The items chosen for the ResQ-Care were identified by means of a literature review that focused on identifying (a) predictors of high stress and high caregiver burden and (b) predictors of resilience. Additionally, we searched for published scales on caregiver burden in the German language and considered constructs underlying high-loading items. A detailed description and justification for each item can be found in the manual in which each item is introduced individually (21).

In the context of health counseling, the ResQ-Care aims at identifying those caregivers at risk of developing stress-related health complaints due to greater experiences of stress and reduced capacities of resilience. Concerning the stress framework, the ResQ-Care model is based on the aforementioned homeostatic model. Concerning the resilience framework, we follow the biopsychosocial model on resilience in caregivers (13), with a particular emphasis on the social dimension of resilience.

Keeping in mind the equilibrium as represented visually by a scale, we aimed at developing a questionnaire that assesses stress and resilience factors in caregivers equally, and allows for them to be related to each other by means of a ratio.

We set out to explore the validity of the ResQ-Care by conducting a survey study in which we presented a total of 291 caregivers with the ResQ-Care in addition to items on resilience, depression, and stress, among others. We aim at performing exploratory factor analysis to examine underlying constructs of the ResQ-Care. Further, we set out to analyze convergent and discriminant validity by hypothesizing that the strain scales correlate positively with scores on depression and stress and negatively with resilience scores, whereas we hypothesize the opposite pattern concerning the resilience scales.

The sample consists of n = 291 informal caregivers (see Figure 1). Caregivers were predominantly female (84.5%, N = 246) and on average 56.60 ± 10.63 years of age. Concerning the relationship to the care recipient, 51.4% (N = 149) of the participants were part of child-parent dyads with the child taking care of their parent, 33.8% (N = 98) were spousal caregivers, and 9.7% (N = 28) were parents taking care of their children.

Concerning characteristics of the care recipient, over half of the care recipients were female (59.1%, N = 172) and the mean age was 72.15 ± 19.61 years. Their level of care was rated at 3.23 ± 1.19 (levels of care range from 1 to 5 in Germany). At the time of study participation, recipients had already been care-dependent for an average of 7.49 ± 6.75 years.

A full review of sociodemographic characteristics can be found in Table 1.

The original German version of the ResQ-Care was administered (21). In German the scale is named “Fragebogen zur Angehörigenbelastung und Resilienz” (FARBE). For the purpose of this publication, the scale was translated into English by a professional translator leading to the title Resilience and Strain Questionnaire in Caregivers (ResQ-Care).

The original German scale was administered in paper-and-pencil format or online in addition to the BRS, GDS-15, and PSS-4, as part of an anonymous survey conducted between October 2020 and January 2021. During this time period for data acquisition, in October 2020 nationwide social restriction rules to contain the COVID-19 pandemic were active in Germany followed by a nationwide lockdown from November 2020 onwards. The survey was embedded in a larger study on caregivers' experience during the COVID-19 pandemic in Germany.

Recruitment for the online version of the survey was performed by distributing the link on various websites, whereas the printed version of the survey, along with the prepaid envelopes, was sent to registered physicians, outpatient clinics, care support offices, and welfare organizations, among others.

The study was approved by the ethics committee of the State Chamber of Physicians of Rhineland-Palatinate (Landesärztekammer RLP, processing number 2020-15286) and was pre-registered at the German Clinical Trials Register (DRKS; registration number DRKS00024621).

The ResQ-Care comprises a total of 20 items covering four scales in addition to a sociodemographic basis scale consisting of four items (Table 2).

The sociodemographic basis scale is conceptualized as a screening scale that allows for the identification of caregivers who are at high risk of adverse health outcomes due to sociodemographic characteristics. These are based on a review by Adelman et al. (2), who identified, among other factors, that caregivers who were female, had low educational attainment (medium-track secondary school diploma or lower), spent many hours per week on caregiving tasks (>21 h/week), and cohabited with the care recipient were at greater risk of higher caregiver burden. The sum score of this scale ranges between 0 and 4 with higher figures indicating higher risk of caregiver burden.

The two resilience scales (“my strength-givers”) focus on the psychological dimension of resilience (“my inner attitude”, IA) and the social dimension of resilience (“my sources of energy”, SE). The items of the scale “my inner attitude” cover self-determination, self-efficacy, growth, and recovery from stress, among others. The items of the scale “my sources of energy” focus predominantly on social support and the scheduling of pleasant activities that exert an antidepressant effect.

The two strain scales (“my strength-sappers”) cover an interpersonal dimension (“difficulties in dealing with the person in need of care”, DIFF) and an intrapersonal dimension (“general burdens of my living situation”, GB) of caregiver burden. The scale “difficulties in dealing with the person in need of care” assesses primarily behavioral and psychological symptoms and impairments in activities of daily living concerning the care recipient. Additionally, dyadic aspects are covered by items on relationship quality. The scale “general burdens of my living situation” focuses on the caregiver him-/herself and comprises items on the caregiver's own health, role conflicts, and general stressors and burdens.

Each item is rated on a scale ranging from 0 (“no”), 1 (“rather no”), 2 (“rather yes”), to 3 (“yes”). The sum score for each scale is calculated and can vary between 0 and 15. All items are positively poled so that higher sum scores on the strain scales represent higher stress whereas higher sum scores on the resilience scales represent higher resilience.

The interpretation of the scale consists of two steps. Initially, the sociodemographic basis scale and the four scales are interpreted individually. For the sociodemographic basis scale, whenever a caregiver fulfills all four criteria, the cut-off values for the ensuing scales are lowered. For the two resilience scales, the cut-off is defined as atleast two items being answered with at least “rather no” or “no.” Concerning the two strain scales, the cut-off is defined as no less than two items being answered with at least “rather yes” or “yes,” respectively. Whenever the cut-off is reached, recommendations on counseling themes for the caregiver are provided (e.g., self-care, social support, psychoeducation, self-help groups). For all caregivers affirming all risk factors on the sociodemographic basis scale (sum score = 4), it is recommended to interpret individually all resilience and strain scales independently of whether the cut-off has been reached or not.

Second, the ratio between stress and resilience factors is interpreted. An automated tool is available to visualize the stress:resilience ratio. Based on this ratio, four types of caregivers with differing needs for support can emerge. However, please note that the interpretation of the stress:resilience ratio is so far based on a visual screening as the validation of the cut-offs is pending. In a next step, this ratio will be evaluated to account for the fact that when relating these values to each other more information is provided in comparison to interpretation of absolute values on strain and resilience scales separately.

The German version of the ResQ-Care scale (FARBE), a detailed user manual (21), and an automated interpretation tool are available online (https://www.zqp.de/frageboegen-farbe/) and can be used free of charge.

To assess trait resilience, we used the German version of the Brief Resilience Scale (22). The BRS was created to assess resilience as the ability to bounce back or recover from stress. It includes six items, half of which are positively worded and the other half reverse-coded. Items are rated on a 5-point Likert scale (11), with higher mean scores representing higher resilience. The reliability and validity of the German adaptation of the BRS has been demonstrated by Chmitorz et al. (22). In our sample, Cronbach's Alpha for the BRS was high with α = 0.840. Additionally, we calculated McDonald's Omega, which led to a similar result of ω = 0.841.

The PSS measures “the degree to which situations in one's life are appraised as stressful” (23). It is a widely used self-report instrument for depicting how participants experience situations as uncontrollable and overloading. The original 10-item scale has proven to be a reliable, valid, and economical instrument for assessing perceived stress (24), with acceptable psychometric qualities for the four-item short version. The PSS-4 consists of four items that are rated on a 5-point Likert scale ranging from 0 = “never” to 4 = “very often”. Higher sum scores represent higher perceived levels of stress. Cronbach's Alpha for PSS-4 measures in our sample was acceptable with α = 0.793, McDonald's Omega was ω = 0.801.

In the present study, we used the Geriatric Depression Scale in its short version (25). All 15 items have a simple dichotomous answer format (“yes” or “no”). Higher sum scores reflect higher levels of depression, with a cut-off of sum score ≥5 being representative of clinically relevant depressive symptoms. Internal consistency as measured by Cronbach's α was high with α = 0.851 and McDonald's Omega was similarly high with = 0.859 in our sample.

Data analysis was performed using JASP (Version 0.15). Missing values were dealt with by pairwise exclusion. To evaluate the structure of the ResQ-Care questionnaire with its four theoretically assumed scales, we performed exploratory factor analysis using maximum likelihood estimation. The number of factors retained was analyzed by parallel analysis. Following the recommendation by Costello and Osborne (26), we used oblique rotation (oblimin) in order to account for the fact that factors correlate with each other. Convergent and discriminant validity were calculated by means of regression analyses, correlating the four ResQ-Care subscales with GDS-15, PSS-4, and BRS scores. For reliability analysis, we report Cronbach's α and MacDonald's ω for all scales. Mean and standard deviation are presented where appropriate. P ≤ 0.05 were considered as significant.

Results of both the Bartlett test (χ² = 2200.565, df = 190, p = <0.001) and the Kaiser-Meyer-Olkin measure of sampling adequacy (KMO = 0.84) verified the adequacy of the analysis. Based on exploratory factor analysis with oblimin rotation, a four-factor solution was proposed explaining 43.3% of variance. This solution supports the designated four ResQ-Care scales for the most part (see Table 3), with few cross-loadings. Factor 1 corresponds to “general burdens of my living situation” (GB) and explains 16.7% of the variance. Factor 2 corresponds to “difficulties in dealing with the person in need of care” (DIFF) and explains 11.7% of the variance. Factor 3 represents “my sources of energy” (SE), making up 9.1% of the total variance in the data. The items in “inner attitude” (IA) load highest on Factor 4, which in turn explains 5.8% of the variance. The ResQ-Care scales were constructed to measure and relate factors of strains (GB and DIFF) and resilience (IA and SE). The explained variance amounts to 28.4% for the strain scales and 14.9% for the resilience scales. Cronbach's α as an indicator of reliability was acceptable for GB, DIFF and SE scales (>0.70) but questionable for the IA scale (=0.67). McDonald's ω was ≥0.68 for all four scales. Reliability analyses of the original allocation of items (5 items for each scale) let to better indicators of reliability concerning the two resilience scales, whereas the two strain scales presented higher reliability resulting from EFA. Only four of the 20 included items do not load highest on their originally designated scales. The item “The person in need of care is physically restricted and needs support to carry out activities of daily living (e.g., getting dressed, washing, movement, eating), which is difficult for me to provide” (DIFF) was assigned to Factor 1 (λ = 0.21), while its loading on Factor 2 (corresponding to the designated scale) is λ = 0.18. Another item from the same DIFF scale (“I can't leave the person in need of care alone for an hour”) loads highest on Factor 1 (λ = 0.34) instead of Factor 2. The item “Despite the increased demands, I manage to pursue my own interests (such as hobbies, sport)” (SE) clearly also loads highest in Factor 1 (λ = −0.53) instead of Factor 3. The fourth item now allocated to Factor 1 (λ = −0.47) instead of its designated scale is “I recover quickly from stress” from the IA scale. The item “I feel joy in my everyday life, e.g., when I take part in pleasant activities” (SE) posed most difficulty in clearly assigning it to one factor, as it loaded similarly high on Factors 1 (λ = −0.23) and 4 (λ = 0.23), additionally to its designated Factor 3. All of these items concerned showed a plausible loading pattern in accordance to the construction of the ResQ-Care, meaning that items originally allocated to the resilience scales showed positive loadings with resilience scales and negative ones with strain scales, and vice versa.

Correlations among ResQ-Care subscales and the BRS, PSS-4, and GDS-15 are presented in Table 4. In the analysis presented we used pairwise deletion; the same pattern of results emerged when using listwise deletion instead. Regarding the BRS and PSS-4, the results confirm the convergent and discriminant validity: The BRS as a measure of resilience correlated positively with the resilience scales and negatively with the strain scales. PSS-4 sum scores correlated negatively with the resilience scales and positively with the strain scales. The GDS-15 sum score, however, was positively correlated with the resilience scales and negatively correlated with the strain scales. Overall, the BRS and PSS-4 were negatively correlated (r = −0.521, p < 0.001) whereas the GDS-15 and BRS were positively correlated (r = 0.510, p < 0.001). The PSS and GDS-15 were negatively correlated (r = −0.690, p < 0.001).

Exploratory factor analysis confirmed the structure of the ResQ-Care scale with its four factors. Indicators of reliability were acceptable. Convergent and discriminant validity were confirmed concerning correlations with well-validated scales on resilience and perceived stress. Validity concerning depressive symptoms as measured by the GDS-15 was counter-intuitive and may be explained by the special lockdown circumstances under which data acquisition took place.

Overall, the structure of the ResQ-Care, with its four assumed scales, was confirmed. Nevertheless, a limited number of items showed cross-loadings and challenge our allocation of items to the respective scales. We designed the ResQ-Care scale to cover five items per scale. This allocation of items has practical reasons as the visualization of the results leads to a scale representing the weight of resilience factors relative to the weight of stress factors. With each scale having the same weight in this visualization, the interpretation of the stress:resilience ratio has high face validity and is intuitive for the user. Especially in view of the fact that the application of the ResQ-Care scale is predestined for professional counseling contexts, it has to be easy and intuitive to use and interpret. Given that the cross-loadings always confirmed the pattern of allocation—items covering resilience scoring high on resilience and low on burden, and vice versa—for practical reasons, we argue that the allocation of items to each scale should be maintained.

Concerning convergent and discriminant validity with the BRS and PSS-4, the validity of the ResQ-Care scale was confirmed. The correlation pattern shows the expected loadings and in particular it demonstrated that the ResQ-Care scales explain additional variance in stress and resilience that are not captured by the BRS and PSS-4. As both the BSS and PSS-4 were designed for very broad populations, it is important to keep in mind that burden and stress factors in informal caregivers share specific components that are not captured by the very general scales. In line with this, Zhou et al. (27) also advocate for a specific definition of resilience in the context of caregiving.

The results of the GDS-15 regression analysis, however, appear counterintuitive at first glance (e.g., the higher the depression score, the higher the resilience and the lower the stress) but may well be explained by the special circumstances under which our study was conducted: Various items of the GDS-15 might describe normal and judicious reactions to a pandemic instead of depressive symptoms in elderly people. For instance, it includes several behavioral items such as “have you given up many activities and interests?” The present survey specifically asked for an evaluation that “corresponded to how you felt during the lockdown.” This framing may have resulted in the questions surveying impairment due to the lockdown rather than actual depressive symptoms. In our sample, for example, the GDS-15 correlated with the subscale “my inner attitude' and the BRS score among others. This correlation is primarily attributable to the fact that none of the participants with a high GDS score considered themselves to be not resilient. Since this distribution pattern seems counterintuitive, it could indicate a selection bias: Individuals who felt very burdened by the COVID-19 lockdown and who were also not very resilient may have been less likely to participate in the study. A high level of stress with a strong “inner attitude” could indeed indicate resilience in this respect: Initial COVID-19 studies indicate that a positive appraisal style in particular might constitute a resilience factor during the pandemic (28). To enable a clear assessment of validity with regard to depressiveness, it would be important to re-administer the questionnaires outside of COVID-19 lockdown episodes.

The finding that the strain scales explained more variance than the resilience scales is highly important, but also unsurprising given that well-validated predictors of caregiver burden cover precisely those factors with a strong negative impact [e.g., cognitive impairments of care recipient, behavioral symptoms, depressive symptoms of caregiver (2, 29–32)]. Furthermore, research has only just begun to examine the role of resilience and other protective factors in understanding caregiver burden. Thus, it will be necessary to further examine which factors best buffer the negative consequences of well-known factors contributing to caregiver burden. A further explanation for this pattern of findings may lie in the well-established negativity bias (33), which refers to people's tendency to better remember “negative” events and also to weigh them more heavily. In retrospective surveys, this might lead to the stressful factors playing a greater role for respondents than the protective factors. Especially in the context of counseling where individuals are seeking help due to stress “the negative” might generally outweigh possible protective factors. At the same time, this negativity bias revers into a positivity bias in older age with older adults showing the capacity for emotion regulation reasons to shift their attention to positive stimuli and remember these better in comparison to younger adults (34). As in this study the major part of caregivers were children taking care of their parents (middle aged adults), it might be an intriguing future research question whether these caregivers differ in their stress:resilience ratio from spousal caregivers (most often older adults).

Although the sample size was adequate to perform EFA, it was not adequate to additionally perform a confirmatory factor analysis or analyses of measurement invariance. Therefore, the results presented here provide first evidence for psychometric good properties of the German version of the ResQ-Care while at the same time future studies are needed to corroborate these findings in larger samples. Additionally, the allocation of items has to be discussed critically as we decided for practical reasons to keep the original allocation although some items scored higher on other factors. However, for feasibility reasons we consider it very important that each scale consists of the same number of items, so that burden and resilience scales can directly be weighed against each other. Concerning the sample size, it has also to be critically discussed that our sample was predominantly female, thus limiting the generalizability to male caregivers. Nevertheless, as the majority of informal caregivers are female (35), the study sample does appear to be representative of the gender distribution of caregiver samples. Furthermore, when interpreting the results, it has to be kept in mind that the survey was conducted during the second wave of the COVID-19 pandemic in Germany and the corresponding lockdown policy. As the pandemic has been associated with increased burden for informal caregivers (36), the responses to each item might be biased by these particular circumstances in which caregivers found themselves. At the same time, this might also have led to a selection bias, with only caregivers who felt well enough to fill in the questionnaire choosing to take part in the survey. Moreover, we conducted the study in German and thus the psychometric properties relate to the original German version. Future studies are necessary to validate the ResQ-Care in different settings, more heterogeneous caregiver subpopulations and also in English language. Nevertheless, the results presented here are encouraging, as they provide confirming evidence for the structure and psychometric qualities of the German version of the scale.