Sex-specific developmental trajectories of eczema from infancy to age 26 years: a birth cohort study

Abstract

Background:

Eczema is a common inflammatory skin disease with varying developmental trajectories/patterns that are influenced by different risk factors. The aim of this study was to investigate eczema development from infancy to early adulthood by identifying distinct developmental trajectories that describe disease patterns over time and evaluate the role of prenatal and early life risk factors.

Methods:

The Isle of Wight Birth Cohort (n=1,456) was prospectively assessed at birth, 1, 2, 4, 10, 18, and 26 years. In all assessments, eczema was defined as chronic or chronically relapsing itchy dermatitis lasting >6 weeks with characteristic morphology and distribution in the past 12 months. Developmental trajectories of eczema between 1-or-2 and 26 years were identified separately for males and females by applying semiparametric mixture models. Associations were assessed by applying a modified Poisson regression to estimate adjusted risk ratios (aRR) and 95% confidence intervals (CI).

Results:

In both males and females, the following eczema developmental trajectories were identified: unaffected/transient (males: 77.7% vs. females: 73.0%), mid-onset late-resolving (males: 7.8% vs. females: 4.4%), late-onset (males: 5.2% vs. females: 9.5%), and early-onset persistent (males: 9.3% vs. females: 5.4%). In females, an additional trajectory was identified: early-onset early-resolving (7.7%). Among males, filaggrin gene (FLG) variants (aRR = 2.45, 95% CI: 1.34-4.46) and paternal eczema (2.66, 1.39-5.08) were associated with the early-onset persistent trajectory. Among females, maternal eczema (2.84, 1.42-5.70) and high birthweight (2.25, 1.08-4.69) were associated with the early-onset persistent trajectory.

Conclusions:

Four and five trajectories represented eczema development among males and females, respectively, with different predisposing risk factors. Our results indicate that males and females may experience a different course of eczema.

Graphical Abstract

Males and females follow different developmental trajectories of eczema from infancy to 26 years of age, with more males than females following an early-onset persistent trajectory and a mid-onset late-resolving trajectory. In contrast, more females than males followed a late-onset trajectory and only females were in the early-onset early-resolving trajectory. Moreover, males and females have distinct risk factors that influence the developmental trajectories of eczema. Hence, it is important to considered sex-stratified clinical management of the disease.

INTRODUCTION

Eczema, also known as atopic dermatitis and atopic eczema, is a common and chronic inflammatory skin disease that often follows a relapsing-remitting course. It has been estimated that up to 20% of children and 10% of adults are affected by eczema.^1,2 The disorder is characterized by a heterogeneous clinical presentation that includes erythematous, scaly, pruritic, and acute oozing and crusting lesions.^3,4 The pathophysiology of eczema is multifactorial, with immune system dysregulation and epidermal barrier dysfunction being the major hallmarks of the disease.^5,6 Although eczema is not life-threatening, it poses a significant impact on healthcare resources and patients’ and their relatives’ overall health, including quality of life and psychosocial wellbeing.^7,8

There is a consensus in the scientific literature that most cases of eczema develop during early childhood^7,9,10; however, the long-term natural history of the disease is less understood. It has been widely accepted that most early childhood cases resolve later in childhood, with few cases developing in adolescence and adulthood, but this view has been challenged as adulthood onset is not rare, and some childhood cases may relapse later in life (relapsing-remitting nature) or even follow a persistent course of the disease.^11–13 A meta-analysis estimated adult-onset eczema to be 26.1% (95% CI: 16.5-37.2%), i.e., around 1 in 4 adults with eczema report adulthood-onset of their disease.¹⁴ Hence, this heterogeneity indicates that different trajectories of eczema development exist that may be influenced by different determinants. One strong risk factor that has been related to eczema development, and a more persistent and severe phenotype is carrying loss-of-function (LOF) variants in the filaggrin gene (FLG), which encodes an important structural protein involved in the epidermal barrier formation and integrity.^15–19

Several birth cohorts have investigated the course of eczema during the first two decades of life, and in a few instances up to early adulthood, as synthesized in a prior systematic review, which concluded that the prevalence of eczema was similar before and after childhood, with some observed heterogeneity between studies.²⁰ Of the five studies reporting eczema prevalence beyond 20 years of age, three studies reported that eczema prevalence is lowest in early adulthood,^21–23 whereas two studies showed that eczema prevalence was higher in early adulthood^13,24 as compared to assessments during adolescence. Few studies have investigated the longitudinal trajectories of eczema development, where distinct developmental trajectories, such as early-onset persistent, early transient, and late-onset, and their determinants have been identified.^25–27 These studies have investigated eczema development up to 6 years,²⁵ 10 years,²⁷ and 16 years²⁶ and hence, there is a need to investigate eczema trajectories beyond adolescence. Another study has incorporated eczema severity phenotypes while investigating eczema development up to age 14 years.²⁸

The participants of the Isle of Wight Birth Cohort (IOWBC) study have been seen on six occasions over the course of 26 years, at 1, 2, 4, 10, 18, and 26 years, with eczema being assessed at each follow-up.²⁹ Using data from the IOWBC, this study aims to investigate the natural course of eczema development from infancy to early adulthood in terms of identifying distinct developmental trajectories that describe disease patterns over time. Given that male and female subjects experience different course of the disease, with eczema prevalence being equal or slightly higher among males during childhood and becoming preponderant among females after puberty,^{23,24,30–32} our analysis was stratified by sex. In addition, effects of prenatal and early life risk factors on eczema trajectories were assessed.

METHODS

Study design and participants

The IOWBC is a whole population-based prospective study that recruited all children (n = 1,536, ~98% Caucasian) born between January 1^st, 1989 and February 28^th, 1990 on the Isle of Wight, UK, to study the natural history and etiology of allergic and respiratory conditions.^29,33 Parental informed consent was obtained to enroll 1,456 newborns to participate in the ensuing follow-ups that occurred at the ages of 1, 2, 4, 10, 18, and 26 years, with >80% retention at all assessments up to 18 years and 71% retention at age 26 years.^29,34 In the current analysis, participants who had three or more valid eczema status information (i.e., not missing) across the follow-ups were included; hence, 1,381 (94.8%) participants satisfied our inclusion criteria. All participants provided informed consent and ethical approval was obtained from the local/national Research Ethics Committee at recruitment and at each assessment. At age 26, ethical approval was granted by National Research Ethics Committee, West Midlands (15/WM/0071). Phenotypic and environmental information were collected using questionnaires and a wide range of clinical assessments and measurements were carried out as detailed by Arshad et al.²⁹

Eczema ascertainment

In all assessments of the IOWBC, investigators determined the presence of eczema based on symptoms over the last 12 months, which was defined as chronic or chronically relapsing itchy dermatitis lasting more than 6 weeks with characteristic morphology and distribution,^29,35 according to the criteria of Hanifin and Rajka.³⁶ Since the 1- and 2-year follow-up data on eczema were collected in a relatively narrow time window, we have combined them for analytic purposes and reported them hereafter as 1-or-2 years.

Prenatal and early life variables

In the current report we have assessed the effect of prenatal and early life risk factors on the development of eczema, including FLG LOF variants, maternal and paternal history of eczema, birthweight (low/normal <4,000 grams; high ≥4,000 grams; the low (<2,500 grams) birthweight group was analyzed with the normal birthweight group due to a small proportion (3.5%, 48/1360) of participants were born with low birthweight), birth order, duration of breastfeeding reported in weeks, and whether the child’s family owned a cat and/or a dog at the time of the child’s birth. The underweight (BMI <18.5) group was analyzed with the normal group due to a small proportion (1.8%) of mothers being underweight. A total of 1150 participants were genotyped for FLG variants R501X, 2282del4, S3247X, 3702delG, and R2447X, with the latter two variants were not informative in our study population due to minor allele frequencies <0.1%. Individuals carrying the minor allele for at least one of the FLG variants R501X, 2282del4, or S3247X were classified as carrying FLG LOF variants. FLG variants genotyping was performed using the GoldenGate Genotyping Assays (Illumina, Inc, SanDiego, CA) on the BeadXpressVeracode platform (Illumina, Inc, SanDiego, CA) per Illumina’s protocol, detailed information is provided by Ziyab et al.³⁷

Measures of disease occurrence

To investigate the natural course of eczema across the five analytical follow-ups (1-or-2, 4, 10, 18, 26 years), different measures of disease occurrence were estimated. The 12-month period prevalence was calculated at each follow-up as the number of existing cases divided by the total number of subjects participating in the respective follow-up. Incidence, to measure newly developed eczema, was defined at the 10, 18, and 26 years follow-ups as the number of new cases at the respective follow-up divided by the total cohort at risk (i.e., disease-free in all preceding follow-ups).³⁸ Remission, to estimate the proportion of cases that outgrew eczema, was defined at the 10, 18, and 26 years follow-ups as the number of cases that outgrew eczema at the respective follow-up divided by the cohort who had eczema in preceding follow-ups.³⁸ Moreover, given the relapsing-remitting course of eczema, disease transition measures were estimated to describe the change in a person’s disease status between two consecutive assessments, regardless of whether it was the first such occurrence or whether prior data were missing.^30,38,39 Hence, four transition periods were considered in our analysis: 1-or-2 to 4 years, 4 to 10 years, 10 to 18 years, and 18 to 26 years. Positive transition (change from disease-free to diseased) was defined as the number of eczema cases at the end of the transition period divided by the number of subjects that were disease-free at the beginning of the transition period. Negative transition (change from diseased to disease-free) was defined as the number of eczema-free subjects at the end of the transition period divided by the number of subjects with eczema at the beginning of the transition period. It is worth noting that positive transition may capture some incident cases and negative transition may capture some remitting cases (i.e., some overlap may exist).

Trajectory analysis

Developmental trajectories (i.e., distinct subgroups within the total sample that follow similar disease patterns over time) of eczema across ages 1-or-2, 4, 10, 18, and 26 years were identified separately for males and females by applying semiparametric mixture models using PROC TRAJ macro in SAS 9.4 (SAS Institute, Cary, North Carolina, USA).^40,41 See the Methods section in this article’s Supporting Information for more details.

Statistical analysis

All statistical analyses were conducted using SAS 9.4. The statistical significance level was set to α=0.05. Descriptive analyses were conducted to calculate frequencies and proportions of categorical variables and median and interquartile range (IQR) were used to describe continuous variables. Chi-square (χ²) tests were used to assess differences in proportions across sex. To control for false positive results, we applied the false discovery rate (FDR) method to estimate adjusted p-values for the sex comparisons.⁴² Adjusted associations were assessed by applying a modified Poisson regression with robust variance estimation to estimate and infer risk ratios (RR) and their 95% confidence intervals (CI).⁴³ Adjusted associations between prenatal and early life risk factors (predictors) and eczema trajectories (outcome variable) were assessed separately for males and females. The “unaffected/transient” trajectory was set as the reference group and a series of three and four regression models were evaluated among males and females, respectively, to assess the associations.

RESULTS

Description of the study population

In total, information on eczema status was available for 1,377 (700 males), 1,214 (619 males), 1,359 (688 males), 1,307 (651 males), and 1,029 (469 males) study participants at ages 1-or-2, 4, 10, 18, and 26 years, respectively. The total enrolled study sample and the analytical study sample were similar with respect to all characteristics under study (Table 1). The combined proportion of carrying the minor (loss-of-function) allele(s) of FLG variants R501X, 2282del4, and/or S3247X was 10.3% (116/1,129) in the analytical study sample. No individuals were homozygous for the minor allele of any FLG variants. One person had a compound heterozygous genotype for R501X and 2282del4, and this person was analyzed with the heterozygotes.

Table 1.

Characteristics of the total enrolled study sample and the analytical study sample

Characteristic	Enrolled study sample	Analytical study sample*
Sex, % (n/total)
Male	51.2 (786/1,536)	50.4 (696/1,381)
Female	48.8 (750/1,536)	49.6 (685/1,381)
FLG genotype, % (n/total)
Loss-of-function	10.3 (118/1,150)	10.3 (116/1,129)
Maternal eczema, % (n/total)
Yes	12.1 (147/1,213)	12.1 (144/1,191)
Paternal eczema, % (n/total)
Yes	6.4 (78/1,213)	6.5 (77/1,191)
Birthweight, % (n/total)
Low/normal†	88.0 (1,329/1,511)	87.6 (1,192/1,360)
High‡	12.0 (182/1,511)	12.4 (168/1,360)
Birth order, % (n/total)
First	42.1 (510/1,212)	42.0 (500/1,189)
Second	34.6 (419/1,212)	34.5 (410/1,189)
Third or more	23.3 (283/1,212)	23.5 (279/1,189)
Household cat at birth, % (n/total)
Yes	32.4 (492/1,516)	32.6 (448/1,374)
Household dog at birth, % (n/total)
Yes	29.0 (440/1,516)	28.8 (396/1,374)
Breastfeeding duration, Median (IQR)
Weeks	8.0 (1.0-26.0)	9.0 (2.0-28.0)

IQR: interquartile range.

^*Includes participants who have three or more valid eczema status information (i.e., not missing) across the five analytical follow-ups.

^†Low/normal birthweight: <4000 grams.

^‡High birthweight: ≥4000 grams

Measures of disease occurrence

Estimates of the 12-month period prevalence, positive transition, incidence, negative transition, and remission of eczema from 1-or-2 to 26 years of age are shown in Table 2. The 12-month prevalence of eczema was highest at 1-or-2 years (14.2%) and lowest at 26 years (10.2%). Among males, eczema was most prevalent at 1-or-2 years (14.6%) and lowest at 18 years (8.3% at 18 years) and 26 years (8.5%). Among females, eczema was most prevalent at 18 years (16.3%) and lowest at 26 years (11.6%; Table 2, Figure S1). Only minor differences in the prevalence were detected between males and females up to 10 years of age; however, eczema was statistically significantly more common among females compared to males at age 18 years (16.3% vs. 8.3%, p <0.001, FDR adjusted p <0.001) and slightly higher at 26 years (11.6% vs. 8.5%, p = 0.104, FDR adjusted p = 0.395; Table 2).

Table 2.

Prevalence (12-month), positive transition, incidence, negative transition, and remission of eczema from infancy to 26 years of age in the total study sample and according to sex

	Age at follow-up (years), % (n/total)
	1-or-2	4	10	18	26
Prevalence
Overall	14.2 (196/1,377)	11.9 (145/1,214)	13.7 (186/1,359)	12.3 (161/1,307)	10.2 (105/1,029)
Male	14.6 (102/700)	12.6 (78/619)	13.4 (92/688)	8.3 (54/651)	8.5 (40/469)
Female	13.9 (94/677)	11.3 (67/595)	14.0 (94/671)	16.3 (107/656)	11.6 (65/560)
P-value†	0.715	0.472	0.733	< 0.001	0.104
Positive transition
Overall		5.9 (57/962)	8.6 (88/1,025)	6.9 (73/1,063)	4.9 (42/850)
Male		6.4 (31/488)	8.3 (43/518)	4.3 (23/532)	4.0 (16/404)
Female		5.5 (26/474)	8.9 (45/507)	9.4 (50/531)	5.8 (26/446)
P-value†		0.569	0.743	0.001	0.209
Incidence
Overall			6.9 (60/865)	5.7 (42/743)	3.0 (16/539)
Male			6.7 (29/435)	3.0 (11/365)	1.6 (4/249)
Female			7.2 (31/430)	8.2 (31/378)	4.1 (12/290)
P-value†			0.754	0.002	0.125
Negative transition
Overall		56.9 (103/181)	43.8 (56/128)	57.5 (96/167)	58.1 (72/124)
Male		55.4 (51/92)	45.8 (33/72)	65.4 (53/81)	51.3 (20/39)
Female		58.4 (52/89)	41.1 (23/56)	50.0 (43/86)	61.2 (52/85)
P-value†		0.685	0.590	0.044	0.300
Remission
Overall			31.4 (22/70)	37.2 (16/43)	33.3 (7/21)
Male			34.2 (13/38)	39.1 (9/23)	20.0 (2/10)
Female			28.1 (9/32)	35.0 (7/20)	45.5 (5/11)
P-value†			0.585	0.780	0.362

^†Calculated using chi-square (χ²) tests to compare proportions across sex. If any cell count was less than 5 observations, Fisher’s exact test was used.

Positive transition (i.e., change in eczema status from ‘no eczema’ to ‘yes eczema’ in two consecutive follow-ups) was highest in males between 4 and 10 years (8.3%) and lowest between 18 and 26 years (4.0%; Table 2). Among females, positive transition was highest between 10 and 18 years (9.4%) and lowest between 1-or-2 and 4 years (5.5%) and 18 and 26 years (5.8%). Positive transition between 10 and 18 years was statistically significantly higher in females compared to males (9.4% vs. 4.3%, p = 0.001, FDR adjusted p = 0.009). Incidence (i.e., the first occurrence of eczema) of eczema decreased over the ages from 6.9% at age 10 to 3.0% at age 26 years. Similar to the finding of positive transition measure, incidence at age 18 years was higher in females than males (8.2% vs. 3.0%, p = 0.002, FDR adjusted p = 0.013), a trend, which partially continued at age 26 years (Table 2).

Negative transition (i.e., change in eczema status from ‘yes eczema’ to ‘no eczema’ in two consecutive follow-ups) was highest in males between 10 and 18 years (65.4%) and highest in females between 18 and 26 years (61.2%; Table 2). On the base of negative transition, males were more likely than females to experience negative transition of eczema between 10 and 18 years (65.4% vs. 50.0%, p = 0.044, FDR adjusted p = 0.187), with no sex differences detected at other negative transition periods. Remission (i.e., outgrowing eczema) was highest in males at age 18 years (39.1%) and highest in females at age 26 years (45.5%; Table 2) without showing statistically significant differences between the sexes.

Developmental trajectories of eczema

Trajectories of eczema development across the first 26 years of age were determined separately for male and female subjects. The optimum number of trajectories was determined based on BIC as shown in Table S1 in this article’s Supporting Information. In males, the development of eczema from infancy to 26 years of age was best represented by four developmental trajectories, labeled as: (1) unaffected/transient (77.7%, n = 541), (2) mid-onset late-resolving (7.8%, n = 54), (3) late-onset (5.2%, n = 36), and (4) early-onset persistent (9.3%, n = 65; Figure 1A). In females, the development of eczema across the first 26 years of life was best represented by five developmental trajectories, labeled as: (1) unaffected/transient (73.0%, n = 500), (2) early-onset early-resolving (7.7%, n = 53), (3) mid-onset late-resolving (4.4%, n = 30), (4) late-onset (9.5%, n = 65), and (5) early-onset persistent (5.4%, n = 37; Figure 1B). The early-onset early-resolving trajectory was identified in females but not in males. Overall, similar proportions made-up the unaffected/transient trajectories in males and females (77.7% vs. 73.0%), but the distribution of the other trajectories were different across sex. For instance, more males than females were in the mid-onset late-resolving trajectory (7.8% vs. 4.4%, p = 0.009) and in the early-onset persistent trajectory (9.3% vs. 5.4%, p = 0.005). Whereas females were more often than males to be in the late-onset trajectory (9.5% vs 5.2%, p = 0.002), which also reflects their higher incidence/positive transition of eczema at age 18 years and the lower negative transition compared to males (Table 2).

Figure 1.

Trajectories of eczema from 1-or-2 to 26 years of age. A) The four trajectories represent the latent growth patterns of eczema from 1-or-2 to 26 years of age in males (n = 696). B) The five trajectories characterize the latent growth patterns of eczema from 1-or-2 to 26 years of age in females (n = 685). The y-axis presents the predicted probability of having eczema among the participants of each trajectory at the different time points.

Prenatal and early life risk factors and trajectories of eczema

Associations between prenatal and early life risk factors and the identified developmental trajectories of eczema were assessed separately for males (Figure 2A, Table S2) and females (Figure 2B, Table S3). Among males, carrying FLG LOF variants was associated with increased risk for being in the early-onset persistent trajectory (aRR = 2.45, 95% CI: 1.34-4.46), so was paternal eczema (2.66, 1.39-5.08). Moreover, FLG LOF showed positive associations with mid-onset late-resolving (1.84, 0.84-4.03) and late-onset 1.97 (0.79-4.94) trajectories, though these associations did not gain statistical significance. Cat-keeping at child’s birth showed an association with the late-onset trajectory (2.79, 1.24-6.27) among male participants (Figure 2A). Among females, carrying FLG LOF variants was not statistically significantly associated with eczema trajectories (Figure 2B); however, indication for possible association was present with mid-onset late-resolving (1.47, 0.54-3.96) and early-onset persistent (1.58, 0.67-3.75) trajectories. Maternal eczema (2.84, 1.42-5.70) and high birthweight (2.25, 1.08-4.69, Figure 2B) were associated with the early-onset persistent trajectory among females.

Figure 2.

Forest plots showing associations between risk factors and trajectories of eczema among males (A) and females (B). ^# Low/normal birthweight: <4000 grams, high birthweight: ≥4000 grams; *Adjusted for all variables shown in the figure. LOF: loss-of-function; FLG: filaggrin gene.

DISCUSSION

The current study described the natural course and developmental trajectories of eczema from infancy to early adulthood and examined factors that are associated with eczema development. The developmental trajectories of eczema were best described by four trajectories among males and five trajectories among females. In general, similar developmental patterns of eczema were observed in males and females, with two exceptions. First, an “early-onset early-resolving” trajectory that was only observed among females. Second, more males than females experienced an “early-onset persistent” and “mid-onset late-resolving” developmental trajectories, whereas females were more often than males to be in the “late-onset trajectory.” In terms of risk factors for eczema trajectories, among males, FLG LOF variants and paternal eczema were associated with increased risk for being in the “early-onset persistent” trajectory. In contrast, among females, maternal eczema and high birthweight were associated with higher risk for being in the “early-onset persistent” trajectory. Such results indicate that males and females may experience a different course of eczema and also sex-specific risk factors.

We have estimated different measures of disease occurrence to better understand the natural course of eczema. The 12-month prevalence captured a decreasing trend in the course of eczema in both males and females, with more females than males being affected at ages 18 and 26 years. Such an observation has been reported in the German birth cohort (the ‘Multicenter Allergy Study’ [MAS]) that showed higher eczema prevalence in females than males at ages 15 and 20 years with an overall decreasing trend.⁴⁴ This female predominance in eczema prevalence that starts in adolescence and persists into adulthood has been described by several prior studies.^32,45,46 In a previous investigation, we reported that more females than males developed eczema between ages 10 and 18 years (positive transition), and more males than females outgrew eczema between age 10 and 18 years (negative transition), which can explain the higher prevalence among adolescent females than males.³⁰ Herein, we showed that between ages 18 and 26 years eczema prevalence did not change among males (8.3% vs. 8.5%), but it decreased among females (16.3% vs. 11.6%). This observation can be explained by the fact that more females than males experienced negative transition and remission at 26 years of age than males. Moreover, newly developed eczema (incident cases) was more common among females than males at ages 18 and 26 years; indicating that females are more likely to experience adolescence- and adult-onset eczema than males. This observation contradicts prior findings that showed adult-onset eczema to be similar among the sexes^14,47 and support the idea that it is more common among females than males.³¹ Moreover, we estimated the incidence of eczema to be 3.0% at age 26 years, which is much less than a previously estimated adult-onset eczema of 26.1%.¹⁴

Although few prior studies have investigated the longitudinal development of eczema^25–27 (summarized in Table S4 in this article’s Supporting Information), our study is the first to extend such trajectories beyond adolescence, to age 26 years. Overall, our identified trajectories were similar in patterns to previously identified trajectories (see Discussion section in this article’s Supporting Information for more details about the prior studies). None of the studies described in this article’s “Supporting Information Discussion section” have considered analyzing the developmental trajectories separately for males and females. Our analysis identified four eczema developmental trajectories among males and five trajectories among females. Although the developmental patterns of four trajectories were similar among both sexes, we identified an ‘early-onset early-resolving’ trajectory only among females. Nonetheless, the distribution of the common trajectories differed according to sex, with more males than females following an early-onset persistent trajectory (9.3% vs. 5.4%) and a mid-onset late-resolving trajectory (7.8% vs. 4.4%). In contrast, more females than males were in the late-onset trajectory (9.5% vs. 5.2%), which may contribute to the observed higher eczema prevalence at ages 18 and 26 among females compared to males (Table 2). It has been suggested previously that sex hormones may contributed to the observed sex dimorphism, with estrogens inducing pro-inflammatory responses and testosterone being immunosuppressive.^48,49 Moreover, we identified different risk factors for male and female trajectories. For example, FLG LOF variants were associated with increased risk for being in the early-onset persistent trajectory among males but such an effect was moderately present among females, which did not gain statistical significance due, possibly, to limited statistical power; this potential sex-specific effect of FLG LOF variants on eczema development has not, up to our knowledge, been reported previously. Such findings indicate that genetic risk factors may influence the long-term development of eczema in a sex-specific manner. Moreover, paternal eczema was associated with early-onset persistent trajectory among males, whereas maternal eczema was associated with early-onset persistent trajectory among females. Hence, indicating sex-dependent effects of parental history of allergy on offspring eczema development. Such a mother-to-daughter and father-to-son effects were reported previously for eczema and asthma in this cohort.⁵⁰ We also observed that high birthweight to be associated with early-onset persistent eczema among females only. Prior studies have shown that elevated birthweight is associated with eczema development in infancy.^51,52 Among males, our analysis identified that cat-keeping at birth was associated with increased risk for late-onset eczema trajectory; this observation has been reported previously.^53–55

Comparing different studies on eczema trajectories (see Table S4 in this article’s Supporting Information) suggests that about 10-40% of the eczema cases that develop early in life remain persistent (Generation R: 2%/24% = 8.3%; IOWBC males: 9.3%/22.3% = 41.7%); 60-90% are resolving in late childhood, adolescence, or early adulthood. It is likely that a family history of eczema/allergy or FLG LOF variants can help to distinguish children who have a higher risk of a persisting disease. However, future studies are needed to corroborate these findings. Moreover, the current IOW analysis and the prior investigations have consistently identified a “late-onset” trajectory, but potential risk factors for this trajectory remain to be identified. When comparing different studies (see Table S4 in this article’s Supporting Information) it is obvious, that long-term cohort studies on eczema mainly originated in Europe. Such research and data is deficient in the United States and other parts of the world.

The prospective long-term follow-ups that covered childhood, adolescence, and early adulthood and the high retention are major strengths to our study. In addition, analyzing males and females separately provided novel insights into sex-specific inferences. Nonetheless, splitting the sample by sex might have magnified the influence of random variations on the assignment of males and females into the developmental trajectories. The trajectories we have identified are similar to those reported by previous investigation^25–27; hence, further strengthening the consistency of our results. Lack of statistical power due to limited subgroup sample sizes was evident in few instances. For example, among males, FLG LOF variants non-statistically significantly increased the risk for “mid-onset late-resolving” and “late-onset” eczema trajectories. A similar pattern of non-statistically significant effects of FLG LOF variants were seen among females, which can be explained by the limited statistical power among the female subsample. Hence, such findings should not be overlooked solely due to the lack of statistical significance. We have used the Hanifin and Rajka³⁶ criteria to define eczema, which has shown high sensitivity (87.9-96.0%) and specificity (77.6-93.8%) as compared with clinical diagnosis as reference standard.⁵⁶ Nonetheless, a limitation of our eczema definition is not being age-adaptive to capture the evolvement of eczema over age. Moreover, the large gaps between the follow-ups is a further limitation to our estimated measures of disease occurrence, as such measures might miss cases that develop and resolve in-between the follow-ups.

In conclusion, we have described the natural course and developmental trajectories of eczema from infancy to early adulthood. Our results showed that males and females may experience different eczema course and different sets of risk factors. Specifically, FLG LOF variants and paternal eczema were associated with a higher risk of an early-onset persistent trajectory among males, whereas maternal eczema and high birthweight were associated with early-onset persistent trajectory among females. Hence, investigations related to the etiology and long-term development of eczema should be sex-stratified to facilitate stratified medicine efforts. Finally, extension of the prior cohorts and their analyses into adulthood and in non-European countries are needed.

Key Messages.

• Males and females experience different developmental trajectories of eczema, with sex-specific risk factors.

• Males are more likely than females to follow an early-onset persistent eczema trajectory.

• Females are more likely than males to developed eczema in adolescence and early adulthood.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.