Introduction
The relevance of family relationships in the outcome of various disorders has been highlighted from different domains, both from Social Support studies (Clinton et al., Reference Clinton, Lunney, Edwards, Weir and Barr1998; Erickson et al., Reference Erickson, Beiser, Iacono, Fleming and Lin1989; Vázquez Morejón et al., Reference Vázquez Morejón, León Rubio and Vázquez-Morejón2018), and from research on expressed emotion (Vaughn and Leff, Reference Vaughn and Leff1976). Empirical studies by Brown et al. (Reference Brown, Birley and Wing1972) on the relationship between the outcome of schizophrenia and various affective dimensions of family relationships have allowed for the identification of particularly relevant aspects: criticism, hostility, emotional over-involvement, warmth and positive comments. These five dimensions, which reflect very relevant characteristics of the emotional climate of the family environment, are integrated into one of the most influential constructs of psychosocial research on schizophrenia, the concept of expressed emotion (EE). This construct refers to interaction patterns that are reflected in attitudes and communication styles of family members. The assessment of these dimensions by means of the Camberwell Family Interview (CFI) allows the identification of family environments of high or low expressed emotion, with particularly the environments of high expressed emotion being associated with the outcome of people with psychosis.
The robust predictive ability for relapses in individuals with schizophrenia has given this construct a special interest (Weintraub et al., Reference Weintraub, Hall, Carbonella, Weisman de Mamani and Hooley2017). Two of these dimensions (criticism and emotional over-involvement) have been considered especially relevant in terms of this predictive capacity (Leff and Vaughn, Reference Leff and Vaughn1985). Subsequent studies have also confirmed the role of this construct in the outcome of other disorders such as depression (Uehara et al., Reference Uehara, Yokoyama, Goto and Ihda1996), anxiety (Chambless and Steketee, Reference Chambless and Steketee1999) and eating disorders (Duclos et al., Reference Duclos, Vibert, Mattar and Godart2012; Rienecke, Reference Rienecke2018).
The assessment of the EE is carried out through the CFI, an interview that in its abbreviated version (Vaughn and Leff, Reference Vaughn and Leff1976) takes an approximate application time of an hour and a half, and two or three hours of evaluation. This extensive duration, as well as the intense training required for its evaluation, significantly diminishes its usefulness and feasibility in clinical practice. For this reason, other assessment instruments have been developed that are shorter and more feasible in clinical contexts, such as the Family Assessment Scale (FAS) (Kavanagh et al., Reference Kavanagh, O’Halloran, Manicavasagar, Clark, Piatkowska, Tennant and Rosen1997) and the Family Questionnaire (FQ) (Wiedemann et al., Reference Wiedemann, Rayki, Feinstein and Hahlweg2002), both based on the family’s perception. However, there are very few instruments that, for the evaluation of the dimensions of EE, are based on the person’s own perception. The Perceived Criticism Scale (PCS) by Hooley and Teasdale (Reference Hooley and Teasdale1989) does so, but in this case, it is exclusively focused on the perception of criticism.
The relative independence of the family perspective and the person’s own perspective (Medina et al., Reference Medina-Pradas, Navarro, Pousa, Montero and Obiols2013) make it necessary to have specific instruments for each perspective, in order to explore the relationships they present with different variables and, very particularly, with the predictive capacity of each perspective. The interest in the patient’s own perspective has been highlighted in various studies, emphasizing the importance of one’s own subjective perception, as some studies indicate that this perspective might have more impact on the stress level of individuals with psychosis and on the risk of relapses (Cutting et al., Reference Cutting, Aakre and Docherty2006; Lebell et al., Reference Lebell, Marder, Mintz, Mintz, Tompson, Wirshing and McKenzie1993; Tompson et al., Reference Tompson, Goldstein, Lebell, Mintz, Marder and Mintz1995; Warner and Atkinson, Reference Warner and Atkinson1988).
The Influential Relationship Questionnaire (IRQ), a modified version of the Parental Bonding Instrument (PBI), developed by Baker et al. (Reference Baker, Helmes and Kazarian1984), includes, in addition to the dimensions of care and over-protection, a third dimension related to criticism. The IRQ takes as a reference period the last year of the parental relationship, allowing the assessment of the basic attitudes of the current parental relationship (a) quickly and therefore of interest to clinical practice, and (b) from a particularly important perspective: the person’s own perception of parental attitudes. These characteristics, along with the inclusion of the criticism dimension, allow this instrument to be proposed as a possible alternative to the CFI.
Incorporating additional dimensions into the IRQ, such as over-protection and, more notably, positive aspects of family relationships like the care dimension, enhances its relevance. This is especially significant given the crucial role these positive affections play as protective factors across various cultures (Ivanović et al., Reference Ivanović, Vuletić and Bebbington1994; Lee et al., Reference Lee, Barrowclough and Lobban2014; López et al., Reference López, Nelson Hipke, Polo, Jenkins, Karno, Vaughn and Snyder2004).
Therefore, it seems of interest to have a Spanish version of the IRQ that confirms its psychometrics characteristics and facilitates studies in the Spanish language in this relevant research area, and to try to develop an abbreviated version of this instrument in order to make it more viable in clinical care contexts.
Method
Participants
One hundred and eighty-eight participants that were treated in a community mental health unit, with a diagnosis of schizophrenia or related disorders according to the ICD-10 criteria, of which 120 were men (63.8%) and 68 were women (36.2%) were selected for this study. Diagnosis was made according to the ICD-10 criteria by specialist in clinical psychology, following a clinical interview that included the collection of clinical history and a psychopathological examination.
The mean age of the participants was 32.05 years (SD=7.86, range 16–52 years). Other sociodemographic and clinical characteristics of the participants are presented in Table 1. One hundred and thirty-six participants provided data related to both father and mother, and 52 only related to mother or father, so the analyses were carried out with a total of 324 questionnaires. For the application of the FAS to relatives, 50 parents participated: 32 mothers and 18 fathers.
Table 1. Sociodemographic information (n=188)
Measures
Influential Relationships Questionnaire (IRQ; Baker et al., Reference Baker, Helmes and Kazarian1984)
This self-report measure consists of 37 items that describe a father or mother’s behaviour or attitude towards their child. Twenty-five of the items originate from the Parental Bonding Instrument (Parker et al., Reference Parker, Tupling and Brown1979) and correspond to the dimensions of over-protection and care, while the remaining 12 were added by Baker et al. (Reference Baker, Helmes and Kazarian1984) and correspond to the dimension of criticism. Each item is rated on a Likert scale ranging from 0 (never) to 3 (frequently). It is completed by the participants in relation to both the father and the mother. The sum of the items of each of the dimensions, after reversing the scoring of the corresponding items, allows obtaining the score corresponding to care, over-protection, and criticism. It displays adequate psychometric characteristics, with internal consistency between .77 in the criticism dimension and .84 in care, and a test–retest reliability ranging between .53 in over-protection and .63 in the criticism dimension. Subsequent studies have confirmed these adequate psychometric characteristics in other populations as well as their predictive validity in relation to relapses in individuals with schizophrenia (Baker et al., Reference Baker, Kazarian, Helmes, Ruckman and Tower1987).
Perceived Criticism Scale (PCS; Hooley and Teasdale, Reference Hooley and Teasdale1989)
The PCS is a simple and quick measure to evaluate perceived criticism, a dimension considered one of the most relevant in studies on EE. It consists of a single question (PCS1) in which the participants are asked to evaluate, on a 10-point Likert scale, to what extent their family member is critical towards him/her. Additionally, a second question (PCS2) is added in which they are asked to value to what extent they consider themselves to have been critical towards their family member, following the same 10-point Likert scale. The correlations found in various studies in relation to EE, assessed using the CFI, and what could be more important, its predictive ability concerning the outcome of various disorders has come to support the interest in this measure (Hooley and Miklowitz, Reference Hooley and Miklowitz2017; Hooley and Teasdale, Reference Hooley and Teasdale1989). It presents satisfactory temporal reliability (r=.75), and its concurrent validity has been confirmed by its correlation with the global level of EE (high-low), measured with the CFI. Its predictive ability concerning the outcome of different disorders gives particular interest to this measure. A third question (PCS3) was added in this study about the degree of satisfaction in their relationship with the family member, with the same scoring procedure as in previous questions: ‘In general, how do you feel about your relationship with your father/mother/relative?’, ranging from 1 (very dissatisfied) to 10 (very satisfied).
Family Attitudes Scale (FAS; Kavanagh et al., Reference Kavanagh, O’Halloran, Manicavasagar, Clark, Piatkowska, Tennant and Rosen1997)
The FAS is an instrument composed of 30 items that measure the emotional family climate in patients with schizophrenia. The scale seeks to provide a valid and cost-effective alternative to assess environmental stress factors (criticism and rejection) associated with the course of schizophrenia. Relatives report how frequently each of the items presented in the scale occurs, answering on a scale ranging from 4 (every day) to 0 (never). In ten of the items, the scores are reversed, and subsequently, a total score is obtained by summing all the items. The total score ranges between 0 and 120, with higher scores indicating higher levels of criticism. The scale shows high internal consistency, and its score correlates with various dimensions of the Camberwell Family Interview, supporting its convergent validity. This study utilizes the Spanish adaptation (Jiménez Garcia-Bóveda et al., Reference Jiménez Garcia-Bóveda, Vázquez Morejón and Vázquez-Morejón2007) in which its reliability and validity are also confirmed.
Procedure
After conducting a double translation of the questionnaire, differences between both translations were reconciled, taking into consideration the Spanish terminology related to studies on EE. Subsequently, a translation into English was performed by an independent translator, and both versions were compared, resulting in minor adjustments in certain terms. The questionnaire was then administered to 10 participants to identify any potential comprehension difficulties, and appropriate corrections were made.
From 2013 to 2017, all patients diagnosed with schizophrenia or related disorders, who lived with family members and were receiving treatment at the community mental health unit, were asked to complete the scales. After being provided with information and upon obtaining informed consent, including permission for the results to be published, the questionnaire was distributed along with routine evaluation scales. Instructions were both orally explained and written on the questionnaire itself, which was to be completed independently by the patients. Simultaneously, data relating to age, marital status, living arrangements, and diagnosis were collected. Of the 324 questionnaires conducted, 183 (56.5%) referred to the mother, and the remaining 141 (43.5%) were related to the father.
Simultaneously, the PCS was applied to 130 participants, and in 50 cases, family members were asked to complete the FAS. In 80 participants, the completion of a second application of the questionnaire was requested within a period of 3 months to explore its temporal reliability. Once the collection was completed, the responses were scored and coded, and statistical analyses were conducted using SPSS v22.
The entire procedure was carried out in accordance with the Helsinki Declaration, and the study was approved by the Andalusian Biomedical Research Ethics Committee under code no. 1384-N-19.
In this adaptation, the possible reduction of the scale was considered by selecting exclusively those items that met saturation criteria for a single dimension, as specified in the ‘Statistical analysis’ section below.
Statistical analysis
Factorial analysis was conducted using the Kaiser-Meyer-Olkin (KMO) test and Bartlett’s test of sphericity. Subsequently, principal components analysis (PCA) with Oblimin rotation was conducted. Items with a saturation greater than .50 and less than .20 in other dimensions were selected, also checking content validity and apparent validity, as well as the effect on Cronbach’s alpha coefficient when eliminating each item. In this way, items that did not meet the appropriate saturation conditions in their factor were eliminated, and a new PCA with Oblimin rotation was conducted using only the selected items.
As the scale does not aim to obtain a global score but scores in different dimensions, the internal consistency and factorial structure of each of the dimensions were explored using Cronbach’s alpha coefficient and omega coefficient in the first case, and through one-factor factorial analysis with Oblimin rotation in the second. For temporal reliability, the intraclass correlation coefficient was used, and for concurrent validity, Pearson’s correlation was employed.
Results
Descriptive statistics
The scores on the 37 items ranged between 0 and 3, with the highest mean score on item 22 (‘Made me feel I wasn’t wanted’) at a mean of 2.3, and the lowest on item 31 (‘Let me go out as often as I wanted’) with a mean of 0.35 (see Table 2).
Table 2. Descriptive statistics and principal component analysis of the 37 IRQ items
Factorial structure
The results of the Kaiser-Meyer-Olkin test (.937), and Bartlett’s test of sphericity (.000), confirmed the appropriateness of the data for factorial analysis. Through an initial PCA, a structure composed of six factors was obtained (see Table 2), with two of the factors made up of a single item. In both cases, these are items worded as negations, and due to their content without clear differentiation, both factors could be considered as artifacts.
After eliminating both items (2 and 13) and those items that did not meet the saturation conditions (>.50 in their own factor and <.30 in other factors), 24 items were selected, and a second PCA with Oblimin rotation was performed from these 24 items (Table 3). This PCA allowed the identification of four factors explaining 61.53% of the total variance and, by the contents of the items, they can be clearly defined and thus termed: criticism (7 items), over-protection (6 items), restriction (5 items), and care (6 items).
Table 3. Principal component analysis of the 24 IRQ items
Factorial validity
The items of each of the four dimensions of the IRQ displayed a single factor explaining between 55.35% in the restriction dimension and 65.49% in the care dimension, with items that saturate above .50 in all cases (see Table 4).
Table 4. Reliability and intercorrelations of the IRQ
** Correlation is significant at the .01 level (2-tailed).
The mean scores of each of the dimensions ranged between 2.82 in restriction and 13.22 in care dimensions (Table 4), indicating in all cases a higher score, reflecting greater intensity in that dimension. The analysis of the distribution of the data showed that they were spread across the entire range of scores, for both the items and the various dimensions.
Internal consistency
The values obtained through Cronbach’s alpha coefficient, as well as through the omega coefficient, revealed high consistency within each of the four dimensions of the IRQ (Table 4), ranging between .80 for restriction and .89 for care in the case of Cronbach’s alpha, and between .81 for restriction and .89 for care with respect to the omega coefficient.
Temporal reliability
Temporal reliability, assessed through the Intraclass Correlation Coefficient, exhibited strong correlations (Table 4), ranging between .73 and .78, all cases above .70, thus confirming the reliability of the various dimensions of the IRQ.
Correlation among dimensions
The four dimensions showed very significant moderate correlations, with positive correlations observed between criticism, over-protection, and restriction, while the care dimension displayed negative correlations with the other three dimensions (see Table 4), as might be expected given the content of each dimension.
Concurrent validity
The scores obtained in the dimensions of the IRQ exhibited significant and strong correlations with those obtained in the PCS (Table 5). As expected, the criticism dimension of the IRQ presented the most intense correlation, with a large effect size, with the score on the perception of criticism in PCS1. The care dimension of the IRQ also showed a significant correlation with the perception of criticism in PCS1, and as expected, of a negative nature, with lower scores in criticism as higher scores in care are observed.
Table 5. Concurrent validity, correlations between IRQ dimensions and PCS
** Correlation is significant at the .01 level (2-tailed).
On the other hand, both the overprotection and the restriction dimensions of the IRQ showed significant correlations with the three PCS scores, positive in relation to the two regarding criticism, PCS1 and PCS2, and negative with the one related to satisfaction (PCS3), with a more intense negative relationship in the case of the restriction dimension.
Construct validity
Construct validity was analysed based on the correlation between the scores obtained in the FAS, an instrument that assesses family criticism through the family’s self-report, and the dimensions of the IRQ in 50 participants. A moderate and significant correlation was observed with the criticism dimension (r=.38, p<.005) and restriction (r=.41, p<.005), and moderate but negative (r=.42, p<.005), between the score in FAS and the care dimension. There was no correlation between the score in FAS and the overprotection dimension of the IRQ.
Discussion
The findings of the present study exhibit satisfactory psychometric characteristics of this abbreviated Spanish adaptation of the IRQ, with data that affirm its reliability and validity.
A primary aspect to highlight is the differences regarding the factorial structure in relation to the original scale. On the one hand, in relation to the number of items ultimately included, various items were discarded due to their lack of clear saturation in any of the dimensions, or because they formed a dimension that was nonsensical in its content and could represent an artifact due to their negative wording. Therefore, it is essential to note that, unlike the original IRQ, this version consists of a reduced 24 items instead of 37. On the other hand, it is necessary to underline that, unlike the factorial structure obtained by Baker et al. (Reference Baker, Helmes and Kazarian1984), which was formed by two dimensions from the PBI, care and over-protection, plus the one added by the authors related to criticism, the present study identified a fourth dimension that, due to its content, can be termed restriction. While the three dimensions of criticism, over-protection, and care are similar to those found in the original scale, the fourth dimension seems to identify a variant of over-protection, related to a type of control that is characterized not by excessive care, but by a restriction and limitation of the patient’s behaviours, impeding their autonomy.
These results align with those observed by other authors. In a Spanish adaptation by Gómez-Beneyto et al. (Reference Gómez-Beneyto, Pedrós, Tomás, Aguilar and Leal1993) and a Japanese adaptation by Sato et al. (Reference Sato, Okada, Morikawa, Nakamura, Yamauchi, Ando and Ozaki2021) of the PBI, a restraint dimension was identified. This third dimension, which is not present in the original version of the PBI (Parker et al., Reference Parker, Tupling and Brown1979), includes the same five items.
Regarding the internal consistency of each of the four dimensions, it must be pointed out that the obtained values confirm the high consistency of the items forming each of these dimensions, with Cronbach’s alpha values ranging between .80 and .89, very similar to those found by Baker et al. (Reference Baker, Helmes and Kazarian1984), with values between .76 and .91, although in this case obtained from the three dimensions found in their study.
Additionally, the results confirm test–retest reliability by the correlations obtained between both applications, with values that in all cases are above the .70 recommended for temporal reliability (Nunnally and Bernstein, Reference Nunnally and Bernstein1994). The effect size is large, even more significant when considering the 3-month interval used in the present study. Baker et al. (Reference Baker, Helmes and Kazarian1984), in the three dimensions, obtained correlations that fluctuated between .53 and .85, although it concerned a very reduced sample (n=26) where higher reliability was observed concerning the second person chosen as most influential by the participants.
As for concurrent validity, it must be noted that the results also support this validity, with significant correlations between the dimensions of IRQ and the scores obtained on items of an evaluation instrument of perceived criticism, such as the PCS. In this sense, the correlations between the criticism dimension of IRQ and the item of perception of criticism of PCS show values with a large effect size, endorsing the concurrent validity of the IRQ. Considering that PCS is an instrument for the assessment of criticism equally assessed from the patient’s perception, it is predictable to expect significant and intense correlations, results that coincide with those obtained in the present investigation. The care dimension, on the other hand, correlates negatively with PCS1 (r=–.37), very similar to the correlation obtained by Lee et al. (Reference Lee, Barrowclough and Lobban2014) (r=–.39).
Likewise, the construct validity is supported by the significant correlation between the dimensions of IRQ and the item related to satisfaction with the relationship, with values that, as expected, are negative in relation to the dimensions of criticism, over-protection, and restriction, and positive with the care dimension.
Furthermore, the construct validity is equally supported by the observed correlation between the criticism dimension of IRQ and the scores obtained on the FAS, correlations of a moderate effect size, as is expected to proceed from a different information source, provided from the perception of the relatives and not by the participants themselves. Equally moderate correlations have been observed between the criticism dimension of CFI and the Level of Expressed Emotion Scale (Cole and Kazarian, Reference Cole and Kazarian1988). Although correlated, various studies show that the perception of the family’s emotional climate by the patients differs from the perception obtained from the parents’ interview (Bachman et al., Reference Bachmann, Bottmer, Jacob and Schröder2006; Medina et al., Reference Medina-Pradas, Navarro, Pousa, Montero and Obiols2013; Onwumere et al., Reference Onwumere, Kuipers, Bebbington, Dunn, Freeman, Fowler and Garety2009). The lack of correlation observed between the FAS score and the over-protection dimension of the IRQ could be due to the different sources of information of the instruments as well as the greater complexity of the over-protection dimension.
Limitations
One primary limitation of this study is its sample composition, which includes only individuals with psychosis. Consequently, the findings cannot be generalized to other diagnostic groups. It should also be noted that it has not been possible to validate in relation to the CFI, the reference instrument for the evaluation of EE. Another important limitation is the absence of data regarding the predictive ability of this adaptation of the IRQ. Moreover, as to the factorial structure of the scale, the results allow for identifying a structure with four relatively independent dimensions, although its confirmation through confirmatory factor analysis remains pending.
Future research
In future studies, therefore, it would be of interest to confirm the validity and reliability of the IRQ in more diverse samples, including other diagnostic groups, particularly in groups where relationships with father and mother are particularly relevant, as is the case with individuals with eating behaviour disorders. Additionally, it is particularly important to confirm the factorial structure in new samples using confirmatory factor analysis and to evaluate the predictive capacity of the IRQ regarding the outcomes of various disorders. This aspect is of significant interest for clinical practice. This was already confirmed by the authors in their original version concerning relapses in a specific group such as people with schizophrenia (Baker et al., Reference Baker, Helmes and Kazarian1984).
Also, it would be of interest to explore the predictive capacity of the care dimension, a particularly intriguing aspect given how scarcely explored this area is and the suggestive results of some studies on the role of positive affections as protective factors in the early stages of psychotic episodes in certain cultural contexts (Butler et al., Reference Butler, Berry, Varese and Bucci2019; González-Pinto et al., Reference González-Pinto, de Azúa, Ibáñez, Otero-Cuesta, Castro-Fornieles, Graell-Berna and Arango2011; Ivanović et al., Reference Ivanović, Vuletić and Bebbington1994; Lee et al., Reference Lee, Barrowclough and Lobban2014; López et al., Reference López, Nelson Hipke, Polo, Jenkins, Karno, Vaughn and Snyder2004).
As Baker et al. (Reference Baker, Helmes and Kazarian1984) show, only the current representation of relationships with significant others shows a relationship with relapses, allowing differentiation between participants who relapse from those who do not experience relapses. Therefore, it is particularly important to have instruments like the IRQ that facilitate the assessment of the perception of the family emotional climate by the patients themselves. As pointed out by Leff and Vaughn (Reference Leff and Vaughn1985), evaluating patients’ perceived attitudes towards the most significant people in their environment is particularly crucial, as these attitudes can exacerbate symptomatology, impacting the outcome of vulnerable individuals.
Data availability statement
Research data are not shared.
Acknowledgements
We would like to thank the authors for granting permission to adapt the scale, and to the patients and their families for their collaboration.
Author contributions
Antonio Vazquez Morejón: Conceptualization (lead), Formal analysis (equal), Investigation (equal), Methodology (equal); Marta Lopez Narbona: Formal analysis (equal), Investigation (equal), Methodology (equal); Miguel Romero González: Investigation (equal), Methodology (equal); Raquel Vazquez-Morejon: Formal analysis (equal), Investigation (equal), Methodology (equal).
Financial support
This research received no specific grant from any funding agency, commercial or not-for-profit sectors.
Competing interests
The authors declare none.
Ethical standards
The study was approved by the Andalusian Biomedical Research Ethics Committee under the code no. 1384-N-19. Participants gave informed consent to participate in the study and for the results to be published.
Open access
Comments
No Comments have been published for this article.