A survey was developed specifically for this study, written in simplified Swedish to accommodate the large part of the workforce with native languages other than Swedish. It was distributed to professional cleaning workers of schools, offices, stores, and hotels (and, to a lesser extent, hospitals, and industrial premises) in Southern Sweden in 2016. The survey covered questions about how often, and which sprays were used, as well as about whether the workers experienced any health-related symptoms in the nose, eyes and throat, coughing or difficulty breathing during spray use or when not using spray. In addition, questions were asked about the workers’ experience of musculoskeletal pain as well as medical background questions (including allergies, physician-diagnosed asthma, and smoking). For the symptom questions, Visual Analogue Scales (VAS) (100 mm) were used with the extremes labelled as “never” (0) and “always” (100). In total, 300 professional cleaning workers from a total of ten cleaning companies were informed about the study. Written informed consent was obtained from all workers that chose to respond, and the study was approved by the regional ethical review board at Lund University, Sweden. The full survey as well as the implementation of the survey is described in more detail elsewhere (in Swedish) .
Human chamber exposure
The study population in the human chamber exposures comprised 19 volunteer subjects. Inclusion criteria were: a) females, b) no current asthma diagnosis, c) non-smoker (for at least six months), d) age range 18–65 years, e) adequate Swedish language skills, and f) written informed consent and voluntary participation. Efforts were made that all subjects should be employed as professional cleaning workers (henceforth denoted “cleaning workers”), recruited among the participants of the survey. Due to recruitment difficulties among this population, 8 of the 19 subjects were recruited using advertising posters and were not professional cleaning workers (denoted “non-cleaning workers”). The principles of written informed consent in the current revision of the Declaration of Helsinki (2013) were implemented in the study. The study was approved by the regional ethical review board at Lund University, Sweden.
The test subjects were introduced to the chamber environment on a separate occasion, prior to study start, to minimize any effect of being unfamiliar with the environment. A pre-study examination described below was performed. A physician checked each subject’s medical history and conduced a physical examination. Spirometry with reversibility test (Bricanyl) was performed and a venous blood sample for Phadiatop allergy screening was obtained. Subjects with a current asthma diagnosis and/or with any regular (not anticonception) medication would have been excluded. No subjects were excluded on these grounds. Table 1 shows the characteristics of all subjects.
Cleaning products were selected for the human chamber exposures based on the observations from the survey. Three frequently used professional cleaning products from an internationally known brand were chosen (here denoted Window, Bathroom (normal), and Bathroom (acidic)). The specific physicochemical characteristics of the aerosol emissions from each of the cleaning products are described in detail in Lovén et al. . During the current study, the total aerosol concentration, both particle and gas phase, was measured continuously. Table 2 lists the ingredients from the Material Safety Data Sheets (MSDS) of the chosen products. Note that only substances with a content over 1% are required to be included in the MSDS. None of the products contained bleach, chlorine, or ammonia. The two bathroom products were provided as concentrates and were manually diluted with water to the recommended concentrations (1%). The two different spray bottles provided for the two bathroom products had adjustable nozzles with no fixed positions. Based on that the spray mists would cover similar target surface areas, nozzle positions of 180° (for Bathroom (normal)) and 360° (for Bathroom (acidic)) from a closed nozzle position were chosen. A foaming nozzle (the same type for all three products) was used for the foam exposures (the different exposure scenarios are described below). This nozzle was also adjustable, and a position of 360° was chosen.
The exposure chamber consists of a 21.6 m3 stainless steel chamber with a glass window and a floor area of 3 × 3 m, the approximate size of a hotel bathroom. The chamber was furnished with a toilet, sink, mirror, and shower (consisting of two tiled walls and two glass doors). A controlled flow of clean air was provided by a separate custom-built air-conditioning system maintained a temperature of 22.4 °C ± 0.7 °C, a relative humidity of 26.0% ± 3.5% (normal range for indoor winter time) and an air exchange rate of 0.9 h−1. The air supplied to the chamber by the conditioning system was filtered through an activated carbon filter (resulting in incoming air VOC concentrations of <0.1 ppm and ozone concentrations of <0.1 ppb). The air also flowed through an ultra-low penetration air (ULPA) filter (resulting in supply air particle concentrations of <100 particles cm−3 for particles <0.5 µm and <1 particle cm−3 for particles >0.5 µm). The chamber is described in detail by Isaxon et al. .
Particle number concentration and size distributions (in the size range 0.5–20 µm) were continuously measured using an Aerodynamic Particle Sizer (APS model 3321, TSI Inc., USA). A Condensation Particle Counter (CPC model 3010, TSI Inc., USA) was used to measure the particle number concentration in the size range >0.01 µm and a VelociCalc (model 9565-P, probe 986, TSI Inc., USA) was used to measure the total VOC gas phase concentration in the chamber. The time resolution of the APS, CPC, and VelociCalc were 5 s. An Ozone Analyzer (model 49i, Thermo Fisher Scientific, USA) was used to monitor the ozone concentration in the chamber with a time resolution of 1 s. In addition, a personal aerosol monitor SidePak (model AM510, TSI Inc., USA) was worn in a belt around the waist of the subjects to estimate the total particle mass concentration of particles in the size range 0.1–10 µm (PM10) in the breathing zone, with a time resolution of 10 s.
Three exposure scenarios were studied: spray – spraying the product onto the surfaces and wiping with pre-moistened microfiber cloths, foam – product application by foam onto the pre-moistened microfiber cloths and wiping with the cloths, and water – wiping only with pre-moistened microfiber cloths. It was randomly assigned to each subject what scenario to conduct on the first exposure day. The microfiber cloths were machine-washed and pre-moistened with water prior to each exposure day. All three cleaning products, denoted as Window, Bathroom (normal), and Bathroom (acidic), were used in scenarios spray and foam, the only difference between these two scenarios was the way the products were applied (either as spray or as foam).
The subjects were provided a protocol with instructions of how to use the cleaning equipment, specifically on the number of pulses used to apply the cleaning product in order to obtain comparable exposure levels. The same cleaning tasks were performed in all three exposure scenarios. The window and mirror in the chamber were cleaned with the Window product, the toilet and sink were cleaned with the Bathroom (normal) product, and the two tiled walls and two glass doors in the shower corner were cleaned with the Bathroom (acidic) product. During scenario water, the same cleaning tasks were performed without using any cleaning products.
Each subject performed one exposure scenario in one day (start ~ 8 am, end ~ 2 pm), for a total of three separate days with 1–3 weeks between scenarios. Two of the subjects only participated in two of the three exposure scenarios due to scheduling issues (one missed scenario foam and the other missed scenario spray). Figure 1 shows a flow chart of the study design for an exposure day. Each day included three 30-min cleaning exposures conducted inside the exposure chamber with a 1.5-h break between them. During the half-hour cleaning exposure, the bathroom inside the exposure chamber was cleaned eight times. Thus, during one exposure day each subject cleaned the bathroom 24 times, well in line with a normal number of bathrooms to clean daily for a hotel cleaning worker.
Biological sampling was performed twice a day, before and after the three exposures, shown in Fig. 1. Nasal lavage sampling was performed by flushing the nasal cavity with a room-temperature saline solution. The cells in the solution were immediately pelleted and the supernatant frozen at -80 °C . The biological sampling procedure was repeated a third time, the morning after exposure, at ~ 8 am. Blood samples and nasal lavage samples were stored at -80 °C until analysis. The blood samples were analyzed for hemoglobin (Hb) as well as for neutrophils, eosinophils, basophils, lymphocytes, monocytes, total leukocytes and C-Reactive Protein (CRP) by standard protocol at Clinical Chemistry at Medical services, Region Skåne. Interleukin 6 and 8 (IL-6 and IL-8) in serum and nasal lavage fluid were analyzed by a multiplexed immunoassay Luminex method according to the manufacturer’s instructions (Bio-Rad Life Science, Hercules, USA). IL-6 and IL-8 were analyzed at the Division of Occupational and Environmental Medicine at Lund University.
Medical assessments were conducted directly before and directly after each half-hour cleaning exposure, a total of six times during an exposure day, as shown in Fig. 1. The physician examined the eyes (redness of conjunctiva, tears), the anterior nose (redness, secretion, blockage) and the throat (redness, secretion). Lung auscultation at normal and forced respiration was performed. Peak nasal inspiratory flow (PNIF) measurements to detect nasal obstruction were performed, as well as non-invasive assessments of the tear film stability by measuring the tear film break-up time (BUT). A PNIF meter (GM instruments, UK) was used together with a reusable mask. An ocular microscope (Keeler TearScope®, Keeler Instruments, UK) was used to assess the BUT [27, 28]. The measurements of PNIF and BUT were repeated three times during each medical assessment and an average value was calculated. Lung function testing, conducted by spirometry, was performed in the morning and the afternoon of the exposure day to record forced vital capacity (FVC) and forced expiratory volume in one second (FEV1) (Fig. 1). The spirometry was performed with SPIRARE 3 (Diagnostica, Norway) according to the European Respiratory Society  protocol. FVC and FEV1 were obtained and compared to reference material (ECSC/ERS 1993).
Subjects also filled in a short self-assessment symptom score with VAS scales during the medical assessments, based on those used by Dierschke et al. . Symptoms from the eyes (itching, running, burning, sensation of dryness), the nose (itching, running, tingling, sensation of dryness, blocked), the pharynx (cough, sensation of dryness) and the lower airways (wheezing, shortness of breath, chest tightness) were registered. A question regarding strong smells was also included. The extremes of the VAS scales were labelled as “none” (0) and “a lot” (100). The symptom score was recorded by each subject six times during an exposure day and once in the following morning.
Heart rate and pulse were continuously monitored for the whole exposure day by a chest belt with a heart rate (HR) transmitter (model RS400, Polar Electronics, Finland). The subjects also wore a pulse watch, which recorded and stored the data.
Technical recordings of physical workload were performed on the eleven right-handed professional cleaning workers participating in the study. The physical workload during cleaning scenarios spray and water was assessed. To avoid aerosol exposure during the spray scenario, the cleaning spray bottle was filled with water instead of cleaning products, with minimal impact on the spraying performance for these recordings.
Postures and movements of the head, upper back and both upper arms were assessed by inclinometry . Reference postures for upright head and back (0° inclination) and for vertical upper arms (0° elevation) were performed according to Dahlqvist et al. . These references were later used to calculate work postures during scenario spray and water. Wrist postures and movements were recorded bilaterally with biaxial flexible electro-goniometers . A reference posture (0° flexion/extension) was recorded according to Gremark Simonsen et al. . The muscular load in the shoulder and forearm muscles was recorded using bipolar surface electromyography (EMG) . The EMG signals were amplified, filtered (10–400 Hz) and sampled at 2048 Hz, and stored in a Mobi-8 data logger (TMS International, Oldenzaal, Netherlands). Further signal processing was then carried out as described in Nordander et al. .
The muscular load (electrical activity) recorded during work was normalized to the activity during maximal voluntary contractions (maximal voluntary electrical activity, MVE), and expressed as %MVE. The maximal voluntary contraction (MVC) for the shoulder muscles was recorded according to Nordander et al. , and for the forearms muscles according to Dahlqvist et al. . Data were presented as group means of the 10th, 50th and/or 90th percentiles of the cumulative distributions of all recordings. Additionally, the recovery time (proportion of time <0.5% of MVE) was calculated as a percentage of the recorded time (% time) during scenario spray and water. To limit the amount of data, we chose to report data from the dominant (i.e. the right) side of the body.
To examine the associations of specific symptoms (nose, eyes, throat, coughing, or difficulty breathing) with different cleaning habits, subjects with symptoms were defined as those who reported self-experienced symptoms “often” or “always” (defined as 51–75 and 76–100 mm, respectively, on the VAS scales), while “never”, “rarely” and “sometimes” (i.e. <51 mm on the VAS scales) were considered as showing no symptom. The notation “any symptoms” was defined as those who “often” or “always” reported one or more specific symptoms. Pearson’s chi-squared (χ2) test was used to compare symptom outcomes and spray/non-spray. Relative risk ratios (RRs) and their corresponding 95% confidence intervals (95% CIs) were calculated. The associations between self-experienced symptoms and age groups, smoking, allergies, and number of working year-groups were also analyzed using Pearson’s chi-squared (χ2) test. Five different age groups (<25, 26–35, 36–45, 46–55, >56 years) and five different number of working year-groups (<5, 6–10, 11–20, 21–30, >31 years) were defined for these analyses. Since a limited number of the respondents were male, all statistical analyses were performed for all workers without stratifying for gender. All analyses were performed with SPSS software (Statistics 24, IBM, USA).
Human chamber exposure
All the medical results were calculated as the changes from each subject’s baseline (i.e. individually normalized values). The baseline values were obtained in the morning of each exposure day (Before 1 in Fig. 1). A linear mixed model was used to analyze the differences in changes of outcomes at scenarios spray and foam, respectively, versus changes at scenario water for the PNIF and BUT measurements, and for the symptom scores. Age and individual baseline values were included in the model. The repeated covariance type chosen was autoregressive (AR(1)) since all the measurements at different times are autocorrelated. For all the other medical results and physical workload, the Wilcoxon signed rank test was used to analyze the difference between each exposure. All analyses were performed using SPSS software (Statistics version 22 and 24, IBM, USA).