Prediction of new-onset asthma and nasal allergy by skin prick test and RAST in a cohort of adults

Descriptive data

Of the 1185 adults asked to participate in ECRHS III, 924 responded (response rate 78.0%). In general, responders and nonresponders did not differ in their basic characteristics, as assessed in 1990–1992 during ECRHS I (table 1). However, responders were more likely to be female, from Erfurt and be of higher social status than nonresponders. In contrast, responders were less likely to own pets and currently smoke.

Outcome data

In total, 50 cases of new-onset asthma and 60 cases of new-onset nasal allergy occurred during the 20-year observation period (1990/1992–2010/2012), which corresponds to 16 072 and 13 668 person-years at risk, respectively.

Main results

The incidence rates for asthma and nasal allergy are presented in table 2. The asthma crude incidence rate was 3.11 per 1000 person-years and that for nasal allergy was 4.39 per 1000 person-years. In Hamburg, the incidence rate was higher than in Erfurt for both outcomes. Rates were also higher for females than males. At the end of the follow-up period, 44% of the new asthma cases and 10.3% of subjects without asthma stated that they had suffered from wheeze in the last 12 months. Additionally, 28 (56%) of the incident asthma cases reported current treatment with asthma medication and thus might have been free of symptoms. Of those not undergoing current treatment with asthma medication, 33.3% of the subjects with new-onset asthma and 9.7% without asthma reported wheeze symptoms in the last 12 months.

The relative risk and 95% confidence intervals for the relationship between each risk factor with new-onset asthma and nasal allergy are given in table 3. These findings differ slightly when the analyses are stratified by centre (online supplementary tables S1a and S2b).

The hazard ratios obtained by Cox regression, which were adjusted for sex, centre, social status, family history and smoking, were higher for each SPT compared with the corresponding RAST for both new-onset asthma and nasal allergy (table 4). A significantly increased hazard for new-onset asthma showed a positive RAST for cat, as well as positive SPTs for cat, D. pteronyssinus and birch. Positive SPTs for cat and birch, as well as a positive RAST for birch were significantly associated with an increased hazard for new-onset nasal allergy. The highest hazard ratio was observed between a positive SPT for cat and new-onset asthma as well as between a positive SPT for birch and new-onset nasal allergy. Any positive SPT and any positive RAST significantly increased the hazard for new-onset asthma and nasal allergy. Thus, for both outcomes, the hazard ratio was stronger for any positive SPT than for any positive RAST. When the analysis was stratified by centre, this trend was similar in Hamburg but reversed in Erfurt. However, the numbers in both of these stratified analyses are small and there is a substantial overlap in the confidence intervals (online supplementary tables S1a–S4b).

As seen in figure 2, there is partial overlap between individuals with new-onset asthma and nasal allergy who were positive to any SPT and any RAST. Among those with either new-onset asthma or nasal allergy, more individuals tested positive for any RAST than for any SPT. In total, 31 (62%) out of 50 individuals with new-onset asthma and 38 (63%) out of 60 individuals with new-onset nasal allergy tested positive for at least one SPT and/or at least one RAST at baseline and, therefore, were atopic. The population attributable fraction for asthma and nasal allergy was 36% and 44%, respectively.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2–

Venn diagram depicting the overlap between any positive skin prick test (SPT) and any positive radioallergosorbent test (RAST) at baseline (1990–1992) in patients with a) a new onset of asthma (N=50) or b) nasal allergy (N=60) in European Community Respiratory Health Survey III. Any positive SPT: at least one positive SPT; Any positive RAST: at least one positive RAST.

Of the 874 nonasthmatic individuals at baseline, 196 were positive to at least one SPT, 21 of whom developed asthma during the 20-year observation period. Similarly, 293 of the 874 nonasthmatics were positive to at least one RAST, 25 of whom developed asthma. With regard to nasal allergy, 754 individuals were without nasal allergies at baseline, 123 of whom were positive to at least one SPT and 24 of whom subsequently developed nasal allergies during the follow-up period. Finally, 216 (of the 754 individuals without nasal allergies) were positive to at least one RAST, 31 of whom developed nasal allergies during follow-up. Sensitivity was higher for RAST than for SPT for both asthma (50.0%, 95% CI 36.1–63.9% versus 42.0%, 95% CI 28.3–55.7%) and nasal allergy (51.6%, 95% CI 39.0–64.3% versus 40.0%, 95% CI 27.6–52.4%). Although the confidence intervals are overlapping specificity was higher for SPT than for RAST for asthma (78.8, 95% CI 76.0–81.6 versus 67.5, 95% CI 64.3–70.7) and for nasal allergy (85.7, 95% CI 83.1–88.3 versus 73.3, 95% CI 70.1–76.6). As seen in table 5, the PPVs for SPT and RAST for asthma were very low and similar and the confidence intervals overlapped (0.11, 95% CI 0.07–0.16 versus 0.9, 95% CI 0.06–0.12). For nasal allergy, the PPVs for SPT and RAST were low and differed more substantially, although again the confidence intervals overlapped (0.2, 95% CI 0.13–0.28 versus 0.14, 95% CI 0.10–0.20). The NPVs for SPT and RAST were high and similar for asthma and nearly identical for nasal allergy.

LR+ and LR- for SPTs and RASTs were rather weak, as LR+ was (with the exception of SPTs in nasal allergy, where the LR+ was 2.8) below 2 and LR- was between 0.5 and 1.0 for both tests (table 5).

The relationship between time to onset of asthma/nasal allergy and new-onset asthma/nasal allergy for subjects with any positive SPT and subjects with any positive RAST at baseline is shown in figure 3. 2 years after the baseline survey (ECRHS I), the probability of new-onset asthma was greater for individuals with at least one positive SPT than for those with at least one positive RAST. For all time-points, the probability of new-onset nasal allergy was higher for individuals with at least one positive SPT than for individuals with at least one positive RAST.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3–

Kaplan–Meier curves for new-onset of asthma and nasal allergy shown as the cumulative proportion of new-onset asthma/nasal allergy at certain times under control against the time to onset of a) nasal allergy or b) asthma. Participants were from Erfurt and Hamburg and were aged 21–47 years at baseline. SPT: skin prick test; RAST: radioallergosorbent test; any SPT positive: at least one positive skin prick test; any RAST positive: at least one positive radioallergosorbent test.

Key results

The present study provides information about the incidence, risk factors and prediction by SPT and RAST of new-onset asthma and nasal allergy in a population of German adults followed for 20 years.

The incidence rates for asthma and nasal allergy were higher in Hamburg than in Erfurt. Females had a higher incidence rate than males for both conditions. The incidence of asthma was related to sex, BHR, nasal allergy, social status, any positive SPT and any positive RAST. The incidence of nasal allergy was associated with eczema, any positive SPT and any positive RAST. SPT was a slightly better predictor for the new onset of asthma and nasal allergy than RAST. An analysis of allergen-specific SPT and RAST identified a positive reaction to the cat allergen as the strongest predictor for new-onset asthma and a positive reaction to the birch allergen as the strongest predictor for new-onset nasal allergy.

Limitations

The main limitation of long-term prospective cohort studies is a loss to follow-up. However, the response rate of our study in ECRHS III can be regarded as high (78% when only eligible subjects are considered). Responders did not differ from nonresponders in many potential risk factors, such as age, nasal allergy, eczema, SPT or RAST. Nevertheless, our sample cannot be considered as completely unbiased. The long observation period of our study, with time intervals of ∼10 years between follow-ups, might lead to recall bias, especially in terms of age of new-onset of disease. We used self-reported data from questionnaires to assign asthma and nasal allergy status, which may lead to misclassification, especially with regard to asthma, given the potential confounding influence of chronic obstructive pulmonary disease. We think asthma medication, as an outcome, is likely to be less subject to recall bias than self-reported physician diagnosis of asthma or nasal allergy. Including more self-reported respiratory symptoms as outcomes would have similar limitations. More “objective” data, such as lung function measurements and BHR testing or a clinical case ascertainment, would diminish the misclassification bias of this study. Unfortunately, this information was not available. Also, it is possible that some of the new-onset asthma cases are attributable to a change in diagnostic labelling.

14 and 46 individuals, respectively, reported that they ever had asthma and nasal allergy in ECRHS I but did not report these symptoms in ECRHS III. These inconsistencies could be due to a true remission in individuals who have had asthma or nasal allergy years ago, but that the disease did not persist. Considering the long observation period of this study, it is plausible that some individuals with new-onset asthma or nasal allergy who had negative SPT and RAST at baseline became sensitised during the observation period. Furthermore, a lower cut-off point for the tests might lead to a higher sensitivity.

The large sample size, international standardised methods of data collection following ECRHS protocols and he population-based setting of this study are unique and important strengths. Moreover, the design of a prospective cohort with a long (20-year) follow-up time, which includes a question on new onset, allows incidence rates to be calculated. The detailed information on health behaviours and risk factors, gained through questionnaires and medical examinations, permits analyses to be adjusted for a large number of potential confounders.

Interpretation

Incidence rates of asthma (crude 3.1, males 2.2 and females 4.1) were within the range of those observed in previous cohort studies (crude 2.2–4.5, males 1.0–4.5 and females 1.3–4.9) [12–14]. As reported by most other studies, a higher incidence rate was found for females than for males [9, 10]. One study reported that only the risk of developing nonallergic asthma differed by sex [15].

In the current study, incidence rates of nasal allergy (crude 4.4, males 3.6 and females 5.3) are below the lifelong incidence reported by Matheson et al. [4] in all 48 centres of the ECRHS study (7.0 males and 7.9 females). This difference might be due to the fact that our study does not report lifelong incidence but rather the incidence of nasal allergy between 1990 and 1992, and 2010 and 2012 only. This shorter timeframe probably leads to less recall and misclassification bias.

Although risk factors for asthma and nasal allergy have been assessed previously, this is the first cohort study to show the effect of potential determinants on new-onset of disease over a 20-year follow-up.

The association between low social status and new-onset asthma found in the present study confirms the results of previous publications that have reported similar associations with the prevalence of asthma [16], with the exception of the study by Montnemery et al. [17]. The association between socioeconomic status and asthma in children is complex [18] and is likely to depend on asthma severity, poverty, many lifestyle factors and potentially also country of residence. While (probably severe) asthma prevalence is higher in children from less privileged families in inner cities, we previously found that asthma is more prevalent in the higher social classes in a large cohort of German children, which is consistent with other German studies [19].

The impact of atopy on asthma remains unclear. 20 years ago, Burrows et al. [20] stated that asthma almost always has an allergic basis in children and adults. However, Anto et al. [3] and Zacharasiewicz et al. [21] found the attributable fraction of atopy on the development of asthma to be in the range of 12% and 30% in adults. The attributable fraction we report (36%) is higher.

Among subjects with new onset of nasal allergy, 63% were atopic (at least one positive SPT or RAST) at baseline. This finding is consistent with data published by Zacharasiewicz et al. [21], who reported that 61% of rhinitis cases were atopic. However, their attributable fraction (53%) is higher than that of the current study (44%). The nonatopic cases may represent rhinitis without an allergic base or may represent cases who were not atopic at baseline but who developed atopy during the observation period. Furthermore, including Timothy grass in the RAST would probably have lead to higher attributable fractions for new-onset asthma and nasal allergy.

SPT and RAST are both significant predictors for new-onset asthma and nasal allergy. The calculation of PPV and NPV demonstrated that positivity to at least one SPT is a better predictor for both conditions than positivity to at least one RAST. However, the overlapping confidence interval of these point estimates indicates that this potential superiority is likely to be minor. The negative prediction of at least one positive SPT and at least one positive RAST were (nearly) equal and performed better than the positive predictions. The sensitivity of RAST for new onset was stronger than of SPT, while specificity showed better results for SPT. Similar results were reported for both conditions for adults by Tschopp et al. [22] using Phadiatop instead of RAST and for children by Schäfer et al. [23] with regard to hay fever.

The likelihood ratio indicates the value of a diagnostic test by comparing the likelihood that a certain test result would be expected in a patient with the definite disorder with the likelihood of the same result expected in a patient without the definite disorder. Here, a distinction is drawn between LR+, concerning positive test results, and LR-, concerning negative test results. The classification of the test value given by the likelihood ratio is determined as excellent (LR+ >10 and LR- <0.10), good (LR+ 5–10 and LR- 0.1–0.2), moderate ( LR+ 2–5 and LR- 0.2–0.5) or weak (LR+ 1–2 and LR- 0.5–1.0) [24]. Therefore, the value of SPT and RAST in terms of prediction of new-onset asthma and nasal allergy is poor. When interpreting the test values of SPT and RAST for new-onset asthma, it must be acknowledged that only a proportion of all asthma cases are atopic. Furthermore, the much lower rate of SPT positivity than RAST positivity might have an effect on these estimates. Sensitisation to cat allergen was most strongly associated with new-onset asthma, followed by mite. Considering the allergens tested in this study, for new-onset nasal allergy, sensitisation to birch allergen was the strongest predictor, followed by cat. Positive SPT of every single allergen was more strongly related to new-onset of both asthma and nasal allergy than a positive RAST. This association might even increase with the use of a lower cut-off value for SPT [23, 25]. Previous studies have stated that the results of SPT and RAST are not interchangeable and that a positive SPT not accompanied by a positive RAST might be due to IgE antibodies that are not detectable by any RAST procedures [26, 27]. As previously shown, there are differences in the prevalence of atopic diseases between east and west Germany [7, 28]. Therefore, as anticipated, the current results differ between Hamburg and Erfurt when the analyses are stratified by centre. This observation supports the hypothesis of site-specific influences on risk factors for the new onset of asthma and nasal allergy. Although the sample sizes are small, there seems to be a generally higher risk of specific allergic sensitisation for asthma onset in Hamburg compared with Erfurt. Similar differences were not found for nasal allergies. As suggested by the International Study of Asthma and Allergies in Childhood [29], there might be substantial variation in the dependency between asthma and atopy, which could explain the differences in the associations between SPT and RAST with new onset of asthma and nasal allergy in Erfurt and Hamburg.

Conclusion

We report that SPT and RAST are both unpromising tests for prediction of new onset of asthma and nasal allergy in a cohort of German adults followed for 20 years, although a positive SPT seems to present a slightly better predictor than any detectable specific IgE. However, the role of atopy on these diseases remains ambiguous, especially with regard to asthma. Further prospective longitudinal studies are needed to determine long-term predictors for new-onset asthma and nasal allergy.