Health Literacy and Preventive Behaviors Modify the Association between Pre-Existing Health Conditions and Suspected COVID-19 Symptoms: A Multi-Institutional Survey

People with pre-existing health conditions (PEHC) are vulnerable to viral infection while health literacy (HL) and preventive behaviors (PB) have been shown to benefit people during the COVID-19 pandemic. The aim of this study was to examine the association between PEHC and suspected COVID-19 symptoms (S-COVID-19-S), and to investigate the modification effect of HL and PB. A cross-sectional study was conducted on 8291 participants visiting outpatient departments at 18 hospitals and health centers across Vietnam from 14 February to 31 May 2020. Data were collected regarding participant’s characteristics, HL, PB, PEHC, and S-COVID-19-S. Regression models were used for analyzing the associations. Results showed that people with PEHC had a 3.38 times higher likelihood of having S-COVID-19-S (odds ratio, OR, 3.38; 95% confidence interval, 95% CI, 3.01, 3.79; p < 0.001). In comparison to participants without PEHC and with the lowest HL score, those with PEHC and one HL score increment had a 7% lower likelihood of having S-COVID-19-S (OR, 0.93; 95% CI, 0.92, 0.94; p < 0.001). In comparison to participants without PEHC and not adhering to mask wearing, those with PEHC and adhering to mask wearing had a 77% lower likelihood of having S-COVID-19-S (OR, 0.23; 95% CI, 0.16, 0.32; p < 0.001). Higher HL and adherence to mask wearing can protect people from having S-COVID-19-S, especially in those with PEHC.

Coronavirus has been indicated to be transmitted via the contact with virus-containing secretions expelled through expiratory human activities of infected persons such as breathing, talking, sneezing, coughing, and singing [12,13]. Thus, it is a matter of great concern and necessary to explore preventive factors that help to reduce the occurrence of viral infection.
Health literacy (HL) is defined as individuals' perceptions, knowledge, and ability to understand, access, appraise, and apply health information into disease prevention and treatment in their life course in order to improve their quality of life [14]. Low HL is associated with worse health decisions and outcomes [15][16][17]. People with low HL have difficulties communicating effectively with health care providers [18]. This requires more efforts from physicians to communicate in colloquial language to promote patients' involvement in shared decision-making [19]. People with high HL have been shown to be more likely to adopt protective methods and have the preparedness to get protected from the COVID-19 infection [15].
Preventive behaviors such as mask wearing, hand washing, and physical distancing have been recommended by World Health Organization (WHO) [12] and Centers for Disease Control and Prevention (CDC) [20] as effective personal methods to protect people from COVID-19 infection. These preventive behaviors have been indicated to significantly reduce the risk of viral transmission and infection in previous studies [21][22][23]. A systematic review and meta-analysis of 172 observational studies across 16 countries and six continents supports the physical distancing and face masking that prove to be beneficial for preventing the infection [24]. In addition, wearing mask and washing hands with soap helps reduce the transmission of viruses [25].
Health literacy and preventive behaviors have shown the protective effects in containing the COVID-19 pandemic. However, their roles on controlling the infection in people with pre-existing health conditions remain to be explored. Therefore, in this study, we examined the association between pre-existing health conditions (PEHC) and suspected COVID-19 symptoms (S-COVID-19-S), and we investigated the effect modification of health literacy (HL) and preventive behaviors (PB) on the association between PEHC and S-COVID-19-S among outpatient department visitors in 18 hospitals and health centers across Vietnam during the pandemic.

Study Design
A cross-sectional study was conducted on adults visiting the outpatient department (OPD) at 18 hospitals and health centers across Vietnam from 14 February to 31 May 2020. The study was reviewed and approved by the Institutional Ethical Review Committee of Hanoi University of Public Health, Vietnam (IRB No. 029/2020/YTCC-HD3 for the first period from 14 February to 31 March 2020, and IRB No. 133/2020/YTCC-HD3 for the second period from 1 April to 31 May 2020).

Study Sample
Adults who visited the OPD, aged from 18 to 85 years old with Vietnamese listening and reading proficiency, were invited to join our survey. We excluded individuals with any emergency condition (e.g., positive with COVID-19, injuries from traffic accident, appendicitis, etc.), having a psychotic disorder, dementia, or blindness. The data collection procedures were described in a previous publication [26].
A total of 8291 participants from 18

Participant Characteristics
Participants were asked for socio-demographic information, including age (18-39, 40-64, 65-80 years old), gender (women, men), education attainment (junior high school or below, senior high school, college/university level or above), occupation (employed, own business, others), ability to pay for medication (very or fairly difficult, very or fairly easy to pay), and social status (low, middle and high). Body height (cm) and weight (kg) of participants was measured. Body mass index was calculated as (weight (kg))/(height (m)) 2 .

Health Literacy
The short version of the health literacy questionnaire (HLS-SF12) consisting of 12 items was used to assess the health literacy of participants in our study. This health literacy questionnaire has been validated and used for the general population (aged ≥ 15 years, both sexes) in Asia [27], including Vietnam [28][29][30][31][32][33][34]. Participants responded to each item on 4-point Likert scales from 1 = very difficult to 4 = very easy. The HL index was standardized to a unified metric from 0 to 50, with a higher score that represents better health literacy, using the following Formula (1): where Index is the calculated HL index, Mean is the mean of all participating items for each individual, 1 is the possible minimum value of the mean (leading to a minimum value of the index equals 0), 3 is the range of the mean, and 50 is the chosen maximum value of the new metric. A higher HL index indicates a better HL level [27,35].

Preventive Behaviors
Participants were asked about their performance of preventive behaviors, including washing hands with soap or sanitizer when touching objects suspected to be contaminated or after going outside [20,36], wearing mask when going outside [20,36], and performing physical distancing (at least 2 m or 6 feet to people around) [37] since the COVID-19 outbreak started from the 23 January 2020 in Vietnam. The frequency of performing each behavior was rated on five levels (never, rarely, occasionally, often, and always/usually). Always/usually performing each preventive behavior during the COVID-19 outbreak was defined as adherence to that behavior [20]. Not always/usually (never, rarely, occasionally, or often) performing a preventive behavior was considered as non-adherence to that behavior.

Pre-Existing Health Condition
Participants self-reported their pre-existing health condition (PEHC). The PEHC were assessed using the Charlson Comorbidity Index items [38], including myocardial infarction, congestive heart failure, peripheral disease, cerebrovascular disease, chronic pulmonary disease, diabetes without complications, depression, taking anticoagulants (for cardiovascular disease, thrombosis), hemiplegia, diabetes with complications, moderate or severe renal disease, tumor without metastasis, moderate or severe liver disease, solid metastatic tumor, and HIV/AIDS. Participants with one or more of these conditions when presented at the OPD were classified as having PEHC.

Suspected COVID-19 Symptoms
Participants reported their health problems that are similar symptoms of COVID-19, including fever, cough, dyspnea, myalgia, fatigue, sputum production, confusion, headache, sore throat, rhinorrhea, chest pain, hemoptysis, diarrhea, and nausea/vomiting [39]. Individuals having one of those symptoms when visiting the OPD were classified as having suspected COVID-19 symptoms or symptoms like COVID-19 [33]. This did not imply that participants were suspected of having COVID-19.

Statistical Analysis
Firstly, the Chi-square test was used to compare the proportion of S-COVID-19-S by different categorical variables, including participant characteristics (age, gender, marital status, education, occupation, social status, ability to pay medication, BMI), PEHC, and PB (mask wearing, hand washing, physical distancing) in studied participants. The analysis of variance (ANOVA) test was used to examine the distribution of continuous variables, including the HL index. Secondly, unadjusted and adjusted logistic regression models were performed to explore the associations of PEHC, PB, and HL with S-COVID-19-S. The factors significantly associated with having S-COVID-19-S at p < 0.05 via bivariate logistic regression were considered as confounding factors and were adjusted in the multivariate model. To avoid multicollinearity, the correlations between covariates were examined. If the Spearman's rho ≥ 0.3 exists between covariates, one representative variable was selected into the multivariate model. Finally, the interactions between HL, PB, and PEHC on having S-COVID-19-S were examined using bivariate and multivariate logistic regression models. All statistical analysis was performed by SPSS Version 20.0 (IBM Corp, Armonk, NY, USA), and a p < 0.05 was set as significance.

Characteristics of Studied Participants
Out of all participants, the proportions of PEHC, adherence to mask wearing, hand washing, and social distancing were 22.6%, 44.2%, 24.3%, and 20.6%, respectively. The proportion of S-COVID-19-S was 37.7% (3129/8291). This proportion was significantly differed by participants' age, marital status, education, occupation, ability to pay, social status, and preventive behaviors. People with S-COVID-19-S had a lower health literacy score than those without S-COVID-19-S (Table 1). Table 1. Characteristics of studied participants (N = 8291). Age

Effect Modifications of Health Literacy and Preventive Behaviors
In comparison to participants without PEHC and the lowest HL score, those with PEHC and lowest HL score had a 16.53 times higher likelihood of having S-COVID-19-S, while a 1-score increment of HL resulted in a 7% lower likelihood of having S-COVID-19-S (OR, 0.93; 95% CI, 0.92, 0.94; p < 0.001). Similarly, in comparison to participants without PEHC and not adhering to mask wearing, those with PEHC and not adhering to mask wearing had a 17.42 times higher likelihood of having S-COVID-19-S, while those with PEHC and adhering to mask wearing resulted in a 77% lower likelihood of having S-COVID-19-S (OR, 0.23; 95% CI, 0.16, 0.32; p < 0.001; Table 3). The modifying effect of adherence to hand washing or social distancing to lower the likelihood of having S-COVID-19-S was not found (p > 0.05; Table 3).

Discussion
In this study, we found that participants with pre-existing health conditions (PEHC) had a significantly higher risk of having S-COVID-19-S than ones without PEHC. Participants, who adhered to performing preventive behaviors and who had higher health literacy scores had a lower risk of having S-COVID-19-S compared to their counterparts. Even though we did not study the symptoms of COVID-19, the symptoms like COVID-19 could possibly help to explore the associations and raise the awareness of the public. A meta-analysis of 1558 patients with COVID-19 showed that patients with certain underlying medical conditions such as hypertension, diabetes, chronic obstructive pulmonary disease (COPD), cardiovascular disease (CVD), and cerebrovascular disease had a higher risk of having COVID-19 than those without these illnesses [6]. Besides, patients with comorbidities, such as diabetes mellitus or chronic obstructive pulmonary disease (COPD), were prone to develop more severe progression of COVID-19 infection [6,9].
Preventive behaviors such as wearing mask, washing hands with soap, and performing physical distancing were proven to be effective in reducing the COVID-19 transmission and infection. A systematic review collecting 172 observational studies across 16 countries and six continents proved that lengthened physical distancing of at least two meters in public places has more effectively reduced the transmission of SARS-CoV-2 and beta coronaviruses (SARS-CoV or MERS-CoV) compared to a closer distance that is less than two meters [24]. Wearing mask and washing hands limit the contact with virus-containing secretion from infected people, thus wearing mask and washing hands with soap have been strongly recommended in the COVID-19 pandemic to reduce the spread of SARS-CoV-2 [40]. Mask wearing and hand washing have also been shown to be preventive measures to reduce the transmission of viruses during the severe acute respiratory syndrome (SARS) [25,41] and novel influenza A (H1N1) pandemic [42].
Regarding health literacy (HL), a study of behavioral reactions to the H1N1 influenza pandemic showed that individuals who had perceptions of the effectiveness of preventive behaviors, such as washing hands with warm water and soap, using a napkin when sneezing, and avoidance responses such as avoiding crowd places and public transportation were more likely have a reduced risk of having the viral infection [43]. The possible explanation was that individuals with a high HL level possibly had more adequate and appropriate information and health knowledge [44][45][46][47], which assisted them to make better health decisions and adopt preventive methods to protect themselves from the disease during the pandemic [48]. Also, HL was indicated to be positively associated with better attitudes and preparedness for the outbreak during the early onset of COVID-19 [15]. In addition, people with higher health literacy had better health-related behaviors that may be a benefit to their health [29].
Importantly, we found that health literacy and preventive behaviors (mask wearing, hand washing, and physical distancing) have modification effects on reducing the risk of having S-COVID-19-S in patients with PEHC. The finding is similar with a previous study that higher health literacy was found to be associated with better adherence to infection prevention and control behaviors, and lower risk of having S-COVID-19-S in health care workers [30]. Health literacy was previously suggested as the driving factor between socioeconomic status and health outcomes [49] and between education level and health status [44]. Noticeably, mask wearing was found to be the most effective modifier to lower the risk of having S-COVID-19-S in patients with underlying medical condition (up to 77% reduction). Even though hand washing and physical distancing were shown to insignificantly modify the risk of having S-COVID-19-S in highly vulnerable people after being adjusted for socio-demographic factors, these preventive behaviors were still shown to have a potentially protective effect to protect people with PEHC from having suspected COVID-19 symptoms.
Our study is the first study to investigate modification effects of preventive factors (such as health literacy, preventive behaviors) on the risk of having S-COVID-19-S in patients with PEHC. However, in our cross-sectional study, it is not possible to explore the causal relationship between underlying medical condition, health literacy, and preventive behaviors and the risk of having S-COVID-19-S in the study. Besides, suspected COVID-19 symptoms and underlying medical condition of participants were self-reported. Although self-reporting is a common approach for collecting data in epidemiological and medical studies, this method may lead to recall bias due to recall error of the past events of participants [50]. Therefore, the findings should be interpreted with caution. Fortunately, a large sample size allows us to examine the associations and interactions that help to raise the phenomenon for the future research. Another limitation was that we cannot identify patients with symptoms of COVID-19 or symptoms like COVID-19 by asking them. Fortunately, there were no new confirmed cases in the studied hospitals and health centers during the data collection period [51].
Despite of remaining limitations, our study helps to explore the modifying effects of preventive factors (health literacy and preventive behaviors) mitigating the negative association between having an underlying medical condition and the risk of having S-COVID-19-S. In the limited scope of one study, we assessed the modifying effect of each preventive behavior separately, but not the combined effects of these behaviors. This approach, however, helps us to figure out the differences of the protective effect of each behavior in people with pre-existing health conditions, and helps to find out the most significantly effective modifier. In further studies, exploring the factors associated with the adherence to preventive behaviors, and the combined effects of preventive behaviors on having suspected COVID-19 symptoms is essential to help researchers and practitioners to have oversight of people's awareness and knowledge during the viral pandemic.

Conclusions
People with pre-existing health conditions had a higher likelihood of having S-COVID-19-S. Fortunately, higher health literacy and better adherence to mask wearing were associated with lower likelihood of having suspected COVID-19 symptoms, especially in those with pre-existing health conditions. Strategic public health approaches are needed to promote health literacy and infection preventive behaviors (e.g., mask wearing, hand washing, and social distancing) that may help to contain the pandemic.  Table S1. Association between socio-demographic characteristics and having S-Covid-19-S in studied participants via logistic regression (N = 8291). Table S2. Spearman correlation among covariates.