Clinical and Experimental Allergy, 1995, Volume 25, pages 296-303

REVIEW

Epidemiology of hay fever: towards a community

diagnosis
D. p. STRACHAN
Department of Public Health Sciences, St George's Hospital Medical School, London, UK

Introduction contact all children with the same birth dates who were
Symptoms of rhinitis (sneezing, running and blocking of resident in Britain during 1965, 1969, 1974, 1981 and
the nose) occur occasionally in all individuals. However, 1991 (at ages 7, 11, 16, 23 and 33 years). The following
the seasonality of allergic rhinitis induced by grass and question was asked in interviews with participants at age
tree pollens is characteristic and much of our under- 23 years (in 1981): have you suffered from hay fever in
standing of the epidemiology of allergic rhinitis relies the last 12 months?
upon self-reported or doctor-diagnosed 'hay fever'. A range of potential explanatory variables were col-
Much less is known about the prevalence or distribution lated from the information collected by midwives at birth
of perennial allergic rhinitis and non-allergic rhinitis [1]. and by interviewers at each of the four follow-ups.
Since exposure to airborne pollen is almost universal in Paternal occupation at the birth of the index child was
countries such as Britain, the distribution and determi- classified according to the Registrar-General's social
nants of hay fever (other than its seasonal pattern) may classes used in 1951 [10]. The child's own current or
provide an insight into the epidemiology of atopy in last occupation at age 23 years was classified by the 1981
general. social classes [11]. The classes used were I (professional
and managerial), II (intermediate), IIIN (skilled non-
This paper explores the association of geographical, manual), HIM (skilled manual), IV (semi-skilled) and V
social and family factors with hay fever as reported by a (unskilled). The numbers of older and younger children
national sample of young British adults who had been in the household were determined at the 1969 follow-up,
followed from birth. The findings are presented in when most families could be assumed to be complete.
greater detail than in earlier reports [2-4] and are Older children of the family who were currently living
discussed in the context of historical and geographical away from home, but aged under 21 years, were included
variations in hay fever prevalence. These epidemiological in the total.
observations ('community presentation') will be used to
construct a working hypothesis ('community diagnosis') Areas of residence at birth and at age 23 years were
to explain why hay fever apparently emerged as a post each classified by the region and degree of urbanization.
industrial revolution epidemic [5] and continues to The local authority areas of birth were classified as
increase in prevalence in modern Britain [6,7]. The conurbations, county boroughs (outside conurbations)
implications for future trends in prevalence ('community or administrative counties (predominantly urban or
prognosis') will be discussed. rural as indicated by their population density at the
1961 census). The counties of residence in 1981 were
classified as conurbations (Greater London and metro-
Methods politan counties) or non-metropolitan counties.
All births in England, Scotland and Wales during 1 week Additional factors which were considered in the
in March 1958 formed the target sample for a perinatal analysis were self-reported cigarette smoking (at age 23
morbidity and mortality survey [8]. The National Child years); parental reports of housing tenure (at age 7
Development Study [9] has attempted to trace and years), shared use of kitchen, bath or toilet with another
household (at age 7 years), housing density (persons per
room at age 11 years), and mode of infant feeding
(reported at the 7-year follow-up); and the following
Correspondence: Dr D. P. Strachan, Department of Public Health features of the birth record: sex, birthweight, maternal
Sciences, St George's Hospital Medical School, Cranmer Terrace,
London SW17 ORE, UK. age, parity and smoking during pregnancy.
Based on an invited contribution to a plenary session at the meeting of
Statistical analysis was performed using SAS [12].
the British Society of Allergy and Clinical Immunology, City Multiple logistic regression models were fitted by the
University, London, UK, 7 September 1994. method of maximum likelihood using the PLR program
296
 Epidemiology of hay fever 297

Table 1. Prevalence (%) of hay fever in the past year at 23, by area of residence at 23

Region Conurbations Other areas Whole region

South Eastern 211 (326/1541) 18-6 (309/1671) 19-8(635/3212)

East Anglia 19 0 (214/1129) 19-0 (214/1129)
South Western 17-5 (143/819) 17-5 (143/819)
Wales 14-7 (96/654) 14-7 (96/654)
West Midlands 13-9 (88/632) 17 9 (100/560) 15-8 (188/1192)
East Midlands 15-7 (148/940) 15-7 (148/940)
Yorkshire and Humberside 13 6 (109/801) 13-6 (24/177) 13-6 (133/978)
North Western 14-9 (141/945) 15-9 (73/460) 15-2 (214/1405)
Northern 10 3 (32/311) 12 6 (64/507) 11-7 (96/818)
Scotland 11-5 (63/549) 161 (106/659) 14-0 (169/1208)
All regions 15-9 (759/4779) 16-9 (1277/7576) 16-5 (2036/12355)

in BMDP [13]. These analyses were restricted to 9356 3009); urban counties (15-6%, 476/3059); county
(76%) of the 12 355 respondents at age 23 years, who boroughs (16-1%, 392/2429); and conurbations other
had full information on all variables included in the than London (14-7%, 272/1847); but was substantially
models. higher among those bom in Greater London (22-5%, 358/
1589). The London excess was mainly attributable to the
northwest to south-east gradient regional gradient by area
Results
of birth [3].
Of the responders at age 23 years, 2919 (25%) had
Geographical variation moved from their region of birth. When adjusted by
Among 12355 cohort members interviewed at age 23, multiple logistic regression for region of birth and region
2036 (16-5%) reported hay fever in the past year. The of residence at 23, migrants as a group had a high
prevalence varied substantially across the regions of prevalence of hay fever at 23 (odds ratio 1-22, 95%
Britain, being higher in the south and lower in the north confidence interval 1-10-1-35, P < 0-001). After
(Table 1). Comparisons within regions between the major adjustment for this general migrant effect, and the risk
conurbations and surrounding non-metropolitan counties associated with the region of birth, the risk associated with
did not reveal any consistent urban-rural variation (Table the region of residence at 23 had a small independent effect
1). When analysed by area of birth, the prevalence differed which was of borderline statistical significance
little among those born in rural counties (16-2%, 486/ (0-05 < i ' < 0-10). In contrast, the influence of region of

Table 2. Prevalence (%) of hay fever in the past year at 23, by father's social class and number of children in the household

Number of other children in the household at age 11 '•

Father's social class at birth 0 1 2 3+ Total

I/II 26-1 23-1 18-6 17-4 20-8

(46/176) (175/759) (100/537) (73/419) (394/1891)
III 19-0 19-6 17-3 11-0 16-4
(125/657) (420/2148) (274/1585) (201/1826) (1020/6216)
IV/V 21-0 14-7 15-3 9-7 13-2
(38/181) (83/563) (79/517) (96/998) (296/2249)
Total 20-6 19-5 17-2 11-4 16-5
(209/1014) (678/3470) (453/2639) (370/3233) (1710/10356)

'Includes members of the family under 21 living away from home in 1969.
298 D. P. Strachan

Table 3. Prevalence (%) of hay fever in the past year at 23, by numbers of older and younger children in the household at age 11

Number of younger children in household at 11

Number of older children
in household at 11~ 0 1 2 3-1- Total

0 20-6 21-8 20-4 162 204

(216/1049) (378/1730) (203/993) (113/698) (910/4470)
1 17-3 159 13-7 10-4 15-7
(307/1776) (169/1063) (72/526) (35/338) (583/3703)
2 141 12-8 69 5-5 116
(89/630) (59/462) (14/204) (10/182) (172/1478)
3-1- 107 69 80 81 85
(31/291) (20/289) (14/175) (14/173) (79/928)
Total 17-2 17-7 160 12-4 165
(643/3746) (626/3544) (303/1898) (172/1391) (1744/10579)

'Includes members ofthe family under 21 living away from home in 1969.

birth was strong and independent of both region at 23 and Although tests of heterogeneity are presented in Table
the general migrant effect (P < 0001). 4, much of the variation in hay fever by paternal social
class and numbers of older and younger children was
attributable to linear trends in risk, each of which were
Family circumstances
highly significant (P < 0 001). The gradients by numbers
Table 2 shows the relationship between hay fever at age of older and younger children in the household were
23, household size at age 11 and paternal social class at significantly different from each other (P < 0 001). The
birth. The prevalence of hay fever declined with increas- effect of maternal age was of lesser significance
ing family size and was lower among the offspring of {P < 0 05).
unskilled and semi-skilled manual workers (IV & V) than
those of skilled workers (III) or professional, managerial
and intermediate groups (I & II). These effects were Factors not associated with hay fever at 23 years
independent of each other, with a 25-fold variation in Other factors were assessed by adding them individually
prevalence across the cells of the Table. to the core model in Table 4. The following variables did
Table 3 shows the joint effects of older and younger not make a significant independent contribution to the
children upon the prevalence of hay fever at age 23 risk of hay fever at 23: male sex (adjusted odds ratio 1 04,
among the index children. Both exerted an independent 95% confidence interval (0-93-1-17), birthweight less
effect, but the number of older children appeared to be than 2500g (1-02, 0-76-1-36), maternal smoking during
the more influential variable. pregnancy (0-91, 0-79-1-03), birth in a conurbation or
county borough (1-02, 0-91-1-15), household crowding
(more than one person per room) at age 11 (101, 0-73-
Independent risk factors 1-38).
Table 4 shows the variables which made a significant Maternal parity and number of older children in the
{P < 0 05) independent contribution to the risk of hay household at age 11 were highly correlated and their
fever. This 'core' model was based upon 9356 subjects independent effects could not be estimated with preci-
with complete data, 1548 (16 5%) of whom had hay sion. However, the number of older children in the
fever. Holding other factors constant, hay fever at 23 was household appeared to be the more influential variable.
more common among breastfed children of younger,
professional parents, born in the southeast, brought up
in small households living in owner-occupied housing Discussion
with sole use of kitchen, bath and toilet. Regular cigar-
ette smokers at 23 reported less hay fever than non- The rising prevalence of hay fever
smokers; subsidiary analyses showed a similar prevalence Anecdotal evidence suggests that hay fever was extremely
in never smokers and ex-smokers at age 23. rare in Britain before the industrial revolution, and
 Epidemiology of hay fever 299

Table 4. Independent risk factors for hay fever at 23

Number of Adjusted~ 95%

subjects odds ratio C.I. df

Regional SE 3082 100 (Reference)

group at SW 2530 078 067-089 44-3 2
birth # N 3744 0-64 056-073
Father's I 414 100 (Reference)
social II 1288 083 0-63-108
class IIIN 965 080 0-60-106
at HIM 4659 0-72 0-56-0-92 17-5 6
birth of IV 1179 058 0-43-0-78
child V 809 0-74 0-54-1-02
Unclassified 42 041 0-14-1-20
Older 0 3942 1 00 (Reference)
children 1 3323 068 0-60-0-77
(under 21) 2 1301 055 0-46-0-66 83-2 4
in household 3 520 0-45 0-32-0-62
at 11 4+ 270 0-36 0-22-0-58
Younger 0 3354 100 (Reference)
children 1 3151 093 0-82-1-07
in 2 1678 085 0-72-1-01 15-4 4
household 3 714 071 0-55-0-90
at 11 4+ 459 064 0-46-0-88
Age of 30+ 3190 100 (Reference)
mother at 25-29 3168 098 (0-85-1-14) 5-34 3
birth of 20-24 2594 085 (0-72-1-01)
index child <20 404 070 (0-50-0-98)
Breastfed (any duration) 6466 M4 1-01-1-29 4-28 1
Rented housing at 7 5313 086 0-76-0-97 6-25 1
Shared amenities* at 7 1595 079 0-68-0-93 8-13 1
Cigarette smoker at 23 3774 071 0-63-0-80 31-6 1
Total included in model 9356

~ Mutually adjusted by multiple logistic regression.

* Use of kitchen, bath or WC shared with another household. SE. South Eastern, East Anglia, East Midlands. SW. South Western,
Wales, West Midlands. N. Yorkshire & Humberside, North Western, Northern, Scotland.

emerged as a complaint of the urban educated classes in and levels of allergen-specific IgE among Japanese
the 19th century [5]. A large population survey in school children [21]. The exception to this general rule
Switzerland in 1926 found only 0-28% of 77000 respon- seems to be in Australia, where there was no appreciable
dents to be suffering from hay fever [14]. At that time, the change in the prevalence of self-reported hay fever from
prevalence in urban areas was 10 times higher than in 1977-1978 to 1989-1990 [22] nor in the degree of skin-
rural areas (1-2% vs 0-13%). A similar survey carried out prick positivity among adults over a similar period [23].
in 1985 found a much higher prevalence (10-0%) and
little evidence of urban-rural variation (8-0% vs 10-5%).
Geographical variations
The substantial increase in prevalence in Switzerland
parallels that described in serial surveys of American Estimates of the prevalence of hay fever in different
college students [15], Swedish army recruits [16] and countries vary widely, from 0-5% to 28% for children
British teenagers [6,7], which suggest a long-standing and from 0-5% to 15% for adults [24]. Much of this
and continuing rise in hay fever prevalence. There is variation may be attributable to the methods of sample
also evidence of a rise in the prevalence of eczema in selection and case definition, but there is also evidence of
Britain [6,7,17] and Denmark [18]; skin-prick reactions real differences, notably between former east and west
to common aeroallergens in Britain [19] and USA [20]; Germany [25]. Regional differences are also apparent
300 D. P. Strachan

within Australia [22], perhaps related to a higher aero- groups was evaluated in a study of adults in south
allergen burden in drier inland districts. London [32] which found little difference in the preva-
The marked geographical variation in the prevalence lence of symptoms or skin-prick reactions by social class,
of hay fever in the British 1958 cohort appears to relate but a greater use of the hay fever label by doctors among
more closely to region of birth than region of residence at patients from more privileged social classes.
23. This may reflect the distribution of the atopic The socioeconomic gradient in self-reported hay fever
genotype, or variations in lifestyle or environmental in the British 1958 cohort parallels that seen for eczema
exposures (including aeroallergens) in early life. A as reported by parents and examined by school medical
study of the birthplace of the parents of children with officers at ages 7, 11 and 16 [4], and for skin-prick
and without hay fever would help to distinguish between positivity to common aeroallergens measured at age
these alternative explanations. 34-35 [Strachan DP, unpublished observations]. This
Studies of urban-rural variation within countries are suggests that it is unlikely to be a reporting artifact and
inconsistent. In Tecumseh, USA, probable allergic probably reflects an association of aflluence with the
rhinitis was found three times more commonly in city underlying atopic tendency. Socioeconomic variations
residents than in the surrounding rural areas (75% vs in sensitization to common aeroallergens in two large
25%) [26], whereas in Switzerland no urban-rural American surveys [33,34] are consistent with this
variation was evident in recent surveys [14]. The 1958 interpretation.
cohort were born at a time when all cities and most major
towns in Britain experienced winter pollution by coal
smoke and sulphur dioxide at levels comparable to those Family structure
in east German cities in the mid-1980s [27]. The absence The strongest association which emerged in this analysis
of substantial urban-rural variation in hay fever pre- was between hay fever and the number of older children
valence in this large cohort and the low prevalence of hay in the household. This appeared to be largely independent
fever and allergic sensitization in previously highly of maternal age and parity, suggesting that postnatal
polluted east German cities [25] are evidence against influences were more important than the intrauterine
the hypothesis that irritant air pollution resulting from environment. This would be consistent with the weaker,
industrialization was a major influence in the emergence but independent, trend by numbers of younger children
of hay fever as a 'post industrial revolution epidemic' in in the household.
Britain and elsewhere [28]. A similar pattern of decreasing prevalence of hay fever
There has been concern that the continuing rise in with increasing numbers of older siblings was found in a
prevalence of allergic disease in recent decades may have later British birth cohort born in 1970. When studied at
been due to increased exposure to vehicle-related pollu- the age of 5 years, 6-0% of the flrstborn children suffered
tants such as nitrogen dioxide, ozone and diesel particu- from hay fever compared with 1-3% of children with
lates [29]. Two studies have suggested a higher four or more older siblings [35]. In a recent survey of over
prevalence of nasal allergy in areas with high traflic 18000 teenage school children in SheflSeld, the preva-
density [30,31]. In modern Britain, ozone levels are lence declined from 18-4% among flrstborns to 11-8%
generally higher in urban areas, whereas nitrogen among children with three or more older siblings
oxides and diesel particulates are urban pollutants. The [Strachan DP, unpublished observations]. Total family
lack of urban-rural variation in the prevalence of hay size is inversely related to the prevalence of skin-
fever may thus conceal subtle but counterbalancing prick positivity among German children [36], but this
influences of different air pollutants. Nevertheless, it is study did not distinguish between older and younger
clear that exposure to outdoor air pollution cannot siblings.
account for the major epidemiological features of hay Within the 1958 cohort, the influence of older siblings
fever which emerge from comparisons within popula- is apparent for hay fever or allergic rhinitis ascertained
tions, by socioeconomic status and family structure. by parental interview at age 11, and for eczematous
rashes occurring in the first year of life [2]. Skin-prick
Socioeconomic variations tests performed on a sub-sample of the cohort at age
34-35 show a strong, graded and statistically significant
Hay fever has been recognized as a complaint of the inverse association between number of older siblings and
wealthier classes since the 19th century [5], and several the prevalence of positive reactions to each of three
studies suggest that this pattern persists in developed common aeroallergens (mixed grass pollen, house
countries [22,26]. The possibility of differential reporting dust mite and cat fur) [37]. These observations
and labelling of nasal symptoms in different social effectively exclude reporting artifacts and suggest
 Epidemiology of hay fever 301

that the relationship between hay fever and house- of its risk factors are shared by infant eczema and skin-
hold size reflects the distribution of atopy in general, prick positivity to common aeroallergens, suggesting
rather than organ-specific effects on the nasal that epidemiological variations in each of these three
mucosa. outcomes reflect the underlying distribution of atopy.
The strongest risk factor to emerge is position in the
household as a child. The apparent protective influence
Infant feeding of older siblings emerges within a few months of birth
There has been considerable interest in the possibility (infant eczema) and persists at least to age 35 (skin-prick
that infant feeding may modulate the development of responses). This may be seen as an example of biological
allergy, in particular that breastfeeding may have a 'programming' [41].
protective effect [38]. This debate emerged some years The precise nature of the programming influence
after the 1958 cohort were born, so it is unlikely that their remains uncertain. A working hypothesis is that allergic
mothers had selectively altered their feeding practice sensitization can be prevented by infections acquired
because of a family history of allergy. As in a later during early childhood [2]. Evidence is emerging from
national cohort studied at 5 years of age [35], breastfeed- in vitro studies that bacterial or viral infections may
ing was actually associated (albeit weakly) with a higher stimulate the development of TH-1 lymphocytes and
prevalence of hay fever. However, these findings from thus prevent the proliferation of TH-2 cell populations
general population samples do not exclude a possible [42]. Early childhood infections are commonly trans-
protective influence of prolonged exclusive breastfeeding mitted by household contact, particularly with older
in a small subgroup of infants at particularly high risk of siblings attending schools and playgroups, but later
allergic diseases. infection or reinfection from younger siblings might
have some additional protective effect. The low
prevalence of hay fever among the offspring of poorer
Smoking families living in rented housing and sharing household
Cigarette smoking has been implicated as a cause of amenities may reflect unhygienic lifestyles or household
increased serum IgE levels but is associated with a lower conditions which promote the spread of infection. Wide-
prevalence of cutaneous reactivity to common aeroaller- spread use of day nurseries from an early age will
gens. This discrepancy has been attributed to a lower promote cross-infection and may explain the unusually
uptake of smoking and a higher rate of smoking cessa- low prevalence of allergic sensitization in former east
tion among allergic individuals [39]. In the 1958 cohort, Germany [36]. The slightly increased risk of subsequent
smoking was less common among respondents with hay hay fever and eczema [17] among breastfed infants may
fever at 23 only if hay fever or allergic rhinitis was reflect a reduced incidence of gastrointestinal infection
reported at age 11 or 16 [Strachan DP, unpublished during breastfeeding, or more hygienic child rearing
analyses]. This suggests that adolescents and young practice in a broader sense.
adults who suffer from hay fever may indeed be less
likely to take up regular cigarette smoking. However, the
similar rates of hay fever among lifelong non-smokers Community prognosis
and former smokers suggest that affected individuals are The hypothesis that early infection may protect against
not more likely to give up smoking once they have atopic disease offers a coherent explanation for both the
started. past and present epidemiological patterns of hay fever.
Higher levels of cord blood IgE in the have been Over the past 200 years, reductions in average family
reported in the offspring of mothers who smoked size, improved household amenities and higher standards
during pregnancy [40], but skin-prick positivity was less of personal cleanliness have reduced the opportunity for
common among German children whose mothers cross-infection within young families. This may have
smoked [36]. There is no evidence in the British birth resulted in more widespread expression of atopy and
cohorts that maternal smoking during pregnancy or the emergence of a 'post industrial revolution epidemic'
childhood is an important determinant of symptomatic of allergic disease, affecting to a greater extent the more
nasal allergy. affluent classes and areas of the country.
If the prevalence of hay fever is increasing as a result of
changes in biological programming in early childhood,
Community diagnosis
then the generations who have been found to be at high
Hay fever displays a remarkable degree of variation risk of allergic disease while young may carry that
within and between families in modern Britain. Most increased susceptibility for much of their lives. Evidence
302 D. P. Strachan

of such a cohort (or generation) effect can be seen in References

observations from Tucson, Arizona, where skin-prick
tests to common aeroallergens were performed on the 1 Sibbald B, Strachan D. Epidemiology of rhinitis. In: Busse
same population on two occasions, 8 years apart [20]. At WW, Holgate ST, eds. Asthma and rhinitis. Oxford: Black-
well, 1994;32-43.
each survey, positive skin tests were most common
2 Strachan DP. Hay fever, hygiene and household size. Br
among teenagers and young adults, thereafter declining
Med J 1989; 299:1259-60
in prevalence with increasing age. However, the preva- 3 Strachan DP, Golding J, Anderson HR. Regional varia-
lence of skin-test positivity was higher on the second tions in wheezing illness in British children: the effect of
occasion in all age groups, so that in each generation, the migration during early childhood. J Epidemiol Community
prevalence of this measure of atopy increased with Health 1990; 44:231-6.
advancing age. It is therefore likely that the reduced 4 Williams HC, Strachan DP, Hay RJ. Childhood eczema:
prevalence of skin-test positivity among the more elderly disease of the advantaged? Br Med J 1994; 308:1132-5.
subjects reflected a lifelong reduction in allergic respon- 5 Emanuel MB. Hay fever, a post industrial revolution
siveness, rather than a biological effect of ageing. epidemic: a history of its growth during the 19th century.
It is likely that over the next few decades hay fever Clin Allergy 1988; 18:295-304.
6 Burr ML, Butland BK, King S, Vaughan-Williams E.
prevalence in older age groups will continue to rise, as
changes in asthma prevalence: two surveys 15 years apart.
more susceptible cohorts replace the generations appar- Arch Dis Child 1989; 64:1118-25.
ently protected by a less affluent upbringing. Future 7 Ninan TK, Russell G. Respiratory symptoms and atopy in
trends among younger age groups are more difficult to Aberdeen school children: evidence from two surveys 25
predict, particularly as living standards and patterns of years apart. Br Med J 1992; 304:873-5.
child care continue to change. With greater use of pre- 8 Butler NR, Bonham DG. Perinatal mortality. First report
school creche and nursery facilities in developed coun- of the 1958 British perinatal mortality survey, under the
tries, it is possible that the risk of allergic sensitization auspices of the National Birthday Trust Fund. Edinburgh:
may decline. As living standards in eastern and western Churchill Livingstone, 1963.
Europe converge, regional variations in the prevalence 9 Shepherd P. The National Child Development Study; an
of hay fever may become less marked among future introduction to the origins of the study and the methods of
data collection. London: NCDS User Support Group, City
generations.
University, 1986.
10 General Register Office. Census of England and Wales
Conclusion 1961. General Report. London: HMSO, 1965.
11 Office of Population Censuses and Surveys. Classification of
Epidemiological variations in reports of hay fever by occupations and coding index. London: HMSO, 1980.
young adults appear to be of broader relevance, point- 12 SAS Institute Inc. SAS User's Guide: Basics, version 5
ing to the underlying distribution of atopy. They sug- edition and Statistics, version 5 edition. Cary, NC: SAS
gest that expression of the atopic phenotype may be Institute Inc., 1985.
programmed by influences operating very early in- 13 Engelman L. PLR: Stepwise logistic regression. In: Dixon
childhood, perhaps before birth. The nature of the WJ ed. BMDP Statistical Software. 1985 printing. Berkeley:
programming influence and the critical period of devel- University of California Press, 1985:330-44.
opment during which it operates can only be deter- 14 Wiithrich B. Epidemiology of the allergic diseases: are they
really on the increase? Int Arch Allergy Appl Immunol
mined by further clinical and epidemiological studies.
1989; 90:3-10.
However, if the working hypothesis of a protective
15 Hagy GW, Settipane GA. Bronchial asthma, allergic rhini-
effect of early infection proves to be correct, the remote tis and allergy skin tests among college students. J Allergy
possibility of preventing allergic diseases by artificial 1969; 44:323-32.
inoculation of an infective agent in early infancy 16 Aberg N. Asthma and allergic rhinitis in Swedish con-
should not be entirely discounted. scripts. Clin Exp Allergy 1989; 19:59-63.
17 Taylor B, Wadsworth M, Wadsworth J, Peckham C.
Changes in the reported prevalence of childhood eczema
Acknowledgements since the 1939-45 war. Lancet 1984; ii: 1255-7.
I wish to thank the Economic and Social Research 18 Schultz-Larsen F, Holm NV, Henningsen K. Atopic derma-
Council Data Archive, University of Essex and the titis. A genetic-epidemiological study in a population-based
twin sample. J Am Acad Dermatol 1986; 15:487-94.
National Child Development Study User Support
19 Sibbald B, Rink E, D'Souza M. Is atopy increasing? Br J
Group, City University, London, UK for their assistance
Gen Pract 1990; 40:338-40.
in accessing the National Child Development Study 20 Barbee RA, Kaltenborn W, Lebowitz MD, Burrows B.
dataset. Longitudinal changes ih~ allergen skin test reactivity in a
 Epidemiology of hay fever 303

community population sample. J Allergy Clin Immunol traffic density on street of residence. Ann Epidemiol 1994;
1987; 79:16-24. 4:79-83.
21 Nakagomi T, Itaya H, Tominaga T, Yamaki M, Hisamatsu 32 Sibbald B, Rink E. Labelling of rhinitis and hayfever by
S, Nakagomi N. Is atopy increasing? Lancet 1994; 343:121- doctors. Thorax 1991; 46:378-81.
2. 33 Gergen PJ, Turkeltaub PC, Kovar MG. The prevalence of
22 Australian Bureau of Statistics. 1989-90 National Health allergic skin test reactivity to eight common aeroallergens in
Survey. Asthma and other respiratory conditions, Australia. the US population: results from the second national health
Catalogue no. 4373.0. Canberra: Commonwealth of and nutrition examination survey. J Allergy Clin Immunol
Australia, 1991. 1987; 80:669-79.
23 Peat JK, Haby M, Spijker J, Berry G, Woolcock AJ. 34 Barbee RA, Lebowitz MD, Thompson HC, Burrows B.
Prevalence of asthma in adults in Busselton, Western Immediate skin test reactivity in a general population
Australia. Br Med J 1992; 305:1326-9. sample. Ann Intern Med 1976; 84:129-33.
24 Montgomery Smith J. Epidemiology and natural history of 35 Golding J, Peters T. Eczema and hay fever. In: Butler N,
asthma, allergic rhinitis and atopic dermatitis (eczema). In: Golding J eds. From birth to five. A study of the health and
Middleton E, Reed CE, Ellis EF, Adkinson NF, Yunginger behaviour of Britain's five-year-olds. Chapter 12. Oxford:
JE, eds. Allergy: principles and practice. St Louis: CV Pergamon Press, 1986: 171-86.
Mosby, 1983:771-803. 36 von Mutius E, Martinez FD, Fritzsch C, Nicolai T, Reitmar
25 von Mutius E, Fritsch C, Weiland SK, Roll G, Magnussen P, Thiemann HH. Skin test reactivity and number of
H. Prevalence of asthma and allergic disorders among siblings. Br Med J 1994; 308:692-5.
children in the united Germany: a descriptive comparison. 37 Strachan DP, Griffiths JM, Anderson HR, Johnston IDA.
BrMedJ 1992; 305:1395-9. Allergic sensitisation and position in the sibship: a national
26 Broder I, Higgins MW, Matthews KP, Keller JB. study of young British adults. Thorax 1994; 49:1053P.
Epidemiology of asthma and allergic rhinitis in a total 38 Saarinen UM, Kajosaari M, Backman A, Siimes MA.
population, Tecumseh, Michigan. III. Second survey of Prolonged breastfeeding as prophylaxis for atopic disease.
the community. J Allergy Clin Immunol 1974; 53:127-38. Lancet 1979; ii: 163-6.
27 Warren Spring Laboratory. The national survey of air 39 Burrows B, Halonen M, Barbee RA, Lebowitz M. The
pollution 1961-71. Volume 1. London: HMSO, 1972. relationship of serum immunoglobulin E to cigarette
28 Finn R. John Bostock, hay fever and the mechanism of smoking. Am Rev Respir Dis 1981; 124:523-5.
allergy. Lancet 1992; 340:1453-5. 40 Magnusson CGM. Maternal smoking influences cord
29 Wardlaw AJ. The role of air pollution in asthma. Clin Exp serum IgE and IgD levels and increases the risk for
Allergy 1993; 23:81-96. subsequent infant allergy. J Allergy Clin Immunol 1986;
30 Ishizaki T, Koizumi K, Ikemori R, Ishiyama Y, Kushibiki 78:898-904.
E. Studies of prevalence of Japanese cedar poUonosis 41 Strachan DP. Is allergic disease programmed in early life?
among the residents in a densely cultivated area. Ann Clin Exp Allergy 1994; 24:603-5.
Allergy 1987; 58:265-70. 42 Romagnani S. Human THl and TH2 subsets: regulation of
31 Weiland SK, Mundt KA, Riickmann A, Keil U. Self- differentiation and role in protection and immunopathol-
reported wheezing and allergic rhinitis in children and ogy. Int Arch Allergy Immunol 1992; 98:279-85.