Workplace Determinants of Endotoxin Exposure in Dental Healthcare Facilities in South Africa

Abstract

Objectives: Aerosols generated during dental procedures have been reported to contain endotoxin as a result of bacterial contamination of dental unit water lines. This study investigated the determinants of airborne endotoxin exposure in dental healthcare settings.

Methods: The study population included dental personnel (n = 454) from five academic dental institutions in South Africa. Personal air samples (n = 413) in various dental jobs and water samples (n = 403) from dental handpieces and basin taps were collected. The chromogenic-1000 limulus amebocyte lysate assay was used to determine endotoxin levels. Exposure metrics were developed on the basis of individually measured exposures and average levels within each job category. Analysis of variance and multivariate linear regression models were constructed to ascertain the determinants of exposure in the dental group.

Results: There was a 2-fold variation in personal airborne endotoxin from the least exposed (administration) to the most exposed (laboratory) jobs (geometric mean levels: 2.38 versus 5.63 EU m⁻³). Three percent of personal samples were above DECOS recommended exposure limit (50 EU m⁻³). In the univariate linear models, the age of the dental units explained the most variability observed in the personal air samples (R² = 0.20, P < 0.001), followed by the season of the year (R² = 0.11, P < 0.001). Other variables such as institution and total number of dental units per institution also explained a modest degree of variability. A multivariate model explaining the greatest variability (adjusted R² = 0.40, P < 0.001) included: the age of institution buildings, total number of dental units per institution, ambient temperature, ambient air velocity, endotoxin levels in water, job category (staff versus students), dental unit model type and age of dental unit.

Conclusions: Apart from job type, dental unit characteristics are important predictors of airborne endotoxin levels in this setting.

INTRODUCTION

Dental unit waterlines (DUWLs) are commonly contaminated with Gram-negative bacteria, which release endotoxin during bacterial growth phase and upon lysis (Fulford et al., 2004; Dutil et al., 2009). A previous study investigating DUWLs in South African dental clinics showed high levels of Legionella, a Gram-negative bacterium (Singh and Coogan, 2005). Despite numerous reports of Gram-negative bacteria in dental water, there are few published reports investigating endotoxin concentrations in DUWLs and the indoor air of dental clinics (Putnins et al., 2001; Fulford et al., 2004; Szymanska, 2005a,b; Huntington et al., 2007; Puttaiah and Cederberg, 2008; Dutil et al., 2009).

During dental procedures with high-speed dental instruments, bioaerosols composed of particles varying in size are produced. Some particles evaporate forming droplet nuclei <5 μm in diameter that remain airborne for many hours (Szymanska, 2005b). Detectable levels of endotoxin have been reported in air samples obtained from dental institutions with some as high as 625 EU m⁻³ (Szymanska, 2005a). However, in the study by Huntington et al. (2007), the airborne endotoxin levels did not exceed 2.7 EU m⁻³ despite the endotoxin levels in the DUWLs being high (15 000 EU ml⁻¹). Similarly, low personal exposure levels (0.63 EU m⁻³) in a simulated dental treatment room were recently reported by Dutil et al. (2009).

Of potentially greater significance than the occasional endotoxin exposure of patients undergoing dental healthcare is the chronic low dose exposure of those involved in the dental profession. Chronic endotoxin inhalation represents an occupational respiratory hazard to this group of professionals (Huntington et al., 2007). Exposure–response relationships between endotoxin and lung function decline have been reported in a number of studies (Kennedy et al., 1987; Sigsgaard et al., 1992; Post et al., 1998; Zock et al., 1998; Smit et al., 2005). Workplace determinants of endotoxin exposure that could be used to set priorities for the reduction of endotoxin exposure in the dental setting have not been previously investigated. A few studies in farming (Preller et al., 1995; Vogelzang et al., 1998) and the domestic environment (Park et al., 2001) have investigated determinants of endotoxin exposure that could be used for application in the development of control measures.

The aim of this study was to conduct a detailed characterization of endotoxin exposures in academic dental care facilities in South Africa and to investigate the determinants of endotoxin exposure variability in these facilities.

METHODS

Sampling strategy

A cross-sectional study of currently employed dental healthcare workers (clinical, administration, laboratory and auxiliary) and registered dental students was conducted in five academic dental institutions in South Africa. There were 454 who were initially selected for personal sampling, representing 25% of all individuals working in these institutions. The institutions are designated as A, B, C, D, E and F to maintain anonymity.

The different dental clinics within the abovementioned institutions surveyed included Maxillofacial and Oral Surgery clinic, Restorative Dentistry or Conservation clinic, Periodontology and Oral Medicine clinic, Prosthetics and Prosthodontics clinic, Orthodontics clinic and Dental Therapy and Oral Hygiene clinic. A total of 14 different job titles were identified. The job titles included the Bachelor of Dental Therapy students (× 3 categories), Bachelor of Dental Science students (× 3 categories), dental assistants, dental technicians, dental nurses, oral hygienists, dentists, administration, security and cleaning staff. Participants were proportionally sampled so that each job type was adequately represented in each exposure group. On average, staff worked 8 h (08:00–17:00 h) a day, whereas the student's day is divided into lectures and clinic sessions, for which the timetable varies for each institution. However, the average clinic session is 2.5 h a day. Students are expected to work in the clinics for an average of 35 h per semester in the third, 39 h in the fourth and 42 h in the fifth years of study. The clinics experience quiescent periods during vacation breaks where major servicing and cleaning of the dental units take place.

Walkthrough inspection

A registered occupational hygienist performed a walkthrough inspection of clinics at each institution. Information regarding the age of the building, recent renovations to the building, model of the dental units, age of dental units, number of floors per dental clinic and type of water supply to the dental unit (municipal or reservoir) were obtained.

Indoor air assessment for microclimatic parameters

Ambient daily temperatures (measured at a height of 1.8 m), percent relative humidity (%RH) values and CO₂ levels were measured using the TSI Q TRAK Model 8551 (SKC Inc. Pittsburg, PA, USA). Air velocity in the work areas was measured using the TSI Velocicalc plus model 8388-M-GB (SKC Inc. Pittsburg, PA, USA). The sampling for Institution A was conducted in summer, Institution B in autumn, Institutions C and D in winter and Institutions E and F in the spring season.

Air sampling

Personal inhalable air samples were collected on dental workers performing clinic sessions during the working day. Samples were collected using preloaded three-piece 37-mm clear styrene closed faced cassettes with pre-sterilized 0.4 μm-polycarbonate (PC) filters and endotoxin-free backing pads. A sampling train was set up with the filter cassette fixed to the lapel of the participants’ laboratory coat or collar. Gillian (SKC Inc., Pittsburg, PA, USA) or Casella Apex (Casella, USA) battery-operated air-sampling pumps set at a flow rate of 2.5 l min⁻¹ were used (Thorne et al., 1997; Technical Committee and CEN/TC 137, 2003). Personal sampling was conducted for ∼150 min depending on the duration of the clinic session. During the sampling period, any unusual occurrences were noted. After sampling, the filters were sealed and stored in plastic bags at ∼4°C for a week until transported to the National Institute for Occupational Health (NIOH) laboratory for immediate extraction and subsequent endotoxin analyses. At least one field blank was taken on each field sampling day and was included in the analyses to assess any contamination of the filter cassettes used. Sampling pumps were calibrated before and after sampling using a Gillibrator bubble flow meter (Gillian Instrument Corp., Wayne, NJ, USA) to verify that the air-sampling rate had been constant (within the 5% allowed variation) throughout the sampling period. If the flow rate was >5% below the initial flow rate, the sample was not included in the analysis (NIOSH, 1998).

Water sampling

Water samples (5–10 ml volume) were aseptically collected in pyrogen-free 10-ml glass bottles (Lonza, Walkersville, MD, USA) from three-in-one dental handpieces and basin taps. A total of 403 water samples were collected. Samples were kept at 4°C and were analyzed within 24 h of receipt by the laboratory.

Laboratory analysis of samples

Extraction procedure.

Air filter samples were extracted in 6 ml pyrogen-free water (Lonza, Walkersville, MD, USA) and 0.05% Tween 20 for 3 h at room temperature with gentle agitation. The samples were filtered using 0.45-μm Acrodisc syringe filters (Pall Corporation, NY, USA). The extracts were then concentrated (Baur, 2002) using Cryodos −50°C freeze dryer (Telstar Cryodos, Spain) and reconstituted in 1-ml pyrogen-free water. The procedure was done according to the method described by Baur (Baur, 2002). The suspension was stored at −70°C for a period of ∼30 days before analysis. According to Spaan et al., prolonged duration of specimen storage does not appear to have a major impact on the integrity of the specimen (Spaan et al., 2007). To test the effectiveness of the extraction procedure, sterile filters were spiked with known concentrations of endotoxin, extracted as described above and then assayed. The results were comparable to the spiked concentration.

Determination of endotoxin concentration.

The chromogenic-1000 limulus amebocyte lysate endpoint assay (Lonza, Walkersville, MD, USA) was used. Throughout the analytical process, pyrogen-free products including water were used. Endotoxin analysis was done according to the manufacturer's instructions. Pyrogen-free water was analyzed as the negative controls and 0.4 EU ml⁻¹ Escherichia coli for the positive control. The concentrations for the standard curve included 0.10, 0.25, 0.50 and 1.00 EU ml⁻¹ and were prepared using the E. coli provided in the kit. The plate was read using a microplate reader (Biotek instruments, VT, USA) at 405 nm. Change in absorbance relative to the assay reagent blank was calculated, and a standard curve of delta absorbance versus endotoxin activity was generated using KC4 EL_x808 software (Biotek instruments, USA). Assays in which the standard curve had a correlation coefficient ≥0.98 and samples with a coefficient of variation ≤10% were accepted. Using the calculated volume of air sampled and endotoxin concentrations, the endotoxin units per cubic meter (EU m⁻³) were calculated. The formula used was: [EU ml⁻¹ × sample volume (ml)]/[time (min) × (rate (l min⁻¹) × 1 m³/1000 l] (Lonza Inc, 2001).

The limit of detection (LOD) = 0.03 EU ml⁻¹ was computed by calculating the mean of the field blanks. The LOD denotes the lowest level above which endotoxin has been detected during sampling. The limit of quantification (LOQ) = 0.08 EU ml⁻¹ was calculated by adding the mean plus three times the standard deviation of the field blanks. LOQ denotes the limit above which there is confidence in the value measured. For data analyses, values for exposure measurements below the LOD were replaced by the analytic LOD divided by the square root of two () (Finkelstein and Verma, 2001). Interference of the samples was assessed by spike recovery of endotoxin in samples that were diluted 10-fold. No interference was noted in the analysis.

Statistical analysis

Statistical analysis was performed using Stata9 computer software (StataCorp, 2007, TX, USA). Descriptive univariate statistics were generated for the total sample distribution. The individual endotoxin data points were log transformed to meet the model assumptions of normally distributed variables with equal variances. Job titles were grouped into similarly exposed groups (SEGs) based on their tasks and associated exposures. Dental Therapy students and Bachelor of Dental Science students (third to fifth year) formed a group; another group included clinical staff (dental assistants, dental nurses, oral hygienists, dentists and dental technicians); administration was a group on its own, whereas security and cleaning staff formed the fourth group. Each exposure group consisted of a number of job titles that were deemed by the study team to represent similar tasks and involve similar exposures based on information from self-administered questionnaires. Analysis of variance (with Bonferroni correction) was used to test for the difference between exposure groups.

Analysis of variance was used to explore the between- and within-variance components of endotoxin levels by institution, department and exposure groups. Univariate and multivariate linear regression was carried out to explore the determinants of variability in endotoxin exposure using log-transformed endotoxin levels as the dependent variable. Independent variables considered for entry (P < 0.1) into the model included microclimatic parameters (temperature and air velocity) and occupational factors (building age, number of dental units, institution, job type, dental unit model and dental unit age). Although not significant, endotoxin levels in water were included in the model as theoretically it could be a strong predictor of airborne endotoxin exposure as reported previously (Szymanska, 2005b).

RESULTS

Characteristics of data

The total number of individuals from each institution that participated in the study was as follows: Institution A, 78; Institution B, 105; Institution C, 57; Institution D, 71; Institution E, 91 and Institution F, 52. Among these 454 individuals, 164 were dental students and 290 were clinical and non-clinical staff. A total of 413 personal air samples were collected from the six sampling sites. Four samples were rejected as the sampling process was interrupted. Ten participants working on prosthetics procedures did not use water during their clinic session and as a result only 403 water samples were collected from the study participant's work area.

Indoor air assessment

The measurements for microclimatic factors during the sampling periods for different institutions are presented in Table 1. Variation in temperature and humidity observed among the institutions may be due to seasonal variability. CO₂ measurements did not vary across the sampling period (data not shown).

Table 1.

Endotoxin levels in air and water and microclimatic air quality in academic dental clinics

Sampling sites	Sampling period	Mean endotoxin levels, GM (GSD)			Ambient parameter: mean (SD)
		Personal air (EU m−3) N = 413	Area air (EU m−3) N = 116	Water (EU ml−1) N = 403	Temperature (°C) N = 183	Humidity (%) N = 183	Air velocity (m s−1) N = 307
Institution A	Summer	1.5 (2.2)	1.0 (2.6)	222 (10.7)	23 (0.96)	48 (1.86)	0.07 (0.02)
Institution B	Autumn	6.5 (4.0)	1.7 (6.2)	1.7 (6.2)	21 (2.10)	47 (4.56)	0.06 (0.02)
Institution C	Winter	2.9 (2.3)	0.8 (5.6)	320 (2.7)	12 (2.89)	15 (3.85)	0.01 (0.003)
Institution D	Winter	3.3 (2.5)	0.5 (3.7)	44 (7.1)	11 (0.92)	22 (1.85)	0.04 (0.01)
Institution E	Spring	8.9 (3.1)	0.5 (3.0)	41 (1.7)	12 (0.32)	17 (1.17)	0.04 (0.01)
Institution F	Spring	4.5 (2.1)	0.6 (2.7)	15 (4.3)	12 (0.52)	21 (1.04)	0.08 (0.04)

GM, geometric mean; GSD, geometric standard deviation; SD, standard deviation. Normal data is presented for ambient microclimatic parameters.

Endotoxin exposure levels

The overall mean airborne endotoxin levels were 4.1 EU m⁻³ (GSD = 3.3 EU m⁻³) and the concentration in water samples was 101 EU ml⁻¹ (GSD = 6.0 EU ml⁻¹) (Table 2). The highest airborne endotoxin levels were observed in the laboratory and clinical departments. Eleven personal air endotoxin samples were above the DECOS recommended exposure limit (50 EU m⁻³) (Heederik and Douwes, 1997).

Table 2.

Characteristics of mean exposure levels of endotoxin in air (EU m⁻³) and water (EU ml⁻¹) for overall exposure and the various departments

Exposure matrix	Personal air samples (EU m−3)				Water samples (EU ml−1)
	n	GM	GSD	Range	n	GM	GSD	Range
Overall	413	4.1	3.3	0.1–555.9	403	101	6.0	1–100000
Clinical	316	4.3	3.4	0.1–117	314	109	6.2	1–100000
Laboratory	22	5.6	5.4	0.4–555.9	19	66	5.3	1–638
Administration	28	2.4	2.7	0.3–20.2	26	50	5.2	2–816
Auxiliary	47	3.6	2.5	0.5–34.5	44	104	5.4	2–74800

n, Number of measurements; GM, geometric mean; GSD, geometric standard deviation; range: natural data.

Table 3 outlines endotoxin levels by department and exposure group for all institutions. Personal endotoxin levels differed significantly by institution (P-value < 0.001) and were highest for Institution E (GM: 8.9 EU m⁻³ and GSD: 3.5 EU m⁻³). The laboratory had the highest mean levels when compared to the other departments.

Table 3.

Endotoxin in air (EU m⁻³) in the various institutions stratified by department and exposure group

Department and exposure group	Personal air endotoxin concentrations (EU m−3)
	Institution A			Institution B			Institution C			Institution D			Institution E			Institution F
	n	GM (GSD)	Range	n	GM (GSD)	Range	n	GM (GSD)	Range	n	GM (GSD)	Range	n	GM (GSD)	Range	n	GM (GSD)	Range
Overall	77	1.5 (2.2)	0.2–27.8	91	6.5 (4.0)	0.3–555.9	49	2.9 (2.3)	0.4–19.3	63	3.3 (2.7)	0.3–53.3	82	8.9 (3.5)	0.1–73.6	51	4.5 (2.2)	1.5–35.5
Clinical	60	1.6 (2.3)	0.2–27.8	63	7.2 (3.7)	0.3–117	38	2.4 (2.2)	0.4–19.3	47	3.6 (2.7)	0.3–53.3	72	9.3 (3.1)	0.1–73.6	36	4.3 (2.1)	1.5–17.5
Maxillo-facial and oral surgery	13	1.4 (1.9)	0.3–3.2	17	5.4 (2.6)	0.8–42	8	2.7 (2.7)	0.9–19.3	10	2.7 (2.2)	0.5–53.3	18	4.9 (3.4)	0.1–31.5	5	3.3 (1.4)	2.4–17.4
Restorative/conservation	30	1.9 (2.1)	0.9–27.8	25	10.9 (3.8)	1.2–117	19	2.6 (2.2)	0.4–7.1	3	6.1 (2.3)	2.5–12.6	27	15.2 (2.6)	2.1–62.7	13	3.5 (2.0)	1.5–16
Periodontology and oral medicine	1	0.8 (—)	—	1	6.6 (—)	—	1	0.9 (—)	—	—	—	—	6	6.1 (3.0)	1.8–24.3	4	3.2 (1.5)	2.3–5.5
Prosthodontics/prosthetics	1	2.4 (—)	—	2	2.1 (1.6)	1.5–2.8	1	3.5 (—)	—	—	—	—	—	—	—	—	—	—
Orthodontics	2	1.4 (2.2)	0.8–2.4	2	8.7 (3.8)	3.4–22.4	2	1.3 (2.0)	0.8–2.0	—	—	—	5	10.8 (4.5)	1.8–24.3	2	10 (2.1)	5.9–17
Dental therapy and hygiene	8	2.2 (2.2)	1.4–14.3	10	9.0 (5.3)	0.3–94.9	5	2.9 (1.5)	1.6–4.9	31	4.2 (2.7)	0.3–47.6	7	14.3 (2.7)	1.6–28.4	6	8.2 (2.1)	2.5–17.5
Sterilization	5	0.7 (3.3)	0.2–2.4	6	3.0 (3.7)	0.8–32.2	2	1.7 (2.3)	0.9–3.1	3	1.4 (2.1)	0.7–2.9	9	6.8 (3.1)	1.4–73.6	6	4.1 (2.5)	1.6–10.9
Laboratory	6	1.1 (2.0)	0.4–2.7	4	43.7 (11.8)	3.9–555.9	2	7.7 (2.7)	3.8–15.6	2	4.2 (1.2)	3.7–4.8	3	6.3 (2.3)	3.1–16.2	5	6.6 (2.7)	2.8–35.5
Technicians lab	6	1.1 (2.0)	0.4–2.7	4	43.7 (11.8)	3.9–555.9	2	7.7 (2.7)	3.8–15.6	2	4.2 (1.2)	3.7–4.8	3	6.3 (2.3)	3.1–16.2	5	6.6 (2.7)	2.8–35.5
Administration	4	0.7 (2.1)	0.3–1.7	9	2.3 (3.1)	0.4–20.2	—	—	—	5	2.6 (2.2)	1.3–9.1	3	3.5 (2.3)	1.5–8.1	7	4.0 (2.0)	1.9–14.8
Clerical	3	0.8 (2.5)	0.3–1.7	7	2.5 (3.3)	0.4–20.2	—	—	—	4	2.9 (2.4)	1.3–9.1	2	5.2 (1.9)	3.4–8.1	5	3.5 (2.7)	2.8–35.5
Reception/switchboard	—	—	—	2	1.7 (3.9)	0.7–4.5	—	—	—	1	1.9 (—)	—	1	1.5 (—)	—	2	5.2 (1.4)	2.5–6.1
Stores	1	0.5 (—)	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—
Auxiliary	7	1.4 (1.7)	1.0–2.7	15	4.7 (2.2)	1.3–20.9	9	4.6 (2.3)	1.6–15.0	9	2.1 (1.8)	0.9–4.9	4	9.6 (2.4)	5.3–34.5	3	4.9 (3.1)	2.5–18.1
Security	—	—	—	2	12.7 (2.0)	7.7–20.9	—	—	—	2	2.4 (1.7)	1.6–3.5	1	8.8 (—)	—	2	6.8 (4.0)	2.5–18.1
Cleaning	7	1.4 (1.7)	1.0–2.7	13	4.0 (2.0)	1.3–13.8	9	4.6 (2.3)	1.6–15.0	7	2.0 (1.9)	0.9–4.9	3	10.0 (2.9)	5.3–34.5	1	2.6 (—)	—

n, Number of measurements; GM, geometric mean; GSD, geometric standard deviation; where one sample was collected the standard deviation was not calculated; range: natural data.

Variance components

Table 4 outlines the variance components of endotoxin levels for the exposure grouping strategies of interest. The between-group variability was significantly lower than the within-group variability for all grouping strategies. Analysis of variability according to department revealed that airborne endotoxin levels were significantly lower in the administration and auxiliary departments as compared to either the clinical department or the laboratory. However, the endotoxin levels were not significantly different between the clinic and laboratory (P > 0.05), as well as the administration and auxiliary departments.

Table 4.

Between-group and within-group Variance components of endotoxin in air based on log-transformed data (n = 413) for various exposure grouping strategies

	Variance components		P-value
	Between group	Within group
Institution	0.46	1.06	<0.001
Department	0.05	1.44	0.036
Exposure group	0.05	1.42	0.016
Job type	0.16	1.30	<0.001
Tasks	0.04	1.42	0.034

Determinants of exposure and exposure variability

Among the occupational factors investigated in the univariate regression models, the following positive relationships were observed. The age of the dental units explained the most variability observed in the personal air samples (R² = 0.20, P < 0.001) increasing in endotoxin with aging dental units, followed by sampling period (R² = 0.11, P < 0.001) where endotoxin is higher in the summer season (Table 5). The institution and total number of dental units per institution also explained a modest degree of variability in airborne endotoxin exposure. In the latter case, endotoxin increases with increasing numbers of dental units. Based on the variance difference between the job types, the 14 job titles were collapsed into four SEGs; students, clinical staff, administrative-related work and auxiliary services. Job titles explained 3% of the variance in endotoxin levels, among which dental students appear to be most exposed to endotoxin. Other explanatory variables such as water source explained a much lower variability in endotoxin concentration.

Table 5.

Exposure determinants of personal airborne endotoxin concentrations (log transformed, ln) among dental healthcare workers in univariate regression models

Variable	Personal endotoxin level, ln (EU m−3)
	β (SE)	R2	P-value
Microclimatic parameters (n = 116)
Temperature (°C)	−0.05 (0.03)	0.02	0.051
Humidity (%)	−0.03 (0.01)	0.05	0.006
Air velocity (m/s)	−8.41 (2.14)	0.05	<0.001
CO2 (p.p.m.)	0.001 (0.001)	0.02	0.155
Season	0.25 (0.04)	0.11	<0.001
Sampling day	0.06 (0.05)	0.01	0.298
Occupational factors (n = 413)
Institution	0.20 (0.03)	0.08	<0.001
Building age (years)	0.01 (0.01)	0.03	<0.001
Departments	−0.10 (0.06)	0.01	0.078
Exposure group	−0.02 (0.02)	0.01	0.159
Job typea	0.24 (0.07)	0.03	<0.001
Tasks	−0.04 (0.02)	0.01	0.043
Dental student versus clinical staffb	0.30 (0.13)	0.01	0.020
Dental unit model	0.20 (0.05)	0.06	<0.001
Total number of dental units per institution	0.01 (0.001)	0.08	<0.001
Age of dental units (years)	0.06 (0.01)	0.20	<0.001
Number of patients seen	−0.01 (0.01)	0.03	0.009
Municipal versus reservoir water source	−0.09 (0.15)	0.01	0.567
DUWL versus tap water	0.25 (0.14)	0.01	0.073
Endotoxin in water (EU ml−1)	−0.05 (0.03)	0.01	0.111

The value in each cell represents a separate regression model.

^aFourteen job types were classified into one of four SEGs (clinical staff, students, administration and auxiliary)

^bClinical staff used as baseline for comparison.

The final multivariate regression model explained 40% of the variability in the level of endotoxin in the personal air samples (model: adjusted R² = 0.40, P < 0.001, n = 100) (Table 6). Variables that explained the greatest variability of airborne endotoxin in this model were the age of institution buildings, total number of dental units per institution, air velocity, temperature, job category (staff versus student), age of the dental unit and the dental unit model. Apart from job category, dental characteristics (e.g. age of dental unit and certain dental unit models) and institutional characteristics (e.g. working in older buildings, number of dental units per institution) were identified as important determinants of airborne endotoxin exposure. Measurements with missing data for a specific factor were omitted from the analysis resulting in a reduction of the total number of measurements in the model. This pertained to jobs with non-clinical functions, as these subjects were not assigned to a dental unit.

Table 6.

Multivariate linear regression for personal airborne endotoxin concentrations (log transformed, ln) among dental healthcare workers (n = 100)

Variable	Personal Endotoxin Level, Ln (EU m−3)
	β (SE)
Intercept	−25.39 (9.00)*
Building age (years)	0.15 (0.06)*
Total number of dental units	0.05 (0.02)**
Air velocity (m s−1)	−22.65 (7.80)*
Temperature (°C)	0.36 (0.17)*
Log (endotoxin in water) (EU ml−1)	−0.06 (0.06)
Dental student versus clinical staffa	1.04 (0.29)**
Dental unit model: Aidecb	8.25 (2.32)**
Dental unit model: Promasb	7.20 (2.30)*
Dental unit model: Rittersb	−15.34 (4.38)**
Dental unit age (years)	0.88 (0.24)**

^aClinical staff used as baseline for comparison.

^bDental unit model: Pelton used as base of comparison (lowest endotoxin levels in DUWLs).

*P < 0.05; **P < 0.001.

The final model satisfied post-regression diagnostic tests for linearity, normality and constant variance. Also, an omitted variable test demonstrated that the model was adequately specified. The model can be expressed as: Log (endotoxin personal concentration) = β0 + β1 building age + β2 total number of dental units + β3 air velocity in work area + β4 ambient temperature in work area + β5 log (endotoxin in water) + β6 job type + β7 dental unit model + β8 age of dental unit.

DISCUSSION

The findings of this study demonstrated considerably high concentrations (100 000 EU ml⁻¹) of endotoxin in the DUWLs as well as airborne endotoxin (555.9 EU m⁻³) in certain areas of the dental clinic environment compared to previous studies (Huntington et al., 2007; Dutil et al., 2009). The average endotoxin level for the administration group was 3.84 EU m⁻³, which corresponds with the mean endotoxin background level of 0.063–3.600 EU m⁻³ reported for background environments (Madsen, 2006). It is evident that the administration department (unexposed grouping descriptor) demonstrated less exposure, which confirms that DUWLs are a potential source of endotoxin exposure. These results are consistent with other studies that demonstrate that DUWLs are an important source of bacterial endotoxin contaminating the dental clinic environment (Fulford et al., 2004; Huntington et al., 2007). The increased levels within the DUWLs may be a result of quiescent periods experienced during weekends, study breaks and vacation, increased temperatures, lack of disinfection procedures, poor water quality, small bore size of DUWLs and low flow rate (<60 ml min⁻¹) (Putnins et al., 2001; Franco et al., 2005).

Although the majority of personal air endotoxin concentrations found in this study were <50 EU m⁻³, exposures considered safe for human health (Heederik and Douwes, 1997), a small proportion (3%) was above this level. It is conceivable that the low-grade concentrations of endotoxin measured in this study may still be clinically relevant as chronic endotoxin inhalation is regarded as an occupational hazard in various settings (Reed and Milton, 2001; Huntington et al., 2007). Endotoxins have not only been implicated in organic toxic dust syndrome but may also act as an adjuvant facilitating sensitization to other allergens or they may increase the severity of allergic disease (Boehlecke et al., 2003; Szymanska, 2005a; Huntington et al., 2007). Furthermore, among the 3% >50 EU m⁻³, five samples were >90 EU m⁻³, the level suggested as a threshold for acute mucous membrane irritation and pulmonary effects among cotton workers (Haglind and Rylander, 1984). These elevated measurements reflect a potential for intermittent peak exposure of dental personnel to levels that may induce acute endotoxin-related effects. While low-grade endotoxin levels were predominant, the within group variability suggests that endotoxins are components of growing microorganisms in a dynamic ecosystem contributing to the variation observed (Spaan et al., 2008). Putin et al. also reported differences in contamination within and between dental units confirming the variability of endotoxin exposure in this setting (Putnins et al., 2001). The levels within DUWLs may decrease during flushing and decontamination treatments; however, the waterlines on the other hand are conducive to the promulgation of microorganisms during favorable conditions, causing the endotoxin levels to rise. For this reason, the relationship between chronic inhalation exposure and respiratory disease in this group of professionals needs further investigation.

An important finding from this study is the identification of factors that determine elevated endotoxin levels in the dental setting. In this study, a number of exposure determinants were significantly associated with increasing the endotoxin levels in dental clinic settings, with the identified factors explaining 40% of variability in airborne endotoxin levels. Determinants of endotoxin levels that have practical implications for the dental clinics are the dental unit characteristics such as the model of the dental units, the age of the dental units and the total number of units per institution, as well as microclimatic factors such as temperature. One of the four dental unit models most commonly used within the dental institutions showed a protective effect. Although the age of dental units is an important determinant of endotoxin exposure, the age of the dental unit in question was similar to the other dental unit models in use (∼25 years old). Therefore, it is likely that other factors regarding the dental unit model (e.g. type of materials used in the tubing, bore size of the tubing and the frequency of use) (Challacombe and Fernandes, 1995) may have influenced the endotoxin levels. These factors were not examined in detail in the current study. However, the information gathered using the building checklist showed that the Ritter model operated at a low flow rate (75 ml min⁻¹) compared to other models (∼150 ml min⁻¹). Therefore, it is possible that less biofilm was flushed out of the waterlines during dental procedures, leading to lower airborne endotoxin concentrations when using the Ritter model. The other factor that might have influenced the protective effect of the latter model is that the waterlines were disinfected quarterly, whereas the other models were only decontaminated on the surfaces; therefore; work practices seem to play a major role.

Another factor with important implications for dental practice involves dental students or apprentices and their work in dental clinics. Students had on average higher levels of endotoxin during their practical sessions than staffs irrespective of whether the latter were clinical or non-clinical in nature. During clinic workplace inspections, it was evident that the students were supplied with inappropriate respiratory protective equipment. Students usually work in pairs, with the assisting student working in close proximity to the DUWL, the latter not being provided with suitable personal protective equipment (PPE). However, there were considerable fluctuations in the levels of endotoxin among the dental students by year of study, which may be related to the type of dental procedure being performed at the time of sampling. Certain tasks such as Restorative dentistry are associated with increased volumes of water being used for certain procedures (Harrel, 2003; Agostinho et al., 2004). It is possible that since clinical staff performs supervisory roles and are not sedentary for most of the clinic session, they may be exposed to lower levels of endotoxin. These findings highlight the need for adequate measures to reduce exposure levels to student trainees. PPE and training on good hygiene practices should be enforced to prevent endotoxin exposure and improve dental healthcare worker behavior.

Another factor for consideration in academic settings, public and military dental clinics are the large number of dental units per clinic since increasing the number of dental units appears to increase the endotoxin levels. Previous studies have demonstrated an increase in bacterial air contamination following dental treatment (Miller et al., 1971; Grenier, 1995). Increasing the number of dental units per clinic would result in an increase in the number of dental treatments being performed in the clinic and result in an increase in airborne microbial load if infection control procedures are not adhered to. In addition, aerosols generated during dental procedures have the capacity to spread throughout multichair dental clinic environments, even into areas where there are no dental activities (Grenier, 1995), thus posing a potential risk to both clinical and non-clinical staff.

In this study, aging dental units and aging buildings were shown to increase the endotoxin levels. It is well known that dental units provide an environment conducive to the rapid proliferation of microorganisms. In the absence of an adequate infection control regimen, the microorganisms would increase to unacceptable levels (Szymanska, 2003). The age of a building may be a proxy for many other exposure determinants. Older buildings may have less efficient ventilation systems, contaminated air-conditioning systems which recirculate microorganisms in the air or they may have old water distribution pipes which may be colonized with microorganisms, providing a niche for endotoxin growth.

A microclimatic factor such as indoor temperature was shown to be a good predictor of airborne endotoxin levels in dental settings, whereas air velocity can be a key element in a reduction strategy to control endotoxin levels. These findings suggest that increasing ventilation (e.g. natural ventilation, well maintained air-conditioning systems) would assist in reducing the endotoxin load in dental setting environments.

The exposure conditions described in this study are reasonably representative of public sector dental clinics in the country since all dental schools in South Africa were selected for this study. To our knowledge, this is the first study to report important dental institutional characteristics that determine endotoxin exposure levels in dental clinic settings. The findings of this study provide a basis for designing dental clinics that take into cognizance the relative importance of certain factors over others in reducing endotoxin exposure. One of the limitations of the study was not being able to do repeat measurements to ascertain the variability within individuals. However, this study provides a good basis for determining inter-group predictors of endotoxin exposure.

In conclusion, this study identified several workplace determinants that predict endotoxin exposure in dental settings. Future studies should assess the value of these identified predictors in reducing the endotoxin levels within dental settings. The relevance of low-grade endotoxin exposure in causing adverse respiratory health effects in this context also warrants further investigation.

FUNDING

Fogarty International Center, University of Michigan, National Institutes of Health (2 D43 TW00812-06); Allergy Society of South Africa-Glaxo Smith Kline research award; National Institute for Occupational Health, National Health Laboratory Services.