Informal caregiving and caregiving-related characteristics

Respondents were defined as informal caregiver when they provided unpaid care in the 12 months preceding the second wave to a family member, partner or friend who needed care because of physical or mental problems, or ageing (n = 1,759). Follow-up questions measured to whom they gave care (first-degree family such as parents, siblings or children; other family such as in-laws or other; partner; friend), whether they lived with the care recipient, reasons for care (physical problems; mental problems; ageing; dementia), kind of care (domestic work; personal care or nursing; emotional support or supervision; practical care), time spent giving care (0–7 hours weekly; 8 or more hours weekly) and duration of care (less than 1 year; 1 year or longer).

Emotional disorders

DSM-IV diagnoses [35] were made using the Composite International Diagnostic Interview (CIDI) 3.0, a fully structured lay-administered diagnostic interview. This instrument was developed and adapted for use in the World Mental Health Survey Initiative [36] The CIDI 3.0 version used in NEMESIS-2 was an improvement of the Dutch one used in this initiative. The disorders considered in this paper include mood (major depression; dysthymia; bipolar disorder) and anxiety disorders (panic disorder; agoraphobia without panic disorder; social phobia; generalized anxiety disorder) in the past 12 months, assessed at the second wave. Clinical calibration studies in various countries [37] found that the CIDI 3.0 assesses mood and anxiety disorders with generally good validity in comparison to blinded clinical reappraisal interviews. For the present study, mood and anxiety disorder were combined into presence or absence of any emotional disorder.

Sociodemographics

Age, gender, educational level (primary, basic vocational or lower secondary education vs higher education), employment status, partner status and having children living at home (vs having children not living at home and having no children) were considered. Age, gender, and educational level were measured at T₀, other sociodemographics were measured at T₁.

Other characteristics

Social support from three resources (partner; family or friends; neighbors) was measured with two questions regarding the extent respondents could rely on the resource for help and could open up to them (four response categories ranging from “not at all” to “a lot”). The mean score on the two questions was used to indicate social support perceived from a particular resource, taking the respondents’ evaluation of it into account. The total social support measure (ranging from 1 to 4) was calculated as the mean score on the perceived social support from at least two resources, because not all respondents had a partner [38]. A dichotomous variable was constructed using a cut-off point of 2.5 (the middle of the scale), thereby separating the lowest twelve percent, a group with quite severe lack of social support, versus the rest. Sensitivity analyses with cut-off points of 2 (4%) and 3 (31%) resulted in similar findings.

A standard checklist assessed presence of 17 chronic physical disorders (asthma; chronic obstructive pulmonary disease; chronic bronchitis; emphysema; severe heart disease; heart attack; hypertension; stroke; stomach or intestinal ulcers; severe intestinal disorders; diabetes; thyroid disorder; chronic back pain; arthritis; migraine; impaired vision or hearing; and another chronic physical disorder), treated or monitored by a medical doctor in the prior 12-months [39].

Analyses

Analyses were performed using Stata version 12.1, [40] which enabled to control for the complex sampling and recruitment procedure of the study. The data were weighted to correct for differences in the response rates in several sociodemographic groups at both waves and differences in the probability of selection of respondents within households at baseline.

Using the complete sample (n = 5,303), characteristics of informal caregiving were examined using frequencies and logistic regression analyses adjusted for age and gender. The association between informal caregiving and presence of 12-month emotional disorder was examined with logistic regression analyses adjusted for all sociodemographics. Sensitivity analyses were performed with stricter definitions of informal care; i.e. eight or more hours weekly, longer than one year, or both.

Further analyses were conducted among the informal caregivers (n = 1,759). That is, potential risk indicators of emotional disorders among informal caregivers were first separately investigated with bivariate logistic regression analyses adjusted for age and gender, and then simultaneously with conventional backstepping procedures to obtain a parsimonious set of risk indicators. No collinearity of risk indicators was observed in the final model. Lastly, the Population Attributable Risk Proportion (PARP) and the Number Needed to Treat (NNT) of the final set of significant risk indicators were calculated in univariable analyses to select useful indicators for targeted prevention [24]. The PARP represents the proportion of emotional disorders in the population of informal caregivers that can be reduced when the adverse effect of the risk indicator is completely eliminated [41]. Although this assumption is not realistic as preventive programs are often unable to completely block the adverse effect of the risk indicator and some risk indicators are unmodifiable (i.e. age, gender), the PARP can be useful as it gives an indication of the highest achievable health gain. The PARP was examined using the punaf procedure of Stata [42]. The NNT indicates how many informal caregivers have to receive an intervention to avoid one case of emotional disorder, assuming that the adverse effect of the risk indicator is completely blocked by the intervention [24,43]. The NNT can best be interpreted as a minimum of the required effort, as interventions are rarely completely effective. This was calculated as the inverse of the risk difference between informal caregivers with and without the risk indicator, which was estimated with generalized linear models with a binomial distribution and an identity link function.

Characteristics of informal caregivers

The association between informal caregiving and having a 12-month emotional disorder

The prevalence of any emotional disorder was similar among informal caregivers and non-caregivers (7.5% vs 8.8%; p = 0.15). Logistic regression analysis adjusted for all sociodemographics (not in table) showed no association between informal caregiving and having an emotional disorder (OR = 0.86; 95% CI = 0.66-1.11). When informal caregiving was defined more strictly, as giving care eight or more hours weekly, longer than one year, or both, no association with an emotional disorder was observed (OR = 0.92; 95% CI = 0.64-1.31, OR = 1.22; 95% CI = 0.92-1.63 and OR = 1.17; 95% CI = 0.73-1.87, respectively).

Risk indicators for 12-month emotional disorder among informal caregivers

Results for the two types of regression models, i.e. bivariate analyses adjusted for age and gender and the final backstep model, are shown in Table 2. Notably, findings are largely similar for both types of models. Therefore, hereafter we will describe the results of the most parsimonious model, the final backstep model. Regarding sociodemographics, emotional disorder among informal caregivers was associated with female gender, being younger than 45, unemployment and living without a partner, but not with educational level and having children at home. Of the caregiving characteristics, giving care to a direct family member, partner or close friend, providing emotional support, and giving care for one year or longer were associated with an emotional disorder. Unexpectedly, providing care to more than one care recipient was associated with a lower chance of an emotional disorder. Living with the care recipient, reason for care, and giving eight or more hours of care weekly were not associated with an emotional disorder. Moreover, neither employment status nor having children at home strengthened the relationship between hours of care provided and having an emotional disorder, that is, no additive interaction-effects were observed (not in table) [44]. Finally, caregivers who perceived a lack of social support had higher chances of experiencing an emotional disorder. No association between chronic physical disorders and emotional disorder was observed.

In the final two columns of Table 2, the PARP and the NNT are displayed for the risk indicators in the final backstep model. Emotional support had the highest PARP (83.0%) and an acceptable NNT (14.4), but the high prevalence rate (89.5%) limits its use for targeted prevention. Another notable risk indicator is lack of social support, with a substantial PARP (23.7%), the lowest NNT (6.3) and a fairly low prevalence rate (11.2%). Finally, being unemployed and living without a partner had considerable PARPs (25.7% and 26.6%, respectively), relatively low NNTs (15.3 and 13.0) and manageable prevalence rates (29.5% and 25.9%).

Key findings

In line with previous cross-national research, [7] we observed that informal caregiving was quite common (31.3%) in the general population. Although no association between informal caregiving and emotional disorders was observed, risk indicators were identified. Specifically, informal caregivers with limited resources (lack of social support, unemployment, living without a partner) and those with a burdensome care situation (giving care to a close loved one or providing emotional support) were at higher risk of an emotional disorder.

Strengths and limitations

The use of a large population-based sample which included different types of informal caregivers, not limited to heavily burdened caregivers, as well as a reference group of non-caregivers, and the use of diagnoses measured with a valid instrument (CIDI 3.0), are important strengths of our study. To our knowledge, this is the first study to examine risk indicators for emotional disorders in a representative sample of informal caregivers and to identify indicators useful for prevention activities.

A few cautionary remarks should be made in relation to our findings. Although the NEMESIS-2 sample was representative of the Dutch population on most parameters, people with an insufficient mastery of the Dutch language were underrepresented. Moreover, although NEMESIS-2 had a broad age range, the maximum age of the sample at the second wave was 68, thereby excluding older caregivers. Hence, our findings are not generalizable to older age groups. Furthermore, non-response may have been higher among heavily burdened caregivers with too little time to participate. The extent to which this occurred is unclear. Another limitation is that only the second wave of NEMESIS-2 could be used and our findings were therefore based on cross-sectional data only. One should thus be careful in interpreting these findings as no causal relationships were examined and reverse-causality may also play a role. That is, an emotional disorder may contribute to presence of the risk indicators.

Findings

Informal caregiving was quite common in the Dutch general population (31.3%). Informal caregivers were more often female, 45 years or older, and had a higher educational level than non-caregivers. Caregivers and non-caregivers did not differ in employment status, partner status, having children at home, social support and presence of a chronic physical disorder.

Providing informal care was not significantly associated with having an emotional disorder and even a lower prevalence of emotional disorder was found among caregivers compared to non-caregivers. In this regard, our findings are at odds with findings from a similar population-based study performed in Canada in 1990, in which informal caregivers were found to have a slightly higher chance of mood and anxiety disorders than non-caregivers. Although that study used a similar age range and the same definition of informal care, [21] the percentage of informal caregivers in the population was much smaller (15%) than in our study (31%). As that study used the same diagnostic instrument to examine mental disorders (though an earlier version), the discrepancy is likely due to societal changes over time (giving informal care might be more common nowadays) or cross-national differences. The fact that we did not observe an association between informal caregiving and emotional disorder, even when stricter definitions of informal care were applied (eight or more hours weekly, longer than one year, or both), might be explained by a continuing selection process, such that those who provide informal care tend to be healthier than those who either do not start or quit caregiving. A similar phenomenon is seen in the workforce, with a lower prevalence of mental disorders among workers than those outside the workforce, [39] dubbed the ‘healthy worker effect’ [45]. Such a phenomenon thwarts examination of the relationship between informal caregiving and emotional disorder in cross-sectional research.

Despite major methodological differences, findings regarding risk indicators of emotional disorders among informal caregivers are largely in line with previous research primarily based on convenience samples [15,16]. Overall, emotional disorders among informal caregivers seem to be related to having limited resources: lack of social support was a strong indicator for emotional disorders among informal caregivers, [17,30-32] as well as unemployment and living alone [46]. Other significant sociodemographic correlates for an emotional disorder were being female and younger age, but not educational level [12,24,46]. Although most of these risk indicators also exist in the general population, [47] they help to identify vulnerable informal caregivers.

Among informal caregivers, duration of care was associated with an emotional disorder, but not time spent giving care. The latter finding was in contrast with previous research [14] and might be explained by positive aspects associated with caregiving, such as increased closeness with the care recipient and personal satisfaction [46,48,49] Such aspects can positively impact mental health, provided that caregiving activities are not too burdensome [46] Our findings could indicate that the positive effect of such rewards offset the burden of much time spent giving care, but does not overcome the burden of a long duration of care. Future research should examine this point. Although a meta-analysis observed a poorer outcome for predominantly spousal caregivers of dementia patients compared with other diseases, [8] we did not observe such a relationship. Although power limitations cannot be excluded, as only 12% of our sample provided care to a dementia patient, there is a more likely explanation: our sample was relatively young and the majority of caregivers of dementia patients provided care to a parent, rather than to a partner. As the relationship with the care recipient in combination with a specific illness may affect the outcome for caregivers, [25] sampling differences (predominantly spousal caregivers vs all types of caregivers) may explain the difference in findings.

Some other caregiving-related indicators are worth noting. Specifically, giving care to a close loved one, i.e. a first-degree relative, partner or close friend, put informal caregivers at higher risk for an emotional disorder [14]. This might be because in such cases the illness of the care recipient is also burdensome. A previous study observed that giving care to a parent was associated with poorer mental health, but giving care to an in-law was not [26]. It was suggested that this difference is due to type of care provided, with less emotional support and more practical care [26]. Similarly, we observed a higher risk of emotional disorder when the caregiver provided emotional support.

Our findings suggest that tailoring prevention measures to people with limited resources (lack of social support, unemployment, living without a partner) can result in health gain (PARP) with relatively low effort (prevalence and NNT). Low-level interventions, such as e-health interventions [50] at the first signs of distress, may suffice to prevent an escalation of problems. As informal caregivers may join the care recipient when visiting the general practitioner (GP) or consult the GP when first signs of distress appear, GPs could play a key role in identifying caregivers at risk of emotional disorders. They should be particularly attentive when caregivers have limited resources. Close and more intense care situations (giving care to a close loved one or giving emotional support) may also be worthy of their attention. However, as such care situations are common, a more general approach may be appropriate, for example providing extra coping tools on websites targeted to informal caregivers.