Atopic dermatitis (AD) is a recurrent chronic inflammatory disease of the skin.1 Its morphology varies with the patient's age, although recurrent forms are predominant, and the most common symptom is pruritus.2 The etiology of AD is complex, involving genetic factors and a combination of allergic factors (80% of patients present increased levels of immunoglobulin E) and nonallergic factors (epidermal barrier dysfunction, biological factors, and environmental factors).1–4AD affects around 10%-15% of children and 2%-7% of adults, especially in developed countries.2,5,6 Fifty percent of cases resolve during adolescence, and the disease can persist in up to 20% of adults.1 The incidence is higher in women, although more males are affected during childhood. Moderate to severe forms account for around 10%-20% of all cases of AD.1,2

AD generates a considerable psychosocial burden for patients and their families.7 Prognosis is poorer in patients with the following characteristics: a family history of AD, late onset, disseminated atopic dermatitis during childhood, female sex, and association with other allergic diseases (asthma and rhinitis).1–4 At present, topical corticosteroids are considered the cornerstone of pharmacologic treatment in moderate cases, whereas severe cases are treated with phototherapy and systemic immunomodulators such as ciclosporinA, methotrexate, mycophenolate mofetil, or azathioprine.2,4 However, ongoing clinical phase III clinical trials in patients with moderate to severe disease are assessing biologics targeting specific aspects of the pathogenic process.8,9

AD generates a high cost burden for patients and their families.10 However, since most studies only evaluate the cost of drug therapy, there is a paucity of evidence, including data on resource usage and costs associated with AD in Spain.11,12 Moreover, there is a growing need for naturalistic studies on the real clinical conditions of health care interventions that appropriately reflect the flow of patients through the health system, consumption of health care and social resources, and the impact on comorbidities. The objective of the present study was to determine the usage of resources and costs generated by AD in adults according to severity (mild, moderate, and severe) under conditions of daily clinical practice.

From an initial selection of 101 834 patients aged ≥18years in the study centers, 6287 were diagnosed with AD. After application of the exclusion criteria, a total of 6186 patients were finally included. Figure 1 shows the flow diagram of patients included in the study after taking into account the inclusion and exclusion criteria set out in Patients and Methods. We established 3 groups according to the severity of AD, as follows: mild, 55.7% (n=3445); moderate, 38.2% (n=2361); and severe, 6.1% (n=380).

Table 4 shows the general characteristics of the study population according to the 3 groups. Mean age was 47.1years, and 61.6% were women. Morbidity associated with AD, measured using RUBs, was 2.3points. The main comorbid conditions in patients with AD were asthma, depressive syndrome, and dyslipidemia (33.7%, 36.3%, and 53.7%, respectively). Comorbidities are also shown in Table 4.

The multivariate model showed that severe AD was associated with age (β=0.159), female sex (β=0.024), and general comorbidity (RUB, β=0.192), specifically asthma (β=0.138), depression (β=0.099), cardiovascular events (β=0.087), obesity (β=0.085), and smoking (β=0.025). The coefficient of determination of the model (R2) was 31.9%.

The distribution of clinical variables and concomitant and specific medication prescribed during the follow-up period is shown by study group in Table 5. The mean time from diagnosis was 32.0years, and mean (SD) concomitant medication was 1.4 (1.0) per patient-year: the most frequently administered drugs were antihistamines (74.0%), anxiolytics (26.1%), and antidepressants (22.7%). As for medication administered specifically for treatment of AD, the most frequent were emollients (87.2%), low- or medium-potency topical corticosteroids (37.0%), high-potency topical corticosteroids (22.8%), and calcineurin inhibitors (19.6%). Patients with severe AD received more concomitant medications than patients with moderate and mild AD (2.2 vs 1.6 and 1.2; P<.001): the main drugs were antihistamines (88.1% vs 74.3% and 70.0%; P<.001), anxiolytics (50.5% vs 32.7% and 18.9%; P<.001), and antidepressants (43.9% vs 28.8% and 16.2%; P<.001), respectively. The most frequently used drugs in patients with severe AD were methotrexate (23.0%) and ciclosporin (65.2%).

Table 6 shows the usage of resources and the associated costs (health care and non–health care) by study group. Patients with AD used more resources, especially for primary care visits (7.8 vs 5.9 and 4.7; P<.001), specialized care visits (5.4 vs 3.0 and 1.3; P<.001), and days unable to work (11.3 vs 5.5 and 1.6; P<.001). The average number of days of hospitalization was 0.30 per patient (n=18). The mean length of stay in patients who were admitted to hospital was 6.6days. No patients died during follow-up.

The total cost of patients included in the study reached €9.3 million, of which 75.5% were health care costs and 24.5% non–health care costs (loss of productivity). Of the total health care costs, 55.7% were generated in primary care and 19.7% in specialized care. Concomitant/specific medication generated the highest share of total costs (42.3%), followed by visits for specialized care (9.5%) (Table 6). The average unit cost as a component of the total cost was €1504/year. The average unit health care costs were higher for patients with severe AD than for patients with moderate and mild AD (€3686 vs €2165 and €811, respectively, P<.001). All components of health care costs were higher for patients with severe AD. The average unit cost as a component of the total cost corrected for covariates (analysis of covariance) was greater in patients with severe AD than in patients with moderate and mild AD (€3397 vs €2111 and €885, respectively, P<.001). These differences were maintained for health care costs (€2340 vs €1572 and €689; P<.001) and non–health care costs (loss of work productivity: €1057 vs €538 and €196, P<.001), respectively.

In the binary correlation model, health care cost was highly correlated with severity of AD (r=0.715), comorbidity (RUB; r=0.455), number of drugs prescribed (r=0.400), and age (r=0.297) (P<.001).

In the multiple linear regression model (stepwise), cost was associated with severity of AD (β=0.398), comorbidity (RUB; β=0.239), asthma (β=0.231), and number of drugs administered (β=0.187) (P<0.01 in all cases). Figure 2 shows the main components of the total cost by study group and disease burden. The proportion of cost in specialized care compared with total cost is greater in moderate to severe AD. The greater health care cost of patients with high comorbidity is noteworthy (€4545 [linear trend]).

Figure 2.

Distribution of the main components of the total cost by study group and disease burden.

Results were statistically significant (P<.05) in all pairwise comparisons.

RUB indicates resource utilization bands; R,2 coefficient of determination (linear model).

(0.3MB).

Figure 3 shows the comparison between health care costs and losses of productivity according to the classification of severity, which was based on the treatment administered (corrected average/unit) (sensitivity analysis). No clinically relevant differences were observed according to the 2 classifications of severe AD analyzed.

Figure 3.

Sensitivity analysis. Comparison between health care costs and loss of productivity according to the classification of severity by treatment administered (average/unit). STU indicates results of study performed. Patients with severe atopic dermatitis were those receiving immunosuppressants or biologics, and/or hospitalized patients (groups: mild, n=3445; moderate, n=2361; and severe, n=380). SEN, sensitivity analysis, which included patients receiving oral corticosteroids in the severe atopic dermatitis group (groups: mild, n=3445; moderate, n=2075; and severe, n=666).

(0.13MB).

AD generates considerable consumption of resources and a marked cost burden for health systems and patients and their families, especially in the case of severe AD.21 The present study showed that the average cost of adult AD was €1504, with notable differences between severe forms, whose cost (€3686) was more than 4-fold higher than that of mild forms (€811). Furthermore, severe AD was associated with age, female sex, and the presence of greater general comorbidity (asthma, depression, cardiovascular events, obesity, and smoking). The weight of these factors in the prognosis of AD has been assessed elsewhere, thus highlighting the coherence of our results.22,23 Expenditure was mostly accounted for by health care costs (75.5%), with the remainder (24.5%) arising from loss of productivity. In the case of the former, most of the cost was generated by drug prescription, followed by specialist care. However, it is important to remember that out-of-pocket expenses for patients and their families—both direct and indirect—were not included. These expenses may be associated with supplementary over-the-counter therapy or with unidentifiable visit-related costs such as days unable to work. Comparisons with similar studies are complex owing to differences in the health care settings and study design. In a study performed in Korea, Kim et al.11 reported an annual direct cost of €2102, of which the part attributable to health care costs and to loss of productivity was similar to that reported in the present study. Fowler et al.24 performed a retrospective study (n=41 247) in which they reported an average cost burden of US$991 per patient-year, highlighting that 38% was attributable to days unable to work. Suh et al.25 reported how costs associated with drugs for AD increased in cases associated with asthma—as in the present study—from US$482 to US$973 when AD was associated with asthma and/or rhinitis. Whiteley et al.26 reported that 30% of costs resulted from annual lost productivity and its impact on quality of life. Finally, in a recent review, Drucke et al.27 report the marked social, economic, and occupational impact of AD, as well as its effect on quality of life. The authors calculated the annual cost of AD from a conservative perspective and found it to be €5297millon in a population of approximately 322 million persons in 2016.

The present study shows that, in addition to the severity of AD, comorbidity and bronchial asthma are associated with greater costs and use of health care resources for the Spanish National Health System. In fact, the health care cost of patients with high comorbidity alone was €4545. This observation enables us to suggest that the AD is more than simply a cutaneous condition and that optimization of both costs and outcomes could benefit from a multidisciplinary approach.

According to the study parameters, the severity of AD was classed as mild in 55.7% of cases (n=3445), moderate in 38.2% (n=2361), and severe in 6.1% (n=380). As this was a population-based study, in which both clinical reporting by physicians and setting varied (primary and specialized care), we chose treatment administered as a parameter for classifying disease severity. Although scales for grading AD (eg, Scoring Atopic Dermatitis [SCORAD], Eczema Area and Severity Index [EASI], and Six Area, Six Sign Atopic Dermatitis Severity Index [SASSAD]) are available, their use is restricted to clinical trials, and they do not form part of daily clinical practice. Several algorithms propose scaling therapy according to severity; therefore, identification of the drugs prescribed, which were in fact recorded reliably, are a marker of the resources used by the physician to manage skin disease.15,28 However, given that systemic immunosuppressants are often restricted to specialized settings, we cannot rule out the possibility that this approach led to an underestimation of severe cases.

While one might think that variations in the course of AD could affect the classification criteria applied during the inclusion period, the estimated frequency of flare-ups, especially in moderate and severe forms, make it unlikely. In Spain, the results of the ACTIDA study,29 which was based on data from 227 dermatologists (N=1441), showed a mean of 3.6flare-ups/year. Most patients (97.2%) stated that they always or occasionally requested a medical evaluation.

Our results show marked heterogeneity in the treatment of AD, especially in the severe form. In addition to treatment for the disease itself, it is worth noting the increased use of concomitant medication such as antihistamines, anxiolytics, and antidepressants, although it is not possible to determine to what extent their prescription is associated with AD. Nevertheless, the association between anxiety, depression, and AD is well recognized.22

The most commonly used drugs prescribed for treatment of AD were systemic corticosteroids, methotrexate, and ciclosporin. CiclosporinA is the only drug approved for the treatment of AD and the drug of choice in various guidelines and recommendations.23 Methotrexate, azathioprine, and mycophenolate mofetil are considered second-line drugs, with a moderate probability of response and a safety profile that often leads treatment to be suspended. Since no specific treatment has priority, choice may be based on prescribing habits at the reference specialist centers in the study.30

Systemic corticosteroid regimens, on the other hand, were in fact a common therapeutic resource. While systemic corticosteroids are not generally considered a first-line regimen—their use is advised against owing to the lack of scientific evidence and poor safety profile—they are commonly used to control exacerbations.31

Given that therapeutic algorithms are heterogeneous in terms of the value of systemic corticosteroids as a marker of moderate or severe forms of AD, we performed a sensitivity analysis in which both possibilities were considered. The results show that the decision on whether or not to prescribe these drugs—which may depend on prescribing habits at specific centers or among specific professionals—has little impact on the final classification (Figure 3).

The present study is subject to a series of limitations. Those associated with the classification of the disease as mild, moderate, and severe have been addressed above. However, weighting severity by treatment clearly reflects the impact of management on costs. Operative measurement of costs is associated with the geographical setting, which may limit extrapolation of our findings to health care systems with a different structure. As this is an observational study, its retrospective nature could favor underrecording of disease or potential variability resulting from differences between professionals and between patients. In this sense, potentially inaccurate coding in the diagnosis of moderate to severe and other comorbidities, or even the absence of variables that could affect the final results (eg, patients’ socioeconomic level, work, drug dose over time, adherence, causes of AD, phenotypes), should be considered a study limitation. However, we must draw attention to the high number of persons included over a limited time period that was well defined in all cases; such a circumstance would hardly be possible in a multicenter retrospective or prospective study carried out in the same setting based on collection of data by clinicians. Similarly, no data were provided on overlaps in medication between flare-ups and maintenance owing to the difficulty in measuring them. In addition, the external validity of the results (representativeness in the regional and national population), the evaluation of indirect costs (only for losses in work productivity), and the difference between moderate and severe AD should be considered possible study limitations.

Treatment of AD is expensive in terms of consumption of resources and social costs. However, the present study was not designed to collect health care outcomes associated with this cost. Several studies conclude that, despite the availability of treatment, some needs are not covered, especially in patients with more severe forms of AD.3,15 Therefore, while somewhat speculative, it seems feasible that resources were consumed inefficiently. Such a circumstance must be thoroughly addressed in the subgroup of patients with severe forms and greater consumption of resources. The arrival of new biologics, which are currently in advanced stages of development, makes it possible to improve the potential response in severe forms of the disease.21

The improvements brought about by these drugs, however, may be associated with an increase in the direct costs of the disease (drug costs), as has been observed in the case of psoriasis or other chronic inflammatory diseases, which should be weighted for improvements in global health care costs (number of visits, admissions to hospital, additional examinations) and social costs.32 It is therefore essential to know the baseline impact of the disease before such a paradigm shift is possible.

The present study reveals the notable impact of AD on health care and social costs, which is more marked in severe forms and in those that manifest alongside other conditions. Future studies should examine health outcomes arising from disease burden and assess uncovered needs in this chronic skin condition.

The study was sponsored by Sanofi.

A. Sicras is an independent consultant who received funding by Sanofi for work on this manuscript. José-Manuel Carrascosa has been a researcher in clinical trials promoted by Sanofi.