Cross-Cultural Adaptation of the General Functioning Subscale of the McMaster Family Assessment Device

Background: The aim of the study was to conduct a cross-cultural adaptation and validation of the General Functioning Subscale (GF12) of the McMaster Family Assessment Device (FAD) into the Malay language. Methods: The translation and adaptation procedure of the Malay GF12 was based on the dual-panel methodology. This involved conducting a bilingual panel (providing the initial translation into Malay language) followed by a lay panel (where items are assessed for comprehension and acceptability). A mixed-methods approach with exploratory sequential study design was employed. Results: Two hundred and fty-one families who have children in pre-schools (Tadikas) answered the Malay GF12. The KMO value for sampling adequacy was 0.81, indicating the sample was adequate for factor analysis. The Malay GF12 demonstrated high internal consistency (Cronbach’s alpha coe�cient > 0.80). In the con�rmatory factor analysis, the original 12-item model did not �t very well. Alternatively, a 2-factor-6-item model showed a su�cient �t. The two constructs comprised Positive Items and Negative Items. Conclusions: The results indicate the feasibility of administering 6 of the 12 items of Malay GF12 (3 positively worded and 3 negatively worded) to measure family functioning in the Malaysian population.


Background
Families play a major role in the expression of variety of ailments including psychiatric and oral disorders 1,2 .3][4] .Nevertheless, the child's oral health and behaviour have received less attention from family functioning and relationships 5 .Family is the primary unit of socialization during childhood, and is central in shaping engagement in health behaviour, including physical activity.Thus, it has been propounded that a child's health status could be directly linked to parents health behaviour including disease and genetic composition 6 .Furthermore, families that are subjected to different ecocultural pathways may re ect altered socio-economic conditions and subsequent variations in resource allocation.As a result, personal interest, time, income generation, educational background and social network activities all play a signi cant role.Thus it becomes imperative to look into the family functioning and how the resources are utilized and their effects on the health 6 .Family functioning may be described as a balance between family cohesion and family adaptability to challenges within the family and the environment 5 .Health literacy of a family may in uence communication of health information to the family, individual health beliefs and health-related behaviour 7 .A child is part of a larger family system and the family, in turn, is part of a broader neighbourhood or community system.Consequently, any changes in the family may affect the child and changes in the child development 8 .Perceived family functioning can be studied extensively using a self-reported assessment tool, Family Assessment Device (FAD) 9 .The FAD is a 60 item measure that is built on McMaster Model of Family Functioning comprises six family functioning domains: general problem solving, communication, roles, affective responsiveness, affective involvement and behaviour control 10,11 .
A shortened version of the FAD comprising 12 items has been formulated to measure general family functioning with excellent psychometric properties 12 .General Functioning subscale (GF12) is a shortened version of FAD with all the domains but reduced items for rating.Each item is rated on a four point-Likert scale with indicators ranging from strongly agree to strongly disagree 13 .The GF12 is adapted into various languages for ease of application and usage with ethnic communities.The prominent translations incorporate Hispanic-Americans 14 , Hawaiian and Japanese American 15 subpopulation and Euro-Asian communities [16][17][18][19][20] .
Various methods have been employed to appropriately translate the instrument for cross-cultural adaptation and implementation with respect to self-reported health outcome questionnaire [21][22][23][24] .
Forward-backward (FB) translation method is the most popular method 25 .Another method utilized a consensus translation methodology with two different panels.This dual panel (DP) approach is rst attempted with panel who are versed with bilingual skills that are native to the target translation and accompanied by a representative of the developers of the instrument.The second panel comprise subjects versed with multi-linguistic abilities but educationally possess average or below average literacy 26 .
An investigation of family functioning in the context of Malaysia's diverse family structures, from urban to rural, single parent to extended families, within a multi-racial society can provide an understanding of the multi-factorial causes of chronic non-communicable oral diseases, including caries.Rapid urbanization, such as that being witnessed in Malaysia, has considerable impacts on living conditions in terms of housing and access to health and social services, which in turn will impact on health outcomes 27 .There is a need to measure the impact of family functioning on the health of the Malaysian population.The multi-dimensional nature and variability associated with the family functioning predisposes a complex background for a deliberate assessment.The GF12 instrument has been widely used in this pursuit, but translation and validation in local language is a requisite.

Methods
The intent of this study was to adapt cross-cultural attributes to the GF12 instrument of the FAD and subsequent psychometric validation in the Malay language.The methodology adopted mechanism elaborated from guidelines and strategies by Guillemin and colleagues 28 .The rst step included a precise translation as outlined by the guidelines.This was followed by a pre-test method which involves a test of the cognitive attribute and subsequently, the process involves pilot testing.The Malay version was transcribed from the English version based on the semantic, operational and measurement equivalences.

Translation of the English GF12 into Malay Language
A dual panel approach was utilized.This method was suggested by Hunt et al 22 and supported by Swaine-Verdier et al for a concept based translation 23 .Linguistic equivalence was not considered in this approach.The contemplation is based on the premise that obtaining "natural" translation for an item in a new language is a di cult proposition.Hence, a word that is equivalent to the conceptual meaning is usually sought.Further, the translation is attempted to harness words that are in common usage.
The dual-panel translation procedure comprised the following main steps: Panel 1: Five bilingual healthcare professionals consisted of a nurse, a public health specialist, a community nutritionist, a dentist and a dental support assistant were invited to attend a one-day discussion forum to produce the rst draft of the Malay GF12 that was conceptually equivalent with the English version.
Panel 2: This panel consisted of a monolingual nurse, a dental surgery assistant and three persons from the targeted community (lay people) with the Malay language as their rst language.A one-day discussion forum was organized within the premise of the International Medical University (IMU).The focus of this group was to review the rst draft of the Malay GF12 to ensure the translations were expressed in a natural, everyday Malay language.This thorough scrutiny of the translation process was facilitated by two members of the research team.At the end of the session, the panel agreed on the draft Malay GF12 with minor changes.

Validation of the draft Malay GF12
A mixed-methods approach with exploratory sequential study design was employed.Face validity and content validity of the draft Malay GF12 was tested using a pre-test qualitative mechanism.The reliability and construct validity in the assessment of psychometric properties was tested using a pilot methodology.At each stage, a signed informed consent was obtained from the participating parent/guardian of the child.The consent form and study information sheet were sent to the parents through the school.Con dentiality and safe record keeping were ensured to the participants at both data collection stages.
Pre-test: The pre-test of the draft Malay GF12 involved an expert panel of eight members (two general dentists, two dental public health specialists, two nurses, and two dental academicians) and a group of parents of preschool children.The objectives of the pre-test were to assess the presence of ambiguities in the translation process, identify items with inappropriate conceptual levels, and identify wordings that were confusing to understand 24 .The questionnaire was emailed to the eight-member expert panel who were bilingual.They were instructed to comment on the conceptual equivalence and wordings of the draft Malay GF12.As for the parents, they attended a focus group discussion (FGD) to assess whether the questionnaire purpose, instructions and items were relevant and easily understood, and the item concepts were culturally relevant.The FGD was audio recorded and eld notes were taken to validate the ndings.Interpretation biases and discrepancy in the observations was tackled using 'member check; approach.
The two pre-test outcomes (expert panel and parents) were reviewed by the research team and the consensus draft of the Malay GF12 was nalized.
Pilot test: A total of 510 parents of 5-6-year-old children from 11 Tadikas (kindergartens) randomly selected in the Petaling district were invited to participate in the pilot test.As per the dictation of the pilot protocol, parents answered question related to their pre-school children and their socio-demographic status.The entire episode was conducted in the presence of a member of the research team.
Subsequently, 15 parents were asked to answer the Malay GF12 questionnaire two weeks later.

Data entry and analysis
Pre-test: The data entry and analysis applied using NVivo-11 software.A textual data was obtained from the recorded data and eld notes from the FGD using verbatim transcription method.A detailed thematic analysis followed the above procedure.
Pilot test: IBM Statistical Package for the Social Science (Chicago, IL, USA) version 25 software was used for data analysis.Scoring of the Malay GF12 was based on a 4-point Likert scale (1 = strongly agree to 4 = strongly disagree).The scores for negative items were reversed and total score was calculated by summing up the scores of the 12 items.Descriptive statistics (frequency, mean and standard deviation) were used to describe parents' socio-demographic details.The psychometric properties of the Malay GF12 were assessed in terms of internal consistency, test-retest reliability, and construct validity.
Cronbach's alpha coe cient measured the internal consistency of the Malay GF12.The homogeneity between the items was explored in this assessment.The statistical computations including intra-class correlation (ICC) coe cient, exploratory factor analysis, con rmatory factor analysis (CFA) were employed to analyse test-retest reliability, construct validity, assessment of the measurement equivalence of Malay GF12, respectively.The analysis examined the instrument factors and factor loadings including expressiveness of factor loadings with respect to the structure of GF12 through convergent and discriminant validity.IBM® SPSS® Amos™ software was utilized to perform the above computations.Amos™ uses the maximum likelihood estimation (MLE) method which requires the assumption of multivariate normality (MVN).Expressiveness of factor loadings, residual variances and modi cation indexes were analysed.The best model t with value < 0.08 is represented as the Root Mean Square Error of Approximation (RMSEA).Good results for the Comparative Fit Index (CFI) and the Tucker-Lewis Index (TLI) are within the range of > 0.90.The Goodness-of-Fit statistic (GFI) calculates the proportion of variance that is accounted for by the estimated population covariance and GFI cut-off point of ≥ 0.90 shows how closely the model comes to replicating the observed covariance matrix.The AGFI adjusts the GFI based upon degrees of freedom, and generally the values of ≥ 0.90 indicate well-tting models.Furthermore, Composite Reliability (CR) values > average variance Extracted for the items (AVE) values indicate good t of the model.To test for discriminant validity, the magnitude of the shared variance between the nal 2 constructs (R 2 ) is checked to be less than the within construct variances (AVEs).A non-signi cant value for χ 2 , values as close as possible to 1.00 for adjusted goodness-of-t index (AGFI), values higher than 0.95 for normed t index (NFI) and comparative t index (CFI), a value as close as possible to zero for standardized root mean square residual (SRMR), and a value lower than 0.05 for root mean square error of approximation (RMSEA) are indicative of a good t between the estimated model and input data 29 .Hu et al noted a value of RMSEA as high as 0.08 and values for CFI ranging from 0.90 to 0.95 for an acceptable t of a con rmatory factor model 30 .

Results
For the pre-test of the draft Malay GF12, all the eight members of the expert panel and the eight parents who were approached responded (100% response rate).All parents were of Malay ethnicity.They found the translated version of GF12 in Malay language easy to adapt and simple.In item 1, the nal Malay word for the word "sukar" (di cult) was changed to "susah" (a synonym of "sukar"), in item 2 the word "krisis" (crisis) was changed to "masalah" (a synonym of "krisis"), and in item 5 the word "gusar" (concern) was changed to "risau" (a synonym of "gusar").Item 12 of the Malay version was rephrased from "Kami bersikap terbuka di antara satu sama lian" (We con de in each other) to "Kami selesa berkongsi masalah antara satu sama lain" which was conceptually equivalent.Table 1 shows the nal Malay translation against the original version of GF12 subscale after nal amendments were made based on decisions from the review committee.Making decisions is a problem for our family Membuat keputusan adalah suatu kesukaran dalam keluarga kami.We are able to make decisions about how to solve problems Kami boleh membuat keputusan bagaimana untuk menyelesaikan masalah.We don't get along well together Kami tidak sehaluan antara satu sama lain.We con de in each other Kami selesa berkongsi masalah antara satu sama lain.
For the pre-test, of the 510 parents who were approached, 251 parents agreed to participate (49.5% response rate).The parents' children were almost equally distributed in terms of gender (male = 50.2%,female = 49.8%) and age (5-year-olds = 46.2%,6-year-olds = 53.8%).Majority of the parents/guardians were working fulltime (56.6%) or were housewives (45.4%) and belonged to a low-income group (67.3%).Table 2 shows the participants' socio-demographic information.The parents were of urban and rural backgrounds.All the respondents were Malays (n = 251) with the majority having education up to secondary school level (90%), and nearly half were working full-time (55.6%).The Cronbach alpha of the Malay GF12 was 0.89.For test-retest reliability analysis, the ICC value was more than 0.75.Factor analysis of the Malay GF12 revealed that for each item the highest correlation with at least one other item in the construct was between 0.3 and 0.9 indicating the items correlated adequately in the construct.The KMO value was 0.8 which is considered good.Two factors emerged from the analysis: factor 1 consisting of positively worded items (2, 4, 6, 8, 10, 12) and factor 2 consisting of negatively worded items (1, 3, 5, 7, 9, 11) (Table 3)..683We don't get along well together. .676 A theoretical model (Fig. 1) allowed for the application of con rmatory factor analysis and the results are shown below.
From the analysis, the value of Chi-square/df was more than 3, the CFI, TLI, GFI, AGFI values less than 0.9 respectively, and the RMSEA value was more than 0.08 indicating the model t was not acceptable.
Based on the factor weights, two factors were extracted (Table 4).Items: 12, 10, 8, 6, 4, 2 are components of factor 2 and 11, 9, 7, 5, 3, 1 are components of factor 1 as per the description in the theoretical model (Fig. 2).In the con rmatory factor analysis, the model of theoretical structure (in Fig. 3), was acceptable.In the nal model, the two-factor-three-item model was found to t well.The rst factor is a measure of Positive Items and the second factor is a measure of Negative Items.The AVE values were 0.736 and 0.732, respectively.The correlation value between the two factors was 0.67, which is less than the square roots of the AVE values.Hence there is su cient discriminant validity between the two constructs.The minimum factor loading was 0.77.

Discussion
The purpose of this study was to explore and undertake the cross-cultural adaptation and validation of the Malay GF12 of the FAD.The study ndings indicate that the Malay GF12 consisting of 2 factors with 3 items in each factor is valid and reliable to assess family functioning in the Malaysian setting.
The translation and adaptation of the English GF12 into Malay version proceeded according to plan 22 .The two panels which comprised bilingual and research experts were able to produce the Malay translation of the GF12 without much di culties.Bilingual panel demonstrated adequate ability in translation with respect to their educational and linguistic capabilities compared to the general population.This nding was in accordance to Streiner and colleagues observations 31 .
In the pre-test, a few Malay items of the GF 12 were revised.Overall, the expert panel recommended that a revision be made to the wordings of Item 12 while the lay panel recommended that changes be made to a few words in item 1, item 5 and item 2. These changes were necessary to promote the readability and understanding of the Malay version by the target population, while at the same time maintaining the conceptual equivalence of the items in the English version.The review panel comprising main investigator and other researchers reviewed and agreed to the changes and adjustments in the translated version.
In the con rmatory factor analysis, the original 12-item model did not produce a well-tted model because the RMSEA value of 0.227 was larger than the recommended value of less than 0.05 for a welltted model 32 .Instead, a two-factor and six-item model showed a more su cient t.The results indicate the feasibility of administering six of the twelve items of the Malay GF12 (3 positively worded and 3 negatively worded items) to measure family functioning in the Malaysian population.Our study ndings are different from the ndings of the Italian GF-12 subscale validation which has a 4-factor model with the following domains: competence, emotional, communication, and centre-on-self 16 .A possible explanation could be attributed to the differences in education levels and cultures between the two populations which conceptualize normal family functioning differently 33 .To examine the latent structure of general functioning of the family (GFF), con rmatory factor analysis was applied to verify the number of underlying dimensions of the instrument (factors) and how the item-factor relationship (factor loadings) is manifested.The results showed that the two-factor-three-item model has good construct validity and reliability in the Malaysian context.Based on the con rmatory factor analysis data, the nal two-factor-three-item model shows a better t, with the exclusion of Item 1: "Planning family activities is di cult because we misunderstand each other", Item 2: "In times of crisis we can turn to each other for support", Item 3: "We cannot talk to each other about the sadness we feel", Item 4: "Individuals are accepted for what they are", Item 5: "We avoid discussing our fears and concerns", and Item 12: "We con de in each other", with residue values that correlate with that of other six items.This may indicate that these items represent similar ideas of the other items, therefore being redundant in this instrument.Different cultural norms regarding family functioning may have affected the results which may also vary according to socioeconomic status.Despite a good correlation, the items 1, 2, 3, 4, 5 and 12 were intended to be maintained and that would allow for decision-making in analysis phase.
The reliability of the Malay GF12 was tested in terms of internal consistency and test-retest reliability whereby ndings were in agreement with the Spanish and French FAD-GF12 studies 34,35 .The Cronbach alpha of the Malay GF12 was 0.89 which indicates a high level of internal consistency supporting the previous ndings of Spanish and Dutch versions 20,36 .For test-retest reliability analysis, the ICC value was more than 0.75 which is considered good and in agreement with the ICC value for the French FAD-GF12 35,37 .Our ndings concurred with similar studies by Georgiadis et al. (0.89) and Zubrick et al. (0.88)   on the English version of GF12 subscale, respectively 38,39 .However, differences between different family members when responding to the questionnaire have been reported previously, therefore limiting the generalizability of our study ndings to other cultures 40 .
Malaysia is a multi-racial country consisting of Malays, Chinese, Indians and indigenous peoples.As the survey had been carried out on a representative sample of the Malay population, the scale may not be suitable for use by other ethnic groups.Further studies are recommended to assess the validity of the Malay GF12 across different cultural settings in the country and also on establishing the cut-off point between "pathological" and "healthy" family functioning in the Malaysian population.

Conclusion
The ndings of this study indicate that 6 (3 positively worded and 3 negatively worded items) of the 12 items of the Malay GF12 have adequate psychometric properties to measure family functioning in the Malaysian context.As family functioning is embedded in the cultural context, further studies should compare groups across different cultural settings to increase the speci city of the index.

Figures Figure 1
Figures

Theoretical
model

Table 1 The
Malay translation items and the corresponding GF12 items.

Table 3
Results from Exploratory Factor Analysis (EFA) of the Malay GF12 items.

Table 4
Factor weights extracted two factors.

Table 5
Final model as per the description in the theoretical model (Fig.3).