determinants of dust mite allergen concentrations in infant bedrooms in tasmania

Determinants of dust mite allergen concentrations in infantbedrooms in Tasmania

D. COUPER, A.-L. PONSONBY and T. DWYER

Menzies Centre for Population Health Research, University of Tasmania, Hobart, Tasmania, Australia

Summary

Background High exposure to house dust mite allergen during the first year of life hasbeen found to increase the risk of subsequent asthma and mite sensitization. Environmentalfactors, home construction and cleaning methods used are associated with levels of dustmites in the home.Objective To investigate determinants of levels of Der p 1 and Der f 1 mite allergens inhomes of infants in southern Tasmania.Methods Dust samples were collected from 72 homes of infants participating in theTasmanian Infant Health Survey (TIHS). The Der p 1 and Der f 1 allergen concentrations inthese samples were measured. The TIHS interviewers obtained information from themothers of the infants via a questionnaire, observed specified aspects of the homeenvironment, and took readings of bedroom temperature and humidity. The effect ofeach item on allergen concentration in dust from bedroom floors was examined in a varietyof ways. Those items which in this study appeared to be significantly related to allergenconcentrations plus items which in other studies have been found to be related to allergenconcentrations were then investigated further in multivariate models.Results Der p 1 allergen concentration (mg/g) and density (mg/m2) in dust from bedroomfloors were found to be related to several home environment factors. In the univariateanalyses, indoor humidity, 24 h maximum temperature, number of residents and acombination of floor covering and cleaning methods appeared to have a significant effecton allergen levels. These factors remained important in the multivariate model except thatindicators for mould in the bathroom and drying washing on an outside line replaced indoorhumidity.Conclusion Features related to home dampness, the number of residents and floorcovering and cleaning were major determinants of Der p 1 levels in the bedrooms studied.

Keywords: allergens, environment, house dust mites, housing

Clinical and Experimental Allergy, Vol. 28, pp. 715–723. Submitted 17 March 1997;revised 21 October 1997; accepted 19 January 1998.

Introduction

The prevalence of childhood atopic disease appears to beincreasing [1]. Environmental factors are thought to beimportant in the development of atopic disease [2]. Highexposure to house dust mite allergen during the first year oflife has been found to increase the risk of subsequent asthmaand mite sensitization [3].

In Tasmania, a cross-sectional asthma survey was con-ducted in 1995 on children who had their seventh birthday inthat year. A subset of these children had been enrolled in theTasmanian Infant Health Survey (TIHS) in 1988. The TIHSincludes data collected on many features of the infant andinfant’s environment at 1 month of age. The two data setshave been record linked to investigate how infant environ-mental factors relate to subsequent asthma.

For a subgroup of the TIHS infants born in 1995, addi-tional information was collected on mite allergen levels inthe homes of participating infant cohort families. The

Clinical and Experimental Allergy,1998, Volume 28, pages 715–723

715q 1998 Blackwell Science Ltd

Correspondence: Dr D. J. Couper, Menzies Centre for Population HealthResearch, University of Tasmania, GPO Box 252–23, Hobart, TasmaniaAustralia 7001.

purpose was to use the comprehensive data on multipleexposures to investigate how various characteristics of theinfant’s home, bedroom and bedding related to Der p 1allergen levels in dust on the infant’s bedroom floor and inbedding in these homes.

Tasmania, the island state of Australia, has a cooltemperate climate. In 1995, the average maximum andminimum summer temperatures were 218C and 128C,respectively, with an average relative humidity at 09.00 hof 66%. The winter values were 128C, 58C and 76%,respectively [4].

Methods

The Tasmanian Infant Health Survey (TIHS)

A prospective infant cohort study was conducted in Tasma-nia from 1988 to 1995 to investigate the aetiology of suddeninfant death syndrome (SIDS). It involved the six majorobstetric hospitals in Tasmania, representing< 93% of livebirths in this state. Infants born in these hospitals wereassessed using a scoring system to predict those at higherrisk of SIDS. The infants were given a composite scorebased on maternal age (score declining linearly with age),birth weight (higher scores for low birth weight categories),season of birth (highest scores for infants born in March orApril), infant sex (higher score for males), duration of thesecond stage of labour (higher scores for Caesarean deliveryor short second stage), and intended infant feeding (higherscore if not intending to breast-feed). The< 20% of liveborn infants who had a score over a specified cut-off pointwere eligible to join the study. All multiple births were alsoincluded in the study. Infants with severe neonatal disease ora major congenital anomaly, infants who would not beresident in Tasmania at 1 month of age, and infants foradoption, were excluded from the study. Of the 1288 infantsin the 1995 birth cohort in the study, 33.9% were born inMarch or April and 29.2% in May, June or July; 68.4% weremale; 20.1% were born to teenage mothers; 23.1% had abirth weight less than 2500 g; and 60.0% of mothers statedthat they intended to breast-feed.

Standard study measurements were collected by researchassistants in three stages. Firstly, a hospital interview wasconducted when the infant was 4 days of age. Secondly, ahome visit was conducted after the infant had reached 4weeks of age — usually during the fifth postnatal week.Thirdly, a phone interview was conducted after the infantreached 10 weeks of age. The study is described in detailelsewhere [5].

At the home visit many of the responses were provided bythe mother, while others were from observations or mea-surements made by the interviewer. The questions ofpotential relevance to the dust mite allergen part of the

study were on construction of the home; outdoor and indoorenvironmental conditions; heating methods; floor coveringand cleaning; number of residents and number of smokers inthe household; sleeping arrangements for the baby; andsundry other factors. The number of external walls in theinfant’s bedroom was noted as this may affect ventilationefficiency and the variation of temperature in the room.Some of the questions were very similar. For instance, aspart of the standard TIHS interview the mother was asked‘Have you noticed mould inside your house (excludingbathroom)?’ and as part of the dust sampling procedurethe mother was asked ‘For the last month has there beenmould in other areas [other than bathroom], i.e. walls/ceiling/window ledges?’ (Of the 69 mothers whoseresponses to both these questions were recorded, nineanswered ‘yes’ to one of these questions but ‘no’ to theother.) In such instances the responses were used as givenand the associations between the questions and mite allergenconcentrations were investigated separately.

Dust sample collection

Permission to collect dust samples was requested from 75mothers of infants born in southern Tasmania betweenOctober and December 1995 who were enrolled consecu-tively in the TIHS. Three mothers refused permission. Twonurse interviewers collected dust samples from 72 homes,corresponding to 80 infants (eight sets of twins). In additionto the dust samples and the standard study protocol for theTIHS home visit, the nurse interviewers also collectedadditional information on factors thought to be related todust levels. The dust sampling procedure followed was oneused previously [6].

Bedroom floor dust sampleDust was collected from the floor of the room where theinfant (s) had spent the previous night. Although this site isreferred to as the bedroom, in at least one case it wasactually the living room of the home. In all cases the roomconcerned was the one in which the baby typically spentmost of each night. Just 8.7% of babies usually slept alone ina bedroom at night while 31.0% had shared a bed with aparent or another child the previous night. An area close tothe side of the cot or bed was used as the site for thecollection of the sample. A template marking the perimeterof a 1 m2 area was laid out. Dust was collected from thesquare metre for 1 min using a modified handheld vacuumcleaner. If insufficient dust was collected, interviewerscollected a sample from an adjacent square metre for afurther minute.

Infant bedding dust sampleThe infant bedding was sampled in the following way. Theitems of underbedding were vacuumed, starting at the

716 D. Couper, A.-L. Ponsonby and T. Dwyer

q 1998 Blackwell Science Ltd,Clinical and Experimental Allergy, 28, 715–723

bottom and working up through the layers, excludingseparate items such as pillows, sheepskins or syntheticpiled underlays. Vacuuming was conducted for 1 min.First, the top surface of the mattress was vacuumed, con-centrating particularly on the area occupied by the baby, andthen the blanket, lower sheet and any mattress liners orblankets. If insufficient dust was obtained, the site wassampled for a further minute. Separate dust samples werecollected for pillows and sheepskins by vacuuming eachitem for 1 min.

Living room floor dust sampleIn the first 15 homes visited an additional sample wascollected in the main living room of the house usingthe same protocol as for the bedroom, except that inthis case the one square metre was marked out in themiddle of the floor. Collection of the living room floorsample was discontinued as the overall sampling procedurewas found to be too onerous for the nurse interviewers andsubjects.

For a small proportion of samples (n¼ 11), the inter-viewers did not complete the time or area sampling infor-mation, but they stated they did not deviate from theprotocol so in these cases the values of 1 min or 1 m2

were assumed. Samples were stored in a freezer at 48C toprevent mite proliferation before being transferred to theUniversity of Sydney for analysis. Samples were sieved and50 mg portions of dust were extracted in 1 mL of phosphatebuffered saline. Der p 1 and Der f 1 allergen concentrationswere measured using a double monoclonal antibody ELISAtest [7].

Statistical methods

For all dust samples the amount of allergen present wasexpressed as a concentration (micrograms of allergen pergram of fine dust). For the samples taken from a definedarea, the density in micrograms of allergen per square metrewas also calculated. In this paper ‘allergen levels’ is usedgenerically to refer to both concentration and density. Theallergen concentration inmg/g was used as the primaryoutcome measure.

For all sites the distributions of allergen levels weremarkedly skewed to the right. This has also been reportedin other studies. Consequently, it has become standardpractice to report geometric means of allergen levels.Because of the extreme skewness, the data were analysedin a variety of ways to check that results obtained were notjust due to violations of the assumptions of a particularmodel. Univariate non-parametric analyses were conductedusing the Kruskal–Wallis [8] test applied to the (untrans-formed) data. For both univariate and multivariate analyses,linear models [8] were applied to the natural logarithms of

the data. (Some other studies have used base 10 logarithms.Geometric means are not affected by the base used forlogarithms.) For calculations using logged data, samples inwhich no Der p 1 allergen was detected were assigned anallergen concentration of 0.05mg/g. This value was chosenbecause the lowest recorded allergen concentration was0.08mg/g and the assumption was made that the allergenconcentrations in the samples concerned were not truly zerobut rather were below the detection limit of the assay. Theeffect of this assumption was examined by repeating theanalyses using 0.005mg/g for the samples with no detectedallergens, as well as by noting the Kruskal–Wallis testresults, which are not affected by this assumption. Thedata were also analysed using logistic regression [8] withthe dependent variable being an indicator of whether or notthe allergen concentration exceeded 2mg/g. These analyseswere repeated using a cut-off of 10mg/g. Except wherethe various approaches yielded conflicting results, reportedP-values are for the test that the coefficient for that variableis 0 in a linear model applied to the logged data with 0 sassumed to be 0.05. No adjustments have been made formultiple testing. However, the numbers of ‘exposures’investigated are reported so that a concerned reader canmake an appropriate adjustment. Statistical analyses wereconducted using SAS version 6.09 [9]. In conductingstatistical tests,P-values less than 0.05 were regarded asbeing statistically significant.

Results

Mite species

Thirty of the dust samples taken from beds were selected atrandom. These were examined for presence or absence ofthe two major dust mite species. In 10 of these samples nomites were found, 15 hadDermatophagoides pteronyssinusmites only, two hadD. farinae mites only and three con-tained both species. The small proportion of samples withD. farinae is unusual for Australia. Just six of 120 samplesanalysed for Der f 1 contained detectable concentrations ofthis allergen. Consequently the rest of this paper focuses onDer p 1 allergen levels.

Mite allergen levels in the bedroom floor dust

As previously noted, the distribution of Der p 1 allergenconcentrations was markedly skewed to the right, having(arithmetic) mean 5.40mg/g (geometric mean 2.04mg/g),median 2.21mg/g, 75th percentile 6.20mg/g and maximumvalue 41.32mg/g. The lowest detected antigen level was0.08mg/g and there were four samples in which no allergenwas detected. The distribution of allergen concentrations forthe bedroom floor samples is shown in Fig. 1.

Dust mite allergen in Tasmania 717


The mite allergen concentration was not strongly relatedto the amount of dust in the sample (Spearman’s correlationcoefficient r ¼ 0.16, P¼ 0.17). The allergen concentrationin the bedroom floor was not significantly related to that inthe infant mattress and underbedding (r ¼ 0.14, P¼ 0.27)nor to that in sheepskins or pillows. The allergen concen-trations in the bedding samples are plotted against those inthe bedroom floor samples on a log scale in Fig. 2. Nomeaningful association is obvious in the figure. Althoughthe allergen concentration in the bedroom floor dust tendedto be greater than that in the bedding (geometric mean2.04mg/g vs 1.56mg/g), the difference was not statisticallysignificant (pairedt-test P¼ 0.28). Floor samples weretaken in the living rooms of 15 homes and the allergenlevels in these samples were also not significantly related tothose in the bedroom floor (r ¼ 0.32,P¼ 0.25).

Univariate analyses

Linear models were used to examine how mite allergenlevels were associated with various factors. Responses to< 70 questions were investigated. For those factors whichwere statistically significant in this data set or which havebeen reported in other studies as being significantly relatedto mite allergen levels, Table 1 contains geometric means ofmite allergen concentration inmg/g and density inmg/m2

and correspondingP-values. Some of the categorizationsreported in the table were constructed after examining moregeneral responses. For instance, for number of residents, thegeometric means were similar for households with two,three, four or five residents so households with 2–5 resi-dents were compared with those with six or more residents.

Factors associated with higher allergen levels were six ormore residents in the household; low maximum temperaturein the infant’s bedroom; high indoor humidity; and frequentvacuum cleaning. Home construction, location and heatingwere not associated with allergen levels. Although thenumber of residents in the household was associated withallergen levels, levels did not vary by the number of roomsin the house nor by occupant density (number in household/number of rooms) nor by the number of occupants in thebedroom in the preceding night, by whether the mother and/or other household residents smoke or whether airfreshnerspray is used in the bedroom.

Humidity and temperatureThe external humidity (range 28 – 100%, measured by asling hygrometer at the home interview) was not linearly



Fig. 1. Distributions of mite allergen concentration (mg/g) anddensity (mg/m2) on infant bedroom floors.

Fig. 2. Mite allergen concentration (mg/g) on bedroom floor vs inbedding.

related to the log of the bedroom floor allergen concentra-tions (P¼ 0.26) and there was no evidence for a non-linearrelationship either. The internal humidity (range 30 – 84%,median 54%) was linearly related to the log of the allergenconcentrations (P¼ 0.037) with higher concentrations asso-ciated with higher humidity. The level of internal humidityon the day of visit did not differ significantly by whether thehome was noted to have mould or not, nor by whetherwashing was dried on an outside line. Neither the differencebetween indoor and outdoor humidity nor the ratio of thesetwo was significantly associated with allergen levels.Twenty-four hour temperatures (measured using a maxi-mum–minimum thermometer left at the home at the time ofthe home visit) were collected at 57 of the 72 homes. Theminimum bedroom temperature (range 7–278C, median158C) was not linearly related to the log allergen concentra-tions (P¼ 0.21), but the maximum temperature (range 17–308C, median 238C) was inversely related (mean log aller-gen concentration decreased by 0.18 with each 18C increasein temperature,P¼ 0.015).

Floor covering and usual cleaning methodBecause of the highly unbalanced distribution of theseexposures, the results are ‘driven’ by a relatively smallnumber of subjects. For instance, just eight of the 72bedrooms did not have carpeted floors. Although carpetedfloors tended to have higher allergen levels than other floors(see Table 1) this association was not statistically significanton its own. Frequent vacuum cleaning (on average at leastonce a week) was associated with significantly higherallergen concentrations than infrequent vacuuming (geo-metric mean of 2.58mg/g vs geometric mean 0.47mg/g,P¼ 0.0016). Other information on cleaning was also con-sidered but the allergen concentrations appear to be mostsuccinctly summarized by the three combinations of floorcovering and vacuum cleaning in Table 1. There was anonsignificant (P¼ 0.24) tendency for the total amount ofdust to be lower in non-carpeted bedrooms (n¼ 8, geo-metric mean 0.06 g) than carpeted (n¼ 64, mean 0.08 g).Frequent vacuuming was not significantly associated withthe total amount of dust present (P¼ 0.09), with slightlymore dust in the frequently vacuumed bedrooms (geometricmean 0.08 g vs 0.05 g).

Multivariate results

Variables which were significant or nearly so in the uni-variate analyses and those which were considered to bepotential confounders of the relationship between thesevariables and mite allergen levels were investigated furtherin multivariate models.

In modelling mite density inmg/g, five factors appeared tobe important. One of these, maximum temperature, had a

substantial number of missing values — information on allfive factors was available in just 56 cases whereas informa-tion on the four factors other than maximum temperaturewas available for 69 of the 72 cases. Table 2 contains resultsof fitting models with and without maximum temperature tothe logged mite allergen concentrations inmg/g.

Using the same variables as in Table 2 to model allergendensity inmg/m2, R2 values of 0.44 or 0.35 were obtaineddepending on whether or not maximum temperature wasincluded.

Infant bedding

The allergen concentration did not vary by type of mattress(P¼ 0.66). The mean mite allergen concentrations weresimilar for underbedding (n¼ 68, geometric mean 1.62mg/g, 95% confidence interval [1.17, 2.22]) and sheepskin andsynthetic underblanket, (n¼ 9, mean 1.75mg/g, 95% CI[0.88, 3.48]) but these allergen concentrations were notsignificantly associated (Spearmanr ¼ – 0.52, P¼ 0.15).Pillows are not commonly used for 1-month-old infants.For the 13 samples from pillows the mean allergen con-centration was 2.71mg/g (95% CI [1.09, 6.71]) and theseconcentrations were also not associated with those in theunderbedding (Spearmanr ¼ 0.25,P¼ 0.40).

Discussion

Der p 1 allergen concentration (mg/g) and density (mg/m2)were found to be related to several home environmentfactors in this study on the homes of 1-month-old infants.

The presence of mould in the home and a large number ofresidents ($ 6) were positively associated with both out-comes. Using an outside line to dry washing was associatedwith lower allergen concentration and density, as was anincrease in maximum 24 h bedroom temperature in therange of 17–258C in these homes, which were sampledduring spring and early summer. Indoor humidity waspositively associated with allergen concentration but notwith allergen density in the univariate analysis. Once othervariables, including the presence of mould, were taken intoaccount, the bedroom measurement of internal humidity hadlittle additional effect. However, it could be that eventhough the level of internal humidity on the day of visitwas not related to the presence of mould, mould may still bean indicator of higher long-term levels of internal humidity.Water vapour generated within the home is derived fromvarious sources, including the drying of clothes [10]. As justsix mothers mentioned a main drying method other than anoutside washing line it was not feasible to compare theseother methods statistically. However, in all these cases themite allergen concentration was higher than the geometricmean allergen concentration for the homes where washing





Table 1.The relationship between various home environment characteristics and the mite allergen concentration and density in bedroomfloor samples. Factors withP-values<0.05 are shown in bold

Mite allergen concentration Mite allergen densityin mg/m2 in mg/m2

Home characterisitic n geometric mean P-value geometric mean P-value

Construction and locationHouse construction 0.25 0.31

weatherboard 31 3.02 0.25brick veneer 33 1.60 0.12double brick 1 4.02 0.09other 7 0.99 0.08

House floor type 0.32 0.36wooden 56 2.37 0.19concrete slab on ground 9 0.82 0.06raised concrete slab 5 2.31 0.09other 2 1.23 0.18

House in windy location 37 1.78 0.48 0.13 0.36not in windy location 35 2.34 0.19

HeatingMain heating electrical or central heating 19 1.42 0.26 0.08 0.06

other forms of main heating 53 2.32 0.20Use a woodheater 45 2.11 0.68 0.18 0.26

do not use woodheater 24 1.78 0.10Heat living room 24h/day 39 2.03 0.98 0.16 0.80

< 24h/day 33 2.05 0.15Indictors of internal humidityMeasured indoor humidity 72 *1.04 0.04 *1.02 0.28

(per 1% increase)Dry washing on outside line 66 1.85 0.05 0.15 0.22

tumbledrier or other drying method 5 8.07 0.42Mother had noticed mould in house 0.05 0.002(not in bathroom) 15 4.21 0.57

mother had not noticed mould 57 1.68 0.11Mould in bathroom in past month 24 3.16 0.07 0.22 0.20

no mould in past month 46 1.50 0.12Mould other than in bathroom in 0.15 0.04past month 13 3.45 0.41

no other mould in past month 56 1.67 0.12Mould observed by interviewer in 0.49 0.14bedroom 7 3.04 0.42

mould not observed 65 1.95 0.14Floor covering and cleaningCarpet observed in bedroom 64 2.25 0.14 0.18 0.06

no carpet 8 0.91 0.05Bedroom has 1 external wall 18 2.08 0.49 0.15 0.76

2 external walls 48 2.22 0.173 external walls 6 0.96 0.09

Clean bedroom regularly? 0.10 0.07vacuum or sweep$ 1/week 64 2.28 0.18clean less frequently 8 0.83 0.05

Sweep bedroom ever 4 0.30 0.02 0.01 0.0009never 64 2.09 0.17

continued over

was dried on an outside line. The type of bedroom floorand frequency of vacuuming frequency were also found tobe associated with mite levels. These factors are discussedlater.

The outcomes studied here included both Der p 1 allergenconcentration inmg/g, which has been used in previous

studies [11,12] and concentration inmg/m2, as the lattertakes into account the amount of dust present in a site aswell as the allergen density within that dust [13]. Data on awide range of home environment variables were available.A high level of agreement between maternal report ofhousing details and interviewer observation of 65 homes



Table 1. continued

Mite allergen concentration Mite allergen densityin mg/m2 in mg/m2

Home characterisitic n geometric mean P-value geometric mean P-value

Vacuum bedroom > 1/week 41 2.46 0.007 0.18 0.0011/week 21 2.82 0.28< 1/week or never 10 0.47 0.02

Carpeting & vacuuming bedroom 0.002 0.0001no carpet & vacuum < 1/week 3 0.16 < 0.01carpet & vacuum < 1/week 7 0.75 0.06

vacuum$ 1/week 62 2.58 0.21MiscellaneousMaternal postnatal smoking 37 2.18 0.71 0.14 0.55

not smoking 35 1.89 0.18$ 6 residents in house 11 5.29 0.03 0.54 0.02

# 5 residents 61 1.71 0.13Birth order 1 or 2 54 1.75 0.18 0.13 0.09

3 or higher 18 3.19 0.30Nurse interviewer 0.90 0.53

nurse interviewer A 36 1.99 0.14nurse interviewer B 36 2.09 0.18

24 h maximum temperature 58 *0.84 0.01 *0.84 0.04(per 18C increase)

* Multiply geometric mean by this value for each 1 unit increase in the variable.

Table 2. Multivariate model for mite allergen concentration

Without maximum temperature With maximum temperature(n¼ 69, R2¼ 0.31) (n¼ 56, R2¼ 0.42)

Effect on geometric mean P-value Effect on geometric mean P-value

Mould in bathroom 2.11 0.048 2.17 0.056Dry washing on outside line 0.26 0.050 0.20 0.033$ 6 residents in home 3.42 0.013 3.69 0.012Floor covering and cleaning 0.014 0.027(baseline: no carpet, vacuum < 1/week)carpet & vacuum < 1/week 4.79 3.49vacuum$ 1/week 10.81 8.33Maximum temperature 0.86 0.020

‘Effect on mean’ is the estimated multiplicative effect this factor has on the geometric mean allergen concentration. For instance, in thismodel after adjusting for the effects of the other factors, homes with mould in the bathroom have mite allergen concentrations 2.11 timesthose without mould. In the case of maximum temperature the effect is of a 18C increase in temperature.

has been found in the infant health survey. In 1993,comparison of data from both sources for 65 homes foundthe level of agreement between the maternal report of thepresence of mould in the baby’s bedroom and the inter-viewer observation was high, with 98% agreement andkappa ¼ 0.531. With regard to bedroom floor covering,the maternal report was in total agreement (100%) withinterviewer observation (unpublished). In addition, mothersof study infants were asked to participate in the dustsampling project at the end of the home interview andthus did not have the opportunity to change bedding orclean the home before the dust samples were collected.

The study has the following limitations. Firstly, thefindings may not apply to other locations with differingclimatic conditions. For this reason, information on theclimate of Hobart, the state’s capital city located in southernTasmania, is provided in the introduction. Secondly, thesample size of the study was small, increasing the likelihoodof type II error. For example, in Table 1, despite a threefoldhigher level of Der p 1 inmg/m2 in carpeted bedroomscompared with uncarpeted bedrooms (n¼ 8), theP-value isonly of borderline significance (P¼ 0.06). A previous largerstudy found carpeted bedrooms to have 6–14 times higherlevels of mite allergens [12] although this has not alwaysbeen found [13]. Most studies have not controlled forpossible confounders, such as dampness in the home. Wehave attempted to control for some possible confounders.However, unmeasured confounders such as thickness andtype of carpet and the use of carpet in other rooms, maypartially explain the unexpected finding that carpeted bed-rooms which were vacuumed at least once a week hadhigher allergen levels and similar amounts of total dust persquare metre compared with those vacuumed less often. Analternative explanation is that vacuuming decreased thenumbers of the Cheyletus mite, which is a predator of theDer p 1 mite [14].

The lack of association between type of infant beddingand mite allergen concentration may be partially due to theage of infants (just 1 month old) in this sample. Some itemsof bedding were new (29.5% of mattresses, 37.5% ofpillows and 43.5% of sheets were no more than 6 monthsold) while many were old and had presumably been used byother infants. The wide variety of types of bedding made itimpractical to try to control for the age of each type.

The optimal conditions for Der p 1 mite growth in cultureare a temperature of< 258C [15] and 75% relative humidity[13,15]. However, an inverse relationship was foundbetween mite allergen levels and maximum 24 h bedroomtemperature (range 17–258C) in these homes. Here atemperature of 258C during the spring season may wellindicate that such homes would be very hot in summer, thuslowering mite populations, or poorly insulated, with lowbedroom temperatures in winter.

Home dampness was a major determinant of allergenlevels in these homes, as in previous studies [13,16–19]. InSweden, the ‘moisture trapping’ characteristic of the home,assessed by the ratio of external to internal humidity, wasassociated with higher allergen levels [20], but this was notobserved in our data. In the univariate analysis, the mean logallergen concentration of bedroom floor dust increased by0.40 (95% C.I. [0.03, 0.76]) for every 10% increase ininternal humidity measured on the day of visit. Possiblemarkers of long-term higher indoor humidity (mould inbathroom, washing not dried on outside line) remainedimportant in the multivariate model. A large number ofresidents in the home was also found to be associated withhigher Der p 1 levels, as reported previously [12,13]. Thelack of correlation between allergen levels on the bedroomfloor, infant bedding and in a subset, lounge room floor,indicates that there is considerable variation within homesand significant cross contamination of consecutive samplesobtained from the same vacuum cleaner did not occur.

In general, the levels of Der p 1 allergens in the bedroomfloor of these homes were lower than the levels reported inthe warmer Australian state of New South Wales. Thegeometric mean Der p 1 concentration in 72 homes testedin Sydney in 1991 was 22.5mg/gm [11] compared with theobserved mean Der p 1 concentration of 2.04mg/g for homesin this study conducted in a location with a cooler climate.Also, the prevalence of asthma was higher in New SouthWales (8.1%) than in Tasmania (6.5%) in the 1989–90National Health Survey of Australians in 22 000 dwellings[21].

Data from the 1995 Tasmanian asthma survey are cur-rently being analysed. The exposures which will be inves-tigated include the factors identified in this study as beingassociated with increased mite allergen levels.

Acknowledgements

The authors would like to thank Ajsa Mahmic of theUniversity of Sydney for the analysis of the dust samplesand Euan Tovey of the University of Sydney for commentson an earlier version of the manuscript.

References

1 Zeiger RS, Heller S. The development and prediction of atopyin high-risk children: follow-up at age seven years in aprospective randomised study of combined maternal andinfant food allergen avoidance. J Allergy Clin Immunol1995; 95:1179–90.

2 Peat JK, Toelle BG, Gray EJ et al. Prevalence and severity ofchildhood asthma and allergic sensitisation in seven climaticregions of New South Wales. Med J Aust. 1995 163:22–6.

3 Sporik R, Holgate ST, Platts-Mills TAE et al. Exposure to



house-dust mite allergen (Der p I) and the development ofasthma in childhood. N Engl J Med 1990 323:502–7.

4 Climate 1995. Commonwealth Bureau of Meteorology. HobartOffice, Australia.

5 Dwyer T, Ponsonby A-L, Newman NM et al. Prospectivecohort study of prone sleeping position and Sudden InfantDeath Syndrome. Lancet 1991 337:1244–7.

6 Marks GB, Tovey ER, Peat JK, Salome CM, Woolcock AJ.Variability and repeatability of house dust mite allergenmeasurements: implications for study design and interpreta-tion. Clin Exp Allergy 1995; 25:1190–7.

7 Luczynska CM, Arruda LK, Platts-Mills TAE et al. A two-sitemonoclonal antibody ELISA for the quantitation of the majorDermatophagoides spp. allergens, Der p I and Der f I. JImmunol Meth 1989 118:227–35.

8 Fisher LD, van Belle G. Biostatistics: a Methodology for theHealth Sciences. New York, John Wiley & Sons Inc., 1993.

9 SAS/STAT, User’s Guide Version. 6, 4th edn, Cary, NC. SASInstitute Inc., 1990.

10 Colloff MJ. Dust mite control and mechanical ventilation:when the climate is right. [Editorial] Clin Exp Allergy 1994;24:94–6.

11 Peat JK, Tovey E, Gray EJ et al. Asthma severity and morbidityin a population sample of Sydney schoolchildren: Part II –Importance of house dust mite allergens. Aust NZ J Med 199424:270–6.

12 Van Strien RT, Verhoeff AP, Brunekreef B et al. Miteantigen in house dust: relationship with different housing

characteristics in The Netherlands. Clin Exp Allergy 1994;24:843–53.

13 Warner JA. Creating optimal home conditions for the housedust mite. Clin Exp Allergy 1994; 24:207–9.

14 Wassenaar DPJ. Effectiveness of vacuum cleaning and wetcleaning in reducing house-dust mites, fungi and mite allergenin a cotton carpet: a case study. Exp Appl Acarol 1988; 4:53–62.

15 Murton JJ, Madden JL. Observations on the biology, behaviourand ecology of the house-dust mite, Dermatophagoides Pter-onyssinus (Trouessart) (Acarina: Pyroglyphidae) in Tasmania.J Aust ent Soc. 1977; 16:281–7.

16 Munir AK, Einarsson R, Kjellman NI et al. Mite (Der p 1,Der f 1) and cat (Fel d 1) allergens in the homes of babies with afamily history of allergy. Allergy 1993; 48:158–63.

17 Munir AK, Bjorkst’en B, Einarsson R et al. Mite allergens inrelation to home conditions and sensitization of asthmaticchildren from three climatic regions. Allergy 1995 50:55–64.

18 Harving H, Hansen LG, Korsgaard J et al. House dust miteallergy and anti-mite measures in the indoor environment.Allergy 1991; 46 (Suppl. II) :33–8.

19 Tovey ER. Allergen exposure and control. Exp Appl Acarol1992 16:181–202.

20 Sundell J, Wickman M, Pershagen G et al. Ventilation inhomes infested by house-dust mites. Allergy 1995 50:106–12.

21 Australian Bureau of Statistics. 1989–90 National HealthSurvey – Asthma and other respiratory conditions, Australia.Catalogue no. 4373.0. Commonwealth Government Printer,Canberra, 1991.



determinants of dust mite allergen concentrations in infant bedrooms in tasmania

Documents