In this follow-up at 32–34 years, we were able to link almost all individuals in the original birth cohort to the Swedish Medical Birth Register. Moreover, 72% also responded to a questionnaire on current symptoms. Elevated CB-IgE and a family history of asthma in infancy were associated with a two- to threefold increased risk of dispensed asthma medication and allergy-related respiratory symptoms. However, CB-IgE or reported heredity in infancy was not a good predictor of medication or symptoms in adulthood. Only 21% of all individuals with elevated CB-IgE and 15% with a family history used asthma medication after three decades. Most of the individuals with dispensed anti-inflammatory asthma medication at 32–34 years of age had not been reported having asthma at previous check-ups of the cohort during childhood. To our knowledge, no other birth cohort has been able to elucidate the association between family history of asthma, CB-IgE and asthma disease in adults aged more than 30 years.
Our findings in this adult population are consistent with the results of similar investigations in paediatric populations 
. Only two studies have investigated the association between CB-IgE, and asthma and allergic disease in adulthood and not beyond the age of 20 years, but their results were contradictory and based on smaller birth cohorts than ours 
. Pesonen et al. demonstrated that CB-IgE above 0.5 kU/l was associated with a two- to threefold increased risk of allergic symptoms 
, whereas Shah et al. failed to find any association between continuous levels of CB-IgE and different clinical outcomes 
Fixed versus continuous cut-off levels of CB-IgE could affect the outcome as could different cut-off levels. In our study the detection level of 0.9 kU/l was given by the calibration of the original laboratory test. As we no longer have access to the blood samples, we were unable to investigate the outcome of continuous levels or a lower cut-off more similar to the one used in several recent studies 
. A cut-off at 0.9 kU/l means that our study has higher specificity but lower sensitivity than studies based on a cut-off at 0.5 kU/l.
High CB-IgE values could be the result of maternofetal contamination. A Danish cohort study of 200 children with maternal asthma suggested that maternal IgE antibodies contributed to high CB-IgE values in about 50% of all individuals with CB-IgE values over 0.5 kU/l. The suspected contamination was based on the detection of IgA and allergen-induced specific IgE in the sampled cord blood 
. In contrary, other studies have suggested that maternofetal contamination of cord blood is uncommon. IgA antibodies were detected in less than 5%, even in mother-child pairs with high IgE levels against the same allergen 
. Low concordance between specific IgE in cord blood and maternal blood for inhalant allergens was demonstrated as an indicator of no significant maternofetal contamination. The agreement between specific IgE for food allergens in cord blood and maternal blood was much better, which the authors interpreted as an indication of sensitization already in utero, especially against food allergens. The high CB-IgE values were considered to be chiefly the result of synthesization by the fetus and not due to maternofetal contamination 
It is not easy to elucidate why these different studies have shown such opposing results. However, they differ in several parameters, such as laboratory methods for detecting the different immunoglobulins, the method of securing adequate cord blood samples from the neonates, and the characteristics of the cohorts, just to mention some of the differences between the studies that might have influenced the different results.
As we did not have access to the original blood samples anymore, as mentioned above, we were unable to control for the proportion of IgA in cord blood, which is more or less standard procedure nowadays when investigating the role of CB-IgE. To account for the risk of maternofetal contamination in our study, we excluded subjects with rather high CB-IgE levels (≥10.0 kU/l) as suggested in a previous study 
. Furthermore, good agreement between high CB-IgE levels and high total IgE levels at the age of 18–24 months was reported in our birth cohort 
indicating that the majority of high IgE levels in cord blood are of fetal origin. This is in accordance with the findings in the previously mentioned Danish study that found maternofetal contamination in up to 50% of the samples. Bønnelykke et al. found that children with elevated CB-IgE levels and indications of maternofetal contamination had significantly lower IgE levels at 6 months of age compared with children with elevated CB-IgE levels without indications of maternofetal contamination. Furthermore, elevated CB-IgE levels were not significantly associated with IgE levels at 6 months of age in samples with indications of maternofetal transfer of IgE after adjustment for maternal IgE levels, which was in contrast to a highly significant independent association for samples in which transfer was not indicated 
Dispensed asthma medication was less common than reported allergy-related respiratory symptoms, and the two outcomes only partly overlapped. The prevalence rates of respiratory symptoms were 10–15%, and only 20–30% of those with reported symptoms had purchased asthma medication. Our definition of allergy-related respiratory symptoms included wheezing, breathlessness or coughing. Some individuals might have reported non-asthmatic symptoms from the airways, and others had mild symptoms with no need of medication. In contrast, allergy-related respiratory symptoms were reported by 50–70% of those who had purchased any asthma medication
. This is in line with reviews indicating that only approximately 50% of all asthma is associated with allergy 
. The prevalence of any asthma medication
use was fairly similar to the prevalence rates of physician-diagnosed asthma in recent studies among young adults in Sweden 
Access to data from the Swedish Medical Birth Register in this study gave us the opportunity to control for numerous potential confounding factors that directly or indirectly could have influenced the outcome. Adjusting for these factors, however, altered the odds ratios only slightly. Elevated CB-IgE and the family history based on information collected in early life were highly significant and independent risk factors for asthma medication and respiratory symptoms in adulthood. The odds ratios for both outcomes were fairly similar even though they were markers of partly different entities. We also investigated whether elevated CB-IgE and a family history of asthma based on information collected in early life could be used for screening. However, and in line with previous studies in younger adults 
, neither elevated CB-IgE nor family history were efficient enough to identify individuals with asthma in adulthood. That did not change either when focusing on the cases with asthma medication at 32–34 years of age who had not been reported to have asthma in childhood, probably as the majority of adult cases with dispensed asthma medication are new cases (). We noticed also a reduced proportion of subjects with elevated CB-IgE among those with dispensed asthma medication as adults compared to the subjects with an asthma diagnosis in childhood, which explains the poor predictive values of these parameters. The predictive ability did not improve even when family history and CB-IgE were combined. Sensitivity, however, did increase to a certain extent when elevated CB-IgE or
a positive family history of asthma were used as a combined information with a sensitivity of 40 for anti-inflammatory medication and only a marginally reduced specificity of 83. Still, the positive predictive value did not increase. In our population-based birth cohort with a prevalence of asthma around 12.0% our investigated parameters do not predict the value of interest in a sufficient way, but might be useful in subgroups with higher asthma prevalence as e.g. high-risk infants.
Further limitations of our study need to be commented on. We have used questionnaire-reported symptoms induced by furred pets or pollen as markers of allergic symptoms from the airways. Other potential allergens as for example molds and house dust mites were not included. House dust mites, however, are no common cause of asthma in this part of Europe. The questionnaire lacked specific questions about physician-diagnosed asthma, which is a weakness we cannot fully compensate for, and which might have contributed to non-overlap between allergen-induced asthma symptoms recorded in the questionnaires and asthma medication data from the drug register. Prescription data probably reflect an asthma diagnosis better than self-reported respiratory symptoms, which more often include other respiratory symptoms than asthma. Furthermore, the design of the follow-up did not include more objective measurements of sensitization such as skin prick tests or specific IgE values. Sensitization to common aeroallergens is a known risk factor for the development of asthma 
. However, atopy is not present in all cases of asthma and the problems of misclassification are obvious 
As misclassification of our outcome variable may reduce the estimated effect, we also investigated the dispensation of anti-asthmatic medication as a proxy for asthma disease similar to other register-based studies investigating asthma in adolescents and young adults 
. Even this outcome variable has certain limitations, mainly related to potential over- or undertreatment of respiratory symptoms 
. However, this seems to be a minor problem in the present study as rather good agreement has been demonstrated between register-based data on dispensed prescriptions of anti-asthmatic drugs and physician-diagnosed asthmatic disease 
. Further, reported anti-asthmatic drug use has been found to be an accurate indicator of both actual use of medication and asthma disease 
. In Sweden, a prescription is valid for one year, with one filling usually covering a 3-month consumption period. Occasionally, however, a physician may prescribe inhaled steroids to be dispensed once only to test non-specific respiratory symptoms, although this is rather unusual in adult patients, who are also unlikely to purchase a drug more than once if it failed to help the first time. We have used at least two purchases of prescribed asthma medication as a proxy for asthma to avoid even this type of bias. The use of so-called objective measurements, such as bronchial hyper-responsiveness, can be questioned, at least in an epidemiological setting, because they do not necessarily increase diagnostic accuracy 
. As there is currently no “golden standard” for the definition of asthma in epidemiological studies, different measurements can and should be used as long as the investigator is aware of each measurement’s particular weaknesses. As prescription data are collected prospectively and independently from the studied individuals, they represent unbiased data. We believe, therefore, that our outcome variable of dispensed asthma medication probably represents current disease more accurately than self-reported data, especially in the case of anti-inflammatory treatment. It is predominantly prescribed to patients with active disease 
and is more likely to be an indicator of clinically significant disease 
Information about asthma heredity was based on data collected by parental questionnaires when the individuals were about 18 months of age. The parents simply answered the question if they, their own parents or any of the child’s siblings ever had asthma. Age and sex of the siblings or numbers of siblings in the family were not recorded which might limit the significance of our variable. The aim of our analysis, however, was to assess if data about asthma disease in the family which can be easily and quickly collected by a physician when meeting a child and its parents could have any diagnostic value for the future.
Adult-onset asthma is known to be more prevalent in females than in males, in contrast to asthma before puberty. The reasons for this are multiple and include sex hormone influences, smoking habits, genetic and immunological factors, among others 
. We found no clear gender difference among those individuals with asthma as adults, neither in the whole group (n
60) nor in the sub-group of 49 individuals developing asthma after the follow-ups in childhood (). The latter group might include individuals who developed asthma even before puberty but who could not be recognized as such due to the lack of a follow-up between the age of 10–11 years and adulthood.
In the group of individuals who no longer have asthma but had asthma at the time of follow-up in childhood, a slight predominance of males over females was seen (34 versus 26). This supports previous discussions about girls having persistent asthma beyond puberty whereas boys more often show a decrease in symptoms with increasing age 
Gender-specific analyses in our study revealed an association between a family history of asthma and anti-inflammatory asthma treatment in adults for men but not for women. The reason for this difference is not obvious. It has been suggested that women have an increased risk of developing non-allergic asthma, which is why the phenotype of asthma in our adult population could influence our findings 
. The nature of our outcome variable, however, does not allow a distinction between non-allergic and allergic asthma. The association between elevated CB-IgE and anti-inflammatory asthma treatment among adults did not to differ greatly between different sexes (Table S2
Elevated CB-IgE and a family history of asthma at birth were associated with dispensed asthma medication and questionnaire-reported allergen-induced respiratory symptoms in adulthood. However, neither family history nor CB-IgE was useful for screening as their predictive ability in this population-based sample was poor.