Factors Associated with Risk Behaviors Towards Hepatitis B Among Migrant Workers: A Cross-sectional Study Based on Theory of Planned Behavior

Background: Rural-to-urban migrant workers are susceptible to hepatitis B (HB) because they lack self-protection awareness and social support. The present study aimed to investigate the current status of risk behaviors (RB) regarding HB among migrant workers and the inuencing factors dened by the Theory of Planned Behavior (TPB). Methods: A cross-sectional study was performed by two-stage cluster sampling from June to December 2018 in Chongqing, China. Logistic regression was adopted to explore factors associated with HB-related RB and behavioral intention (BI). Results: A total of 1299 migrant workers were recruited in the surveys, among whom 29.56% respondents have performed HB-related RB and 85.53% had the BI. 58.19%of respondents having sexual activities never wore a condom. The risk scores of attitudes towards behavior (AB), subjective norms (SN), experience of behavior (EB) and regret feeling (RF) were positively associated with BI, while the sores of AB, EB and BI were positively associated with RB. Conclusions: A considerable proportion of migrant workers have had HB-related risk behaviors or had the behavioral intention. Theory-grounded education, focused on the identied TPB variables, may play a signicant role in improving the cognition and behaviors towards HB.


Introduction
Hepatitis B (HB), a leading cause of human's liver cirrhosis and liver cancer, remains a major threat to global public health, particularly in the Asian-Paci c and sub-Saharan Africa regions [1]. There are nearly 2.57 billion hepatitis B surface antigen (HBsAg)-positive people, and 887,000 HB cases dying of liver damage and complications worldwide according to the data from the World Health Organization (WHO) in 2015,the WHO also urged countries to invest in eliminating hepatitis, especially low-and middle-income countries [2].As one of the countries with high HB endemicity reported by WHO, China has more than 90 million HBpatients and 100,000 annually new HBV infections [2]According to the National Hepatitis Seroepidemiology Survey in 2006,although the (HBsAg)-positive rate has already declined to 2.08%among Chinese aged from 1 to14,the (HBsAg)-positive rate in populations aged between 15 to 59 was 8.57% [3]. Nevertheless, the overall (HBsAg)-positive rate was still over 8% for Chinese adults in 2016 [4].
Hepatitis B virus (HBV)is mostly transmitted through blood or body uids of infected individuals [4], and having sex contact and using contaminated injector are known as the major routes for HBV infection among adults [5,6]. Therefore, risk behaviors related to HB refer to unprotected sex, whoring, multi-partner sex, sharing injectors for drug use and so on [6,7].
Rural-to-urban migrant workers, almost accounting for the whole of internal migrant population in China, were commonly de ned as people who left rural area where they used to live for towns or cities to seek better employment opportunities and higher-incomes [8].The latest Migrant Workers Monitoring Investigation Report of China showed that the number of migrant workers has increased to 288.36 million by 2018 [9]. Most migrant workers are basically low educated, engage in low-incomes and low-skilled jobs, and most of them live a stressful life [10]. They are mostly at sexually active ages but commonly being single or away from their spouses [10]. And the majority of them have poor sex-related to knowledge and little self-protection awareness, insu cient social support and limited access to health care [8]. Therefore, migrant workers are more likely to have HB-related behavior mentioned above, in which case the risk of HBV infection will be increased among them. Previous surveys showed that construction workers had unprotected sex, commercial sex, casual sex, blood selling and drug use, accounting for 14.9%, 7.9%, 8.4% and 0.7% respectively [11,12]And previous studies also found that migrant workers had relatively higher susceptibility to HBV infection compared with non-migrants and local dwellers [13].As summarized in a systematic review and meta-analysis by Zou et.al, the prevalence of viral hepatitis among rural to urban migrant workers was 0.45%and 38.5 higher odds of infection than general populations in China [13].In addition, the frequent ow of migrant workers is possible to enlarge HBV spread to general population facilitate the regional transmission across China [14].It is therefore important to know the status of HB-related risk behaviors of migrant workers and have insight into the determinants.
One of the most recognized theories to understand behaviors is the theory of planned behavior (TPB, Ajzen, 1991). TPB states three conceptual modules determine behavior:(1)attitude toward the behavior (AB), refers to favorable/unfavorable appraisal of the behavior;(2)subjective norms (SN), refers to perception of social pressure to perform/not perform the behavior;(3) perceived behavioral control (PBC), refers to the perceived ease or di culty to perform the behavior. TPB postulates that AB, SN, and PBC lead to the formation of a behavioral intention. The more favorable the AB and SN are, and the greater the PBC is, the stronger would be a person's intention to perform the behavior in question, and the higher possibility that the person would go into action [15,16]In addition, Previous studies also argued that some independent variables, like experience and/or regret of performing a behavior, would directly or indirectly in uence the behavioral intention and worth being taken into account to improve the TPB framework [17,18]. To address the issue of health behaviors, TPB variables has been widely adopted to interpret HIV/AIDS-related behaviors, particularly for the highly susceptible groups such as commercial sex workers and men who have sex with men (MSM) [19,20] And previously TPB was also applied to explore determinants on smoking, drinking and health-seeking behaviors of people [21,22]. Nevertheless, the study that examined HB-related risk behaviors of rural-to-urban migrants was rarely reported.
The present study was the rst attempt to (1) understand the status of risk behaviors related to HB that performed by migrant workers and (2) detect what and how the factors motivate and in uence the behavioral intention and practical behaviors on the basis of TPB theory.

Study sites and sampling
Chongqing, located in southwestern China, is the largest municipality directly under the Chinese central government. It is regarded as "miniature China" because its geographic characteristics, urban-rural distribution and social-economic pro le are close to the national average [23]. The city area of Chongqing, one of the busiest places Chinese migrant workers keep owing in, consists of nine administrative districts with an area of 5472.82 Km 2 and a population of 8.65 million, among which immigrants take about 23.5% [23,24].The (HBsAg)-positive rate among migrant workers in the city area of Chongqing was 8.6% [25].And there were nearly 26,000 new infections of viral hepatitis in 2016 in terms of the Health Statistic Yearbook of Chongqing [24]. Two-stage strati ed cluster sampling was performed to recruit participants between June 2018 and January2019. Firstly, nine districts of Chongqing's city area were categorized into three strati cations-more developed, medium developed and less developed-by economic development, geographic background and population density, and then three districts were randomly selected to represent for each strati cation respectively. Secondly, two enterprises were purposively sampled in each district, including the manufacturing, construction, wholesale and retail industry, transportation industry, hotel and catering industry and community services. Local Center for Disease Control and Prevention, Health Supervision Institute, and Urban-rural Development Committee helped to coordinate with sampled units. The inclusion criteria of target individuals were (1) 18 years and above, (2) having been in the Chongqing's city area for at least six months, (3) not registered as Chongqing urban resident, (4) engaging mainly in the secondary or tertiary industry. Given a considerable part of migrant workers were low-educated or illiterate, trained investigators assisted to explain the questions in mandarin and Chongqing dialect. Subjects who were unable to understand the questionnaire items or refused to be surveyed were excluded. Participants were reassured that all responses would be anonymous and written informed consent was granted by each respondent. Surveys were conducted in relatively undisturbed environment and peak working hours were avoided to guarantee the quality of surveys. Completeness of each questionnaire was double-checked by investigators.

Study Instrument
The questionnaire was constructed basically in terms of the TPB on health-related to behaviors and perceptions in published studies [19,20]. Experts in epidemiology and hepatology were involved to modify the logic and wordings of the questionnaire. A pilot survey was conducted with 90 migrant workers in nearby restaurants and construction sites. The nal version consisted of nine modules with Cronbach's Alpha coe cients ranging from 0.759 to 0.968, and con rmatory factor analysis (CFA) showed a good tting degree (χ 2 /df = 1.859, RMSEA = 0.039, GFI = 0.900, AGFI = 0.883, CFI = 0.969, IFI = 0.969).TPB variables was assessed using a ve-point semantic differential scale, and the average item score for each module was computed to be used as the scale score. The higher the scores are, the more risky the respondents would be. The de nition of each modules and variable scales were shown in Table 1.The items of different dimensions and positive and negative items were sorted at intervals to prevent the respondents from having a Socially Desirable Responding or picking an initial scale for each item. Unprotected sexual behaviors (never or rarely wear a condom), casual sexual behavior, commercial sexual behavior, homosexual behavior, IDUs, blood selling /transfusion illegally, and sharing toothbrushes/towels. RB questions were valued by 5-point semanticdifferential scale: from "never" (1) to "often" (5). An example is "Have you had casual sex in the last six months?" Behavioral intention (BI) BI refers to person's readiness to perform a given behavior. BI questions were measured by 5-point semantic differential scale: from"absolutely impossible"(1)to "absolutely possible"(5).An example is: "Is it possible for you to have commercial sexual behavior? " Attitudes toward a behavior (AB) AB refers to the degree to which performance of the risk behavior is positively or negatively valued.
AB questions were measured by 5-point semantic differential scale: from "very much unsafe"(1) to"very much safe"(5). An example is: "Do you think it is safe to have commercial sex?" SN refers the perceptions of social pressure to engage or not to engage in risk behaviors.
SN questions were measured by5ponitsemantic differential scale: from"strongly agree"(1) to "strongly disagree"(5). An example is: "Do you agree with your friends if they advise you not to actany of the above behaviors?" PBC refers a person feels she/he is in control of a given risk behaviors. PBC questions were measured by 5-point semantic differentialscale:from"verymuchable"(1) "very much unable"(5). An example is: "Are you able to decide whether or not have casual sex by yourself?"

Experience of behavior (EB)
EB was referred the safe or unsafe perception towards related to risk behaviors. EB questions were measured by 5-point semantic differential scale: from stronglydisagree(1)tostrongly agree(5). An example is: "I had behaviors mentioned above to meet physiological needs." Regret feeling (RF) RF refers to an individual's psychological regret and shame on HB risk behaviors that he ever did and/or he plans to have. And RF includes retrospective regretand prospective regret.
RF questions were measured by 5-point semantic differential scale: from "very much"(1) to"not at all"(5). Examplesare: "Did you regret after having commercial sex behavior?" and "Will you regret if you had commercial sex?"

Data analysis
Survey data were double-checked and entered into a database by Epidata 3.1 (The EpiData Association, Odense, Denmark).All data were analyzed using IBM SPSS 22.0 (SPSS Institute, Chicago, USA). Categorical data were assessed by the number and proportion of responders. Continuous variables of sociodemographics, such as age, years of being a migrant worker, working hours per day, were converted into categorical variables and then described by number and proportion of responders. Knowledge levels of responders were divided into poor, medium and good by scores less than 7, 8 ~ 10, 11 ~ 13respectively [26]. TPB variable were involved as continuous variable with average scores. With regard to the association analyses, independent variables were identi ed as the variables in the modules of socio-demographics, HB knowledge and TPB framework, and the dependent variables were identi ed as behavioral intention and HB-related risk behaviors. For the dependent variables, behavioral intention was dichotomized into "never had an intent" and "had an intent for at least a behavior" and HB-related risk behaviors were dichotomized into "never had risk behavior" and "had at least once risk behavior". Univariate analyses were performed with independent variables of socio-demographic and HB knowledge with the two dependent variables by Chi-square tests. And independent variables with P-values less than 0.10 in the univariate analyses were subsequently inputted into the logistic regression models (α = 0.05, β = 0.10) along with variables in TPB modules to detect the possible in uence factors for the two dependent variables. Binary logistic regressions were tted with the dependent variables by entering three blocks of variables: block I, socio-demographics and knowledge level; block II, TPB variables; block III, demographics, knowledge level and TPB variables. BI was not included in the block II and block III when it was regarded as dependent variable. Dummy variables were coded for variables with more than two values, and variables stepwise entered into the models. Adjusted odds ratios (ORs) and 95% con dence intervals (CIs) were computed, and P-values less than 0.05 were deemed statistically signi cant.

In uencing factors associated with behavioral intention
As be positivelyassociated with the behavioral intention (Table 5).  (Table 5).

Discussion
The Action Plan for Prevention and Treatment of Viral Hepatitis in China (2017-2020) has underlined the signi cance to prevent and control of viral hepatitis,particularly for those who are susceptible to the disease and possible to spread the transmission [27]. Rural-to-urban migrants may play a crucial role in the cross-regional diffusion of HBV in China due to their frequently seasonal moves over workplaces and homelands [13]. Existing evidence showed that the proportion of migrant workers who performed risk behaviors related to sexually transmitted diseases (STDs) were relatively high [13]. Our ndings, verifying this point, detected that about one-third of participants had at least once of HB-related risk behaviors during in the last six months, although the proportion of migrant workers' causally extramarital sex (10.24%) and commercial sex (4.7%) were lower than those reported in studies conducted in Shanghai (15.22%) and Zhejiang (5.7%) [28].In addition, 2.80% of migrant workers had admitted to have homosexual behaviors, which reminds us to also pay attention because MSM have been referred to be highly risky for STDs [19].
Although the Chinese Ministry of Health has advocated condom use to prevent STD ssince 2006 [29],in our study a considerably big proportion of migrant workers (58.19%)never wore a condom when having any types of sexual behaviors. This nding was in line with the high rate of condom nonuse for migrant workers in Hefei, China (52.68%%) [12].Regarding the potential barriers for condom use, having taken other contraception methods, uncomfortable to wear and no requirement by the partner took the top three positions. Those indicate that wearing a condom was only viewed as a contraception way but not understood as a protection from STDs by migrant workers, and the pleasant sensation outdid the perception of disease infection. All those have pointed to the weak knowledge and awareness of STDs among migrant workers, which is right consistent with the generally low level of HB knowledge for respondents in our study. In fact, protected sex with a condom has been proved to be signi cantly practical and cost-effective to prevent STD transmission [19],and therefore, extensive publicity for condom usage is in need to be strengthened among migrant workers. Meanwhile, vending machines for condoms (better for free) could be set near by migrant workers' living areas to cope with the embarrassed feeling mention by the respondents [30].
Similar as we though, there was a small part of respondents having a history of injector sharing for drug use (1.00%) and illegal blood selling/transfusion (1.23%), in line with studies with the migrant workers in eastern China [12,28]. Blood transmission, a signi cant route of HBV infection as well as sexual transmission, should not be neglected. In addition, one-fth of participants in our study were detected sharing personal hygiene products like toothbrushes and/or towels, which would amplifythe possibility of HBV infection through the broken skin [31]. Therefore, health educationtargeting those issues need to be enhanced with migrant workers.
Approximately one-third of migrant workers admitted having had HB-related risk behaviors,moreover, nearly 90% of respondents expressed that they would like to perform some risk behaviors sometime. This is a potential threat as behavioral intention indicates a possibility for a person to have actual behavior.
Therefore, educational intervention on altering behavior intention and self-protection cognition is crucial in addition to directly regulating on risk behaviors.
Similar with previous studies, logistic regressions suggested that migrant workers who were male, at younger age, with lower educational background and at lower knowledge level would be more intended to act HB-related risk behaviors [17]. Compared with females, most males are less perceived with disease risk and overestimate their own health status [32]. Compared with older ones, young people are right in a sexually active period and will probably have more sexual demands [28]. Compared with higher educated ones, people with lower education tend to be less cognitive for disease prevention [33].Compared with individuals at a higher knowledge level, those with weak HB knowledge may lack understanding of HB and be less aware of self-protection against the disease [32]. Consistent with the study on HIV-related behaviors in northwest Ethiopia, there was a positive association between drinking and performing risk behaviors. It is to some extent because drinking probably results in more opportunities for casual sex and unprotected sex for migrant workers [34].
As adjusted by socio-demographics, migrant workers scored higher points ofAB and SN were more intended to act risk behaviors, and those scored higher points of BI were more likely to have performed HB-related risk behaviors. All these can be exactly interpreted by the typical TPB framework -AB and SN, produced from behavioral beliefs and normative beliefs, will act on BI, and then work together with BI to trigger the behavior [16]. That is, if migrant workers have a more favorable attitude towards HB-related risk behaviors and less perceived social pressure upon acting the behaviors, they would be more intended to do so, and also be more likely put into practice [15] To strengthen the interpretability of the practical behaviors of migrant workers, two socio-psychological modules -EB and RF -were introduced into the typical TPB framework. As we expected, the two variables were positively associated with both behavioral intention and practical risk behaviors. Previous studies argued that daily decision making would be affected by the actual emotional experience, and successful implementing of risk behaviors in the past appears to render migrant workers more likely to repeat in the future [35]. Besides, the stronger regret feeling migrant workers have, the less intended they would be and also less likely to perform risk behaviors. It is because regret representing a negative consciousness and an emotional reaction to persons' intention or behaviors [17].Given migrant workers showed low perception and poor self-protection against HBV infection, health educational campaigns are necessary to improve their cognition and behaviors as mentioned above. Considering friends/family members, television/radio and internet /cell phone APPs are widely popular sources for migrant workers to gain health knowledge and information, peer education will work, and the health education through the combination of new and traditional media will be also encouraged. Besides, only half of migrant workers have inoculated HB vaccine in our study. It is partially because free HB immunization program is not offered to people aged above 15 currently in China[3]. Therefore, a sound compensation system is in expectation to provide extra nancial support for HB-susceptible adults including migrant workers to expand the coverage of the HB vaccine.
There were some limitations necessary to be noticed. Firstly, causal inference based on the associations observed in our study might be more or less limited as it was in cross-sectional study design. Secondly, selection bias, giving rise to an imbalance of occupation distribution between the sampled participants and the whole population of Chongqing's migrant workers, might be introduced as it was non-random sampling. In addition, report bias, mostly assumed as an underestimate of acting risk behaviors, might be inevitable due to the personal privacy and social desirability, although anonymity of respondents was reassured.

Conclusions
There was one-third of migrant workers having had HB-related risk behaviors and nearly 90% of migrant workers intended to act risk behaviors in the present study. The Condom was poorly used among migrant workers for STDs prevention. Migrant workers who were male, less educated, poorer in HB knowledge and at a younger age have the stronger intention of HB-related risk behaviors, while those who were drinking alcohol were more likely to have performed risk behaviors. As completed with innovative variables of EB and RF, the TPB framework played well in interpreting the in uencing factors, showing that migrant workers have a more positive attitude and less subject norm towards risk behaviors would be more intended and then be more likely to act. Accordingly, more attention should be paid to both improving the disease perception and self-protection awareness, as well as helping migrant workers to regulate their behaviors. Meanwhile, theory-grounded health educational interventions, targeting the key in uencing factors proposed by the analyses, are in need of peer education and the combination of both new and traditional media.