Validation of the ACPQ-M) consists of two parts: translation of the validated English Advance Care Planning Questionnaire (ACPQ) to Malay, and its validation and reliability testing.
Translation of the Advance Care Planning Questionnaire (ACPQ)
The English ACPQ was translated according to the Principles of Good Practice for the Translation and Cultural Adaptation Process (Figure 1) (22). No problems were encountered during the pilot test. Hence, version 4 of the ACPQ-M was used [Table 1].
Validation of Psychometric properties of the Malay Advance Care Planning Questionnaire (ACPQ-M)
Study design and setting
This validation study was conducted from January to June 2018 at an urban primary care clinic and a tertiary education institution in Malaysia.
Community dwellers who were ≥21 years and able to understand Malay were recruited. Community dwellers with mental illnesses such as dementia or psychosis were excluded. Our aim was to recruit community dwelling adults, regardless of their health status as the next phase of our study was to assess the readiness of community dwelling adults for advance care planning.
Sample size calculation
The sample size was calculated based on rule of thumb of 10 participants per item to perform factor analysis (23, 24). The number of items in the ACPQ that could be validated was 16 as the items had a 5-point Likert-scale response (Table 1). Therefore, the total number of participants required was 16*10=160 participants.
The Malay Advance Care Planning Questionnaire (ACPQ-M)
The ACPQ-M consists of four sections and 60 items (29 items were measure on nominal scale, whilst 31 items were measured on a 5-point Likert scale). Participants are required to answer all the items in section A, B and C of the ACPQ-M. As for section D of the ACPQ-M, those who were in in favour of advance care planning were asked to answer the items in the domain “justifications for advance care planning”; while those who were not in favour of advance care planning were asked to answer the items in the domain “justifications for not having advance care planning: avoid thinking about death” and “justifications for not having advance care planning: fate and religion”.
Eligible participants were approached, and the purpose of the study was explained to them using the participant information sheet. Written informed consent was obtained for those who agreed to participate. A researcher administered the ACPQ-M to participants via a face-to-face interview as this questionnaire contained some medical terms which may not be understood by the lay person. Each interview took between 10 to 20 minutes. The retest was performed two weeks later over the phone.
Data analyses were performed using the Statistical Package for Social Sciences (SPSS) version 23.0 (Chicago, Illinois, USA). As normality could not be assumed, the central tendency was described as median and interquartile range (IQR), whilst descriptive data was presented as number and frequency. Confirmatory factor analysis (CFA) was performed using Lavaan package in R software for statistical computing and graphics version 3.5.1 (R Foundation, Vienna, Austria) (25).
Face and content validity were verified by an expert panel (which consisted of two primary care physicians and three academic pharmacists). Flesch reading ease, to test the readability of the instrument was not applied to ACPQ-M because the computer calculated score was not developed and validated for use in Malay language (26).
Factor analysis was used to determine the construct validity. Exploratory factor analysis was performed to explore the dimensionality of the ACPQ-M by computing the percentage of total variance explained, the number of factors and factor loadings to determine the degree of agreement between observed scores and latent variables (27). Bartlett’s test of sphericity was used to test for intercorrelations between all the variables within the correlation matrix (28). Sampling adequacies were determined using the Kaiser-Meyer-Olkin criterion and communalities of the variables. Kaiser-Meyer-Olkin, factor loading values and communalities of at least 0.6, 0.4 and 0.4, respectively, were deemed as having good construct validity (29).
Confirmatory factor analysis was conducted to verify the factor structure of the ACPQ-M (30). The factor structure was examined by computing the model fit indices such as comparative fit index, Tucker Lewis Index; standardised root mean square residual and root mean square error of approximation using diagonally weighted least square method for categorical variables (27). Comparative fit index and Tucker Lewis Index values of at least 0.95 and standardised root mean square residual value of ≤0.09 indicates goodness of fit (27). Additionally, root mean square error of approximation p-value of ≥0.05 also indicates acceptable level of model fit with the degree of fit being interpreted as such: root mean square error of approximation values <0.05 close fit, 0.05 to < 0.08 reasonably good, 0.08 to <0.10 mediocre, and ≥0.10 unacceptable (27, 31).
Cronbach’s α was used to assess the internal consistency of the items in the ACPQ-M. Cronbach’s α values between 0.70–0.90 implied adequate internal consistency (32). Corrected item-total correlations were used to identify items which are inconsistent with other items in the ACPQ-M. Corrected item-total correlation values <0.2 were deemed as unacceptable (32).
Test-retest reliability was assessed using quadratic weighted Cohen’s kappa coefficient as the items for validation were measured as ordinal data on a 5 point Likert-scale (33). Kappa values can range from +1 to -1 (34). The value of +1 indicates complete inter-rater agreement for the categorical items, whereas -1 represents disagreement at the other end of the continuum. Kappa values can be interpreted as follows: <0 less than chance agreement, 0.01–0.20 slight agreement, 0.21–0.40 fair agreement, 0.41–0.60 moderate agreement, 0.61–0.80 substantial agreement and 0.81–1.00 almost perfect agreement (35).