Occupational exposure to disinfectants and asthma control in US nurses

Discussion

In this study of 4102 US female nurses with asthma, disinfection tasks, in particular disinfection of medical instruments, were associated with poor asthma control. We found increased risks of poor asthma control associated with exposure to glutaraldehyde, formaldehyde, enzymatic cleaners, hypochlorite bleach and hydrogen peroxide, in particular for nurses exposed to several of these products. Exposure to quats and alcohol was not associated with poor asthma control.

To the best of our knowledge, this is the first time that asthma control has been studied in relation to occupational exposure to disinfectants among a large cohort of healthcare workers. Asthma was well-characterised in this population of nurses; a previous validation study in NHSII indeed confirmed 95% of the nurses' reports of doctor-diagnosed asthma [17]. Moreover, we evaluated asthma control using a standardised and validated definition (ACT), integrating the main domains of asthma control (symptoms, use of rescue therapy, sleep interference and activity limitations) [1, 22]. Many studies on work-related asthma have focused on distinguishing risk factors for occupational asthma (i.e. new-onset asthma caused by occupational exposure) and work-exacerbated asthma (worsening of pre-existing asthma) [23, 24]. In the current study, the association between use of disinfectants and poor asthma control was similar among nurses with childhood-onset (i.e. pre-existing) and adult-onset (i.e. potentially caused by occupational exposures) asthma. Although the distinction between occupational and work-exacerbated asthma is important for legal considerations (workers' compensation), it may be less relevant from a public health perspective as both occupational and work-exacerbated asthma have long-term socioeconomic and health consequences [25]. To minimise risk of exacerbations and accelerated lung function decline in workers, experts have called for research on occupational exposures contributing to poor asthma control [6]. Our results are consistent with a few recent studies showing that exposure to occupational asthmagens [5, 26, 27], and in particular cleaning agents [5, 28], is associated with uncontrolled asthma. Mechanisms by which disinfectants and cleaning products impact respiratory health remain unclear and require further research [11]. Some products (e.g. enzyme-based products) have sensitising potential, but most agents are assumed to act as respiratory irritants [11], and may cause injury of the airway epithelium, oxidative stress and long-lasting neurogenic inflammation [23, 24, 29]. Regardless of the exact mechanisms, our results support the need to consider occupation and potential exposure to disinfectants in clinical practice to improve management of patients with asthma [6, 30].

Identifying the specific tasks and agents associated with poor asthma control is of particular importance to improve asthma management among healthcare workers. We found that medical instrument disinfection was associated with poor asthma control, pointing toward a role of high-level disinfectants used on critical or semicritical items [15]. We did not find evidence of an association between use of disinfectants to clean surfaces and asthma control, although this type of task has been associated with increased risk of physician-diagnosed asthma in a study of healthcare workers in Texas in the USA [14]. Few epidemiological studies have investigated the relationship between specific cleaning agents and disinfectants and asthma outcomes in healthcare workers, and they were limited in terms of number/variety of specific products studied and attempts to control for correlated exposures [10, 12, 13]. In the current study, although we could not fully study the independent effect of each chemical (because we did not find subgroups of nurses with exposure to a single product), we attempted to identify combinations of exposures associated with greater risk of poor asthma control. We found particularly increased risk among nurses with exposure to aldehydes (formaldehyde or glutaraldehyde), hypochlorite bleach or hydrogen peroxide and enzymatic cleaners, based on the JTEM estimates. Formaldehyde has been known as an asthmagen for a long time [31]. Glutaraldehyde and hydrogen peroxide also have been suggested as agents implicated in occupational asthma among healthcare workers in a smaller study in Canada or case reports in France and the USA [32–34]. Associations between the use of hypochlorite bleach and asthma outcomes have been reported in European studies [11]. Enzymatic products are used to clean items before high-level disinfection or sterilisation. A role of enzymatic cleaners in asthma among healthcare workers was suspected based on case reports in healthcare settings and earlier reports in the detergent industry in the UK [35, 36]. We believe this is the first time these agents have been implicated in an epidemiological study. Finally, a strong association between exposure to quats and physician-diagnosed asthma was reported in a study of 543 French healthcare workers, in contrast to our results [12]. Exposure to quats is particularly difficult to evaluate [21]. In our study, although we integrated information collected on product brand names and review of the corresponding safety date sheets, as recommended [12], when creating the JTEM, exposure to quats may still be underestimated. However, any trend for an association between exposure to quats and poor asthma control disappeared after controlling for exposure to other products.

Apart from the active ingredients, specific types of tasks and product application procedures may influence respiratory health effects and are interesting targets for interventions [15]. Spraying has been associated with increased risk of asthma or respiratory symptoms, likely because of the higher potential for inhalation exposure when using sprays [13]. Consistently, we found in the current study that use of sprays was associated with poor asthma control.

Major strengths of our study included the large sample size and the use, for the first time, of a nurse-specific JTEM. Most existing studies on the relationship between the use of specific cleaning/disinfecting products and asthma relied on self-reported exposures, raising the issue of both nondifferential and differential misclassification bias, potentially leading to bias either toward or away from the null [21]. Use of job–exposure matrices (JEMs) is generally favoured in occupational epidemiology for exposure assessment in large populations [37]. As nursing jobs are heterogeneous (i.e. nurses with the same job title may perform different tasks), we showed that taking into account disinfection tasks (within job variability) in exposure assessment by creating a JTEM provides better exposure estimates than a JEM, i.e. reduces exposure misclassification [16]. In the present study, we observed associations between exposure to several products, as evaluated by the JTEM, and asthma control. However, as we used questionnaires to evaluate disinfection tasks, some of our results may be subject to differential misclassification bias (i.e. differential recall of exposure according to asthma outcomes, possibly leading to spurious associations). To address this question, we performed a sensitivity analysis studying prospectively the associations between disinfection tasks in 2013 and asthma control in 2014–2015, and found similar results, suggesting that differential misclassification bias is not likely a major explanation for our findings regarding disinfection tasks. Potential for recall bias in the association between self-reported cleaning/disinfection tasks and asthma has been formally evaluated in the Texas healthcare workers study; although the authors warn against the possibility of such bias, only a slight effect was reported [38]. Finally, other studies where evaluation of disinfectant exposure was based on expert assessment (i.e. independent of participant recall) have reported associations with current or physician-diagnosed asthma [12, 13].

The cross-sectional design of our study also prevents addressing the question of a healthy worker effect, i.e. a tendency of workers with a history of asthma or more severe asthma to avoid or leave jobs involving harmful exposures such as disinfectants. Such an effect has been suggested in a previous analysis of this cohort of women who have been in nursing jobs for several decades and may cause underestimation of some associations in the current study [19]. However, it is notable that we observed strong positive associations between exposure to several disinfectants and poor asthma control despite a potential healthy worker effect, which will typically bias associations toward the null [19]. In future studies, assessing the association between disinfectant exposures and asthma control among healthcare workers at an earlier career stage, and the impact of exposure duration, would be of particular interest.

Finally, in our study, exposure to eight specific disinfectants could not be evaluated using the JTEM as exposure prevalence was too low (<10% with weekly exposures in all nursing job types) to create such estimates, which was a limitation [16]. However, we evaluated risks associated with exposure to major disinfectants and cleaning agents used in healthcare settings [39]. Our results also confirmed that latex exposure was no longer the major concern for respiratory health among healthcare workers, as suggested recently [8], most likely because of the reduction in the use of powdered latex gloves.

In our study, the use of respiratory protection devices when handling disinfectants was limited (in accordance with a previous report [39]), although it was more frequent among nurses with poor asthma control. Even with more than 4000 nurses, because of low numbers, we could not determine whether use of respiratory protection devices had an impact on the association between disinfectant use and asthma control. However, use of personal protective equipment is generally not considered as the most effective measure of work-related asthma prevention. Elimination of hazardous substances and replacement with safer alternatives is preferred [15]. In accordance with earlier reports, we found that many chemicals in healthcare settings may have an impact on respiratory health, including products (e.g. hydrogen peroxide) used as alternatives to known asthmagens (e.g. aldehydes) [39]. Our results thus support the investigation of emerging nonchemical technologies for disinfection (e.g. steam, ultraviolet light) as a potential alternative to chemical disinfection and further research on green cleaning, integrating health risk reduction among the products' standards [15, 40].

The USA has more than 3 million registered nurses and over 12 million workers are employed in the healthcare industry [41]; in Europe, healthcare workers represents ∼10% of the workforce [8]. We found that the use of several specific disinfectants by nurses was associated with poor to very poor asthma control. Our findings highlight the urgency of integrating occupational health considerations in guidelines for cleaning and disinfection in healthcare [15]; they suggest targets for future efforts to improve asthma management in a large group of workers.