Eczema prevalence in the United States: Data from the 2003 National Survey of Children’s Health

Abstract

Using the 2003 National Survey of Children’s Health (NSCH) sponsored by the federal Maternal and Child Health Bureau, we calculated prevalence estimates of eczema nationally and for each state among a nationally representative sample of 102,353 children 17 years of age and under. Our objective was to determine the national prevalence of eczema/atopic dermatitis in the United States pediatric population and to further examine geographic and demographic associations previously reported in other countries. Overall, 10.7% of children were reported to have a diagnosis of eczema in the last 12 months. Prevalence ranged from 8.7% to 18.1% between states and districts, with the highest prevalence reported in many of the East Coast states, as well as Nevada, Utah, and Idaho. After adjusting for confounders, metropolitan living was found to be a significant factor in predicting a higher disease prevalence with an OR of 1.67 (95% confidence interval of 1.19-2.35, p=0.008). Black race (OR 1.70, p=0.005) and education level in the household greater than high school (OR 1.61, p=0.004) were also significantly associated with a higher prevalence of eczema. The wide range of prevalence suggests social or environmental factors may influence disease expression.

Introduction

Atopic dermatitis (AD) is a global public health concern considering its increasing prevalence and mounting financial costs to health systems (Carroll CL, et al, 2005; Ellis C, et al, 2002; Lapidus CS, et al, 1993). The International Study of Asthma and Allergies in Childhood (ISAAC) revealed that AD affects children across the globe, although the disease prevalence varies substantially between countries (Asher MI, et al, 2006). The prevalence of AD is alsoincreasing, especially in developing countries (Asher MI, et al, 2006; Williams H, et al, 2008). The factors that underlie disease prevalence, geographic variability, and secular trends are unknown (Burney PG, Chinn S, Rona RJ, 1990; Williams H, et al, 2008), although industrialization and urban living have correlated with elevated rates of eczema (Addo Yobo EO, et al, 1997; Keeley DJ, Neill P, Gallivan S, 1991; Mercer MJ, et al, 2004; Yemaneberhan H, et al, 1997).

Most data regarding AD prevalence in industrialized countries are derived from the study of European populations. Only three previous studies of AD prevalence reported data from a United States population (Asher MI, et al, 2006; Hanifin JM, et al, 2007; Laughter D, et al, 2000), with the scope of two of these limited to one state. The third and most recent study did not examine geographical trends (Hanifin JM, et al, 2007). Further data regarding disease prevalence, geographic variation, and risk factors are needed from the United States.

The primary objective of the current study was to determine the prevalence of AD in the United States using data from the National Survey of Children’s Health, a large population-based survey of over 100,000 families representing all 50 states. We examined the geographical distribution of the disease and explored whether certain risk factors and associations previously reported in Europe and Asia were also present in the U.S. population.

Results

Univariate analyses

Overall, a total of 9,752 children had a diagnosis of eczema, which translated to a 10.7% national prevalence of eczema in children under 18 years of age. The disease prevalence ranged from 8.7% to 18.1% between states and districts. The Figure and Table 1 present state prevalence estimates for United States children (0-17 years of age) who were reported to have a diagnosis of eczema in the last 12 months. The highest state prevalence values were reported in many East Coast states, as well as Utah, Idaho and Nevada. The lowest state prevalence values were in the middle and southwestern parts of the country (Table 1).

Figure 1.

Eczema prevalence in the United States shows a trend towards higher disease prevalence’s in the East Coast states.

Table 1.

Eczema prevalence by state

State/District	Freq*	%**	95% Confidence Interval
West Virginia	159	8.69	(7.21, 10.18)
South Dakota	136	8.69	(6.93, 10.46)
California	180	8.74	(7.26, 10.22)
New Mexico	142	8.74	(7.08, 10.39)
Iowa	159	8.78	(7.27, 10.30)
Vermont	154	8.92	(7.30, 10.54)
Arkansas	139	9.02	(7.31, 10.73)
Florida	159	9.07	(7.34, 10.80)
Mississippi	147	9.37	(7.56, 11.19)
Wisconsin	162	9.39	(7.78, 11.00)
Texas	174	9.69	(8.08, 11.30)
Pennsylvania	200	9.70	(8.18, 11.23)
Hawaii	173	9.73	(8.04, 11.43)
Nebraska	147	9.88	(8.09, 11.66)
New Hampshire	179	9.88	(8.35, 11.41)
Missouri	189	9.94	(8.36, 11.52)
Oklahoma	165	10.03	(8.36, 11.71)
North Dakota	154	10.12	(8.31, 11.93)
Tennessee	176	10.26	(8.49, 12.03)
Kansas	156	10.26	(8.40, 12.11)
Illinois	203	10.29	(8.61, 11.98)
Wyoming	168	10.35	(8.67, 12.02)
Indiana	164	10.66	(8.84, 12.48)
Washington	190	10.72	(9.07, 12.37)
Arizona	152	10.79	(8.86, 12.71)
Alaska	165	10.93	(9.08, 12.78)
Colorado	194	10.93	(9.21, 12.65)
Maine	187	11.14	(9.36, 12.92)
Montana	185	11.17	(9.38, 12.96)
Delaware	228	11.26	(9.67, 12.86)
South Carolina	208	11.30	(9.60, 13.00)
Ohio	220	11.32	(9.66, 12.97)
Minnesota	168	11.36	(9.39, 13.32)
Oregon	188	11.49	(9.69, 13.29)
North Carolina	204	11.51	(9.75, 13.27)
Connecticut	241	11.56	(9.93, 13.20)
Kentucky	204	11.57	(9.80, 13.33)
Alabama	216	11.63	(9.84, 13.41)
Michigan	222	11.68	(9.98, 13.39)
Virginia	220	11.73	(9.97, 13.49)
New York	222	11.75	(10.01, 13.49)
Idaho	169	11.82	(9.93, 13.71)
Georgia	192	11.93	(10.00, 13.86)
New Jersey	270	13.14	(11.36, 14.91)
Maryland	261	13.20	(11.37, 15.03)
Massachusetts	265	13.44	(11.65, 15.23)
Utah	169	13.52	(11.39, 15.65)
Rhode Island	258	13.56	(11.68, 15.45)
Louisiana	250	13.70	(11.74, 15.66)
Nevada	226	14.17	(12.17, 16.18)
District of Columbia	293	18.05	(15.64, 20.45)

^*Raw frequency of surveyed subjects with eczema

^**Weighted percent of state pediatric population with eczema

Of those children with eczema, 30.7% reported concurrent hay fever and 22.8% reported concurrent asthma consistent with similar AD populations in Europe (Asher MI, et al, 2006; Van der Hulst A, Klip H, Brand P, 2007). As expected, age of the child was a significant determinant of eczema prevalence given the natural course of the disease (Table 2). There was a significant effect of the highest reported education level in the household on eczema prevalence, with those households reporting education levels greater than high school having the greatest prevalence of eczema (Table 2). Other significant demographic variables showing positive associations with disease prevalence included living in a metropolitan area (defined by using Rural-Urban Commuting Area [RUCA] codes), speaking English as the primary language, and being of black or multiple race (Table 2).

Table 2.

Eczema prevalence stratified by primary demographic variables

Variable	Subgroup	Freq*	%**	95% Confidence Interval	p-value†
Age	< 4 years	2977	13.92	(13.12, 14.73)	<.0001
	4-8 years	2623	10.63	(9.98, 11.27)
	9-12 years	1862	9.96	(9.23, 10.68)
	13-17 years	2290	8.56	(7.97, 9.16)
Gender	Male	4874	10.52	(10.04, 11.01)	0.3507
	Female	4867	10.85	(10.36, 11.34)
Highest education level completed by parent	<HS	278	6.95	(5.63, 8.27)	<.0001
	HS	1721	9.61	(8.89, 10.33)
	>HS	7721	11.47	(11.06, 11.88)
Residence in metropolitan area	No	1442	8.53	(7.90, 9.16)	<.0001
	Yes	5161	10.99	(10.55, 11.43)
Primary language spoken in home	English	9273	11.15	(10.78, 11.51)	<.0001
	Any other	474	6.91	(5.89, 7.94)
Race	White only	6770	9.70	(9.34, 10.05)	<.0001
	Black only	1464	15.89	(14.64, 17.14)
	Multiple race	550	15.03	(12.97, 17.10)
	Other	470	10.08	(8.36, 11.80)
Household income	0-99% FPL	1037	10.38	(9.40, 11.37)	0.0357
	100-199% FPL	1732	11.09	(10.18, 12.00)
	200-399% FPL	3135	10.21	(9.65, 10.77)
	≥400% FPL	3024	11.53	(10.91, 12.15)

^*Raw frequency of surveyed subjects with eczema

^**Weighted percent of subgroup population with eczema

^†Rao-Scott chi-square test for equal proportions

Birthplace of parents or child was associated with disease prevalence. Children or parents born outside the United States reported a lower prevalence of eczema (Table 3). A significant association was also found with health insurance status. Children with health insurance had greater eczema prevalence than those without (10.9% vs. 8.2%, p=0.0004), possibly reflecting healthcare access disparities.

Table 3.

Eczema prevalence stratified by birthplace variables

Variable	Subgroup	Freq*	%**	95% Confidence Interval	p-value†
Child’s mother born in US	No	912	9.08	(8.09, 10.07)	0.0004
	Yes	8352	11.14	(10.76, 11.52)
Child’s father born in US	No	739	9.27	(8.15, 10.39)	0.0297
	Yes	6508	10.66	(10.24, 11.07)
Child born in US	No	233	6.80	(5.28, 8.32)	<.0001
	Yes	9431	10.84	(10.49, 11.19)

^*Raw frequency of surveyed subjects with eczema

^**Weighted percent of subgroup population with eczema

^†Rao-Scott chi-square test for equal proportions

Eczema prevalence showed an association with family structure, with single mothers reporting the highest prevalence (Table 4). Single child homes had a higher prevalence than families with more than one child, but birth order did not seem to influence disease prevalence. Children reported to regularly receive child care had a significantly higher prevalence of eczema than those who did not (Table 5), with the highest prevalence being seen in those who attended child care outside of the home. Smoking in the home showed no association with eczema prevalence.

Table 4.

Eczema prevalence stratified by family structure variables

Variable	Subgroup	Freq*	%**	95% Confidence Interval	p-value†
Number of children in household	1 child	4149	11.82	(11.30, 12.35)	0.0039
	2 children	3700	10.75	(10.25, 11.24)
	3 children	1382	10.07	(9.29, 10.86)
	4 or more children	521	9.78	(8.52, 11.04)
Birth order in families with 2 or more children	Oldest child	2182	9.72	(9.10, 10.33)	0.1098
	2nd oldest child	2590	10.97	(10.34, 11.59)
	3rd oldest child	640	10.07	(8.88, 11.26)
	4th oldest child	191	11.14	(8.75, 13.52)
Family structure	Two parent biological/adopted	6378	10.68	(10.26, 11.10)	0.0013
	Two parent stepfamily	680	9.97	(8.70, 11.23)
	Single mother/no father present	2179	11.42	(10.65, 12.20)
	Other	296	7.68	(6.14, 9.22)

^*Raw frequency of surveyed subjects with eczema

^**Weighted percent of subgroup population with eczema

^†Rao-Scott chi-square test for equal proportions

Table 5.

Eczema prevalence stratified by environmental variables

Variable	Subgroup	Freq*	%**	95% Confidence Interval	p-value†
During the past month did child regularly attend child-care center?	No	2508	11.60	(10.89, 12.31)	<.0001
	Yes	1598	15.41	(14.14, 16.67)
Does anyone in the household use cigarettes, cigars, or pipe tobacco?	No	5717	10.40	(9.96, 10.83)	0.8966
	Yes	2413	10.45	(9.78, 11.12)

^*Raw frequency of surveyed subjects with eczema

^**Weighted percent of subgroup population with eczema

^†Rao-Scott chi-square test for equal proportions

Multivariate analysis

We developed a logistic regression model to better explain the relationship between area of residency (metropolitan area versus rural area) and eczema prevalence. After adjusting for potential confounders including race and age of child, parental education level, household income, and health insurance coverage status, metropolitan living continued to be a significant factor in predicting a higher disease prevalence with an OR of 1.67 (95% confidence interval of 1.19-2.35, p=0.008) compared to rural living. Black race (OR 1.70, p=0.005) and education level in the household greater than high school (OR 1.61, p=0.004) were also significantly associated with a higher prevalence of eczema compared to white race and education level less than high school, respectively (Table 6).

Table 6.

Subgroup comparisons of variables included in multivariate model

Variable	Contrast	Odds ratio	Standard Error	95% Confidence Interval	p-value*
Residence in metropolitan area	Metro vs. Rural	1.67	0.29	(1.19, 2.35)	0.0079
Race	Black vs. White	1.70	0.29	(1.22, 2.37)	0.0048
	Multiple race vs. White	0.84	0.18	(0.56, 1.27)	0.5033
	Other vs. White	0.95	0.22	(0.60, 1.49)	0.8513
Age category	<4 yrs vs. 13-17 yrs	1.77	0.11	(1.57, 2.00)	<.0001
	4-8 yrs vs. 13-17 yrs	1.27	0.08	(1.13, 1.43)	0.0006
	9-12 yrs vs. 13-17 yrs	1.15	0.08	(1.01, 1.31)	0.0640
Highest education level completed by parent	HS vs. <HS	1.34	0.20	(1.00, 1.78)	0.0799
	>HS vs. <HS	1.61	0.23	(1.21, 2.13)	0.0038
Income as a percent of poverty level	100-199% vs. 0-99%	1.11	0.09	(0.94, 1.31)	0.2938
	200-399% vs. 0-99%	0.99	0.08	(0.84, 1.15)	0.8513
	≥400% vs. 0-99%	1.13	0.09	(0.96, 1.32)	0.2304

^*Wald chi-square test adjusted for multiple comparisons by False Discovery Rate method

In addition to adjusting for the main effects of potential confounders, interactive effects of insurance coverage, race, and metropolitan residency were included in the final model to better adjust for possible inequities in healthcare access. Statistically significant interaction terms included insurance status by metropolitan residency (p=0.047), and the three-way interaction between insurance, race, and residency (p=0.04), suggesting that uninsured and insured, as well as the different racial subgroups, may have experienced differences in healthcare access depending on their residency status.

Discussion

Our large population-based study found the prevalence of AD in the United States to be approximately 10.7% with a significant variation between states and districts. Urban living and being of black race were significantly associated with a higher prevalence of eczema after controlling for possible confounders. A general geographic trend toward higher disease prevalence in the East Coast states was also found. We confirmed known demographic AD associations previously observed only in European populations including the association of AD with higher education levels, higher household incomes, and smaller family sizes. Notable associations not observed in our study included a lack of association with smoking in the household, breast feeding, birth order, gender, or body mass index (BMI). The lack of a correlation between BMI and eczema is supportive of current studies that show no relationship between BMI and eczema (Leung TF, et al, 2009; Van Gysel D, et al, 2009).

Our findings of an AD prevalence of 10.7% in US children 0-17 years of age agrees with reported estimates from the three prior US-based studies of AD prevalence. A study by Hanifin reporting the results of a 1998 survey found 17.1% of the study population had at least one of four eczematous symptoms, while 10.7% of respondents reported empirically defined eczema (Hanifin JM, et al, 2007). Laughter’s study reported in 2000 of 1465 Oregon schoolchildren 5-9 years of age found a prevalence of 11.8% based on the question, “Has a doctor ever said that your child has eczema?” (Laughter D, et al, 2000). Using the self-administered Schultz Larsen questionnaire, a 17.2% lifetime prevalence was found in that study. The global ISAAC study, where the US was represented by a sample of 2,422 children from one medical center in Seattle, found a prevalence of eczema symptoms to be 8.3% (Lapidus CS, et al, 1993). Our study estimate was slightly higher, with a prevalence estimate of 10.7% in the state of Washington.

Similar to the ISAAC study, which revealed striking world-wide geographic variability in AD prevalence, our data revealed significant geographic variability in disease prevalence within the United States with a higher prevalence in the East Coast states. The reason for this variability is not known and is likely multi-factorial. One explanation may be the presence of a higher number of metropolitan centers in the Eastern versus Western United States. Our data revealed a higher eczema prevalence in metropolitan areas even when controlling for confounders. Several previous studies of atopic disease reported a similar increase in disease prevalence in metropolitan/urban areas compared to rural areas (Addo Yobo EO, et al, 1997; Keeley DJ Neill P, Gallivan S, 1991; Laughter D, et al, 2000; Mercer MJ, et al, 2004; Yemaneberhan H, et al, 1997). Potential explanations for this phenomenon include metropolitan-related environmental factors such as exposure to environmental pollution (Asher MI, et al, 2006). For example, an increased prevalence of allergic disease in Ethiopia was associated with the use of modern fuels, particularly kerosene use in homes (when compared to other biomass fuel) (Venn AJ, et al, 2001). Another possibility noted by von Hertzen was the heavy exposure to microorganisms in soil and vegetation when living in rural farming areas (Haahtela T, 2006, and von Hertzen L). Cultural and behavioral factors that affect the skin barrier may also play a role. Sheriff found a correlation between an increased hygiene score (that included the frequency of washing/wiping hands and faces and bathing practices of young children) and subsequent eczema risk (Sherriff A, et al, 2002). Whether skin care practices vary between rural and metropolitan inhabitants is not known.

An unexpected association in our study was the greater prevalence of eczema in black and multi-race populations compared to whites. Hanifin did not find statistically significant differences between various race populations and their prevalence of eczema. A few prior studies have reported racial disparities in eczema prevalence (Davis LR, Marten RH, Sarkany I, 1961; Schachner L, Ling LS, Press S, 1983; Williams HC, et al, 1995). In the most recent, Williams found a higher prevalence of AD in black Carribeans in London compared to whites (Williams HC, et al, 1995). Using medical care usage as a proxy for disease prevalence, Horii (Horii KA, et al, 2007) reported an increased use of medical care for atopic dermatitis by Blacks and Asian/Pacific Islanders when compared to whites. It is not known whether these racial differences derive from environmental or genetic influences. There are no large studies of the prevalence of common filaggrin mutations in an African population. Studies in asthma have also reported similar racial disparities and differences in socioeconomic status and air quality have been proposed as the possible explanations (Gorman BK, 2009).

A significant limitation of our study was that we could not be certain whether geographic differences in disease prevalence reflected differences in access to medical care of dermatologic specialty care. There are fewer dermatologists per capita in rural areas compared to urban areas, although wait time to be seen by a dermatologist was not statistically different between urban and rural areas (Uhlenhake E, Brodell R, Mostow E, 2009). Our regression model controlled for this issue but this does not eliminate the potential bias completely. Another limitation of this study was the nature of the self-reported survey data collection. Diagnoses were not confirmed by a chart review or direct examination of the patients. Single questions addressing parent recall of physician-diagnosed eczema that have been validated and used in other prevalence studies reported a high concordance between using a similar single question (“Has a doctor ever said that your child has eczema?”) with direct clinical examination and questionnaire diagnosis of atopic dermatitis (Laughter D, et al, 2000). Another study from Germany tested the validity of the diagnosis of AD using the question, “Has a physician ever diagnosed eczema in your child?” It showed 63% sensitivity and 88% specificity using dermatologic exam as the gold standard (Kramer K, et al, 1998). Based on the results of these studies, the wording of the question in this survey has adequate sensitivity and specificity to provide meaningful data on eczema prevalence. Finally, this survey data is now seven years old.

Materials and Methods

Data source

We used data from the 2003 National Survey of children’s Health (NSCH) survey of 102,353 households, which was designed to estimate the prevalence of various child health issues including physical, emotional, and behavioral factors. The NSCH was sponsored by the Maternal and Child Health Bureau and the U.S. Department of Health and Human Services. The National Center for Health Statistics conducted a total of 102,353 interviews using the State and Local Area Integrated Telephone Survey (SLAITS) program. The telephone numbers were chosen at random, followed by identification of the households with one or more children under the age of 18. Subsequently, one child was randomly selected for interview. The survey results were weighted to represent the population of non-institutionalized children nationally and in each state. Using the data from U.S. Bureau of the Census, weights were adjusted for age, sex, race, ethnicity, household size, and educational attainment of the most educated household member to provide a dataset that was more representative of each state’s population of non-institutionalized children less than 18 years of age. The National Center for Health Statistics of Center for Diseases Control and Prevention oversaw sampling and telephone interviews. More detailed information on the survey has been previously published (Blumberg SJ, et al, 2005).

Study variables

We calculated the period prevalence of atopic dermatitis/eczema using the NSCH question, “During the past 12 months, have you been told by a doctor or other health professional that [child’s name] had eczema or any kind of skin allergy?” To limit the effect healthcare access may have on the results, we excluded all subjects who responded “no” to the question, “During the past 12 months, did (child) see a doctor, nurse, or other health care professional for any kind of medical care, including sick-child care, well-child check-ups, physical exams, and hospitalizations?” We also included health care insurance status in our final regression model when we examined the role of metropolitan living on AD prevalence.

NSCH data were interpreted to calculate the national prevalence of eczema for the United States as well as for each state. Further investigation into the influences of race, geography, socioeconomic status, education levels, family size, place of residence, and birth order was performed based on previously described associations in the literature found in European populations (Hanifin JM, 2009).

Statistical methods

Analyses were performed using SURVEY procedures in SAS version 9.2. Univariate associations were tested by Rao-Scott Chi-square method. Multivariate results were obtained by logistic regression for domains of weighted survey data. Regression analysis did not include data from many states (including Alaska, Connecticut, Delaware, Hawaii, Idaho, Maine, Maryland, Massachusetts, Montana, Nevada, New Hampshire, North Dakota, Rhode Island, South Dakota, Vermont, and Wyoming) for which metropolitan residency status was unavailable. The regression model used residency status (metropolitan versus rural) to predict diagnosis of pediatric eczema while controlling for potential demographic confounders, including race and age and health insurance coverage status. The number of children living in the home was not significantly associated with eczema diagnosis, so this variable was removed to simplify the model. Interactive effects between race, insurance status, and metropolitan residency were included in an attempt to better control for inequity in healthcare access between races and areas of residency. Odds ratios for specific demographic comparisons were determined using the final multivariate model, and their p-values were adjusted for multiple comparisons using the False Discovery Rate method (Table 6).

Abbreviations used

(AD)

Atopic dermatitis

(BMI)

Body Mass Index

(CAHMI)

Child and Adolescent Health Measurement Initiative

(ISAAC)

International Study of Asthma and Allergies in Childhood

(NSCH)

National Survey of Children’s Health

(RUCA)

Rural-Urban Commuting Area

(SLAITS)

State and Local Area Integrated Telephone Survey program