Food Allergens in Mattress Dust in Norwegian Homes - A Potentially Important Source of Allergen Exposure

Abstract

Background

Sensitization to food allergens and food allergic reactions are mostly caused by ingesting the allergen, but can also occur from exposure via the respiratory tract or the skin. Little is known about exposure to food allergens in the home environment.

Objective

To describe the frequency of detection of allergens from fish, egg, milk, and peanut in mattress dust collected from homes of 13 year old adolescents, and secondly to identify home characteristics associated with the presence of food allergen contamination in dust.

Methods

Food allergens were measured by dot blot analysis in mattress dust from 143 homes in Oslo, Norway. We analyzed associations between home characteristics (collected by parental questionnaires and study technicians) and food allergens by multivariate regression models.

Results

Fish allergen was detected in 46%, peanut in 41%, milk in 39%, and egg allergen in 22% of the mattress dust samples; only three samples contained none of these allergens. All four food allergens were more frequently detected in mattresses in small dwellings (<100m²) than larger dwellings (≥130 m²); 63-71% of the small dwellings (n=24) had milk, peanut, and fish allergens in the samples compared to 33-44% of the larger dwellings (n=95). Milk, peanut, and egg allergens were more frequently detected in homes with bedroom and kitchen on the same floor as compared with different floors; with odds ratios of 2.5 (95% confidence interval (CI): 1.1, 5.6) for milk, 2.4 (95% CI: 1.0, 6.1) for peanut, and 3.1 (95% CI: 1.3, 7.5) for egg allergens.

Conclusions and clinical relevance

Food allergens occurred frequently in beds in Norwegian homes, with dwelling size and proximity of kitchen and bedroom as the most important determinants. Due to the amount of time children spend in the bedroom, mattress dust may be an important source of exposure to food allergens.

Introduction

Ingestion of food items is thought to be the most common route of exposure leading to food allergen sensitization in the general population [1]. Household consumption of peanuts is suggested to represent a risk factor for the development of peanut allergy in infants [2]. Allergy development caused by exposure to food allergens via the airways or the skin is mainly observed in occupational settings and experimental models [3-5], but could lead to asthmatic reactions or eczema in highly sensitized individuals [4, 6-8].

Children spend much of their time indoors, including in the bedrooms where mattresses can be important dust and allergen reservoirs. In 2001 it was estimated that on average, Americans spent 87% of their leisure time indoors [9], and Germans spent 75% of their time indoors at home [10]. Thus, our home environment is a potentially important source for allergen exposure. Little is, however, known about sources and prevalence of food allergens in environmental dust samples.

Egg and milk protein levels in 11 house dust samples from the Netherlands were reported to be high enough to elicit allergic symptoms in food sensitized patients [11]. Peanut protein levels measured in dust samples from 45 British homes were found to be related to household peanut consumption [12]. In Northern Norway, codfish and egg allergens were measured in dust from 38 homes and 7 schools. Codfish allergens were found in dust samples from living rooms, mattresses and class-rooms, while egg protein was less common in all locations [13]. Codfish allergen in schools was hypothesized to derive from passive exposure via clothing, but in general little is known about how food allergens may contaminate indoor environment.

We aimed to describe the frequency of allergens from fish, egg, milk and peanut in mattress dust collected from homes of Norwegian adolescents. Secondarily, we aimed to investigate if gender or home characteristics were associated with the presence of food allergens in mattresses.

Material and Methods

Study design and subjects

The present study is based on a 13-year follow-up of a subgroup of participants from The Environment and Childhood Asthma (ECA) study in Oslo, Norway [14]. Home visits were performed by study investigators from November 2005 to March 2007 (excluding spring and summer seasons). The 147 children with current asthma defined at the 10-year follow up [15], and 163 controls without lower respiratory disease were invited to participate in the present study. All participating families were offered individual reports on the outcome of the home inspection. Among the 146 participating families (see flow-chart of inclusion, Figure 1), 71 families had a child with current asthma diagnosed at the 10-year follow-up investigation. No asthma or allergy re-assessment was made at age 13 when the home inspections were performed.

Figure 1.

Flow chart of inclusion and exclusion of study participants

Home inspections

The home inspections were performed according to a pre-established checklist and included dust sampling, ventilation measurements, floor plan drawings with location of bedrooms, kitchen and dining room, descriptions of dwelling conditions, building materials, and visible signs of dampness and mould. The protocols for home inspections were adopted from the protocols used in the second step of the “Dampness in Buildings and Health” study performed in Sweden in 2001-2002 [16]. In addition, the parents also completed a questionnaire with information on frequency of cleaning, bed-making procedures, ventilation system, pet keeping and dwelling size.

Mattress dust sampling

Families were asked not to vacuum or wet clean floors and furniture in the 3-4 days prior to the inspection and not to replace the linen of the adolescent's bed during the week prior to home inspections. Dust from the mattresses (without sheets) were collected with vacuum cleaners (1600 W effect) equipped with special dust collectors (P-B Miljø A/S, Bjerringbro, Denmark). The whole surface area of the mattress was vacuumed for four minutes and the size of the mattress was recorded. The dust samples were kept frozen at −20 °C until further handling. The dust samples were weighed and extracted in phosphate-buffered saline with 0.05% Tween-20 (PBT) at the Norwegian Institute of Public Health according to protocols as previously described for allergen analyses [17]. Three samples had insufficient amounts of dust, resulting in a total of 143/146 dust samples available for food allergen analyses.

Determination of total protein and food proteins in dust

Total protein concentration in dust and detection of food proteins were performed at the Norwegian Veterinary Institute. Total protein concentrations were determined according to the Lowry method (DC protein assay, BioRad, Hercules, California, USA) using bovine serum albumin (BSA) as a standard. Absorbance was measured at 660 nm with a 1420 VICTOR²™ multilabel plate counter (Wallac, Turku, Finland).

Food protein detection in mattress dust was performed by semi-quantitative dot blot analyses with specific antibodies. Based on the measured protein content in the samples, two spots containing 1μg or 2μg of total protein were applied on the nitrocellulose membrane (BioRad, Hercules, California, USA) and allowed to dry. Four replicates of each blot were produced allowing for the determination of four different food proteins. Additionally, 1 μg of matched positive control proteins were applied on each blot. The control proteins were standards of hen's egg ovomucoid, peanut, and a 6-species fish preparation (cod, salmon, mackerel, herring, bluefin tuna and sprat) produced at the Norwegian Veterinary Institute as well as Hammerstein casein from cow's milk (Nutritional Biochemicals Corporation, Cleveland, OH, USA). In the following immunoblot analyses, tris-buffered saline (TBS) containing 0.1% Tween20 pH 7.6 was used as washing buffer, and after adding 3% BSA, also used as blocking and assay buffer. After blocking for 60 min, the dot blots were incubated under gentle shaking overnight at 4°C with the specific polyclonal antibodies.

Egg proteins were detected using a 1:1 mixture of anti-ovalbumin antibodies (Riedel-de-Haën, Seelze, Germany) and anti-ovomucoid antibodies (produced at the Norwegian Veterinary Institute) in 1:5×10⁶ dilution. Milk proteins were determined with 1:10⁵ diluted anti-total bovine casein antibodies (Pharmacia Diagnostics, Uppsala, Sweden). Fish proteins were detected with a 1:10⁴ dilution of polyclonal antibodies that had been raised in a Chinchilla rabbit against the 6-species fish preparation at the Norwegian Veterinary Institute. The antibodies detect a variety of fishes by binding to the fish allergen parvalbumin as well as several other fish proteins, but do not cross-react with other species. Peanut protein was analyzed by using 1:5×10⁶ diluted polyclonal anti-peanut antibodies, which had been raised against a total peanut protein extract. The anti-peanut antibodies bind peanut allergen (confirmed by Western blotting), but do not cross-react with other legumes.

After overnight incubation with the respective primary antibodies, blots were washed (3×15 min) and incubated for 1 h with HRP-conjugated goat anti-rabbit antibody (1:5000, Zymed, San Francisco, CA, USA) at room temperature. After a final wash, the membranes were developed with TMB substrate solution (Invitrogen, Carlsbad, CA, USA) until signals of satisfactory intensity appeared (1 to 8 min). The signal intensities of the dots were compared to control dot containing 1 μg of reference protein (Figure S1, online supplement). We grouped the “weak” and “definite” signals in a “probable detection” category and the “distinct” and “strong” signals into a “confirmed detection” category to avoid over-interpretation. The antibody dilutions had been optimized using the respective reference proteins in different dilutions, with the lower limit of detection at 10 ng protein for a weak signal.

Explanatory variables

Dwelling and household characteristics potentially influencing food allergens in mattress dust were retrieved from parental questionnaires or recorded by study technicians during home visits and included mattress age, use of house dust mite protection or other mattress coverings, bed making procedures, open window ventilation, mechanical ventilation system, cleaning procedures of floor and furniture in the adolescent's bedroom, carpet on the bedroom floor, pet keeping, dwelling size, dwelling type (apartment, townhouse, or single-family house), and the location of the bedroom relative to the kitchen (same versus different floors).

Statistical analyses

Pearson's χ² test was used to examine the association between frequency of mattresses with no, probable, and confirmed detection of food allergen with home characteristics and gender. Analysis of variance (ANOVA) was used to assess associations between frequency of mattresses with food allergens and mattress dust load in mg/m² (log₁₀-transformed). Multivariate regression models were applied separately for each of the four food allergens. We used multinomial models with the outcome of the dotblot analyses categorized in the three categories (“no”, “probable”, and “confirmed” detection of food allergens). The independent variables were gender, dust load, mattress age, bed making procedures, open window ventilation, ventilation system, wet floor cleaning, pet keeping, dwelling size, and location of the bedroom relative to the kitchen. All analyses were performed with Statistical Packages for Social Sciences (SPSS, version 19.0, SPSS InC. Chicago, IL, USA).

Results

The presence of milk allergen was confirmed in 39% of the 143 mattress dust samples, peanut in 41%, egg in 22% and fish allergen in 46%. Additionally, milk allergen was detected at the “probable” level in 29%, peanut in 34%, egg in 27% and fish allergen in 29% of the samples (Figure 2). Co-occurrence of the four food allergens differed considerably between the dust samples. Four, three and no food allergens were confirmed in 10 (7%), 21 (15%) and three samples, respectively. Total protein concentrations in the extracted mattress dust varied about 20-fold from 0.10 to 2.03 mg/ml, mean: 1.01 mg/ml (standard deviation: ± 0.43 mg/ml).

Figure 2.

Percentages of mattress dust samples (n=143) with no, probable, and confirmed detection of milk, peanut, egg, and fish allergens

Peanut allergen was more frequently found in mattresses of girls than boys (52% with confirmed detection for girls and 31% for boys, p=0.03), and similar for egg allergen (32% for girls and 14% for boys, p=0.03) (Table 1). The findings were confirmed in the multivariate logistic regression models (Table 2), whereas milk and fish allergen detection was similar between boys and girls (Table 1).

Table 1.

Percentages of mattresses (n=143) with probable and confirmed detection of milk, peanut, egg, and fish allergens by gender and home characteristics.

Characteristics	N	Milk (%)		Peanut (%)		Egg (%)		Fish (%)
		Probable detection	Confirmed detection	Probable detection	Confirmed detection	Probable detection	Confirmed detection	Probable detection	Confirmed detection
Gender
Male	78	19	37	37	31*	30	14*	24	44
Female	65	23	42	31	52	26	32	32	48
Mattress age
≤2 yrs	34	24	50*	35	44	29	24	21	41
>2 yrs, - 10 yrs	74	22	42	35	39	27	22	30	46
≥10 yrs	26	19	20	35	31	27	15	35	46
Bed usually made during the day
No	90	22	31*	32	40	30	19	30	44
Yes	49	20	53	35	45	27	29	22	51
Open window ventilation
< twice a week	33	15	42	24	42	18	27	30	52
≥ twice a week	107	23	38	42	41	32	21	26	45
Wet floor cleaning
< twice a month	61	20	31	31	36	31	21	26	49
≥ twice a month	79	23	46	35	46	27	23	28	44
Vacuum cleaning
< once a week	39	26	33	36	41	31	26	23	49
≥ once a week	100	20	42	33	42	28	21	29	45
Dry or wet clean furniture
< twice a month	98	24	39	33	41	27	22	27	45
≥ twice a month	42	17	41	36	43	26	21	29	50
Furry pets
No	78	24	39	39	40	35	19	28	47
yes	62	18	40	27	44	21	26	26	45
Ventilation system
None	11	27	55	55	36	9	27	9	55
Kitchen fan only	92	19	37	30	41	33	23	30	47
Other mechanical ventilation	35	26	40	37	43	23	20	23	43
Housing type
Apartment	25	12	56	24	64	24	48*	24	72*
Townhouse	41	20	39	42	32	27	22	32	42
Single family house	77	25	34	34	38	30	14	27	39
Dwelling size
<100 m2	24	8	63*	21	63	21	38	17	71*
≥100-129 m2	21	24	43	33	33	33	24	48	29
≥130 m2	95	24	33	37	38	30	18	25	44
Bedroom and kitchen on the same floor
No	91	25	31*	40	32*	29	15*	31	41
Yes	52	14	54	25	56	27	35	23	54

^*Chi-square test for differences in frequency of mattresses with no, probable, and confirmed detection of food allergens by gender and home characteristics (p≤ 0.05).

Table 2.

Multinomial logistic regression models for association between gender and home characteristics^* and food allergen detection in mattress dust. Only the variables that were statistically significant associated with the food allergens are reported.

Allergen	Characteristic	OR (95% CI) of probable detection relative to no detection	OR (95% CI) of confirmed detection relative to no detection
Milk
	Bed made during the day (no = ref)	1.7(0.6, 4.5)	2.6 (1.1, 6.0)
	Kitchen and bedroom on the same floor (no = ref)	0.8 (0.3, 2.1)	2.5 (1.1, 5.6)
Peanut
	Girls (ref: boys)	1.6 (0.6, 3.9)	3.0 (1.3, 7.5)
	Kitchen and bedroom on the same floor (no = ref)	0.9 (0.3, 2.4)	2.4 (1.0, 6.1)
Egg
	Girls (ref: boys)	1.2 (0.5, 2.6)	2.9 (1.2, 7.2)
	Kitchen and bedroom on the same floor (no = ref)	1.4(0.6, 3.1)	3.1 (1.3, 7.5)
Fish
	Dwelling size
	≥130 m2	1 (ref)	1(ref)
	≥100 -129 m2	2.4 (0.73, 8.0)	0.83 (0.23, 3.0)
	<100 m2	1.6 (0.33, 7.9)	3.9 (1.05, 14.6)

^*Variables included in the initial models: gender, dust load, mattress age, bed making procedures, open window ventilation, wet floor cleaning, furry pets, mechanical ventilation system, dwelling size, and bedroom and kitchen on the same floor.

The dust load increased with mattress age (p=0.012, ANOVA), and was lower in mattresses in bedrooms that were vacuum cleaned at least once a week, geometric mean (GM): 173 mg/m² (95% CI: 143, 208), as compared to less frequent vacuum cleaning, GM: 233 mg/m² (95% CI: 201, 272), p=0.06. Fish allergen was more often detected in mattresses with high versus low dust load (GM: 219 mg/m² (95% CI: 188, 254) versus 163 mg/m² (95% CI: 131, 203), respectively, p=0.04). Milk allergens were more often detected in newer (≤ 2 years) than older mattresses (≥ 10 years), p=0.006 (Table 1).

Food allergens and home characteristics

Frequencies of vacuum cleaning and cleaning of furniture in the adolescents’ bedroom were not associated with detection of allergens from milk, peanut, egg, and fish in mattresses (all p≥0.6) (Table 1). Adolescents, who usually had their bed made (covered) during the day, had a higher probability of having food allergen present in the mattresses, statistically significant for milk allergen only (Tables 1 and 2).

Type of housing, dwelling size, and having bedroom and kitchen on the same floor were statistically significant associated with detection of peanut, egg, and milk allergens in mattress dust (Tables 1 and 2), wheras detection of fish allergens were inversely related to dwelling size (Table 2).

Only four adolescents had house dust mite covers on their mattress and eight adolescents had carpets on the bedroom floor. Due to the low frequency these variables were not included in the statistical models.

Eight of the 13 children allergy towards (any) nuts at 10 years of age had detectable levels of peanut allergens in their beds (Supplemental Table S1).

Discussion

Food allergens from milk, peanut, egg, and fish were frequently detected in dust samples from mattresses of Norwegian adolescents, more often in the beds of girls than boys. Only 3/143 dust samples had none of the four food allergens present. The home characteristics most consistently associated with food allergen detection in mattress dust were dwelling size, and co-location of kitchen and bedroom on the same floor.

Detection of food allergens in mattresses of adolescents has to our knowledge not previously been demonstrated. However, one study has described peanut protein in dust collected from mattresses of infants and their parents [12], and egg, milk, and fish allergens have previously been described in dust samples collected from floors [11, 13]. The most commonly detected food allergens in the present study were fish allergens followed by peanut, milk and egg allergens. In contrast, according to Statistics Norway, the average relevant annual food consumption per person in 2005-2007 were 92 liters per person per year of milk (not including cheese and curd), 17.2 kg fish (all types, fresh and prepared) and 7.1 kg egg [18], whereas no information is available for peanut consumption. However, peanut protein levels in house dust have by others been reported to be related to household peanut consumption [12].

The higher prevalence of fish than egg allergen in mattress dust, are also in line with a previous publication from Norway, where codfish allergens were found in dust samples from both living-rooms and class-rooms [13] indicating that fish proteins could be spread passively via clothing. Fish may be prepared and cooked at home, and fish proteins get easily aerolized during processing and cooking [19]. Also, fish is often used on sandwiches such as in smoked salmon and mackerel in tomato paste. Parvalbumin, the main fish allergen, is quite resistant to degrading conditions and is stable in different environments. Peanut allergen was confirmed in 41% of the mattress dust samples. Although the major peanut allergens do not easily become airborne [20], they are stable in the environment for some time [21], and peanut-containing foods tend to be sticky and could be spread passively by the consumer on skin or clothing [12]. The major milk and egg allergens are heterogeneous conglomerates of proteins with different properties regarding stability and ability to migrate in the environment [22, 23].

The higher frequency of peanut and egg allergens in girls’ compared to boys’ mattresses is to our knowledge not previously reported, and the reason for this finding is not known. However, girls may be more likely than boys to eat in their bedroom or have more decorative pillows and stuffed toys that might act as dust and allergen reservoirs [24]. A Swedish study in schools demonstrated that furnishing and textiles in the classrooms acted as reservoirs for allergens and irritants [25]. Allergens have been found to rapidly accumulate in stuffed toys, with an increasing allergen load with increased age of the stuffed toys [26], and regularly cleaning or removal of stuffed toys reduced the load and concentrations of allergens (house dust mite) in children's beds [27]. In the present study, the study technicians noted that girls tended to have more decorative pillows and stuffed toys in their bed as compared to the boys. Although we are not able to quantify the difference, it is a possibility that the stuffed toys may have contributed to the observed differences between boys and girls in the frequency of food allergen detection in their beds. The use of cosmetics with peanut oil may be higher among girls than boys, but because peanut oils usually do not contain significant levels of proteins [28], this is unlikely to be the main source of the peanut allergens in the dust samples.

Little is known about the sources of contamination of environmental samples by food allergens. Interestingly, in one study from Finland, casein was reported to be abundant in house dust samples and to exceed the levels of animal and house dust mite allergens [29]. The main source of casein was believed to be indoor plasters, as high casein levels were also found in sites where milk was not used or allowed (laboratories). In Finland, casein had been used in indoor plasters since the 1960s to improve material properties. It is unknown whether casein is used in indoor construction materials in Norway. However, our study indicates that the presence of food allergens, including casein, was associated with having bedroom and kitchen on the same floor, pointing to the kitchen as the source.

Carpeting is not common in Norwegian homes and thus wet mopping of floors is common. In homes with a high frequency of wet mopping, there was a tendency towards a higher prevalence of milk and peanut allergens, than in bedrooms with less frequent cleaning, although none of the differences reached statistical significance. Frequent cleaning could cause greater turbulence of sedimented dust and allergenic particles. It is known that vacuum cleaning increases airborne dust and allergen concentrations [30], whereas one study found that classrooms that were mainly cleaned by wet mopping had less settled dust but more airborne viable bacteria than classrooms mainly cleaned by dry methods [25]. The major peanut allergen, Ara h 1, was found not to become airborne under various simulated environmental conditions [20]. Another possible source of the food allergens in the bedrooms is cleaning equipment that has been used in the kitchen. We have analyzed dust samples that had been collected from vacuum cleaner bags from Norwegian homes, and milk allergen was detected in 86%, peanut in 43%, egg in 71% and fish allergens in 29% of the 14 samples (unpublished data). The vacuum cleaner bag samples represent the overall exposure in the home [17]. Allergens may leak from vacuum cleaner bags [31] and, thus, vacuuming may be one contributing source of food allergens in the mattress dust.

The findings that co-location of bedroom and kitchen on the same floor or a small-sized dwelling (typically apartments) were the most important factors for the presence of food allergens in mattresses suggests that airborne dust and cooking vapors are important modes of transportation. Alternatively, adolescents with bedroom on the same floor as the kitchen may be more likely to bring food to their bedroom. The higher frequency of milk allergens in mattresses in beds that had been made (covered) during the day may be explained by an increased probability of the allergens to be retained in the mattresses if the bed is covered rather than becoming airborne during disturbances in the room, although this would be likely to also affect the other food allergens.

The present study was not designed to assess food avoidance or possible effects of allergens on disease severity. However, it may be inferred from detecting peanut levels in more than half of the children who reported nut allergy, that avoidance of proving allergens was not consistent. It must, however, be noted that we did not analyze for tree-nuts, and it is not known which nuts (peanuts or tree-nuts) the children reported allergies towards.

Few studies have addressed whether food allergens in dust in the home may cause worsening of disease. One food bronchial challenge study reported exposure to aerosolized allergens from fish, chickpeas and buckwheat to trigger symptoms in asthmatic children with food allergy [32], but direct skin contact to allergens in dust is unlikely to result in severe or systemic reactions [33, 34].

Our study is the first to demonstrate allergens from fish, peanut, egg, and milk to be frequently present in household dust collected from mattresses. Although skin or inhalational exposure may only induce limited reactions in highly sensitive individuals, mattress dust is a previously little explored food allergen reservoir that is important to consider, because children spend so much time in the bedroom.