Skulberg, Knut R.*; Skyberg, Knut*; Kruse, Kristian*; Eduard, Wijnand*; Djupesland, Per†; Levy, Finn†; Kjuus, Helge*
Health problems related to the indoor environment have received increasing attention during the past 20 years. In 1982, a World Health Organization expert group described “the Sick Building Syndrome” as a combination of general symptoms, mucosal irritation symptoms, and skin symptoms. Several large cross-sectional studies reporting symptom prevalence and risk factors in office workers have been published.1,2 Female sex, atopy, low job status, unfavorable psychosocial load, and work with video display units are some of the risk factors for the syndrome. However, Sick Building Syndrome is not an accepted clinical syndrome, and its causes and biologic mechanisms are unclear.3
Exposure to dust in the indoor office environment could be one of the causes of mucosal irritation symptoms. Exposure chamber challenge of humans has shown a relation between airborne dust and symptoms.4 An association between office dust and eye complaints has also been found.5 Saraf et al.6 observed that house dust induced the production of interleukin-8 and interleukin-6 in lung epithelial cells in vitro. An association between the amount of settled dust and symptoms was found in the large cross-sectional “Danish Town Hall Study.”1 In an extension of the same cross-sectional study, Gyntelberg and coworkers7 found correlations of Sick Building Syndrome with Gram-negative bacteria, settled dust concentration, large dust particles. and volatile organic compounds in dust. Wålinder et al.,8 using acoustic rhinometry in a cross-sectional study, observed that nasal mucosal swelling in school workers was related to a low air exchange.
In a controlled and blind intervention study, Kemp et al.9 found that Sick Building Syndrome symptoms (in particular, irritation of the mucosal membranes) could be reduced by using higher performance carpet cleaning. However, this study lacked measurements of exposure to airborne dust and had no objective health indicators. Franke et al.10 found that a very thorough cleaning of offices could reduce the airborne dust concentration, total volatile organic compounds, and culturable fungi and bacteria. Kildesø et al.11 compared several cleaning methods in an intervention study. They found that cleaning had a small effect on the dust level and composition of the dust. However, other studies have shown that cleaning can reduce the indoor dust concentration and change the composition of the dust.9,12,13
The aim of this study was to test the hypothesis that the removal of dust by cleaning offices can reduce workers’ mucous membrane symptoms and nasal congestion. The study was carried out as a double-blind controlled intervention study, and objective methods were used to assess dust exposure. We measured airborne dust with stationary and portable person-borne equipment before and after a comprehensive cleaning. To improve the reliability of the study, we also recorded nasal congestion as an objective health indicator.
This study was performed in 1998 in 1 large office building holding approximately 1600 employees in Oslo, Norway (Fig. 1). The building was erected in 1986 and is located approximately 6 km from the center of Oslo, close to a road with heavy traffic. Offices in the building had a median dust concentration of 75 μg/m3 (particle size range 0.5–40 μm diameter, measured stationary with an optical particle counter).14 Site intervention was carried out in the late winter season (January through March). All measurements were done before the birch pollen season started, but after the start of hazel and alder pollen release. The study was approved by the Regional Medical Ethics Committee. All participants received a written description of the study and informed consent was collected. Participation was voluntary.
A short screening questionnaire, recording age, sex, smoking status, and mucosal irritation symptoms, was distributed to all employees. Participants in the intervention study were selected from among the 967 employees who completed the questionnaire. Inclusion criteria were nonsmoker, single occupant in an office, and moderate to severe mucosal irritation symptoms (at least 3 points on a 0–8 mucosal irritation scale). In addition, persons taking part should not plan to be absent from the workplace for more than 3 days during the intervention period.
Sample sizes for 2 levels of statistical power were calculated, using the intraindividual differences in symptoms before and after intervention. For 80% statistical power, 28 participants were needed to detect a 0.5 point change in the mucosal irritation symptom index; the corresponding figure for 95% statistical power and a 0.5 point change was 61 participants (2-sided test).
A total of 118 employees participated in the study (Table 1). Fourteen participants did not complete the study because of severe upper respiratory infections (8 persons), moving out of the office (3 persons), and absence as a result of business trips (3 persons).
The participants were randomly allocated to an intervention group or a control group using group level matching by sex, level of irritation symptom index, and allergy status. The distribution of age was similar in the 2 groups. The participants and the field researchers were blinded to the group status of the participants. Information was given to all participants that cleaning would be performed. They did not know that 2 different types of cleaning would be used and that 1 group would receive “placebo” treatment. In addition, the cleaning was done in the evening after the employees had left the building. Exposure and health data were collected before the intervention and again 3 weeks after the intervention.
The intervention group got a comprehensive cleaning of all surfaces in their offices after the removal of all small objects. Carpets were vacuum-cleaned using a canister with a rotating brush head that had an ordinary bag filter and microfilter. The microfilter had a filtration efficiency of 99.98% of particles larger than 0.3 μm in diameter. The suction velocity was 31 l/sec and the negative pressure was 200 mbar. Estimated duration of floor vacuum cleaning was 10 to 15 seconds/m2. The vacuum cleaner was mounted with an “open head” for cleaning of small objects, furniture, curtains, bookshelves, and so on, done according to methods described by Schneider et al.15 A thorough cleaning of all walls, the ceiling, desks, bookshelves, and windows was also done. Any manmade mineral fiber insulation mats above the suspended-panel ceilings were removed before vacuum cleaning and mopping. Objects temporarily removed were put back in their original place after vacuuming.
The placebo group got only a superficial cleaning. This included wet-wiping of the office desk and other hard surfaces, and removal of any visible stains on carpets and furniture, but no vacuuming. Small objects and any manmade mineral fiber insulation were left untouched. Study participants were not aware of the group design and consequently were not aware of their group status.
Dust concentrations were measured in 2 ways, using personal and stationary measurements. Personal dust measurements were performed with portable equipment carried by the participants. We used a PS101 dust pump (developed at the National Institute of Occupational Health, Norway) with a flow rate of 2 l/min, connected to PAS 6 cassettes with Teflon filters. The employees carried the pumps with a filter cassette near the breathing zone during 1 ordinary working day. This equipment has been shown to collect inhalable aerosols in indoor environments.16
Seventy-seven personal dust measurements were obtained. On average, almost 40% of the participants’ working day was spent outside their own office. Mean total dust exposures were 65 μg/m3 before intervention/placebo, and this was reduced by approximately 10 μg/m3 in both groups. When restricting the analysis to employees spending most of their working day inside their own office, we found that the mean dust concentration was reduced by 25 μg/m3 in the intervention group compared with 10 μg/m3 in the control group (not shown in tables).
Stationary measurements in the offices were performed with a GRIMM optical particle counter (type 1.105, Grimm Labortechnik GmbH, Ainring, Germany).17 The optical particle counter was placed on the office desk approximately 50 cm from the video display unit, representing the working zone of the employee. Because the activity in the office can influence dust concentration,18 the measurements were executed under standardized conditions. These conditions were: windows and door were closed, the study participant was alone in the room, measurement time was 10 minutes, and half of the time was video display unit work and the rest of the time was paperwork.
Dust concentrations measured with stationary equipment are given in Table 2. We obtained measurements both before and after intervention or placebo for 63 of the 104 offices. There was an increase in the dust level of the control group and a reduction in the dust level of the intervention group. The mean difference in change between intervention and the control group before and after intervention was 37 μg/m3 (95% confidence interval [CI] = 13–61). This difference was more pronounced for offices with dust concentrations above 50 μg/m3 before the cleaning. No effect of cleaning on dust concentration was found for offices with a low dust level before the cleaning.
Twenty dust samples were analyzed for microorganisms by fluorescence microscopy.19 No bacteria or fungi were observed in samples collected before and after the cleaning. Cultivation of microorganisms in dust was not performed.
Health Effects Monitoring
A screening allergy blood test (Phadiatop, Pharmacia Diagnostica, Uppsala, Sweden) for the 9 most common respiratory allergens (birch, timothy, mugwort, house dust mite, dog, cat, horse, and 2 molds: Aspergillus and Cladosporium) was performed on 98 participants. Forty participants were Phadiatop-positive (IgE >100). Single allergen reaction tests were analyzed for those who had a positive reaction to this screening test. Most of the allergic participants reacted to pollen, whereas only 15 of 40 showed a reaction to common indoor allergens (mite, mold, cat, and dog). All participants were asked about rhinitis caused by allergens, asthma, and allergy in the family. Risk estimates between the Phadiatop test and the subjective allergic questions were calculated (see “Statistical Methods” section). The risks for subjective allergy if the Phadiatop test was positive were 22 for rhinitis (CI = 5.9–82), 3.3 for asthma (CI = 0.8–14), 3.2 for allergy in the family (CI = 1.4–7.5), and 11 for asthma and/or rhinitis (CI = 3.8–32).
Subjective health effects were recorded using a questionnaire.20 Phrasing of the questions on symptoms was: “During the last 3 weeks have you had any of the following symptoms?” Three symptom indices were constructed from the questionnaire in the following way. Each question was given a value according to the frequency of symptoms: 0 points for “never,” 1 point for “sometimes,” and 2 points for “often.” The value from each question was summarized in a skin symptom index (dry, red, itching or burning sensation of facial skin), mucosal membrane symptom index (irritation of nose, throat, eyes, or cough), and a general symptom index (fatigue, heavy-headed, headache, dizziness, or concentration problems). The possible index sum values were 0 to 8, except for general symptoms, which were 0 to 10.
Nasal congestion was measured by acoustic rhinometry.8,21–24 The following variables were recorded: the minimum cross-sectional areas and volumes of the anterior 22 mm of the nasal cavity and of the deeper cavity between 22 and 52 mm from the nostril. Measurements of nasal cavity dimensions were carried out with the participants sitting in their own offices holding their breath. Measurements were repeated until 3 similar measurements were obtained. The mean value of the 3 accepted measurements was used in the statistical analysis.
The relation between the subjective allergy recording and the Phadiatop test was quantified by odds ratios (χ2 test). Exposure and health effects changes were calculated using arithmetic means and quartiles. Multivariate analysis was done using logistic regression. The dependent variables in the models were the intraindividual differences before and after intervention for symptom indices and for nasal congestion measurements. The independent variables tested were age, sex, allergy test, group category, and precleaning values. The dependent variables were dichotomized using 2 cut-off points. Odds ratios and 95% CIs are given.
There was a reduction in the irritation and general symptom indices in the intervention group compared with the control group after the intervention (Table 3). The median decline in irritation symptoms in the intervention group was 1 point, compared with no change in the control group. The median change in the general symptom index was similar, but with a large variation in the difference, as shown by a wide interquartile range. The median change in both groups was 0 for the skin symptom index.
The measurements of nasal dimensions revealed an increase in all parameters in the intervention group (Table 4). However, the pretreatment values were systematically higher in the control group than in the intervention group. The median values for the deeper nasal cavities were decreased in control subjects, whereas there was a small increase in the median values for the anterior cavities after treatment. Mean values showed similar results as the medians.
After the last health effects were recorded, participants from both groups were asked about their opinion on the cleaning. Only 2 persons had comments about it; 1 person mentioned that the cleaning had not been done very well (control) and the other person was very impressed by the result (intervention).
In the multivariate analysis for “mucosal irritation symptoms,” the intervention showed a strong effect (Table 5). The outcome in this model was the individual difference in symptoms before and after the intervention. The odds ratio for achieving a reduction of the mucosal irritation symptoms was 3.5 in the intervention group compared with the control group. This was the result for a symptom index reduction of 1 point or more, and the same odds ratio was found for a 2 points or more symptom index reduction. In all the models, the variables that explained most of the variance were the precleaning symptom index values (not shown in tables); participants with the highest number of symptoms before intervention showed the largest reduction in the symptom index values. Another variable explaining some variance was the allergy indicator. No interaction between the allergy indicator and the intervention was found (not shown in tables).
Table 6 shows the results of adjusted logistic regression analyses of the acoustic rhinometry measurements. An odds ratio of 2.4 for volume of the deeper nasal cavity was observed in the first analysis (with median reduction as the cutpoint), but the CI included the reference value. When the cutpoint was the 70th percentile of the reduction, the odds ratios for the deeper nasal volume measure increased to 4.2 and the CI did not include the reference value. The cross-sectional area of the deeper nasal cavity also showed an elevated odds ratio for reduction (1.4), but the CI included unity. The unadjusted results are similar.
This controlled intervention study gives further support to the hypothesis that indoor air dust may cause mucosal irritation symptoms. This finding confirms the results from cross-sectional studies1,7 as well as other intervention studies.9,12 In addition, we observed that nasal congestion (measured objectively) was reduced after cleaning. In contrast to other studies, we found only a tendency to reduced general symptoms from the cleaning intervention (Table 5).1,5,8 The use of rather broad questions with a few reply categories could be debatable. We used mostly the same questionnaire as in other large Scandinavian indoor air studies, and this has the advantage of thorough validity testing. Also, the symptom prevalence could be compared with those in other studies. A further discussion of the questionnaire is available in Skyberg et al.25 Visual analog scales are an alternative approach. However, our experience is that when indexes based on intraindividual change are computed, simple categories provide a sufficient contrast.14
The questions used were originally developed for large screening studies of the prevalence of possible indoor air related symptoms. We used the same questions but with minor changes. Questions were mostly constructed so that they contained more than 1 symptom from 1 target organ (eg, the question on irritation and/or congestion of the nose), and these questions were combined into indices. Thus, the outcome “mucosal irritation index” could represent a health effect other than the measured nasal dimensions.
The strength of the intervention model is based on a sufficient difference in exposure between the intervention and the control groups. We observed that the mean dust reduction in the intervention group was 17 μg/m3 compared with a 21 μg/m3 increase in the control group. The increased airborne dust in the control group was not part of the design. Theoretically, this increase might contribute to the differences in outcomes between the intervention and control groups. However, we observed only minor changes in the irritation and general symptom indices in the control group (Table 3). In contrast, the 2 most important nasal dimension indicators (measuring the deeper nasal cavities) showed an increased congestion in the control group. This could be partly the result of the observed increase in dust concentration in this group.
In general, personal exposure measurements provide the best indicator of the individual exposure load. However, because the intervention was done in single-occupant offices and not in corridors, resting areas, and meeting rooms, the effect of the intervention was better monitored by the stationary measurements. Thus, the effect of the intervention was “diluted” by the fact that the participants spent several hours outside their own office during a workday. An alternative intervention approach would have been to clean larger areas of the building so that each participant would spend a larger fraction of the workday within a cleaned area. This might have introduced problems with confounding and blinding. The stationary measurements showed (Table 2), that the fraction of larger particles (>10 μm) were responsible for 73% of the airborne dust reduction. As expected, the cleaning removed mostly the larger particles. A reduction of the larger particles is also in agreement with health improvements, mainly in the upper airways.
Confounding can be a serious problem in intervention studies, because other conditions could have changed during the intervention. During this study, there could have been changes in environmental factors. However, the relatively short follow-up period of the study limited this type of confounding. In addition, a control group was used, and any environmental changes (such as season or pollen exposure) would be expected to affect both groups, because the offices in the study were spread around the building. Subjects with a respiratory infection could also be a problem; such subjects were evenly distributed between the intervention and control groups and were excluded from the data analysis.
When using a parallel block design, the precleaning health variables could be unequally distributed in the intervention and control group by chance. Consequently, the regression-toward-the-mean effect could give misleading results in simple tabular analysis. An illustration of this was that the precleaning values of the acoustic rhinometry were quite different when comparing the control group and the intervention group. However, pretreatment values were adjusted for in the multivariate analyses.
Schneider et al.26 concluded that there is inadequate scientific evidence that mass concentration can be used as a risk indicator of health effects in nonindustrial buildings. Despite this, according to 3 studies, airborne particles in office air are often considered important exposure factors.5,9,27 The present study gives further support to a possible relation between airborne particles and health effects in nonindustrial buildings. We propose that low levels of allergens, toxins, or irritants among the airborne particles or specific components in the dust rather than the mass concentration of particles could be responsible for the observed health effects.
Office cleaning may change dust composition as well as dust level.10,11 In a 1-group experimental study in a school environment, improved health was observed after fixing moisture problems.28 This implies that mold spores could be responsible for dust-related health problems in some buildings. We found no visible bacteria or fungi by fluorescence microscopy, but this method was probably too insensitive to detect microorganisms in the office dust samples. Culture-based methods can measure substantially lower levels than nonculture-based methods, but could underestimate the exposure by missing nonculturable microorganisms.29
Mucosal irritation symptoms can be caused by an allergic reaction or physical irritation. This study indicates that both allergic and nonallergic participants experienced a reduction of irritation symptoms after the intervention. Other studies also indicate that nonallergic persons could show an irritation response to airborne house dust.4,27 Saraf et al.6 observed that house dust induced the production of “nonallergic” interleukins. The present study was not designed to describe mechanisms of the relation between exposure to airborne indoor particles and health. However, it supports the possibility that nonallergic persons could respond to office dust.
In conclusion, a cleaning intervention caused a change in dust concentration between the intervention group and the control group, leading to reduction in both subjective and objective indicators of mucosal irritation. We do not know whether the change in dust concentration, dust composition, or other factors related to the intervention caused the health effect changes. There could have been factors associated with intervention that we have not investigated such as cleaning detergents, other volatile organic compounds, and gases. However, it seems unlikely that something other than the cleaning intervention was responsible for the health improvement in the intervention group.
We appreciate the participation of the occupational health services and the employees at Norsk Hydro. We also thank Steinar Nilsen, ISS (Integrated Service Solutions), for the planning and implementation of the cleaning intervention. Per Ole Huser, Lene Madsø, and Claudia Hauge at the National Institute of Occupational Health, Oslo (NIOH) performed dust measurements and dust analyses. Lars Ole Goffeng, NIOH, helped with acoustic rhinometry measurements. Patricia Flor provided linguistic help.
1.Skov P, Valbjørn O, Pedersen BV. Influence of personal characteristics, job-related factors and psychosocial factors on the sick building syndrome. Scand J Work Environ Health. 1989;15:286–295.
2.Stenberg B, Mild KH, Sandström M, et al. A prevalence study of the sick building syndrome (SBS) and facial skin symptoms in office workers. Indoor Air. 1993;3:71–81.
3.Schneider T, Skov P, Valbjørn O. Challenges for indoor environment research in the new office. Scand J Work Environ Health. 1999;255(special issue):574–579.
4.Pan Z, Mølhave L, Kjærgaard SK. Effects on eyes and nose in humans after experimental exposure to airborne office dust. Indoor Air. 2000;10:237–245.
5.Wargocki P, Lagercrantz L, Witterseh T, et al. Subjective perceptions, symptom intensity and performance: a comparison of two independent studies, both changing similarly the pollution load in an office. Indoor Air. 2002;12:74–80.
6.Saraf A, Larsson L, Larsson BM, et al. House dust induces IL-6 and IL-8 response in A549 epithelial cells. Indoor Air. 1999;9:219–225.
7.Gyntelberg F, Suadicani P, Nielsen JW, et al. Dust and sick building syndrome. Indoor Air. 1994;4:223–238.
8.Wålinder R, Norbäck D, Wieslander G, et al. Nasal mucosal swelling in relation to low air exchange rate in schools. Indoor Air. 1997;3:198–205.
9.Kemp PC, Dingle P, Neumeister HG. Particulate matter intervention study: a causal factor of building related symptoms in an older building. Indoor Air. 1998;3:153–171.
10.Franke DL, Cole EC, Leese KE, et al. Cleaning for improved indoor air quality: an initial assessment of effectiveness. Indoor Air. 1997;7:41–54.
11.Kildesø J, Tornvig L, Skov P, et al. An intervention study of the effect of improved cleaning methods on the concentration and composition of dust. Indoor Air. 1998;8:12–22.
12.Raw GJ, Roys MS, Whitehead C. Sick building syndrome: Cleanliness is next to healthiness. Indoor Air. 1993;3:327–345.
13.Sundell J, Lindvall T, Stenberg B, et al. Sick building syndrome (SBS) in office workers and facial skin symptoms among VDT-workers in relation to building and room characteristics: two case-referent studies. Indoor Air. 1994;4:83-94.
14.Skulberg KR, Skyberg K, Eduard W, et al. Effects of electric field reduction in visual display units on skin symptoms. Scand J Work Environ Health. 2001;27:140–145.
15.Schneider T, Nilsen SK, Dahl I. Cleaning methods, their effectiveness and airborne dust generation. Building and Environment. 1994;29:369–372.
16.Kenny LC. Developments in workplace aerosol sampling. Analyst. 1996;121:1233–1239.
17.Vincent JH. Aerosol Science for Industrial Hygienists. New York: Elsevier Science; 1995;313:
18.Micallef A, Caldwell J, Colls JJ. The influence of human activity on the vertical distribution of airborne particle concentration in confined environment: Preliminary results. Indoor Air. 1998;2:131–136.
19.Palmgren U, Ström G, Blomquist G, et al. Collection of airborne microorganisms on Nuclepore filters: estimation and analysis—CAMNEA method. J Appl Bacterial. 1986;61:401–406.
20.Andersson K, Fagerlund I, Stridh G, et al. The MM-Questionnaires. A Tool When Solving Indoor Climate Problems. Örebro: Örebro Medical Center Hospital; 1993.
21.Mayhew T, O’Flynn P. Validation of acoustic rhinometry by using the Cavalieri principle to estimate nasal cavity volume in cadavers. Clin Otolaryngol. 1993;18:220–225.
22.Fischer EW, Scadding GGK, Lund VJ. The role of acoustic rhinometry in studying the nasal cycle. Rhinology. 1993;31:57–61.
23.Hilberg O, Pedersen OF. Acoustic rhinometry: recommendations for technical specifications and standard operating procedures. Rhinology. 2000;16:3–17.
24.O’Flynn P. Posture and nasal geometry. Acta Otolaryngol. 1993;113:530–532.
25.Skyberg K, Skulberg KR, Eduard W, et al. Symptoms prevalence among office employees and associations to building characteristics. Indoor Air. 2003;13:246–252.
26.Schneider T, Sundell J, Bischof W, et al. ‘EUROPART.’ Airborne particles in the indoor environment. A European interdisciplinary review of scientific evidence on associations between exposure to particles in buildings and health effects. Indoor Air. 2003;13:38–48.
27.Mølhave L, Kjærgaard SK, Attermann J. Sensory and other neurogenic effects of exposures to airborne office dust. Atmospheric Environment. 2000;34:4755–4766.
28.Åhman M, Lundin A, Musabađiã V, et al. Improved health after intervention in a school with moisture problems. Indoor Air. 2000;10:57–62.
29.Eduard W, Heederik D. Methods for quantitative assessment of airborne levels of non-infectious microorganisms in highly contaminated work environments. Am Ind Hyg Assoc J. 1998;59:113–127.
© 2004 Lippincott Williams & Wilkins, Inc.