insert Table 1 about here >
3.3 Informal caregiver burden
The QASCI univariate values are presented in Table 2. The overall score in our study ranged between 48 to 155. The mean value was 99.6±17.8, corresponding to 53 (when the values are normalized to 0-100) meaning moderate burden. Regarding the seven dimensions of QASCI, the domain of financial burden was the one that reached high levels (73±26), revealing an extremely severe overload situation. The followed dimensions with higher scores were the emotional burden (66±21) and implications in personal life (62±21), meaning moderate burden. The remain four dimensions of QASCI presented lower values meaning slight burden regarding the reactions to demands, perception of efficacy and control mechanism, family support and satisfaction with caregiver role and their relative with dementia. In sum, the included caregivers presented high levels of burden regarding financial, emotional and personal life issues.
< Insert Table 2 about here >
3.4 Impact of dementia according to informal caregivers and PwD related factors
Briefly, the results of the instruments used in this study are presented in Table 3 according to the caregiver’s characteristics (familiar relationship with the PwD, professional situation, educational level, parenthood, years of care, hours of care). This table also integrates the characteristics of the PwD (daily occupation and educational level). Regarding the familiar relationship, the wife’s were the ones with lower QoL and higher burden (0.232±0.24 EQ-5D-3L, 50.9±18.5 EQ-VAS, 57±15 QASCI), followed by the daughters (0.359±0.27 EQ-5D-3L, 62.2±19.4 EQ-VAS, 53±12 QASCI). The husbands presented slightly high scores regarding QoL than wife’s but experienced lower levels of burden. The sons were the ones with high QoL and less burden. Other familiar relationships (grandchild, daughter-in-low…) presented the highest scores of QoL, however, the QASCI score was as high as the wife’s, being higher than daughters, husbands, and sons.
Regarding the professional occupation of the caregivers, the ones that were professional active presented better QoL and less overall burden (slight overload) than the ones that were professionally inactive (moderate overload).
The educational level revealed that for the caregivers that attend school for less or during six years there was no big differences regarding the QoL (by EQ-5D-3L) and burden experienced, however, more years of education revealed more QoL (by EQ-VAS). The caregivers that attended school over 9 years revealed highest levels of EQ-5D-3L than the previous ones, but the lower EQ-VAS score (53.7±19.7). Finally, caregivers that had 12 or more years of education revealed high scores of EQ-5D-3L and EQ-VAS, meaning higher QoL than caregivers with lower education. Moreover, the caregivers with more years of education presented lower scores in QASCI score, which means lower burden, still, moderate burden, despite the level of education.
The fact that caregivers had or not children did not affect QoL since the obtained values regarding the EQ-5D-3L index, EQ-VAS and QASCI score were very similar in both situations.
The years of care for the PwD indicated that more years of care corresponded to poor QoL and higher overload to the caregiver. When the duration of care was less than 2 years comparing to 3 years of care, the QoL of life increased (0.365±0.31 EQ-5D-3L, 66.2±19.7 EQ-VAS comparing to 0.531±0.24, 72.6±16.2, respectively). However, the QASCI score was very similar in the two situations, revealing slight overload. The period including four to fifteen years showed similar EQ-5D-3L index values and QASCI scores (moderate burden), but the EQ-VAS was high in when caregivers were caring for ten to fifteen years. Finally, when caring included more than fifteen years, only in nine cases, the EQ-5D-3L index decrease a lot comparatively to the remain situations and the QASCI score was the highest level when considering only the years of care (59±10). The same occurred with hours of care, where more hours corresponded to less QoL and higher overload. The impact of hours of care in decreasing QoL and increasing caregiver’s burden was noticed when caregivers spend more than five hours in caring tasks, being more evident when the hours of care increased to 7 or ten hours a day. So, the caregivers that were caring for their relatives for more than fifteen years and during more than ten hours a day (n=5) were extremely affected in both QoL and overall burden (0.05±0.21 EQ-5D-3L, 50±12.3 EQ-VAS and 67±5 QASCI score).
Considering the PwD characteristics, the resort to day centers during the week resulted in a slight increased QoL for the caregiver, particularly, in the EQ-VAS, however in both situations caregivers revealed moderate burden (56±15 if PwD visited the day center and 52±13 if PwD did not attended to day centers). If the PwD educational level is considered, more years of education resulted in better QoL for the caregivers (0.350±0.27 EQ-5D-3L, 59.4±20.8 EQ-VAS in caregivers of PwD that attended less than 6 years of school, comparing to 0.411±0.27 EQ-5D-3L, 64.0±17.7 EQ-VAS when PwD that attended to school for more than 12 years). The burden of the caregiver was nearly the same in all educational levels, being moderate.
< Insert Table 3 about here >
3.5 Correlations between QoL, overall burden and sociodemographic variables of informal caregiver and their relatives with dementia
The correlations for EQ-5D-3L and QASCI with each other and with sociodemographic variables of the informal caregivers and their relatives with dementia are presented in Table S1 of supplementary material. As expected, the convergent validity of the EQ-5D-3L index with QASCI score was moderate and negative, (rh0=-0.476, p<0.01), meaning that the increase of the burden leaded to low QoL. Regarding the QASCI seven dimensions, only two (emotional burden and implications in personal life) correlated in a moderate negative way with EQ-5D-3L, while the financial burden, and perception of efficacy and control mechanisms showed a weak negative correlation, meaning that burden decreased caregivers QoL. The EQ-VAS showed a strong correlation with EQ-5D-3L (rh0=0.610, p<0.01).
In what concerns to informal caregivers’ characteristics, the majority of the correlations founded were weak, however age (rh0=-0.314, p<0.05), educational level (rh0=0.334, p<0.01) and hours of care (rh0=-0.306, p<0.05) showed moderated correlations with QoL. The EQ-5D-3L correlated weakly with the majority of sociodemographic variables of the PwD (gender, age, educational level, EQ-5D-3L, mobility, anxiety/depression, EQ-VAS, QoL-AD, BDRS score and one of the BDRS domains) (see Table S1 of supplementary material).
The QASCI score correlated mainly weakly with the informal caregivers’ characteristics, but moderated correlations were founded for hours of care (p=0.351, p<0.01) and some PwD characteristics such as EQ-5D-3L (p=-0.363, p<0.01), and some of its dimensions (mobility, self-care, pain/discomfort and anxiety/depression (weak correlation) and usual activities (moderate correlation). Contrary to EQ-5D-3L, the QASCI score correlated moderately to QoL-AD score (p=-0.471, p<0.01), BDRS score (p=0.401, p<0.01) and with the three dimensions of BDRS score (changes in everyday life, changes in habits and changes in personality).
No statistically significant correlations were found for both instruments (EQ-5D-3L and QASCI) with the familiar relationship of daughter and husband, years of PwD diagnosis, PwD attend to day centers and MMSE scores.
Factors potentially associated with caregivers’ QoL and overall burden
The data analysis included linear regression models to identify the factors related to informal caregivers QoL and burden (Table S2 (model 1) and Table S3 (model 2) of supplementary material, respectively).
The model for QoL is well adjusted (F test: p<0.001) and explained about 39% of the observed variance in the value of the dependent variable (EQ-5D-3L score). Results showed inexistence of autocorrelation (Durbin Watson= 1.831), and inexistence of heteroscedasticity (Breusch–Pagan test, p>0.05). The model for overall burden is well adjusted (F test: p<0.001) and explained about 57% of the observed variance in the value of the dependent variable (QASCI score). Results showed inexistence of autocorrelation (Durbin Watson= 2.208), and inexistence of heteroscedasticity (Breusch–Pagan test, p>0.05).
In the first model, regarding informal caregivers QoL, the age of caregiver was statistically associated with changes in EQ-5D-3L index, despites its small effect, (B=-0.005, p<0.05). The familiar relationship of the caregiver with the PwD presented statistical differences when comparing husbands/wife’s, sons/daughters, brothers/sisters with other family members. Being brother or sister of the PwD resulted in better QoL, followed by being nephew, grandchild or son/daughter-in-law. The husband and wives were the ones where the increase on EQ-5D-3L index was smaller, as well as sons/daughters. Regarding the effect of burden in QoL, two of the QASCI domains were included in this model, with only one being statically associated with EQ-5D-3L index: implication in caregivers’ personal life (B=-0.49, p<0.05), meaning that the impact of dementia in caregivers life decreased its QoL. The characteristics of the PwD that were statically associated with caregivers EQ-5D-3L index included the age (B=-0.005, p<0.05) and the pain/discomfort where the increase of PwD problems and age resulted in decreased caregiver QoL. The caregiver gender and PwD mobility were not statistically associated with changes in EQ-5D-3L score, but the presence of these variables increased the model quality.
In what concerns to the informal caregiver overall burden (model 2, Table S3 of supplementary material), the pain/discomfort felt by caregivers, their familiar relationship with the PwD and professional situation were the caregivers-related factors statistically associated to its burden. In detail, the familiar relationship where the burden was more inflected (by increasing the QASCI score) was the nephew, followed by sons/daughters-in-law, grandchild, husband/wife’s, brothers/sisters and finally, the sons/daughters. This demonstrates that there were differences regarding the familiar relationship in caregivers perceived burden. For caregivers, being professional occupied in full or part-time resulted in less increased QASCI score, meaning lower overall burden comparatively to domestic or unemployed caregivers. The PwD characteristics statically associated with informal caregivers’ burden were the ability to perform self-care tasks (B=0.084, p<0.001), were more difficulties of the PwD resulted in higher burden, and changes in PwD personality (BDRS domain) (B=0.015, p<0.05). The QoL of the PwD (QoL-AD score), was also statically associated with informal caregivers’ burden, despites its low effect (B=-0.01, p<0.001).
The caregiver ability to perform daily activities, anxiety/depression, hours of care, educational level and the PwD mobility, pain/discomfort were not statistically associated with changes in QASCI score, but the presence of these variables increased the model quality. A briefly analysis of the effect of these variables showed that more hours of care and higher educational levels of the caregivers were associated to higher burden (Table S3 of supplementary material).