Predicting COVID-19 Preventive Behaviors based on Protection Motivation Theory in Hormozgan, Iran


 Background: The present research aimed to predict the preventive behaviors of COVID-19 in the light of the Protection Motivation Theory (PMT).Methods: The present cross-sectional research was conducted in 2 months (March 2020-April 2020) in Hormozgan Province of Iran. The research population comprised all residents above 15 years of age. They received an online questionnaire survey in two parts, demographic information along with PMT constructs. To analyze the data, independent t-test and one-way anova test, Multiple linear regression and the statistical technique of path analysis and structural equation modeling was used. All statistical calculations and hypothesis testing were performed using spss21 and Amos21 software and a significant level of hypothesis testing was considered to be 0.05Results: A total number of 2,032 subjects participated in this research, the mean age of whom was 34.84±9.8 (R=15-98). The majority of participants were between 31 - 40 years old, female (60.4%), married (72%), urban residents (87.3%), holding a Bachelor’s degree or higher (58.8%) and employed (58.8%). Statistically significant positive correlation coefficients were estimated between preventive behaviors of COVID-19 and perceived vulnerability (r=0.192, p<0.001), perceived severity (r=0.092, p<0.001), response efficacy (r=0.398, p<0.001) ,self efficacy (r=0.497, p<0.001) and protection motivation (r=0.595, p<0.001). Statistically significant negative correlations were found between behavior and maladaptive behavior rewards (r=-0.243, p<0.001) and perceived costs (r=-0.121, p<0.001). Conclusions: The present findings showed that maladaptive behavior reward and fear predicted protective behaviors negatively; response efficacy and self-efficacy predicted protective behaviors positively. The effect of self-efficacy was the strongest among all. The information provided in this research can contribute to policy-making in the country.


Background
Today, the prevalence of infectious diseases is considered a global issue (1) .Respiratory diseases are widespread and epidemic and account for a high rate of mortality in communities (2).
In December 2019, a new Coronavirus named as nCoV-2019 appeared as pneumonia pandemic in Wuhan (originating from sea food markets) and soon spread in whole China (3). On January 30, 2020, the pandemic of the new Coronavirus was announced by the World Health Organization (WHO) as a public medical emergency (4)(5)(6)(7) . COVID-19 is a serious infectious respiratory disease and is currently considered the most fatal threat to human population. This disease has infected large populations so far and has accounted for many mortalities (8).
The most frequent symptoms of the disease include fever, dry coughs and fatigue (9,10) Fortunately, this disease can be prevented by taking good care of health. Preventive measures exempli ed by WHO were avoiding close contact with the a icted, washing hands recurrently and avoiding unprotected contacts with domestic or wild animals. Moreover, those with the symptoms of acute respiratory infection should take certain protective measures (e.g. keeping a 1-2-meter distance, covering face while coughing or sneezing with tissue-paper or cloth, washing hands recurrently and not touching the mouth, nose and eyes. Moreover, stress and anxiety management can help prevent the disease (8,11). Until July 28, 2020, global statistics of COVID- 19 show that a total number of 215countries from all continents are a icted. Iran ranks eleventh among all. The total number of a icts at a global scale exceeds 16899316. The mortality rate is 663540 cases. In Iran, 296273 people are a icted while 16,147 people died. In Hormozgan, these are, respectively, 19,934 and 516 cases (12).
In Iran, On February 19, 2020, two patients were diagnosed with COVID-19 in Qom city. Afterwards, the disease quickly spread to the neighboring provinces including Tehran, Markazi, Isfahan and Semnan.
Soon, the disease was spread to 31 provinces (11).
As this disease is truly contagious, it is rapidly widespread and it a icts so many people; the adverse effects are severe especially the acute respiratory syndrome and the mortalities induced; no de nite cure or vaccine has ever been developed; thus, protective measures seem to be essential (8,9).
Healthy behavior theories can help better perceive the effective factors involved in protective behaviors to plan for health promotion. The Protection Motivation Theory (PMT) was introduced by Rogers (1975) and has been widely used as a framework to predict protective behaviors (13).
PMT assumes that adopting a protective behavior recommended against health threats is a direct act of personal motivation for self-protection. In PMT, fear is appraised to predict and encourage people to adopt protective behaviors and explain the intermediary cognitive processes for threat appraisal and coping appraisal (14).
Threat and coping appraisals can lead to the intention of adaptive response (as a protective motivation) or can lead to maladaptive responses, which are threats to one's health (15).
In PMT, threat appraisal depends on: 1. one's belief in the severity of the problem (perceived severity) 2. one's estimation of the chances of being infected with the disease (perceived vulnerability) 3. one's belief in the positive aspects of unhealthy behavior (perceived rewards). Thus, if perceived severity and vulnerability are high and perceived rewards is low, there is a stronger motivation for engagement in health promoting behaviors. In PMT, coping appraisal includes: 1. one's evaluation of the extent to which the protective behavior is effective in coping with the threat (response e cacy) 2. one's belief in one's own capability of managing protective behaviors (self-e cacy) 3. one's estimation of any cost including money, time, energy and efforts spent on showing protective behaviors (perceived response cost).
Response e cacy and self-e cacy are expected to reinforce coping appraisal. To the contrary, response cost is expected to lower it ( Figure 1) (16) Fear is a moderating variable between perceived vulnerability, severity and threat appraisal. Thus, if one feels vulnerable to a serious health threat, the level of fear is increased and one is further motivated to adopt a preventive/protective behavior. These two appraisal processes are integrated to form protection motivation (17).
A body of related research shows that PMT is very well capable of predicting the adoption or nonadoption of protective behaviors (2,18,19).
PMT has been used to investigate behaviors such as the reception of in uenza vaccine (20), preventive behaviors of H1N1 pandemic (2) ,cancer preventive behaviors and sun-safe behavior (21,22) SARS preventive behaviors (23) and preventive behaviors for infectious diseases and skin cancer preventive behaviors (24)(25)(26). Yet, this theory has not been employed to explore protective behaviors of COVID-19 in Hormozgan Province population.

Research Design:
The present cross-sectional research was conducted within a period of 2 months (March 2010 to April 2020) in Hormozgan Province. The target population comprised residents of Hormozgan Province aged above 15 years. Hormozgan is located in a Southern province in Iran located on Persian Gulf coastline.
This province includes 13 counties. The population of this province, according to the census in 2017 was estimated at 177641500. To avoid the probability of transmitting the disease through paper-and-pencil surveys, online survey was preferred. The link of the survey was shared via social networks or emails with participants and they were kindly requested to respond to the questions. Inclusion criteria were the minimum age of 15, literacy and residence in Hormozgan Province. The exclusion criterion was unwillingness to participate. The data collection instrument was a researchermade questionnaire, the content validity of which was tested and approved.
The content of the questionnaire was derived from credible sources including relevant books and academic papers, and it was approved by a panel of 3 experts (health education and promotion specialists) the required changes were both qualitatively and quantitatively made. The former involved asking experts to rate items in terms of the words chosen, the order of presentation and appropriate scoring. They provided the feedback and the corrections were made accordingly.
The reliability of the questionnaire was checked through Cronbach's alpha test and was approved. The questionnaire consisted of two parts, the rst of which addressed demographic information (i.e. age, gender, marital status, education, occupation, background disease, history of smoking tobacco). The second part dealt with the PMT constructs. All constructs of PMT except for behavior were rated on a 5point Likert scale (1 to 5) ranging from strongly agree to strongly disagree. The behavior construct was rated on a 5-point scale (0-4) ranging from never to always.
Perceived vulnerability (e.g. "I may also get a icted with Coronavirus.") was tested with 4 items and a score ranging from 4 to 20. Perceived severity (e.g. "If a icted with Coronavirus, there are chances of early death.") was tested with 6 items and a score ranging from 6 to 30. Perceived reward of maladaptive behavior (e.g. "It is easier to breathe without a mask.") was tested with 7 items and a score ranging from 7 to 35. Perceived response e cacy (e.g. "Recurrent washing of hands with water and soap for at least 20 seconds can protect me against Coronavirus.") was tested with 7 items and a score ranging from 7 to 35. Perceived self-e cacy (e.g. "I can adequately and appropriately disinfect contaminated or suspicious things and areas.") was tested with 6 items and a score ranging from 6 to 3o. Perceived response cost (e.g. "Disinfecting contaminated things and areas is time-consuming.") was tested with 7 items and a score ranging from 7 to 35. Fear (e.g. "When I think about Coronavirus, I feel anxious.") was tested with 6 items and a score ranging from 6 to 30. Protection motivation (e.g. "I have decided not to travel until the disease is over.") was tested with 7 items and a score ranging from 7 to 35. Finally, behavior (e.g. "I avoid kissing or shaking hands with others.") was tested with 10 items and a score ranging from 0 to 40.

Data Analysis:
Frequency, percentage, average and standard deviation indices will be used to describe the data. To test the hypotheses and explore the correlation of COVID-19 preventive behaviors and the demographic variables, T-test and one-way ANOVA test were used. To test the correlation between the protection motivation theory constructs and preventive behaviors, multivariate linear regression analysis was used.
Moreover, to test the interaction of constructs within the model and accept/reject the conceptual model of COVID-19 preventive behaviors, path analysis and structural equation modeling were used. All statistical analyses and hypothesis testing was done in SPSS21 and AMOS21. The signi cance level was set at 0.05.
As the results showed, the mean score of protective behaviors was signi cantly higher among participants who were above 15 years of age, female, married, holding a Bachelor's degree or higher, employed, health staff, urban residents and non-smokers. The demographic information of the participants and the mean scores of their protective behaviors are summarized in Table 1. ( Table 1) Table 2 includes Cronbach's alpha and the range of scores for PMT constructs. The crude mean scores, adjusted mean scores, percentages as well as Pearson correlation coe cients of PMT constructs are included concerning the target protective behaviors.
R2 for protection motivation is .330, which indicates that 33% of variance in the dependent variable (motivation) is explained by the independent variables (coping appraisal, fear and threat appraisal).
Two variables, rewards (Beta = -0.046) and Fear (Beta = -0.078) showed to be negatively correlated with behaviour. A higher score of these two variables was associated with a lower score of COVID-19 preventive behaviour. ( Table 4) In Table 5, the distribution (frequency and percentage) of protective behaviors reported by the participants are summarized. The results show that 79.8% of the participants always avoid unnecessary travel; 78.6% always avoid kissing or hand-shakes; 74.4% wear masks if they approach a suspicious case of the disease. (Table 5)

Discussion
The present research aimed to predict the adoption of COVID-19 preventive behaviors based on the Protection Motivation Theory among Hormozgan residents, those above 15 years of age. The results revealed that maladaptive behavior rewards, response e cacy, self-e cacy and fear predicted the protective behaviors. Among these constructs, self-e cacy was the strongest predictor. The mean score for COVID-19 preventive behaviors in the target population was above the average. This can be due to the adequate awareness-raising and management of the disease. Among preventive behaviors, avoiding unnecessary travel, avoiding kissing/shaking hands and wearing masks showed to be the most prevalent preventive behaviors.
Moreover, the results showed a statistically signi cant correlation between age and protective behavior of the disease. The higher the age, the higher the frequency of adopting protective behaviors. This can be due to a higher awareness and perceived threat at the higher age. Older people are more prone to background diseases, which can be a reason for more protection too. Perceived severity is dependent on age, and is higher among the elderly than the youth (27).
The present nding is consistent with some similar research about unsafe driving in Yazd (28), which showed that a higher age is associated with more protective behaviors. Yet, the results reported by Lowe et al. among Australian university students showed that a higher age was followed by fewer protective behaviors against sunlight (22). This is divergent from the present nding and the reason can be the different cultures and protective behaviors binvolved. As the results showed, it is essential to plan for the youth's awareness-raising and to encourage them to adopt more protective behaviors.
The overall score showed statistically signi cant differences for preventive measures and education levels. Those holding a Bachelor's degree or higher adopted more protective behaviors, which can be due to their higher level of knowledge. It is, thus, necessary to plan for making the required interventions to raise the awareness of less educated residents of the province.
The results also showed that women adopted more preventive behaviors than men. Several other studies in other ethnicities also con rmed such differences between male and female in terms of health beliefs and healthy behaviors. These ndings show that certain gender-speci c plans need to be made to reinforce preventive behaviors for COVID-19.
Pearson's correlation coe cients showed statistically signi cant positive correlations between preventive behavior and perceived severity, vulnerability, response e cacy, self-e cacy and protection motivation.
Signi cant negative correlations were found between behavior and perceived reward and costs. These are consistent with a number of studies (21,29).
The present ndings also revealed that COVID-19 preventive behavior was signi cantly and positively correlated with perceived vulnerability. This nding shows that if people perceive themselves vulnerable to the disease, they adopt more protective behaviors. These ndings are consistent with the results reported by Babazadeh et al (30) and Mohammadi et al (31).
The present research also found that protective behaviors and perceived severity were positively and signi cantly correlated. Thus, if people are made aware of the health consequences of the disease, they adopt more protective behaviors. This is in agreement with Tazval et al (32) yet different from Hadi et al (33).
A statistically signi cant positive correlation was also observed between perceived vulnerability and severity of the disease. This nding is consistent with a number of studies including Barati et al (34)., Zare et al (21) and Park et al (35). These ndings show that for a better perception of the risk of a iction with the disease, a higher level of perceived vulnerability is needed, as such a perception can strongly and positively affect the perceived threat of COVID-19.
A statistically signi cant positive correlation was found between perceived response e cacy of preventive behaviors and self-e cacy and protection motivation. This shows that an increased perception of the effectiveness of protective behaviors is followed by an increased level of self-e cacy, and vice versa. Moreover, an increased perception of the effectiveness of protective behaviors is associated with a higher motivation for protection, and vice versa. This is consistent with Zare Sakhvidi et al. (21).
The present research also showed that COVID-19 preventive behavior was signi cantly and positively correlated with protection motivation. This nding diverged from that of Kaviani et al. (36).
Variables subsumed under coping appraisals were the strongest predictors of protective behaviors in the target population. Self-e cacy was the strongest predictor of the behavior. Only next was the response e cacy, similar to several other studies(21, 37) .Those enjoying a higher level of self-e cacy perceive themselves capable of making achievements. Thus, they show a tendency to preventive behaviors of the disease. Both response e cacy and self-e cacy are subsumed under coping appraisal in PMT. Response e cacy deals with the fact that the suggested coping response may be effective in lowering threat. Selfe cacy deals with the perceived capability of making a coping response. Regression coe cients of these two constructs show that a higher level of self-e cacy and response e cacy can help increase motivation for preventing COVID-19. These ndings are consistent with previous studies which also adopted PMT as their theoretical framework.
Available coping strategies to participants are more effective in the frequency of adopting protective behaviors. High response e cacy strengthens their intention of self-protection and belief in the effectiveness of the protective behavior (38,39). The effect size of coping variables or threat depends to some extent on the nature of the health issue. In the light of the related literature, threat appraisal variables showed to be the strongest predictor of cancer preventive behaviors. Yet, for smoking, coping appraisals were the strongest predictor of preventive behaviors (29). The high education level of the participants can be part of the reason for this (40). In the present research, the majority of participants were between 31 and 40 years old. This can be the main reason for the unprecedented effect of perceived severity and vulnerability in PMT and preventive behaviors. Higher self-e cacy can improve preventive behaviors in vulnerable population (41,42). Health promotion programs with the aim of improving selfe cacy in COVID-19 preventive behaviors might improve behaviors and promote health at a social level.
Maladaptive behavior reward and fear were negatively correlated with preventive behaviors. Thus, it can be concluded that the higher the level of fear and maladaptive behavior reward, the lower the probability of adopting protective behaviors. Therefore, awareness of the rewards of maladaptive behaviors is more important than the perceived cost of healthy behaviors. This nding is consistent with the use of condom in some other research (43). However, in an investigation of cancer preventive behaviors at workplace, perceived costs showed to be more important than perceived reward (21).
In the present research, protective behaviors showed to be at a desirable level, which is consistent with some research by Barati et al about COVID-19 protective behaviors among hospital staff (34). Moreover, in the present research among preventive behaviors, avoiding unnecessary travel was found to be the most prevalent protective behavior, which seems to result from limits set by social distancing. Avoiding kissing/shaking hands and wearing masks ranked second and third among protective behaviors. In another study in Saudi Arabia on MERS, hand-wash showed to be the commonest protective behavior along with wearing masks while coughing/sneezing (44). In a study in China, the most prevalent protective behaviors included avoiding crowds, going out less frequently, wearing a mask while going out, and not entering closed and crowded places (45). Considering the fact that COVID-19 is a new-coming disease and has a icted all countries all over the world, the information sources are constantly being updated and, thus, adopting protective behaviors at an acceptable level is not far-fetched.
As regular washing of hands is the foremost protective behavior, yet is less adopted by the target population, it seems that more extensive awareness-raising and relevant education is of an incomparable importance.
Behavioral change is a process and it takes time to initiate and develop a new behavior. According to the theory adopted in this research, coping appraisal responses that lead to protection motivation are made after the threat appraisal process. That is because a threat needs to be identi ed before the appraisal of coping strategies to choose from. Evidently, before things get uncontrollable, attempts should be made to improve public awareness, create mutual trust, promote effective coping responses and contribute to the achievement of plans.

Strengths of study:
The present research is among the very few cases of applying health promotion models to explore preventive behaviors of COVID-19. This research was conducted with a large sample and within the shortest time possible.

Limitations of study:
Online surveys can be only used by those able to read and write and having access to the net. The majority of participants in this research were young urban residents. Thus, the generalizability of ndings is limited.

Conclusion
The present ndings revealed that the reward of maladaptive behaviors and fear were the negative predictors of protective behaviors. Response e cacy and self-e cacy were the positive predictors of protective behaviors. The effect of self-e cacy was the strongest. The participants' score for COVID-19 preventive behaviors was above average. This can be due to adequate awareness-raising and management of the issue. Among preventive behaviors, avoiding unnecessary travel, avoiding kissing/shaking hands and wearing masks were the most prevalent. The present ndings can illuminate policy-making in the target population.

Declarations
Ethics approval and consent to participate The study was approved by the ethics committee of Hormozgan University of Medical Sciences (ethical code: IR.HUMS.REC.1398.479). The ethics committee has approved to conduct the study online and obtain online consent. All participants who provided online consent for the study, were assured that participation was voluntary, they could leave the study at any time, the data were anonymised, securely stored and analyzed for publication. People under 16 years of age have completed the questionnaires with parental consent.

Consent for publication
Not applicable.

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

Competing interests
The authors declare that they have no competing interests.

Funding
This study received nancial support from deputy of Research, Hormozgan University of Medical Sciences (grant no.980470). Study design, data collection and analysis processes were developed independently of the funder.
Authors' contributions TA contributed to the study design, SHM analyzed data, RER wrote the manuscript AND conducted data collection, NSH and MHA conceived the study, developed the study protocol and questionnaires, and supervised data collection, FN conducted data collection, HKT Design online questionnaire. All authors have read and approved the manuscript.

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