Adaptation and validation of a questionnaire measuring knowledge, awareness and practice regarding familial hypercholesterolaemia among primary care physicians in Malaysia

Background: Primary care physicians (PCP) play an important role in early detection of Familial Hypercholesterolaemia (FH). However, knowledge, awareness and practice (KAP) regarding FH among Malaysian PCP are not well established, and there was no validated tool to assess their FH KAP. Therefore, this study aimed to adapt an FH KAP questionnaire and to determine its validity and reliability among Malaysian PCP. Methods: This cross-sectional validation study involved PCP with ≥1-year working experience in the Malaysian primary care settings. In Phase 1, the original 19-item FH KAP questionnaire was content validated and adapted by 7 experts. The questionnaire was then converted into an online survey instrument and was face validated by 10 PCP. In Phase 2, the adapted questionnaire was distributed via e-mail to 1500 PCP for self-administration. Data were collected on their KAP, demography, qualification and work experience. The construct validity was tested using known-groups validation method. PCP with postgraduate qualification (PCP-PG-Qual) were hypothesized to have better FH KAP compared with PCP without postgraduate qualification (PCP-noPG-Qual). The internal consistency reliability was tested using Kuder Richardson formula-20 (KR-20) and test-retest reliability was tested on 26 PCP using kappa statistics. Results: During content validation and adaptation, 10 items remained unchanged, 8 items were modified, 1 item was moved to demography and 7 items were added. The adapted questionnaire contained 25 items (11 knowledge, 5 awareness and 9 practice items). A total of 130 out of 1500 PCP (response rate: 8.7%) completed the questionnaire. The mean percentage knowledge score was significantly higher in PCP-PG-Qual compared to PCP-noPG-Qual (53.5, SD±13.9 vs. 35.9, SD±11.79), t(128)=6.90, p<0.001. The median percentage awareness score was significantly higher in PCP-PG-Qual compared to PCP-noPG-Qual (15.4, IqR±23.08 vs. 7.7, IqR±15.38),

Conclusion: The robust validation methods and findings suggest that the adapted 25-item questionnaire is a valid and reliable tool to measure FH KAP among Malaysian PCP. This would aid to improve FH care in the community.

Background
Familial Hypercholesterolaemia (FH) is a genetic disorder characterized by severely elevated low density lipoprotein cholesterol (LDL-c) that leads to atherosclerosis, resulting in an increased risk for premature coronary artery disease (CAD) [1,2]. It is one of the most common forms of inherited conditions with an autosomal mode of inheritance [2].
Mutations in several genes such as LDLR, APOB100 and PCSK9 have been strongly linked to FH [2]. Clinically, this condition presents in the form of either heterozygous FH (HeFH) or homozygous FH (HoFH). HeFH is more common with an estimated 70-90% of FH cases resulting from heterozygous pathogenic variants [2,3]. HeFH accounts for 2-3% of CAD in individuals below 60 years of age [1,3]. In contrast, most individuals with HoFH experience severe CAD by their mid-20s and the rate of either death or coronary bypass surgery by the teenage years is very high [1,3]. Early detection and treatment of FH through cholesterol-lowering therapies can effectively prevent premature CAD [4].
Globally, the prevalence of HeFH is estimated to be 1 in 200 to 1 in 500 in various populations [5][6][7]. In Malaysia, the prevalence of clinically diagnosed FH has been reported at 1 in 100, which is one of the highest in the world [8]. Therefore, with a population of 32 million, it is estimated that 320,000 individuals may be affected by HeFH [8]. However, like in most countries, the majority of these cases are still undiagnosed, resulting in lost opportunities to prevent premature CAD [9,10]. This has undoubtedly contributed to the high prevalence of premature CAD among Malaysians which accounted for 10-15% of acute coronary syndrome (ACS) [11]. A recent national report found that the mean age of individuals with ACS at admission in Malaysia was 58.6 years old, of which 23.8% were under the age of 50 years [12]. This is younger compared with our Asian counterparts in neighbouring countries [12,13].
Improving identification of FH, particularly in primary care, enables early treatment of these individuals which is crucial to reduce their risk of premature CAD [14]. In Malaysia, primary care physicians (PCP) are well positioned in the front line of primary care service which is delivered by the public and private sectors [15]. They are already managing common cardiovascular risk factors such as diabetes, hypertension and hypercholesterolaemia [16]. However, only a handful of PCP in Malaysia hold formal postgraduate (PG) qualification in primary care, while the majority do not [15,16]. The situation in Malaysia is similar to many other developing countries where doctors without primary care qualification are allowed to practice as PCP [17]. This is in contrast to some developed countries such as the United Kingdom (UK) and Australia where PCP are required to have mandatory PG training and qualification in primary care.
Numerous studies have shown that there were gaps in knowledge, awareness and practice (KAP) regarding FH among PCP in various parts of the world [18][19][20][21][22][23][24][25], especially in developing countries [25]. Pang et al. recently assessed FH KAP among PCP in several Asia-Pacific countries in the 'Ten Countries Study' [25]. Their KAP have been found to be suboptimal where less than half of the PCP were aware of FH clinical guidelines; and their knowledge of prevalence, inheritability, diagnostic criteria and CAD risk of FH were also found to be low [25]. Addressing these gaps is essential for effective implementation strategies to improve management of FH among PCP [26].
In the FH 'Ten Countries Study', the 19-item FH KAP questionnaire designed by Bell et al. in 2014 [18] was utilized to assess the FH KP among PCP [25]. This questionnaire was initially developed with the aim to determine the KAP regarding FH among PCP in Western Australia [18]. Since

Methods
This study was undertaken under the aegis of the FH 'Ten Countries Study' which investigated several pertinent areas of FH care in the Asia-Pacific Region [25]. It was conducted in two parts. The first part involved adaptation and validation of the FH KAP questionnaire; and the second part was determination of the FH KAP among Malaysian PCP using the adapted and validated questionnaire. This paper presents the detailed methods and findings of the first part of the study. The second part of the study and its results was already published in 2018 [27]. The sampling for the first and second parts was mutually exclusive i.e. participants who were recruited in the first part of the study were excluded from the second part.

Study design and participants
This was a cross-sectional questionnaire validation study conducted in two phases. Phase 1 was the content validation, adaptation and face validation of the FH KAP questionnaire. Phase 2 was the field testing and psychometric evaluation.
This study was conducted between January 2016 and January 2017 among a group of PCP in Malaysia consisted of Medical Officers (MO), General Practitioners (GP) and Family Medicine Specialists (FMS). Those registered with Malaysian Medical Council and have one or more years of working experience in the Malaysian primary care settings were included.
Those who were working as locum doctors in primary care clinics or those who were from medical specialties other than family medicine were excluded. In Phase 2, the psychometric evaluation involved construct validity and reliability testing.
Construct validity was conducted by testing differences between two groups with expected differences to establish known-groups validity [28]. PCP-PG-Qual was hypothesized to have better KAP regarding FH compared to PCP-noPG-Qual. Reliability testing involved internal consistency analysis using Kuder Richardson formula-20 (KR-20) reliability coefficient [29]; and test-retest reliability using Cohen's kappa statistics to measure the stability of the responses to the questionnaire over time [30]. Figure 1 outlines the two phases of the adaptation and validation processes.

Study instrument
The FH KAP questionnaire used in this study was originally designed by Bell et al. [18].
Written permission to adapt and validate the questionnaire was obtained from the questionnaire developer via e-mail prior to the conduct of the study. The questionnaire was developed in the English language and consisted of 19 items divided into three domains i.e. knowledge, awareness and practice. The questionnaire items and responses were presented in various forms which included 7-point Likert scale, single best answer, multiple answer, 'Yes/ No/ Don't know' and free text answer. Most of the questions have predetermined correct answers.
There were seven items in the knowledge domain covering the following areas: i) description of FH; ii) identification of lipid profile in FH; iii) prevalence of FH in Australia; iv) inheritance of FH in first-degree relatives; v) CAD risk in untreated FH; vi) age threshold for premature CAD; vii) role of genetic testing in FH. The awareness domain consisted of three items covering the following areas; i) familiarity with FH; ii) Australian clinical guideline on FH; iii) lipid specialist service. There were nine items in the practice domain covering the following areas; i) assistance in FH detection; ii) number of FH cases under care; iii) screening of relatives in FH cases; iv) family screening of FH among premature CAD; v) preference on effective healthcare provider in FH detection; vi) age for FH screening among young individuals in a family with premature CAD; vii) referral of FH patients to lipid specialist; viii) pharmacological agents used in hypercholesterolaemia; ix) combination of pharmacological agents used in severe hypercholesterolaemia. Conversion of the questionnaire into an online survey instrument The adapted paper-based FH KAP questionnaire was then converted into an online survey questionnaire using the Google® Forms [31]. Relevant information about the study was given on the first page of the online survey and also in the email containing the link to the survey. The information given was the same as the paper-based information sheet, containing the identity of the researchers, contact details, the reason for conducting the survey and how the data would be used. The study information sheet also highlighted the right of the participants to anonymously and voluntarily participate in the study and the right to withdraw from the study at any time without any reason and without penalty.
Informed consent was obtained via the online questionnaire when the respondent had to click the 'Consent' button before the next page of the questionnaire could be accessed.
The questionnaire was divided into three sections. Section A contained questions on the inclusion and exclusion criteria. Participants could only proceed to Section B and C, if they fulfil the eligibility criteria. Section B contained questions covering the KAP domains.
Section C included questions on demographic characteristic, qualification and work experience. Participants could omit to answer any of the items as none of them required an obligatory response (other than those relating to the consent) before they could proceed until the end of the questionnaire.

Face validation
Face validation of the adapted FH KAP questionnaire was conducted among 10 PCP who fulfilled the inclusion and exclusion criteria, and who were naive to the study. The 10 PCP were given both the paper-based and online versions of the questionnaire. The questionnaire was self-administered, and participants were requested to take note of the time taken to answer the questionnaire, clarity of the content, language and wording used and the general structure of the questionnaire. Their opinions on understanding the instructions, contents, wording and general structure of the questionnaire were assessed and recorded. The result was discussed among the panel of experts. Minor correction and fine tuning of the questionnaire were addressed according to their comments and suggestions.

Phase 2: Field testing and psychometric evaluation
The adapted FH KAP online questionnaire was field tested amongst PCP who fulfilled the same inclusion criteria as in Phase 1. However, PCP who participated in Phase 1 and 2 were mutually exclusive, as those who participated in Phase 1 were not recruited for Phase 2.

Sample size
Two sample sizes were calculated for this study, one for the known-groups validity and the other for the reliability testing. The known-groups validity compared mean or median percentage score of FH KAP between PCP-PG-Qual and PCP-noPG-Qual. Therefore, the sample size for each group was calculated using the OpenEpi software for comparison of two means formula [32]. As there was no previous study comparing KAP between PCP-PG-Qual and PCP-noPG-Qual related to FH, the calculation was therefore made based on the previous study by Mosli et al., which compared KAP regarding colorectal cancer screening among two groups of PCP in Saudi Arabia [33]. Family Medicine trained physicians had higher mean knowledge score compared to physicians with MBBS only (4.93 ± 2.29 vs. 3.23 ± 1.88, P < 0.01) [33]. Based on the difference between the two means in this study, 95% Confidence Interval (2 sided), 80% power and a ratio of 1:1 between groups, the minimum sample size required for each group was 24 participants.
The sample size for the internal consistency reliability analysis using KR-20 was calculated based on the subject to item ratio for which a subject to item ratio of 5:1 was used [29]. Therefore, a minimum of 125 participants were needed (25 items×5 = 125).
Considering estimated response rate of 10-30% for an online survey [34], the questionnaire was planned to be distributed to at least 1500 participants. Regarding testretest reliability, sample size for testing the Cohen's kappa agreement was determined to be 26, which was 20% of the total number of participants [35].

Sampling method
A link to access the adapted FH KAP online questionnaire was sent via email to 1500 PCP in the e-mail lists of two major professional bodies for PCP in Malaysia. The email contained information on the study background, purpose and benefits, participation in the study, study procedure, confidentiality as well as informed consent. They were invited to open a link to the online questionnaire.
The questionnaire consisted of three sections as previously described. Those who consented to the study and fulfilled the study inclusion and exclusion criteria in Section A were able to proceed to Section B and C of the online form. Once the questionnaire was completed and submitted, no modification was allowed. The participants had the right to withdraw from the study at any time without penalty. To avoid repeated response from the same participant, the questionnaire contained an item asking whether they have answered the questionnaire previously. For test-retest reliability testing, those who have responded were contacted via e-mail to obtain their second response after two weeks of their first response.

Questionnaire interpretation, coding and scoring
For item no. 1 which assessed FH familiarity, Likert scale scores of 1 to 4 were interpreted as 'unfamiliar' and coded as '0' while scores of 5 to 7 were interpreted as 'familiar' and coded as '1'. The interpretation was done in accordance to the study by Rangarajan et al.
[23] and Pang et al. [25]. For all of the items that have predetermined correct answers, incorrect response was coded as '0' while correct response was coded as '1'. The

Construct validation
The construct validity of the adapted FH KAP questionnaire was conducted using knowngroups validation method in view of the dichotomous nature of the questionnaire responses rather than numerical or continuous form [28]. The questionnaire would be considered valid if it is able to significantly discriminate across groups of subjects that have been predicted or 'known' to differ from each other [28]. In this study, PCP-PG-Qual was hypothesized to have better KAP regarding FH compared to PCP-noPG-Qual. Therefore, the FH KAP questionnaire would be considered valid if it could significantly discriminate the mean or median FH KAP scores between the two groups.

Reliability testing
With regards to the internal consistency reliability testing of the adapted FH KAP questionnaire, KR-20 reliability coefficient which is a special form of Cronbach's alpha was carried out in view of the dichotomous nature of the questionnaire's responses [29]. KR-20 reliability coefficient of < 0.50 was interpreted as low, 0.50 -0.80 was considered moderate and > 0.80 was interpreted as high [29].
For the test-retest reliability, 26 participants (20%) were requested to answer the questionnaire again after two weeks interval. Cohen's kappa statistics, which is a robust statistical method, was used for the reliability testing because of the dichotomous nature of the questionnaire responses [30]. The kappa result was interpreted as follows: values ≤ 0 indicated no agreement, 0.01 -0.20 as none to slight, 0.21 -0.40 as fair, 0.41 -0.60 as moderate, 0.61 -0.80 as substantial, and 0.81 -1.00 as almost perfect agreement [30].

Statistical analysis
Data were analysed using the SPSS software version 24.0 (SPSS Inc., Chicago, IL, USA).
Missing data were treated using discrete value. Descriptive statistics were performed to depict the demographic background and practices information of the participants. Data were reported and presented as frequency and percentage. Before conducting the knowngroups validity testing of FH KAP between PCP-PQ-Qual and PCP-noPG-Qual, normality of data distribution and equality of variance were examined. For the normally distributed data, independent t-test was applied to compare the mean percentage score of FH KAP between the 2 PCP groups, and for the non-normally distributed data, Mann Whitney u-test was applied. The statistically significance difference between the two groups was reported using P-value of < 0.05. For the internal consistency reliability testing, the KR-20 coefficient [36] for each KAP domain and the overall KR-20 coefficient were calculated.
For the test-retest reliability testing using Cohen's kappa statistics [30], the kappa value of each KAP item and the average kappa value were calculated.

Ethical considerations
The ethical approval for this study was obtained from the Research Ethics Committee (REC), Universiti Teknologi MARA (600-IRMI (5/1/6). The designing of the online survey questionnaire using the Google® Forms complied with the British Psychological Society Ethics Guidelines for Internet-Mediated Research, 2013 [37]. The study information provided on the first page of the questionnaire highlighted the right of the participants to withdraw from the study at any time without any reason and without penalty. Informed consent was obtained via the online questionnaire when the respondent had to click the 'Consent' button before the next page of the questionnaire could be accessed.
Participants could omit to answer any of the items as there was no item (other than those relating to the consent) required an obligatory response before they could proceed until the end of the questionnaire. The survey was made anonymous by switching the option to collect computer IP addresses to 'No'. To ensure confidentiality, the password to the Google® Forms account was only known to the researcher and data were not stored within a shared account.

Content validation, adaptation and face validation
During the content validation and adaptation process, a consensus decision was made by the panel of experts whereby 10 items remained unchanged, one item needed to be moved to the demography section, eight items were modified or rephrased to suit the

Construct validity using known-groups validation
The data for knowledge and practice score were normally distributed hence independent ttest was used to compare the mean percentage scores, and the data for awareness score were not normally distributed thus Mann Whitney u-test was applied to compare the median percentage scores.

PLEASE INSERT TABLE 3 HERE
Regarding awareness, the median percentage score was significantly higher in PCP-PG-Qual (15.4, IqR ± 23.08) compared to PCP-noPG-Qual (7.7, IqR ± 15.38), p = 0.030. The difference in the median percentage score in awareness regarding FH between the two groups was 7.69. The results are shown in Table 4.

Internal consistency reliability using Kuder
Richardson formula-20 The KR-20 internal consistency reliability coefficient of the FH KAP questionnaire was 0.53, 0.76 and 0.61 for knowledge, awareness and practice domains, respectively. The overall KR-20 coefficient for the FH KAP questionnaire was 0.79 which indicated moderate reliability [29]. The internal consistency reliability results of the FH KAP questionnaire are shown in Table 5.

PLEASE INSERT TABLE 5 HERE
Test-retest reliability using Cohen's kappa statistics Out of 130 participants, 26 (20%) participants completed the questionnaire again at two weeks interval. The kappa values were 0.818, 0.810 and 0.760 for knowledge, awareness and practice domains, respectively. The average kappa value for the FH KAP questionnaire was 0.796, which indicated substantial agreement [30]. The test-retest reliability results of the FH KAP questionnaire are shown in Table 6.

PLEASE INSERT TABLE 6 HERE
This study has produced a valid and reliable 25-item FH KAP questionnaire, and the final version is supplied in Supplementary File 3.

Discussion
To the best of our knowledge, this is the first study in Malaysia which has adapted a questionnaire to assess KAP among PCP to suit the local primary care setting. Malaysian Overall, 10 items remained unchanged, 1 item was moved to demography, 8 items were modified or rephrased and 7 new items were added during the content validation and adaptation process. With regards to the knowledge domain, 3 items remained unchanged, 1 item was modified, 3 were rephrased for clarity and 4 items were added, giving a total of 11 items. The item assessing knowledge on prevalence of FH in Australia was found to be irrelevant to our PCP population and was therefore modified to global prevalence of FH. Regarding the practice domain, 5 items remained unchanged, 1 was moved to demography, 3 items were rephrased for clarity and 1 item was added, giving a total of 9 items. The item assessing the number of FH patients under care was moved to demography, as it was perceived to fit better in the demographic section. An item assessing practice in CAD risk stratification among FH patients was added. This is crucial as FH is known to carry a high mortality risk from CAD thus it is pivotal to assess whether PCP would still risk stratify patients with FH. Risk stratification in patients with FH is not recommended as they should be classified as high risk irrespective of other risk factors [47].
With regards to construct validity, known-groups validation was conducted in view of the dichotomous nature of the responses in the FH KAP questionnaire [28]. Exploratory factor analysis which is commonly used for numerical or continuous responses items is not recommended to assess construct validity of a questionnaire which is dichotomously scored [28]. In this study, PCP-PG-Qual had significantly higher mean percentage scores for knowledge and practice compared with PCP-noPG-Qual. Similar trend was also found with regards to the median percentage score for awareness. This is in keeping with the hypothesis of this study where PCP-PG-Qual was expected to perform better in all KAP domains. This finding indicated that the adapted 25-item FH KAP questionnaire was a valid tool to be used for determination of FH KAP among PCP in Malaysia. Furthermore, the second part of our study which had used the 25-item adapted and validated questionnaire to determine the FH KAP among Malaysian PCP showed substantial gaps in FH-KAP among Malaysian PCP, with PCP-PG-Qual having better knowledge, awareness and practice than PCP-noPG-Qual [27].
Regarding internal consistency, the KR-20 reliability coefficient which is an alternative to Cronbach's alpha was carried out in this study because of the dichotomous nature of the questionnaire responses [29]. The KR-20 internal consistency coefficients for all the three Regarding test-retest reliability analysis, Cohen's kappa coefficient was appropriately used instead of intra-class correlation coefficient (ICC) in view of the dichotomous nature of the questionnaire responses [30]. Cohen's kappa statistic is a form of reliability coefficient to determine the degree of agreement between two different evaluations from dichotomous variables [30]. In this study, the kappa values were almost perfect for the knowledge (0.818) and awareness (0.810) domains while the practice domain showed substantial agreement (0.760). The average kappa coefficient of the FH KAP questionnaire was 0.796, indicating substantial agreement [30]. Therefore, this questionnaire was considered reliable and stable over time to be used among Malaysian PCP. In comparison to the study by Batais et al.,, they reported an average kappa of 0.85 [24], which was comparable to our study.

Strength and limitation
The strength of this study included the robust adaptation and validation methods used to determine the validity and reliability of this questionnaire among the Malaysian PCP. The adapted and validated 25-item questionnaire is more comprehensive as it contains areas of FH care not included in the original 19-item questionnaire. These include pertinent areas such as assessing awareness of FH diagnostic criteria and assessing practice on CAD risk stratification for FH patients.
There were several inevitable limitations in this study which include the low response rate, which is expected for an online questionnaire disseminated through emails [34].
However, the sample size for psychometric evaluation was adequate to ensure validity and reliability of the FH KAP questionnaire in this study. Another limitation was the possibility of response bias in this study. The self-selected group that responded to the questionnaire may reflect those with more interest and knowledge in lipid disorders. The use of online questionnaire was also vulnerable to response bias in which the PCP could get the information easily from the internet. It is well recognised that a more representative and unbiased sampling method would be to conduct a multistage probability sampling which would give all the PCP in Malaysia an equal chance to be selected.

Implications for future research and clinical practice
This study has produced a valid and reliable paper-based as well as online version of the questionnaire which can be used to determine the KAP regarding FH among PCP in Malaysia. However, it is recommended that researchers should at least perform the KR-20 internal consistency analysis to ensure that the questionnaire is reliable in their  [37]. Information about the study and participant's right to withdraw from the study at any time was provided in the English language on the front page of the paper-based and online versions of the questionnaire.
Informed consent was obtained when participants clicked on the consent button in order to proceed to the various sections of the questionnaire. The data collected was kept anonymous as IP address was not collected. Confidentiality was ensured by keeping a single account protected by a password which was only known to the researcher and data was not stored within a shared Google® Forms account.

Consent for publication
Written permission was obtained from the original developer of the FH KAP questionnaire (GFW) to adapt and validate this questionnaire in Malaysian PCP. Permission was also obtained to publish the finding. Participants' consent for publication is not applicable as participants' individual data was neither provided nor presented in the manuscript. Each author has participated sufficiently in the work to take public responsibility for appropriate portions of the content as described above. All authors agreed to be accountable for all aspects of the work to ensure that questions related to the accuracy or integrity of any part of the work would be appropriately investigated and resolved.        Figure 1 Flow chart of the conduct of the study