Factors associated with caregiver burden among adult (19–64 years) informal caregivers – An analysis from Dutch Municipal Health Service data

Abstract Due to the ageing population and the rising prevalence of chronic diseases, it is expected that the demand on informal caregivers will increase. Many informal caregivers experience burden, which can have negative consequences for their own health and that of the care recipient. To prevent caregiver burden, it is important to investigate factors associated with this burden. We aimed to identify factors associated with caregiver burden in adult informal caregivers. Among a sample of adult informal caregivers (n = 1,100) of the Dutch region of Zaanstreek‐Waterland, perceived caregiver burden, demographic factors, caregiving situation, health‐related factors and socio‐financial factors were measured as part of the national Health Survey in 2016. Using univariate and multivariate logistic regression analysis, for which a backward selection method was applied, associations with caregiver burden were studied. In the multivariate model, time spent providing informal care was significantly associated with perceived caregiver burden, with an odds ratio (OR) [95% confidence interval] of 7.52 [3.93–14.39] for those spending >16 hr compared to 1–2 hr on informal care. Also providing care to their child(ren) (OR: 2.55 [1.51–4.31]), poor perceived health (OR: 1.80 [1.20–2.68]) and loneliness of the caregiver (OR: 2.05 [1.41–2.99]) were significantly associated with caregiver burden. To possibly prevent and reduce informal caregiver burden, factors associated with such burden should be intervened on. As such, special attention should be paid to caregivers who provide many hours of care or provide care to their child(ren), as well as those who have a poor perceived health themselves and/or experience feelings of loneliness.

to decentralise care provision (Eurocarers, 2017). As such, the responsibility for long-term care has partly shifted from healthcare facilities to family and friends of patients, which has resulted in a growing need for these informal caregivers to maintain a sustainable healthcare system (Arno, Levine, & Memmott, 1999;Deeken et al., 2003). In the Netherlands, responsibility for supporting and guiding informal caregivers has been allocated to municipalities, and many governments of other countries also aim to decentralise care provision (Eurocarers, 2017). This policy change is likely to have important effects on the burden of informal caregivers (Kelders et al., 2016). In the Netherlands it was shown that 14% of the population (i.e. those aged ≥19 years) provided informal care in 2016, of whom 14% feel 'heavily burdened' as a consequence of providing this care (Statistics Netherlands, 2017).
Caregiver burden has been defined as the distress informal caregivers experience as a result of providing care, influenced by characteristics of the patient, the individual caregiver and the environment (Grant et al., 2013). Previous studies have reported that a perceived high burden among informal caregivers can have negative consequences for their own mental and physical health (Cuijpers, 2005;Schulz & Beach, 1999;Schulz, O'Brien, Bookwala, & Fleissner, 1995;Shiba, Kondo, & Kondo, 2016), as it has been associated with depression (Schoenmakers, Buntinx, & Delepeleire, 2010), stress (Vitaliano, Zhang, & Scanlan, 2003), and low self-efficacy and well-being (Pinquart & Sorensen, 2003). Moreover burdened informal caregivers are often impaired in their caregiving, which could have an impact on the health of the care recipient (Shiba et al., 2016).
Previous studies have identified a number of factors associated with burden among those who provide informal care to people with a chronic illness (Karakis et al., 2014;Pinquart & Sorensen, 2011;Vaingankar, Subramaniam, Abdin, He, & Chong, 2012). This includes demographics (e.g. gender, age, marital status, educational level, employment status and household composition), caregiving situation (e.g. the number of hours spent at providing care and the type of relationship with the care recipient), health-related factors (e.g. perceived health, chronic diseases and mental health) and socio-financial factors (e.g. loneliness, financial difficulties, mastery of life, contact with neighbours and social cohesion in neighbourhood). However, information on factors associated with caregiver burden in the general adult population, in particular from data collected after recent policy changes, is scarce.
The present cross-sectional study aims to investigate which factors are associated with high-perceived caregiver burden of adult (19-64 years) informal caregivers. In order to do so, we studied a range of factors in a large sample of adults from a population-based cohort of adult inhabitants of Zaanstreek-Waterland, the Netherlands. Insights from this study might guide policy and interventions aimed at prevention of caregiver burden.

| Study design and participants
Conducting and reporting of this study were done in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement guidelines (von Elm et al., 2007). The medical ethical committee of Amsterdam UMC, location AMC has declared that the Dutch Medical Research Involving Human Subjects Act (WMO) does not apply to this study and that official approval of this study is not required by the medical ethical committee is thus not required.
This study was performed using cross-sectional data, gathered by the Health Survey of the Public Health Services of Zaanstreek-Waterland and Statistics Netherlands. The Health Survey is a nationwide survey conducted by Dutch Public Health Services in collaboration with the National Institute for Public Health and the Environment (RIVM) and Statistics Netherlands, with the purpose to get insight into the health of citizens on a local, regional and national level. The Health Survey is repeated every 4 years. For the Health Survey in Zaanstreek-Waterland (a region existing of eight municipalities north of Amsterdam, the Netherlands), a random sample of adult (19-64 years) citizens in the Zaanstreek-Waterland area received an information letter as well as a hardcopy version of the questionnaire and a code to access the digital version of that same questionnaire in the autumn of 2016. While a comparable questionnaire was sent out to older citizens (>64 years), there were some important differences in the questions asked making it difficult to combine these data with those from the adult questionnaire. Therefore, only participants aged 19-64 years and providing informal care are included in the present study.

| Measures
Self-reported data on informal care and caregiver burden, as well as associated factors and descriptive variables were obtained. The list of

What is known about this topic
• Demands on informal caregivers are likely to increase as a result of the ageing population and increasing prevalence of chronic disease.
• Informal caregiver burden can have negative consequences for caregiver's own health and that of their care recipient.
• Up to date information on factors associated with caregiver burden in the general adult population is scarce.

What this paper adds
• We identified factors associated with caregiver burden in Dutch adult informal caregivers.
• We found that time spent providing informal care, providing care to child(ren), poor perceived health and loneliness were significantly associated with perceived caregiver burden.
• These factors should be intervened on in order to prevent and reduce informal caregiver burden. associated factors was composed based on the apparent association with caregiver burden as obtained from the literature (Karakis et al., 2014;Pinquart & Sorensen, 2011;Vaingankar et al., 2012).

| Informal care and caregiver burden
According to the Dutch Public Health Service and RIVM (RIVM, 2016), informal caregiving is defined as the unpaid nonprofessional care that people provide to a person from his or her social network, such as a partner, child or friend, if this person is sick, disabled and needs help for a longer period. Here, 'a longer period' is defined as providing care for at least 3 months and 8 hr per week (Statistics Netherlands, 2017). Participants were asked whether they were providing informal care in the past 12 months (yes/no), if they provided informal care right now (yes/no), for how many hours they were providing informal care per week and for how long they provided care (at least 3 months or less). Only participants who were currently providing informal care for a 'longer period' were included for further analyses.
The main dependent variable, perceived caregiver burden, was measured according to methods of Statistics Netherlands and the Dutch Public Health Services (RIVM, 2016), using a 5-point Likert scale (Allen & Seaman, 2007). Informal caregivers who reported that they were 'not (really) burdened' or 'somewhat burdened' were defined as not burdened. Informal caregivers who filled in that they were 'quite heavily burdened', 'very heavily burdened' or 'overburdened' were defined as burdened. This dichotomised variable was used in the analysis.

| Demographic factors
The following demographic factors were collected: gender (male/ female), age (categorised into 19-34, 35-49 and 50-64 years), marital status (dichotomised into living together with a partner vs. living alone). Moreover the highest completed level of education was assessed according to the Standard Education Classification of Statistics Netherlands (i.e. primary, secondary and tertiary level education). Employment status was assessed by asking participants whether they have paid work, and if so, for how many hours per week. Responses were dichotomised into ≥20 hr/week of paid work and <20 hr/week of paid work or being unemployed. Finally, household composition was questioned with outcome categories dichotomised into single parent families (living together with child(ren), but not with their partner, parent(s) or other adult(s)) versus non-single parent families.

| Caregiving situation factors
Caregiving situation was characterised by time (in hours) invested in providing informal care and the type of relationship between the informal caregiver and the care recipient. Participant's average time invested in providing informal care (including travel time) was categorised into 1-2, 3-4, 5-8, 9-16 and >16 hr per week. The informal caregiver's self-reported relationship with their care recipient was dichotomised into informal caregivers who provided care to their child(ren) (in law) versus recipients with other relationship types (spouse/partner, parents (in law), other family members and neighbours, friends and acquaintances). This categorisation was made since caregivers who provide care to their child(ren) (in law) stood out with respect of the other categories in terms of their caregiver burden (Table 1). Informal caregivers who provided care to more than one care recipient were excluded, to be able to assess the independent association of the relationship between informal caregivers and their recipient on caregiver burden. However, there were no clear differences in participant characteristics and caregiver burden between informal caregivers who provided informal care to either one or more than one care recipients (data not reported in this paper).

| Health-related factors
Regarding health-related factors, perceived health, having chronic diseases and mental health were assessed. Participants were asked to rate their perceived health, using a 5-point Likert scale (Oldenkamp, Wittek, Hagedoorn, Stolk, & Smidt, 2016) ranging from 'very poor health' (1) to 'very good health' (5). The highest two categories were defined as good perceived health, and the other three categories as poor perceived health. Participants were also asked whether they had any chronic disorder(s) (yes/no). To assess mental disorders, the Kessler Psychological Distress Scale (K10; Kessler et al., 2003) was used to assess anxiety and depressive symptoms in the past 4 weeks.

| Socio-financial factors
The 11 statements defined by Jong-Gierveld were used to measure loneliness, with participants responding by 'yes' (1), 'more or less' (0) or 'no' (0). Summary scores were dichotomised into not lonely (0-2) and lonely (3-11;de Jong Gierveld & van Tilburg, 2010). Participants were also questioned about having financial difficulties by asking whether they had difficulties in getting by. Participants who filled in 'not at all' or 'no difficulties but have to pay attention to expenses' were categorised as having no financial difficulties, while participants who filled in that they had 'some difficulties' or 'great difficulties' were categorised as having financial difficulties. The Pearlin Mastery Scale (PM; Pearlin & Schooler, 1978) was used to measure an individual's level of mastery, which is the extent to which someone regards his/her life chances as being under his/her own control. The PM is a 7-item scale including five negatively worded items and two positively worded items with response options from 'strongly disagree' to 'strongly agree', resulting in TA B L E 1 Characteristics of the study population (column 1), the informal caregivers (column 2) and the burdened informal caregivers (column 3). The study sample comprised 8,544 inhabitants of ZW (column 1), of whom approximately 16% were informal caregivers (column 2) and 15% of these informal caregivers were burdened (column 3) (1) Total study population 100% n = 8,544 (2) Informal caregivers 16.2% of (1) n = 1,289 (3) Burdened informal caregivers a 15.2% of (2) n = 196

| Statistical analysis
Of all variables, descriptive statistics were provided. Logistic regression was used to assess the association of aforementioned sociodemographic, caregiver situation, health-related and sociofinancial factors, that have shown to be associated with caregiver burden in previous research (Buchanan, Radin, & Huang, 2011;Karakis et al., 2014;Pinquart & Sorensen, 2011;Vaingankar et al., 2012), with caregiver burden. We developed a prediction model using three steps, according to earlier work (Huysmans et al., 2017). In the first step, a univariate analysis was applied to examine the association between each of the aforementioned variables and caregiver burden individually. Variables associated with caregiver burden with a conservative level of significance of p < .2 were retained. In the second step, a backward stepwise regression analysis was applied within each of the four categories (i.e. sociodemographic, caregiver situation, health-related and socio-financial) of factors, retaining variables being associated with caregiver burden (with level of significance p < .1). In the third step, all remaining variables from step two were analysed together, again using a (1) Total study population 100% n = 8,544 (2) Informal caregivers 16.2% of (1) n = 1,289 (3) Burdened informal caregivers a 15.2% of (2) n = 196 Note: Number of respondents might vary between variables due to item non-response a Burdened informal care givers are defined as informal caregivers who filled in that they were 'quite heavily burdened', 'very heavily burdened' or 'overburdened'. b Cases where informal caregivers filled in more than one option for care recipient type were not taken into regression analysis.

TA B L E 1 (Continued)
backward stepwise regression analysis with removal set at p < .1 (Tran, Arabnia, Mohammed, Naugler, & Far, 2015). Prior to step two and three, correlation of all variables were calculated. In cases where two variables were strongly correlated (with a Spearman r ≥ .7), only the variable with the largest standard deviation was retained to avoid multi-collinearity.
Effects were presented in odds ratios (OR) with 95% confidence intervals (95% CI). Goodness-of-fit was calculated by Nagelkerke's R 2 to evaluate explained variance of the final model. All statistical analyses were performed using SPSS (IBM Statistics version 24).

TA B L E 3 (Continued)
Our results of an association between time spent providing care and caregiver burden are in line with previous research (Karakis et al., 2014;Oliva-Moreno et al., 2018;Piran et al., 2017;Prevo et al., 2018;Schulz & Sherwood, 2008;Tadema & Vlaskamp, 2010). For example, Kenny, King, and Hall (2014) found that providing informal care for ≥20 hr per week was associated with worse physical and mental health (Kenny et al., 2014). Another study showed that the capacity of informal caregivers to continue to care and to pursue other activities, such as work and leisure activities, can be disrupted when providing too many hours of care (Scherpenzeel & Storms, 2016).
We also found higher caregiver burden among informal caregivers who provide informal care to their own child(ren), compared to informal caregivers who provide care to other care recipients.
This confirms earlier research (Piran et al., 2017;Raina et al., 2005), showing that taking care for child(ren) with a chronic disease is a main challenge for involved parents (Tadema & Vlaskamp, 2010).
Moreover providing such care may impact on the physical and psychological health of the informal caregivers and affect the whole family function (Raina et al., 2005). Our findings are in contrast, however, with earlier work showing that burden was highest among those providing care to partners (van Groenou, de Boer, & Iedema, 2013). It is important to note that the latter study was conducted among older (>55 years) adults, which could explain the different findings.
Our result of an association between poor perceived health and caregiver burden is in line with previous studies (Cormac & Tihanyi, 2006;Karakis et al., 2014;Rowe, McCrae, Campbell, Benito, & Cheng, 2008 illness, compared to non-caregivers (Shaw et al., 1999). In line with this vision, poor perceived health could be interpreted as a negative consequence of caregiving (Buyck et al., 2011). However, some informal caregivers might already have had poor health before they started to provide care, as there are indications that unhealthy persons will engage more often in providing informal care than healthy persons (Brouwer et al., 2004;Savage, Fisher, & Birch, 2007).
The current study confirms that there is more loneliness among burdened caregivers compared to non-burdened caregivers (Beeson, 2003;Shah, Wadoo, & Latoo, 2010;Vasileiou et al., 2017). Due to our cross-sectional design, loneliness can be both an antecedent and a consequence of providing care. Providing informal care has been shown to be a commonly perceived stressor (Bevan, 2015), which can increase the desire for a supportive social network (Clyburn, Stones, Hadjistavropoulos, & Tuokko, 2000). When the social network is insufficiently supportive, feelings of loneliness may arise (van den Broek & Grundy, 2018). There are indications that informal caregivers socially isolate themselves, due to limited time and energy to invest in maintaining other relationships (Rand, Malley, Netten, & Forder, 2015;van de Maat, Vermaas, & Kruijswijk, 2012). In the current study, we analysed informal caregivers who were 19-64 years of age. This group, compared to older caregivers, is more likely to engage in other societal participation activities, including work. The combination of caregiving and other societal participation activities could be an extra burden, but could also provide possibilities to reduce the burden of these care- givers. For example, the workplace setting of the caregiver may be an important support network to deal with loneliness and thereby caregiver burden.
Unlike earlier research, the current study did not find that demographic factors, such as gender, age, marital status, educational level, employment status and household composition, were associated with caregiver burden (Karakis et al., 2014;Kenny et al., 2014;McCullagh, Brigstocke, Donaldson, & Kalra, 2005;Piran et al., 2017;Prevo et al., 2018;Schrank et al., 2016). These previous studies were, however, limited to certain cultural groups, age groups (i.e. mainly older adults), educational level and patient groups with certain conditions. Moreover the current study is conducted with data after the legislation changes in the Netherlands.

| Strengths and limitations
An important strength of this study is the use of a large-scale sample of adults from the general population. It is the first study with data, after the decentralisation of the social domain in the Netherlands in 2015, examining the association of a range of demographic, caregiving situation, health-related and socio-financial factors with caregiver burden. Since many governments of countries also aim to decentralise care provision (Eurocarers, 2017), this study provides worthy results that could be used to support policy and intervention development.
Moreover data of the regional Health Survey were used (Statistics Netherlands, 2017), with widely accepted and validated questionnaires and definitions that are used by public health services throughout the Netherlands, Statistics Netherlands and the RIVM.
Another strength of this study is that the prediction model development was carried out in three steps with conservative cut-off points for p-values (Huysmans et al., 2017). By this procedure, variables were not unintentionally removed from the model.

TA B L E 3 (Continued)
Nevertheless, the final model, based on aforementioned four fac-

| Practical implications
The findings of the current study provide insights that could help early identification of informal caregivers who may experience caregiver burden. For reducing burden of informal caregivers, policy and interventions should focus on the amount of time spent providing informal care, informal caregivers providing care to their own child(ren), perceived health, and loneliness of the informal caregivers.
Facilities for 'respite care', a temporary or complete takeover of the care with the aim to reduce the burden on informal caregivers, could be improved. Respite care could reduce the number of hours providing informal care among informal caregivers and thereby provide a better 'caring -life balance' (Kenny et al., 2014). In addition, more leisure time could increase the time for social contacts which might reduce social isolation and loneliness (Vasileiou et al., 2017) and could provide the informal caregiver with the time to pay attention to their own health. To this end, availability, accessibility and transparency of respite care should be improved, for example by supporting informal caregivers in applying for respite care. Also task division (e.g. by using existing mobile applications) could be encouraged among family, friends and neighbours to help the informal caregiver with certain tasks and thereby reduce the number of hours they spent on providing informal care (Schrank et al., 2016).
To maintain the social network of the informal caregiver, it is also important that social contacts of the informal caregiver understand the impact of providing informal care. Interventions based on stimulating the social network of the informal caregivers might decrease loneliness (and withdrawal from society) among informal caregivers.

| CON CLUS ION
Time spent at providing informal care, providing informal care to their own child(ren), poor perceived health and loneliness of the informal caregiver were found to be associated with caregiver burden. Due to the rising demands on the healthcare system, the need for informal caregivers is likely to increase in coming years. To support the informal caregivers, policy and intervention programs should be developed that support informal caregivers by reducing the number of hours providing informal care, support parents who care for their child(ren), maintain the health and reduce loneliness among informal caregivers.

ACK N OWLED G EM ENTS
The authors thank GGD GHOR Nederland (the Association of Dutch Community Health Services) and the Netherlands National Institute for Public Health and the Environment (RIVM) and Statistics Netherlands (CBS) for developing and supporting the 'Health Survey -Adults 2016' study. The 'Health Survey Adults 2016' study in Zaanstreek-Waterland is predominantly financed by the Local health authorities in region Zaanstreek-Waterland. The authors also would like to thank Femmy Bijnsdorp for her help with critically reviewing our work.

CO N FLI C T O F I NTE R E S T
The authors declare no conflicts of interest.