Why is the composition of older adults’ care network associated with psychological wellbeing: an application of the self-determination theory

Abstract

Objectives

Older care recipients have different types of care networks, varying from spouse-only to large mixed care networks, that add to different levels of wellbeing. Applying Self-Determination Theory (SDT) to the care context, we argue that the care network composition may foster or hamper the three basic needs for wellbeing: relatedness, autonomy and competence.

Method

Data are from ten observations between 1992 and 2022 of the Longitudinal Aging Study Amsterdam (N = 18,434 observations from 4,837 older Dutch adults). Five care network types are used: no care, partner, informal, formal or privately paid care. Mixed-hybrid-multilevel regression analysis of depressive symptoms as measure of wellbeing is applied on care network type and loneliness, mastery and care sufficiency as indicators of the three basic needs for wellbeing.

Results

Receiving care from a partner care network is, compared to the formal care network, the most negatively associated with depressive symptoms, followed by informal care and privately paid care. Differences in care network types existed in loneliness and care sufficiency, but not in mastery, and in part explained the association between care network types and depressive symptoms. Results of between and within effects are comparable.

Conclusion

Using a rich data set and advanced methodology support the hypotheses that formal care networks hamper wellbeing due to insufficient care and increased loneliness, in particular compared to partner and informal care. The role of mastery was less important, possibly because it does not measure care related level of control.

Keywords:

• Caregiving
• informal care
• care networks
• self determination theory
• long term care

Introduction

Population ageing creates an increasing demand for long-term care (LTC). LTC is any assistance in personal care or household activities due to the care recipients’ health-related inability to perform these tasks and is generally provided by (a mix of) informal and formal caregivers (Colombo & Mercier, 2012). Health impairment contributes to LTC use (Kwak et al., 2014) ánd to low psychological wellbeing (Steptoe et al., 2015), but the association between care use and wellbeing is less clear. Moreover, the association seems to differ by type of caregiver. In general, the use of formal care contributes to lower wellbeing, whereas the use of partner care and privately paid care contributes to higher wellbeing (Broese van Groenou, 2020; Pepin et al., 2017). But in contrast, other studies report that formal care use increases levels of wellbeing, while informal care use lowers wellbeing (Lee et al., 2018; Wylie et al., 2024). In both cases, the explanation is that the care use may lead to a loss of autonomy and control and becoming a burden to others (Kwak et al., 2014; Morgan & Brazda, 2013). These studies point at the importance of self-determination in the context of care use. More insight in underlying mechanisms of the association between informal and formal care use and wellbeing is thus needed. It is important to understand how we can organise the care in such a way that it contributes to the wellbeing of the older care recipient.

Our study builds on a previous, cross-sectional study (Broese van Groenou, 2020) and asks ‘How is the composition of the care network associated with the wellbeing of the older care recipient?’ We will expand current insights by applying the Self-Determination Theory (SDT) to the context of care, and by using a longitudinal dataset covering 30 years of observations, allowing for differentiation between and within subject effects. The presence of depressive symptoms is used as indicator of a lack of psychological wellbeing. We use an a prior set of care network types as it is the mere presence of specific caregiver type that seem to contribute to wellbeing: (1) a partner care network, (2) an informal care network, (3) a formal care network (with or without other caregivers), and (4) a privately paid care network only.

Theoretical framework

SDT posits that three psychological needs are innate and universally required for wellbeing (Ryan & Deci, 2017): relatedness, autonomy, and competence. The need for relatedness captures the wish to interact with others, the ability to feel connected and the capacity to experience caring from and for others (Ryan & Deci, 2020). People feel relatedness most typically when they feel supported by others, as this fosters a sense of connectedness, love, and understanding within relationships (Vansteenkiste et al., 2010). Autonomy refers to being the perceived source of one’s own behaviour, whereas competence indicates feeling effective in one’s ongoing interactions with the social environment and experiencing opportunities to exercise and express one’s capacities (Ryan & Deci, 2020).

When any of these three basic psychological needs is frustrated or neglected the individual will show diminished wellbeing (Ryan & Deci, 2017). We use the SDT as an older adult with care need is at risk of not fulfilling these basic needs. Our central argument is that care network types differ in the extent they support or frustrate the needs of the care recipient, with consequences for their wellbeing.

Hypotheses

The concept of relatedness suggests that the bond with the caregivers is important for wellbeing. The degree to which the care recipient is able to develop and maintain supportive relationships with caregivers could foster feelings of relatedness. Relatedness reflects that one is embedded in a care network, is in contact with others and feels important to others (Ashida & Heaney, 2008). It is more likely that care provision from the partner and, to a lesser extent, an informal care network, may in particular reflect feelings of relatedness, which may be less the case when receiving care from formal or privately paid care networks (McCamish-Svensson et al., 1999). Although it is possible that formal caregivers and care receivers develop warm and friendly relationships, it is overall less likely that one feels related or connected to care networks which mainly consist of formal or privately paid caregivers (Barbieri & Ghibelli, 2020). A lack of informal care implies that core social relationships are lacking or not well functioning. We use loneliness as an indicator of a lack of relatedness, as loneliness indicates a low functioning or absent core social network (O’Rourke et al., 2018). Therefore, we hypothesize that older adults that receive care from a partner or, to a lesser extent, an informal care network type, have fewer depressive symptoms than those with a formal or privately paid care network type, due to lower levels of loneliness (H1).

Autonomy and feeling competent in the care situation may be the result of the care recipient being capable to define when, what type and from whom care is used. This thus refers on the one hand to skills or capabilities of the care recipient oneself, who is able to control the care situation and receive care according to one’s standards and needs. In Western cultures, where autonomy and self-reliance are highly valued, care receipt may lead to the loss of one’s self-perceived identity as an independent person (Kwak et al., 2014). We use perceived mastery as an indicator of being in control in the care situation (Lee et al., 2018; Wylie et al., 2024). For the levels of autonomy and competence, receiving sufficient care is important (Kadowaki et al., 2015). It can be acknowledged that the care network may also be an important factor here: providing care in such a way that the care recipient feels in control and receives sufficient care, which will also foster feelings of autonomy and competence. This indicates that one is able to receive the care that one needs.

We assume that a partner and privately paid care network help the most in promoting autonomy and competence. Care from the partner, especially after their retirement, can be received 24/7 and a privately paid caregiver can be hired when needed. This suggests that these types of care are more easy to control and provide more often the care that is needed (Potter, 2019). In contrast, formal care has restrictions and rules, such as limited working hours. Informal caregivers other than the partner may have other duties, as work and family life, making them less available to provide care (Steiner & Fletcher, 2017). So, these types of care are less easy to control and may contribute to insufficient care. Some studies report that formal care contributes to a higher level of control, as formal care lowers the dependency on informal care (Lee et al., 2018; Wylie et al., 2024). Either way, mastery and care sufficiency could explain differences in wellbeing among care network types. Therefore, we hypothesize that differences in depressive symptoms among care network types are due to differences in mastery and care sufficiency as reported by the care recipient (H2).

Other factors that may influence the care network type used and depressive symptoms

In addition to the need factors of the SDT on the level of the care network types, there are individual factors that impact on both depressive symptoms and the type of care network used. First, care recipients’ health limitations that necessitate caregiving tend to be conflated with their wellbeing (Kwak et al., 2014; Lin & Wu, 2011) as well as with the care network type used. If the need is high and the care is demanding, the likelihood of receiving formal care (in addition to informal care) is higher than when the care tasks are less complex and higher care needs are associated with higher depressive symptoms (Steptoe et al., 2015). Therefore, we added physical and cognitive functioning, number of chronic diseases and age to the models. Second, concerning gender, on average, men receive care from smaller non-kin care networks than women and women report higher depressive symptoms (Andersson & Monin, 2018). Third, higher education is associated with higher income levels and thus may facilitate buying privately paid care and also higher education is associated with less depressive symptoms (Smith & Wesselbaum, 2024). Finally, as we cover a period of 30 years, many societal changes may have taken place over the years. These changes concern retreat of the welfare state, individualization, digitalization and economic crises. As these macro level changes are beyond the focus of the study, we only control for time in our analyses, and refer to this limitation in the discussion.

Method

Data

Data were used from the Longitudinal Aging Study Amsterdam. LASA is an ongoing longitudinal study of older adults (aged 55+) in the Netherlands that started in 1992 and focuses on the physical, emotional, cognitive and social functioning. Respondents were randomly selected from the registers of 11 municipalities in three regions in the Netherlands that vary in terms of religion and level of urbanisation, in such a way that the sample is representative of the Dutch older adult population in the Netherlands. In 1992/93 the baseline sample, containing 3107 respondents aged 55 to 84 years, was drawn from both urban and rural regions throughout the Netherlands. In 2002 and 2012 two additional cohorts of 1002 and 1023 respondents were drawn, from the same sampling frame but of different birth cohorts. Every three or four years additional measurement waves were conducted. Respondents were visited at home by trained interviewers who collect the data. For detailed information on LASA, the topics, measurement moments and methods, response rates and reason for drop outs, see the website and cohort papers (https://lasa-vu.nl/; Hoogendijk et al., 2016, 2020; Huisman et al., 2011).

Analytical sample

We used the data from the main face-to-face interviews in ten waves (1992 till 2022): 19,872 observations from 5132 respondents (on average 3.8 observations per respondent). In this sample some data were missing because of a short or terminated interview, or too many missing items to construct a scale. First we excluded those observations without valid data on our outcome variable (244 observations from 45 respondents). Second, we imputed the missing data with data from the earlier observation, if the network type in that measurement moment was the same, and if not available or not the same network type, we excluded those observations (in total 4,6% observations and 4% of the respondents). We excluded those older adults that were institutionalized (471 observations from 103 respondents). That resulted in a sample with 18,434 observations from 4,837 respondents. From the 4,837 respondents, 2,499 were women (9,795 observations) and 2,338 men (8,639 observations). 74% of the respondents (3,582) received care on at least one observation (7,762 observations).

Measurements

Depressive symptoms are measured with the Center for Epidemiologic Studies Depression Scale (CES-D; Radloff, 1977). The answers ranged from 0 ‘rarely or never’ to 3 ‘mostly or always’. CES-D can be understood as measuring a bipolar continuum that ranges from wellbeing to depression (Siddaway et al., 2017). After coding all the items in the same direction, all 20 items were summed, with higher scores indicating a higher number of depressive symptoms (range 0-60).

Care network type: Participants were asked whether they received help (no, yes), with at least one of two tasks: personal care or household care. And if so, from who they receive that care: the partner, other informal caregiver(s) (children, other kin, neighbours/friends/acquaintances), publicly paid caregiver(s) (district nurse, help at home), or privately-paid caregiver(s) (private help or personnel at home) The following exclusive categories were created: (1) ‘no care’; (2) ‘partner care network’ (care from the partner, with or without others; no formal care), (3) ‘informal care network’ (informal with or without privately paid care use; no partner and no formal care), (4) ‘mixed formal care network’ (formal care use, with or without other caregivers) and (5) ‘privately paid care network only’ (no other caregivers).

Loneliness is measured with the 11-item loneliness scale (De Jong Gierveld, 1998) and recoded on a scale from ‘0′ to ‘11′ with higher score meaning less lonely.

Mastery is measured by a five-item version of the Pearlin Mastery Scale (Pearlin & Schooler, 1978), including questions like ‘I have little control over things that happen to me’ and ‘I often feel helpless dealing with the problems of life’. Response categories range from 1 = strongly disagree to 5 = strongly agree. A higher score indicates a greater sense of mastery (range 5-25).

Care sufficiency: Participants were asked whether overall the amount of care received was sufficient, with (0) indicating ‘not sufficient’, which was given when respondents rated their care either insufficient or ‘in between sufficient and insufficient’ and (1) as sufficient.

Physical functioning is measured with six questions about activities of daily living based on Katz et al. (1963). Sum scores range between 6 and 30. A higher score means a better physical functioning; Chronic diseases as the sum of seven major chronic diseases (range 0-7): lung disease, heart disease, disease of the arteries, diabetes, cardiovascular accidents, rheumatic disease, and cancer; Cognitive functioning is measured using the Mini-Mental State Examination scale (MMSE; Folstein et al., 1975). The scale scores range from 0 to 30, with higher scores indicating better cognitive functioning; Year of observation (range 0—30 years) is created using the exact date of the interview; Age in years; Gender (1 = female; 0 = male); Level of education varies from “not completed” to “university” education and was converted to years of education (range 5-18 years).

Analysis

We described the data with the aggregated average of all variables from the ten observation moments for the different network types and total sample (Table 1). For this descriptive table we tested differences between the different network types using ANOVA and post hoc tests. Because of the longitudinal structure of the data (respondents at level 1 nested in time at level 2) we used hybrid multi-level models to disentangle the between-subjects and within subject parts of the longitudinal relationship (Twisk, 2019). Between- and within-person effects were estimated separately. The between-person part of the association compared depressive symptoms between the different respondents and was based on the individual mean values of the time varying independent variables. The between-subject part is the relationship between the mean value of the particular independent variable for each respondent and the repeatedly measured outcome variable (Twisk, 2019; $Y_{\mathit{\text{it}}}={\beta }_0+{\Sigma }_{j=1}^J{\beta }_{\mathit{\text{Bj}}}(X_{\mathrm{\text{ijt}}}-{\overline{X}}_{\mathit{\text{ij}}})+{ϵ}_{\mathit{\text{it}}}).$ The within person part of the association provided insight into the changes in depressive symptoms when care network types changed for one person (e.g. when an individual changed from one network type to another) and was based on the differences between the observations and the individual mean value. The within subject part of the relationship is often called the fixed effect $(Y_{\mathit{\text{it}}}={\beta }_0+{\Sigma }_{j\text{=1}}^J{\beta }_{\mathit{\text{wj}}}\left(X_{\mathit{\text{ijt}}}-{\overline{X}}_{\mathit{\text{ij}}}\right)\hspace{0.17em}+\hspace{0.17em}{\mathit{ϵ}}_{\mathit{\text{it}}}$). Random intercepts were included in all models to adjust for the correlated observations within the respondents. All variables were entered as time varying besides gender and level of education. The xtreg (be/fe) procedure in STATA V.17 was used for the analyses. The number of within person changes in network type can be found in the supplementary material, Table S2.

Table 1. Description of study variables, the aggregated average from the ten observations between 1992 and 2022, for the different network types and the total sample (N = 18,434 observations from 4,837 respondents).

Care network type	No care	Partner	Informal	Formal	Private	Total	Post-hoc tests
Average ten observations	59%	10%	6.3%	11%	13%	100%
Number of respondents	1067.20	216.30	118.20	199.90	241.80	1843.40
Depressive symptoms (0-30)	6.65	8.19	9.80	12.56	8.18	7.76	^a^,b,c,d,e,f,g,h,i
Less loneliness (0-11)	9.28	9.47	8.81	7.62	8.91	9.05	^b^,c,d,e,f,h,i,j
Mastery (5-25)	18.17	17.27	16.61	15.73	17.76	17.70	^a^,b,c,e,f,g,h,i
Sufficient care	95%	89%	86%	73%	89%	91%	^a^,b,c,d,f,g,i
Physical functioning (6-30)	28.68	25.89	24.83	21.93	27.41	27.37	^a^,b,c,f,g,h,i,j
Cognitive functioning (0-30)	27.87	27.51	26.78	26.38	28.05	27.63	^b^,c,e,f,h,i
Number of chronic diseases	1.07	1.40	1.62	1.94	1.27	1.25	^b^,c,f,h,i
Age	69.13	69.89	77.32	80.08	73.53	71.45	^b^,c,d,e,f,g,h,i,j
Female	49%	57%	65%	65%	61%	53%	^b^,c
Education (years)	10.21	9.76	8.72	8.83	11.92	10.19	^b^,c,d,h,i,j
Partner in household	73%	100%	29%	32%	54%	66%	^a^,b,c,d,e,f,,h,i,j
Total observations
Percentage of respondents	58%	12%	6.4%	11%	13%	100%
Personal care	–	3%	0%	1%	0%
Household care	–	91%	96%	77%	99%
Personal and household care	–	6%	3%	22%	0%

Post-hoc tests p < 0.05.

^a No care versus partner.

^b No care versus informal.

^c No care versus formal.

^d No care versus private.

^e Partner care versus informal care.

^f Partner care versus formal care.

^g Informal versus formal care.

^h Informal versus private.

ⁱ Formal versus private care.

^j Partner care versus private care.

Note: For more detail per measurement moment, see supplementary material, Table SI.

First, we added the dummies for network types to the model, to test the bivariate association, followed by inclusion of control variables: year, physical functioning, cognitive functioning, number of chronic diseases, age, gender and educational level (Model 1 and 2, Table 2). Thereafter, we added loneliness, mastery and care sufficiency one by one in Models 3, 4 and 5 (Table 2) to see the contribution of explanation of each individual variable. The differences between the coefficients of care network types in Model 2 and the Models 3, 4 and 5, were compared and percentages of change were calculated, to see the extent that these variables influence the associations between the different network types on depressive symptoms. Model 6 is the full model with all variables included. For all analysis the formal care network type was used as the reference category, as analysis showed that receipt of formal care is most distinct from the other types.

Table 2. Mixed-model analysis on the role of loneliness (Model 3), mastery (Model 4) and sufficient care (Model 5) in the associations of the different network types on the number of depressive symptoms (N = 18,434 observations from 4,837 respondents). Between (BP) and within person (WP) effects.

	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6
	BP	BP	BP	BP	BP	BP
No care^a	−6.53***	−2.71***	−1.78***	−2.16***	−1.48***	−1.14**
	(0.35)	(0.39)	(0.34)	(0.34)	(0.39)	(0.32)
Partner care network^a	−5.77***	−3.24***	−1.15**	−3.17***	−2.28***	−1.40***
	(0.45)	(0.45)	(0.40)	(0.38)	(0.44)	(0.36)
Informal care network^a	−3.27***	−2.12**	−0.79	−2.25***	−1.09*	−0.90*
	(0.53)	(0.50)	(0.44)	(0.43)	(0.49)	(0.4)
Privately paid care network^a	−4.92***	−1.68**	−0.92*	−1.24**	−0.76	−0.45
	(0.45)	(0.45)	(0.4)	(0.39)	(0.45)	(0.36)
Physical functioning (6-30)		−0.34***	−0.27***	−0.23***	−0.28***	−0.18***
		(0.02)	(0.02)	(0.02)	(0.02)	(0.02)
Cognitive functioning (0-30)		−0.28***	−0.10**	−0.16***	−0.27***	−0.09*
		(0.04)	(0.02)	(0.03)	(0.04)	(0.03)
Number of chronic diseases		1.08***	0.92***	0.72***	0.95***	0.64***
		(0.09)	(0.08)	(0.08)	(0.09)	(0.08)
Age		−0.10***	−0.11***	−0.12***	−0.09***	−0.11***
		(0.01)	(0.01)	(0.01)	(0.01)	(0.01)
Female		1.63***	1.69***	1.22***	1.51***	1.32***
		(0.18)	(0.15)	(0.15)	(0.17)	(0.14)
Education (years)		−0.03	−0.03	0.03	−0.04	0.01
		(0.03)	(0.02)	(0.02)	(0.03)	(0.02)
Year		−0.02	0.00	0.01	−0.03*	0.02
		(0.01)	(0.01)	(0.01)	(0.01)	(0.01)
Less loneliness (0-11)			−1.29***			−0.84***
			(0.03)			(0.03)
Mastery (5-25)				−1.11***		−0.83***
				(0.03)		(0.03)
Sufficient care (yes)					−6.36***	−2.80***
					(0.40)	(0.33)
Constant	13.36***	33.05***	38.15***	46.51***	35.34***	47.47***
R-squared	0.07	0.20	0.39	0.42	0.24	0.51

Table

	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6
	WP	WP	WP	WP	WP	WP
No care^a	−3.34***	−1.45***	−1.05***	−1.19***	−1.36***	−0.82***
	(0.18)	(0.19)	(0.19)	(0.19)	(0.19)	(0.18)
Partner care network^a	−3.19***	−1.16***	−0.68**	−0.89***	−1.10***	−0.47*
	(0.21)	(0.23)	(0.22)	(0.22)	(0.23)	(0.22)
Informal care network^a	−1.65***	−0.50*	−0.30	−0.39	−0.49*	−0.22
	(0.25)	(0.25)	(0.24)	(0.24)	(0.25)	(0.23)
Privately paid care network^a	−2.04***	−0.55*	−0.30	−0.39	−0.50*	−0.17
	(0.22)	(0.23)	(0.22)	(0.22)	(0.22)	(0.21)
Physical functioning (6-30)		−0.30***	−0.27***	−0.25***	−0.28***	−0.22***
		(0.02)	(0.02)	(0.02)	(0.02)	(0.02)
Cognitive functioning (0-30)		−0.13***	−0.11***	−0.10***	−0.13**	−0.09***
		(0.03)	(0.02)	(0.02)	(0.03)	(0.02)
Number of chronic diseases		0.24***	0.27***	0.22***	0.22**	0.24***
		(0.07)	(0.07)	(0.07)	(0.07)	(0.06)
Age		7.18	4.89	7.45	7.01	5.32
		(4.70)	(4.55)	(4.56)	(4.69)	(4.43)
Year		−7.14	−4.87	−7.43	−6.98	−5.33
		(4.70)	(4.55)	(4.56)	(4.68)	(4.43)
Less loneliness (0-11)			−0.79***			−0.68***
			(0.03)			(0.03)
Mastery (5-25)				−0.53***		−0.46***
				(0.02)		(0.02)
Sufficient care (yes)					−1.60***	−1.12***
					(0.16)	(0.15)
Constant	10.52***	−391.07	−254.25	−398.85	−380.62	−272.69
R-squared	0.03	0.07	0.13	0.13	0.08	0.17

^a Formal care network is reference.

Standard errors in parentheses.

*** p < 0.001, **p < 0.01, *p < 0.05.

Additionally, we tested the relationship between the network types and loneliness, mastery and care sufficiency (supplementary material, Table S3) and between loneliness, mastery and care sufficiency and the number of depressive symptoms (supplementary material, Table S4) separately. Finally, there are more factors influencing the care network type used and depressive symptoms that could have such a high impact that additional stratified analysis are necessary. First, having a partner is a precondition of care receipt from a partner care network and decreases the likelihood of other forms of care (Jutkowitz et al., 2022), and having a partner decreases depressive symptoms (Braithwaite et al., 2010). However, not all older adults with a partner receive care from that partner, e.g. the partner him-/or herself may have a health impairment (Swinkels et al., 2022), what in turn also will influence depressive symptoms. Second, the kind and amount of tasks the care network type provides, personal or household tasks, can influence who gives the care and also the depressive symptoms of the care recipient. Therefore sensitivity analyses are added to check the importance of partner status and type of care: We applied stratified analysis for a sample with older adults with a partner and without a partner, and for a sub sample with only household care use (supplementary material, Table S5).

Results

Over all ten observations, on average 41% of the respondents received some kind of care: 10% from a partner care network, 6% from an informal care network, 11% from a formal care network and 13% from a privately paid care network (Table 1; for all observations separately see Table SI). The depressive symptoms were highest among those with care receipt from a formal care network, followed by an informal, a partner and a privately paid care network respectively and lowest among those without care use. Those with care receipt from a partner care network were the least lonely, followed by no care, privately paid care and informal care use respectively, those with formal care use were most lonely. Mastery was highest for those with no care and a privately paid care network type, followed by those receiving care from a partner, informal and formal care network respectively. Compared to older adults that receive care from a formal care network, all other care users received more often sufficient care, and had better physical and cognitive functioning.

Care network types and number of depressive symptoms

Results showed that compared to formal care use all other groups report lower depressive symptoms (Table 2, Model 1BP). After adding the control variables, this difference became smaller, but still existed: no care and care from a partner care network showed the lowest level of depressive symptoms compared to the formal care network type and the informal care network type and privately paid care network type lie more in between (Table 2, Model 2BP). The within person results (Table 2, Model 1 and Model 2WP) showed the same differences.

Loneliness

If loneliness is added to Model 2 (Table 2, Model 3BP), the coefficients of alle network types decreased. This decrease was the highest among those who received care from a partner (65%) and an informal care network (63%), followed by a privately paid (45%) care network and no care users (34%). The within person results (Table 2, Model 2 and Model 3WP) also showed a decrease in effect size of care network type on depressive symptoms when loneliness was added. The R-squared from the between model increased from 0.20 to 0.39 by adding loneliness and for the within Models from 0.07 to 0.13. Care network types differed in loneliness as expected and being less lonely was negatively related to depressive symptoms (Tables S3 and S4). These results showed that differences in care recipients’ depressive symptoms of care use between different care networks can in part be explained by the level of loneliness of the care recipient (supporting hypothesis 1). However our expectation that this would also count for privately paid care is not supported.

Mastery

If mastery was added to the model 2 (Table 2, Model 4BP), the coefficients of those with no care use and care receipt from a partner or privately paid care network decreased a little compared to formal care. The association of care receipt from an informal care network increased which indicates a suppression effect. For the within effects (Table 2, Model 2 and 4WP) all coefficients decreased lightly compared to formal care use if mastery is added to the model. The R-squared from the between model increased from 0.20 to 0.42 by adding mastery, and for the within models from 0.07 to 0.13. Additionally, no differences between the different care network types were found concerning mastery, but having a higher mastery was strongly and negatively related to depressive symptoms (Tables S3 and S4). These results showed that differences in care recipients depressive symptoms of care use between different care networks cannot be explained by different levels of mastery (not supporting hypothesis 2).

Care sufficiency

If the sufficiency of care was added to the model 2 (Table 2, Model 5), the coefficients of all network types decreased compared to formal care. This decrease was the highest for care receipt from a privately paid (55%), an informal (49%) care network, and no care use (45%) and the lowest for those that received care from a partner care network (30%). The within person effects showed a small decrease in coefficients if sufficiency of care is added. The R-squared from the model increased from 0.20 to 0.24 by adding care sufficiency and for the within models from 0.07 to 0.08. Care network types differed in care sufficiency as expected and receiving sufficient care was negatively related to depressive symptoms (Tables S3 and S4). These results showed that differences in CR’s depressive symptoms of care use between different care networks can partly be explained by care sufficiency (supporting hypothesis 2).

Full Model

Results (Table 2; Model 6) showed that older adults who received care from a privately paid care network did not significantly differ from those in a formal care network concerning depressive symptoms, and older adults with partner care or no care use showed the lowest depressive symptoms, informal care use in between, controlled for all variables. The within results showed about the same results. The R-squared for this full hybrid, between model was 0.51 and for the within model 0.18.

Sensitivity analysis

The sensitivity analyses (Table S5), highlighted one meaningful difference: For the group without a partner those that use care from an informal care network show the least depressive symptoms compared to all other groups. The analyses with only household care use showed approximately the same results. Thus the found differences between the care network types cannot be attributed to the kind of care used.

Conclusion and discussion

This study assessed the explanation of the differences in number of depressive symptoms related to the composition of different care network types. Guided by the SDT we argued that differences between the level of wellbeing (measured by the absence of depressive symptoms) could be explained by differences in the extent that the care network fostered feelings of relatedness, autonomy and competence, indicated by loneliness, mastery and care sufficiency. The results revealed that those with a formal care network show the most depressive symptoms, and those with a partner care network, or if no partner is present an informal care network, have the least depressive symptoms. Adjusted for the control variables, in particular loneliness and care sufficiency, but not mastery, explained care network differences in depressive symptoms. Our results are consistent with former research that suggests that older adults who receive support from close ties may be better able to cope with stress and with functional disability than those who do not receive this support (Umberson et al., 2010). Also care sufficiency is an important factor in understanding the relation between care use and wellbeing.

Our study corroborates several empirical studies that successfully applied SDT in the context of health and wellbeing within the general population (Gillison et al., 2019) and recently for healthy aging of older adults (Ng & Ho, 2020) and care use among older adults (Broese van Groenou, 2020; Djundeva et al., 2015). We add to these studies several important insights. First, the use of a priori care network types is more informative than using the traditional distinction of informal, formal and mixed care. We based our typology on the assumption that many older adults receive care from a diverse set of caregivers, and that composition matters. The findings corroborate that the mere presence of specific caregiver types is more important for wellbeing than others, which is clear from the differences in depressive symptoms between the partner and the formal care network type. In our typology, in particular the formal care network type included also other types of helpers. Given its mix with other types of caregivers, in particular informal caregivers, this may have lowered the impact on depressive symptoms. If this is true the formal care users are even worse off than our results indicated.

Second, the concept of relatedness appeared a crucial ingredient for wellbeing that may be hampered in the context of receiving care. Our findings imply that formal care cannot compensate for informal care (potential) in terms of social connectedness or loneliness. Studies suggesting that formal care is preferred due to the lower dependency on informal care (Wylie et al., 2024) did not take into account that formal care is used because informal care is lacking. Using loneliness as an indicator of the quality of the care network, shows that this is more likely to be the case. What is lacking still, however, is information on how related or connected the care recipient is to formal caregivers. A proper indicator of how well one bonds with informal, formal and privately paid caregivers would improve the testing of this SDT concept in the context of care.

Third, care sufficiency is an indicator of the functioning of the care network that largely distinguished between the formal and the other types of care networks, and proved important for the differences in wellbeing. This implies that formal caregivers are not able to provide care in the way that is desired or needed, whereas a spouse or informal or privately paid caregiver appears able to do. However, it is not clear how care sufficiency is associated with the capabilities/characteristics of the care recipient or of the care network, so more information is needed about how care sufficiency is reached as it is obviously an important contributor to wellbeing. Other indicators of the functioning of the care network (e.g. levels of disagreements on care provision) are needed to expand our insights on why care networks contribute to wellbeing.

Fourth, we need more indicators of perceived control in the care context instead of general indicators of perceived control in life (mastery in our case). Mastery proved to differentiate very little amongst the care network types, although it showed a strong association with depressive symptoms. Feeling autonomous and competent in the care context may be better indicated by directly measured indicators as feeling in control over care (Broese van Groenou, 2020), or feeling a burden to others (Lee et al., 2018). We feel that autonomy is still a very important trigger of wellbeing in the care context, but that our indicator may have missed some of it.

This study has several strengths, particularly the large sample and longitudinal design. Both, between and within effects, lead to nearly the same conclusions, which underlines the robustness of the model. However, some limitations should be acknowledged as well. First, it is not certain that other measurements of wellbeing would have given the same results (Padayachey et al., 2017). Whether wellbeing and the absence of depressive symptoms can be considered as two sides of the same coin is widely debated (Baselmans & Bartels, 2018). Depression is a specific dimension of wellbeing, and many studies show the positive correlation with care use (Garrido et al., 2009). Yet, the CES-D is a multidimensional scale of depressive symptoms and combines positive and negative affect with biological and cognitive symptoms. The low scores in our sample indicate that very few reached the threshold of 16, indicating clinical depression, which support the use of this scale as an indicator of psychological wellbeing. Second, the data were collected over 30 years of time, wherein changes in policies and norms regarding care use took place (Da Roit, 2012). Although we controlled for the year of data collection, it could be that the impact of, for example the use of formal care, has changed. Future research could make a comparison between different time periods in care network type used and associations with wellbeing in order to test the robustness of our explanatory model. It was too comprehensive to take these differences also into account for this study.

There are two important implications for policy. First, because care use from a partner (and if no partner is present the informal care network), is impactful for the wellbeing of the care recipient, this reinforces the importance of support for the informal caregivers, partners and children in order for them to continue providing care and avoiding overburden (Liang et al., 2022). Thus support to this caregiver helps to make this care more sustainable, but also may improve caregivers’ wellbeing as well as care recipient’s wellbeing (Ferraris et al., 2022). Second, attention is needed for the way we organize our formal care services as they are less able to provide care that is perceived as sufficient and to create strong bonds with the care recipient. Formal caregivers themselves perceive little autonomy in how they spend their time and work pressure is high as they are assigned to many home care recipients on one day. Joining forces with informal caregivers may be a way to relieve work pressure and to create time to bond with the care recipient. In the formal care network, there are very often also informal caregivers, and more information is needed on how they collaborate in caregiving and discuss issues of dependency, care preferences and feeling in control. This may allow care recipients to arrange the care they need themselves as well as to relieve the stress on the care supply site (Han et al., 2023).

To conclude, our contribution in this field is that wellbeing of the care recipient is dependent on the care network type and that SDT can be helpful to understand these differences. The study corroborates that a formal care network is the least beneficial and a partner care network the most beneficial for the care recipient. In addition to individual factors as health and age, this can be explained by the structure and functioning of the care network.

Authors contributions

J.C. Swinkels planned the study, performed all statistical analyses and wrote the paper. M. Broese van Groenou and J. Abbing helped to plan the study, supervised the data analysis and contributed to revising the paper.

Acknowledgements

LASA data collections have been approved by the Medical Ethical Committee of the VU University Medical center, number 2012.361.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

For the purpose of replication, data supporting the findings of this study are available upon reasoned request at https://lasa-vu.nl/en/request-data/.

Additional information

Funding

The Longitudinal Aging Study Amsterdam is supported by a grant from the Netherlands Ministry of Health, Welfare and Sport, Directorate of Long-Term Care.The data collection in 2012-2013 was financially supported by the Netherlands Organization for Scientific Research (NWO) in the framework of the project “New Cohorts of young old in the twenty first century” ‘(grant number 480-10-014)’. The study is supported by a grant from the Netherlands Organization for Scientific Research (NWO) for the project “FAMCARE: The changing role of families in care networks and wellbeing of older adults in the Netherlands: (grant number 406.21.SW.027).