A cross-sectional study was carried out in order to develop and explore the validity and reliability of the questionnaire on knowledge, attitude and practice of drowning prevention and water safety among primary school children’s parents/guardians in Selangor, Malaysia. Ethical approval was obtained from the University Research Ethics Committee (PPI/111/8/JEP-2016-594). With regards to acquiring the consultation and opinions from the panel experts, the professionals were given the following materials for evaluation: an invitation letter, informed consent form, and a folder with the questionnaire. Based on their personal willingness, participants took part in this study to be the respondents by completing the informed consent forms that were distributed to each one of them.

Documents analyses and literature review were conducted to gain a thorough understanding of the views and ideas relevant to the topic, as well as to figure out the items of KAP of drowning prevention and water safety in the instrument. A comprehensive search of existing literature was done through a journal database by searching several keywords related to the research topic, such as “drowning”, “drowning awareness”, “water safety”, “drowning prevention”, “water safety awareness”, “KAP towards drowning” and “Health Belief Model (HBM) on drowning”. The search returned a considerable number of articles (abstract) in the first instance (N=12,810), and the process of extracting relevant articles were subsequently conducted based on several exclusion criteria, namely publication years (only articles published in 2006-2016 were reviewed), languages other than English, and no access or eligibility of full-text articles. After a detailed assessment was performed, eight articles [7,12,21-26] were finally selected as the guide in designing the questionnaire’s construct. Additionally, existing documents related to drowning in Malaysia such as reports, statistical data, guidelines and books were analysed in identifying the factors associated with the issues of water safety and drowning prevention [27-29]. Parents/guardians were selected based on convenience sampling and they were required to share their understanding about the general concepts or specific terminologies related to the topic and how they perceive the questions in terms of sensitivity or difficulty. The informants varied in gender, age, place of residence, and occupation.

The overall information from the literature, documents analyses, and parents/guardians’ opinions were gathered to develop a preliminary framework for characterising main domains and detailing the items of the KAP survey. The assumption of HBM was used as a base for the theoretical framework upon which the instrument was developed. It has been extensively demonstrated that HBM framework is applicable in developing an instrument and intervention messaging in injury prevention [30-32]. The initial draft of the questionnaire comprised of the socio-demographic section with 19 items (children’s swimming activity and swimming ability questions in an open-ended format), knowledge domain with 35 items, attitude domain with 28 items and practice domain with 34 items. The items in the knowledge domain were designed to identify the parents’/guardians’ level of knowledge related to drowning risk, drowning prevention, water safety, first-aid and CPR skills.

Content validity: In the second phase, content validation of the developed questionnaire was done through the process of critical reviews and assessment of CVI. The questionnaire’s draft was critically examined by the supervisory committee to confirm the content of the questionnaire. The supervisory committee consists of academicians with expertise in public health for >10 years. The comments and suggestions from the supervisory committee were applied in designing the questionnaire, culturally appropriate for the target group.

With regards to the content validity, six experts of the professional team comprising of public health lecturers, clinicians, and professional swimming instructors were appointed to review and confirm the content validity of the developed questionnaire. Approaches of CVI i.e., item level (I-CVI) and scale level (S-CVI/ Ave) were employed to validate the content of the tool [33,34]. Each questionnaire was reviewed individually, followed by the experts’ rating on the content based on four characteristics, namely relevance, clarity, simplicity and ambiguity. The Likert’s scale was used with four scoring systems of 1 to 4 (1=not relevant, 2=item needs some revision, 3=relevant but needs minor revision, and 4=very relevant). The content validity of the questionnaire was then determined as a proportion of items, and Scale level (S-CVI/Ave) subsequently described as an average proportion or agreement of the item on the scale scored 3 or 4 by experts. As suggested by a previous study, for panel experts more than 5, the I-CVI less than 0.78 should be revised or deleted [33]. As for the Scale level (S-CVI), the score was considered excellent for value >0.9 [33].

Construct validation: A cross-sectional study was conducted among 130 parents/guardians in one randomly selected primary school in Selangor to examine the construct validity of the questionnaire on KAP of drowning prevention and water safety. An exploratory factor analysis was conducted (for attitude and practice items) using the principal component method and varimax rotation was applied to determine whether the questionnaire was presented in a multidimensional structure or simple structure [35,36]. The recommended sample size for factor analyses was a range of absolute number from 100 to >1000 or a minimum of 3 to 20 times the figure of the variables [36]. A total set of 156 questionnaires were distributed to the primary school children’s parents/guardians with 130 responses (83.3%) asserting their willingness to participate in the study.

As for the knowledge domain, the items were calculated in terms of item difficulty and item discrimination. The item difficulty index presented the percentage of respondents who answer an item correctly. It should be noted that the smaller percentage value of difficulty index indicated a higher level of difficulty [37].

Face validity: Face validity was conducted among fifteen randomly selected parents/guardians (who were not involved in the cross-sectional study) to verify the developed questionnaire materialise in terms of their appropriateness relating to the general formatting, clarity of the language used, readability, and feasibility [38,39]. Based on the assessment and comments obtained from the process of face validation, the items in the questionnaire were either edited, removed or unchanged.

Reliability: The Kuder-Richardson Formula (KR-20) was the formula used to measure internal consistency for a dichotomous choice of the items [40]. Therefore, item reliability in the knowledge domain was explained by KR-20 and other items in attitude and practice domains were determined by Cronbach’s alpha value [20,41,42].

Several analyses approach is involved in this phase. For the face validity and content validity, the analysis of the result was used percentages and CVI. The calculation of CVI was included I-CVI and S-CVI. For the construct validation of the questionnaire, the analysis of numerical used Statistical Packages for the Social Sciences (SPSS, version 22.0) for Exploratory Factor Analysis (EFA), internal consistency (Cronbach Alpha) and the relationship of domains (Pearson’s correlation). Collected data were managed, checked and cleaned with the removal of any subjects with greater than 10% of the necessary data elements missing or any impossible values.

Application of the Health Belief Model (HBM) to the construction items of the questionnaire on drowning prevention and water safety context are presented in [Table/Fig-1]. The exploration of parents/guardians knowledge and activities in the context of drowning risk, drowning prevention strategies, drowning awareness, water safety skill, competencies, and barriers were utilising the conceptualisation of the HBM.

The draft of the instrument-KAP questionnaire on drowning prevention and water safety was structured in five sections; socio-demographic section, children’s swimming activity and ability section and three domains; knowledge, attitude and practice as detailed in [Table/Fig-2]. The items in knowledge domain have three selection of answers; “Yes”, “No” and “Not Sure” which correct answer received one point, incorrect and “Not Sure” answers received zero points. The possible scores for knowledge domain ranged from 0 to 35. Attitude domain consists of 28 items using five points Likert’s scale. The scoring systems of “Strongly Agree”, “Agree”, “Not Sure”, “Disagree” and “Strongly Disagree” with numerical scores 5, 4, 3, 2 and 1 will be given to each answer respectively which respondent can indicate their degree of agreement towards the statement given. The possible scores for attitude domain ranged from 28 to 140, and this will determine the respondent’s opinion and belief about drowning risk, drowning prevention, awareness and socio-cultural perspectives on water safety. Thirty-five items of the practice domain were hoped to correspond with the respondent’s practice towards drowning prevention and water safety such as action is taken to avoid drowning occurrence, especially in risky places. This domain assessed using a four-point Likert’s scale of “Always”, “Sometimes”, “Often” and “Never” category with numerical scores 4, 3, 2 and 1 will be given to each answer respectively, and possible scores ranged from 35-140.

Content validity: The item CVI (I-CVI) was computed for all items in each domain (knowledge, attitude, practice), respectively using the four parameters; relevance, clarity, simplicity and unambiguity. The agreement of the panel experts on the adequacy of the instrument was obtained; thus items were revised or considered for changes based on the I-CVI score (<0.83). The scale-level content validity (S-CVI/Ave) are shown in [Table/Fig-3]. The characteristics of the content validity were showed high and accepted in the knowledge domain with >90% of relevancy, clarity, simplicity and unambiguity of each item in the questionnaire [41]. However, in the attitude domain, the S-CVI for ambiguity criteria showed a lower score and in the practice domain, a low S-CVI in criteria for relevancy and clarity. Thus the items in the respective domain either were revised or deleted based on the I-CVI score.

Construct validity (Factor analysis): A total of 130 (83.3% response rate) parents/guardians of primary school children participated in the study. As shown in [Table/Fig-4], the result obtained responses predominantly from Malay (96.2%), female caregivers (54.6%) with age ranged from 21 to 56 years with mean age was 36.4 (±6.657).

Construct validation were conducted for attitude and practice domain. Factor analysis resulted with Kaiser-Meyer-Olkin (KMO) test was 0.771, and Bartlett’s test of sphericity was significant (χ²=1248.72, df=378, p-value <0.001), indicated the adequacy of sample size and items in attitude domain that met the factor analysis requirement. For items in the practice domain, the KMO test was 0.773, and Bartlett’s test of sphericity was significant (χ²=2257.08, df=595, p-value <0.001), also showed an appropriate data for factor analysis. From the factor analysis, 10-factor solutions explained by 69.44% total variance obtained from total 28 items in attitude domain and 9-factor solutions with 65.75% explained total variance from total 35 items in practice domain.

Item analysis was conducted on the knowledge domain, and the items difficulty and items discrimination indexes were calculated. The items difficulty indexes were ranged from 0.02 to 1.00 and items less than 0.2 were revised as it is considered to be difficult [37]. Four items were calculated to have an index less than 0.2, therefore they were reviewed accordingly. Eleven items were having an index of more than 0.9 and were considered very easy items and still be revised in view of their importance and relatedness to the study. Twenty questions in a range of 0.2 to 0.9 were maintained. The items discrimination was ranged from 0 to 1. Item discrimination index with value of more than 0.2 were maintained and items below the value were revised. Twelve items were found to have a low discrimination index and were revised. The other 23 items were within range of 0.25-1.00 indicated an ordinarily regarded satisfactory for use [37]. The items knowledge domain was considered to have an optimal level of difficulty and able to discriminate performance of good and poor knowledge of the respondent.

Reliability: KR-20 reliability coefficient score was 0.58 for knowledge and reliability coefficients (Cronbach’s alpha)’s scores were 0.81 and 0.84 for attitude and practice respectively.

Face validity: From 15 parents/guardians participated in rating the instrument, 73.3% agreed that the title and objectives of the questionnaire on the front page were easy to understand. About 86.7% of respondents also agreed on the font writing style, the font size and spacing of the text. Majority of the respondents also indicated that the structure and formatting of the questionnaire are appropriate and suitable (83.3%) and easily understandable and relevant (80%).

The mean scores in knowledge, attitude and practice domains are presented in [Table/Fig-5]. The correlation of questionnaire domains showed a positive relationship, moderate in strength and statistically significant [Table/Fig-6].

The final questionnaire consisted of 111 items; with 22 items of respondent’s socio-demographic, 32 items in knowledge domain (3 items were edited and shifted to the respondent’s background section), 25 items in attitude domain and 32 items in practice domain (with 3 items were deleted in each domain respectively) due to the low CVI score. The children’s swimming activity and swimming ability items is made as an open-ended format.

The development of instrument utilised face and content validity processes to establish construct validity in ensuring that the instrument is psychometrically sound. The final questionnaire of knowledge, attitude and practice concerning drowning prevention and water safety has demonstrated adequate construct validity and good internal reliability. The knowledge items were evaluated using item analysis and the present study has found that the knowledge items had the appropriate levels of difficulty and discrimination index as suggested in a previous study [37]. As such, four items displayed a low discrimination index but were subsequently revised, retained, and still considered vital to measure the knowledge of drowning prevention and water safety. This finding is similar to that of other researches that examined the development of instruments; the item analysis was essentially useful to guide the evaluation of each item and reflect only the items’ internal consistency and not the items’ validity [41,43]. Furthermore, it was found that 12 items displayed low difficulty index, implying that the parents/guardians were lacking in knowledge related to drowning prevention and water safety. As suggested by earlier researcher, these items were also revised and retained in the questionnaire in order to identify the specific topics that justify the need for an education programme that can be targeted to the parents/ guardians [41].

In terms of the content validity, the agreement of the judges on the adequacy of the instrument obtained an average I-CVI value of 0.95, showing a statistically excellent content validity as it has I-CVIs of higher than 0.83 as suggested in a previous study [33]. In comparison with other researchers that performed content validity using CVI on instrument development, present study showed that the items in the questionnaire had high scales of simplicity, relevance, clarity, and unambiguity, comprehensive and representative [14,19,35]. In addition, findings from face validity supported the content validity of the questionnaire and indicate its readability and feasibility. Present study showed similar findings as the previous study on items validity agreement of at least 75% positive responses [43].

With regards to the reliability determination, the value of Cronbach’s alpha for attitude and practice domains showed internal consistency of the items in the questionnaire which, therefore, confirmed the adequacy of these scales’ of internal consistencies. Nevertheless, the knowledge domain resulted in a low value (KR-20=0.579) which signalled a need for improvement [44]. A single approach to examine the content validity was suggested to be insufficient; thus, a construct validity approach had been tested for the multiple items in the instrument to tackle this issue [34]. In the present study, Bartlett’s test of sphericity was found significant, and the KMO test demonstrated the sampling adequacy. The result of exploratory factor analysis displayed that the factor loading for all factors in attitude and practice domains were statistically significant and higher than 0.4 as suggested in most literature [44,45]. Studies have suggested that the value of loading factors provided an overview of how much the variables contributed to the factor and showed a clear association between the dimensions of attitude and practice respectively [46,47].

In comparison to other studies on instrument development and validation, the present questionnaire was developed to measure knowledge, attitude and practice of drowning prevention and water safety. It has successfully demonstrated good reliability, acceptable in readability and feasibility and high content validity [9, 41, 47, 48]. Therefore, present study managed to develop a valid questionnaire and it is ready to measure KAP drowning prevention and water safety for primary school children’s parents/guardians of Malaysia.

The limitation of the study was small sample size. Further studies with larger and more diverse samples need to be conducted for scale standardisation in various settings.

Discovering the KAP of drowning prevention and water safety among parents and guardians offers a broad overview of the current awareness pertaining to drowning issues in malaysia. The present study allows the exploration and a clearer understanding of the parents’/guardians’ existing knowledge, skills and practices on drowning prevention and water safety. This method can serve as a suitable format to evaluate public health intervention related to the implementation of water safety recommendations and drowning prevention measures. In a nutshell, the initial assessment of KAP will lead to better planning of health programmes that are designed to raise the awareness about water safety as well as to impart the necessary drowning prevention education and skills among parents and guardians.

In conclusion, the survey instrument employed in this study in examining knowledge, attitude and practice should be useful in screening parents and guardians to determine their understanding and willingness to improve their awareness of drowning prevention and water safety. The psychometrics of the instrument was empirically reliable for internal consistency in addition to the evidence of strong reliability estimates on the three scales across a representative and diverse sample.