Effects of variations in access to care for children with atopic dermatitis

Background An estimated 50% of children in the US are Medicaid-insured. Some of these patients have poor health literacy and limited access to medications and specialty care. These factors affect treatment utilization for pediatric patients with atopic dermatitis (AD), the most common inflammatory skin disease in children. This study assesses and compares treatment patterns and healthcare resource utilization (HCRU) between large cohorts of Medicaid and commercially insured children with AD. Methods Pediatric patients with AD were identified from 2 large US healthcare claims databases (2011–2016). Included patients had continuous health plan eligibility for ≥6 months before and ≥12 months after the first AD diagnosis (index date). Patients with an autoimmune disease diagnosis within 6 months of the index date were excluded. Treatment patterns and all-cause and AD-related HCRU during the observation period were compared between commercially and Medicaid-insured children. Results A minority of children were evaluated by a dermatology or allergy/immunology specialist. Several significant differences were observed between commercially and Medicaid-insured children with AD. Disparities detected for Medicaid-insured children included: comparatively fewer received specialist care, emergency department and urgent care center utilization was higher, a greater proportion had asthma and non-atopic morbidities, high- potency topical corticosteroids and calcineurin inhibitors were less often prescribed, and prescriptions for antihistamines were more than three times higher, despite similar rates of comorbid asthma and allergies among antihistamine users. Treatment patterns also varied substantially across physician specialties. Conclusions Results suggest barriers in accessing specialty care for all children with AD and significant differences in management between commercially and Medicaid-insured children. These disparities in treatment and access to specialty care may contribute to poor AD control, especially in Medicaid-insured patients. Supplementary Information Supplementary information accompanies this paper at 10.1186/s12895-020-00114-x.


Background
An estimated 50% of children in the US are Medicaidinsured [1,2]. Access to care for patients enrolled in Medicaid programs is an ongoing concern in the United States (US) [3]. Several studies have shown that patients enrolled in Medicaid are less likely to gain outpatient access to specialty providers [4][5][6][7]. A number of factors, including unfavorable fee-for-service reimbursement, longer wait times for payments, and higher clinic nonattendance rates, contribute to the dearth of specialists accepting Medicaid patients [5,8].
Skin disease is very common in children, prompting up to 30% of all primary care pediatric visits [9]. Atopic dermatitis (AD), a chronic inflammatory skin disease characterized by eczematous lesions and intense pruritus [10][11][12][13], is the most common inflammatory skin disease in children, with an estimated prevalence in the United States of about 11-13% among children less than 18 years of age [14,15]. Up to one-third of these patients are estimated to have moderate-to-severe disease [16], along with a higher risk of atopic and non-atopic morbidities compared with children without AD [13,[17][18][19]. The burden of AD is substantial, especially in children with moderate-to-severe disease and their caregivers. Chronic sleep disturbance related to persistent pruritus profoundly affects daily functioning, quality of life (QoL), and psychosocial health [20][21][22][23][24][25][26]. AD in children is also associated with poorer performance in school, difficulties forming social relationships and participating in sports, and increased rates of anxiety, depression, and even suicidal ideation [20,22,24].
While there are currently no treatment guidelines for pediatric AD, the standard of care includes a combination of maintenance skin care and topical medication to prevent and treat flares [27]. Topical medication options include corticosteroids (TCS), calcineurin inhibitors (TCI), and a PDE4 inhibitor. Until the recent approval of dupilumab in adolescents (12-17 years old) with moderate-to-severe AD (March 2019), treatment options among pediatric patients with AD whose disease was not adequately controlled with topical therapies were limited [28]. In fact, for patients with moderate-tosevere AD with uncontrolled symptoms, phototherapy, oral immunosuppressants, and systemic corticosteroids (SCS) are often used off-label, despite potential side effects [29].
Management of AD has been observed to vary greatly across physician specialties [30]. Diagnosis and followup care are usually provided by the primary healthcare providers, most often pediatricians and family practitioners [31,32]. However, primary care training does not include requirements for care of pediatric skin disease, and most trainees have limited exposure to dermatology specialists [9,33]. Current recommendations for managing pediatric AD suggest dermatology referral for patients with moderate-to-severe and/or refractory, poorly controlled, or generalized AD [34,35]. Pediatric dermatologists are most experienced in the evaluation and treatment of children with AD, but their limited workforce makes subspecialist access challenging [36].
Restricted access for pediatric Medicaid-insured patients impacts referral to subspecialty care and utilization patterns, particularly for patients needing dermatologic care. A minority of children are successful in obtaining a new patient dermatology appointment [5,37,38]. The relapsing nature of AD often prompts patients to seek same-day evaluation at urgent care centers and emergency departments (EDs). These visits often prompt a specialty clinic referral [39], but the outcome of this tactic is unclear.
Although the challenges related to obtaining specialist evaluation and treatment for Medicaid patients are well documented, no study has specifically investigated the global healthcare disparities for children with AD. Understanding and addressing access disparity in pediatric patients with AD is especially critical given the significant healthcare burden for patients, as well as their caregivers. The objective of this study is to assess and compare treatment patterns and healthcare resource utilization (HCRU) of two large cohorts of Medicaidinsured and commercially insured children with AD.

Methods
This retrospective cohort analysis relied on administrative medical and pharmacy claims from the IBM® Mar-ketScan® Commercial Database and the Multi-State Medicaid Database covering the period from January 1, 2011 to December 31, 2016. Both databases include deidentified patient-level claims and comply with the Health Insurance Portability and Accountability Act (HIPAA) of 1996. Institutional review board approval was not necessary for this study.
Pediatric patients with AD were identified using the following criteria: (1) ≥1 medical claim with a diagnosis of AD (International Classification of disease, ninth revision [ICD-9] code 691.8; ICD-10 codes L20.x), (2) less than 18 years of age on the first observed AD diagnosis (defined as the index date), and (3) continuous health plan eligibility ≥6 months pre-index date (baseline period; up to 6 months for infants <1 year of age) and ≥12 months post-index date. Patients who received a diagnosis of an autoimmune condition (listed in Table 1) during the baseline period or on the index date were excluded. This criterion was applied to help exclude use of treatments of interest for conditions other than AD. The observation period spanned the index date to the end of health plan continuous eligibility or end of data availability, whichever occurred first.

Outcomes
Baseline characteristics included age, gender, type of healthcare provider seen on the index date, and ADrelated comorbidities evaluated during the 6-month baseline period and on the index date. Treatment patterns considered included the number of prescriptions per year, the proportion of patients with ≥1 combination therapy (overlap ≥3 months between ≥2 distinct AD treatments), and the proportion of patients with ≥1 prescription filled for the selected AD medications among patients with at least one treatment for AD during their observation period (treated patients). Medications assumed to be prescribed to treat AD included: TCS, TCI, antihistamines (topical and oral; sedating and nonsedating), montelukast sodium, SCS, immunosuppressants (azathioprine, cyclosporine A, methotrexate, mycophenolate mofetil, interferon gamma), intravenous immunoglobulin (IVIG), and phototherapy. Although topical and oral antibiotics are often prescribed for infected AD, they are also used for various unrelated, common childhood infections, and so were not included. We were unable to assess the use of crisaborole, approved in December 2016, at the end of our available data. Similarly, dupilumab was not included in the list of selected AD treatments, as it was not yet approved for AD in adolescents over the period covered by the data. Both all-cause and AD-related HCRU were assessed (including inpatient stays, ED visits, outpatient visits, urgent care center visits, and other resources [primarily patient home, independent laboratory, and other unlisted facilities]). AD-related HCRU was defined as encounters with a diagnosis of AD, an AD-related condition (ICD-9 codes 690.8, 692.9, 705.81; ICD-10 codes L25.9, L30.0, L30.1, L30.3, L30.8, L30.9), a selected skin infection (ICD-9 codes 041.10, 054.0, 686.09, 686.8, 686.9; ICD-10 codes B00.0, B95.8, L08.08, L08.9), or an IVIG or phototherapy procedure. Emergency department reliance (EDR), defined as the ratio of ED and urgent care visits divided by the sum of all ambulatory visits (urgent care + outpatient + ED), was also assessed. The EDR allows distinguishing between patients with frequent ED or urgent care episodes due to an increased need for care, and patients who rely on ED or urgent care center visits to gain access to other outpatient resources. The proportion of patients with high EDRdefined as the percentage of ambulatory visits occurring in the ED setting greater than 33% (i.e., EDR >0.33)was also evaluated. The cutoff value defining high EDR was determined in accordance with the findings of previous studies [40][41][42]. Treatment patterns, HCRU, and EDR were evaluated during the entire observation period.

Statistical analyses
Patient demographics, clinical characteristics, and AD treatment patterns (among treated patients) were summarized and compared between Medicaid-insured patients (Medicaid patients) and commercially insured patients (Commercial patients) using Chi-square and Wilcoxon Mann-Whitney tests for categorical and continuous variables, respectively. An alpha level of 0.05 was used to declare statistical significance. Treatment patterns were stratified by type of healthcare provider seen on the index date (dermatologist, allergist/immunologist [A/I], pediatrician, and "other" provider types [mainly primary care providers (PCPs), nurse practitioners, and acute care providers]) and by age groups (< 2 years, 2-5 years, 6-11 years, and 12-17 years). To account for the varying length of observation, HCRU was reported on a per 1000 patient-year basis. Incidence rate ratios (IRR) of HCRU between Medicaid and Commercial patients were estimated using generalized linear models with a log link and negative binomial distribution, adjusting for baseline demographics and clinical characteristics ( Table 2). All analyses were performed using SAS version 9.4 (SAS Institute, Cary, NC).

Sample
A total of 268,580 and 338,678 patients were identified in the Medicaid and Commercial cohorts, respectively. These relative numbers reflect the sizes of the Commercial and Medicaid samples in the IBM MarketScan databases over the 2011-2016 period, rather than the  On the index date, the largest subset of patients (43%) were seen by non-specialist providers (other provider types), followed by pediatricians (38%), dermatologists (12%), and A/I (6%) ( Table 2). On the index date, Medicaid patients were less likely to be seen by a pediatrician (25.1% Medicaid vs. 49.0% Commercial), a dermatologist (3.2% vs. 18.7%), or an A/I (2.8% vs. 9.3%; p < 0.001 for all). However, a greater proportion of Medicaid patients saw a non-specialist provider (other provider types) on the index date (68.9% compared to 22.9% Medicaid). Medicaid patients were also less likely with subsequently visit a dermatologist or an A/I during their observation period, except for those who were seen by a dermatologist or an A/I on their index visit.

Treatment patterns
During the observation period, Medicaid patients were more likely to be prescribed ≥1 treatment for AD compared with commercially insured patients (89.6% vs. 69.9%; p < 0.001). All treatment patterns were assessed and compared in this sample of treated patients with Medicaid versus Commercial coverage.
Compared with commercially insured patients, the proportion of Medicaid-insured patients who received at least 1 prescription for a systemic antihistamine was more than three times higher (70.9% vs. 21.9%; p < 0.001). Among those who were prescribed antihistamines, almost half of Medicaid patients and almost twothirds of commercially insured patients were prescribed a sedating antihistamine (Table 3). Among patients prescribed oral antihistamines in both cohorts, most did not have comorbid asthma or allergies during the baseline period (Medicaid: 60.3%; Commercial: 67.5%; p < 0.001). A slightly smaller proportion of Medicaid patients received SCS (22.8% vs. 25.2%; p < 0.001).
Among patients with a dermatologist visit on the index date, Medicaid patients were more likely to be prescribed high-potency TCS ( 27.5%; p < 0.001 for all). Regardless of the payor type, high-potency TCS and TCI were most often prescribed by dermatologists and montelukast sodium by A/I. Significant differences in antihistamine prescribing patterns were also observed across provider types. Systemic antihistamines were most often prescribed to patients who saw a non-specialist provider (other providers) on the index dateprescribed in more than half of these patients. Differences were also observed with regard to the proportion of sedating antihistamines prescribed (Table 4). Among patients who were prescribed a systemic antihistamine, sedating antihistamines were prescribed in 72.9% of patients who saw a dermatologist on the index date, compared with around 50% for those who saw the other types of provider. Among commercially insured patients who were prescribed systemic antihistamines, the majority received sedating antihistamines, reaching up to 80% of patients among those who saw a dermatologist on the index date.

Healthcare resource utilization
Although all-cause HCRU was higher among Medicaid patients than among Commercial patients for all types of encounter (adjusted IRR; inpatient: 1.31; ED: 2.25; outpatient: 1.08; urgent care centers: 1.93; other: 4.21; all p < 0.001; Fig. 1), the rate of AD-related outpatient visits

Discussion
Using administrative healthcare claims data, this study aimed to compare real-world patterns of care, medications prescribed, and HCRU between two large cohorts of children with AD covered by Medicaid and by Commercial insurance plans. Access to medical care, and especially subspecialty care, for pediatric patients is an issue that is well recognized, but not well studied. Most publications have used surveys of either providers or caregivers for broad therapeutic areas [3,17,19,24]. Few reports have focused on variations in treatment patterns of pediatric patients with AD observed across different providers [5-7, 37, 39]. This study provides a unique portrait of patterns of care for AD derived from databases including large samples of Medicaid-and commercially insured children. Moreover, analyses of treatments and HCRU stratified by provider type provide additional insights about the nature of potential healthcare disparities.
The type of clinician seen by patients for their initial AD-related visit varied greatly between Medicaid and Commercial patients. Although a small proportion of all pediatric patients, regardless of insurance type, were seen by a dermatologist or A/I on their index visit, the proportion of Medicaid patients who did was significantly lower than for commercially insured patients (dermatologist: 3.2% vs. 18.7%; A/I: 2.8% vs. 9.3%). Furthermore, Medicaid patients were three times as likely to be seen by a provider type other than a dermatologist, allergist, or pediatrician, compared with Commercial patients (68.9% vs. 22.9%). These results support, but exceed, those of a retrospective study using data from the 2008 to 2012 Medical Expenditure Panel Survey, documenting a <50% likelihood of receiving a diagnosis for a skin-related disease from a dermatologist, for Medicaidinsured compared with privately insured individuals [43]. The large proportion of Medicaid patients seen by other provider types may reflect evaluation/management of Medicaid patients by a non-specialist provider in an ED or urgent care setting.
Other studies have also documented low acceptance rate of new Medicaid patients by specialists. One 2004 Fig. 1 Incidence Rate Ratios of All-Cause HCRU of Patients with AD -Medicaid vs. Commercial (Reference Group). Notes: AD: atopic dermatitis; ED: emergency department; EDR: emergency department reliance; HCRU: healthcare resource utilization. 1. Estimated using generalized linear models with a log link and negative binomial distribution, adjusting for baseline demographics and clinical characteristics. 2. High EDR is defined with a percentage of ambulatory visits occurring in the ED of at least 33% (i.e., EDR >0.33). 3. Includes primarily patient home, independent laboratory, and other unlisted facilities. *P-value <0.05 survey of 612 dermatologists in the US found acceptance rates to be much lower for Medicaid patients than for privately insured patients (32% vs. 87%) [7]. In a 2011 US study focused on access to ambulatory dermatology care for new pediatric patients with AD, the average Medicaid acceptance rate was 19% among 471 dermatologists surveyed [37]. Our larger databases included even fewer referrals to specialists following the initial ADrelated visit. Among patients initially seen by a provider type classified as "other", only 2% of Medicaid patients versus 12% of Commercial patients had a subsequent visit with a dermatologist or an A/I. This is particularly surprising given the significantly greater proportion of Medicaid versus Commercial patients with extracutaneous morbidities, suggesting a greater need for subspecialty care in this cohort.
The majority of Medicaid patients were seen by other types of providers (68.9% vs. 22.9% Commercial), mainly PCPs, nurse practitioners, and acute care providers, all of whom have been found to approach treatment of AD with different prescribing patterns [30]. High-potency TCS and TCI were most often prescribed for patients who saw a dermatologist on their index visit. This finding supports greater comfort using higher potency agents among dermatologists, who may be more familiar with the principles of topical treatment and the low risk of side effects when these medications are used as recommended [30,44]. Overall, Medicaid-insured children were less likely to be prescribed high-potency TCS, SC, and TCI. Lower TCI utilization among Medicaid patients may be related to formulary restrictions and more limited access to dermatologists, the provider type most often prescribing TCI.
Stratification analyses by provider type showed that, for the small proportion of Medicaid-insured patients who saw a dermatologist or A/I on the index date, a similar or even higher proportion of patients received high-potency TCS and SCS compared with commercially insured patients seen by the same type of specialists. This suggests that fewer disparities between Medicaid and commercially insured patients remain when patients are managed by dermatologists or A/I. These results also support the fact that the limited access to dermatologists for Medicaid patients could, in part, explain their overall lower utilization of higher potency treatments. The higher proportion of patients who were prescribed high-potency TSC and SCS among Medicaid patients seen by a dermatologist may potentially be explained by their heavier comorbid profile and, given their more limited access to specialists, the fact that only the most severe AD cases are referred to/managed by dermatologists [5,7,37,38].
Oral antihistamines are regularly used in AD, despite their reported lack of efficacy beyond promoting sleep and their potential adverse effects [45,46]. Our results reflected this trend. Although antihistamines are sometimes prescribed to treat atopic comorbidities (e.g., urticaria and allergic rhinitis) and sedating antihistamines to treat AD for their soporific effect [47][48][49], the difference in the utilization of all antihistamines by Medicaid-and commercially insured patients was striking (70.9% vs. 21.9%)despite similar rates of comorbid asthma or allergies (60.3% and 67.5%). Moreover, the proportion of Medicaid patients receiving sedating antihistamines was more than twice that of Commercial patients (32.5% vs. 15.7%; of whom 64.9% and 71.5%, respectively, did not have comorbid asthma/allergies). Two common overthe-counter (OTC) antihistamines, loratadine and diphenhydramine, were prescribed 61 and 75 times more often to Medicaid-than commercially insured children (28.2% vs. 0.5% and 10.6% vs. 0.1%, respectively). Major factors that could affect antihistamine prescribing include need for a prescription as well as insurance coverage (regardless of prescription-restricted status). Several sedating antihistamines are available by prescription only; however, some are not, such as diphenhydramine, which is available without a prescription. All nonsedating antihistamines are available without a prescription, but many Medicaid plans and some Commercial plans will cover the cost of OTC antihistamines with a prescription, including loratadine. Accordingly, OTC antihistamines may be commonly used by commercially insured patients, but not detectable in the database because the costs are not covered by the Commercial health plans but are covered by Medicaid programs. Widespread use of antihistamines by these patients, especially those who are Medicaid-insured, also highlights possible over use. More limited Medicaid formularies that restrict access to many topical corticosteroids and TCIs may contribute to this health disparity.
We also detected a difference in antihistamine prescribing patterns among specialties. The majority of antihistamine prescriptions (72.9%) among patients who saw a dermatologist on the index date were for a sedating product, compared with 50.1% for those who saw an A/I, and 49.3% and 50.6% for those who saw a pediatrician and other provider types. This may be related to better understanding about histamine's lack of contribution to dermatitis itch and use of sedating antihistamines for their soporific effects. Among the few patients who were treated with immunosuppressants (Medicaid: 0.2%; Commercial: 0.3%), methotrexate was the preferred choice, regardless of the payor type. This is consistent with the results of a survey of pediatric dermatologists in the US and Canada that identified methotrexate as one of the top first-and second-line choices of immunosuppressants [50].
A potential explanation for the overall higher HCRU of Medicaid versus Commercial patients may be their heavier comorbid profile (atopic comorbidities: 33.2% vs. 26.0%; other comorbidities: 30.7% vs. 22.3%). It may also be the result of disparities in access to care, and more specifically to specialty care. Medicaid patients were found to be more than twice as likely to have an ADrelated visit to the ED compared with Commercial patients, a finding corroborated by an analysis of the 2006-2012 Nationwide Emergency Department Sample [51].
The EDR provided further understanding into the differences in ED and urgent care center use between the two populations. Medicaid patients were close to three times as likely to have a high EDR (EDR >0.33) compared with Commercial patients, suggesting that Medicaid patients rely more heavily on ED and urgent care center visits for ongoing care.
Qualitative interviews of 26 specialists and 14 PCPs indicated that pediatric patient referral through the ED was common practice for PCPs in order to obtain access to specialists for patients under public insurance coverage, regardless of the condition [39]. Another study looking at the use of ED among children with AD found that most ED visits were by publicly insured patients [52]. These findings corroborate those of numerous other studies, suggesting a need for improved utilization and convenient access to primary care and specialty outpatient care for Medicaid recipients [53,54].
A few limitations affected this large administrative healthcare claims analysis. One challenge is related to nonspecific use of the term "eczema", a larger group of dermatoses that includes AD as well as a wide range of other ICD diagnostic codes [55]. In order to limit our cohort to patients with AD and exclude those with other forms of eczema, only AD-specific ICD codes were used to identify relevant patients. We also conducted sensitivity analyses for a larger cohort that included a broader range of eczema-related diagnostic codes, with similar results. An additional limitation is failure to capture patients with AD who did not seek care for their symptoms, potentially skewing the study sample towards patients with more severe disease. Finally, race information was unavailable for Commercial patients (available only for Medicaid patients) and is thus not included. A growing body of data indicates differences among various racial AD patient groups, including greater severity among Black and Hispanic patients with AD [15]. Thus, some of the differences observed between Medicaid and Commercial patients may be due to disparities in the race distribution among patients in the two samples.

Conclusion
Results from this claims data analysis comparing two large pediatric AD cohorts, Medicaid-and commercially insured children, indicate that an overall minority of patients were seen by a specialist. Non-specialist providers saw a considerably greater proportion of Medicaid patients, compared with specialists, with dermatologists seeing the very smallest proportion. Accordingly, it was not surprising that Medicaid patients had a higher reliance on ED and urgent care centers, especially for ADrelated care, with a rate of ED visits more than twice as high for Medicaid compared with Commercial patients, highlighting the importance of barriers to accessing outpatient and specialist care. Finally, antihistamines were prescribed more than three times more often to Medicaid patients. There is currently no well-established standard-of-care or pediatric-specific guideline accepted by clinicians for AD, and treatment approaches vary greatly across physician specialties. These variations are amplified by disparities in access to specialty care, exacerbating the unmet treatment needs for children with AD. A more consistent and concerted approach is needed to treat this chronic condition. Long-term disease control has the potential to alleviate the direct burden of AD and impede the risk of developing atopic and non-atopic comorbidities and, in turn, may help reduce the utilization of healthcare resources in this patient population.
are not publicly available. Access to the IBM® MarketScan® Commercial Database and the Multi-State Medicaid Database can be requested by contacting Truven Health Analytics.
Ethics approval and consent to participate Not applicable. Data are de-identified and comply with the Health Insurance Portability and Accountability Act (HIPAA) of 1996. Accordingly, this study did not require approval from an institutional review board or collection of informed consent.

Consent for publication
Not applicable.