Introduction

Chronic kidney disease (CKD) negatively impacts patients’ lives in various spheres (social, economic, psychological, work, and in the family nucleus) (^{Cukor, Cohen, Peterson, & Kimmel, 2007}). These psychopathologies are most common in patients with CKD in hemodialysis-based substitute treatment, and impact their morbi-mortality (^{Cukor, Coplan, Brown, Friedman et al., 2008}; ^{Cukor, Rosenthal, Jindal, Brown, & Kimmel, 2009}).

Resilience is defined as a person’s ability to develop healthily and successfully, despite being exposed to situations that jeopardize his/her integrity (^{Cukor et al., 2009}). When resilience is seen from this positive approach, it is possible to understand the close link between health psychosocial adaptation and successful recovery from adverse experiences (^{Cabanyes Truffino, 2010}), particularly illness, due to the presence of protective factors learned in childhood (^{Rodriguez-Piaggio, 2009}). Resilience makes it easier for patients to perceive a low degree of physical interference and improve their future health perspectives, since they remain more optimistic, stave off illness, and perceive more control over their behavior and emotional states, which in turn encourages their treatment adherence and improves their overall quality of life (^{Vinaccia & Quiceno, 2011}). Conversely, various sustained, unresolved psychosocial stressors in kidney patients may be associated with low resistance or allostatic burden, when applied to health and illness settings, particularly CKD (^{Cukor et al., 2007}). For example, it is known that the time of diagnosis of an illness is inversely related to resilience (^{Quiceno & Vinaccia, 2012}), and that depression and anxiety symptoms are associated with certain distorted thoughts in patients with CKD (^{Lerma, Salazar, Perez-Grovas, Bermudez et al., 2012}). It is also know that modifying these distorted thoughts makes it possible to reduce depression and anxiety symptoms and improve the quality of life of these patients (^{Lerma, Perez-Grovas, Bermudez, Peralta-Pedrero, Robles-García et al., 2017}). Multicultural studies of kidney patients point to the need to explore their resilient response from the context of their ethnic-family origin (^{White, Bichter, Koeckeritz, Lee, & Munch., 2002}). In transplant situations, resilience is associated with aspects of positive or negative emotional adjustment, depending on patients’ perception of the disease (guilt, fear, gratitude, physical limitation, independence) (^{Schipper, Abma, Koops, Bakker et al., 2014}), as well as social support, confidence, self-esteem, devaluation, and self-efficacy (^{Tong, Cheung, Nair, Kurella et al., 2014}).

Against this background, it is important to be able to reliably measure resilience in this population with CKD, since it has been evaluated using questionnaires in Spanish in other non-Mexican populations (^{Vinaccia et al., 2011}). There is also a questionnaire to evaluate resilience that has been validated in Mexican population as a whole (^{Palomar-Lever & Gómez-Valdéz, 2010}), but there is no information to date on its validity or reliability in the specific context of patients in chronic hemodialysis.

Nor have researchers evaluated whether resilience factors in kidney patients are associated with depression or anxiety symptoms, or their distorted thoughts. Having a valid, reliable questionnaire to assess resilience in this population would make it possible to identify the relationship between resilience and the development of depression and anxiety and key clinical repercussions (such as morbidity, hospitalizations, and premature mortality). Likewise, a reliable evaluation instrument would make it possible to test the effects of potential therapeutic interventions that include resilience as a modifying variable that prevents or reduces the depression and anxiety symptoms of kidney patients.

The objective was to evaluate the validity and reliability of a psychological instrument to measure the resilience of Mexican patients diagnosed with chronic kidney disease under treatment with hemodialysis, and to evaluate the correlation between resilience factors and depression and anxiety symptoms and distorted thoughts.

Method

Results

The study included 280 patients who had been in hemodialysis for some time (median [25th - 75th percentile]) = 33 (16 - 50) months. Patients were aged between 18 and 85, 40% of whom had diabetes mellitus. The main occupations were: retired or unemployed (33%), housewife (28%), or employee (17%). Educational attainment was as follows: secondary (66%), basic (21%), and university (11%) education. The respondents’ marital status was as follows: 39% lived alone (single, widowed, divorced), while 61% are partnered (married or living with their partners).

Confirmatory analysis

The fit of the five-factor model obtained through the exploratory analysis was evaluated through a confirmatory factor analysis (CFA) using the maximum likelihood method, with the following steps:

1. Identification of the model: For the model analyzed, there were 46 endogenous variables (41 involving measures and five involving structures) out of a total of 93 variables. There were 52 non-observed variables (total parameters to be estimated by the model), 46 of which were exogenous (41 measurement errors, five residual errors), out of a total of non-redundant elements of the matrix equal to 1081. Since the parameters to be estimated are clearly lower than the total number of non-redundant elements in the matrix, we can say that the model is over-identified and can be identified. Since there are no correlations between the residuals, it is a recursive model.

2. Specification of the model: Based on the exploratory analysis of reliability and the factor analysis previously presented, a model for Figure 1 was designed whereby the circumferences or ovals represent latent variables and the rectangles represent the observable variables. Measurement and residual errors are shown as smaller ovals.

3. Estimate of parameters: The AMOS program was asked to apply the maximum likelihood method, with standardized estimators, R2 estimator (multiple squared correlations), covariances of the estimators, indices to be modified and critical proportions for the differences.

4. For the evaluation of fit (the sample exceeds n = 200), as a first step, all the results were examined to ensure that the estimated coefficients did not exceed the acceptable limits (that there were no negative or insignificant error variances, standardized coefficients over 1, or excessively high standard errors related to an estimated coefficient). No collinearity was observed in the variables measured, since the correlations were less than .3 (^{Kline, 2005}) or extreme or univariate or multivariate scores (^{George & Mailery, 2010}). Moreover, excellent asymmetry was observed in all the variables (values not over ± 1.00), (± 10 points) except in the Family Support and Social Support variables, yet still within adequate parameters (^{Tabachnick & Fidell, 2001}).

5. Table 3 shows that as regards the absolute measures of overall adjustment, the chi square value was 1755.32 (p = ≤ .001, 766 gl), while the chi square ratio and degrees of freedom (CMIN/DF = 2.29) is excellent, indicating that mistakes in the variances and covariances of the model are null and adjusted to samples of over 200 data as recommended by the experts (^{Byrne, 2010}).

Figure 1. Confirmatory model of five resilience factors in Mexican kidney patients (n = 280). Chi square = 1755.32, 766 gl, p ≤ .001; CMIN/gl = 2.292; RMR = .031; GFI = .769; CFI = .881; RMSEA = .068 (.064 - .072). 

Table 3.  Goodness-of-fit indices of the confirmatory model resulting from 5 factors in kidney patients (n = 280)  

Statistics	Desirable criterion	Value in the present study	Interpretation
Absolute fit X2/gl (CMIN/gl)	Less than 2 or 3	1755.32/766 gl = 2.292	The errors of the model are null with the
			used and the absolute fit is excellent
Goodness of fit index (GFI)	> .90 Preferential > .95	GFI = .769	Good fit
Comparative goodness of fit index	> .90 Preferential > .95	CFI = .881	Acceptable comparative fit
(CFI)
Root Mean Square Residual (RMR)	Near zero	RMR = .031	Model error close to zero, almost perfect fit
			of model to data
Root mean square error of approx	Less than .08, close to zero	RMSEA = .068 (.064 - .072)	Model error close to zero, almost perfect fit
-imation			of model to data

The comparative measures of global adjustment, and the comparative fit index (CFI), are very close to the ideal (CFI = .88), as is the Tucker-Lewis index, which takes into account the complexity of the model (TLI = .87), indicating an acceptable fit with the data, since at least 87% of the covariance in the data can be reproduced by the model (^{Byrne, 2010}).

Since the parsimony indices (PCFI = .823) are very close to the optimal value (.900), the model can be described as complex ‒ (^{Hu & Bentler, 1998}), as the goodness of fit statistics show to an acceptable extent (GFI = .769, AGFI = .740, PGFI = .684), in addition to the Root Mean Square Error of Approximation (RMSEA = .068, CI_95% = .064 - .072), which penalizes the increase in the complexity of the model and indicates an excellent fit with the data, since it is less than .08 and is confirmed by the value in the Residual Mean Error (RMR = .031), since this fulfils the requirement of being below the threshold value, indicating nearly zero error and an almost perfect fit (less than .05) (^{Byrne, 2010}).

Discussion and conclusion

The results of this study show that the instrument evaluated is reliable and valid for measuring resilience in patients with CKD. This specific version retained 41 of the 43 original items, with the same number of original subscales, all associated with depression and anxiety symptoms as well as distorted thoughts. In the confirmatory analysis, a theoretical structural model was obtained, which acceptably fits the complexity of the data in this specific sample with kidney patients, according to the most important reliability indices shown and suggested by experts (^{Byrne, 2010}; ^{George et al., 2010}; ^{Hu & Bentler, 1998}; ^{Kline, 2005}; ^{Tabachnick et al., 2001}). The model contains five factors with a strong association between them, as the exploratory analysis showed, which allows us to recommend this instrument to assess resilience in patients with chronic kidney insufficiency (CKI). The most powerful indicators of the fit of the model’s structure (CFI, RMR, and RMSEA) suggest that it is a balanced, parsimonious model (^{Ullman, 2006}).

This adjusted version for measuring psychological resilience could be used to complement information on psychological aspects of kidney patients. The psychological vulnerability of these patients to the disease could be detected by the factorial components of the adapted resilience scale. In future studies, the clinical usefulness of resilience could be evaluated as part of new strategies to reduce dysfunctional behaviors (e.g., dropping out of treatments) and the negative physical (e.g., impaired health), mental (depression and anxiety) and social consequences (such as social abandonment and functional dependence).

This psychological instrument evaluates resilience in patients with CKD, which would enable new studies to understand, analyze, and explore how they use their internal and external resources to positively cope with illness, because some previous studies mainly focus on detecting the negative effects of the disease (^{Cukor et al., 2009}; ^{Dobrof, Dolinko, Lichtiger, Uribarri, & Epstein, 2002}; ^{Elliott, Gessert, Larson, & Russ, 2014}; ^{Ma, Chang, Liu, Hsieh et al., 2013}; ^{Nicholas, Picone, & Selkirk, 2011}; ^{Shidler, Peterson, & Kimmel, 1998}; ^{Tong et al., 2014}; ^{Wick, Bauer, Malessa, Settmacher, Strauss, 2015}).

This study confirms the relationship between distorted thoughts and depression and anxiety symptoms (^{Lerma et al., 2012}) and is now combined with the measurement of resilience in kidney patients. We regard these associations as empirical evidence that support the development of cognitive-behavioral strategies combined with these variables in order to impact psychological and clinical aspects in this population (^{Tong, Lesmana, Johnson, Wong et al., 2013}). For example, the reliable quantitative evaluation of resilience in these patients would make it possible to evaluate the effectiveness of psychological interventions that include resilience to reduce depression, anxiety, and the presence of distorted thoughts, all of which reduce the quality of life (^{Cukor, 2007}; ^{Cukor et al., 2009}; ^{Shidler et al., 1998}). Likewise, this validated instrument would make it possible to study the relationship between resilience and other variables such as family ethnicity (^{White et al., 2002}), time of diagnosis (^{Quiceno & Vinaccia, 2012}), perception of disease (^{Schipper et al., 2014}), social support, devaluation, self-efficacy, and self-esteem (^{Tong et al., 2013}) so that they can be explored in studies of diagnosis, prognosis, and psychological intervention.

Although the constructs of resilience and depression and anxiety belong to different theoretical models ‒with the former coming from a positive psychosocial approach (^{Cabanyes Truffino, 2010}; ^{Rodriguez-Piaggio, 2009}; ^{Vinaccia et al., 2011}) which attempts to identify protective or risk elements or factors‒, whereas depression and anxiety are more closely linked to the medical risk model (^{Cukor et al., 2009}; ^{Ma et al., 2013}; ^{Wick et al., 2015}) (for example, age, sex, education, negative attitudes, or perception of disease) ‒ the relationship found between them in our research suggests the need for new studies to explore and empirically test the mechanisms that explain this relationship from the point of view of their pertinence and clinical interest for measuring the quality of life in patients with CKD. The role of psychological resilience in these patients, as well as the impact it has on the emotional state and quality of life has barely been researched. The instrument validated in the present study could be used to evaluate whether low resilience in a population predisposes it to greater hypersensitivity and inadequate responses to the presence of adverse symptoms of the disease (^{Cukor et al., 2007}; ^{Cukor et al., 2009}; ^{Lerma et al., 2017}; ^{Shidler et al., 1998}) (e.g., hypotension during hemodialysis).

As suggested by previous research (^{Miranda-Zapata, Riquelme-Mella, Cifuentes-Cid, & Riquelme-Bravo, 2014}) our study was conducted with a multicentric sample of patients with CKD, from secondary and tertiary level hospitals (in kidney substitution treatment), which strengthens our confidence in the external validity of the scale used. However, more extensive, heterogeneous studies are required nationwide to increase the generalization of the results that will facilitate the construction of standardized scales of this instrument. The present study did not evaluate the associations between resilience and sociodemographic and clinical variables of patients, which is undoubtedly an important research topic. However, the evaluation of these associations goes beyond the scope of the objectives of this paper and future studies are required to address the issue comprehensively and sufficiently. Another limitation of the study is that it failed to evaluate the cognitive capacity of respondents, one of the factors that can be altered by disease or age. Lastly, the results of the validation, including the structural equations model obtained in the confirmatory analysis, are applicable to patients with CKD only. Other studies must be undertaken to evaluate this scale in patients with other chronic diseases (such as diabetes mellitus) where it is known that there is also high comorbidity of anxiety and depression symptoms.