environmental asthmatic symptoms and volatile organic

Occupational and Environmental Medicine 1995;52:388-395

Asthmatic symptoms and volatile organiccompounds, formaldehyde, and carbon dioxide indwellingsD Norbdck, E Bj6rnsson, C Janson, J Widstr6m, G Boman

AbstractObjectives-As a part of the worldwideEuropean Community respiratory healthsurvey, possible relations between symp-toms of asthma, building characteristics,and indoor concentration of volatileorganic compounds (VOCs) in dwellingswere studied.Methods-The study comprised 88 sub-jects, aged 20-45 years, from the generalpopulation in Uppsala, a mid-Swedishurban community, selected by stratifiedrandom sampling. Room temperature,air humidity, respirable dust, carbondioxide (CO2), VOCs, formaldehyde, andhouse dust mites were measured in thehomes of the subjects. They underwent astructured interview, spirometry, peakexpiratory flow (PEF) measurements athome, methacholine provocation test forbronchial hyperresponsiveness, and skinprick tests. In addition, serum concen-tration of eosinophilic cationic protein(S-ECP), blood eosinophil count, and totalimmunoglobulin E (S-IgE) were measured.Results-Symptoms related to asthmawere more common in dwellings withhouse dust mites, and visible signs ofdampness or microbial growth in thebuilding. Significant relations were alsofound between nocturnal breathlessnessand presence of wall to wall carpets, andindoor concentration of C02, formalde-hyde, and VOCs. The formaldehyde con-centration exceeded the Swedish limitvalue for dwellings (100,uglm3) in onebuilding, and CO2 exceeded the recom-mended limit value of 1000 ppm in 26%of the dwellings, showing insufficientoutdoor air supply. Bronchial hyper-responsiveness was related to indoorconcentration of limonene, the mostprevalent terpene. Variability in PEF wasrelated to two other terpenes; a-pinenand 6-karen.Conclusion-Our results suggest thatindoor VOCs and formaldehyde maycause asthma-like symptoms. There is aneed to increase the outdoor air supplyin many dwellings, and wall to wall car-peting and dampness in the buildingshould be avoided. Improved indoorenvironment can also be achieved byselecting building materials, buildingconstruction, and indoor activities on theprinciple that the emission of volatileorganic compounds should be as low asreasonably achievable, to minimize

symptoms related to asthma due toindoor air pollution.

(Occup Environ Med 1995;52:388-395)

Keywords: volatile organic compounds; terpenes;formaldehyde; asthma

During recent decades, concern has increasedabout possible health effects resulting fromindoor air pollution.' Asthma is the mostcommon lung disease associated with indoorpollutants,2 and affects about 5% of the adultSwedish population. There have been reportsof increased morbidity and mortality fromasthma and allergies in several countries,4-6and it has been suggested that this increasemay be due to increased concentrations of in-door pollutants in modern buildings.4 As thepopulation in the industrialized world spendsabout 65% of their lives in their homes,7 expo-sures in the home environment might have aprofound impact on respiratory health.

Indoor air may contain many different airpollutants, including microorganisms, aller-gens from mites or furry animals, nitrogenoxides, and volatile organic compounds(VOCs). Symptoms related to asthma causedby exposure to nitrogen dioxide (NO2) fromgas cooking have been studied extensively inchildren. A meta-analysis of several studiesconcluded that gas cooking increased the longterm exposure to NO2 by 30 ,ug/m,3 and thisexposure results in a 20% increase in risk oflower respiratory tract illness.8 A relationbetween lower airway symptoms and type offuel used for cooking and heating has alsobeen found in adults in northern Italy.9 Thetype of heating system and fuels for cookingmay, however, differ greatly between differentcountries. In Scandinavia, most cooking isdone by electric stoves, and most heating is bymeans of water borne central heating.

Indoor air may contain different types ofVOCs, emitted from various sources-forexample, building materials, consumer prod-ucts, and human activities. Rarely, asthma iscaused by sensitisation to specific indoorVOCs. As an example, one case of asthmawas recently described where the cause wasindirect exposure to a biocide used in a floorcleaner.'0 Formaldehyde is a reactive volatilecompound that may induce airway irritationat low concentrations. Due to differencesin measurement techniques, formaldehyde,obviously a volatile compound, is notincluded in the common definition ofVOCs. Recent prevalence studies have shown

Department ofOccupational andEnvironmentalMedicineD NorbdckJ WidstromDepartment ofLungMedicine and AsthmaResearch Centre,Uppsala University,Akademiskasjukhuset, S-751 85Uppsala, SwedenE Bj6rnssonC JansonG BomanCorrespondence to:Dr Dan Norback,Department of Occupationaland EnvirommentalMedicine, UppsalaUniversity, Akademiskasjukhuset, S-751 85Uppsala, Sweden.Accepted 13 February 1995

388

on October 10, 2021 by guest. P

rotected by copyright.http://oem

.bmj.com

/O

ccup Environ M

ed: first published as 10.1136/oem.52.6.388 on 1 June 1995. D

ownloaded from

http://oem.bmj.com/

Asthmatic symptoms and volatile organic compounds, formaldehyde, and carbon dioxide in dwellings

correlations between symptoms of asthma anddamp housing conditions." 1-2 Dampness inbuildings favours growth of microorganismsand house dust mites within the building.'3Another source of airborne indoor microor-ganisms is air humidifiers, which may in rare

cases cause asthma. 14 Besides the increasedrisk of microbial growth and mites, dampnessin buildings may also increase the emission ofVOCs, due to degradation of building material,or microbial activity.'5 One example of thefirst is hydrolysis of phthalates in PVC floormaterial, where an emission of 2-ethyl-hexa-nol can be detected.'5 By contrast, 1-octen-3-olis emitted from the metabolism of microor-ganisms but not from building materials.'5Allergy to house dust mites is a well knowncause of asthma related to buildings.'6 Fewstudies are, however, available where relationsbetween asthmatic symptoms and VOCs havebeen investigated.Our aim was to study possible relations

between symptoms and clinical signs relatedto asthma in the general adult population andexposure to volatile organic indoor pollutantsin dwellings.

Materials and methodsSTUDY POPULATIONThe European Community respiratory healthsurvey is a multicentre study on the preva-lence of allergies and asthma found in 48 cen-

tres in 23 countries throughout the world.'7Each centre covers a source population ofabout 150 000 inhabitants within a definedgeographical and administrative area. Swedencontributes with data from three such areas,

one being the municipality of Uppsala.The source population was all subjects

living in the community of Uppsala in 1990,a mid-Swedish urban commune with a

total population of 160 000 inhabitants. InDecember 1990, a screening questionnairewas mailed to a random sample of 3600 men

and women, aged 20-44 years, selected fromthe population register of Uppsala. A randomsubsample of 600 of these 3600 subjects was

further examined for signs of atopy andbronchial hyperresponsiveness at the depart-ment of lung medicine. Blood samples were

obtained and all subjects were interviewedand examined by specially trained nurses. Allsubjects gave their informed consent.The study population of our study con-

sisted of two groups of subjects further selectedfrom the subsample. The first group was allsubjects (n = 74) who had reported at least oneof the following symptoms in the screeningquestionnaire: attacks of asthma during thepast 12 months, nocturnal breathlessness inthe past 12 months, or current use of asthmamedication. Also, 80 subjects who gave nega-tive answers to all these three questions were

randomly selected, without matching.These 154 subjects were contacted by post

and by phone, and were offered air monitor-ing of their dwelling. These field measure-

ments were performed from October 1991 to

April 1992, as were the clinical investigations.

The study was blinded, to the extent thatinformation from the exposure measurementswas not linked to the medical information,until the data collection was completed. Theprotocol of the study was approved by theEthics Committee of the Medical Faculty ofUppsala University.

ASSESSMENT OF EXPOSURERoom temperature, air humidity, VOCs, res-pirable dust, and carbon dioxide concentra-tion (CO2) were measured in the living roomand the bedroom in the houses of all partici-pants. In the bedroom, additional measure-ments of formaldehyde, and guanine fromhouse dust mites were performed. Measure-ments in the living room were performed inthe centre of the room, 1 0 m above the floor.Bedroom measurements were performed besidethe pillow of the bed, at the same height abovethe floor as the pillow. The bedroom door wasclosed, and no subject stayed in the bedroomduring the VOC, CO2, respirable dust, andformaldehyde measurements.Room temperature and air humidity were

recorded with an Assman psychrometer.Concentrations of respirable dust were mea-sured by a direct reading instrument based onlight scattering (Sibata P-5H2, SibataScientific Technology, Japan). The instru-ment was calibrated by the manufacturer to0 3 ,um particles of stearic acid. Indoor CO2concentration was measured by a direct read-ing infrared spectrometer (Rieken RI-4 11A,Rieken Keini, Japan), calibrated by standardgases containing known concentrations ofCO2. The average CO2 concentration in thedwelling was calculated, by taking the averageof 30 minutes registration in the bedroom and30 minutes registration in the living room.

Indoor concentrations of formaldehydewere measured with glass fibre filters impreg-nated with 2,4-dinitro-phenylhydrazine.'8 Theair sampling rate was 0-25 1/min for twohours. The filters were analysed by liquidchromatography. Volatile organic com-pounds, other than formaldehyde, were mea-sured both in the bedroom and the livingroom by sampling on charcoal sorbenttubes.19 The air sampling rate was 1 1/min fortwo hours. The charcoal tubes were desorbedwith 1 ml of carbon disulphide before analy-sis, which was performed within one weekfrom the sampling day with a gas chromato-graph equipped with a flame ionisation detec-tor. Sixteen common solvents were identifiedand quantified by the external standard tech-nique, by comparing the retention times ontwo different columns. When quantifying lowboiling uncalibrated compounds (C3-C 12)the response factor of n-decane was used; highboiling unknown compounds used theresponse factor of a mixture of high boilinghydrocarbons (dodecylbenzenes). The total con-centration of the identified and unidentifiedVOCs expressed as ,ug/m3 was calculated.'9

Settled bedroom dust was collected bystandardised vacuum cleaning of both the bedmattress and the bedroom floor, with thesame vacuum cleaner in all homes (ELRAM

389



.bmj.com

/O

ccup Environ M


ownloaded from

http://oem.bmj.com/

Norback, Bjornsson, J7anson, Widstrdm, Boman

HSS 09, ELRAM, Sweden). In each bed-room, a sampling time of one minute/M2 (fiveseconds/foot2) was applied to both the floorand the mattress. Presence of house dustmites in collected dust was detected by thesemiquantitative Acarex-test,20 which gives ascore from 0-3. It has been validated against aquantitative immunological test with mono-clonal antibody against house dust mites.20According to the suggestions of van derBrempt et a120 an ACAREX score of 0 or 1was regarded as negative, and a score of 2 or 3was regarded as positive for the presence ofallergenic concentrations of house dust mites.The test is based on a reaction to guanine infecal pellets from dust mites, and is thereforenot specific for house dust mites.2' Through-out this text, however, a positive ACAREX scoreis considered to be evidence ofhouse dust mites.

ASSESSMENT OF SYMPTOMS AND PERSONALFACTORSThe screening and interview questionnaireused in the European Community respiratoryhealth survey was a modified version of theInternational Union against Tuberculosis andLung Disease (IUATLD) questionnaire.22 23All questions were translated to Swedish andthen back to English, to minimise translationbias. The recall period for airway symptomswas 12 months. The questionnaire also con-tained questions on the dwelling, socialsituation, education, and environmental expo-sures. Symptoms related to asthma were de-fined as reported in the interview to have had inthe past 12 months: (a) wheezing or whistlingin the chest; or (b) at least one daytime attackof shortness of breath during exercise or atrest; (c) at least one wakeful night because ofbreathlessness or tightness in the chest.

In addition, information on age, sex, andsmoking habits was collected from the screen-ing and interview questionnaire. Currentsmoker was defined as a subject who reportedcurrent smoking in the interview, or hadstopped smoking less than a year ago.

Assessment of clinical signsAtopy-Skin prick tests were carried out in astandardised way, by means of lancets coatedwith allergen (Phazets, PharmaciaDiagnostics, Uppsala, Sweden). The follow-ing allergens were tested: Dermatophagoidespteronyssinus (house dust mite), cat, dog,birch, Cladosporium, olive, ragweed, mug-wort, timothy, and Alternaria. Histamine wasused as a positive control. As proposed byDreborn, atopy was defined as a prick testreaction to at least one of the allergens, with amean diameter of 3 mm or greater.24 A negative control was used, and its mean diameterwas subtracted from that of the allergens.

LUNG FUNCTION AND BRONCHIALHYPERRESPONSIVENESSForced expiratory volume in one second(FEV,) was measured with the Spiro Medicscomputerised dry rolling seal spirometer sys-tem 2130 (Sensor Medics, Anaheim,California, USA). The predicted value for

each subject was calculated.25 Peak expiratoryflow (PEF, best of three measurements) wasrecorded twice daily for one week with aMini-Wright peak flow meter (ClementClark, London). Peak flow variability was cal-culated by dividing the difference between thehighest and the lowest daily PEF reading bythe daily mean PEF value. The index used isthe one suggested by Higgins et al, who foundthis way of measuring peak flow to be themost sensitive when comparing asthmatic andnon-asthmatic subjects.26 This index has alsobeen described in a subsequent paper by thesame group.27 Methacholine challenge wasperformed by a MEFAR inhalation dosimeter(MEFAR, Brescia, Italy).28 Bronchial hyper-responsiveness was defined as a positivemethacholine test-that is, a reduction inFEVy by at least 20% with an accumulateddose of less than or equal to 2 mg (PD20).

BIOCHEMICAL MARKERS OF INFLAMMATIONAND ALLERGYFrom each volunteer, 35 ml of venous bloodwas collected, just before the methacholineprovocation test. The same investigationsequence was applied to all participants.Blood eosinophil counts were analysed on aHemalog 2R (Technicon ChemicalsCompany, Tournai, Belgium) in blood (5 ml)supplemented with ethylene diaminetetraacetic acid (EDTA, 034 mol/l). Theconcentration of eosinophilic cationic protein(S-ECP) was measured by means of a doubleantibody radioimmunoassay (PharmaciaDiagnostics AB, Uppsala, Sweden).29 Thecoefficients of variation within and betweenassays are less than 11 %, and the detectionlimit is less than 2 ,ug/l. The 20 ml blood sam-ple used to measure S-ECP was allowed tocoagulate for 60 minutes at room temperatureafter sampling. After that it was centrifugedfor 10 minutes, the serum was poured into anew tube and then centrifuged again for 10minutes. The serum was collected thereafterand kept frozen at - 70°C until analysed.Also, serum from another 10 ml of blood wascollected, and analysed for total serumimmunoglobulin E (S-IgE) by the PharmaciaCAP system (Pharmacia Diagnostics AB,Uppsala, Sweden).

STATISTICAL METHODSAn unpaired t test was used to study relationsbetween asthmatic symptoms and indepen-dent continuous variables that were roughlynormally distributed (room temperature, airhumidity, and carbon dioxide concentration).The Mann-Whitney U test was applied for thecrude analysis of relations between symptomsand other exposures that were not normallydistributed (VOCs, respirable dust, andformaldehyde). The x2 test or Fisher's exacttest, depending on the number in the cells,was used for analysis of the relation betweenbinary dependent and independent variables,Kendal rank correlation test (Tau-beta) wasused when testing correlation between clinicalsigns and measured environmental exposures.Multivariate statistical analysis was performed

390



.bmj.com

/O

ccup Environ M


ownloaded from

http://oem.bmj.com/


by multiple logistic regression with theSPIDA statistical package (the StatisticalLaboratory, Macquarie University, Australia).30Logarithmic values of formaldehyde andVOCs were applied in the model. Thecollinearity diagnostics described in theSPIDA manual were applied, and adjustedodds ratios (ORs) with a 95% confidenceinterval (95% CI) were calculated. Adjust-ments were made for potential confounders:age, sex, and current smoking. To reducedetected collinearity problems, the data onformaldehyde and VOCs were centred bysubtracting the mean concentration from eachperson. In all statistical analyses, two tailed testsand a 5% level of significance was applied.

ResultsThe total response rate in the initial selfadministered questionnaire survey was 87%.In the second part, the response rate was 81%(figure). Of the 154 subjects chosen for thisstudy, 26 (17%) were excluded because ofmigration. They had either moved to anotherdwelling during the study period (n = 19), orwere working or living outside the municipalityof Uppsala during the field study period (n =7). The non-participants (n = 40) consisted of14 subjects who did not come to the medicalinvestigation, 12 who could not be contactedat their home address, and 14 refused to par-ticipate in the exposure measurements. Theremaining 88 subjects (57%) participated inthe medical interview, had lived in the samedwelling during the study period, and gavetheir approval for the measurements in theirhome. The non-participants did not differfrom the participants for age, reported symp-toms, smoking, bronchial hyperresponsive-ness, or atopy. From the 88 participants,

Uppsala municipality

Random samplesent postal questionnaire +3600

Response rate87%

3146 men and women

Random sample interviewed,examined, blood tests,prick tests, bronchial provocation600

Response rate81%

488 men and women

All who answered yes tothree questions on asthma

Random sample of thosewho answered no to threequestions on asthma

I4

Response rate64%

Asthmaticsymptoms

4 4 Response rate57%

m 4I No asthmaticLiii 41 symptoms

information was obtained as follows: interviewquestionnaires from 88 subjects, spirometry in82, prick tests in 77, methacholine challengetest in 73, a peak flow diary in 68, and bloodsamples from 74 subjects.

PERSONAL CHARACTERISTICSMean (SD) age of all participants was 32 (7)years, and the mean (range) duration in thepresent dwelling was six (0-5-31) years. Themean (SD) FEV, was 106% (13%) of the pre-dicted value, and the mean (range) variabilityin PEF was 5% (1-18%). Subjects with symp-toms related to asthma had a higher preva-lence of asthmatic medication, bronchialhyperresponsiveness, and were more oftenwomen. In contrast, the proportion of currentsmokers, and subjects with atopy, were similaramong subjects with and without asthmaticsymptoms (table 1). One of 17 people whohad cats at home had a positive prick test tocat allergen. None of the seven subjects whohad dogs were allergic to dog. Four subjectshad a positive prick test to house dust mites(Dermatophagoides pteronyssinus). None ofthese had a positive ACAREX test for housedust mites in their home.

CHARACTERISTICS OF BUILDINGSMost of the buildings were not situated nearroads with heavy traffic, and about halfthe buildings (49%) were built after 1970(table 2). Half the buildings (51%) wereapartments in larger houses, and 49% weresingle family houses. Most of the buildingswere heated by a water borne central heatingsystem (87%), a few (9%) were equipped withducted air heating. The source of the heat wasmainly hot water distributed by the communethrough underground ducts (73%), or electricradiators (24%). In a few of the buildings,heat was produced by combustion of organicmaterials, mainly wood (25%), and rarelycentral heating by oil (5%) or keroseneheaters (5%) were used. Some buildings had acombination of different heating systems. Allbuildings were equipped with electric stovesonly, and none had a gas stove or any othertype of gas heater. Indoor tobacco smokingwas reported to occur in 23% of the homes,and was found during the measurements infour dwellings (5%). Presence of wall to wallcarpets was noticed in 18% of the dwellings,

Table 1 Prevalence of demographic and medical data forsubjects with and without symptoms related to asthma(n = 88)

Symptomt No symptom(n = 47) (n = 41)

Characteristics ()/0 (%/0)Women 72 32***Wheeze 33 0***Day time breathlessness 28 0***Nocturnal breathlessness 25 0***Asthma medication 30 0***Current tobacco smoker 23 22Atopy 48 38Bronchial hyperresponsiveness 32 9*

*P < 0 05, ***P < 0-001tSubjects reported wheezing or whistling in the chest, daytimeattacks of shortness of breath during exercise or at rest, or wak-ing at night because of breathlessness or tightness in the chest.The study population and response rates.

391



.bmj.com

/O

ccup Environ M


ownloaded from

http://oem.bmj.com/

Norback, Bjornsson, Janson, Widstrom, Boman

Table 2 Characteristics of dwellings ofsubjects with and without symptoms related toasthma (n = 88)

Symptoms No symptomCharacteristics (n = 47) (%) (n = 41) (Oo)

Age of the dwelling:< 1960 32 171961-70 27 241971-80 17 29> 1980 24 29

Type of house:Detached 32 29Terrace 15 22Apartment 49 53

Tobacco smoking indoors 23 22Cats or dogs at home 28 24Birds at home 6 5Presence of house dust mites* 21 5*Presence of wall to wall carpets 19 17Visible signs of dampness in the building: 23 7*Ventilation system

Natural ventilation only 30 22Mechanical exhaust air only 32 41Mechanical supply and exhaust air 38 37

Dwelling situated near (< 50 m) heavy traffic 4 10

*P < 0 05 by x2 test; tas for table 1; observed or measured at inspection.

and 16% had visible signs of dampness in thebuilding or microbial growth. Most of thehouses had mechanical exhaust air ventila-tion, but 26% had natural ventilation only.Presence of house dust mites, defined as score2 or 3 on the ACAREX test, was detected in13% of the bedrooms (table 2).

SYMPTOMS RELATED TO ASTHMA IN RELATIONTO CHARACTERISTICS OF THE BUILDINGSPresence of house dust mites was significantlyrelated to asthmatic symptoms: crude OR =5-3 (95% CI 1-2-22-8). A significant relationbetween visible signs of dampness in thebuilding or microbial growth and asthmaticsymptoms was also found: crude OR = 3 9(95% CI 1-1-14-1). None of the other charac-teristics of the buildings differed significantlybetween subjects with and without asthmaticsymptoms, when subjects who reported anytype of asthmatic symptoms were comparedwith subjects without asthmatic symptoms(table 2). For nocturnal breathlessness, how-ever, a significant relation was found for wallto wall carpets in the dwelling: crude OR =

Table 3 Indoor concentration of selected volatile organic compounds (VOCs) (pg/m3) indwellings ofsubjects with and without nocturnal attacks of breathlessness

Nocturnal breathlessness or chest tightness

Absence of (n = 62) Presence of (n = 26)Type of VOC mean (range) mean (range)

Toluene, bedroom 21 (3-230) 120 (1-2330)**Toluene, living room 21 (1-230) 120 (3-2250)**C8-Aromaticst, bedroom 23 (4-190) 55 (5-690)**C8-Aromaticst, living room 23 (2-190) 46 (6-470)**n-Alkanest, bedroom 19 (2-150) 39 (2-290)**n-Alkanest, living room 20 (2-120) 39 (2-270)**Terpenes§, bedroom 52 (5-350) 96 (5-580)*Terpenes§, living room 60 (5-960) 130 (7-1010)*Butanols¶, bedroom 10 (< 1-43) 19 (1-90)*Butanols¶, living room 10 (< 1-45) 18 (2-84)*LVOCII, bedroom 99 (9-2200) 370 (21-5520)*LVOCII, living room 130 (9-1030) 380 (17-5010)**Total VOCjt, bedroom 310 (70-2920) 790 (90-9380)*Total VOCft, living room 300 (70-1670) 780 (70-8350)**

*P < 0-05; **P < 0-01 by Mann-Whimey U test.tSum of ethylbenzene, m-sylene, o-xylene, and p-xylene.tSum of n-octane, n-nonane, n-decane, and n-undecane.§Sum of a-pinene, 6-carene, and limonene.¶Sum of n-butanol and iso-butanol.ISum of unidentified compounds with a retention time below benzene.ftSum of all identified and unidentified compounds.

4-2 (95% CI 1A-12-3). The influence of car-peting was significant even after adjustmentfor possible influence of age, sex, currentsmoking, and other characteristics of thebuilding by multiple logistic regression:adjusted OR= 9-7 (95% CI 2-1-45-3). Nosignificant relation was found between noc-turnal breathlessness and signs of dampnessin the building or microbial growth. No signif-icant relations were found between daytimeattacks of shortness of breath and any charac-teristics of the buildings. Moreover, no signifi-cant relation was found between any type ofsymptoms related to asthma, and exposure toenvironmental tobacco smoke, or living near(< 50 m) heavy traffic.

SYMPTOMS RELATED TO ASTHMA ANDMEASURED EXPOSURESHouse dust mites were detected in 13% of thebedrooms, and a significant relation betweenpresence of these and nocturnal breathless-ness was found. The influence of dust miteswas significant even after adjustment for pos-sible influence of age, sex, current smoking,and other characteristics of the building:adjusted OR=8-2 (95% CI 1-4-48-1).Furthermore, the average concentration ofCO2 was significantly higher in the homes ofthose people who reported nocturnal chesttightness, compared with subjects withoutsuch symptoms (1020 ppm and 850 ppmrespectively, P < 0-01). Also, a relationbetween formaldehyde concentration in thebedroom, and nocturnal symptoms was found(P < 0-01). The mean (range) concentrationof formaldehyde was 29 (< 5-110) pg/m3 inhomes of subjects with nocturnal breathless-ness, compared with 17 (< 5-60) yug/m' inhomes of subjects without nocturnal breath-lessness. The formaldehyde concentrationexceeded the Swedish limit value fordwellings (100 pug/mi) in one building, butCO2 exceeded the recommended limit valueof 1000 ppm in 26% of the dwellings. For res-pirable dust concentration, no significant rela-tion to nocturnal breathlessness could beshown. The room temperature (range 17-8-22-80C) and relative air humidity (range 33-75%) was not significantly different in subjectswith and without nocturnal symptoms.

In the initial analysis, a significant relationbetween any type of symptom related toasthma and concentration of total VOCs wasfound (P < 0-01). The relation between car-peting, CO2, formaldehyde, and total VOCsand asthmatic symptoms was most pro-nounced for nocturnal dyspnoea (table 3).Wheezing or whistling in the chest and day-time attacks of shortness of breath showed nosignificant relations to any measured expo-sures. Significant relations between nocturnalbreathlessness and various subclasses ofVOCs were also detected (table 3). The aver-age concentrations of most compounds weresimilar in bedrooms and the living room, and alarge proportion of the VOCs were low boil-ing VOCs that consisted of unidentified com-pounds with a boiling point lower thanbenzene (table 3).

392



.bmj.com

/O

ccup Environ M


ownloaded from

http://oem.bmj.com/


Table 4 Adjusted ORs (95% CI) for nocturnalbreathlessness in relation to CO2, formaldehyde*, andsignificant types of VOCs*

Type ofcompound Adjusted OR (95% CI)

CO2t 20 0 (2-7-146)Formaldehyde 12 5 (2 0-77 9)Toluene 4-9 (1 1-22 8)C8-Aromaticst 6-7 (1-0-45-1)Terpenes§ 4 0 (1-2-13-4)LVOC¶ 4 4 (1-3-15-0)TVOCI 9.9 (1 7-58-8)

*Logarithmic values used in the regression models. Odds ratiocalculated for a 10-fold increase of the indoor concentrationadjusted for age, sex, current smoking, presence of wall to wallcarpets, and presence of house dust mites.tOdds ratio for an increase of the indoor concentration by1000 ppm CO2 adjusted for age, sex, current smoking, pres-ence of wall to wall carpets, and presence of house dust mites.tSum of ethylbenzene, m-xylene, o-xylene, and p-xylene.§Sum of a-pinene, 5-carene, and limonene.¶Sum of unidentified compounds with a retention time belowbenzene.ISum of all identified and unidentified volatile organic com-pounds.

Some subclasses of VOCs were stronglyrelated to each other and to total VOCs, par-ticularly toluene, C8-aromates (xylene andethylbenzene), n-alkanes, and low boilingVOCs. In contrast, butanols or terpenes werenot significantly related to total VOCs.Significant relations between presence ofhouse dust mites and the concentration oftotal VOCs, and some subclasses of VOCs,were also found. The total concentration ofVOCs was 1050,ug/m' in buildings that hadhouse dust mites, and 350 yug/M3 in buildingswithout house dust mites (P < 0 05), but norelation was found between total VOCs andsigns of microbial growth or dampness in thebuilding. Carbon dioxide, a surrogate mea-sure of the rate of air exchange, was related tototal VOCs, but not related to formaldehydeor the presence of house dust mites.Formaldehyde concentration was higher indwellings with wall to wall carpets (P <0 05), but no relation was found between car-peting and total VOCs.To further evaluate the effect of formalde-

hyde and different VOCs statistical modellingby multiple logistic regression analysis wasapplied. Those seven classes of VOCs wherethe concentration in both the bedroom andthe living room were significantly related tosymptoms in the crude analysis were selectedfor further testing. The relations between noc-turnal symptoms and CO,, formaldehyde,toluene, C8-aromates, terpenes, low boilingVOCs, and total VOCs were also significantin the logistic model (table 4). Due to collin-earity between formaldehyde and volatile

Table 5 Correlation coefficientst between exposure toterpenes, and clinical signs of obstructive airway reactions

Type ofcompound BHRt PEFI FEV,¶

a-Pinene, bedroom 0-01 0 26*** -005a-Pinene, living room 0-01 0-21** -0 056-Karene, bedroom 0 08 0-18* -0 013-Karene, living room 0 09 0 19* -0 03Limonene, bedroom 0 17* 0-08 0 09Limonene, living room 0-16* 0-12 0-07

*P< 005; **P <0-01; ***P <0-001.tMeasured by Kenals rank correlation test (Tau-beta);bronchial hyperresponsiveness; Svariability of peak expiratoryflow (%); Tforced expiratory flow in one second.

organic compounds, however, these expo-sures could not be kept in the models simulta-neously. The effects of house dust mites andcarpeting on nocturnal chest tightness were,however, significant even when the effects offormaldehyde and VOCs were controlled.

Finally, relations between clinical signs andindoor exposures were also investigated. Nosignificant relations were found betweenbronchial hyperresponsiveness, variability inPEF, FEV,%, and presence of house dustmites, signs of dampness in the building, ormicrobial growth. For measured exposures,only compounds significantly related to noc-turnal breathlessness in the logistic regressionanalysis were evaluated. Bronchial hyper-responsiveness was significantly related to theindoor concentration of limonene, the mostprevalent terpene. Variability in PEF wasrelated to the two other measured terpenes;a-pinene and 3-karen (table 5). On average,64% of the terpenes consisted of limonene,15% of 3-karen, and 21% of ca-pinene. Noactivities known to emit terpenes, such ascleaning or peeling of citrus fruits, wereobserved during the measurements. No rela-tions were found between bronchial hyper-responsiveness, variability in PEF, FEV,%,and the indoor concentration of formalde-hyde, n-alkanes, aromatic compounds, ortotal VOCs. Furthermore, no significant rela-tions were found between serum concentra-tion of eosinophilic cationic protein, bloodeosinophilic count, total serum immunoglob-ulin E concentration, and either carpeting,house dust mites, formaldehyde, or any of theVOCs significantly related to nocturnalbreathlessness.

Discussion and conclusionWe have shown significant relations betweenasthmatic symptoms and dampness in thebuilding, house dust mites, and indoor con-centration of VOCs and formaldehyde. Also,significant relations between some clinicalsigns related to airway obstruction and theindoor concentration of terpenes were found.The design was cross sectional and in such

studies, selection effects may cause false nega-tive results. This could be of particular signifi-cance for asthma, which is a severe condition.The failure to find a relation between pets andsensitisation is most probably due to healthbased selection that would cause sensitisedsubjects to avoid house pets. Moreover, thecross sectional design does not enable us todifferentiate between exacerbation of symp-toms in subjects with asthma, and inductionof new asthma.Many methodological problems of internal

validity are inherent in an epidemiologicalstudy. In this particular study, selection biasdue to low response rate is unlikely as the par-ticipation rate in the initial postal question-naire was high (87%). Also, the participantsand non-participants in the clinical study didnot differ in age, atopy, smoking habits, or air-way symptoms. Response bias due to aware-ness of exposure may cause a general

393



.bmj.com

/O

ccup Environ M


ownloaded from

http://oem.bmj.com/

Norback, Bjornsson, J7anson, Widstrom, Boman

overreporting of symptoms in exposed groups.Also, a time difference between clinical inves-tigations and exposure measurements mayinduce a possible bias. In our study, however,the exposure measurements were performedin parallel with the medical investigations, andcould not influence the reporting of symptoms.

Despite the limitations of the cross sec-tional design and the study size, we couldshow significant relations between VOCs andsymptoms related to asthma. A large numberof statistical tests were made, but most of therelations were significantly below the 1 % level,and similar results were obtained both in thecrude analysis, and by logistic multiple regres-sion analysis. Thus, we do not think that ourconclusions are seriously biased by selectionor response errors, or due to chance findings.We found a relation between measured

exposure and nocturnal symptoms, but notbreathlessness during the daytime. One poss-ible explanation could be that most people arenot at home during the daytime, and if thereis a cumulative effect of indoor pollutants athome it would be most likely to occur at theend of the night. Another explanation couldbe that the exposures are higher during thenight than during the day, or that subjects aremore susceptible to air pollutants duringsleep. For practical reasons, our exposuremeasurements were performed during theday, without any subject in the bedroom. Thisenabled us to isolate the contribution of thebuilding and its interior to the indoor concen-trations of VOCs. During sleep however,human emissions are added to the back-ground concentrations of VOCs, which couldresult in peak exposures higher than suggestedby our measurements.We found that the average indoor concen-

tration of CO, was above the comfort value of1000 ppm in many dwellings in Sweden. Thisinadequate outdoor air supply leads toincreased concentrations of volatile organiccompounds, as well as other indoor pollu-tants. Either the rate of outdoor air supply, orthe indoor concentration of carbon dioxide,are used as criteria of ventilation by most ven-tilation standards,'1132 and a comfort value of1000 ppm of CO2 has been adapted for workenvironments in Sweden.'2 There is, however,sparse information in the scientific literatureon the medical significance of poor outdoorair supply. Our result, however, has shownthat nocturnal symptoms related to asthmaare more common if the outdoor air supply isinadequate.We showed a relation between asthmatic

symptoms and signs of dampness in the build-ing or microbial growth, which is in agree-ment with earlier observations."1-2 A relationbetween house dust mites and nocturnal asth-matic symptoms was also shown, even thoughmost (95%) of the subjects had a negativeprick test to Dermatophagoides pteronyssinus.Moreover, none of the four subjects with thatpositive prick test had house dust mites intheir home, which suggests that they had beensensitised in earlier dwellings. A possibleexplanation for our findings could be that thepresence of house dust mites in dwellings is

related to other indoor pollutants that triggerasthmatic symptoms.

Respiratory effects of exposure to nitrogendioxide (NO2) from gas cooking have beenextensively studied in other countries.8 9 Ourstudy was performed in a region where no gascooking or gas heating was used. This made itpossible to perform a study on respiratoryeffects of VOCs, which was not confoundedby the emission of both VOCs and NO2 fromgas combustion. Unfortunately, we did notmeasure indoor NO2 in our study. As possiblesources of indoor NO2 in our study wouldhave been traffic exhaust gases and indoortobacco smoking, we expected the indoorconcentrations of NO2 to be below theobserved effect level of 30 ,Ug/m3 NO2.8 Thisassumption is supported by data from a recentstudy from mid-Sweden, where indoor con-centrations of NO2 were between 4 and 18,ug/m3 in urban dwellings and 1-10,ug/m3 inrural dwellings."We found significant relations between

measured indoor concentrations of variousVOCs in dwellings and symptoms related toasthma. To our knowledge, relations betweenasthmatic symptoms and measured VOCshave not been reported earlier in any epidemi-ological study. There are, however, some ear-lier studies that support the hypothesis thatasthmatic symptoms may be related to indoorVOCs. In one population study, presence ofnewly painted surfaces indoors was related toasthmatic symptoms.34 Also, two experimentalstudies have shown that even moderate con-centrations of VOCs may cause inflammationand obstructive reactions in the airways." 36 Inone study, it was shown that a four hour expo-sure of humans to 25 mg/M3 of a mixture ofVOCs induced an inflammatory response inthe nose, as shown by a significant increase ofgranulocytes in nasal lavage." In the otherstudy, the same mixture of VOCs, caused adecline of FEVy among asthmatic subjectsafter a 15 hour exposure to 25 mg/m3 ofVOCs.36 To our knowledge, no exposurechamber studies of the effect on health relatedto asthma ofVOCs below 25 mg/m3 are avail-able. The average exposure in our study waswell below 25 mg/m3n, but a maximum con-centration of 9 mg/m3 was measured in thebedroom of one symptomatic subject.Moreover, the exposure time is much longerin dwellings than in exposure chamber stud-ies, and the types of VOCs in our dwellingsdiffered from the mixture of VOCs used instudies mentioned above. Thus we do notconsider it unlikely that the concentrations ofVOCs measured in our study may contribute toasthmatic symptoms in some sensitive subjects.Our findings that indoor concentrations of

formaldehyde are related to asthmatic symp-toms agree with another study that suggeststhat dwellings with more than 30,g/m3 offormaldehyde had a higher proportion of chil-dren (< 15 years) with abnormal variability inPEF."7 Because of covariation betweenformaldehyde and VOCs, we were not able toseparate the effect of these volatile com-pounds in our statistical analysis. Terpenes,however, were the only volatile compounds

394



.bmj.com

/O

ccup Environ M


ownloaded from

http://oem.bmj.com/


significantly related to both symptoms andclinical signs, including variability in PEF andbronchial hyperresponsiveness. We admit thatit would have been better if we had measuredpeak flow four times a day, and a more fre-quent measurement of PEF may haveincreased the likelihood of detection of furtherhealth effects of indoor VOCs. Terpenes areusually considered to be harmless naturalproducts, emitted from wood, paint, food,cleaners, and other consumer products. Amaximum indoor concentration of IP0 mg/m3of terpenes was measured in our study, abouttwo thirds being limonene. No activitiesknown to emit terpenes, such as cleaning, orpeeling of citrus fruits, were found that couldexplain the high values. Obstructive airwayreactions to terpenes have earlier been shownat higher concentrations in sawmills.38 Ourinvestigation, however, suggests that somesensitive subjects may show bronchialobstructive reactions at low exposures to ter-penes in the indoor environment.To improve management of asthma, and to

counteract the increase in asthma, the impor-tance of the indoor environment should notbe neglected. Our results suggest that indoorexposure to VOCs may affect the airways andcause asthmatic symptoms. Improved indoorenvironment can be achieved by variousmeans-for example, selecting building mater-ial, building construction, and indoor activi-ties on the principle that the emission ofVOCs should be as low as reasonably achiev-able. There is also a need to increase the out-door air supply in many dwellings in Sweden.Finally, wall to wall carpeting, which could actas a depot for various types of pollutants, anddampness in the building should also be avoided.

This study was supported by grants from the SwedishAssociation against Asthma and Allergy, The Swedish MedicalResearch Council, The Swedish Society of Medicine, TheSwedish Heart and Lung Foundation, The Bror HjerpstedtsFoundation, Pharmacia and the County Council of Uppsala.

1 Samet JM, Marbury MC, Spengler JD. Health effects ofindoor air pollution. Part II. Am Rev Respir Dis 1989;137:221-42.

2 Samet J. Environmental controls and lung disease. Am RevRespir Dis 1990;142:915-39.

3 Boman G. Epidemiology of obstructive lung diseases inthe Nordic countries. European Respiratory Reviews 1991;1:316-20.

4 Aberg N. Asthma and allergic rhinitis in Swedish con-scripts. CGin Exp Allergy 1989;19:59-63.

5 Burr ML, Butland BK, King S, Vaughan-Williams E.Changes in asthma prevalence: two surveys 15 yearsapart. Arch Dis Child 1989;64:1452-6.

6 Sears MR. Epidemiological trends in bronchial asthma. In:Kaliner MA, Barnes Pi, Persson CG, eds. Asthma. Itspathology and treatment. New York, Dekker, 1991:1-50.

7 Moschandreas DJ. Exposure to pollutants and daily timebudgets of people. Bull NYAcad Med 1981;57:845-59.

8 Hasselblad V, Eddy DM, Kotchmar DJ. Synthesis of envi-ronmental evidence: nitrogen dioxide epidemiologystudies. JAir Waste Manage Assoc 1992;42:662-71.

9 Viegi G, Paoletti P, Carrozzi L, Vellutini M, Ballerin L,Biavati P, et al. Effects of home environment on respira-tory symptoms and lung function in a general populationsample in North Italy. Eur Respir_ 1991;4:580-6.

10 Burge PS, Richardson MN. Occupational asthma due toindirect exposure to lauryl dimethyl nenzyl ammoniumchloride used in a floor cleaner. Thorax 1994;49:842-3.

11 Waegemaekers M, Van Wageningen N, Brunekreef B,Bolsei JSM. Respiratory symptoms in damp houses.Allergy 1989;44:192-8.

12 Platt SD, Martin CJ, Hunt SM, Lewis CW. Damp hous-ing, mould growth and symptomatic health state. BMJ1989;298:1673-8.

13 Wickman M, Nordvall SL, Pershagen G, Sundell J,Schwartz B. House dust mite sensitization in childrenand residential characteristics in a temperate region.JAlUergy Clin Immunol 1991;88:89-95.

14 Burge PS, Finnegan MJ, Horsfield N, Pickering CAC.Occupational asthma in a factory with a contaminatedhumidifier. Thorax 1985;40:248-54.

15 Norback D, Edling C, Wieslander G, Ramadhan S.Exposure to volatile organic compounds (VOC) in thegeneral Swedish population and its relation to perceivedindoor air quality and sick building syndrome. In:Jaakola JJK, Ilmarinen R, Seppanen 0, eds. Proceedingsof indoor air '93 vol 1, July 1993, Helsinki. Helsinki,Finland: Laboratory of Heating, Ventilation and AirConditioning, Helsinki University of Technology, 1993:573-8.

16 Sporik R, Holgate ST, Platts-Mills TEA, Cogswell JJ.Exposure to house-dust mite allergen (der p I) and thedevelopment of asthma in childhood. N Engl J Med1990,323:502-7.

17 Burney PGJ, Luczynska CM, Chinn S, Jarvis D. TheEuropean Community respiratory health survey. EurRespirJ 1994;7:954-60.

18 Andersson K, Hallgren C, Levin J-O, Nilsson C-A.Chemosorbtion sampling and analysis of formaldehydein air: influence of recovery during the simultaneoussampling of formaldehyde, phenol, furfural, and furfurylalcohol. ScandJ3 Work Environ Health 1981;7:282-9.

19 NorbAck D, Torgen M, Edling C. Volatile organic com-pounds, respirable dust, and personal factors related toprevalence and incidence of sick building syndrome inprimary schools. BrJ Ind Med 1990;47:733-4 1.

20 Van der Brempt X, Haddi E, Michel-Nguyen A, Fayon JP,Soler M, Charpin D, et al. A comparison of theACAREX test with monoclonal antibodies for the quan-tification of mite allergens. J Allergy Clin Immunol 1991;87:130-2.

21 Chapman MD. Guanine-an adequate index of miteexposure? Allergy 1993;48:301-2.

22 Burney PGJ, Chinn S. Developing a new questionnaire formeasuring the prevalence and distribution of asthma.Chest 1987;19(suppl):79-92.

23 Burney PGJ, Laitinen LA, Perdrizet, et al. Validity andrepeatability of the IUATLD (1984) bronchial symp-toms questionnaire: an international comparison. EurRespirJ 1989;2:940-5.

24 Dreborn S. Skin tests used for epidemiological studies.Allergy 1989;44(suppl):52-9.

25 European Community for Coal and Steel. Standardizationof lung function tests. Clinical Respiratory Physiology1983;19(suppl 5):22-7.

26 Higgins BG, Britton JR, Chinn S, Jones TD, Jenkinson D,Burney PG, Tattersfield AE. The distribution of peakexpiratory flow variability in a population sample. AmRev Respir Dis 1989;140:1368-72.

27 Higgins BG, Britton JR, Chinn S, Cooper S, Burney PGJ,Tattersfield AE. Comparison of bronchial reactivity andpeak expiratory flow variability measurements for epi-demiological studies. Am Rev Respir Dis 1992;145:588-93.

28 Knox AJ, Wisniewski A, Cooper S. Tattersfield AE. Acomparison of the Yan and a dosimeter method formethacholin challenge in experienced and inexperiencedsubjects. Eur RespirJ 1991;4:497-502.

29 Peterson CGB, Enander I, Nystrand J, Andersson AS,Nilson L, Venge P. Radioimmunoassay of humaneosinophilic cationic protein (ECP) by an improvedmethod. Establishment of normal levels in serum andturnover in vivo. Clin ExpAllergy 1991;21:561-7.

30 Gebski V, Leung 0. McNeil D, Lunn D. SPIDA usersmanual, version 6. Eastwood, Australia: StatisticalComputing Laboratory, 1992.

31 ASHRAE. ASHRAE standard 62-89. Ventilation foracceptable indoor air quality. Atlanta: American Societyof Heating, Refrigeration, and Airconditioning.Engineers Inc Atlanta, 1989.

32 National Swedish Board of Occupational Safety andHealth. Ventilation and quality of air, AFS 1993;5.Stockholm: NSBOSH 1993. (In Swedish.)

33 Berglund M, Briback L, Bylin G, Jonson J-O, Vather M.Personal NO2 exposure monitoring shows high exposureamong ice-skating schoolchildren. Arch Environ Health1994;49:17-24.

34 Norbick D, Edling C, Wieslander G. Asthma symptomsand the sick building syndrome (SBS)-the significanceof microorganisms in the indoor environment. In:Samson RA, Flannigan B, Flannigan ME, Verhoeff AP,Adan OCG, Hoekstra ES, eds. Health implications offungi in indoor environments. Air quality monographs, vol 2.Amsterdam: Elsevier, 1994:229-39.

35 Koren HS, Graham DE, Devlin RB. Exposure of humansto volatile organic mixture III. Inflammatory response.Arch Environ Health 1992;47:39-44.

36 Harving H, Dahl R, Molhave L. Lung function andbronchial reactivity in asthmatics during exposure tovolatile organic compounds. Am Rev Respir Dis 1991;143:751-4.

37 Quackenboss ni, Lebowitz MD, Michaud JP, BronnimanD. Formaldehyde exposure and acute health effectsstudy. Environmental International 1989;15: 169-79.

38 Johard U. The wood trimmer's lung. Clinical and experi-mental studies focusing on inflammatory reactions.Stockholm: Unit of Lung Medicine, Institution ofMedicine, Karolinska Hospital, Karolinska Institute,1994. (Thesis.)

395



.bmj.com

/O

ccup Environ M


ownloaded from

http://oem.bmj.com/

environmental asthmatic symptoms and volatile organic

Documents