The Burden of Childhood Atopic Dermatitis in the Primary Care Setting

A Report From the Meta-LARC Consortium

Jinan Al-naqeeb, MD, MPH; Sankirtana Danner, MA, CCRP; Lyle J. Fagnan, MD; Katrina Ramsey, MPH; LeAnn Michaels; Julie Mitchell; Kelsey Branca, MPH; Cynthia Morris, PhD, MPH; Donald E. Nease, Jr., MD; Linda Zittleman, MSPH; Barcey Levy, MD, PhD; Jeanette Daly, RN, PhD; David Hahn, MD, MS; Rowena J. Dolor, MD, MHS; Hywel C. Williams, DSc, FMedSci; Joanne R. Chalmers, PhD, BSc; Jon Hanifin, MD; Susan Tofte, RN, FNP; Katharine E. Zuckerman, MD, MPH; Karen Hansis; Mollie Gundersen; Julie Block; Francie Karr; Sandra Dunbrasky, MD; Kathy Siebe, CPNP; Kristen Dillon, MD; Ricardo Cibotti, PhD; Jodi Lapidus, PhD; Eric L. Simpson, MD, MCR


J Am Board Fam Med. 2019;32(2):191-200. 

In This Article


Study Design, Population, and Setting

This study, named the Community-Based Assessment of Skin Care, Allergies and Eczema (CASCADE) study, was a cross sectional survey study conducted in 5 US states. CASCADE was a planning study to determine the feasibility of conducting a large, 5-year, community-based pragmatic randomized controlled clinical trial to test the hypothesis that certain skin care practices can prevent or delay AD and allergic comorbidities. Study participants were dyads of parents or guardians and children 0 to 5 years old attending 1 of 10 community-based pediatric (n = 6) and family medicine (n = 4) clinics located in Oregon, Wisconsin, Colorado, North Carolina, and Iowa. These clinics were all members of a practice-based research network (PBRN) within their respective state and were a mix of rural and suburban practices. All participating PBRNs collaborate via the Meta-network Learning and Research Center (Meta-LARC) consortium, an administrative structure funded by the Agency for Health care Research and Quality encompassing almost 1,000 primary care practices and 7,000 clinicians (, accessed April, 2018). PBRNs serve as essential partners in translating academic research advances into real-world health improvements in the general ambulatory care population.[22] The study was approved by the institutional review board (IRB#11116) of Oregon Health and Science University and recruited participants from April 2015 through January 2016.

Inclusion required being a parent or legal guardian, aged ≥18 years, of a child between the ages of 0 and 5 years who was a current patient at the participating clinic; respondents also needed to be able to read and write in either English or Spanish. Potential respondents were excluded if unable to complete the questionnaire due to mental or cognitive capacity, if they or another of the child's parents had already completed the questionnaire (ascertained by self-report), or if the child had been born preterm at less than 25 weeks of pregnancy.


The Iowa PBRN recruited participants by mail only, whereas all other PRBNs used a combination of the following methods to capture as wide a sample as possible: in clinic while waiting for their appointments, mailed surveys, and electronic surveys via Research Electronic Data Capture (REDCap) hosted by Oregon Health and Science University. This flexibility in recruitment methods allowed for a minimal impact on clinic workflow that is critical to performing research in this setting. Clinic staff were instructed to broadly distribute the surveys to all eligible patients attending the clinic during the enrollment period, without selecting for any demographic or medical history to minimize selection bias. Due to the nature of the survey distribution in a busy practice setting, refusals were not recorded; thus, we are unable to report a participation rate.


The questionnaire was completed by the child's caregiver and included questions about AD history, symptoms, age of onset, presence of other atopic disorders, medications use, and skin care and bathing practices (Supplemental Figure 1.) The questionnaire was adapted from previous childhood AD surveys that have been used and validated in a community setting to measure the prevalence of AD.[2,9,23]

AD and Severity Assessment

A history of AD was determined by a positive response to the question "Have you ever been told by a health care provider that your child has eczema or atopic dermatitis?" A similar question has been shown to have adequate sensitivity and specificity to estimate the prevalence of AD in the United States.[24–26] We assessed the effect of AD on the child's sleep by asking how many nights in the past week the child's sleep had been disturbed because of a red rash or eczema. AD severity was assessed by asking respondents to rate the rash or eczema as mild, moderate, or severe. This question has been used previously in epidemiologic studies to assess severity and found to be a good indicator of childhood AD burden.[2,9,27]

Comorbidities and Family History

A history of asthma and wheezing were measured using the questions "Has your child ever been diagnosed with asthma by a health care provider?" and "Has your child ever had wheezing?" Food allergies were measured by asking "Has your child ever been diagnosed with a food allergy by a health care provider?" Family history of any atopic condition was assessed by asking the question "Has at least 1 of your child's parents or older brothers or sisters (related by blood) ever had any of the following conditions: eczema, asthma, or hay fever/spring-time allergies?" Parental history of asthma was considered positive if at least 1 of the parents had asthma. The questions were adapted from previously validated questions used in epidemiologic studies which measured comorbidities.[28,29]

Skin Care and Bathing

Moisturizer use was assessed by answering the question "do you use a moisturizer/lotion/oil on your child's skin?" An affirmative response to the above question was followed by secondary questions about the type of moisturizer and the application site—all over the body or just on dry areas. Moisturizer type was assessed by asking the following question: "Which moisturizer(s) did you use on your child?" The answer choices included the most commonly used commercial moisturizers brands: CeraVe cream, Cetaphil cream, Vaseline/petroleum jelly, Sunflower seed oil, Aveeno, Aquaphor, Vanicream, and Johnson's baby lotion. If the moisturizer that had been used was not listed in the answer choices that were given, the parents were instructed to check "other" and write what they were using. For a better understanding of current trends in moisturizer use, more than 1 answer was acceptable for this question. We stratified moisturizers based on their content: lotion, cream, ointment, or liquid oil. If the parents did not specify the type (ie, lotion, ointment, cream, or liquid oil), and the brand product could represent more than 1 type of moisturizer, the answer was excluded from analysis.

Moisturizer frequency and bathing or shower frequency was measured using the following questions: "Over the past 3 months, on average how many days per week was a moisturizer/lotion/oil applied to your child's skin" and "Over the past 3 months, on average how many days per week did your child receive either a bath or shower?" Moisturizer frequency was asked only in those participants who were using moisturizers, whereas bathing frequency was asked of the whole sample. We categorized frequency of moisturizer use and bathing into 2 categories: <4 and 4 days or more per week. Because the biological effect of most moisturizers lasts more than 24 hours, we considered 4 days a week or more to be frequent use.

Sample Size and Statistical Analysis

A sample size of approximately 250 was estimated to provide a reasonably precise sample of disease prevalence from age 0 to 5 years with a 95% CI within 5 percentage points. Our actual sample was significantly larger to obtain an adequate sample from all age groups and study sites. We excluded 9 respondents who failed to provide the child's age (n = 5) and history of provider diagnosis of AD (n = 3), or both (n = 1); our final dataset included 652 children.

We calculated simple descriptive statistics overall and for AD and non-AD groups and tested differences with χ 2 tests. To estimate age-specific characteristics of AD, we used predictive margins from a logistic regression model with clustered standard errors to account for correlation between respondents from the same clinic. Similarly, estimates of comorbid conditions and skin care practices resulted from logistic or log-binomial (relative risk) models with clustered standard errors and including age, in months, as a covariate to adjust for this effect. All analyses were performed using Stata SE for windows version 14 (Stata Corp, College Station, Texas).