Page 102 - IJPS-7-2

International Journal of
Population Studies Knowledge, attitudes, practices, and COVID-19 in Lebanon

(1= Yes and 0= No); multiple-option questions were 2.2.2. Statistical analysis
considered separate variables. The correct answers to the Data were analyzed on SPSS software version 24.
multiple-choice questions were scored 1 point for yes and A descriptive analysis was performed using absolute
0 points for no. The total score was created by summing frequencies and percentages for categorical variables
the correct answers to the multiple choice with those of and means and standard deviations (SD) for quantitative
the binary variables. The total score ranged from 0 to 22, measures. Construct validity of the knowledge, attitude, and
where a higher score indicated higher knowledge about practice scales was evaluated using principal component
COVID-19 (Supplementary File, Table S1). analysis (PCA). This method shares many similarities with

2.2.1.1.2. Attitude scale exploratory factor analysis to determine the efficacy of
the model and the validity of KAP scales. Kaiser–Meyer–
Six questions measured attitudes toward preventive Olkin’s measure of sampling adequacy and Bartlett’s test of
measures, adherence to government actions, and sphericity were calculated to ensure the model’s adequacy.
adaptation toward COVID-19. All are graded on a 3-point Factors with eigenvalues values larger than 1 were retained,
Likert scale from 1 (disagree) to 3 (agree). The total score and the scree plot method was used to determine the
calculated by summing the six responses ranged from 6 number of components to extract (Kanyongo, 2005). Only
to 18. A higher score indicated a more acceptable attitude items with a factor loading >0.4 were considered (Ellis,
toward COVID-19 (Supplementary File, Table S2). 2017). This procedure helped generate the KAP scales.
Cronbach’s alpha was also used to evaluate the internal
2.2.1.1.3. Practice scale
consistency of the studied scales; values of ≥0.9, 0.8 –
Twenty-three items assessed good practice and behavior 0.9, 0.7 – 0.8, 0.6 – 0.7, and <0.6 indicate that reliability
regarding preventive measures against COVID-19. All is excellent, good, good to acceptable, acceptable, and not
were graded on a 5-point Likert scale from the worst acceptable, respectively (Ursachi et al., 2015).
(1 = never) to the best (5 = always). The total practice score
calculated by summing the 23 items ranged from 23 to Bivariate analyses were then conducted. For each
115, with higher scores indicating good practice toward independent variable, medians (interquartile range [IQR])
were used to compare between categories: Due to non-
COVID-19 (Supplementary File, Table S3).
normal distribution of the outcome continuous variables
2.2.1.2. Fear of COVID-19 scale (FCV-19S) (KAP scores), non-parametric tests were used to compare
the ranks of these outcome continuous variables (Kruskal–
The FCV-19S is a 7-item scale that evaluates the fear of
COVID-19 among the general population (Ahorsu et al., Wallis to compare between three groups and Mann–
Whitney to compare between two groups). In addition, the
2020). The Arabic-validated version of the FCV-19S was Spearman test was used to correlate between continuous
used in this study (Alyami et al., 2020). Items are rated variables. Significant results were defined as p <0.05.
on a 5-point Likert scale from 1 (strongly disagree) to 5
(strongly agree). A higher score indicates higher fear of Afterward, since the transformation of the KAP scores
COVID-19 (Ahorsu et al., 2020). did not normalize their distribution, they were dichotomized
into two groups at the median level of each scale, since they
2.2.1.3. Coronavirus Anxiety Scale (CAS) were not normally distributed. Dichotomization was done
The CAS is a 5-item self-report scale rating the as follows. First, the knowledge score was dichotomized
frequency of physiologically-based symptoms caused by into low (values less than 17.00) versus high knowledge
COVID-19-related facts and thoughts (Lee, 2020). The (values greater than or equal to 17.01). Second, the attitude
survey is graded on a 5-point Likert scale from 0 (not at all) score was dichotomized into fearful (values less than 17.00)
to 4 (extremely). The overall score is obtained by summing versus acceptable attitude (values greater than or equal to
the five items, with higher scores indicating higher anxiety 17.01). Third, the practice score was dichotomized into
caused by COVID-19 (Lee, 2020). flawed (values less than 104.00) versus good practice (values
greater than or equal to 104.01). Other categorizations of
2.2.1.4. Translation procedure each scale only altered the results slightly, indicating that
The used scales were translated from English to Arabic using our current classification for three scales is relatively valid.
the forward and backward translation process except for the Four logistic regressions were performed taking the
FCV-19S scale. The translation from English to Arabic was above-mentioned dichotomized dependent variables
done by two authors, and the back-translation was done to conduct multivariable analyses. In the first logistic
by two others. Discrepancies between the original English regression, knowledge was taken as the dependent variable
version and the translated edition is resolved by consensus. and sociodemographic characteristics as the independent

Volume 7 Issue 2 (2021) 96 https://doi.org/10.36922/ijps.v7i2.342