Hand washing and social distancing as mitigation strategies

Both hand washing and social distancing have been identified as non-pharmaceutical public health interventions to combat the spread of COVID-19. International studies have long shown that proper and frequent hand washing is a simple, effective health behavior that can mitigate the risk of acquiring several viral and bacterial infections transmitted through skin-to-skin contact, respiration, or food [9, 10]. Godoy and colleagues [11] found that frequent hand washing and the use of alcohol-based hand sanitizers were protective factors against influenza acquisition requiring hospitalization, and other scholars noted the utility of frequent hand washing with soap and water to protect against cholera in Nigeria [12]. Although prior research has demonstrated notable increases in perceived efficacy of hand washing to prevent communicable diseases [13], compliance with proper hand washing remains low [14–16]. As such, more investigations are needed to identify mechanisms that contribute to disparities of proper hand hygiene in local and cross-cultural contexts.

Like effective hand washing, social distancing (i.e., maintaining a distance of at least six feet, or two meters, from individuals with whom one does not live) has been shown to be effective against the spread against certain respiratory viral conditions. Recent studies in international contexts confirm that social distancing may be effective in mitigating COVID-19 risk [17–19]. In the Kingdom of Saudi Arabia, for example, researchers found that taking a multi-faceted and systemic approach to facilitating social distancing measures may potentially mitigate the spread of COVID-19, especially in communities and cultures that highly value interconnectedness as social norms [17]. To be sure, successful implementation of widespread social-distancing measures require coordinated, multicomponent efforts across multiple ecological levels and community sectors. Nevertheless, the extant literature provides consensus on the potentially mitigative effects of proper social distancing on attenuating the spread of COVID-19.

Sociodemographic determinants of social distancing and hand washing

To date, several studies have examined sociodemographic determinants of social distancing and hand washing in the COVID-19 context. For instance, older age is significantly associated with COVID-related preventive measures such as social distancing and hand washing [20–22]. Although age effects are typically measured in years, one study [20] found that the Baby Boomer generation (aged 56–74 years) endorsed greater social distancing than Millennials (aged 24–39 years). This approach adds a nuanced understanding of the relationship between age and COVID-related preventive measures, suggesting a potential generational effect. More broadly, the extant literature suggests that across international samples, age is a consistent predictor of COVID-related protective measures.

Similarly, investigations have demonstrated educational attainment and employment status impact social distancing and hand washing behaviors. Lüdecke and von dem Knesebeck [21] found that in a German sample, individuals with lower educational attainment were less likely to engage in social distancing or engage in appropriate hand washing behavior—although the observed effect sizes were small to negligible. Those findings were consistent with other work [23], which found that lower educational attainment and employment in agricultural and domestic labor (compared to business sector) predicted lower levels of social distancing and hand washing among Iranian adults.

Although research is consistent regarding the predictive effects of age, education, and employment on social distancing and hand washing, the same cannot be said for gender and urbanicity. Suen and colleagues [10], for example, found that women were more likely to perform proper hand hygiene techniques compared to men. Their findings were echoed in research among Bangladeshi men who have been found to have significantly greater odds of experiencing challenges with practicing COVID-related preventive behaviors [23]. However, in Saudi Arabian samples, men were more likely to wash their hands with soap and water than women [24]. Together, these studies point to inconsistent gender effects on hand washing behavior related to COVID-19 prevention.

Numerous scholars have highlighted the urban-rural disparities in COVID-19 morbidity and mortality [7, 8, 25]; however, limited work has examined urbanicity in relation to social distancing and hand washing. The extant literature suggests, similar to gender effects, that the evidence for the effect of urbanicity remains inconclusive. On the one hand, rural residents have shown to endorse lower levels of social distancing and hand washing [23], even though they present with greater knowledge of COVID-related prevention strategies [26]. On the other hand, urban dwellers have endorsed greater intent to practice social distancing and hand washing relative to their rural-dwelling counterparts [26]. Still, urban dwellers living in poverty may experience unique challenges in practicing social distancing and hand washing [27], suggesting that the syndemic nature of urbanicity and socioeconomic status may play a more substantial role in COVID-19 prevention than previously conceived. Taken together, the role of specific social determinants of health on social distancing and hand washing remains largely inconclusive. As such, the implementation of broader ecological approaches to mitigate COVID-19 risk is limited.

Instruments and measures

Social distancing was defined as engaging in socially distancing at least 6 feet (2 meters) from others, and adequate hand washing was defined as washing hands with soap and water for at least 20 seconds each time. Social distancing was measured using a COVID Protective Measures Scale developed by the author. The scale comprised four items (i.e., “I avoid large groups,” “I stay at least 2 meters (6 feet) away from others,” “I do not hug people or shake people’s hands,” “I wash my hands with soap and water for 20 seconds or more”) and was measured using Likert scaling (i.e., 1 = “strongly disagree” and 5 = “strongly agree”). Likewise, hand washing was measured on a Likert scale (1 = “strongly disagree” and 5 = “strongly agree”) using the item “I wash my hands with soap and water for 20 seconds or more.” Overall, the four items showed good internal consistency in the sample (Cronbach’s α = .83), and social distancing questions showed similar internal consistency (Cronbach’s α = .82). All items were translated and back-translated into Italian, Spanish, Arabic, and Hindi for participants residing in Italy, Spain, the Kingdom of Saudi Arabia, and India, respectively. The English version of the scale, along with the translations, are provided as S1 File.

Sexual orientation was a categorical variable measured with the item "Which of the following best describes your sexual orientation?" Response options included heterosexual, lesbian, gay, bisexual, and pansexual or other. Participants self-reported their gender identity as one of the following categories: cisgender man, cisgender woman, transgender man, transgender woman, nonbinary or gender fluid, and other gender identity. They were also given the option to select “I don’t know,” to fill in their gender identity, or decline to answer. Because only 27 participants identified as transgender, nonbinary, genderfluid, other gender identity, or unknown gender identity, we did not include those individuals in the final analysis due to inadequate sample size. Other categorical predictor variables included education, employment status; and age in years was entered as a continuous variable. Locality was a categorical variable measured using the item "What type of community best describes where you live?" Response options included: (a) “I live in a large (or major) city,” (b) “I live near a large (or major) city,” (c) “I live in a rural town or village,” and (d) “I decline to answer.”

Data analysis

The full data set is available through the Open Science Foundation (https://osf.io/gdy3z/). Data were analyzed using Stata/SE version 15.1 (StataCorp, College Station, TX). Descriptive statistics were used to show sociodemographic characteristics of respondents overall and stratified by country. Chi-square tests were used to examine differences by country, and one-way ANOVA was used to identify whether significant age differences existed between countries. Statistically significant variables (i.e., p < .05) were screened and included in the multivariable linear regression analysis. Strengths of associations were assessed using both standardized and unstandardized estimates as well as standard errors, which are presented in the Results.

Ethical considerations

The Michigan State University Institutional Review Board determined the study was exempt from review. All approvals and permits to conduct human subjects research in each participating country were obtained through Qualtrics as part of their process for administering research surveys to international participants. Prior to agreeing to participate, all respondents were provided with detailed informed consent forms regarding the purpose of the study, their responsibilities, risks and benefits of participating, and the privacy and protection of their data. All procedures for this study complied with university, local, state, federal, and international regulations on research with human participants. Each participant had the option of downloading a PDF version of the detailed informed consent form in their language for their records.

Social distancing

Table 2 shows the unstandardized and standardized effects of sociodemographic factors on practicing COVID-related social distancing in the overall sample and stratified by country. In the overall sample, age was significantly associated with social distancing (b = 0.03, β = 0.16, p < .001) with a small effect size. Country-stratified analyses showed similar effects of age in India (b = 0.04, β = 0.15, p = .002), Saudi Arabia (b = 0.04, β = 0.22, p < .001), Spain (b = 0.07, β = -0.28, p < .001), and Italy (b = 0.03, β = 0.13, p = .023). Women were also more likely than men to socially distance (b = -1.04, β = -0.16, p < .001), with greater effect sizes found in Italy (b = -1.04, β = -0.16, p < .001) and Spain (b = -1.04, β = -0.16, p < .001). No significant effects of gender on social distancing were found in the United States, Saudi Arabia, or India.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Unstandardized and standardized multivariable linear regression weights for social distancing, stratified by country.

https://doi.org/10.1371/journal.pone.0264820.t002

Differences by sexual orientation were also found but only in the overall sample. Specifically, lesbian and gay participants endorsed less social distancing compared to their heterosexual peers (b = -0.75, β = -0.05, p < .001). No significant differences were found in the overall sample with respect to employment status; in Saudi Arabia, however, both college graduates (b = -0.62, β = -0.12, p = .029) and those with a high school education or less (b = -1.10, β = -0.15, p = .009) endorsed less social distancing compared to postgraduates. Compared to adults who completed postgraduate school, those who completed some college (b = -0.38, β = -0.05, p = .003) and high school or less (b = -0.45, β = -0.07, p = .003) were less likely to socially distance from their peers.

Although we found no significant effects of employment status in the overall sample, stratified analyses showed that retired adults in Spain were significantly less likely to engage in social distancing compared to those employed full-time (b = -2.74, β = -0.17, p = .006). In the United States, part-time employees were also significantly less likely to engage in social distancing compared to full-time workers (b = -0.76, β = -0.11, p = .041), although the effects were smaller than those for retired Spanish respondents. We also found that both suburban (b = -0.38, β = -0.01, p = .005) and rural (b = -0.72, β = -0.11, p < .001) respondents were significantly less likely to practice social distancing compared to their urban counterparts, with stronger effects shown for rural respondents. Similar effects were shown among Indian respondents living in suburban (b = -0.84, β = -11, p = .034) and rural (b = -1.45, β = -0.14, p = .020) settings compared to their urban-dwelling peers. Furthermore, rural residents in Saudi Arabia were significantly less likely to practice social distancing than their counterparts in urban settings (b = -1.45, β = -0.13, p = .013).

To assess the specific effects of sexual orientation and locality on social distancing, we conducted a three-block hierarchical linear regression model (Table 3). In the first model (R² = 0.002, p = .023), sexual minority status was associated with less social distancing (b = -0.16, β = -0.07, p = .023). The addition of urbanicity and country of residence in Model 2 added 2.2% (p < .001) to the variance in Model 1, bringing the total variance explained to 2.5% (p < .001). In Model 2, sexual minority status remained significantly associated with less social distancing (b = -0.20, β = -0.06, p = .004). Moreover, living in nonurban settings was negatively associated with social distancing (b = -0.30, β = -0.09, p < .001). However, participants who lived outside the United States endorsed more social distancing overall compared to United States residents (b = 0.18, β = 0.10, p < .001).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Hierarchical linear regression of social distancing on sociodemographic predictors with standardized and unstandardized coefficients.

https://doi.org/10.1371/journal.pone.0264820.t003

The final model added 3.5% to the variance reported in Model 2 (p < .001), with a total of 5.9% of the variance explained by the full model (p < .001). In the full model, urbanicity (b = -0.31, β = 0.10, p < .001) and country of residence (b = 0.21, β = 0.11, p < .001) remained significant predictors of social distancing. Additionally, older adults (b = 0.03, β = 0.17, p < .001), women (b = 0.56, β = 0.11, p < .001), and those with greater educational attainment (b = 0.14, β = 0.06, p = .019) were more likely to engage in social distancing compared to younger adults, men, and those who completed high school or less, respectively. However, there were no significant effects of sexual orientation or employment status in the full model.

Hand washing

Table 4 displays the unstandardized and standardized effects of social cofactors on practicing hand washing during COVID in the overall sample and stratified by country. Older age was significantly associated with greater hand washing in the overall sample (b = 0.01, β = 0.07, p = .006) and among Spanish respondents (b = 0.01, β = 0.004, p = .003). We also found that women endorsed more hand washing than men (b = 0.12, β = 0.07, p = .004) and in the Indian subsample (b = 0.19, β = 0.09, p = .047). Differences by sexual orientation were also found; lesbian and gay adults in the full sample were less likely to engage in hand washing relative to their heterosexual peers (b = -0.35, β = -0.07, p = .015). Among Italian respondents, adults who identified as bisexual reported greater hand washing compared to heterosexuals (b = 0.56, β = 0.10, p < .001), yet those who identified as and pansexual or other sexual orientation were less likely to practice adequate hand washing compared to heterosexuals (b = -1.02, β = -0.13, p < .001). Additionally, adults working part-time were less likely to engage in hand washing relative to adults working full-time (b = -0.11, β = -0.04, p = .047). Although no significant differences were found among respondents in India, Saudi Arabia, Spain, or Italy, part-time workers in the United States were significantly less likely to engage in proper hand washing compared to those working full-time (b = -0.33, β = -0.15, p = .011). Finally, suburban (b = -0.12, β = -0.06, p = .014) and rural (b = -0.13, β = -0.05, p = .022) respondents were significantly less likely to practice appropriate hand washing compared to their urban peers. Subgroup analyses showed that Indian suburban (b = -0.31, β = -0.11, p = .026) and rural residents (b = -0.44, β = -0.17, p = .009) were significantly less likely than urban dwellers to engage in hand washing. Among Saudi Arabian respondents, rural residents endorsed significantly less engagement in proper hand washing compared to their urban peers (b = -0.61, β = -0.14, p = .005).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Unstandardized and standardized multivariable linear regression weights for hand washing behavior, stratified by country.

https://doi.org/10.1371/journal.pone.0264820.t004

Table 5 presents the results from a three-block hierarchical linear regression model regressing handwashing on sociodemographic predictors. In the first model (R² = 0.002, p = .064), sexual orientation was not a significant predictor of hand washing behavior (b = -0.05, β = -0.04, p = .064). However, in the second model (R² = .012, p < .001), sexual minority status (b = -0.05, β = -0.04, p = .045) and non-urban residence (b = -0.10, β = -0.08, p < .001) were negatively associated with hand washing as a protective behavior. Country of residence was not a significant predictor in the second model, regardless of participants’ sexual orientation or urbanicity.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 5. Hierarchical linear regression of hand washing on sociodemographic predictors with standardized and unstandardized coefficients.

https://doi.org/10.1371/journal.pone.0264820.t005

The final model explained 5.9% of the overall variance (p < .001) and represented a 3.5% increase in the variance explained in Model 2. In the final model, non-urban residence was associated with less hand washing (b = -0.09, β = -0.08, p = .001). However, older age (b = 0.01, β = 0.07, p < .001), identifying as a woman (b = 0.14, β = 0.07, p = .001), and greater educational attainment (b = 0.5, β = 0.06, p = .017) were all positively associated with hand washing in our sample. We did not find significant effects for country of residence, sexual minority status, or employment status on hand washing in the full model.

Sexual orientation and gender

In the hierarchical linear regression models, we did not find evidence for sexual minority adults experiencing disparately less social distancing or hand washing behavior compared to heterosexuals. However, analyses by country revealed that lesbian women and gay men endorsed greater levels of social distancing and hand washing than their heterosexual peers. Furthermore, Italian bisexual, pansexual, and other sexual minority adults reported significantly less hand washing compared to Italian heterosexuals. However, the effects for sexual minority adults were small to negligible. Thus, our hypothesis was partially supported.

Research suggests that community connectedness may suppress the deleterious effects of minority stress on health-related outcomes [29, 30]. However, the COVID-19 pandemic presents unique public health limitations on social gatherings that may run counter to the protective nature of close gatherings among sexual minority populations. Yet, sexual minority individuals in our sample may have utilized online communities to foster connectedness, support, and disseminating COVID-related mitigation strategies such as social distancing and hand washing. To this end, future investigations can investigate the comparative effects of in-person versus online connectedness to the LGBTQ+ community with respect to COVID conditions.

Overall, women reported greater social distancing and adherence to proper hand washing compared to men. Stratified analyses showed that Italian and Spanish women were significantly more likely to practice social distancing than their male peers, and Indian women were significantly more likely to engage in proper hand washing compared to Indian men. The observed effect of gender may be because in many cultures, women are often the family caregivers, and the COVID-19 pandemic may have increased the sense of cultural responsibilities for women to keep their families safe [31]—particularly women in India, Spain, and Italy. Additionally, the act of social distancing and hand washing may have been protective measures against psychological distress for women in our sample. In a recent systematic review of the literature [32], scholars have found that women experienced higher levels of anxiety, depression, and stress related to the COVID-19 pandemic. Thus, engaging in mitigation strategies may be a direct or indirect coping strategy for COVID-related psychological distress.

Urbanicity

There was also an effect of urbanicity on social distancing and hand washing, but multivariable analyses by country showed significant effects only for respondents living in India and Saudi Arabia. Non-urban dwellers in these two countries may not have the ability or financial means to live in areas that are conducive to social distancing. Although awareness of social distancing and hand washing during the COVID-19 pandemic have been documented in Saudi Arabia [17, 24], other scholars have found disparities in implementing mitigation strategies in the Saudi population [33]. In India, social distancing and proper hand washing are not as prevalent as other mitigation strategies [34]. In rural areas in India, clean water is not as readily available to residents as in more urbanized areas [35, 36]; as such, environmental disparities may exacerbate existing disparities in hand hygiene.

Furthermore, rural residents may not have equitable access to reputable public health information compared to their urban-dwelling peers. It may also be the case that some respondents, particularly men, living in rural areas provide the majority of household income and may not be able to socially distance at work. Because the research in this area is new, additional investigations are needed to examine the mechanisms that drive disparities in COVID-related mitigation strategies among suburban and rural adults, especially in areas such as India and Saudi Arabia. Such research could further support public health efforts in promoting social distancing and proper hand washing among nonurban residents in these countries.

Education and employment status

Results from the hierarchical regression models demonstrate that greater educational attainment was associated with greater social distancing and hand washing. Although the effects were negligible in the overall sample, stratified analyses revealed that slightly larger effect sizes were found among Saudi adults with a high school education or less. Moreover, employment was not a significant predictor of social distancing or hand washing in the hierarchical regression model. That said, part-time workers in the United States experienced significantly less social distancing and proper hand washing than adults working full-time. Although the literature on the specific contributions of educational attainment and employment status on social distancing and hand washing is limited, our findings correspond with studies demonstrating significant associations between social determinants of health, such as education and employment, and practicing mitigation strategies to minimize risk of respiratory-related communicable diseases [21, 26, 37–39]. Overall, our results, interpreted in context of the existing literature, suggest that the impact of education and employment on COVID-related social distancing and hand washing may be more nuanced than the literature may indicate. Furthermore, our findings warrant greater scientific investigations in the disparities experienced by residents of Saudi Arabia in an effort to reduce COVID infection and enhance their existing efforts to promote public health initiatives to address COVID-19.

Limitations

Interpreting this study’s results comes with extreme caution due to its limitations. First, the cross-sectional nature of the study prohibits causal inference. Second, the self-report nature of the study increases the risk of both social desirability and recall biases. To address recall bias, future research should consider and integrate behavior-analytic technology, which has been shown to be effective at increasing COVID-related protective behaviors in other populations and settings [40–42]. Third, despite the study drawing from a large sample of adults across five countries, Internet-based recruiting may have limited efforts to recruit from geographic locations that lack high-speed Internet access. Given the stigma of being lesbian, gay, or bisexual that remains in many countries, participants in this study may have been reluctant to identify as a sexual minority and therefore indicated a heterosexual sexual orientation for the study. Moreover, due to international privacy concerns, we did not include variables such as household size, marital status, and other associated variables.

At the time of the study’s implementation, wearing face masks and other mitigation strategies were not universally required across all five countries in the study; therefore, we lack the data to draw inferences on such strategies. However, future research should investigate disparities in such mitigation strategies within and across marginalized populations. Finally, we did not analyze race/ethnicity in this international study because the categorization of race/ethnicity varies widely across countries, and many do not document race/ethnicity/nationality in the same ways as the United States. Therefore, it would have been impossible to accurately and completely document the nuanced racial/ethnic identities of the target populations.