Of the 10 plagues visited on the biblical Egyptians, the last was the most terrible; after the rain of frogs, the plague of boils, and the hailstorms came the indiscriminate slaughter of all firstborn animals including children. Infanticide of this degree is thankfully rare—but is it possible that the author(s) of Exodus were expressing a subtler truth?

ASSOCIATIONS BETWEEN BIRTH ORDER/SIBSHIP AND DISEASE

Studies of birth order—or sibship size—as a risk factor have a long history and have examined a wide variety of diseases. Thus, for example, the rates of Hodgkin’s lymphoma in young adults,^1,2 HBsAg+ hepatocellular cancer,³ acute lymphoblastic leukaemia,⁴ and type I diabetes mellitus^5,6 all appear to fall with increasing birth order. In each case the pattern has been assumed to reflect the relatively late age at which children of low birth order (or their mothers during pregnancy) acquire common infections. A similar (but opposite) reasoning has been applied to the observations that children of low birth order are at reduced risks of non-Hodgkin’s lymphoma,⁷ schizophrenia,⁸ gastric carcinoma and ulcer,⁹ acute myeloblastic leukaemia,¹⁰ and some congenital heart defects.¹¹ Children of low birth order are also more likely to have infantile pyloric stenosis, to be taller,¹² to be right handed¹³ and, if they are male, to be heterosexual;¹⁴ these are less easily attributed to patterns of early infection.

Nowhere, however, are the patterns of birth order/sibship clearer than with the childhood respiratory allergies. First observed by Butler and Golding in 1986,¹⁵ reductions in the risks of hay fever, eczema, atopy and, less consistently, asthma with increasing birth order or sibling numbers have been reported in at least 30 studies and usefully reviewed by Karmaus and Botezan.¹⁶ As with most of the diseases above, these observations have generally been attributed to different rates and timings of early (unspecified) childhood infection. Indeed, they form the cornerstone of the “hygiene hypothesis” whereby it is proposed that the risks of atopic disease are reduced by early contact with infection,¹⁷ a proposal bolstered by the more recent suggestion that children born to Alpine farmers are protected in a similar manner.

ASSOCIATION BETWEEN SIBSHIP SIZE/BIRTH ORDER AND ALLERGIC DISEASE

In this issue of Thorax, Kinra et al¹⁸ provide an historical perspective on the associations between allergic disease and sibship size/birth order. Their population comprised 14 000 students, predominantly male and about 50% of those eligible, who were screened at Glasgow University between 1948 and 1968. The students had a mean age of 19 years and were born between 1918 and 1952; for the purposes of this analysis, they were divided into three equally spaced birth cohorts. Intriguingly, there was no increase in the prevalence of self-reported allergic disease across the time frame of the three cohorts, although it is difficult to judge how representative this finding might be. The authors found clearly decreasing trends in reported allergic diseases (especially hay fever) with both increasing family size and birth order. Although the differences were small and are readily explained by chance, the effects were strongest in the earliest cohort. If this is explained by greater interfamily differences in infection in that era, then the findings overall are compatible with a “hygiene explanation”. Further support is provided by the finding that allergic diseases were less common among students from less affluent backgrounds, and by a (weak) interaction between birth order (but not family size) and childhood socioeconomic status. The report is valuable because it suggests that the sibship/birth order phenomenon is temporally robust and thus indicative of a relatively era independent mechanism.

HYGIENE HYPOTHESIS

As an explanation of family size effects (especially where these have been related to older siblings) and of other aspects of the epidemiology of childhood allergies (notably its geographical distribution), the hygiene hypothesis is both plausible and parsimonious.¹⁹ There are, however, increasing doubts that it is a complete explanation. First is the very tenacity of the birth order/sibship observation. This, of course, is not in itself a counterargument, but consistency is unusual in the shifting sands of asthma epidemiology. If early contact with infection alone is crucial in the development of childhood allergy, then one might expect the effects of the surrogate of family size to have changed more clearly with shifts in the patterns of early infection. This does not appear to be the case. It is not even clear whether the rates and timings of childhood infections are indeed closely related to family structure in the heavily immunised small family societies where allergic diseases are so prevalent. Even in rural Poland where the prevalence of atopy is extremely low—presumably as a result of extensive microbial contact in early childhood—we have recently observed a clearly increased risk among firstborn children. Attempts to explain sibship effects by direct measurement of early infection have not been successful,^20,21 although this may reflect the poverty of our methods of measuring the subtleties of “infective load”.

Equally as striking as its consistency is the strength of the effect. With the exception of family history and, in certain communities, a farming childhood, no other risk factors have been identified as being so powerful. There is, furthermore, a paucity of supporting analogous evidence. Studies of child care, for example—a surrogate for a large family—have produced inconsistent results,²² and those that have examined the early use of antibiotics have found no consistent evidence that these are harmful.²³ Interestingly, even in classical infectious disease epidemiology the logic that birth order determines age of first infection and thus outcome is not always clear.²⁴

Birth order and family size are inevitably correlated, especially when average family sizes are small. Distinguishing the effects of the two has proved to be very difficult and probably requires the use of large populations. Nonetheless, they may be indicative of different mechanisms. Children at the head of a sibship differ from others in ways that do not necessarily reflect their infectious experience. On the whole they are more often male, are born to younger parents, are of lower birth weight, and are more frequently from “abbreviated” families. Traditionally, children from small families have been from higher socioeconomic groups. Patterns of breast feeding may be related to birth order or family size as, intriguingly, may early exposures to domestic allergens.²⁵ Parents’ reporting of their children’s illnesses is likely to be heavily influenced by their level of previous experience. Studies of sibship effects need to consider each of these potentially confounding variables and to be aware that, on cross sectional analysis (the most common design used in asthma epidemiology), the probability of a child being in any particular birth position is dependent on time related changes in family size.²⁶

OTHER POSSIBLE EXPLANATIONS

If variations in the pattern of early infection are not the whole explanation of the sibship effects in childhood allergies, what other explanations are there? Levels of IgE in cord blood are lower in multiparous women, as is the specific reactivity of cord blood mononuclear cells.^27,28 These findings may reflect higher rates of in utero infection of women who already have other children or, at least in the latter instance, higher maternal age. Alternatively, they may reflect the maternal experience of pregnancy itself. Pregnancy is an immunological challenge to the mother and child and it may be that the complex mechanisms associated with fetal survival vary systematically with increasing parity. These may, in turn, be reflected by shifts in maternal immunity. Women (but not men) who have had several children are less likely to be atopic, an observation that is not explained by age,²⁹ and they are less likely than nulliparous women to report symptoms consistent with allergic rhinitis³⁰ or conjunctivitis.³¹ When studied prospectively, women who have had more pregnancies are more likely to “lose” their atopic state and their hay fever than are women who have had no (further) pregnancies.³² Some have suggested that these apparent shifts in maternal immunity are hormone related.³³ It remains unclear how and whether they are transmitted to successive children, although there is some evidence that firstborn children respond differently to respiratory allergens encountered in the first weeks of life.³⁴

Further attention to non-infectious explanations for the sibship effects in childhood allergy seems warranted. A useful starting point might be the experience of other diseases: for many years it has been known that the risks of pre-eclampsia fall sharply with parity but return to baseline when a woman is pregnant through a new partner. Paternity has not, as far as I am aware, been examined in the development of respiratory allergies. If nothing more, it might shed some light on the muddled area of parental inheritance.

The public health implications of a sibship effect that is not attributable solely to “hygiene” are not easily envisaged. After all, firstborn children cannot be avoided, but it may be that a better understanding of the nature of their risk will allow the more effective targeting of preventive strategies; and it may be time to revisit the discredited argument that changes in family structure have been an important part of the late 20th century epidemic of childhood allergies.