Characterization of Exposure to Cleaning Agents Among Health Workers in Two Southern African Tertiary Hospitals

Abstract

Background

Whilst cleaning agents are commonly used in workplaces and homes, health workers (HWs) are at increased risk of exposure to significantly higher concentrations used to prevent healthcare-associated infections. Exposure assessment has been challenging partly because many are used simultaneously resulting in complex airborne exposures with various chemicals requiring different sampling techniques. The main objective of this study was to characterize exposures of HWs to various cleaning agents in two tertiary academic hospitals in Southern Africa.

Methods

A cross-sectional study of HWs was conducted in two tertiary hospitals in South Africa (SAH) and Tanzania (TAH). Exposure assessment involved systematic workplace observations, interviews with key personnel, passive personal environmental sampling for aldehydes (ortho-phthalaldehyde—OPA, glutaraldehyde and formaldehyde), and biomonitoring for chlorhexidine.

Results

Overall, 269 samples were collected from SAH, with 62 (23%) collected from HWs that used OPA on the day of monitoring. OPA was detectable in 6 (2%) of all samples analysed, all of which were collected in the gastrointestinal unit of the SAH. Overall, department, job title, individual HW use of OPA and duration of OPA use were the important predictors of OPA exposure. Formaldehyde was detectable in 103 (38%) samples (GM = 0.0025 ppm; range: <0.0030 to 0.0270). Formaldehyde levels were below the ACGIH TLV-TWA (0.1 ppm). While individual HW use and duration of formaldehyde use were not associated with formaldehyde exposure, working in an ear, nose, and throat ward was positively associated with detectable exposures (P-value = 0.002). Glutaraldehyde was not detected in samples from the SAH. In the preliminary sampling conducted in the TAH, glutaraldehyde was detectable in 8 (73%) of the 11 samples collected (GM = 0.003 ppm; range: <0.002 to 0.028). Glutaraldehyde levels were lower than the ACGIH’s TLV-Ceiling Limit of 0.05 ppm. p-chloroaniline was detectable in 13 (4%) of the 336 urine samples (GM = 0.02 ng/ml range: <1.00 to 25.80).

Conclusion

The study concluded that detectable exposures to OPA were isolated to certain departments and were dependent on the dedicated use of OPA by the HW being monitored. In contrast, low-level formaldehyde exposures were present throughout the hospital. There is a need for more sensitive exposure assessment techniques for chlorhexidine given its widespread use in the health sector.

What is important about this paper?

Cleaning agents are commonly used in healthcare workers to prevent healthcare-associated infections, but these chemicals have been associated with a variety of adverse health outcomes. To the best of our knowledge, this is the first study in Africa to conduct quantitative exposure assessment for aldehydes (ortho-phthalaldehyde, glutaraldehyde, and formaldehyde) and to have conducted biological monitoring for chlorhexidine exposures in health workers. Exposure patterns differed between the two hospitals studied, but were below occupational exposure limits.

Introduction

Various studies have demonstrated an association between exposure to cleaning agents and adverse health effects such as rhinitis, asthma, and contact dermatitis (Jeebhay et al., 2014; Folletti et al., 2017; Prasad et al., 2018; Dumas et al., 2019; Rosenman et al., 2020). Although exposure to cleaning agents is common in different workplaces, health workers (HWs) in particular are at increased risk, since higher concentrations of a wide range of cleaning agents are used to prevent healthcare-associated infections; and cleaning agents use was enhanced with the COVID-19 pandemic (Arif and Delclos, 2012; Mwanga and Jeebhay, 2020; Zheng et al., 2020).

Fixed surface cleaning is one of the major tasks performed in healthcare settings to prevent healthcare-associated infections. Several studies have reported increased risk of adverse health effects such as asthma, rhinitis and dermatitis associated with exposure to cleaning agents used for surfaces, such as ammonia and bleach (Buck et al., 2000; Medina-Ramón et al., 2003; Vizcaya et al., 2015; Svanes et al., 2018; Rosenman et al., 2020). Chloramines released when bleach is mixed with cleaning products containing ammonium salts have also been reported to cause occupational asthma (Thickett et al., 2002; Quirce and Barranco, 2010).

Aldehydes, such as ortho-phthalaldehyde (OPA) and glutaraldehyde, are the most common high-level disinfectants used for heat-sensitive reusable semi-critical medical instruments such as endoscopes. Glutaraldehyde was used for over 40 years as a high-level disinfectant but it has been banned in some countries, e.g. the United Kingdom, due to its causal association with occupational asthma and allergic contact dermatitis (NIOSH, 2001; Henn et al., 2015; Walters et al., 2018). OPA was considered a safer replacement for glutaraldehyde in some healthcare settings, but has recently been reported to cause occupational asthma, contact dermatitis, and anaphylaxis (Pala and Moscato, 2013; Mwanga and Jeebhay, 2020).

Hand hygiene and skin/wound disinfection of patients in hospital settings have been effected using chlorhexidine (Nagendran et al., 2009; Wittczak et al., 2013; Dumas et al., 2018). Since chlorhexidine is also present in several common household products, it is ubiquitous in the domestic setting. Chlorhexidine is well known for its sensitizing and irritating properties to both the skin and airways (Nagendran et al., 2009; Wittczak et al., 2013; Dumas et al., 2018). Previous studies have reported cases of occupational asthma and dermatitis due to chlorhexidine (Nagendran et al., 2009; Wittczak et al., 2013; Dumas et al., 2018; Chan et al., 2019) and anaphylaxis among patients undergoing surgery/invasive procedures (Garvey et al., 2001; Stephens et al., 2001).

Exposure assessment for cleaning agents has historically posed a challenge since many cleaning agents are used simultaneously, resulting in airborne exposures that are a complex mixture of various chemicals with different physico-chemical properties; assessing such exposures requires multiple sampling techniques (Gerster et al., 2014; LeBouf et al., 2014; Saito et al., 2015; Su et al., 2018; Virji et al., 2019; Garrido et al., 2021). Another challenge has been that the product type, frequency of use and duration of use vary depending on the specific cleaning task performed (Gerster et al., 2014; Saito et al., 2015; Garrido et al., 2021). HWs may repeat a cleaning task several times a day, or use the same cleaning agents in different ways, resulting in variable degrees of chemical exposures.

The literature on development of sampling and analytical methods for the determination of OPA concentrations in air is scant (Fujita et al., 2006; Uchiyama et al., 2006; Miyajima et al., 2010; Tucker, 2014, 2008; NIOSH, 2015). Despite the commercial availability of passive samplers and their use with formaldehyde and glutaraldehyde (Levin et al., 1989; Wellons et al., 1998), to our knowledge, no studies have used passive sampling methods for determining airborne OPA exposures. With respect to chlorhexidine, biomonitoring has successfully identified chlorhexidine and/or its metabolites (p-chloroaniline and 1-chloro-4-nitrobenzene) in biological fluids, but challenges exist with specific reference to the lack of simple validated methods with high accuracy, in quantifying exposure (Wainwright and Cooke, 1986; Below et al., 2004; Fiorentino et al., 2010).

This study was conducted to characterize exposures of HWs to aldehydes and chlorhexidine, two major categories of cleaning agents, in two large tertiary hospitals in South Africa and Tanzania, and to identify important exposure determinants. Another separate communication focuses on the findings of the detailed health outcome assessments performed in the epidemiological study of this group of HWs (Mwanga et al., 2022).

Methods

Study population

A cross-sectional study including different categories of HWs was conducted in two large tertiary hospitals, one in South Africa (SAH) and one in Tanzania (TAH). Following meetings with several key stakeholders and walk-through inspections of both hospitals by the investigators, specific departments were identified as potentially high-risk exposure settings for cleaning agents owing to amount and frequency of cleaning agent use. In the SAH, 36 departments were included from the following sections: out-patient clinic; intensive care units (ICUs); operating theatres; emergency units; ear, nose and throat (ENT) ward; vascular radiology and the haemodialysis unit. In the TAH, 13 departments were included from the following sections: out-patient clinics, ICUs, operating theatres, emergency unit, Central Sterile Services Department (CSSD), and haemodialysis unit. Ethics approval was obtained from the Human Research Ethics Committee (HREC) of the University of Cape Town (HREC Ref: 212/2013), Muhimbili University of Health and Allied Sciences (MUHAS) Institutional Review Board, and University of Michigan Medical School Institutional Review Board (HUM00083115). The study was conducted prior to the COVID-19 pandemic.

General exposure assessment

A list of cleaning agents used in the two hospitals was obtained from the respective supply chain departments. Information was also obtained about the type and volume of chemicals used in the various departments of the hospital to confirm departments identified a priori as high risk for cleaning agent exposures. The most recent safety data sheets of the cleaning products were obtained from the supply chain departments and/or from suppliers/manufacturers directly. Walk-through surveys were conducted in both hospitals using a proforma checklist (see S1, available at Annals of Occupational Hygiene online) based on a National Institute for Occupational Safety and Health (NIOSH) questionnaire to collect information regarding cleaning products used, tasks performed, frequency and duration of use (Saito et al., 2015). The surveys were conducted by a team that included an occupational hygienist and occupational medicine specialist. The team members also conducted short interviews with the operational managers of the respective work area. During the walk-through survey, each member of the team conducted their own independent evaluation (duration of observation = 25% of the working time). After evaluating each specific work area (department), the team members convened for a short discussion to have a consensus decision on the relevant exposures pertaining to that department. Discrepancies in findings were resolved by consensus and the findings recorded on the checklist. The research team also communicated with the supply chain departments and nurse managers on a regular basis to ensure that any new products that were introduced during the study period also formed part of the assessment.

Environmental sampling of aldehydes

Environmental sampling of aldehydes was conducted in the SAH since HWs complained of work-related asthma and skin symptoms to the nurse manager when using medical instrument cleaning agents, especially OPA (see passive sampling checklist in S2, available at Annals of Occupational Hygiene online). A preliminary study in the TAH revealed very low levels of OPA and glutaraldehyde and, as a result, extensive environmental sampling was not done in this hospital. In the SAH, measurements of aldehydes were conducted in the 17 departments where OPA and enzymatic cleaners were used for cleaning and disinfection of medical instruments. This decision was informed by the initial walk-through survey findings. Selection of workers for OPA monitoring was limited to those working on the day shift for logistical reasons, and deliberately sampled workers that directly handled OPA or worked in the close proximity to the medical instrument cleaning activity. The selection of workers was based on using sample size calculations to select the top 20% of highly exposed individuals, employing a 95% confidence level (Leidel et al., 1977).

Passive sampling used TraceAir^® AT580 monitors (Assay Technology, Livermore, CA), which have a maximum sampling rate for OPA = 54.8 ml/min and for glutaraldehyde = 60.2 ml/min. Field blanks were included in every day of sampling. All samples were stored at 4°C after collection. Analysis was conducted within four weeks of sample collection.

Samples were analysed for OPA, glutaraldehyde and formaldehyde using OSHA method 64 and NIOSH method 2016 in a South African National Accreditation System (SANAS) accredited laboratory. The detailed analytical methods are available in S3, available at Annals of Occupational Hygiene online.

Biomonitoring for chlorhexidine

Chlorhexidine biomonitoring was conducted in the SAH only since none of the HWs in the TAH used chlorhexidine. Spot urine samples (50 ml) from 336 participants were collected from the study participants during their health outcome assessment visit to the study venue, in a clean indoor toilet using a plastic container topped with a plastic cap. To avoid contamination, participants were instructed on specific precautions on washing hands before handling containers; not touching the inside of the container; collecting the midstream urine, and covering the containers immediately after producing the sample. The samples were stored in the refrigerator at 4°C at the study venue and then transported on dry ice to the permanent storage facility on the same day of sample collection. The samples were stored at ‐80°C before being analysed at the Clinical Pharmacology laboratory at the University of Cape Town using an LC–MS/MS method developed in-house based on the detection of p-chloroaniline (PCA). The detailed analytical methods are available in S4, available at Annals of Occupational Hygiene online. Briefly, the samples were thawed at room temperature and extracted with 4 volumes of a 1:1 mix of 0.1% formic acid and acetonitrile, containing 62.5 ng/ml p-bromoaniline as internal standard. The supernatant following centrifugation was transferred to a 96-well plate for LC–MS/MS analysis. Calibration standards and quality control samples were prepared by spiking PCA in blank urine to give final concentrations between 1 and 3125 ng/ml. These were then extracted as described and analysed together with the patient samples to provide a standard curve from which patient PCA levels were determined. LC-MS/MS was performed on an ABSciex 4000Qtrap^® mass spectrometer coupled to an Agilent 1200 Rapid Resolution HPLC system. Chromatography was achieved using a Kinetex F5 column (100 × 4.6 mm, 2.6 µm) using 0.1% formic acid as the aqueous mobile phase and 0.1% formic acid in acetonitrile as the organic phase. An isocratic method at 0.8 ml/min was run, with a 1:1 split between the MS and waste. Carry over was avoided using a needle wash consisting of water, acetonitrile, methanol, isopropanol and formic acid (30:30:30:10:0.1). Analyst 1.6 software was used for instrument control, data acquisition and analyte quantitation.

Statistical analyses

All data analyses were performed using statistical package STATA version 14 (StataCorp, College Station, Texas). Frequencies of categorical variables such as major categories of cleaning/disinfecting tasks, specific control measures uptake and common cleaning products used were calculated. Numerical variables were summarized using median and range where data were not normally distributed. Predictably, exposure data followed a log-normal distribution, and as a result geometric mean and geometric standard deviation were used to summarize the measured concentrations of aldehydes and PCA. Chi-squared test and Wilcoxon sum rank test were used to examine the association between the outcomes of interest (aldehyde levels and PCA) and the predictor variables (e.g. job title, department). Analysis of this left-censored data used the β-substitution method (Ganser and Hewett, 2010; Huynh et al., 2014). OPA data were reported as a binary variable (detectable versus undetectable) due to the high percentage of censored data.

Results

General exposure assessment

Results from walk-through surveys demonstrated that the major categories of cleaning-related tasks performed in these two hospitals included: medical instruments cleaning and disinfection, fixed surface cleaning and disinfection, floor finishing tasks (stripping, waxing, and buffing), specimen preparation, patients’ skin/wound cleaning and disinfection and hand washing/sanitising (Table 1). It was also observed that in most departments sampled (100% in TAH and 89% in SAH), all groups of HWs used more wipes than aerosolized sprays when applying cleaning agents.

Table 1.

Cleaning-related tasks and control measures reported by a tertiary hospital in South Africa and in Tanzania.

	South Africa n (%)	Tanzania n (%)
No. of health workers (HWs) in the selected departments	759	560
No. of departments investigated	36	13
Cleaning-related tasks performeda
Hand washing/ sanitizing	36 (100)	13 (100)
Fixed surfaces cleaning and disinfection	35 (97)	13 (100)
Medical instruments cleaning and disinfection	20 (56)	13 (100)
Patients’ skin/wound cleaning and disinfection	31 (86)	12 (92)
Floor finishing tasks (stripping, waxing, buffing)	36 (100)	0 (0)
Specimen preparation	19 (53)	8 (62)
The manner of cleaning products use: used more sprays or more wipes  a
More wipes than sprays	32 (89)	13 (100)
More sprays than wipes	3 (8)	0 (0)
Use sprays and wipes about equally	1 (3)	0 (0)
Control measures uptake
Engineering controls presenta
Extractor fans in the ceiling	30 (83)	2 (15)
Local exhaust ventilation system	0 (0)	0 (0)
Administrative controls present
Training on adverse health effects due to cleaning agentsb	16 (44)	3 (23)
Availability of standard operating procedures document/s on how to use cleaning agentsa	13 (36)	11 (85)
Housekeeping: chemical spill/ release observeda	0 (0)	1 (8)
Personal protective equipment usea
Gloves	36 (100)	13 (100)
Protective clothing (aprons, overalls)	29 (81)	3 (23)
Eye protection (goggles, face shields)	2 (6)	0 (0)
Foot protection (safety/gum boots)	1 (3)	3 (23)
Air purifying half face respirator	0 (0)	0 (0)
Medical surveillance programmeb	0 (0)	0 (0)
Number of departments that reported at least one HW with adverse health effects associated with cleaning agents in the last 12 months (N = 36; N = 13)b	8 (22)	7 (54)
Number of HWs that reported to their manager to have experienced adverse health effects (skin, ocular-nasal or chest symptoms) associated with cleaning agents in the last 12 monthsb	16 (2)	39 (7)

^aInformation from walk-through surveys.

^binformation from interviews with nurse managers.

It was also noted during walk-through surveys that there were no local exhaust ventilation (LEV) systems present for medical instrument cleaning and disinfection tasks in both hospitals. A greater proportion of departments in the SAH (83%) than the TAH (15%) used ceiling extractor fans. Interviews with nurse managers revealed that a very low proportion of HWs (44% in SAH and 23% in TAH) had received training on adverse health effects due to cleaning agents. Furthermore, nurse managers at both hospitals reported that there were no specific medical surveillance programs for HWs working with cleaning agents.

Among the 36 nurse managers in the SAH and 13 nurse managers that were interviewed in the TAH, 8 (22%) and 7 (54%), respectively, reported at least one HW in their department with adverse health effects caused by cleaning agents in the last 12 months, as reported to them by symptomatic HWs. The reported number of HWs that experienced these adverse health effects to their nurse manager was higher in the TAH, 39 (7%), when compared with the SAH, 16 (2%). While nurse managers in TAH stated that HWs reported mainly ocular symptoms, nurse managers at SAH stated that airway (nasal, throat, and chest) and skin symptoms were more commonly reported. The cleaning agents suspected of being responsible for these adverse health effects included products commonly used for medical instrument and fixed surfaces cleaning and disinfection. Enzymatic cleaners and OPA were identified as associated with symptoms at both hospitals. Chlorhexidine, quaternary ammonium compounds and a high-level disinfectant mixture containing acetic acid, peracetic acid and hydrogen peroxide were only reported at the SAH. Glutaraldehyde and chlorine-based products (sodium dichloroisocyanurate “Troclosene sodium” tablets and liquid bleach—3.5% sodium hypochlorite) were only reported at the TAH.

The most common high-level disinfectant used in the departments for heat-sensitive medical instruments in SAH was OPA (36%) followed by hydrogen peroxide (14%) (Table 2). Whilst OPA (23%) was also used in the TAH, glutaraldehyde (31%) was more commonly used. Glutaraldehyde was not used for medical instruments cleaning and disinfection in the SAH. Enzymatic cleaners were also used in both hospitals for cleaning medical instruments prior to disinfection. Although quaternary ammonium compounds (QACs) were not used in the TAH, three departments (8%) in the SAH used a product containing QACs for high-level disinfection of medical instruments. A high-level disinfectant containing acetic acid, peracetic acid, and hydrogen peroxide was used only in departments in the SAH.

Table 2.

Commonly used cleaning products in departments according to duration of use reported by a tertiary hospital in South Africa and in Tanzania.

	South Africa N= 36 departments			Tanzania N = 13 departments
Medical instruments cleaning and disinfection	N = 20	n (%)	Duration of use Min/week median (range)	N = 13	n (%)	Duration of use Min/week median (range)
	Enzymatic cleaners	15 (42)	300 (8–1800)	Bleach	10 (77)	248 (70–490)
	Ortho-phthalaldehyde	13 (36)	75 (5–1800)	All-purpose cleaner	10 (77)	368 (95–1500)
	Alcohols	7 (19)	300 (2–600)	Enzymatic cleaners	5 (39)	215 (70–770)
	Chlorhexidine	6 (17)	255 (70–1200)	Glutaraldehyde	4 (31)	70 (20–750)
	Hydrogen peroxide	5 (14)	210 (25–400)	Ortho-phthalaldehyde	3 (23)	100 (30–500)
Fixed surfaces cleaning and disinfection	N = 35	n (%)		N = 13	n (%)
	Bleach	34 (94)	1020 (140–4200)	Bleach	13 (100)	1200 (900–1800)
	Ammonia	33 (92)	600 (50–2520)	All-purpose cleaner	13 (100)	900 (150–1800)
	Alcohols	31 (86)	300 (50–2520)	Glass cleaner	7 (54)	150 (30–150)
	Dishwashing liquid	29 (81)	840 (50–4200)	Floor cleaner	2 (15)	120 (120–120)
	Air freshener	21 (58)	9 (2–150)
Floor finishing tasks (stripping, waxing, and buffing)	N = 36	n (%)		N = 0	n (%)
	Floor stripperb	36 (100)
	Floor waxa	36 (100)	150 (30–900)
Specimen preparation	N = 19	n (%)		N = 8	n (%)
	Formalin 10% solution	18 (50)	33 (2–300)	Formalin 10% solution	8 (62)	6 (2–50)
	Cytological fixative spray	14 (39)	28 (2–300)
	Alcohol solution	3 (8)	20 (2–50)
	Glutaraldehyde	1 (3)	2 (2–2)
Patients’ skin/wound cleaning and disinfection	N = 31	n (%)		N = 12	n (%)
	Alcohols	27 (75)	200 (7–1000)	Alcohols	12 (92)	150 (25–300)
	Chlorhexidine	22 (61)	130 (3–700)	Povidone iodine	11 (85)	150 (30–300)
	Povidone iodine	15 (42)	144 (2–350)	Chloroxylenol	4 (31)	113 (75–300)
	Ether	6 (17)	60 (12–280)	Hydrogen peroxide mouthwash	1 (8)	50 (50–50)
Hand washing/sanitizing	N = 36	n (%)	Times/ay median (range)	N =12	n (%)	Times/day median (range)
	Chlorhexidine	36 (100)	55 (3–120)	All-purpose cleaner (diluted)	12 (92)	30 (10 – 40)
	Liquid hand soap	33 (92)	15 (2–100)	Alcohol sanitizer	8 (62)	23 (5–35)
	Alcohol sanitizer	27 (75)	60 (10–140)	Hand wash liquid soap	1 (8)	10 (10–10)
	Povidone iodine	6 (17)	38 (15–320)

N, number of departments assessed; n, number of departments conducting the task/s.

^aDuration of buff sprays (diluted floor wax).

^bFloor strippers were used only once a year in each department in the SAH.

Chlorine-based bleach (chlorine granular powder at the SAH and 3.5% sodium hypochlorite solution at the TAH) was the most common product used for fixed surfaces cleaning and disinfection in both hospitals. Bleach also had the longest average duration of use for fixed surfaces cleaning and disinfection in both hospitals (Table 2). Floor finishing tasks were not performed in the TAH. While floor strippers and waxes were used only once a year in each department in the SAH, buff sprays (diluted floor waxes) were used more frequently (median: 2 times/week; range: 1–5 times/week). Formalin (10%) solution was commonly used in both hospitals for specimen preparation (tissue fixation).

Alcohols and povidone iodine were used commonly in both hospitals for disinfection of patients’ surfaces before a surgical or instrument procedure or for wound care. Notably, chlorhexidine containing products were commonly used (61%) in the SAH for patients’ surfaces disinfection and wound care but not in TAH. HWs in the emergency units of both hospitals had the longest average duration of alcohol usage for disinfection of patients’ surfaces and wound care compared with other departments (see Table S5.1A–D, available at Annals of Occupational Hygiene online). Liquid products for hand washing and/or sanitising were used quite commonly by HWs in all the departments studied in both hospitals although South African HWs reported much higher frequency of use.

Environmental sampling for aldehydes

A total of 269 full-shift passive personal samples were collected from 164 HWs selected from 17 different departments at SAH. Among the 164 workers selected, 70 (43%) were sampled once, 83 (50%) sampled twice, and 11 (7%) were sampled thrice. The median sampling duration was 441 minutes (interquartile range: 386–477 min). The limits of detection (LOD) were 0.0001 ppm for OPA, 0.003 ppm for formaldehyde and 0.002 ppm for glutaraldehyde.

Formaldehyde was detectable in 103 of the 269 (38%) collected samples (GM = 0.0025 ppm; range: < 0.0030 to 0.0270) (Table 3). The 95th percentile formaldehyde concentration (0.0073 ppm) was 10-fold lower than the ACGIH TLV-TWA (0.1 ppm) but almost approached the NIOSH recommended exposure limit (REL) of 0.016 ppm TWA. Formaldehyde exposure was positively associated with working in an ENT ward (P = 0.002). Job title, individual HW formaldehyde use and its duration of use were not associated with detectable formaldehyde levels (Table 4).

Table 3.

Personal formaldehyde exposure levels per department and job title in a South African tertiary hospital.

	K	N	Formaldehyde (ppm)
			Detectable samples
			n (%)	AM	GM	GSD	Range
Department
Out-patient clinics	75	147	51 (35)	0.0028	0.0024	1.5959	<0.0030 to 0.0080
Operating theatres	59	79	25 (32)	0.0029	0.0023	1.6968	<0.0030 to 0.0200
Emergency units	3	3	3 (100)	0.0040	0.0039	1.2919	<0.0030 to 0.0050
ENT ward	7	13	10 (77)	0.0057	0.0042	2.0660	<0.0030 to 0.0270
Vascular radiology	20	27	14 (52)	0.0041	0.0031	1.9828	<0.0030 to 0.0180
Job title
Nurses (RN, EN, and NA)	94	160	60 (38)	0.0030	0.0025	1.7082	<0.0030 to 0.0200
Sterilizing operators	6	10	3 (30)	0.0024	0.0021	1.4515	<0.0030 to 0.0040
Doctors	5	7	4 (57)	0.0067	0.0039	2.7116	<0.0030 to 0.0270
Cleaners	19	32	10 (31)	0.0026	0.0023	1.5551	<0.0030 to 0.0070
Othersa	40	60	26 (43)	0.0032	0.0026	1.7228	<0.0030 to 0.0180
Overall	164	269	103 (38)	0.0031	0.0025	1.9196	<0.0030 to 0.0270

K, no. of workers sampled; N, no. of samples collected; n, no. of samples with detectable levels; AM, arithmetic mean; GM, geometric mean; GSD, geometric standard deviation; ENT, ear, nose and throat; ppm, parts per million; RN, registered nurse; EN, enrolled nurse; NA, nurse assistant.

^aOthers (clerks, porters, technologists, ECG technicians and radiographers); LOD for formaldehyde = 0.003 ppm.

Table 4.

Predictors of detectable personal formaldehyde exposure levels in a South African tertiary hospital.

	Formaldehyde
	Overall	Detected n (%)	Undetected n (%)	P-value (Chi-squared test)
Samples (n)	269	103	166
Department				0.002
Out-patient clinics	147	51 (35)	96 (65)
Operating theatres	78	25 (32)	53 (68)
Emergency units	3	3 (100)	0 (0)
ENT ward	13	10 (77)	3 (23)
Vascular radiology	27	14 (52)	13 (48)
Job title				0.582a
Nurses (RN, EN, and NA)	160	60 (38)	100 (62)
Doctors	7	4 (57)	3 (43)
Cleaners	32	10 (31)	22 (69)
Sterilizing operators	10	3 (30)	7 (70)
Othersb	59	26 (44)	33 (56)
Formalin used in the department	183 (68)	66 (64)	117 (71)	0.242
Formalin used by health worker	19 (7)	7 (7)	12 (7)	0.882
Duration of formalin use during specimen preparation task [(median (IQR)] (min/day)	2 (1–6)	2 (1–2)	2 (1–6)	0.392c

IQR, interquartile range; ppm, parts per million; ENT, ear, nose and throat; RN, registered nurse; EN, enrolled nurse; NA, nurse assistant; LOD for formaldehyde = 0.003 ppm.

^aFisher’s exact test.

^bOthers (clerks, porters, radiographers, technologists, and electrocardiogram technicians).

^cWilcoxon sum rank test.

Among all 269 samples obtained, 62 (23%) were collected from HWs that used OPA during specific tasks on the day of sampling. OPA was detectable in 6 (2%) of all samples analysed and none of the field blanks had detectable aldehyde levels. All the OPA detections occurred in the GI unit of SAH, among sterilizing operators and nurses that used OPA on the day of sampling. Sterilizing operators had significantly greater odds of having detectable OPA exposures than nurses (P < 0.001) (Table 5). Detectable OPA levels were positively associated with individual HW OPA use and longer duration of OPA use (P < 0.001). Overall, department, job title, individual HW’s use of OPA and duration of OPA use were the significant predictors of detectable OPA exposures (Table 5).

Table 5.

Predictors of detectable personal ortho-phthalaldehyde (OPA) exposure levels in a South African tertiary hospital.

	Ortho-phthalaldehyde
	Overall	Detected n (%)	Undetected n (%)	P-value (Fisher’s exact test)
Samples (n)	269	6	263
Job title				<0.001
Sterilising operators	10	3 (30)	7 (70)
Nurses(RN, EN, and NA)	160	3 (2)	157 (98)
Othersa	98	0 (0)	98 (100)
OPA used by health worker	62 (23)	6 (100)	56 (21)	<0.001
Duration of OPA use during high-level disinfection task [(median (IQR)] (min/day)	4 (2–10)	20 (14–30)	4 (2–6)	<0.001b

^aOthers (cleaners, clerks, porters, radiographers, technologists, electrocardiogram technicians, and doctors).

^bWilcoxon sum rank test; IQR, interquartile range; ppm, parts per million; RN, registered nurse; EN, enrolled nurse; NA, nurse assistant; LOD for OPA = 0.0001 ppm.

Glutaraldehyde was not detected in the SAH. In the preliminary sampling conducted in the TAH, glutaraldehyde was detectable in 8 of 11 (73%) samples (GM = 0.003 ppm; range: <0.002 to 0.028). The 95th percentile glutaraldehyde concentration (0.024 ppm) was lower than the ACGIH’s TLV-Ceiling Limit of 0.05 ppm.

Biomonitoring for chlorhexidine

PCA, a metabolite of chlorhexidine, was detected in 13 of 336 (4%) urine samples (GM = 0.02 ng/ml range: <1.00 to 25.80) and did not appear to differ by department or job type (see Tables S6.1 and S6.2, available at Annals of Occupational Hygiene online). The highest concentration (25.80 ng/ml) measured was obtained from a technician in the Haemodialysis unit, and next highest from a registered nurse (6.47 ng/ml) in the Vascular Radiology department.

Discussion

To the best of our knowledge, this is the first study in Africa to conduct quantitative exposure assessment for aldehydes (OPA, glutaraldehyde and formaldehyde) and biological monitoring for chlorhexidine exposures in HWs. The study demonstrated that a wide variety of agents are used for cleaning and disinfection in hospital settings located in Southern Africa, though the agents are similar to those used in healthcare settings elsewhere, despite the fact that some agents are not used for similar tasks. OPA was the most common high-level disinfectant used in the SAH while was glutaraldehyde dominant in the TAH. Measurable exposures to OPA were isolated to certain departments and were dependent on job title, regular use of OPA and duration of use, while low-level formaldehyde exposures present throughout these hospitals.

In contrast to other hospital settings (El-Helaly et al., 2016; Quinot et al., 2017), formaldehyde was not used for medical instrument cleaning and disinfection in either the SAH or the TAH. While quaternary ammonium compounds are commonly used for cleaning and disinfection in other healthcare settings, they were not used in the TAH and were only used in three departments at the SAH for medical instrument disinfection (Dumas et al., 2012; Gonzalez et al., 2014; Saito et al., 2015; El-Helaly et al., 2016; Zheng et al., 2020),.

In this study, chlorine-based bleach was the most common cleaning product used, which also had the longest average duration of use for fixed surface cleaning and disinfection in both hospitals. This finding is consistent with studies reporting bleach as commonly used for cleaning and disinfection in both the domestic and hospital settings (Medina-Ramón et al., 2005; Quinot et al., 2017; Garrido et al., 2021). Furthermore, alcohols were commonly used for several cleaning tasks. These findings corroborate results from other countries reporting alcohols as one of the most common ingredients of cleaning products used in hospital settings (Bello et al., 2009; Dumas et al., 2012; Saito et al., 2015).

Workplace engineering controls recommended for reducing exposure to cleaning agents (ACGIH, 2013; NIOSH, 2015) were either lacking or present in only a few areas of both hospitals. Control of workplace exposures to cleaning agents has posed challenges not only in lower income healthcare settings but also in the industrialized world (Nayebzadeh, 2007). A Canadian study reported lack of LEV in all the locations where glutaraldehyde was used (Nayebzadeh, 2007). It may also be challenging to implement engineering controls (e.g. LEV) for some cleaning tasks in locations such as fixed surfaces cleaning throughout the hospital. However, it is possible to install LEV systems in areas where specific tasks are performed such as medical instrument cleaning and disinfection, decanting or dilution of cleaning products. A recent study reported better workplace controls in areas with negative ventilation where most nurses (96.4%) worked (El-Helaly et al., 2016).

Formaldehyde levels (GM = 0.0025 ppm) recorded in the current study were, on average, 10-fold lower compared with the median formaldehyde concentration reported by Lee et al. (0.04 ppm using the active method and 0.05 ppm for the passive method) (2017). The most likely reason could be the exclusion of laboratory workers from this study, as this group is known to utilize formaldehyde for specimen preparation and were studied by Lee et al. Formaldehyde levels observed in the current study appear to be more comparable to average levels in US general buildings (U.S. Environmental Protection Agency, 2001), suggesting that the formaldehyde could have resulted from the widespread use of formaldehyde-10% solution in most clinical departments for specimen preparation, as observed during walk-through survey in addition to emission from formaldehyde-contaminated surfaces and other general indoor sources, like building materials. However, the contribution of ambient formaldehyde levels cannot be excluded.

The highest glutaraldehyde concentration measured during the preliminary sampling conducted in the TAH was 0.028 ppm. This particular measurement was collected from a nurse working in an endoscopy unit with the highest duration (60 min) of glutaraldehyde usage. The median glutaraldehyde levels (GM = 0.003 ppm) in this study were similar to those reported by the Marena et al. (mean = 0.005 ppm), slightly higher than the NIOSH study (range: not detected—0.005 ppm), but lower than levels measured by the Nayebzadeh et al. (GM = 0.025 ppm) (Marena et al., 2003; Nayebzadeh, 2007; NIOSH, 2015). Although glutaraldehyde levels in the current study were below the ACGIH’s TLV-Ceiling Limit, HWs in the TAH continue to report work-related symptoms due to glutaraldehyde, underscoring the need to lower the regulatory exposure standards for this chemical.

The results of this study continue to highlight the challenges in developing analytical methods for biological monitoring of chlorhexidine and its metabolites (PCA and 1-chloro-4-nitrobenzene), including lack of sensitivity, specificity, accuracy and lengthy sample extraction processes (Fiorentino et al., 2010). Since there are no exposure assessment studies that have reported chlorhexidine metabolite levels among HWs, the relevance of the levels found in the current study with regard to the risk of adverse health effects cannot be interpreted with certainty. Nevertheless, the findings of this study could be a useful comparator for future studies of PCA measurements in urine. Furthermore, given the fact that chlorhexidine is a common ingredient in many household products, it would be difficult to establish with certainty the contribution of occupational exposure as opposed to the non-occupational personal exposures to the PCA detected.

In this study, passive sampling methods were used for environmental exposure assessment of the aldehydes. It is possible that the sampling or analytical method could have impacted on the exposures measured in this study. Lower aldehyde levels than measured in this study have been reported previously when active sampling methods were used (Tucker, 2014, 2008). Environmental conditions such as temperature, relative humidity, ozone and air movements are well known factors that can affect the performance of passive samplers in measuring airborne aldehyde concentrations but were unlikely to have systematically biased results in this study since the measurements were collected on workers who worked indoors with stable ambient environmental conditions (temperature 22–30°C, relative humidity 40–68%). While ozone measurements have also been conducted in some studies of exposures to cleaning agents to assess the contribution of ozone interference during sampling (Mullen et al., 2013), these were not done due to resource constraints. Furthermore, since it is possible that information obtained from interviewing nurse managers regarding skin and asthma symptoms of individual workers may be subject to information reporting bias, a more detailed epidemiological study of these HWs provides comprehensive information on work-related symptoms and their relationship to OPA sensitization (Mwanga et al., 2022).

In conclusion, this study demonstrated a wide variety of chemicals used for cleaning and disinfection in Southern African hospital settings, some of which are known to cause or aggravate respiratory and skin diseases such as asthma and contact dermatitis. The study also confirmed that workplace controls for reducing exposure to cleaning agents were sub-optimal. The mean detectable exposures to OPA were more isolated to certain departments and are dependent on the HW’s dedicated use of OPA. This highlights the need to promote safer disinfectant use in these settings (Gamba et al., 2020). Furthermore, there is a need for more standardized, sensitive and validated assays for the determination of chlorhexidine and its metabolites in biological monitoring of HWs given its widespread use in the sector.

Funding

South African Medical Research Council, Millennium Promise Programme (University of Michigan/Fogarty International Center), Allergy Society of South Africa & National Research Foundation of South Africa.

Conflict of Interest

The authors declare no conflict of interest.

Data Availability

The anonymous data underlying this article will be shared on reasonable request to the corresponding author.