Recruitment strategies in marginalized industries for occupational health research: an example in a pilot study of cleaning staff during COVID-19

Abstract

This commentary describes challenges in recruiting workers from marginalized industries with examples from a pilot study of janitors, custodians, and maids and their experiences of cleaning and disinfection protocols during COVID-19 and potentially associated respiratory symptoms. Recruitment strategies included contacting a local hospital, national workers’ unions, and a large Arizona employer; using online Facebook groups; and contacting Arizona maid service companies and a school district. English and Spanish online and hard copy surveys about cleaning protocols and respiratory symptoms were used. Participants could also participate in online/phone interviews. Worker’s compensation, liability, and confidentiality were concerns across organizations. Online surveys yielded unreliable data. Hard copy surveys were used due to workers’ limited technology access. We reflect upon these challenges and discuss other strategies for recruiting from marginalized populations for occupational health research. Building trust with organizations and workers and considering technology access may ultimately increase recruitment feasibility.

During COVID-19, cleaning and disinfection protocols intensified despite relatively low transmission risks from surfaces (Centers for Disease Control and Prevention 2021). Some described these dramatic cleaning demonstrations as “hygiene theater,” especially within the context of fogging machines (Thompson 2020), with protocols designed to increase the public’s confidence in returning to these spaces. In 2020, there was a notable increase in calls to U.S. Poison Centers regarding exposures to cleaners and disinfectants (Kuehn 2020). Occupational exposure to cleaning and disinfection products has long been associated with increased risks of asthma-related outcomes and chronic obstructive pulmonary disease, especially for those in healthcare and cleaning industries (Zock et al. 2010; Dumas et al. 2019; Romero Starke et al. 2021). However, it is unknown how changes in COVID-19 cleaning and disinfection protocols affected workers in this industry.

In addition to likely higher exposures to cleaning and disinfection products among those in the cleaning industry during COVID-19, this occupational population tends to be understudied and composed of marginalized individuals. In 2017, the Pew Research Center reported that immigrants make up more than a third of those in building/grounds cleaning and maintenance industries (Desilver 2017). The Economic Policy Institute reported in 2020 that over half of house cleaners were Hispanic women (Wolfe et al. 2020). Marginalized populations are traditionally difficult to recruit, and continued efforts to determine the most effective ways to reach and serve marginalized communities through research are needed. Some workers may not have the awareness, language, writing, and computer skills necessary to access primarily virtual resources to protect their health (Okun et al. 2001; Brunette 2005). Due to their marginalized status, they may distrust government agencies and labor unions and are unlikely to push back against employers’ intensified cleaning and disinfection protocols even if they have concerns that it is affecting their health (Azaroff et al., 2011).

We piloted various recruitment strategies for administering a survey and conducting interviews with those in the cleaning industry (e.g. environmental services personnel, janitors/custodians, maids) regarding changes in cleaning and disinfection protocols during the COVID-19 pandemic and self-reported asthma-related symptoms, with our final recruitment strategy and survey results described elsewhere (Wilson et al. 2023). Our initial recruitment approach was to contact the occupational health specialist at a local hospital. While we gained written and signed support, after review by the workers’ compensation and legal departments, the project was deemed too risky due to concerns it would lead to “false” worker’s compensation claims. This threatened the hospital’s goal of decreasing the number of claims. One member of workers’ compensation gave an example of a person once reporting a reaction to a cleaning and disinfection product, but others who were exposed did not, implying this claim could have been false. This demonstrated a lack of knowledge regarding how exposures to cleaning and disinfection products could affect one worker and not the other due to sensitization effects (Dang et al. 2022), revealing a need for training on asthma-related outcomes and exposures. An alternative option that was discussed included access to workers’ compensation records to analyze exposures, as to avoid additional claims. However, this was not approved by the legal department, and there was uncertainty regarding who managed this data and would ultimately give permission for its use and data access. These experiences brought to light that the concern may be more than an increase in “false” claims but rather an increase in claims in general, leading to increased liability and economic risk. It should be noted that hospitals at this time (Fall 2021) were experiencing retention challenges, and there were broad concerns regarding worker satisfaction (American Hospital Association 2021; Johnson 2022).

Our next approach was to reach out to a large Arizona employer with many custodial staff who clean a variety of indoor environments. There was initially positive reception from the director of facilities management, but there was uncertainty as to what the benefits would be for the company. They stated that they use “safe” products and have “safe” protocols for their workers, implying that the protocols and product choices, alone, likely meant there were no health issues. This revealed a need for education among facilities managers regarding human error and barriers to using cleaning products in intended ways or carrying out cleaning protocols correctly. After more conversations about the potential benefits (e.g. understanding barriers to workers following safe protocols), they agreed to consider the opportunity. The employer’s legal team was then involved, and they expressed concerns regarding implications in the case of discovering occupational health issues in the surveys. They said that they could support recruitment from the company if anyone was technically eligible to participate, i.e. there would be no way of knowing which data originated from their workers versus others’ workers. The director of facilities management felt this would make the results no longer useful for improving the health and safety of their protocols because there would be no way to evaluate whether issues elucidated by the survey reflected their workers’ experiences, specifically.

The third approach was to call and email workers’ organizations, such as unions for those in the cleaning industry. No one was reached via email or phone despite multiple attempts and voicemails. A nurses’ union was contacted with connections to another union with environmental services personnel members, but the nurses’ union expressed concerns regarding the trust of researchers and whether the results of the project would have any direct impacts on union members. Due to the timeline of the project (Fall 2021–Fall 2022), a new recruitment approach was explored.

Next, we used private and public Facebook groups for those in janitorial, custodial, and maid services industries, using a reCAPTCHA at the beginning of the survey to prevent bots from filling out the survey. In some groups, users commented that the post was spam and not to be trusted, despite responses from the research team addressing concerns that the post was legitimate. Some Facebook group administrators categorized the post as promotion/advertisement and removed our access to the Facebook group. In one case, an individual, multiple individuals, or possibly online bots, appeared to take the anonymous online survey multiple times, entering identical responses in some cases to previously submitted surveys. New email addresses appeared to be created for each survey submission to maximize compensation. Responses to some questions appeared illogical, as if answering a question that was not being asked. Emails were then sent to research staff repeatedly demanding compensation, some from email addresses that were not recorded as participants but with first and last names that were spelled in varying arrangements that mimicked those that were suspected of being unreliable. Based on advice from the University of Arizona’s Human Subjects Protection Program, anyone who we had a record of participating was compensated since it could not be completely confirmed whether a participant was submitting reliable data or not. We were encouraged to state in our consent form that only those who provide “reliable” responses would be compensated. After the online survey and interview opportunities were then halted, emails demanding compensation from email addresses of which we have no record continued nearly 2 years after posting of the survey link. All those we had record of participating based on email address were compensated. In consultation with other local research groups who had experienced similar challenges with online surveys during the pandemic, it was recommended that either no compensation be given for publicly available online surveys, or that a smaller amount (e.g. $5) be provided, to demotivate those seeking ways to maximize profit from online surveys.

We refocused our recruitment efforts on potential local partners. We contacted local maid service companies and school districts to gauge interest in participating. We highlighted the benefits of the research for the cleaning industry to protect the workforce. Some companies described low retention since the start of the pandemic and expressed interest in learning whether workplace exposures could be in part to blame. Two maid service companies out of more than approximately ten that were contacted provided written approval to let us recruit their workers. A school district in Arizona with over 300 janitors/custodians also provided written approval and district Institutional Review Board approval. We were advised that hard copy surveys would be easier to administer due to the lack of technology access for the workers. Businesses did express concerns about what may be revealed through the survey. We reassured them that we would not ask about site of employment and would not analyze by employer, both to protect workers and employers. This recruitment strategy resulted in 59 completed surveys out of 150 that were administered. In some cases, multiple surveys were returned in a single packet. Extra postage had been provided on envelopes that resulted in packets successfully arriving to the research team. In a small number of instances, it appeared as though an individual helped another individual fill out a survey (multiple surveys in a single packet completed with similar handwriting but with unique and seemingly reliable answers), which indicated that some participants may not have been able to write their responses in Spanish and English. Upon reflection, the research team agreed that in-person surveys may be more successful in the future to increase accessibility for all community members. However, due to pandemic restrictions, this approach was not feasible for this project.

The recruitment challenges in this study are generalizable to the many challenges that can be encountered when conducting occupational health research, especially when attempting to access a workforce composed of marginalized populations. Recruitment may be especially challenging when the occupational group has low technology access and potential challenges with writing one’s own answers. While workers’ organizations or unions may be the most reliable way to recruit workers, building trust with these organizations is a long process that requires collaboration from the beginning of the project, ensuring that results have direct application to improving working conditions for organization members. Additionally, these populations tend not to be part of worker organizations or unions, in part leading to their marginalization. In some cases, these organizations may have no interest in research due to previous experiences in which members were exploited for research purposes that did not serve a benefit to the organization. They may also have concerns that research will undercut their efforts and support employers in gathering evidence that exposures are not of concern. While access to workers through employers may be the most feasible, this depends upon employer approval. Businesses may be reluctant to allow researchers to recruit their staff for research participation unless the legal risks for the employer are perceived as low and the benefit to the business is clear, such as protecting the workforce long-term and/or increasing retention. If businesses do approve, another challenge may arise: the reluctance of workers to participate due to fears of retaliation from employers if negative health outcomes associated with the job are revealed. Strategies are needed to increase the workers’ confidence and trust in the confidentiality and privacy of their participation. One way to alleviate these concerns is to provide hard copy surveys with a return mailer and postage that can be mailed to researchers independently, removing the employer as a contact at that point in the project. Even with this approach in the case of our study, 22% of participants reported “prefer not to answer” whether the frequency of respiratory symptoms at work during COVID-19 had increased, decreased, or stayed the same (Wilson et al. 2023).

A strategy we did not implement in this study but has been used for marginalized populations includes the “snowball” strategy (Sadler et al. 2010), in which personal contacts are told about the project and then disseminate knowledge about the project among their personal contacts. Another strategy includes connecting with trusted members of a community, such as engaging with promotores, or community health workers, who have long-standing relationships with community members. While promotores have been involved in environmental and public health research (Victory et al. 2017, 2022), their role has had limited utilization in occupational health research, with the exception of research on farmworkers and low wage workers in home-based and small businesses, such as auto repair and beauty salons (Ingram et al. 2007, 2021; Ramírez et al. 2015).

Occupational health research for occupations composed of marginalized individuals may present access and recruitment challenges, but research on the health of these workers is an important public health effort. With more than 3,000,000 workers in the cleaning industry in the United States (statista 2023), the health of this workforce has major implications for the maintenance and hygiene of built environments across the country and has suffered increased and sometimes unnecessary exposures due to the pandemic. The collective knowledge of successes and failures in the recruitment of the cleaning workforce for occupational health research advances our ability to protect their health and give voice to their health concerns.

Conflict of interest statement

AW worked on a project funded by an unrestricted grant from Reckitt, partially overlapping with the duration of the analysis stage of this research. The authors declare no other conflict of interest relating to the material presented in this article.

Data availability

No data were used in this study.