Clinical predictors of wheeze trajectories and associations with allergy in Asian children

Abstract

Background

Childhood wheezing is a highly heterogeneous condition with an incomplete understanding of the characteristics of wheeze trajectories, particularly for persistent wheeze.

Objective

To characterize predictors and allergic comorbidities of distinct wheeze trajectories in a multiethnic Asian cohort.

Methods

A total of 974 mother-child pairs from the prospective Growing Up in Singapore Towards healthy Outcomes (GUSTO) cohort were included in this study. Wheeze and allergic comorbidities in the first 8 years of life were assessed using the modified International Study of Asthma and Allergies in Childhood questionnaires and skin prick tests. Group-based trajectory modeling was used to derive wheeze trajectories and regression was used to assess associations with predictive risk factors and allergic comorbidities.

Results

There were 4 wheeze trajectories derived, including the following: (1) early-onset with rapid remission from age 3 years (4.5%); (2) late-onset peaking at age 3 years and rapidly remitting from 4 years (8.1%); (3) persistent with a steady increase to age 5 years and high wheeze occurrence until 8 years (4.0%); and (4) no or low wheeze (83.4%). Early-onset wheezing was associated with respiratory infections during infancy and linked to subsequent nonallergic rhinitis throughout childhood. Late-onset and persistent wheeze shared similar origins characterized by parent-reported viral infections in later childhood. However, persistent wheezing was generally more strongly associated with a family history of allergy, parent-reported viral infections in later childhood, and allergic comorbidities as compared with late-onset wheezing.

Conclusion

The timing of viral infection occurrence may determine the type of wheeze trajectory development in children. Children with a family history of allergy and viral infections in early life may be predisposed to persistent wheeze development and the associated comorbidities of early allergic sensitization and eczema.

Introduction

Wheezing is one of the most common symptoms in infants and young children. The global prevalence of wheezing in infants is estimated at 36%,¹ whereas the cumulative prevalence of wheezing and asthma in schoolchildren aged 4 to 6 years in Singapore is 27.5% and 11.7%, respectively.² Childhood wheezing is highly heterogeneous,3, 4, 5, 6 and a better understanding of wheezing phenotypes and their associated risk factors is critical to develop interventions. For instance, compared with transient wheezing patterns, recurrent or persistent wheeze patterns are more likely to be associated with asthma and the development of abnormal pulmonary function in later life.^7,8

The first study classifying longitudinal observations on wheezing in relation to asthma was conducted by the Tucson Children's Respiratory Group, which proposed 4 trajectories: (1) no wheeze, (2) transient wheeze, (3) late-onset wheeze, and (4) persistent wheeze.³ Subsequently, analytical approaches were extended to the use of unsupervised data-driven statistical methods such as group-based trajectory modeling and latent class analysis. These do not rely on a priori definitions to derive homogenous trajectories from longitudinal data.9, 10, 11 Using a data-driven approach, intermediate and nonpredefined phenotypes can be derived (studies described in eTable 1).^5,7,12,13 Current studies on wheeze trajectories are mainly conducted in non-Asian countries,¹⁴ but genetic and environmental drivers of wheeze may differ in Asian countries. For example, asthma candidate genes identified in European counties found no associations with asthma in Asian countries,¹⁵ suggesting ethnic differences in susceptibilities to allergies. There is only 1 study conducted among Japanese schoolchildren, which used group-based trajectory modeling to cluster wheeze data.¹⁶ In the current study, we used group-based trajectory modeling to determine whether distinct wheeze trajectories existed in the Growing Up in Singapore Towards healthy Outcomes (GUSTO) cohort, an Asian multiethnic mother-offspring cohort, using wheeze data in the first 8 years of life. We evaluated the risk factors and associations with allergic comorbidities including eczema, rhinitis, and allergic sensitization for each trajectory. Our overall hypothesis was that there are distinct wheeze trajectories, each with particular risk factors.

Methods

GUSTO Study Design

The detailed methodology for the GUSTO study was described by Soh et al.¹⁷ Trained interviewers gathered information on demographic characteristics, family history of allergy, social data, and lifestyle factors. Ethics approval was obtained from the Domain Specific Review Board of Singapore National Healthcare Group (D/2009/021) and the Centralized Institutional Review Board of SingHealth (2018/2767). The conduct of this study was on the basis of the guidelines in the Declaration of Helsinki. Informed written consent was obtained from all participants.

Definition of Allergic and Viral Outcomes

The modified International Study of Asthma and Allergies in Childhood questionnaire was used to evaluate offspring allergic symptoms at ages 3, 6, 9, 12, 15 months and 1.5, 2, 3, 4, 5, 6, 7, and 8 years. Wheeze with use of nebulizer was defined by positive responses to the questions: “Has your child ever wheezed?” and “Has your child ever been prescribed with nebulizer or inhaler treatment?”; eczema was defined as a positive response to the question: “Has your child ever been diagnosed with eczema?”; rhinitis was defined by a positive response to the question: “Has your child had runny nose, blocked or congested nose, that has lasted for 2 or more weeks duration?” Cumulative eczema or rhinitis by 1.5, 3, and 5 years was classified as “yes” when a patient answered “yes” by the time point and “no” if the patient answered “no” at all time points. Cumulative eczema or rhinitis by 8 years was classified as “yes” when a patient answered “yes” by 8 years and “no” if the patient answered “no” at 8 years (with a maximum of 2 missing responses for previous time points). Additional analyses were performed with atopic eczema, atopic rhinitis, and eczema with steroid use. Atopic eczema or rhinitis was defined as positive when the child was reported to have both the allergic condition and a positive skin prick test (defined below) at the same time point.

Viral infections in the first 8 years of life were defined as having parent-reported croup (data available up to 24 months) or bronchiolitis or bronchitis, similar to other birth cohort studies.^18,19 They were assessed using the questions “Has your child ever had any episodes of croupy cough or been diagnosed with croup?” and “Has your child ever been diagnosed with bronchiolitis or bronchitis?” Respiratory infections in the first 8 years of life were defined as having parent-reported viral infections or pneumonia.

Allergic Sensitization

Skin prick testing was conducted at ages 1.5, 3, 5, and 8 years, and included cow's milk, egg, peanut, and house dust mites, Dermatophagoides pteronyssinus, Dermatophagoides farina (Greer Laboratories, Lenoir, North Carolina) and Blomia tropicalis (developed in-house)²⁰ at 1.5 and 3 years; and all the above plus crab and shrimp (Greer Laboratories, Lenoir, North Carolina) at 5 and 8 years. At 8 years, skin prick testing for cockroach allergens Blattella germanica and Periplanata americana were also performed. It was defined as positive at a time point when any of the allergens tested positive at the time point with an average wheal size of at least 3-mm greater than the negative control saline. We have previously identified allergic sensitization trajectories in the GUSTO cohort using latent class analysis—early food and mite sensitization (16.2%), late mite sensitization (24.2%), and no or low sensitization (59.6%).²¹

Statistical Methods

Patients who had at least 50% follow-up data on wheeze were included in the analysis. Group-based trajectory modeling using the traj command in the Stata plugin²² was used to derive wheeze trajectories on the basis of questionnaire data. We fitted models from 1 through 5 trajectories and calculated the respective Bayesian information criterion (BIC) and the change in BIC between models (∆BIC). BIC has been widely used in similar studies evaluating wheeze trajectories.^{5,7,12,13,16,23} The optimal number of trajectories was chosen on the basis of the largest number of trajectories with 2∆BIC greater than 10 and the clinical interpretability of the model.²⁴ Posterior probabilities were generated to each child, and the child was subsequently allocated to the trajectory with the highest posterior probability.

Variables were summarized by frequency (percentage), median (interquartile range), mean (SD), or geometric mean as appropriate. As questionnaires were administered at a higher frequency in the first 2 years of life, the number of viral and respiratory infections were compared across the wheeze trajectories and not across time periods. χ2 and Fisher's exact tests were used to compare categorical variables. Continuous variables with approximately normal distribution were compared by independent 2-sample t test, whereas those with skewed distribution such as the number of viral infections were compared by Wilcoxon rank-sum test. Associations between predictors and wheeze trajectories were evaluated by multivariable multinomial logistic regression. The predictors include prenatal factors, such as ethnicity and child's sex, and early life environmental factors, such as type of feeding. The type of feeding was categorized as follows: (1) mainly breastfeeding (full breastfeeding for at least 4 months with continued breastfeeding for at least 6 months); (2) mainly formula (breastfeeding below 3 months or no breastfeeding); and (3) mixed feeding (between mainly breastfeeding and mainly formula categories). The association between the number of viral and respiratory infections experienced by each child and wheeze trajectory was evaluated by multivariable multinomial logistic regression. The cluster-robust SE was implemented to account for the correlation of multiple observations from the same child. Relationships between wheeze trajectories and allergic comorbidities (defined as binary outcomes) were assessed by multivariable Poisson regression with robust error variance adjusted for confounders. The Benjamini-Hochberg procedure was implemented to reduce type I errors associated with multiple comparisons. The above statistical analyses were performed in Stata SE 16.1 for Windows (StataCorp LLC, College Station, Texas), assuming 2-sided tests with a 5% significance level.

Results

Out of 1237 children, we included 974 with at least 50% follow-up data on wheeze in the analysis. Most of the mothers were of Chinese ethnicity (557 [57.3%]), had at least college or higher education (684 [70.9%]), and had a family history of allergy (479 [54.6%]). There were, in total, 510 (52.4%) boys, and 298 (30.6%) children born by means of cesarean section. A description of the study population is presented in Table 1. There were no differences in child's sex, family history of allergy, parity, mode of delivery, cat or dog ownership by the first year, childcare attendance by the first year, and rhinitis or respiratory infection by 6 months between included and excluded participants; a higher proportion of excluded mothers were of Malay ethnicity, had less than college education, were exposed to tobacco smoke during pregnancy and a higher proportion of excluded children had lower birthweight, were mainly formula-fed, and had eczema in early life (eTable 2).

Table 1.

Characteristics of Participants According to Wheeze Trajectory

Risk factors	Total (n = 974)	No or low wheeze (n = 812)	Early-onset (n = 44)	P	Late-onset (n = 79)	P	Persistent (n = 39)	P
Prenatal factors, n (%)
Sex				.08		.18		.009
Female	464 (47.6)	404 (49.8)	16 (36.4)		33 (41.8)		11 (28.2)
Male	510 (52.4)	408 (50.3)	28 (63.6)		46 (58.2)		28 (71.8)
Ethnicity				<.001		.70		.48
Chinese	557 (57.3)	475 (58.6)	16 (36.4)		47 (59.5)		19 (48.7)
Indian	179 (18.4)	152 (18.7)	6 (13.6)		12 (15.2)		9 (23.1)
Malay	237 (24.4)	184 (22.7)	22 (50.0)		20 (25.3)		11 (28.2)
Maternal education level				.70		.98		.22
Less than college	281 (29.1)	239 (29.6)	11 (26.8)		23 (29.5)		8 (20.5)
College and higher	684 (70.9)	568 (70.4)	30 (73.2)		55 (70.5)		31 (79.5)
Family history of allergy	479 (54.6)	368 (50.6)	28 (68.3)	.03	54 (73.0)	<.001	29 (82.9)	<.001
Parity				.95		.50		.55
0	438 (45.0)	364 (44.8)	20 (45.5)		37 (46.8)		17 (43.6)
1	327 (33.6)	275 (33.9)	14 (31.8)		22 (27.9)		16 (41.0)
≥ 2	209 (21.5)	173 (21.3)	10 (22.7)		20 (25.3)		6 (15.4)
Environmental factors, n (%)
Tobacco exposure during pregnancy	339 (36.5)	268 (34.5)	23 (54.8)	.007	33 (45.2)	.06	15 (40.5)	.43
Birth weight (kg), mean (SD)	3.10 (0.44)	3.11 (0.44)	3.01 (0.54)	.25	3.13 (0.48)	.69	3.07 (0.33)	.51
Cesarean section	298 (30.6)	252 (31.0)	14 (32.6)	.83	22 (27.9)	.56	10 (25.6)	.48
Cat ownership by 1 y	32 (4.0)	27 (4.1)	2 (5.7)	.65	2 (2.9)	>.99	1 (3.5)	>.99
Dog ownership by 1 y	60 (7.6)	49 (7.4)	1 (2.8)	.51	10 (14.9)	.03	0 (0.0)	.26
Childcare attendance by 1 y	72 (13.0)	52 (11.2)	13 (48.2)	<.001	3 (7.3)	.60	4 (18.2)	.30
Type of feeding				.66		.14		.91
Mainly breastfeeding	119 (12.7)	105 (13.4)	4 (9.8)		5 (6.7)		5 (13.2)
Mixed feeding	424 (45.2)	345 (44.0)	21 (51.2)		40 (53.3)		18 (47.4)
Mainly formula	396 (42.2)	335 (42.7)	16 (39.0)		30 (40.0)		15 (39.5)
Eczema in early life by 6 mo	74 (8.6)	53 (7.3)	3 (7.5)	>.99	8 (11.6)	.20	10 (29.4)	<.001
Rhinitis in early life by 6 mo	233 (26.8)	173 (23.8)	24 (60.0)	<.001	22 (31.0)	.18	14 (43.8)	.01
Respiratory infections by 6 moa	51 (5.6)	30 (4.0)	14 (34.2)	<.001	3 (4.0)	>.99	4 (11.1)	.06

NOTE. Benjamini-Hochberg procedure with a 5% false discovery rate was applied.

^aRespiratory infections include bronchiolitis, croup, and pneumonia.

Identification of Wheeze Trajectories

In this study, 234 (30.5%) children reported at least 1 occurrence of wheezing in the first 8 years of life. Group-based trajectory modeling classified the children into 4 wheeze trajectories: (1) early-onset with rapid remission from age 3 years (44 [4.5%]); (2) late-onset peaking at age 3 years, and rapidly remitting from 4 years (79 [8.1%]); (3) persistent with a steady increase to age 5 years and high wheeze occurrence until 8 years of age (39 [4.0%]); and (4) no or low wheeze (812 [83.4%]) (Table 1, eTable 3, and Fig 1). The no or low wheeze trajectory, which included children who had less than 1% probability of wheeze development throughout the 8 years, was used as the reference trajectory.

Figure 1.

Wheeze trajectories based on children with at least 50% follow-up data (n = 974).

Risk Factors

The characteristics and risk factors for each trajectory are presented in Tables 1 and 2, respectively. Compared with the no or low wheeze trajectory, the odds of early-onset wheeze were higher with maternal tobacco exposure during pregnancy, childcare attendance during infancy, and rhinitis and respiratory infections in the first 6 months of life. The odds of late-onset wheeze were higher with mixed feeding, and those of persistent wheeze were higher for those who had eczema in the first 6 months of life. A family history of allergy was associated with higher odds of both late-onset and persistent wheeze, but the association was stronger for persistent wheeze. As childcare attendance had the most number of missing responses in the multivariable multinomial logistic regression, a sensitivity analysis was performed using imputation to simulate the extreme case scenarios (all missing responses were either “did not attend” or “attended”). The directions of associations remained similar (data not shown), hence complete case analysis results are presented.

Table 2.

Associations Between Participant Characteristics and Wheeze Trajectories by Multivariable Multinomial Logistic Regression (n = 431)

	Early-onset		Late-onset		Persistent
Risk factors	AOR (95% CI)a	P	AOR (95% CI)a	P	AOR (95% CI)a	P
Prenatal factors
Sex
Female	Ref	—	Ref	—	Ref	—
Male	0.76 (0.23-2.49)	.65	0.59 (0.26-1.34)	.21	1.71 (0.51-5.72)	.39
Ethnicity		.18		.22		.44
Chinese	Ref	—	Ref	—	Ref	—
Indian	0.83 (0.09-7.92)	.87	0.42 (0.09-1.98)	.28	2.79 (0.58-13.38)	.20
Malay	3.44 (0.89-13.32)	.07	0.41 (0.12-1.37)	.15	1.24 (0.31-5.02)	.76
Maternal education level
Less than college	Ref	—	Ref	—	Ref	—
College and higher	7.94 (0.83-76.17)	.07	0.53 (0.20-1.41)	.20	0.89 (0.24-3.22)	.85
Family history of allergy	1.14 (0.34-3.89)	.83	5.90 (2.12-16.46)	.001	6.50 (1.31-32.08)	.02
Parity		.88		.60		.98
0	Ref	—	Ref	—	Ref	—
1	1.12 (0.34-3.68)	.85	1.63 (0.63-4.19)	.31	1.08 (0.31-3.77)	.90
≥ 2	0.69 (0.11-4.18)	.68	1.37 (0.44-4.21)	.59	0.93 (0.18-4.77)	.94
Cesarean section	1.64 (0.50-5.36)	.42	0.78 (0.30-1.99)	.60	1.40 (0.44-4.44)	.57
Birth weight	1.50 (0.40-5.55)	.55	1.25 (0.47-3.34)	.65	0.39 (0.11-1.36)	.14
Tobacco exposure during pregnancy	3.78 (1.05-13.61)	.04	1.24 (0.51-3.04)	.64	1.20 (0.31-4.63)	.79
Environmental factors
Cat ownership by 1 y	0.20 (0.02-2.17)	.19	0.73 (0.08-7.06)	.79	1.09 (0.11-11.14)	.94
Dog ownership by 1 y	0.81 (0.07-9.60)	.87	2.90 (0.87-9.63)	.08	—	—
Childcare attendance by 1 y	5.39 (1.66-17.54)	.005	0.17 (0.02-1.37)	.10	0.95 (0.21-4.25)	.95
Type of feeding		.44		.01		.69
Mainly breastfeeding	Ref	—n	Ref	—	Ref	—
Mixed feeding	1.33 (0.29-6.13)	.71	4.89 (1.27-18.83)	.02	1.98 (0.35-11.15)	.44
Mainly formula	0.60 (0.11-3.31)	.56	1.51 (0.35-6.45)	.58	1.37 (0.20-9.48)	.75
Eczema in early life by 6 mo	0.81 (0.16-4.05)	.80	0.58 (0.12-2.75)	.49	5.98 (1.72-20.77)	.005
Rhinitis in early life by 6 mo	3.39 (1.06-10.80)	.04	1.60 (0.61-4.16)	.34	1.88 (0.58-6.10)	.30
Respiratory infections by 6 mob	11.72 (3.16-43.43)	<.001	0.80 (0.08-7.97)	.85	1.73 (0.26-11.36)	.57

Abbreviations: AOR, adjusted odds ratio; CI, confidence interval.

^aNo or low wheeze trajectory was the baseline group for odds.

^bRespiratory infections include bronchiolitis, croup, and pneumonia.

Viral and Respiratory Infections

The associations of the number of parent-reported viral and respiratory infections with wheeze trajectories are presented in Table 3. A higher proportion of children with early transient wheeze had pneumonia infections from 0 to 2 years as compared with those with no or low wheeze (eTable 4). With reference to no or low wheeze, viral and respiratory infections from 0 to 2 years, 2 to 4 years, 4 to 6 years, and 6 to 8 years were associated with higher odds of late-onset wheeze and persistent wheeze, with stronger associations throughout the 8 years for persistent wheeze as compared with late-onset wheeze, whereas such associations were observed for only 0 to 2 years and 2 to 4 years for early-onset wheeze.

Table 3.

Associations Between the Number of (A) Viral (N=851) and (B) Respiratory (N = 853) Infections Experienced by Each Child and Wheeze Trajectory

(A)
Viral infections	Early-onset		Late-onset		Persistent
	AOR (95% CI)a	P	AOR (95% CI)a	P	AOR (95% CI)a	P
Number of viral infections from 0 to 2 y	4.97 (3.43-7.21)	<.001	1.75 (1.24-2.45)	.001	2.39 (1.60-3.58)	<.001
Number of viral infections from 2 to 4 y	7.37 (4.12-13.19)	<.001	9.22 (5.42-15.68)	<.001	10.33 (5.23-20.40)	<.001
Number of viral infections from 4 to 6 y	1.28 (0.31-5.18)	.73	8.01 (4.23-15.16)	<.001	12.80 (5.90-27.74)	<.001
Number of viral infections from 6 to 8 y	1.52 (0.27-8.73)	.64	3.66 (1.55-8.63)	.003	10.60 (4.02-27.94)	<.001
(B)
Respiratory infections	Early-onset		Late-onset		Persistent
	AOR (95% CI)a	P	AOR (95% CI)a	P	AOR (95% CI)a	P
Number of respiratory infections from 0 to 2 y	4.96 (3.40-7.22)	<.001	1.78 (1.28-2.47)	.001	2.46 (1.65-3.69)	<.001
Number of respiratory infections from 2 to 4 y	5.73 (3.26-10.08)	<.001	6.95 (4.09-11.81)	<.001	7.94 (4.22-14.94)	<.001
Number of respiratory infections from 4 to 6 y	1.03 (0.27-3.89)	.97	5.84 (3.21-10.60)	<.001	9.74 (4.69-20.24)	<.001
Number of respiratory infections from 6 to 8 y	1.15 (0.24-5.59)	.87	2.56 (1.21-5.43)	.01	6.55 (2.84-15.09)	<.001

Abbreviations: AOR, adjusted odds ratio; CI, confidence interval.

^aReference: No or low wheeze trajectory. Sex, ethnicity, maternal education level, and family history of allergy were adjusted with cluster-robust standard errors.

Allergic Comorbidities

The proportions and associations of allergic comorbidities with wheeze trajectories are presented in eTable 5 and Table 4, respectively. Children with early-onset wheeze had higher risks of rhinitis at all time points as compared with those with no or low wheeze. Children with late-onset wheeze had higher risks of developing eczema, atopic eczema, and eczema with steroids use in the first 8 years and allergic sensitization in the first 5 years than no or low wheeze. Similarly, those with persistent wheeze had higher risks of eczema, atopic eczema, eczema with steroids use, and allergic sensitization development in the first 8 years than no or low wheeze. Further analysis found that persistent wheeze was associated with a higher risk of atopic eczema by 3 and 5 years and allergic sensitization by 5 years than late-onset wheeze (eTable 6). An additional analysis was performed to differentiate between the risk of developing food and dust mite sensitization as compared with no or low wheeze (eTable 7). Late-onset wheeze was associated with a higher risk of sensitization to food allergens by 8 years and dust mites in the first 5 years of life, whereas persistent wheeze was associated with higher risks of both food and dust mite sensitization in the first 8 years of life than no or low wheeze.

Table 4.

Associations Between Wheeze Trajectory and Allergic Comorbidities

Allergic comorbidities	n	Early-onset		Late-onset		Persistent
		ARR (95% CI)a	P	ARR (95% CI)a	P	ARR (95% CI)a	P
Allergic sensitization by 1.5 y	760	0.21 (0.03-1.46)	.12	2.28 (1.49-3.48)	<.001	2.41 (1.37-4.25)	.002
Allergic sensitization by 3 y	705	1.32 (0.80-2.18)	.28	1.93 (1.45-2.58)	<.001	2.04 (1.39-3.01)	<.001
Allergic sensitization by 5 y	660	1.16 (0.81-1.64)	.42	1.34 (1.07-1.67)	.009	1.82 (1.46-2.28)	<.001
Allergic sensitization by 8 y	660	1.16 (0.96-1.40)	.12	1.11 (0.95-1.30)	.18	1.26 (1.08-1.46)	.003
Eczema by 1.5 y	709	1.75 (1.02-3.00)	.04	1.85 (1.27-2.70)	.001	2.46 (1.68-3.60)	<.001
Eczema by 3 y	672	1.42 (0.87-2.33)	.16	1.75 (1.28-2.41)	<.001	2.10 (1.46-3.01)	<.001
Eczema by 5 y	612	1.31 (0.82-2.10)	.26	1.62 (1.21-2.17)	.001	2.11 (1.58-2.83)	<.001
Eczema by 8 y	713	1.43 (0.95-2.15)	.09	1.70 (1.31-2.19)	<.001	1.80 (1.32-2.46)	<.001
Rhinitis by 1.5 y	729	1.55 (1.29-1.86)	<.001	1.20 (0.96-1.51)	.12	1.27 (0.97-1.65)	.08
Rhinitis by 3 y	723	1.34 (1.13-1.59)	.001	1.16 (0.97-1.39)	.10	1.25 (1.02-1.53)	.03
Rhinitis by 5 y	698	1.24 (1.05-1.47)	.01	1.15 (0.97-1.35)	.10	1.20 (0.99-1.45)	.07
Rhinitis by 8 y	766	1.27 (1.07-1.52)	.007	1.20 (1.03-1.40)	.02	1.18 (0.96-1.44)	.12
Atopic eczema by 1.5 y	645	—	—	2.37 (1.00-5.60)	.05	5.59 (2.58-12.12)	<.001
Atopic eczema by 3 y	610	2.05 (0.80-5.24)	.14	2.60 (1.39-4.88)	.003	5.42 (2.94-9.98)	<.001
Atopic eczema by 5 y	533	1.90 (0.91-3.95)	.09	2.26 (1.39-3.67)	.001	4.47 (2.82-7.07)	<.001
Atopic eczema by 8 y	651	1.47 (0.75-2.89)	.26	1.75 (1.14-2.68)	.01	2.76 (1.85-4.12)	<.001
Atopic rhinitis by 1.5 y	653	0.34 (0.05-2.29)	.27	2.17 (1.13-4.14)	.02	2.54 (1.25-5.16)	.01
Atopic rhinitis by 3 y	642	1.38 (0.64-2.96)	.41	2.58 (1.67-3.99)	<.001	2.45 (1.41-4.26)	.002
Atopic rhinitis by 5 y	585	1.55 (0.92-2.60)	.10	1.64 (1.14-2.36)	.008	1.92 (1.25-2.95)	.003
Atopic rhinitis by 8 y	649	1.50 (1.03-2.18)	.03	1.19 (0.84-1.68)	.34	1.43 (1.03-2.00)	.04
Eczema with steroids use by 1.5 y	667	1.79 (0.82-3.94)	.15	2.09 (1.19-3.66)	.01	3.58 (2.11-6.07)	<.001
Eczema with steroids use by 3 y	617	1.35 (0.62-2.92)	.45	1.95 (1.20-3.18)	.007	2.80 (1.72-4.56)	<.001
Eczema with steroids use by 5 y	544	1.32 (0.63-2.74)	.46	1.89 (1.20-2.96)	.006	2.70 (1.75-4.17)	<.001
Eczema with steroids use by 8 y	686	1.37 (0.75-2.51)	.31	1.70 (1.15-2.50)	.007	2.13 (1.40-3.24)	<.001

Abbreviations: ARR, adjusted relative risk; CI, confidence interval.

NOTE. Poisson regression with robust error variance was implemented by adjusting for sex, ethnicity, maternal education level, and family history of allergy for each allergic comorbidity.

^aReference: No/low wheeze trajectory.

Comparisons of Allergic Sensitization Trajectories and Family History of Allergy Between Late-Onset and Persistent Wheeze Trajectories

To further differentiate the late-onset and persistent wheeze trajectories, their allergic sensitization trajectories were compared. Higher proportions of children with persistent wheeze had early food and mite and late mite sensitization as compared with those with late-onset wheeze (eTable 8). A higher proportion of children with persistent wheeze had a family history of allergy as compared with those with late-onset wheeze (P = .02).

Discussion

In this study, group-based trajectory modeling of data collected over multiple time points in the first 8 years of life generated 4 distinct wheeze trajectories (no or low, early-onset, late-onset and persistent), and their associations with clinical risk factors were assessed. This is the first study of data-driven wheeze trajectories in a multiethnic population-based birth cohort from Asia. The derived trajectories are clinically relevant as similar wheeze trajectories were reported by the Millennium Cohort Study, Sibilancias de Lactante y Asma de Mayor cohort, and Columbia Center for Children's Environmental Health birth cohort study.^23,25,26

On the basis of our findings, early-onset wheeze was likely nonallergic and was instead linked to early respiratory infections. Respiratory infections in infancy, coupled with other environmental risk factors linked to viral infections—that is, childcare attendance, and tobacco exposure during pregnancy—were associated with early-onset wheeze. These findings are supported by other studies; the Childhood Origins of ASThma cohort reported that rhinovirus infection in the first year of life was the strongest predictor of wheezing at age 3 years²⁷ whereas Rusconi et al²⁸ reported that early childcare attendance increased the risk of transient early wheezing. Tobacco exposure during pregnancy has been found to affect fetal lung development²⁹ and increase the risk of infections.³⁰ Moreover, early-onset wheeze was associated with a higher risk of rhinitis but not atopic rhinitis development in later childhood. This suggests a shared pathology with nonallergic rhinitis, sometimes termed “united airway disease.”³¹ Similarly, the European Community Respiratory Health Survey found that rhinitis and asthma were associated independently of allergic sensitization.³² Wheeze or asthma and rhinitis often co-occur and are characterized by lower and upper airway inflammation, respectively,³³ and may be allergic or nonallergic in nature. Nonallergic wheeze and rhinitis can result from viral infections because of viral hypersensitivity.^31,34

Viral infections later in childhood may also trigger the development of late-onset and persistent wheeze. However, unlike early-onset wheeze, these 2 trajectories were in addition associated with higher risks of eczema and allergic sensitization, suggesting atopy. This follows the classic atopic march model in which early sensitization and eczema onset are associated with skewed T_H2 immunity and the subsequent development of wheeze later in life.³⁵ Supportive evidence is provided by the Manchester Asthma and Allergy Study and Avon Longitudinal Study of Parents and Children cohorts, which identified an atopic march disease class featuring persistent eczema with later onset of persistent wheeze.³⁶ The associations with atopy and family history of allergy suggest genetic origins for both trajectories. Indeed, several genes that influence susceptibility to wheeze and eczema are expressed in the epithelium.^37,38

As compared with the late-onset wheeze trajectory, we observed that persistent wheeze was associated with higher risks of atopic diseases. Higher proportions of children with persistent wheeze had early food and mite sensitization and late mite sensitization as compared with those with late-onset wheeze, hence, indicating the involvement of early mite sensitization in driving wheeze persistence. In addition, a family history of allergy, early-onset eczema, and early viral infections were stronger risk factors for persistent wheeze than for late-onset wheeze. These findings are supported by the Children's Health Study, which reported that parental or sibling history of asthma had a stronger association with early-onset persistent asthma as compared with early-onset transient or late-onset asthma³⁹; Project Viva, which reported that early eczema by 6 months increased the risk of persistent wheeze⁴⁰; and a study involving US infants aged 2 to 18 months, which reported that bronchiolitis can trigger eosinophilic inflammation and increase the risk of persistent wheezing at 7 years.⁴¹

Taken together, we propose that the timing of viral infection occurrence is a key determinant of the type of wheeze trajectory development. Viral infections in infancy may increase susceptibility to early-onset wheeze. Conversely, acquiring viral infections later in childhood may confer a higher risk of late-onset or persistent wheeze that is differentiated by the genetic risk of allergy. A family history of allergy coupled with the development of early viral infections may drive persistence in wheeze that is comorbid with allergic sensitization and eczema. Although high proportions of children with late-onset wheeze also had atopy and viral infections, we postulate that they might have experienced remission because of weaker associations with a family history of allergy and fewer early viral infections.

Strengths of this study include interviewer-administered questionnaires at regular time intervals to collect data on wheeze and other allergic outcomes, allowing assessment of longitudinal changes. We were able to evaluate a large number of risk factors for the wheeze trajectories because of extensive data collection. Moreover, we incorporated analysis with allergic sensitization trajectories in this study and were able to elucidate the importance of mite sensitization, a key allergen in tropical countries, in wheeze development. This study was performed in a multiethnic cohort and accounted for possible genetic and lifestyle differences through adjustment for ethnicity, hence, the results are more representative of the diverse Southeast Asian population.

A limitation is the use of parental reports to assess allergies, but recall bias was minimized by regular follow-up with structured interviewer-administered questionnaires. In addition, viral infections were ascertained through parental reports and not viral cultures, which may also be subjected to recall bias. In addition, pet ownership is low in our cohort and may preclude meaningful study of the potential effects of pet ownership on wheeze development.

Conclusion

In conclusion, this study derived 4 distinct wheeze trajectories using group-based trajectory analysis and identified different risk factors and allergic comorbidities in an Asian cohort with ethnic diversity. The timing of viral infection occurrence may determine the type of wheeze trajectory development in children. Viral infections in infancy were linked to nonatopic early-onset wheeze that coexists with nonallergic rhinitis. Acquiring viral infections later in childhood was linked to late-onset and persistent wheeze development, which were differentiated by genetic risk to allergy. A family history of allergy coupled with the development of early viral infections may drive persistence in wheeze that is comorbid with allergic sensitization and eczema.

eReferences

• 1.Oksel C, Granell R, Haider S, Fontanella S, Simpson A, Turner S, et al. Distinguishing wheezing phenotypes from infancy to adolescence. A pooled analysis of five birth cohorts. Ann Am Thorac Soc. 2019;16(7):868-876.
• 2.Duijts L, Granell R, Sterne JAC, Henderson AJ. Childhood wheezing phenotypes influence asthma, lung function and exhaled nitric oxide fraction in adolescence. Eur Respir J. 2016;47(2):510.
• 3.Henderson J, Granell R, Heron J, Sherriff A, Simpson A, Woodcock A, et al. Associations of wheezing phenotypes in the first 6 years of life with atopy, lung function and airway responsiveness in mid-childhood. Thorax. 2008;63(11):974.
• 4.Savenije OE, Granell R, Caudri D, Koppelman GH, Smit HA, Wijga A, et al. Comparison of childhood wheezing phenotypes in 2 birth cohorts: ALSPAC and PIAMA. J Allergy Clin Immunol. 2011;127(6):1505-1512.e14.
• 5.Caudri D, Savenije OEM, Smit HA, Postma DS, Koppelman GH, Wijga AH, et al. Perinatal risk factors for wheezing phenotypes in the first 8 years of life. Clin Exp Allergy. 2013;43(12):1395-1405.
• 6.Yang L, Narita M, Yamamoto-Hanada K, Sakamoto N, Saito H, Ohya Y. Phenotypes of childhood wheeze in Japanese children: A group-based trajectory analysis. Pediatr Allergy Immunol. 2018;29(6):606-611.
• 7.Kotecha SJ, Watkins WJ, Lowe J, Granell R, Henderson AJ, Kotecha S. Comparison of the associations of early-life factors on wheezing phenotypes in preterm-born children and term-born children. Am J Epidemiol. 2019;188(3):527-536.
• 8.Belgrave DCM, Simpson A, Semic-Jusufagic A, Murray CS, Buchan I, Pickles A, et al. Joint modeling of parentally reported and physician-confirmed wheeze identifies children with persistent troublesome wheezing. J Allergy Clin Immunol. 2013;132(3):575-583.e12.
• 9.Cano-Garcinuño A, Mora-Gandarillas I, SLAM Study Group. Wheezing phenotypes in young children: an historical cohort study. Prim Care Respir J. 2014;23(1):60-66.
• 10.Chen Q, Just AC, Miller RL, Perzanowski MS, Goldstein IF, Perera FP, et al. Using latent class growth analysis to identify childhood wheeze phenotypes in an urban birth cohort. Ann Allergy Asthma Immunol. 2012;108(5):311-315.e1.
• 11.Lodge CJ, Zaloumis S, Lowe AJ, Gurrin LC, Matheson MC, Axelrad C, et al. Early-life risk factors for childhood wheeze phenotypes in a high-risk birth cohort. J Pediatr. 2014;164(2):289-294.e1-2.
• 12.Tse SM, Rifas-Shiman SL, Coull BA, Litonjua AA, Oken E, Gold DR. Sex-specific risk factors for childhood wheeze and longitudinal phenotypes of wheeze. J Allergy Clin Immunol. 2016;138(6):1561-1568.e6.
• 13.Depner M, Fuchs O, Genuneit J, Karvonen AM, Hyvärinen A, Kaulek V, et al. Clinical and epidemiologic phenotypes of childhood asthma. Am J Respir Crit Care Med. 2014;189(2):129-138.

eTable 1.

Studies Using Unsupervised Data-Driven Methods to Study Wheeze Trajectories

Study	Study population	Wheeze data collection time points	Wheeze trajectories	Key findings
Oksel et al,1 2019	STELAR consortium of 5 birth cohorts	• Infancy: 0.5-1 y • Early childhood: 2-3 y • Preschool or early school age: 4-5 y • Middle childhood: 8-10 y • Adolescence: 14-18 y	• Never or infrequent wheeze • Early-onset preschool remitting wheeze • Early-onset mid-childhood remitting wheeze • Persistent wheeze • Late-onset wheeze	• All trajectories were associated with higher risks of asthma development and poorer lung function • Persistent trajectory was linked to the highest risk of asthma
Duijts et al,2 2016	ALSPAC cohort	• 6, 18, 30, 42, 54, 69 and 81 mo	• Never or infrequent wheeze • Transient early wheeze • Prolonged early wheeze • Intermediate-onset wheeze • Late-onset wheeze • Persistent wheeze	• All trajectories were associated with higher risks of asthma development. • Intermediate-onset and persistent wheeze were linked to the highest risk of asthma and poor lung function.
Henderson et al,3 2008	ALSPAC cohort	• 6, 18, 30, 42, 54, 69 and 81 mo	• Never or infrequent wheeze • Transient early wheeze • Prolonged early wheeze • Intermediate-onset wheeze • Late-onset wheeze • Persistent wheeze	• Intermediate-onset, late-onset, and persistent wheeze were linked to a higher risk of any skin prick sensitivity. • Maternal history of asthma and allergy increased the risk of developing all trajectories (highest risk for persistent wheeze).
Savenije et al,4 2011 Caudri et al,5 2013	ALSPAC and PIAMA cohorts	• ALSPAC: 6, 18, 30, 42, 54, 69, 81, and 91 mo • PIAMA: 3, 12, 24, 36, 48, 60, 72, 84, and 96 mo	ALSPAC: • Never or infrequent wheeze • Transient early wheeze • Prolonged early wheeze • Intermediate-onset wheeze • Late-onset wheeze • Persistent wheeze PIAMA • Never or infrequent • Transient early wheeze • Intermediate-onset wheeze • Late-onset wheeze • Persistent wheeze	PIAMA • Intermediate-onset and persistent wheeze were linked to the highest risk of asthma and poorer lung function. • Intermediate-onset wheeze was associated with the highest risk of allergic sensitization at 4 and 8 y. • Consistent with results from ALSPAC. • Male sex, parental allergy, young maternal age, higher maternal BMI, low gestational age, smoke exposure during pregnancy, having older siblings, and childcare attendance increased the risk of transient early wheeze. • Male sex, parental allergy, and low breastfeeding increased the risk of persistent wheeze. • Low birthweight increased the risk of intermediate-onset wheeze. • Maternal allergy increased the risk of late-onset wheeze.
Yang et al,6 2018	Hospital-based birth cohort in Tokyo	• Every year from 1 to 9 y	• Never or infrequent wheeze • Transient early wheeze • School-age onset wheeze • Early-childhood onset remitting wheeze • Persistent wheeze	• Persistent wheeze had the highest proportion of children with a family history of allergy. • Smoke exposure during infancy was linked to a higher risk of transient early and persistent wheeze.
Kotecha et al,7 2019	Millennium Cohort Study	• 9 months and at 3, 5, 7, and 11 y	• Never or infrequent wheeze • Early wheeze • Persistent wheeze • Late wheeze	• Formal childcare was associated with early wheeze development. • Maternal atopy increased the risk of developing all trajectories, especially for persistent wheeze.
Belgrave et al,8 2013	MAAS	• 1, 3, 5, and 8 y	• No wheezing • Transient early wheeze • Late-onset wheeze • Persistent controlled wheeze • Persistent troublesome wheeze	• Children with persistent troublesome wheeze were the most vulnerable to hospitalizations, poor lung functions, and airway hyperreactivity.
Cano-Garcinuño et al,9 2014	SLAM	• 12 three-month periods between 0 and 36 mo	• Never or infrequent wheeze • Transient wheeze • Persistent wheeze • Late wheeze	• All 3 wheeze phenotypes were linked to a higher occurrence of active asthma at 6 y.
Chen et al,10 2012	CCCEH	• In-person interview: 6, 12, 24, 36, 60, 84, and 108 mo • Telephone interview: 3, 9, 15, 18, 21, 30, 48, and 72 mo	• Never or infrequent wheeze • Early transient wheeze • Early persistent wheeze • Late-onset wheeze	• Maternal asthma increased the risk of all phenotypes as compared with never/infrequent. • Male children had a higher risk of early persistent wheeze. • Early persistent wheeze was more common during cold/flu season.
Lodge et al,11 2014	MACS (high-risk children)	• Every 4 wk from birth to age 15 mo, once at age 18 mo, and yearly at age 2-7 y (23 time points)	• Never or infrequent wheeze • Early transient wheeze • Early persistent wheeze • Intermediate-onset wheeze • Late-onset wheeze	• LRTI and childcare attendance in the first year of life and low breastfeeding were risk factors for early transient wheeze. • Aeroallergen sensitization was a risk factor for early persistent wheeze. • LRTI, eczema, aeroallergen, and food sensitization were risk factors for intermediate-onset wheeze whereas exposure to dogs and having no older siblings were protective. • High exposure to parental smoke at birth and low breastfeeding were risk factors for late-onset wheeze.
Tse et al,12 2016	Project Viva	• Yearly from 1 to 9 y	• Never or infrequent wheeze • Early transient wheeze • Persistent wheeze	• Maternal asthma, early bronchiolitis, and eczema increased the risk of persistent wheeze.
Depner et al,13 2014	PASTURE	• 2, 12, 18, 24, 36, 48, 60, and 72 mo	• No or infrequent wheeze • Transient wheeze • Intermediate wheeze • Late-onset wheeze • Persistent wheeze	• Persistent wheeze was linked to asthma locus on chromosome 17q21. • Late-onset wheeze was linked to atopic sensitization at 6 y.

Abbreviations: ALSPAC, Avon Longitudinal Study of Parents and Children; BMI, body mass index; CCCEH, Columbia Center for Children's Environmental Health birth cohort study; LRTI, Lower respiratory tract infection; MACS, Melbourne Atopy Cohort Study; MAAS, Manchester Asthma and Allergy Study; PASTURE, Protection against Allergy–Study in Rural Environments; PIAMA, Prevention and Incidence of Asthma and Mite Allergy; SLAM, Sibilancias de Lactante y Asma de Mayor; STELAR, Study Team for Early Life Asthma Research.

eTable 2.

Comparison of Characteristics of Participants Included and Excluded in the Study

Risk factors	Analysis 1			Analysis 2
	Included (n = 974)	Excluded (n = 263)	P	Included (n = 431)	Excluded (n = 806)	P
Prenatal factors, n (%)
Sex			.45			.73
Female	464 (47.6)	88 (44.7)		206 (47.8)	346 (46.8)
Male	510 (52.4)	109 (55.3)		225 (52.2)	394 (53.2)
Ethnicity			.03			.001
Chinese	557 (57.3)	134 (51.0)		270 (62.7)	421 (52.3)
Indian	179 (18.4)	44 (16.7)		59 (13.7)	164 (20.4)
Malay	237 (24.4)	85 (32.3)		102 (23.7)	220 (27.3)
Maternal education level			.001			.007
Less than college	281 (29.1)	101 (39.6)		114 (26.5)	268 (34.0)
College and higher	684 (70.9)	154 (60.4)		317 (73.6)	521 (66.0)
Family history of allergy	479 (54.6)	46 (54.8)	.97	245 (56.8)	280 (52.7)	.20
Parity			.07			.31
0	438 (45.0)	98 (49.5)		185 (42.9)	351 (47.4)
1	327 (33.6)	72 (36.4)		152 (35.3)	247 (33.3)
≥ 2	209 (21.5)	28 (14.1)		94 (21.8)	143 (19.3)
Cesarean section	298 (30.6)	52 (26.3)	.22	121 (28.1)	229 (31.0)	.30
Birth weight (kg), mean (SD)	3.10 (0.44)	2.97 (0.55)	.001	3.11 (0.44)	3.06 (0.48)	.07
Tobacco exposure during pregnancy	339 (36.5)	98 (49.0)	.001	151 (35.0)	286 (41.0)	.047
Environmental factors, n (%)
Cat ownership by 1 y	32 (4.0)	0 (0.0)	.40	22 (5.1)	10 (2.5)	.046
Dog ownership by 1 y	60 (7.6)	3 (6.8)	>.99	31 (7.2)	32 (7.9)	.71
Childcare attendance by 1 y	72 (13.0)	4 (23.5)	.26	61 (14.2)	15 (10.8)	.31
Type of feeding			<.001			.002
Mainly breastfeeding	119 (12.7)	3 (3.3)		68 (15.8)	54 (9.0)
Mixed feeding	424 (45.2)	22 (24.4)		188 (43.6)	258 (43.1)
Mainly formula	396 (42.2)	65 (72.2)		175 (40.6)	286 (47.8)
Eczema in early life by 6 mo	74 (8.6)	10 (19.6)	.008	43 (10.0)	41 (8.4)	.42
Rhinitis in early life by 6 mo	233 (26.8)	23 (32.4)	.31	104 (24.1)	152 (29.8)	.051
Respiratory infections by 6 mo	51 (5.6)	3 (4.4)	>.99	24 (5.6)	30 (5.5)	.97

NOTE. Analysis 1 is a group-based trajectory model that was used to derive wheeze trajectories. Analysis 2 is a multivariable multinomial logistic regression model, which assessed the association between predictive risk factors and wheeze trajectory.

eTable 3.

Derivation of Wheeze Trajectories Using Group-Based Trajectory Modeling

Number of groups	BIC	∆BIC	2(∆BIC)
1	−2101.24	—	—
2	−1780.54	320.70	641.40
3	−1737.70	42.84	85.68
4	−1729.83	7.87	15.74
5	−1738.53	−8.70	−17.40

Abbreviation: BIC, Bayesian information criterion.

NOTE. Boldface represents the optimal number of trajectories chosen based on 2∆BIC > 10 and the clinical interpretability of the model.

eTable 4.

Number of Pneumonia Infections Experienced by Each Child in Each Wheeze Trajectory

Pneumonia infections	Total (n = 974)	No or low wheeze (n = 812)	Early-onset transient (n = 44)	P	Late-onset transient (n = 79)	P	Persistent (n = 39)	P
Pneumonia infections from 0 to 2 y				.005		.11		.13
0	684 (97.3)	569 (98.3)	30 (88.2)		55 (94.8)		30 (93.8)
1	19 (2.7)	10 (1.7)	4 (11.8)		3 (5.2)		2 (6.3)
Pneumonia infections from 2 to 4 y				.21		.22		.45
0	802 (97.8)	665 (98.2)	38 (95.0)		68 (95.8)		31 (96.9)
1	17 (2.1)	11 (1.6)	2 (5.0)		3 (4.2)		1 (3.1)
2	1 (0.1)	1 (0.2)	0 (0.0)		0 (0.0)		0 (0.0)
Pneumonia infections from 4 to 6 y				>.99		.59		.38
0	756 (98.6)	632 (98.6)	36 (100.0)		58 (98.3)		30 (96.8)
1	11 (1.4)	9 (1.4)	0 (0.0)		1 (1.7)		1 (3.2)
Pneumonia infections from 6 to 8 y				>.99		>.99		>.99
0	762 (99.2)	634 (99.1)	34 (100.0)		61 (100.0)		33 (100.0)
1	5 (0.7)	5 (0.8)	0 (0.0)		0 (0.0)		0 (0.0)
2	1 (0.1)	1 (0.2)	0 (0.0)		0 (0.0)		0 (0.0)

NOTE. Benjamini-Hochberg procedure with a 5% false discovery rate was applied.

eTable 5.

Participants With Allergic Comorbidities in Each Wheeze Trajectory

Allergic comorbidities	Total (n = 974)	No or low wheeze (n = 812)	Early-onset transient (n = 44)	P	Late-onset transient (n = 79)	P	Persistent (n = 39)	P
Allergic sensitization by 1.5 y	115 (13.9)	84 (12.1)	2 (5.3)	.30	18 (27.7)	<.001	11 (32.4)	.002
Allergic sensitization by 3 y	231 (29.6)	169 (26.2)	13 (34.2)	.28	31 (47.0)	<.001	18 (58.1)	<.001
Allergic sensitization by 5 y	351 (47.8)	270 (44.9)	18 (48.7)	.66	37 (56.9)	.07	26 (83.9)	<.001
Allergic sensitization by 8 y	515 (70.0)	411 (68.3)	27 (75.0)	.40	45 (72.6)	.49	32 (88.9)	.009
Eczema by 1.5 y	166 (21.5)	112 (17.6)	12 (32.4)	.02	24 (36.4)	<.001	18 (52.9)	<.001
Eczema by 3 y	209 (28.4)	146 (24.3)	13 (35.1)	.14	31 (47.0)	<.001	19 (59.4)	<.001
Eczema by 5 y	226 (33.5)	158 (28.8)	15 (42.9)	.08	32 (51.6)	<.001	21 (70.0)	<.001
Eczema by 8 y	266 (33.8)	192 (29.6)	17 (46.0)	.04	36 (53.7)	<.001	21 (61.8)	<.001
Rhinitis by 1.5 y	397 (49.7)	300 (45.9)	33 (84.6)	<.001	40 (58.8)	.04	24 (63.2)	.04
Rhinitis by 3 y	489 (61.1)	380 (57.8)	34 (85.0)	.001	48 (70.6)	.04	27 (77.1)	.02
Rhinitis by 5 y	509 (65.9)	399 (62.9)	34 (85.0)	.005	49 (75.4)	.046	27 (79.4)	.051
Rhinitis by 8 y	548 (64.1)	432 (61.3)	35 (81.4)	.008	54 (77.1)	.009	27 (73.0)	.15
Atopic eczema by 1.5 y	40 (5.7)	24 (4.1)	1 (2.9)	>.99	6 (10.5)	.03	9 (30.0)	<.001
Atopic eczema by 3 y	70 (10.4)	40 (7.3)	6 (16.7)	.04	11 (18.6)	.003	13 (46.4)	<.001
Atopic eczema by 5 y	93 (15.8)	56 (11.7)	8 (23.5)	.04	15 (28.3)	.001	14 (58.3)	<.001
Atopic eczema by 8 y	143 (20.0)	101 (17.0)	7 (21.9)	.48	17 (29.8)	.02	18 (56.3)	<.001
Atopic rhinitis by 1.5 y	61 (8.6)	42 (7.1)	2 (5.9)	>.99	9 (16.1)	.02	8 (23.5)	.001
Atopic rhinitis by 3 y	108 (15.2)	71 (12.1)	8 (21.1)	.11	18 (31.0)	<.001	11 (36.7)	<.001
Atopic rhinitis by 5 y	144 (22.2)	100 (18.8)	13 (37.1)	.008	18 (34.0)	.008	13 (48.2)	<.001
Atopic rhinitis by 8 y	236 (32.8)	183 (30.6)	15 (45.5)	.07	21 (38.2)	.25	17 (51.5)	.01
Eczema with steroids use by 1.5 y	92 (12.7)	61 (10.1)	7 (20.0)	.07	13 (22.0)	.006	11 (36.7)	<.001
Eczema with steroids use by 3 y	112 (16.7)	76 (13.7)	7 (20.6)	.27	16 (29.1)	.002	13 (46.4)	<.001
Eczema with steroids use by 5 y	120 (20.2)	82 (16.7)	7 (23.3)	.35	18 (36.0)	.001	13 (52.0)	<.001
Eczema with steroids use by 8 y	161 (21.4)	114 (18.4)	10 (27.8)	.16	22 (34.4)	.002	15 (45.5)	<.001
Sensitization to food by 1.5 y	38 (4.6)	27 (3.9)	1 (2.6)	>.99	4 (6.2)	.33	6 (17.7)	.003
Sensitization to food by 3 y	45 (5.9)	33 (5.2)	2 (5.4)	>.99	4 (6.5)	.56	6 (20.0)	.006
Sensitization to food by 5 y	74 (11.1)	55 (10.0)	4 (11.4)	.77	8 (14.8)	.27	7 (28.0)	.01
Sensitization to food by 8 y	117 (19.6)	80 (16.2)	9 (28.1)	.08	15 (31.9)	.007	13 (54.2)	<.001
Sensitization to dust mite by 1.5 y	95 (11.4)	70 (10.1)	1 (2.6)	.16	16 (24.2)	.001	8 (23.5)	.02
Sensitization to dust mite by 3 y	212 (27.1)	155 (23.9)	12 (31.6)	.29	29 (43.9)	<.001	16 (53.3)	<.001
Sensitization to dust mite by 5 y	334 (45.6)	255 (42.5)	18 (48.7)	.46	36 (55.4)	.047	25 (83.3)	<.001
Sensitization to dust mite by 8 y	502 (68.8)	400 (66.9)	27 (75.0)	.31	43 (71.7)	.45	32 (88.9)	.006

NOTE. Sample used to study the associations was smaller because of missing values. Benjamini-Hochberg procedure with a 5% false discovery rate was applied.

eTable 6.

Associations Between Wheeze Trajectory and Allergic Comorbidities

Allergic comorbidities		No or low		Early-onset		Late-onset
	n	ARR (95% CI)a	P	ARR (95% CI)a	P	ARR (95% CI)a	P
Allergic sensitization by 1.5 y	760	0.41 (0.24-0.73)	.002	0.09 (0.01-0.65)	.02	0.94 (0.50-1.78)	.86
Allergic sensitization by 3 y	705	0.49 (0.33-0.72)	<.001	0.64 (0.35-1.17)	.15	0.95 (0.61-1.46)	.80
Allergic sensitization by 5 y	660	0.55 (0.44-0.69)	<.001	0.63 (0.43-0.93)	.02	0.73 (0.56-0.97)	.03
Allergic sensitization by 8 y	660	0.79 (0.68-0.92)	.003	0.92 (0.74-1.15)	.47	0.88 (0.73-1.07)	.21
Eczema by 1.5 y	709	0.41 (0.28-0.60)	<.001	0.71 (0.39-1.28)	.26	0.75 (0.47-1.20)	.23
Eczema by 3 y	672	0.48 (0.33-0.68)	<.001	0.68 (0.39-1.18)	.17	0.84 (0.55-1.27)	.41
Eczema by 5 y	612	0.47 (0.35-0.63)	<.001	0.62 (0.37-1.03)	.06	0.77 (0.54-1.09)	.14
Eczema by 8 y	713	0.56 (0.41-0.76)	<.001	0.79 (0.49-1.27)	.34	0.94 (0.66-1.35)	.74
Rhinitis by 1.5 y	729	0.79 (0.60-1.03)	.08	1.22 (0.91-1.63)	.18	0.95 (0.68-1.31)	.74
Rhinitis by 3 y	723	0.80 (0.65-0.98)	.03	1.07 (0.84-1.36)	.57	0.93 (0.73-1.19)	.56
Rhinitis by 5 y	698	0.84 (0.69-1.01)	.07	1.04 (0.82-1.31)	.75	0.96 (0.76-1.21)	.73
Rhinitis by 8 y	766	0.85 (0.69-1.04)	.12	1.08 (0.84-1.39)	.53	1.02 (0.80-1.29)	.87
Atopic eczema by 1.5 y	645	0.18 (0.08-0.39)	<.001	—	—	0.42 (0.16-1.09)	.08
Atopic eczema by 3 y	610	0.18 (0.10-0.34)	<.001	0.38 (0.14-0.99)	.049	0.48 (0.24-0.97)	.04
Atopic eczema by 5 y	533	0.22 (0.14-0.35)	<.001	0.43 (0.20-0.90)	.03	0.51 (0.29-0.87)	.01
Atopic eczema by 8 y	651	0.36 (0.24-0.54)	<.001	0.53 (0.26-1.10)	.09	0.63 (0.38-1.06)	.08
Atopic rhinitis by 1.5 y	653	0.39 (0.19-0.80)	.01	0.13 (0.02-0.99)	.049	0.85 (0.35-2.05)	.72
Atopic rhinitis by 3 y	642	0.41 (0.23-0.71)	.002	0.56 (0.23-1.36)	.20	1.05 (0.57-1.94)	.87
Atopic rhinitis by 5 y	585	0.52 (0.34-0.80)	.003	0.81 (0.44-1.48)	.48	0.85 (0.52-1.40)	.52
Atopic rhinitis by 8 y	649	0.70 (0.50-0.97)	.04	1.05 (0.66-1.66)	.85	0.83 (0.53-1.29)	.40
Eczema with steroids use by 1.5 y	667	0.28 (0.16-0.47)	<.001	0.50 (0.21-1.18)	.11	0.58 (0.30-1.14)	.11
Eczema with steroids use by 3 y	617	0.36 (0.22-0.58)	<.001	0.48 (0.21-1.11)	.09	0.70 (0.38-1.27)	.24
Eczema with steroids use by 5 y	544	0.37 (0.24-0.57)	<.001	0.49 (0.22-1.07)	.07	0.70 (0.41-1.20)	.20
Eczema with steroids use by 8 y	686	0.47 (0.31-0.71)	<.001	0.64 (0.32-1.27)	.21	0.80 (0.48-1.32)	.38

Abbreviations: ARR, adjusted relative risk; CI, confidence interval.

NOTE. Poisson regression with robust error variance was implemented with adjustment for sex, ethnicity, maternal education level, and family history of allergy for each allergic comorbidity.

^aReference: Persistent wheeze trajectory.

eTable 7.

Associations Between Wheeze Trajectory and Food and Dust Mite Sensitization

Allergic sensitisation	n	Early-onset		Late-onset		Persistent
		ARR (95% CI)a	P	ARR (95% CI)a	P	ARR (95% CI)a	P
Sensitization to food by 1.5 y	760	0.64 (0.09-4.39)	.65	1.68 (0.62-4.56)	.31	4.20 (1.72-10.25)	.002
Sensitization to food by 3 y	693	0.53 (0.08-3.65)	.52	1.43 (0.53-3.87)	.48	4.02 (1.65-9.78)	.002
Sensitization to food by 5 y	605	0.95 (0.32-2.80)	.93	1.69 (0.85-3.35)	.14	2.85 (1.37-5.94)	.005
Sensitization to food by 8 y	545	1.70 (0.93-3.08)	.08	2.15 (1.35-3.42)	.001	3.20 (2.04-5.02)	<.001
Sensitization to dust mite by 1.5 y	762	—	—	2.39 (1.51-3.79)	<.001	2.30 (1.20-4.42)	.01
Sensitization to dust mite by 3 y	707	1.33 (0.77-2.28)	.30	1.97 (1.45-2.68)	<.001	2.07 (1.36-3.15)	.001
Sensitization to dust mite by 5 y	659	1.23 (0.86-1.74)	.26	1.37 (1.09-1.71)	.007	1.91 (1.52-2.40)	<.001
Sensitization to dust mite by 8 y	655	1.18 (0.98-1.42)	.09	1.11 (0.95-1.30)	.20	1.27 (1.09-1.47)	.002

Abbreviations: ARR, adjusted relative risk; CI, confidence interval.

NOTE. Poisson regression with robust error variance was implemented with adjustment for sex, ethnicity, maternal education level, and family history of allergy for each allergic comorbidity.

^aReference: No or low wheeze trajectory.

eTable 8.

Number of Allergic Sensitization Trajectories in Each Wheeze Trajectory

Allergic sensitisation	Total (n = 974)	No or low wheeze (n = 812)	Early-onset (n = 44)	Late-onset (n = 79)	Persistent (n = 39)	P
Allergic sensitization trajectory						.03
No or low sensitization	558 (58.6)	494 (62.3)	20 (46.5)	36 (46.2)	8 (21.1)
Late mite sensitization	237 (24.9)	191 (24.1)	13 (30.2)	18 (23.1)	15 (39.5)
Early food and mite sensitization	157 (16.5)	108 (13.6)	10 (23.3)	24 (30.8)	15 (39.5)

NOTE. P value is for the comparison between late-onset and persistent trajectories.