Structural equation modelling of COVID-19 knowledge and attitude as determinants of preventive practices among university students in Ghana

Coronavirus disease (COVID-19) has distorted the economic development activities of many countries across continents. This undesirable tragedy has highly affected the educational system, which majorly contributes to the wellbeing of an individual and the economy as a whole. The study aims to explore the determinants of COVID-19 preventive practices among students considering their knowledge about COVID-19 and attitudes toward the disease. The data for the study were collected through an online questionnaire survey involving university students. The relationship between students’ knowledge, attitude and their preventive practices towards COVID-19 were investigated using structural equation modelling. The results indicated that most students demonstrated substantial knowledge on COVID-19, moderate to strongly agree attitude towards COVID-19, and sometimes practiced COVID-19 preventive and safety protocols. In addition, a positive relationship between knowledge and attitudes towards COVID-19 was established. Also, a positive effect was established for students’ knowledge about COVID-19 and preventive practices, whilst an adverse effect was confirmed for attitudes towards COVID-19 and practices to avoid spreading the COVID-19 disease.


Introduction
Coronavirus disease 2019  is the world's current greatest problem. The spread of COVID-19 began in Wuhan, a city in China and has currently been transmitted to almost all countries worldwide [1] . COVID-19 affected persons show symptoms of dry cough, short breath, myalgia, fatigue and fever, with an advance stage characterized by septic shock, respiratory distress syndrome, bleeding and coagulation dysfunction [2] . The effect of the virus has posed a psychological threat to daily human activities following the declaration of the disease as a public health emergency of international concern and a pandemic on January 30, 2020, and March 11, 2020, respectively [ 3 , 4 ]. This global crisis has forced countries to implement different strategies to reduce the rate of transmission and produce vaccines to provide immunity against the disease [ 5 , 6 ]. The pandemic has posed several challenges on human lives following the introduction of preventive measures such as national quarantines and stay at home measures which have led to largely unknown mental health implications concerning people's living environment, especially in big cities [ 7 , 8 ].
A total of over 266 million confirmed cases of COVID-19 and about 5.3 million deaths have been reported as of November 29, 2021 [9] . Currently, many vaccines have been produced and distributed across countries to curb the transmission rate of coronavirus disease. However, the battle against COVID-19 exists in most countries where people are exposed to the threats of this dangerous disease. Although territories like Marshall Island, Solomon Island, Saint Helena and a few more have recovered all active cases from COVID-19 infection, most countries are experiencing different waves of the COVID-19 pandemic [9] . Recently, different variants of the COVID-19 virus have also been discovered in many countries, thereby creating fear and panic among people worldwide.
The education system is one of the most affected sectors in the COVID-19 pandemic era owing to the large number of students who visit educational facilities to obtain training. Due to crowding and close contact situations, initiatives taken by governments directed schools to shut down totally because of the increasing numbers of confirmed cases reported each day. The situation led most educational institutions to move academic activities online to reduce face-to-face sessions as part of the interventions [10] . However, this shift in the educational system has resulted in different challenges to students, including lack of access to online learning aids such as computers, and stable internet accessibilities, thereby making online learning a challenging task during this period [11] . Due to these challenges, most educational institutions have resumed face-to-face class sessions with imposed governments restrictions on COVID-19 safety protocols.
To enhance a safe and continuous academic life in schools, it is essential to assess students' understanding and state of mind that affect their preventive practices against the coronavirus disease. Several studies [12][13][14][15] have been conducted to assess the knowledge, attitudes and practices (KAP) among students. Other studies [ 10 , 16 , 17 ] have also focused on the challenges, and the negative psychological impact students encounter in the COVID-19 pandemic. Research findings from these studies could be used for decision making and policy implementation. However, studies from specific countries may be necessary due to the diversification of culture and behavioural differences among students from different geographical locations.
This study seeks to determine whether the practices performed by students to avoid COVID-19 infection are influenced by their knowledge and attitudes using structural equation modelling (SEM) techniques. The conceptual model developed in the study exhibits the importance of knowledge and attitude as predictors of practices ( Figure 1 ). The model is used to test the following hypothesis: H1 : Knowledge about COVID-19 and attitudes are related. H2 : Knowledge about COVID-19 influences preventive practices. H3 : Attitude towards COVID-19 influences preventive practices.

Survey design
The data used in this study were obtained through a questionnaire survey. The questionnaire is a modified version of the instrument used by [18] to assess COVID-19 related KAP among tertiary students. The questionnaire consisted of 41 questions and was divided into four sections: demographics (6 questions), knowledge (17 questions), attitude (6 questions), and practices (12 questions). Under the COVID-19 knowledge section, optional answers of "yes", "no", or "don't know" were provided to the respondents. These answers restrained participants from selecting the correct answer by chance. Under the attitude section, participants were asked to rate a 5 point Likert scale questions ranging from strongly disagree to strongly agree. Also, participants were asked to choose from options of "always", "sometimes", and "never" practiced in the practices section. A pilot survey involving 50 participants was conducted in the early stages of the study to assess the validity of the questionnaire and revise where necessary. The responses from the pilot survey were not included in the analysis.

Participants and data collection
The participants involved in the survey were university students (i.e. both undergraduates and postgraduates) in Ghana. The study was limited to students at Kwame Nkrumah University of Science and Technology (KNUST). KNUST is the second largest public university in Ghana and is well known for quality research and higher education. The questionnaire was selfadministered through an online survey system (Google Forms) which lasted for 10 days between 16 th to 25 th October 2021. The students were targeted through social media class groups including WhatsApp and Telegram which is commonly used by them in the university. A message comprising of the objectives and importance of the survey, a link to the questionnaire and a consent to fill the questions were sent to almost all social media platforms of student groups. Participants were encouraged to recommend the link to other coursemates and friends in the university. A total of 1075 students responded to the questionnaire.

Structural equation model
Structural equation modelling (SEM) is a widely used multivariate technique for investigating the direct and indirect effect of relationships between observed and latent variables. In SEM, an observed variable is a variable that has been directly measured and contributes to the composition of a latent variable. On the other hand, latent variables are unobserved variables that cannot be measured directly. The theory of SEM simplifies complex relationships between variables by utilizing a path model or analysis for explaining effects resulting from observed and latent variables [ 19 , 20 ]. The technique clearly determines the relationship between cause-and-effect variables. The SEM aims to generalize a confirmatory factor analysis (CFA) model to assess the relationship between latent variables and how they affect each other in many ways. The SEM summarises a linear structural relationship into a measurement model (CFA model) and structural model [21] . The measurement model is defined as: where x 1 ( r × 1 ) and x 2 ( s × 1 ) are random vectors of the observed variables that explains endogenous (dependent) η and exogenous (independent) ξ latent variables, 1 (r × q 1 ) and 2 (s × q 2 ) are loading matrices, and ε 1 (r × 1) , and ε 2 (s × 1) are random vectors of measurement errors. The measurement model can also be expressed in matrix form as: On the other hand, the structural model can be defined as: where B ( q 1 × q 2 ) and ( q 1 × q 2 ) are unknown matrices of regression coefficients that describes the cause-and-effect relationship between η and ξ , and δ( q 1 × 1) is a random vector of residuals or error of endogenous variables. Thus, the SEM is expressed as a CFA model whose latent factors represent a linear structural equation. The model aids in finding a direct and indirect effect among variables and the measurement errors. Statistically, SEM depicts an extended form of a generalized linear model like the multiple regression model and analysis of variance [ 20 , 22 ].
The performance of the measurement and structural model was evaluated using root mean square error approximation (RMSEA), comparative fit index (CFI), tucker-lewis index (TLI), goodness of fit index (GFI), and adjusted goodness of fit index (AGFI). The RMSEA, CFI, and TLI can be expressed as [23] : T LI = with their corresponding degrees of freedom. Also the GFI and AGFI are given as [24] : where W is a weight matrix condition on the method of model parameter estimation, tr is the trace or addition of diagonal elements contained in the matrix, ˆ is the covariance matrix of the observed variables in the restricted model, S is the unrestricted sample covariance matrix relating to the saturated model, and k represents the number of observed variables.

Results
The demographic profile of the students involved in the study shows participation of 61% males and 39% females ( Table 1 ). About two-thirds of the students who participated in the study were more than 20 years. Most of the students were studying in regular mode and 67% resided in accommodations outside the university campus. In addition, almost 40% of the students were pursuing science related academic programmes whilst 5% of them were pursuing agricultural and natural resources related programmes.
The frequencies and percentages distribution of correctly answered questions in the knowledge section among students are displayed in Table 2 . About 90% of the students demonstrated a good understanding of what causes COVID-19, symptoms and knowledge about early treatment of COVID-19 patients as an effective step to help patients recover as well as avoid further transmission. A little over half of the students (55%) correctly answered questions on knowledge about the spread of COVID-19 through buried deceased COVID-19 patients. Also, 47% of the students knew that COVID-19 does not only spread through surfaces of materials. Correct answers on COVID-19, symptoms, handling of deceased, transmission and knowledge on preventive measures have been supported in existing literature [ 1 , 25-31 ].
About 40% of the students were highly concerned about COVID-19 confirmed cases and 42% were anxious about government's call to reduce the rate of transmission in the communities ( Table 3 ). A little over 49% of the students believe that COVID-19 infected persons who isolate themselves show a responsibility in preventing the transmission of COVID-19. In addition, 39% of the students were scared and worried with 19% expressing extreme fear about the total number of COVID-19 cases reported globally each day.
The characteristics of the preventive practices against COVID-19 is presented in Table 4 . It was observed that majority of the students (73%) always wear nose mask while 54% sometimes implemented social distancing protocols in crowded  places. Also, 24% of the student always educate other people on COVID-19 preventive measures. About 35 -43% of the students implemented basic COVID-19 safety protocols such as washing of hands with soap under running water, using of hand sanitizers and observing social distancing protocols. The diagonally weighted least square (DWLS) estimator was used in the estimation of the factor loading of the SEM parameters since all observed variables considered in the study were categorically scaled. The DWLS yields more accurate factor loading estimates than maximum likelihood estimation with such data [32] . Table 5 shows the estimated factor loadings of the initial model and final model achieved. In the initial stage, the estimated model exhibited a poor fit as the RMSEA = 0.067, CFI = 0.846, and TLI = 0.836 were below their respective cut-off points ( Table 6 ). Hence, there was a need for model improvement by implementing the theory of trimming in modifying the initial model. This method helps in eliminating variables that are non-significant and contribute poorly to the fitness of the initial model developed. The observed variables K6, K7, K8, K10, K11, K12, A3, A4 and P1 were excluded in the determination of the final model due to low and    non-significant factor loadings in the initial model generated. The final model selected was found to be acceptable as the fit indices meet the minimum criteria ( Table 6 ). From the developed model ( Figure 2 and Table 7 ), there exist a significant positive relationship between knowledge and attitudes that students exhibit towards COVID-19 (r = 0 . 404 ; p < 0 . 001) , explaining the hypothesis (H1). The second hypothesis H2 initiates a claim that knowledge about COVID-19 affects preventive practices. A significant positive effect (β = 0 . 192 ; p < 0 . 001) was established for H2, whiles there existed a significant negative effect (β = −0 . 046 ; p = 0 . 007) between attitude and safety practices that prevents coronavirus disease (H3).

Discussion
The COVID-19 pandemic impact have posed a major distortion on the daily life activities of individuals worldwide. This has significantly change the way of living among persons and has extremely affected the education sector which majorly contributes to the development and wellbeing of an individual. The diversion of normal life activities combined with fear of a deadly and highly transmissible disease, poor performance of economies, and an uncertain future have kept many anxious about the consequences of the COVID-19 pandemic [ 33 , 34 ].
In this study, the effect of knowledge and attitude on COVID-19 preventive practices among university students in Ghana was explored. Also, the relationship between knowledge and attitude was tested. Findings from this study revealed that most students strongly believed that achieving knowledge and well informed updates on number of reported cases of COVID-19 as well as information about COVID-19 preventive efforts are very relevant to living in their various communities.
With the exception of students having little knowledge about handling of COVID-19 deceased person, they exhibited substantial understanding of the causes of COVID-19, symptoms an infected person is likely to show, conditions of infected children and elderly, information on transmission and transmissibility as well as preventive practices and knowledge about asymptomatic persons. According to [35] , addressing student knowledge and understanding of the disease should have been the starting point prior to returning to school. This will also allow them to give better information (i.e. it will reduce the myth) about the disease to others who are ignorant of it. [36] also argued that individuals access to information about a particular health issue influences the level of knowledge. The study found a significant positive impact of knowledge on COVID-19 preventive practices among students. Thus, student's preventive practices turn to get better if they have substantial knowledge about the disease. This results is consistent with the findings in existing studies [ 36 , 37 ]. Thus, lack of knowledge is likely to contribute to the high risk and easy spread of the disease.
In relation to student's attitude towards COVID-19, many of them were against the view that the infected persons were those who violated government's effort s in the implementation of policies to prevent the transmission of COVID-19. However, a little over 5% were in favour of stigmatization of COVID-19 infected and recovered persons. The act of stigmatization can psychologically affect the wellbeing of COVID-19 affected persons in their surroundings. This can also lead to further transmission of the virus and subsequent death since infected persons who show symptoms would avoid visiting healthcare facilities for disease treatment [38] . Contrary to positive relationship between knowledge and preventive practices, a significant negative relationship was found between student's attitudes and COVID-19 preventive practices. That is, the more an individual's attitude towards COVID-19 improves, he or she is less likely to implement COVID-19 preventive practices. This result contradict the findings in the existing studies [ 14 , 39 ] which revealed that attitude towards COVID-19 has a positive impact on preventive practices. These differences may be influenced by the concept of optimistic bias which occurs when an individual felt shielded against risk. This type of relationship may also demand further investigation.
Knowledge about COVID-19 and attitude towards the disease were positively correlated. This indicates that students' attitude may improve on the basis of having a good understanding of their environment and information they receive about the disease. This type of relationship is reciprocal; thus, being knowledgeable about the disease can improve an individual's attitude towards it. This result is in agreement with the findings in existing literature [39][40][41] . However, irrespective of student's awareness and eagerness to acquire information about COVID-19, more than 50% of the students expressed fear and anxiety from daily reported cases of COVID-19. This indicates the importance of policy makers to implement interventions to support students and the general public during this crisis.

Conclusion
From the study, students are likely to exhibit good preventive practices when knowledgeable about COVID-19. Thus, the rate of disseminating information on COVID-19 should be intensified particularly those dealing with safety protocols. Because lack of knowledge on the disease constitute the major obstacle to preventive practices. Educating students and the general public on the need to implement safety protocols at all times may help to minimised the risk of infection and spread. Such education can also help to explain the myth the public have on the disease.

Funding
No funding was received for this study.

Ethical approval
The study was approved (document no. CHRPE/AP/491/21) on 15 th October 2021 by the Committee for Human Research, Publications and Ethics (CHRPE) of the School of Medicine and Dentistry at the Kwame Nkrumah University of Science and Technology. The committee assessed whether the study was in accordance with ethical standards.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.