Using the convenience sampling method, recruitment was carried out in multiple centers in China, including Shanghai, Anhui, Gansu, Sichuan, Chongqing, Guizhou, and other places. By contacting community workers, investigators visited surrounding communities from January 2021 to March 2022 to recruit disabled elderly individuals and their caregivers who met the inclusion criteria and who were willing to participate in the study.

The inclusion criteria for disabled elderly individuals were as follows: (1) age ≥ 60 years old and classified as mildly or more severely physically disabled by the Activities of Daily Living scale; (2) primarily received home care; (3) able to communicate verbally; and (4) understood the purpose of the research and participated in the research voluntarily. The exclusion criteria were as follows: (1) serious physical diseases or extreme weakness; and (2) a mental disorder or cognitive impairment.

The inclusion criteria for primary caregivers were as follows: (1) age ≥ 18 years old; (2) responsible for the primary care of elderly family members; if there are several caregivers, the person who cared for the elderly individual for the longest time was chosen; (3) cared for the elderly individual for at least 1 month; (4) clear awareness, normal communication, and understanding skills; and (5) understood the purpose of the research and volunteered to participate in the research. The exclusion criteria were as follows: (1) mental and cognitive impairments; and (2) formal caregivers who received financial compensation.

The sample size was determined using a subject-to-item ratio of 5–10:1 (Lu et al., 2015), based on the WFRQ-C of 26 items. A sample size of 260 to 520 was determined for exploratory factor analysis (EFA) and confirmatory factor analysis (CFA). Forty elderly individuals were selected and completed the questionnaire again after 2 weeks to evaluate the retest reliability. A total of 607 questionnaires were distributed, and 566 valid questionnaires were recovered (mainly distributed in five provinces across China: Shanghai 24.4%, Anhui 27.6%, Guizhou 25.4%, Chongqing 7.8%, Gansu 5.8%), and the recovery rate was 93.2%.

A questionnaire was used as the survey instrument for this study, including General Information Questionnaire for elderly individuals, Activity of Daily Living Scale (ADL) to selected participants, the Walsh Family Resilience Questionnaire Chinese Version tested among elderly individuals, the Family Care Capacity Scale for Elderly Patients used as a concurrent validation tool for the caregivers, and the Psychological Resilience Scale and the Social Support Assessment used as concurrent validation tools for both the elderly individuals and the caregivers.

The validation of this study started by conducting personal cognitive interviews with 20 community-dwelling elderly individuals based on the WFRQ-C. Then, formal investigation was performed. In the homes of the disabled elderly individuals or in the offices of community service center, the subjects (including disabled elderly individuals and primary caregivers) were asked to complete the WFRQ-C together and independently complete their own questionnaires. Disabled elderly individuals completed the CD-RISC-10 and SSRS-10, while caregivers completed the CD-RISC-10, FCCSE, and SSRS-10. Under normal circumstances, self-reports were used, but when the subjects could not complete the questionnaires due to educational level, visual impairment, disease or other reasons, the investigators read the questions one by one with a neutral and non-judgmental attitude, and the research subjects answered orally. All participants provided informed consent. The data were maintained securely, and only researchers had access to them. To compensate respondents for their time, a gift costing approximately 2 dollars was provided.

Statistical analysis was performed using SPSS 26.0 and AMOS 23.0 software. SPSS 26.0 software was used to randomly divide the total sample of disabled elderly individuals into two groups at a ratio of approximately 1:1 for analysis. For sample 1 (281), the item-total score Pearson correlation coefficient, item critical ratio value (CR value), EFA, Cronbach alpha coefficient, and Guttman split-half reliability coefficient analysis were used. For sample 2 (285), we used CFA. The total sample (566) was included to examine calibration-related validity. Test-retest reliability was calculated for sample 3 (40), which included participants who were retested after 2 weeks. A two-sided test level of a = 0.05 and P < 0.05 were considered statistically significant. The overall process can be seen in Figure 1.

This study was approved by the Institutional Review Board of Research Institutions (IRB). Participation was entirely voluntary, verbal and written consent were obtained prior to all interviews, and the informants’ anonymity and right to withdraw from the study prior to analysis were guaranteed. The interviews were conducted when the participants felt alert and ready. Contact information for psychological assistance was provided if needed. All participant information collected in the context of this study was anonymous and stored securely by the researchers.

The characteristics of disabled elderly individuals are shown in Table 1. The age of disabled elderly individuals ranged from 60 to 102 years old. Of the participants, 78.3% experienced mild dependence, 15.7% medium dependency, and 6.3% heavy dependency. There were no significant differences between the two randomly selected samples in terms of sociodemographic and family characteristics.

Cronbach’s α coefficient for the WFRQ-CE total scale was 0.93, and Cronbach’s α coefficients for the four factors (i.e., family belief, organization and problem solving, family communication, and utilization of external resources) was 0.87, 0.83, 0.89, and 0.65, respectively.

The retest reliability of the total scale was 0.96, and the retest reliability of the four factors (i.e., family belief, organization and problem solving, family communication, and utilization of external resources) was 0.95, 0.92, 0.92, and 0.95, respectively.

The split-half reliability of the WFRQ-CE total scale was 0.85, and the split-half reliability of the four factors (i.e., family belief, organization and problem solving, family communication, and utilization of external resources) was 0.81, 0.78, 0.79, and 0.68, respectively.

From a statistical point of view, the Pearson correlation coefficients of the item-total score of WFRQ-CE ranged from 0.50 to 0.80 except for item 8 (r = 0.183). When the high and low groupings were taken as 26% before and after, the CR values of the items reached a significant level (P < 0.05). If the standard deviation of item 8 > 1, the discrimination is acceptable. There was no ceiling effect for each item (the first and last options were selected >50%) (Jingnan et al., 2021). The factor analysis indicated that the factor loading for item 8 was 0.689 (>0.4), which indicated an acceptable level of representativeness (Hao and Jingwu, 2006). The 26 items on the WFRQ-CE are reserved; this indicates that all 26 items on the WFRQ-C are suitable for community-dwelling disabled elderly individuals in China.

Reliability, also known as accuracy, reflects the measurement results stability of a research tool (Hao and Jingwu, 2006). Cronbach’s α coefficients above 0.7 are generally considered acceptable, and the Cronbach’s α coefficients for subscales are preferably above 0.6. The stability of the scale is better when the split-half and retest confidences are above 0.7 (Hao and Jingwu, 2006; Minglong, 2010). The reliability coefficients of WFRQ-CE are all within the acceptable range, indicating that the WFRQ-CE exhibits good reliability.

Validity refers to the closeness of the measurement results of the research tool to the target content, reflecting the authenticity and accuracy of the research tool (Hao and Jingwu, 2006). Walsh pointed out that the nine core processes of the family resilience theoretical framework are recursive, and the key processes have synergistic effects, which can reasonably be clustered in different ways in different environments and populations (Walsh, 2016a). In this study, the EFA yielded four common factors with eigenvalues > 1, and the cumulative variance contribution rate was 56.94%. Compared with the general community residents in China (Wang and Lu, 2022), the item attribution is the same as that of the family belief and external support domains but the factor of “communication and resolution” in the community resident group. The results of the test for disabled elderly individuals divided this part into two factors, namely, “organization and problem solving” and “family communication” in this study. This is consistent with the model of coping with “disability” in families of the elderly in the qualitative study, where problem solving and communication processes are relatively independent (Abulaiti et al., 2022). Compared with the three domains and nine processes of the Walsh family resilience model developed by the original scale (Walsh, 2016a), the items included in F1 (family belief) in this study are all derived from its family belief system items. The items included in F2 (organization and problem solving) involve family organizational processes and communication and problem-solving processes, which mainly emphasize the core process of family flexible coping and problem solving. F₃ (Family Communication) contains items that are communication/problem solving processes and focuses on the expression of ideas and emotions between families. F₄ (utilization of external resources) contains items related to family belief systems and family organization processes that measure the support of religion, community, and friends. Each of the adaptive indicators in the adjusted CFA is within a reasonable range, indicating that the internal structure of the scale is relatively stable in the families of disabled elderly individuals. The factor structure and item levels of the WFRQ-CE were different from those of the WFRQ-IT (Rocchi et al., 2017), which was used in families with chronic childhood diseases. They mainly indicated that family belief and communication had interaction. This may be related to different sociocultural environments and groups. This is similar to the distinction between solving somatic problems and psychological communication for the elderly (Hsu, 2009). Family resilience may be understood in unique ways across families, but the common metric helps to understand variations across contexts.

The correlation coefficient between the tested scale and the calibration scale should be between 0.4 and 0.8 (Hao and Jingwu, 2006). Previous research has found that characteristics of individual resilience such as intrinsic control trajectories, emotional regulation, belief systems, and self-efficacy are protective factors that promote family resilience at the individual level (Meihua, 2013). Families have large social networks, and the more resources to which they have access, the stronger their resilience and the more easily they can adjust the family model and division of human resources flexibly to suit the situation, which allowed the family to adapt to adversity. The type of resources could be physical, financial, natural, traditional, social capital, and human resources (Yang et al., 2021). In this study, the CD-RISC-10, FCCSE, and SSRS-10 were used as the efficacy criteria. The correlations between these scales and the WFRQ-CE total scale ranged from 0.35 to 0.60, which indicates a moderately positive correlation between family resilience and personal resilience, caring ability, and social support in families of disabled elderly individuals. Moreover, family resilience includes other elements, which is the important theoretical basis for families’ ability to cope and adapt to adversity (Walsh, 2002). These findings also show that the WFRQ-CE has a suitable concurrent validity, which contains the primary source of adaptive capacity for the elderly and their caregivers in the face of disabling adversity.