Knowledge, attitudes and practices towards COVID-19 in Saudi Arabia: Planning implications for public health pandemics

Background: Saudi Arabian authorities have implemented a number of preventive measures to confront COVID-19, including complete public lockdown, limitations on many services and a public awareness campaign. The success of these preventive measures is highly dependent on individuals’ compliance, which is influenced by their knowledge, attitudes and practices towards this disease. This study aims to assess the knowledge, attitudes and practices towards COVID-19 among the population of Saudi Arabia.MethodsThis is a cross-sectional design study conducted between May 4 and May 21, 2020, using an online survey of 1135 participants. The survey instrument consisted of demographic characteristics, 19 items on knowledge, 4 items on attitudes and 6 items on practices. Descriptive statistics, independent sample t -test, analysis of variance, Mann–Whitney U test, Kruskal–Wallis test, standard multiple regression analysis and ordinal logistic regression analysis were conducted. ResultsThe majority of the study participants were knowledgeable about COVID-19, with an overall correct rate of 77%. Most participants were worried of contracting COVID-19 (96.3%), and they held positive attitudes towards the intent to use a vaccine for COVID-19 if it was available (81%), informing health authorities if they developed symptoms of COVID-19 (98.1%) and confidence in the government efforts during the COVID-19 pandemic (90%). Most participants were also taking precautions such as keeping social distancing (99.2%), avoiding leaving the house (98.6%) and wearing face masks in public (92.2%). Nearly all of the respondents stated that they were avoiding shaking hands (96.8%), avoiding touching their face with unwashed hands (96.7%) and washing/rubbing hands as recommended (96.6%). Male participants with higher education level and high income who had attended a health education activity were more likely to have more COVID-19 knowledge.ConclusionsThe findings of this research suggest that the Saudi population are generally knowledgeable and have good attitudes and practices regarding COVID-19. Health authorities in Saudi Arabia may benefit from the findings of this research, as the results could be used to improve planning for COVID-19-related efforts and to plan for future health crises. Such planning will help local authorities react promptly to prevent any health threats to the Saudi population and to those who visit Islamic holy places in Saudi Arabia.

To successfully control COVID-19, people must adhere to the public health preventive measures. The adherence behavior of a population is largely affected by their knowledge, attitudes and practices (KAP) towards a disease [13]. According to KAP theory, "the changes of human behavior are divided into three successive processes: the acquisition of knowledge, the generation of attitudes and the formation of behavior. … Knowledge is the foundation of behavior change, and belief and attitudes are the driving force of behavior change." [14]. Studying the population KAP towards COVID-19 will help in understanding the current situation and in directing resources and measures of the health sector to the right areas. Lessons learned from this pandemic may improve the planning for and management of future public health crises.
This study aims to assess the KAP towards COVID-19 among the population of Saudi Arabia and to re ect on the efforts implemented by authorities to manage the crisis.

Design and sample
This study applied a cross-sectional design to assess the KAP of the Saudi Arabian population towards COVID-19. An online survey was conducted from May 4 to May 21, 2020. All citizens and residents of Saudi Arabia who lived in the country during the COVID-19 pandemic and who were aged 18 years or older were eligible to participate in the study.
The protocol of this study was approved by the Research Ethics Committee of Jazan University, Saudi Arabia (REC41/5/095). Participants were provided with all required information about the voluntary nature of their participation in the study, con dentiality measures applied to protect the collected data, and instructions on how to complete the online survey. Although the participation was anonymous, an informed consent of willingness to participate voluntarily was obtained through a yes-no question. Based on the answer, participants were directed to complete the survey or to stop at that point.

Questionnaire
This questionnaire was developed by the author and consisted of two parts: demographics and KAP. Demographic variables included gender, age, marital status, education, occupation, income per month, place of residence, nationality, source of information and whether the participant had attended a health education activity. The KAP questions were developed based on information published by the WHO, the Ministry of Health (MOH) in Saudi Arabia, the Saudi Centre for Disease Prevention and Control, and the United States Centers for Disease Control and Prevention (CDC). The questionnaire had 29 items: 19 for knowledge (Table 2), 4 for attitudes (Table 5), and 6 for practices (Table 8). Knowledge questions 1-9 were answered with 'yes', 'no' or 'not sure'. Knowledge questions 10-19 were answered by choosing the appropriate answer/answers. Each correct answer carried 1 point and incorrect or 'I do not know' answer carried 0 points. This gave a score range of 0-40, with a higher score indicating a better knowledge of COVID-19. Attitudes towards COVID-19 were measured through four questions using an ordinal ve-point Likert scale. The possible answers were 'strongly agree', 'agree', 'not sure', 'disagree' and 'strongly disagree'. The practices of respondents regarding COVID-19 were assessed using six yes-no questions concerning their behaviours during the pandemic (KAP Questionnaire of COVID-19 in Additional le 1).
Before the distribution of the survey, the questionnaire was reviewed by two public health specialists. Based on this review, modi cations of the number and content of questions were made. An Arabic version of the questionnaire was developed by the author, and its clarity was ensured by two bilingual researchers. A pilot study of 15 individuals was carried out to ascertain the clarity and applicability of the study tool and to identify the obstacles and problems that may be encountered during data collection. Based on the ndings obtained in the pilot study, some modi cations were made, including rewriting two questions to improve clarity.The Cronbach's alpha coe cient of the knowledge questionnaire in this study was .83, indicating reliable internal consistency [15]. For the attitude and practice questions, the inter-item correlations were .3 and .2, respectively. The optimal mean inter-item correlation values range from .2 to .4 [15].

Statistical analysis
For analysis of data, IBM SPSS statistics software, version 26, was used. Initially, data were entered, cleaned up, coded and tested for normality.
Frequencies and descriptive statistics of KAP of participants were calculated. Knowledge scores were compared across demographics using the independent samples t-test and one-way between-groups analysis of variance (ANOVA). Multiple regression was used to identify factors associated with knowledge scores, considering the demographics as a set of independent variables. For attitudes and practice questions, the Mann-Whitney U and Kruskal-Wallis tests were used to compare the demographics. Ordinal logistic regression analysis was used to identify demographic variables (as a set) associated with attitudes. A p-value ≤ .05 (two-tailed) for all tests was considered statistically signi cant.

Results
A total of 1144 participants responded to the survey questionnaire. However, the nal sample consisted of 1135 participants, as nine incomplete or duplicated questionnaires were excluded. The average age was 36.93 years (SD = 10.25, range 18-77). The majority of respondents were males (843; 74.3%), Saudi (1042; 91.8%), married (855; 75.2%), from the southern region (739; 65.1%) and held a university degree (738; 65%) with a non-health job (557; 49%) and used social media as a primary source of COVID-19 information (706; 62.2%) ( Table 1). Table 1 Demographic characteristics of participants Knowledge The majority of the study participants were knowledgeable about COVID-19. The mean COVID-19 knowledge score was 30.77 (SD = 5.39, range 7-40); thus, the mean percentage score was 77% (Table 2). An independent samples t-test was conducted to compare scores for the categories of gender, marital status, and having attended a health education activity on COVID-19. There were signi cant differences in scores for these demographics (all p < .05), with very small effect sizes (Table 3). A one-way between-groups ANOVA was conducted to explore the impact of age, education level, occupation and income per month on the COVID-19 knowledge scores. There were statistically signi cant differences in knowledge scores for categories of all demographics, with medium effect sizes for education degree (r = .07) and income per month (r = .06). Post hoc comparisons using the Tukey HSD test indicated that the mean knowledge scores for those >40 years of age were signi cantly different from other age groups and those with a health-related occupation were signi cantly different from other occupation groups, and that the knowledge scores for all education level groups as well as income groups were signi cantly different from each other (Table 3). Table 2 Questionnaire items of knowledge regarding COVID-19 Table 3 Knowledge by demographic variables using independent sample t-test and ANOVA Multiple regression analysis was used to estimate the possible effect of the demographic variables as a set on COVID-19 knowledge scores. It was also used to identify the unique contribution of each factor to the prediction of knowledge scores. Taken as a set, the predictors explain 12% of the variance in knowledge scores. The overall regression model was statistically signi cant F(7,1118) = 21.56, p < .001, R 2 = .12. However, only four variables made statistically signi cant unique contributions to the equation at p < .05: gender, education level, income per month and having attended a health education activity. The education level had the largest unique contribution (beta = .174), followed by the income per month (beta = .135) ( Table 4). Table 4 Summary of coe cients for the standard multiple regression of the demographic predictors on the knowledge scores

Attitudes towards COVID-19 by demographic variables
Ordinal logistic regression analysis was performed to assess the impact of a number of factors on the likelihood that respondents would report that they were worried about COVID-19. The model contained ve independent variables (gender, age, education level, occupation and income per month).
The full model containing all predictors was statistically signi cant, X 2 (5, N=1135) = 44.61, p < .001, indicating that the model was able to distinguish between respondents who reported and did not report worrying about COVID-19. Only three of the independent variables made a unique statistically signi cant contribution to the model: gender (p = .002, OR = .65), age (p = .033, OR = .77) and education level (p < .001, OR = .54). These indicated that for a one-unit decrease in each of these variables, the odds of worrying about COVID-19 is increased by .65, .77 and .54, respectively, given that all of the other variables in the model are held constant (Table 7). Table 7 Summary of results of ordinal logistic regression analysis on factors signi cantly associated with attitudes towards COVID-19.
Regarding the 'intention to use vaccine for COVID-19 if it was available' (Attitude2), 81% of respondents agreed to use the vaccine. Using Mann-Whitney U and Kruskal-Wallis tests, Attitude2 was statistically signi cant at p < .05 in gender only, with small effect sizes (r = .07; Table 6).
The majority of respondents (98.1%) agreed to 'inform the health authorities if they developed symptoms of COVID-19' (Attitude3). No statistically signi cant differences were revealed in this attitude item according to demographic variables (Table 6).
Regarding the 'con dence in the government efforts during COVID-19' (Attitude4), 89.8% of respondents agreed that they had con dence in these efforts. Using Mann-Whitney U and Kruskal-Wallis tests, Attitude4 was statistically signi cant at p < .05 across age, occupation and income per month. Post hoc Mann-Whitney tests using a Bonferroni-adjusted alpha level of .017 (.05/3) were used to compare all groups in the variables in three categories. The differences were signi cant in age (between <30 years and >40 years, p < .001, r = .138), occupation (between non-Health related and Unemployed, p = .002, r = .100) and income per month (between >10,000 to 20,000 and >20,000, p = .014, r = .099; Table 6). Ordinal logistic regression analysis was performed among the variables for which a statistically signi cant difference in Attitude4 had been shown (age, occupation and income per month) and Attitude4. The full model containing all predictors was statistically signi cant, X 2 (3, N=1135) = 9.031, p = .029. Only the age variable made a unique statistically signi cant contribution to the model (p = .006, OR = 1.29). This indicated that for a one-unit increase in age, the odds of trust in the government efforts is increased by 1.29, given that all of the other variables in the model are held constant (Table 7).

Practices
The mean COVID-19 practice score was 5.80 (SD = .603, range 1-6) indicating a 96.7% rate of positive practices ( . Mann-Whitney U and Kruskal-Wallis tests at p ≤ .05 revealed no signi cant differences in the practices items Practice1, Practice3 and Practice4 across demographic variables of gender, age, marital status, education level, occupation and income per month. A signi cant difference in Practice2 was revealed in gender (p = .014), with small effect sizes (r = .073; Table 9). For Practice5, statistically signi cant differences were found at p < .05 across age, marital status and occupation, with small effect sizes (r < .3). Post hoc Mann-Whitney tests using a Bonferroni-adjusted alpha level of .017 (.05/3) were used to compare all groups of age and occupation. The difference was signi cant in age only (between <30 years and >40 years, p = .014, r = .096; Table 9). Similarly, for Practice6, there were statistically signi cant differences at p < .05 in occupation only. Post hoc Mann-Whitney tests using a Bonferroni-adjusted alpha level of .017 (.05/3) were used to compare all groups of age and occupation. With these tests, the differences were no longer statistically signi cant (Table 9).
Based on these ndings, there was no need for further regression analysis. Table 8 Questionnaire items of practice regarding COVID-19 Table 9 Practices regarding COVID-19 by demographic variables Discussion COVID-19 is a major global threat impacting public health, social life, business, investments, political affairs and almost all other aspects of our life. Thus far, no vaccine or medication has been developed to control this infectious pandemic disease. The only effective way to protect the public from COVID-19 is for all to adhere to proper preventive measures. However, these measures are in uenced by the KAP towards COVID-19 in the population [16,17]. This study has assessed the KAP towards COVID-19 among the population of Saudi Arabia. It also discusses the efforts implemented by authorities to manage the crisis.
Many public health activities have been undertaken by the MOH to increase people's awareness and preventive practices of COVID-19. An important activity is the public awareness campaign through all available channels of information. The participants of this study stated that they receive COVID-19-related information from social media (62%), MOH mobile messages (60%), the MOH website (50%) and the public news (43%). Other important activities by the MOH include producing public health education guides in 25 languages, more than 26 scienti c guides for health practitioners, tens of educational videos, more than 45 prevention protocols for all aspects of life and daily public conferences [18][19][20]. The government authorities also suspended visas to the country and locked down all cities, schools, universities, public transportation, ights, public departments and malls, with exceptions for pharmacies and supermarkets [8,11,12]. Holy and public mosques were closed for the rst time in the recent history of the nation. These awareness and preventive efforts have resulted in a good level of knowledge regarding COVID-19 among the Saudi population.
The majority of the participants in this study (77%) were knowledgeable about COVID-19 in general. Although this nding is lower than that of a COVID-19 KAP study in China (90%) [13], it is similar to other ndings from Saudi Arabia (81.6%) [17] and Malaysia (81.5%) [21]. Differences in measurement scales limit accurate comparison of knowledge levels across these studies [21].
In this study, the vast majority of the participants correctly answered the knowledge questions. This could be attributed to the characteristics of the study participants; availability of information; modern technologies, such as the Internet and smartphones; and the public health efforts by the government authorities. However, substantial portions of the participants answered a number of knowledge questions incorrectly. For example, 40% of the participants were not aware of the MOH self-assessment tool for COVID-19, and 24% believed that there is licensed vaccine for COVID-19 available.
Nearly 30% of the participants believed that all patients with COVID-19 develop severe symptoms, while only 38.3% of the participants ranked pregnant women as a group at increased risk for COVID-19. According to a recent study by CDC researchers, pregnant women are at an increased risk compared to non-pregnant women. Pregnant women with COVID-19 are more likely to be hospitalised and admitted to the ICU for mechanical ventilation than non-pregnant women [22,23]. Although the majority of participants knew that there is no speci c medication for COVID-19 (63%) and that using medical and nursing care with supportive medications can help most patient recover (62%), there were still large portions who were not aware of this information (37% and 38%, respectively). This nding is different from prior studies, in which participants indicated higher knowledge on these topics [13,17]. More than one-third of the participants were not aware that in patients with more severe COVID-19, the infection may cause pneumonia and acute respiratory distress syndrome. Overall, 66% of the participants knew that during the COVID-19 pandemic, anyone with fever, cough or shortness of breath must wear a face mask in addition to isolating himself or herself and contacting the MOH hotline. However, 34% did not know the importance of wearing a face mask. These knowledge areas may require special attention through intensive awareness interventions.
A signi cant association was found between knowledge scores and the demographics of gender, age, marital status, education level, occupation, income per month and having attended a health education activity. Testing these predictors as a set, only four predictors made signi cant unique contributions to the participants' knowledge: gender, education level, income per month and having attended a health education activity. Male participants with a higher education level and high income who attended a health education activity were more likely to have more COVID-19 knowledge. This is consistent with prior research, except for the nding on gender [13,17,21]. Public health policymakers may bene t from these ndings, as they could be used to direct their awareness interventions to suitable population groups.
Regarding attitudes, the majority of participants held positive attitudes towards COVID-19 measures. Although 96% feel worried about having COVID-19, which is important for promoting preventive behaviors, 89.8% are con dent that the Saudi government measures are su cient to control the COVID-19 pandemic. This level of trust could be attributed to the substantial COVID-19-related preventive measures and actions taken by the government authorities. This nding is similar to those of other studies from Saudi Arabia [17], China [13] and Malaysia [21]. Signi cant predictors of being worried about COVID-19 in this study were gender, age and education level. Younger, male and less educated participants are more worried about having COVID-19. Concerning the participants' con dence in the government efforts, only age (being older) made a signi cant unique contribution after controlling for all associated predictors. Interestingly, when asked whether they would take a vaccine if available, 19% either were not sure or disagreed. The reason for this is not clear, but it may be due to misinformation among the public. In contrast to the ndings of prior research, this attitude is found to be signi cantly associated with male participants of this study [24]. Most of the participants stated that they will inform the health authorities if they experience symptoms of COVID-19. Unfortunately, 2% were still not willing to disclose their symptoms, if developed. This attitude could be attributed to their being worried about isolation time and/or due to their worrying about social stigma related to COVID-19. Social stigma can negatively affect those with diseases and their family, friends, caregivers and even communities [25]. Such attitudes may increase the risk of COVID-19 spreading throughout the population. Hopefully, this attitude will change with time in response to the current intensive public awareness campaign concerning COVID-19.
Most participants of this study reported taking precautions to protect themselves and others from COVID-19. Nearly all participants reported that they maintain social distancing (99.2%) and avoid leaving the house except for necessities (98.6%). These practices re ect their desire to prevent this infectious disease. Interestingly, 92% of the participants indicated that they wear a face mask in public. This could be attributed to the good understanding of the high infectiousness of COVID-19 and their ability to afford the cost of face masks. At the time of data collection, the MOH and WHO urged people not to use face masks, except for those with respiratory symptoms or who were caring for COVID-19 patients [26,27]. The WHO had stated that using face masks by healthy people to prevent COVID-19 may create a false sense of security, result in unnecessary costs, and take masks away from healthcare workers who need them most [26]. It is possible that the 8% of respondents who did not wear face masks in public were following the previous instruction by the MOH and the WHO. Currently, wearing a face mask in public is mandatory, as the authorities of Saudi Arabia imposed nancial nes on those who do not comply [28]. A signi cant association at p < .05 was found between wearing a face mask in public and gender, as females showed more commitment. One of the unexpected results is that 97% of the participants avoid shaking hands. Prior research reported a lower percentage among the Saudi population [17]. Shaking hands is a crucial public practice in Saudi Arabia that is seen as a sign of respect. Avoiding shaking hands during this pandemic is a major behavioral development that indicates public seriousness in dealing with COVID-19.
The results also indicate that 97% of the participants maintained proper hand hygiene (according to the recommendations) and only touched their faces with clean hands. This nding is higher than that of other studies [17,21]. Avoiding of touching the face with unwashed hands was found to be signi cantly associated with age (being older).
Generally, the good practices of the participants towards COVID-19 could be attributed to the modern life in Saudi Arabia and the major preventive efforts and strict control measures taken by the government during the COVID-19 pandemic.
Recommendations to the health system Public health as a discipline in Saudi Arabia has faced many challenges over the past decades, as most support is directed to clinical medicine. COVID-19 has made public health and clinical medicine again work side by side in order to protect the society. This collaboration needs to be strengthened, with more focus on public health, as proposed by the Saudi Vision 2030.
The MOH has issued a number of health awareness guides in many languages. However, the availability of mobile and social media communications for expatriate workers in Saudi Arabia who are non-Arabic or English speakers appears to be less than supposed. It is necessary to pay attention to this category of workers because they constitute a large segment of the Saudi population.
Participants mentioned that they derive most of their information on COVID-19 from four sources: social media, mobile messages from the MOH, the MOH website and public news, respectively. It would therefore be appropriate to use these methods in future awareness campaigns. In addition, it is necessary to gather all COVID-19 efforts by various parties in a single website. This step will make it easier for researchers, planners and those interested to follow the latest developments and measures of the pandemic.
The enactment of laws and regulations to control COVID-19, along with focused awareness campaigns, have shown positive results regarding the level of knowledge and people's attitudes and practices. In support of awareness campaigns during future epidemic outbreaks, health authorities may need to impose supportive laws and regulations to encourage people to adhere to preventive measures.
Drawing on its considerable experience during COVID-19, the MOH, in collaboration with other bodies, should develop a comprehensive control system to manage emerging health epidemics. Health leaders and managers at all levels must be trained to lead during health pandemics and crises.

Study Strengths And Limitations
This study has a number of limitations. Due to the limited time and the urgency of the pandemic, the study used a convenience sampling method. The data were collected from participants who were willing to participate in the study. Although all regions of Saudi Arabia were included, the voluntary sampling methodology and variance in the number of participants may limit the ability to generalise the ndings to the Saudi population. In addition, the study used an online survey to collect the data, leaving the interpretation to the participants. Participants may also have given socially acceptable responses, rather than their real attitudes and practices. A further limitation of the present study is that it focuses on educated people who have internet access. Vulnerable populations of Saudi Arabia, such as rural communities, older people, expatriate labor workers and those who cannot read or use technology, need more consideration in future research. Despite these limitations, the ndings of the study provide an important contribution to the existing research on the KAP of COVID-19, particularly in Saudi Arabia. The ndings could assist healthcare leaders and policymakers in understanding the KAP and experience of the Saudi population regarding COVID-19. This may help health authorities plan for future preventive efforts and health crises.

Conclusion
The ndings of this study suggest that the Saudi population are generally knowledgeable and having good attitudes and practices regarding COVID-19.
In particular, male participants with higher education level and high income and who attended a health education activity are more likely to have more COVID-19 knowledge. Saudi Arabia has experience in dealing with epidemiological challenges that emerge at mass gathering activities such as Hajj. In addition to this experience, the ndings of this research could be used to develop appropriate strategies to improve COVID-19-related KAP and to plan for future health crises such as pandemics. Such planning will help local authorities react promptly in order to prevent any health threats to the Saudi population and those who visit Islamic holy places in Saudi Arabia.

Availability of data and materials
The datasets used and/or analyzed during the current study are available from the corresponding author upon reasonable request.

Competing interests
The author declares that he has no competing interests.

Funding
None.

Author Contributions
Mohammed J. Almalki is the author responsible for conception, design, acquisition of data, or analysis and interpretation of data, drafting the article or revising it critically for important intellectual content, gave nal approval of the version to be published, and agreed to be accountable for all aspects of the work. Note. * Others in the marital status include divorced and widowed. ** Participants can select more than one source, MOH: Ministry of Health, WHO: World Health Organization, Internet search engines such as Google. Note. N=the total number of the sample; answer options = yes, no, not sure (K.1-K.9); yes answer score=1, not sure=0 and no=0, K.4 and K.5 are reversed scored items as yes=0, not sure=0 and no=1; n=number of respondents chose yes; % = percentage of the "yes answer" compared to the total sample. Note. * p < .05. ** p < .001. Size effect using r 2 : .01 = Small effect, .06 = Medium effect, .14 = Large effect.  Note. N=the total number of the sample; answer options= Agree (Agree & Agree Strongly), Not Sure, and Disagree (Disagree and Strongly Disagree); n=number of respondents to each item; % = percentage of the answers to each item compared to the total sample.   Note. N=the total number of the sample; answer options = yes, no; n=number of respondents to each item; % = percentage of the answers to each item compared to the total sample.  Note. P-value < .05 is statistically signi cant; Mann-Whitney U Test for two categorical independent variables; Kruskal-Wallis Test for three or more categorical independent variable; n=the number of respondents in each group; effect size was calculated using eta squared, post-hoc tests and Bonferroni adjustment as appropriate and presented in the result section; for more information on Practice1(P.1) -Practice 6 (P.6), see Table 8.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.