Peter A. Lio, MD

There is a well-established connection between the mind and the skin, and it is clear that this relationship is bidirectional—not only does skin disease increase the risk for depression, anxiety, sleep disturbance, and suicidality, but psychologic stress actually can worsen skin disease through multiple mechanisms, including direct damage to the skin barrier.^1,2 Psychologic stress also impacts the microbiome, another critical driver of skin disease.^3,4 The concept of the itch-scratch cycle vividly illustrates the vicious interplay between the mind and body in atopic dermatitis (AD).

However, patients with AD are not the only ones impacted—caregivers also experience psychologic stress. Remarkably, one study of patients with AD and their caregivers found that the caregivers actually reported significantly worse mental health and anxiety (P=.01 and P=.03, respectively) than patients themselves, even when controlling for the severity of disease.⁵

Thus, it would seem obvious for mental health to be a central component of AD care—to improve patient and caregiver quality of life while also improving symptoms. Research has actually borne this out, with one systematic review and meta-analysis concluding that psychological intervention has a beneficial effect on AD,⁶ and another that the addition of psychological and educational interventions to conventional treatment provided better therapeutic results in alleviating eczema severity and psychological symptoms.⁷ One study demonstrated that patients with AD who received cognitive behavioral therapy via the internet displayed a statistically significant improvement in their disease (P<.001) as measured by the Patient-Oriented Eczema Measure compared with those in the control group who received standard care alone. They also reported improvements in perceived stress, sleep problems, and depression in the intervention group that were sustained at 1-year follow-up.⁸ These findings are particularly impactful because clinical results were achieved while leveraging an internet-based approach to therapy.

Regrettably, despite the preponderance of evidence supporting the connection between mental health and AD, there remain considerable unmet needs. A recent cross-sectional survey of 954 adults with AD and caregivers of children with AD (N=954) conducted by the National Eczema Association found that half of patients were never asked about mental health during any of their visits, and of those referred for mental health resources, only 57% utilized the recommended services.⁹ Importantly, patients aged 18 to 34 years reported wanting to be asked about mental health. Of those who did receive referrals, most were for counseling services (23%), followed by alternative mental health therapy such as music or art therapy (15%), cognitive behavioral therapy (13%), or peer/social support groups (12%). Approximately 10% reported receiving a pamphlet or a brochure only.⁹

Physicians who treat patients with AD can and must do better, but first we must explore why these referral rates are so low. As with many complex problems, there is unlikely to be one simple unifying reason. As expected, the answer is nuanced and multifaceted, and—most importantly—staggeringly incomplete.

For starters, mental health interventions rarely are as easy as applying a cream or taking a pill. Hedman-Lagerlöf et al⁸ specifically pointed out that although their approach—using internet-based cognitive behavioral therapy—was explicitly designed to be more accessible with fewer resources, it required approximately 35 hours of treatment over 12 weeks, requiring both substantial time and commitment from patients who often are already burned out and exhausted due to AD. They even underscored that the most commonly reported adverse effect of therapy was increased stress or worry, making it a difficult sell.⁸

Even before most patients have a chance to consider the time required and the potential adverse effects of mental health interventions for AD, greater hurdles exist. Finances, medical insurance, and wait times were highlighted as barriers to care in a systematic review.¹⁰ These are deep-seated problems in the United States; while they may be surmountable in certain geo­graphic areas, the frequency with which these concerns arise means that it does not take too many failed attempts at referring patients for mental health services before clinicians just give up—similar to any form of operant conditioning.

A more elusive concept is stigmatization. Although it may not be quantifiable, the idea is that patients may encounter additional challenges when seeking mental health care, either because the interactions themselves may worsen their symptoms (eg, increased anxiety) or they may be more likely to have a negative perception of the experience.¹¹ A 2020 systematic review of barriers to addressing common mental health problems found that stigma was the most prominent barrier in adolescents, with the second most prominent being negative attitudes and beliefs about mental health services and professionals.¹² As a clinician, I can attest that I have sometimes detected skepticism when I have suggested mental health services to patients and have even been asked outright if I thought the problem was all in their head. My patients with AD generally have been much more open to the idea of mental health support, especially after I explain the powerful mind-body connection, than patients with other conditions—most notably delusions of parasitosis—who have been much more dismissive of such overtures. An oft-cited paper from 1976 frames the problem perfectly, describing what can happen after a referral for mental health services.¹³ The authors stated that the suggestion of mental health makes patients feel that the dermatologist does not believe them in the first place. Beyond this, the authors pointed out that referring the patient elsewhere reduces their hopes for dermatologic treatment.¹³

Knowing now—perhaps more than ever before—that the mind and skin are intimately connected compels us to solve these problems and find ways around these obstacles. Selecting the optimal forms of mental health services for each patient, having the structural support of the health care system, and winning the trust of patients and caregivers while combating stigma are undoubtedly tall orders; however, understanding the stakes for patients with AD, their caregivers, and society as a whole should inspire us to keep pushing forward.

There is a well-established connection between the mind and the skin, and it is clear that this relationship is bidirectional—not only does skin disease increase the risk for depression, anxiety, sleep disturbance, and suicidality, but psychologic stress actually can worsen skin disease through multiple mechanisms, including direct damage to the skin barrier.^1,2 Psychologic stress also impacts the microbiome, another critical driver of skin disease.^3,4 The concept of the itch-scratch cycle vividly illustrates the vicious interplay between the mind and body in atopic dermatitis (AD).

However, patients with AD are not the only ones impacted—caregivers also experience psychologic stress. Remarkably, one study of patients with AD and their caregivers found that the caregivers actually reported significantly worse mental health and anxiety (P=.01 and P=.03, respectively) than patients themselves, even when controlling for the severity of disease.⁵

Thus, it would seem obvious for mental health to be a central component of AD care—to improve patient and caregiver quality of life while also improving symptoms. Research has actually borne this out, with one systematic review and meta-analysis concluding that psychological intervention has a beneficial effect on AD,⁶ and another that the addition of psychological and educational interventions to conventional treatment provided better therapeutic results in alleviating eczema severity and psychological symptoms.⁷ One study demonstrated that patients with AD who received cognitive behavioral therapy via the internet displayed a statistically significant improvement in their disease (P<.001) as measured by the Patient-Oriented Eczema Measure compared with those in the control group who received standard care alone. They also reported improvements in perceived stress, sleep problems, and depression in the intervention group that were sustained at 1-year follow-up.⁸ These findings are particularly impactful because clinical results were achieved while leveraging an internet-based approach to therapy.

Regrettably, despite the preponderance of evidence supporting the connection between mental health and AD, there remain considerable unmet needs. A recent cross-sectional survey of 954 adults with AD and caregivers of children with AD (N=954) conducted by the National Eczema Association found that half of patients were never asked about mental health during any of their visits, and of those referred for mental health resources, only 57% utilized the recommended services.⁹ Importantly, patients aged 18 to 34 years reported wanting to be asked about mental health. Of those who did receive referrals, most were for counseling services (23%), followed by alternative mental health therapy such as music or art therapy (15%), cognitive behavioral therapy (13%), or peer/social support groups (12%). Approximately 10% reported receiving a pamphlet or a brochure only.⁹

Physicians who treat patients with AD can and must do better, but first we must explore why these referral rates are so low. As with many complex problems, there is unlikely to be one simple unifying reason. As expected, the answer is nuanced and multifaceted, and—most importantly—staggeringly incomplete.

For starters, mental health interventions rarely are as easy as applying a cream or taking a pill. Hedman-Lagerlöf et al⁸ specifically pointed out that although their approach—using internet-based cognitive behavioral therapy—was explicitly designed to be more accessible with fewer resources, it required approximately 35 hours of treatment over 12 weeks, requiring both substantial time and commitment from patients who often are already burned out and exhausted due to AD. They even underscored that the most commonly reported adverse effect of therapy was increased stress or worry, making it a difficult sell.⁸

Even before most patients have a chance to consider the time required and the potential adverse effects of mental health interventions for AD, greater hurdles exist. Finances, medical insurance, and wait times were highlighted as barriers to care in a systematic review.¹⁰ These are deep-seated problems in the United States; while they may be surmountable in certain geo­graphic areas, the frequency with which these concerns arise means that it does not take too many failed attempts at referring patients for mental health services before clinicians just give up—similar to any form of operant conditioning.

A more elusive concept is stigmatization. Although it may not be quantifiable, the idea is that patients may encounter additional challenges when seeking mental health care, either because the interactions themselves may worsen their symptoms (eg, increased anxiety) or they may be more likely to have a negative perception of the experience.¹¹ A 2020 systematic review of barriers to addressing common mental health problems found that stigma was the most prominent barrier in adolescents, with the second most prominent being negative attitudes and beliefs about mental health services and professionals.¹² As a clinician, I can attest that I have sometimes detected skepticism when I have suggested mental health services to patients and have even been asked outright if I thought the problem was all in their head. My patients with AD generally have been much more open to the idea of mental health support, especially after I explain the powerful mind-body connection, than patients with other conditions—most notably delusions of parasitosis—who have been much more dismissive of such overtures. An oft-cited paper from 1976 frames the problem perfectly, describing what can happen after a referral for mental health services.¹³ The authors stated that the suggestion of mental health makes patients feel that the dermatologist does not believe them in the first place. Beyond this, the authors pointed out that referring the patient elsewhere reduces their hopes for dermatologic treatment.¹³

Knowing now—perhaps more than ever before—that the mind and skin are intimately connected compels us to solve these problems and find ways around these obstacles. Selecting the optimal forms of mental health services for each patient, having the structural support of the health care system, and winning the trust of patients and caregivers while combating stigma are undoubtedly tall orders; however, understanding the stakes for patients with AD, their caregivers, and society as a whole should inspire us to keep pushing forward.

There is a well-established connection between the mind and the skin, and it is clear that this relationship is bidirectional—not only does skin disease increase the risk for depression, anxiety, sleep disturbance, and suicidality, but psychologic stress actually can worsen skin disease through multiple mechanisms, including direct damage to the skin barrier.^1,2 Psychologic stress also impacts the microbiome, another critical driver of skin disease.^3,4 The concept of the itch-scratch cycle vividly illustrates the vicious interplay between the mind and body in atopic dermatitis (AD).

However, patients with AD are not the only ones impacted—caregivers also experience psychologic stress. Remarkably, one study of patients with AD and their caregivers found that the caregivers actually reported significantly worse mental health and anxiety (P=.01 and P=.03, respectively) than patients themselves, even when controlling for the severity of disease.⁵

Thus, it would seem obvious for mental health to be a central component of AD care—to improve patient and caregiver quality of life while also improving symptoms. Research has actually borne this out, with one systematic review and meta-analysis concluding that psychological intervention has a beneficial effect on AD,⁶ and another that the addition of psychological and educational interventions to conventional treatment provided better therapeutic results in alleviating eczema severity and psychological symptoms.⁷ One study demonstrated that patients with AD who received cognitive behavioral therapy via the internet displayed a statistically significant improvement in their disease (P<.001) as measured by the Patient-Oriented Eczema Measure compared with those in the control group who received standard care alone. They also reported improvements in perceived stress, sleep problems, and depression in the intervention group that were sustained at 1-year follow-up.⁸ These findings are particularly impactful because clinical results were achieved while leveraging an internet-based approach to therapy.

Regrettably, despite the preponderance of evidence supporting the connection between mental health and AD, there remain considerable unmet needs. A recent cross-sectional survey of 954 adults with AD and caregivers of children with AD (N=954) conducted by the National Eczema Association found that half of patients were never asked about mental health during any of their visits, and of those referred for mental health resources, only 57% utilized the recommended services.⁹ Importantly, patients aged 18 to 34 years reported wanting to be asked about mental health. Of those who did receive referrals, most were for counseling services (23%), followed by alternative mental health therapy such as music or art therapy (15%), cognitive behavioral therapy (13%), or peer/social support groups (12%). Approximately 10% reported receiving a pamphlet or a brochure only.⁹

Physicians who treat patients with AD can and must do better, but first we must explore why these referral rates are so low. As with many complex problems, there is unlikely to be one simple unifying reason. As expected, the answer is nuanced and multifaceted, and—most importantly—staggeringly incomplete.

For starters, mental health interventions rarely are as easy as applying a cream or taking a pill. Hedman-Lagerlöf et al⁸ specifically pointed out that although their approach—using internet-based cognitive behavioral therapy—was explicitly designed to be more accessible with fewer resources, it required approximately 35 hours of treatment over 12 weeks, requiring both substantial time and commitment from patients who often are already burned out and exhausted due to AD. They even underscored that the most commonly reported adverse effect of therapy was increased stress or worry, making it a difficult sell.⁸

Even before most patients have a chance to consider the time required and the potential adverse effects of mental health interventions for AD, greater hurdles exist. Finances, medical insurance, and wait times were highlighted as barriers to care in a systematic review.¹⁰ These are deep-seated problems in the United States; while they may be surmountable in certain geo­graphic areas, the frequency with which these concerns arise means that it does not take too many failed attempts at referring patients for mental health services before clinicians just give up—similar to any form of operant conditioning.

A more elusive concept is stigmatization. Although it may not be quantifiable, the idea is that patients may encounter additional challenges when seeking mental health care, either because the interactions themselves may worsen their symptoms (eg, increased anxiety) or they may be more likely to have a negative perception of the experience.¹¹ A 2020 systematic review of barriers to addressing common mental health problems found that stigma was the most prominent barrier in adolescents, with the second most prominent being negative attitudes and beliefs about mental health services and professionals.¹² As a clinician, I can attest that I have sometimes detected skepticism when I have suggested mental health services to patients and have even been asked outright if I thought the problem was all in their head. My patients with AD generally have been much more open to the idea of mental health support, especially after I explain the powerful mind-body connection, than patients with other conditions—most notably delusions of parasitosis—who have been much more dismissive of such overtures. An oft-cited paper from 1976 frames the problem perfectly, describing what can happen after a referral for mental health services.¹³ The authors stated that the suggestion of mental health makes patients feel that the dermatologist does not believe them in the first place. Beyond this, the authors pointed out that referring the patient elsewhere reduces their hopes for dermatologic treatment.¹³

Knowing now—perhaps more than ever before—that the mind and skin are intimately connected compels us to solve these problems and find ways around these obstacles. Selecting the optimal forms of mental health services for each patient, having the structural support of the health care system, and winning the trust of patients and caregivers while combating stigma are undoubtedly tall orders; however, understanding the stakes for patients with AD, their caregivers, and society as a whole should inspire us to keep pushing forward.

It is unsurprising that food frequently is thought to be the culprit behind an eczema flare, especially in infants. Indeed, it often is said that infants do only 3 things: eat, sleep, and poop.¹ For those unfortunate enough to develop the signs and symptoms of atopic dermatitis (AD), food quickly emerges as a potential culprit from the tiny pool of suspects, which is against a cultural backdrop of unprecedented focus on foods and food reactions.² The prevalence of food allergies in children, though admittedly fraught with methodological difficulties, is estimated to have more than doubled from 3.4% in 1999 to 7.6% in 2018.³ As expected, prevalence rates were higher among children with other atopic comorbidities including AD, with up to 50% of children with AD demonstrating convincing food allergy.⁴ It is easy to imagine a patient conflating these 2 entities and mistaking their correlation for causation. Thus, it follows that more than 90% of parents/guardians have reported that their children have had food-induced AD, and understandably—at least according to one study—75% of parents/guardians were found to have manipulated the diet in an attempt to manage the disease.^5,6

Patients and parents/guardians are not the only ones who have suspected food as a driving force in AD. An article in the British Medical Journal from the 1800s beautifully encapsulated the depth and duration of this quandary: “There is probably no subject in which more deeply rooted convictions have been held, not only in the profession but by the laity, than the connection between diet and disease, both as regards the causation and treatment of the latter.”⁷ Herein, a wide range of food reactions is examined to highlight evidence for the role of diet in AD, which may contradict what patients—and even some clinicians—believe.

No Easy Answers

A definitive statement that food allergy is not the root cause of AD would put this issue to rest, but such simplicity does not reflect the complex reality. First, we must agree on definitions for certain terms. What do we mean by food allergy? A broader category—adverse food reactions—covers a wide range of entities, some immune mediated and some not, including lactose intolerance, irritant contact dermatitis around the mouth, and even dermatitis herpetiformis (the cutaneous manifestation of celiac disease).⁸ Although the term food allergy often is used synonymously with adverse food reactions, the exact definition of a food allergy is specific: “adverse immune responses to food proteins that result in typical clinical symptoms.”⁸ The fact that many patients and even health care practitioners seem to frequently misapply this term makes it even more confusing. 

The current focus is on foods that could trigger a flare of AD, which clearly is a broader question than food allergy sensu stricto. It seems self-evident, for example, that if an infant with AD were to (messily) eat an acidic food such as an orange, a flare-up of AD around the mouth and on the cheeks and hands would be a forgone conclusion. Similar nonimmunologic scenarios unambiguously can occur with many foods, including citrus; corn; radish; mustard; garlic; onion; pineapple; and many spices, food additives, and preservatives.⁹ Clearly there are some scenarios whereby food could trigger an AD flare, and yet this more limited vignette generally is not what patients are referring to when suggesting that food is the root cause of their AD.

The Labyrinth of Testing for Food Allergies

Although there is no reliable method for testing for irritant dermatitis, understanding the other types of tests may help guide our thinking. Testing for IgE-mediated food allergies generally is done via an immunoenzymatic serum assay that can document sensitization to a food protein; however, this testing by itself is not sufficient to diagnose a clinical food allergy.¹⁰ Similarly, skin prick testing allows for intradermal administration of a food extract to evaluate for an urticarial reaction within 10 to 15 minutes. Although the sensitivity and specificity vary by age, population, and the specific allergen being tested, these are limited to immediate-type reactions and do not reflect the potential to drive an eczematous flare.

The gold standard, if there is one, is likely the double-blind, placebo-controlled food challenge (DBPCFC), ideally with a long enough observation period to capture later-occurring reactions such as an AD flare. However, given the nature of the test—having patients eat the foods of concern and then carefully following them for reactions—it remains time consuming, expensive, and labor intensive.¹¹ 

To further complicate matters, several unvalidated tests exist such as IgG testing, atopy patch testing, kinesiology, and hair and gastric juice analysis, which remain investigational but continue to be used and may further confuse patients and clinicians.¹²

It is useful to first separate out the classic IgE-mediated food allergy reactions that are common. In these immediate-type reactions, a person sensitized to a food protein will develop characteristic cutaneous and/or extracutaneous reactions such as urticaria, angioedema, and even anaphylaxis, usually within minutes of exposure. Although it is possible that an IgE-mediated reaction could trigger an AD flare—perhaps simply by causing pruritus, which could initiate the itch-scratch cycle—because of the near simultaneity with ingestion of the offending food and the often dramatic clinical presentations, such foods clearly do not represent “hidden” triggers for AD flares.³ The concept of food-triggered AD (FTAD) is crucial for thinking about foods that could result in true eczematous flares, which historically have been classified as early-type (<2 hours after food challenge) and late-type (≥2 hours after food challenge) reactions.^13,14 

A study of more than 1000 DBPCFCs performed in patients with AD was illustrative.¹⁵ Immediate reactions other than AD were fairly common and were observed in 40% of the food challenges compared to only 9% in the placebo group. These reactions included urticaria, angioedema, and gastrointestinal and respiratory tract symptoms. Immediate reactions of AD alone were exceedingly rare at only 0.7% and not significantly elevated compared to placebo. Just over 4% experienced both an immediate AD exacerbation along with other non-AD findings, which was significantly greater than placebo (P<.01). Although intermediate and late reactions manifesting as AD exacerbations did occur after food ingestion, they were rare (2.2% or less) and not significantly different from placebo. The authors concluded that an exacerbation of AD in the absence of other allergic symptoms in children was unlikely to be due to food,¹⁵ which is an important finding.

A recent retrospective review of 372 children with AD reported similar results.⁴ The authors defined FTAD in a different way; instead of showing a flare after a DBPCFC, they looked for “physician-noted sustained improvement in AD upon removal of a food (typically after 2–6-wk follow-up), to which the child was sensitized without any other changes in skin care.” Despite this fundamentally different approach, they similarly concluded that while food allergies were common, FTAD was relatively uncommon—found in 2% of those with mild AD, 6% of those with moderate AD, and 4% of those with severe AD.⁴ 

There are other ways that foods could contribute to disease flares, however, and one of the most compelling is that there may be broader concepts at play; perhaps some diets are not specifically driving the AD but rather are affecting inflammation in the body at large. Although somewhat speculative, there is evidence that some foods may simply be proinflammatory, working to exacerbate the disease outside of a specific mechanism, which has been seen in a variety of other conditions such as acne or rheumatoid arthritis.^16,17 To speculate further, it is possible that there may be a threshold effect such that when the AD is poorly controlled, certain factors such as inflammatory foods could lead to a flare, while when under better control, these same factors may not cause an effect.

Finally, it is important to also consider the emotional and/or psychological aspects related to food and diet. The power of the placebo in dietary change has been documented in several diseases, though this certainly is not to be dismissive of the patient’s symptoms; it seems reasonable that the very act of changing such a fundamental aspect of daily life could result in a placebo effect.^18,19 In the context of relapsing and remitting conditions such as AD, this effect may be magnified. A landmark study by Thompson and Hanifin²⁰ illustrates this possibility. The authors found that in 80% of cases in which patients were convinced that food was a major contributing factor to their AD, such concerns diminished markedly once better control of the eczema was achieved.²⁰

This brings us to what to do with an individual patient in the examination room. Because there is such widespread concern and discussion around this topic, it is important to at least briefly address it. If there are known food allergens that are being avoided, it is important to underscore the importance of continuing to avoid those foods, especially when there is actual evidence of true food allergy rather than sensitization alone. Historically, elimination diets often were recommended empirically, though more recent studies, meta-analyses, and guidance documents increasingly have recommended against them.³ In particular, there are major concerns for iatrogenic harm. 

First, heavily restricted diets may result in nutritional and/or caloric deficiencies that can be dangerous and lead to poor growth.²¹ Practices such as drinking unpasteurized milk can expose children to dangerous infections, while feeding them exclusively rice milk can lead to severe malnutrition.²² 

Second, there is a dawning realization that children with AD placed on elimination diets may actually develop true IgE-mediated allergies, including fatal anaphylaxis, to the excluded foods. In fact, one retrospective review of 298 patients with a history of AD and no prior immediate reactions found that 19% of patients developed new immediate-type hypersensitivity reactions after starting an elimination diet, presumably due to the loss of tolerance to these foods. A striking one-third of these reactions were classified as anaphylaxis, with cow’s milk and egg being the most common offenders.²³

It also is crucial to acknowledge that recommending sweeping lifestyle changes is not easy for patients, especially pediatric patients. Onerous dietary restrictions may add considerable stress, ironically a known trigger for AD itself. 

Finally, dietary modifications can be a distraction from conventional therapy and may result in treatment delays while the patient continues to experience uncontrolled symptoms of AD. 

Final Thoughts

Diet is intimately related to AD. Although the narrative continues to unfold in fascinating domains, such as the skin barrier and the microbiome, it is increasingly clear that these are intertwined and always have been. Despite the rarity of true food-triggered AD, the perception of dietary triggers is so widespread and addressing the topic is important and may help avoid unnecessary harm from unfounded extreme dietary changes. A recent multispecialty workgroup report on AD and food allergy succinctly summarized this as: “AD has many triggers and comorbidities, and food allergy is only one of the potential triggers and comorbid conditions. With regard to AD management, education and skin care are most important.”³ With proper testing, guidance, and both topical and systemic therapies, most AD can be brought under control, and for at least some patients, this may allay concerns about foods triggering their AD. 

It is unsurprising that food frequently is thought to be the culprit behind an eczema flare, especially in infants. Indeed, it often is said that infants do only 3 things: eat, sleep, and poop.¹ For those unfortunate enough to develop the signs and symptoms of atopic dermatitis (AD), food quickly emerges as a potential culprit from the tiny pool of suspects, which is against a cultural backdrop of unprecedented focus on foods and food reactions.² The prevalence of food allergies in children, though admittedly fraught with methodological difficulties, is estimated to have more than doubled from 3.4% in 1999 to 7.6% in 2018.³ As expected, prevalence rates were higher among children with other atopic comorbidities including AD, with up to 50% of children with AD demonstrating convincing food allergy.⁴ It is easy to imagine a patient conflating these 2 entities and mistaking their correlation for causation. Thus, it follows that more than 90% of parents/guardians have reported that their children have had food-induced AD, and understandably—at least according to one study—75% of parents/guardians were found to have manipulated the diet in an attempt to manage the disease.^5,6

Patients and parents/guardians are not the only ones who have suspected food as a driving force in AD. An article in the British Medical Journal from the 1800s beautifully encapsulated the depth and duration of this quandary: “There is probably no subject in which more deeply rooted convictions have been held, not only in the profession but by the laity, than the connection between diet and disease, both as regards the causation and treatment of the latter.”⁷ Herein, a wide range of food reactions is examined to highlight evidence for the role of diet in AD, which may contradict what patients—and even some clinicians—believe.

No Easy Answers

A definitive statement that food allergy is not the root cause of AD would put this issue to rest, but such simplicity does not reflect the complex reality. First, we must agree on definitions for certain terms. What do we mean by food allergy? A broader category—adverse food reactions—covers a wide range of entities, some immune mediated and some not, including lactose intolerance, irritant contact dermatitis around the mouth, and even dermatitis herpetiformis (the cutaneous manifestation of celiac disease).⁸ Although the term food allergy often is used synonymously with adverse food reactions, the exact definition of a food allergy is specific: “adverse immune responses to food proteins that result in typical clinical symptoms.”⁸ The fact that many patients and even health care practitioners seem to frequently misapply this term makes it even more confusing. 

The current focus is on foods that could trigger a flare of AD, which clearly is a broader question than food allergy sensu stricto. It seems self-evident, for example, that if an infant with AD were to (messily) eat an acidic food such as an orange, a flare-up of AD around the mouth and on the cheeks and hands would be a forgone conclusion. Similar nonimmunologic scenarios unambiguously can occur with many foods, including citrus; corn; radish; mustard; garlic; onion; pineapple; and many spices, food additives, and preservatives.⁹ Clearly there are some scenarios whereby food could trigger an AD flare, and yet this more limited vignette generally is not what patients are referring to when suggesting that food is the root cause of their AD.

The Labyrinth of Testing for Food Allergies

Although there is no reliable method for testing for irritant dermatitis, understanding the other types of tests may help guide our thinking. Testing for IgE-mediated food allergies generally is done via an immunoenzymatic serum assay that can document sensitization to a food protein; however, this testing by itself is not sufficient to diagnose a clinical food allergy.¹⁰ Similarly, skin prick testing allows for intradermal administration of a food extract to evaluate for an urticarial reaction within 10 to 15 minutes. Although the sensitivity and specificity vary by age, population, and the specific allergen being tested, these are limited to immediate-type reactions and do not reflect the potential to drive an eczematous flare.

The gold standard, if there is one, is likely the double-blind, placebo-controlled food challenge (DBPCFC), ideally with a long enough observation period to capture later-occurring reactions such as an AD flare. However, given the nature of the test—having patients eat the foods of concern and then carefully following them for reactions—it remains time consuming, expensive, and labor intensive.¹¹ 

To further complicate matters, several unvalidated tests exist such as IgG testing, atopy patch testing, kinesiology, and hair and gastric juice analysis, which remain investigational but continue to be used and may further confuse patients and clinicians.¹²

It is useful to first separate out the classic IgE-mediated food allergy reactions that are common. In these immediate-type reactions, a person sensitized to a food protein will develop characteristic cutaneous and/or extracutaneous reactions such as urticaria, angioedema, and even anaphylaxis, usually within minutes of exposure. Although it is possible that an IgE-mediated reaction could trigger an AD flare—perhaps simply by causing pruritus, which could initiate the itch-scratch cycle—because of the near simultaneity with ingestion of the offending food and the often dramatic clinical presentations, such foods clearly do not represent “hidden” triggers for AD flares.³ The concept of food-triggered AD (FTAD) is crucial for thinking about foods that could result in true eczematous flares, which historically have been classified as early-type (<2 hours after food challenge) and late-type (≥2 hours after food challenge) reactions.^13,14 

A study of more than 1000 DBPCFCs performed in patients with AD was illustrative.¹⁵ Immediate reactions other than AD were fairly common and were observed in 40% of the food challenges compared to only 9% in the placebo group. These reactions included urticaria, angioedema, and gastrointestinal and respiratory tract symptoms. Immediate reactions of AD alone were exceedingly rare at only 0.7% and not significantly elevated compared to placebo. Just over 4% experienced both an immediate AD exacerbation along with other non-AD findings, which was significantly greater than placebo (P<.01). Although intermediate and late reactions manifesting as AD exacerbations did occur after food ingestion, they were rare (2.2% or less) and not significantly different from placebo. The authors concluded that an exacerbation of AD in the absence of other allergic symptoms in children was unlikely to be due to food,¹⁵ which is an important finding.

A recent retrospective review of 372 children with AD reported similar results.⁴ The authors defined FTAD in a different way; instead of showing a flare after a DBPCFC, they looked for “physician-noted sustained improvement in AD upon removal of a food (typically after 2–6-wk follow-up), to which the child was sensitized without any other changes in skin care.” Despite this fundamentally different approach, they similarly concluded that while food allergies were common, FTAD was relatively uncommon—found in 2% of those with mild AD, 6% of those with moderate AD, and 4% of those with severe AD.⁴ 

There are other ways that foods could contribute to disease flares, however, and one of the most compelling is that there may be broader concepts at play; perhaps some diets are not specifically driving the AD but rather are affecting inflammation in the body at large. Although somewhat speculative, there is evidence that some foods may simply be proinflammatory, working to exacerbate the disease outside of a specific mechanism, which has been seen in a variety of other conditions such as acne or rheumatoid arthritis.^16,17 To speculate further, it is possible that there may be a threshold effect such that when the AD is poorly controlled, certain factors such as inflammatory foods could lead to a flare, while when under better control, these same factors may not cause an effect.

Finally, it is important to also consider the emotional and/or psychological aspects related to food and diet. The power of the placebo in dietary change has been documented in several diseases, though this certainly is not to be dismissive of the patient’s symptoms; it seems reasonable that the very act of changing such a fundamental aspect of daily life could result in a placebo effect.^18,19 In the context of relapsing and remitting conditions such as AD, this effect may be magnified. A landmark study by Thompson and Hanifin²⁰ illustrates this possibility. The authors found that in 80% of cases in which patients were convinced that food was a major contributing factor to their AD, such concerns diminished markedly once better control of the eczema was achieved.²⁰

This brings us to what to do with an individual patient in the examination room. Because there is such widespread concern and discussion around this topic, it is important to at least briefly address it. If there are known food allergens that are being avoided, it is important to underscore the importance of continuing to avoid those foods, especially when there is actual evidence of true food allergy rather than sensitization alone. Historically, elimination diets often were recommended empirically, though more recent studies, meta-analyses, and guidance documents increasingly have recommended against them.³ In particular, there are major concerns for iatrogenic harm. 

First, heavily restricted diets may result in nutritional and/or caloric deficiencies that can be dangerous and lead to poor growth.²¹ Practices such as drinking unpasteurized milk can expose children to dangerous infections, while feeding them exclusively rice milk can lead to severe malnutrition.²² 

Second, there is a dawning realization that children with AD placed on elimination diets may actually develop true IgE-mediated allergies, including fatal anaphylaxis, to the excluded foods. In fact, one retrospective review of 298 patients with a history of AD and no prior immediate reactions found that 19% of patients developed new immediate-type hypersensitivity reactions after starting an elimination diet, presumably due to the loss of tolerance to these foods. A striking one-third of these reactions were classified as anaphylaxis, with cow’s milk and egg being the most common offenders.²³

It also is crucial to acknowledge that recommending sweeping lifestyle changes is not easy for patients, especially pediatric patients. Onerous dietary restrictions may add considerable stress, ironically a known trigger for AD itself. 

Finally, dietary modifications can be a distraction from conventional therapy and may result in treatment delays while the patient continues to experience uncontrolled symptoms of AD. 

Final Thoughts

Diet is intimately related to AD. Although the narrative continues to unfold in fascinating domains, such as the skin barrier and the microbiome, it is increasingly clear that these are intertwined and always have been. Despite the rarity of true food-triggered AD, the perception of dietary triggers is so widespread and addressing the topic is important and may help avoid unnecessary harm from unfounded extreme dietary changes. A recent multispecialty workgroup report on AD and food allergy succinctly summarized this as: “AD has many triggers and comorbidities, and food allergy is only one of the potential triggers and comorbid conditions. With regard to AD management, education and skin care are most important.”³ With proper testing, guidance, and both topical and systemic therapies, most AD can be brought under control, and for at least some patients, this may allay concerns about foods triggering their AD. 

It is unsurprising that food frequently is thought to be the culprit behind an eczema flare, especially in infants. Indeed, it often is said that infants do only 3 things: eat, sleep, and poop.¹ For those unfortunate enough to develop the signs and symptoms of atopic dermatitis (AD), food quickly emerges as a potential culprit from the tiny pool of suspects, which is against a cultural backdrop of unprecedented focus on foods and food reactions.² The prevalence of food allergies in children, though admittedly fraught with methodological difficulties, is estimated to have more than doubled from 3.4% in 1999 to 7.6% in 2018.³ As expected, prevalence rates were higher among children with other atopic comorbidities including AD, with up to 50% of children with AD demonstrating convincing food allergy.⁴ It is easy to imagine a patient conflating these 2 entities and mistaking their correlation for causation. Thus, it follows that more than 90% of parents/guardians have reported that their children have had food-induced AD, and understandably—at least according to one study—75% of parents/guardians were found to have manipulated the diet in an attempt to manage the disease.^5,6

Patients and parents/guardians are not the only ones who have suspected food as a driving force in AD. An article in the British Medical Journal from the 1800s beautifully encapsulated the depth and duration of this quandary: “There is probably no subject in which more deeply rooted convictions have been held, not only in the profession but by the laity, than the connection between diet and disease, both as regards the causation and treatment of the latter.”⁷ Herein, a wide range of food reactions is examined to highlight evidence for the role of diet in AD, which may contradict what patients—and even some clinicians—believe.

No Easy Answers

A definitive statement that food allergy is not the root cause of AD would put this issue to rest, but such simplicity does not reflect the complex reality. First, we must agree on definitions for certain terms. What do we mean by food allergy? A broader category—adverse food reactions—covers a wide range of entities, some immune mediated and some not, including lactose intolerance, irritant contact dermatitis around the mouth, and even dermatitis herpetiformis (the cutaneous manifestation of celiac disease).⁸ Although the term food allergy often is used synonymously with adverse food reactions, the exact definition of a food allergy is specific: “adverse immune responses to food proteins that result in typical clinical symptoms.”⁸ The fact that many patients and even health care practitioners seem to frequently misapply this term makes it even more confusing. 

The current focus is on foods that could trigger a flare of AD, which clearly is a broader question than food allergy sensu stricto. It seems self-evident, for example, that if an infant with AD were to (messily) eat an acidic food such as an orange, a flare-up of AD around the mouth and on the cheeks and hands would be a forgone conclusion. Similar nonimmunologic scenarios unambiguously can occur with many foods, including citrus; corn; radish; mustard; garlic; onion; pineapple; and many spices, food additives, and preservatives.⁹ Clearly there are some scenarios whereby food could trigger an AD flare, and yet this more limited vignette generally is not what patients are referring to when suggesting that food is the root cause of their AD.

The Labyrinth of Testing for Food Allergies

Although there is no reliable method for testing for irritant dermatitis, understanding the other types of tests may help guide our thinking. Testing for IgE-mediated food allergies generally is done via an immunoenzymatic serum assay that can document sensitization to a food protein; however, this testing by itself is not sufficient to diagnose a clinical food allergy.¹⁰ Similarly, skin prick testing allows for intradermal administration of a food extract to evaluate for an urticarial reaction within 10 to 15 minutes. Although the sensitivity and specificity vary by age, population, and the specific allergen being tested, these are limited to immediate-type reactions and do not reflect the potential to drive an eczematous flare.

The gold standard, if there is one, is likely the double-blind, placebo-controlled food challenge (DBPCFC), ideally with a long enough observation period to capture later-occurring reactions such as an AD flare. However, given the nature of the test—having patients eat the foods of concern and then carefully following them for reactions—it remains time consuming, expensive, and labor intensive.¹¹ 

To further complicate matters, several unvalidated tests exist such as IgG testing, atopy patch testing, kinesiology, and hair and gastric juice analysis, which remain investigational but continue to be used and may further confuse patients and clinicians.¹²

It is useful to first separate out the classic IgE-mediated food allergy reactions that are common. In these immediate-type reactions, a person sensitized to a food protein will develop characteristic cutaneous and/or extracutaneous reactions such as urticaria, angioedema, and even anaphylaxis, usually within minutes of exposure. Although it is possible that an IgE-mediated reaction could trigger an AD flare—perhaps simply by causing pruritus, which could initiate the itch-scratch cycle—because of the near simultaneity with ingestion of the offending food and the often dramatic clinical presentations, such foods clearly do not represent “hidden” triggers for AD flares.³ The concept of food-triggered AD (FTAD) is crucial for thinking about foods that could result in true eczematous flares, which historically have been classified as early-type (<2 hours after food challenge) and late-type (≥2 hours after food challenge) reactions.^13,14 

A study of more than 1000 DBPCFCs performed in patients with AD was illustrative.¹⁵ Immediate reactions other than AD were fairly common and were observed in 40% of the food challenges compared to only 9% in the placebo group. These reactions included urticaria, angioedema, and gastrointestinal and respiratory tract symptoms. Immediate reactions of AD alone were exceedingly rare at only 0.7% and not significantly elevated compared to placebo. Just over 4% experienced both an immediate AD exacerbation along with other non-AD findings, which was significantly greater than placebo (P<.01). Although intermediate and late reactions manifesting as AD exacerbations did occur after food ingestion, they were rare (2.2% or less) and not significantly different from placebo. The authors concluded that an exacerbation of AD in the absence of other allergic symptoms in children was unlikely to be due to food,¹⁵ which is an important finding.

A recent retrospective review of 372 children with AD reported similar results.⁴ The authors defined FTAD in a different way; instead of showing a flare after a DBPCFC, they looked for “physician-noted sustained improvement in AD upon removal of a food (typically after 2–6-wk follow-up), to which the child was sensitized without any other changes in skin care.” Despite this fundamentally different approach, they similarly concluded that while food allergies were common, FTAD was relatively uncommon—found in 2% of those with mild AD, 6% of those with moderate AD, and 4% of those with severe AD.⁴ 

There are other ways that foods could contribute to disease flares, however, and one of the most compelling is that there may be broader concepts at play; perhaps some diets are not specifically driving the AD but rather are affecting inflammation in the body at large. Although somewhat speculative, there is evidence that some foods may simply be proinflammatory, working to exacerbate the disease outside of a specific mechanism, which has been seen in a variety of other conditions such as acne or rheumatoid arthritis.^16,17 To speculate further, it is possible that there may be a threshold effect such that when the AD is poorly controlled, certain factors such as inflammatory foods could lead to a flare, while when under better control, these same factors may not cause an effect.

Finally, it is important to also consider the emotional and/or psychological aspects related to food and diet. The power of the placebo in dietary change has been documented in several diseases, though this certainly is not to be dismissive of the patient’s symptoms; it seems reasonable that the very act of changing such a fundamental aspect of daily life could result in a placebo effect.^18,19 In the context of relapsing and remitting conditions such as AD, this effect may be magnified. A landmark study by Thompson and Hanifin²⁰ illustrates this possibility. The authors found that in 80% of cases in which patients were convinced that food was a major contributing factor to their AD, such concerns diminished markedly once better control of the eczema was achieved.²⁰

This brings us to what to do with an individual patient in the examination room. Because there is such widespread concern and discussion around this topic, it is important to at least briefly address it. If there are known food allergens that are being avoided, it is important to underscore the importance of continuing to avoid those foods, especially when there is actual evidence of true food allergy rather than sensitization alone. Historically, elimination diets often were recommended empirically, though more recent studies, meta-analyses, and guidance documents increasingly have recommended against them.³ In particular, there are major concerns for iatrogenic harm. 

First, heavily restricted diets may result in nutritional and/or caloric deficiencies that can be dangerous and lead to poor growth.²¹ Practices such as drinking unpasteurized milk can expose children to dangerous infections, while feeding them exclusively rice milk can lead to severe malnutrition.²² 

Second, there is a dawning realization that children with AD placed on elimination diets may actually develop true IgE-mediated allergies, including fatal anaphylaxis, to the excluded foods. In fact, one retrospective review of 298 patients with a history of AD and no prior immediate reactions found that 19% of patients developed new immediate-type hypersensitivity reactions after starting an elimination diet, presumably due to the loss of tolerance to these foods. A striking one-third of these reactions were classified as anaphylaxis, with cow’s milk and egg being the most common offenders.²³

It also is crucial to acknowledge that recommending sweeping lifestyle changes is not easy for patients, especially pediatric patients. Onerous dietary restrictions may add considerable stress, ironically a known trigger for AD itself. 

Finally, dietary modifications can be a distraction from conventional therapy and may result in treatment delays while the patient continues to experience uncontrolled symptoms of AD. 

Final Thoughts

Diet is intimately related to AD. Although the narrative continues to unfold in fascinating domains, such as the skin barrier and the microbiome, it is increasingly clear that these are intertwined and always have been. Despite the rarity of true food-triggered AD, the perception of dietary triggers is so widespread and addressing the topic is important and may help avoid unnecessary harm from unfounded extreme dietary changes. A recent multispecialty workgroup report on AD and food allergy succinctly summarized this as: “AD has many triggers and comorbidities, and food allergy is only one of the potential triggers and comorbid conditions. With regard to AD management, education and skin care are most important.”³ With proper testing, guidance, and both topical and systemic therapies, most AD can be brought under control, and for at least some patients, this may allay concerns about foods triggering their AD. 

In patients with persistent atopic dermatitis (AD) who are taking dupilumab, is there benefit of patch testing to determine if allergic contact dermatitis (ACD) also is contributing to their disease? Results of patch testing are likely be influenced by the immunomodulatory effects of dupilumab. Similar to the recommendation for patients to refrain from using topical or systemic corticosteroids for 1 week or more prior to patch testing to eliminate false negatives, we reviewed the literature to create practice guidelines for dermatologists regarding patch testing while a patient is taking dupilumab.

Pathophysiology and Pathomechanism

Dupilumab functions through the blockade of T helper 2 (T_H2) cells; ACD is propagated through the T helper 1 (T_H1) cellular pathway. However, patients with ACD that is unresponsive to allergen avoidance and traditional therapies, such as topical and oral corticosteroids, have responded to dupilumab. The more common reports of this responsiveness are with fragrances; multiple case series described patients with ACD to fragrance mix I¹ and balsam of Peru^1,2 who improved on dupilumab when other treatments failed. There also are reports of response when ACD was secondary to nickel,^2,3p-phenylenediamine,¹ Compositae,⁴ and non–formaldehyde-releasing preservatives (non-FRPs).⁵ Therefore, not all ACD is propagated through the T_H1 cellular pathway.

As noted in these cases, ACD can be a response to an allergen whose pathogenesis involves the T_H2 pathway or when patient characteristics favor a T_H2 response. It has been suggested that AD patients are more susceptible to T_H2-mediated contact sensitization to less-potent allergens, such as fragrances.⁶

Patch Test Results

Positive patch test results for allergens have been reported while patients are on dupilumab therapy, including a few studies in which results prior to starting dupilumab were compared with those while patients were on dupilumab therapy. In a retrospective chart review of 48 patients on dupilumab for AD with persistent disease, 23 patients were patch tested before and during dupilumab therapy. In these patients, the majority of contact allergies were persistent and only 10% (13/125) of patch test–positive results resolved on dupilumab therapy.⁷ Contact allergies that resolved included those to emulsifiers (propylene glycol, Amerchol L101 [lanolin-containing products found in cosmetics and other goods], dimethylaminopropylamine), fragrances (fragrance mix I, balsam of Peru), sunscreens (sulisobenzone, phenylbenzimidazole-5-sulfonic acid), and metals (vanadium chloride, phenylmercuric acetate).⁷ The following results observed in individual cases demonstrated conflicting findings: persistence of allergy to non-FRPs (methylisothiazolinone [MI]) but resolution of allergy to formaldehyde⁸; persistence of allergy to corticosteroids (budesonide and alclometasone)⁹; persistence of allergy to an antibiotic (neomycin sulfate) but resolution of allergies to a different antibiotic (bacitracin), glues (ethyl acrylate), bleach, and glutaraldehyde⁹; persistence of nickel allergy but resolution of allergies to fragrances (cinnamic aldehyde, balsam of Peru) and non-FRPs (methylchloroisothiazolinone or MI)¹⁰; and persistence of allergies to non-FRPs (MI) and FRPs (bronopol) but resolution of allergies to nickel, fragrances (hydroperoxides of linalool), and Compositae.¹¹ Additional case reports of positive patch test results while on dupilumab but with no pretreatment results for comparison include allergies to rubber additives,^12-14 nickel,¹⁴ textile dyes,¹⁴ cosmetic and hair care additives,^12,14,15 corticosteroids,¹⁵ FRPs,¹⁵ fragrances,^15,16 emulsifiers,¹⁶ and non-FRPs.¹⁷

An evident theme in the dupilumab patch-testing literature has been that results are variable and case specific: a given patient with ACD to an allergen will respond to dupilumab treatment and have subsequent negative patch testing, while another patient will not respond to dupilumab treatment and have persistent positive patch testing. This is likely because, in certain individuals, the allergen-immune system combination shifts ACD pathogenesis from a purely T_H1 response to at least a partial T_H2 response, thus allowing for benefit from dupilumab therapy. T helper 1 cell–mediated ACD should not be affected by dupilumab; therefore, reliable results can be elucidated from patch testing despite the drug.

Final Thoughts

We propose that AD patients with residual disease after taking dupilumab undergo patch testing. Positive results indicate allergens that are not inhibited by the drug. Patients will need to follow strict allergen avoidance to resolve this component of their disease; failure to improve might suggest the result was a nonrelevant positive.

If patch testing is negative, an alternative cause for residual disease must be sought. We do not recommend stopping dupilumab prior to patch testing to avoid a disease flare from AD or possible T_H2-mediated ACD.

In patients with persistent atopic dermatitis (AD) who are taking dupilumab, is there benefit of patch testing to determine if allergic contact dermatitis (ACD) also is contributing to their disease? Results of patch testing are likely be influenced by the immunomodulatory effects of dupilumab. Similar to the recommendation for patients to refrain from using topical or systemic corticosteroids for 1 week or more prior to patch testing to eliminate false negatives, we reviewed the literature to create practice guidelines for dermatologists regarding patch testing while a patient is taking dupilumab.

Pathophysiology and Pathomechanism

Dupilumab functions through the blockade of T helper 2 (T_H2) cells; ACD is propagated through the T helper 1 (T_H1) cellular pathway. However, patients with ACD that is unresponsive to allergen avoidance and traditional therapies, such as topical and oral corticosteroids, have responded to dupilumab. The more common reports of this responsiveness are with fragrances; multiple case series described patients with ACD to fragrance mix I¹ and balsam of Peru^1,2 who improved on dupilumab when other treatments failed. There also are reports of response when ACD was secondary to nickel,^2,3p-phenylenediamine,¹ Compositae,⁴ and non–formaldehyde-releasing preservatives (non-FRPs).⁵ Therefore, not all ACD is propagated through the T_H1 cellular pathway.

As noted in these cases, ACD can be a response to an allergen whose pathogenesis involves the T_H2 pathway or when patient characteristics favor a T_H2 response. It has been suggested that AD patients are more susceptible to T_H2-mediated contact sensitization to less-potent allergens, such as fragrances.⁶

Patch Test Results

Positive patch test results for allergens have been reported while patients are on dupilumab therapy, including a few studies in which results prior to starting dupilumab were compared with those while patients were on dupilumab therapy. In a retrospective chart review of 48 patients on dupilumab for AD with persistent disease, 23 patients were patch tested before and during dupilumab therapy. In these patients, the majority of contact allergies were persistent and only 10% (13/125) of patch test–positive results resolved on dupilumab therapy.⁷ Contact allergies that resolved included those to emulsifiers (propylene glycol, Amerchol L101 [lanolin-containing products found in cosmetics and other goods], dimethylaminopropylamine), fragrances (fragrance mix I, balsam of Peru), sunscreens (sulisobenzone, phenylbenzimidazole-5-sulfonic acid), and metals (vanadium chloride, phenylmercuric acetate).⁷ The following results observed in individual cases demonstrated conflicting findings: persistence of allergy to non-FRPs (methylisothiazolinone [MI]) but resolution of allergy to formaldehyde⁸; persistence of allergy to corticosteroids (budesonide and alclometasone)⁹; persistence of allergy to an antibiotic (neomycin sulfate) but resolution of allergies to a different antibiotic (bacitracin), glues (ethyl acrylate), bleach, and glutaraldehyde⁹; persistence of nickel allergy but resolution of allergies to fragrances (cinnamic aldehyde, balsam of Peru) and non-FRPs (methylchloroisothiazolinone or MI)¹⁰; and persistence of allergies to non-FRPs (MI) and FRPs (bronopol) but resolution of allergies to nickel, fragrances (hydroperoxides of linalool), and Compositae.¹¹ Additional case reports of positive patch test results while on dupilumab but with no pretreatment results for comparison include allergies to rubber additives,^12-14 nickel,¹⁴ textile dyes,¹⁴ cosmetic and hair care additives,^12,14,15 corticosteroids,¹⁵ FRPs,¹⁵ fragrances,^15,16 emulsifiers,¹⁶ and non-FRPs.¹⁷

An evident theme in the dupilumab patch-testing literature has been that results are variable and case specific: a given patient with ACD to an allergen will respond to dupilumab treatment and have subsequent negative patch testing, while another patient will not respond to dupilumab treatment and have persistent positive patch testing. This is likely because, in certain individuals, the allergen-immune system combination shifts ACD pathogenesis from a purely T_H1 response to at least a partial T_H2 response, thus allowing for benefit from dupilumab therapy. T helper 1 cell–mediated ACD should not be affected by dupilumab; therefore, reliable results can be elucidated from patch testing despite the drug.

Final Thoughts

We propose that AD patients with residual disease after taking dupilumab undergo patch testing. Positive results indicate allergens that are not inhibited by the drug. Patients will need to follow strict allergen avoidance to resolve this component of their disease; failure to improve might suggest the result was a nonrelevant positive.

If patch testing is negative, an alternative cause for residual disease must be sought. We do not recommend stopping dupilumab prior to patch testing to avoid a disease flare from AD or possible T_H2-mediated ACD.

In patients with persistent atopic dermatitis (AD) who are taking dupilumab, is there benefit of patch testing to determine if allergic contact dermatitis (ACD) also is contributing to their disease? Results of patch testing are likely be influenced by the immunomodulatory effects of dupilumab. Similar to the recommendation for patients to refrain from using topical or systemic corticosteroids for 1 week or more prior to patch testing to eliminate false negatives, we reviewed the literature to create practice guidelines for dermatologists regarding patch testing while a patient is taking dupilumab.

Pathophysiology and Pathomechanism

Dupilumab functions through the blockade of T helper 2 (T_H2) cells; ACD is propagated through the T helper 1 (T_H1) cellular pathway. However, patients with ACD that is unresponsive to allergen avoidance and traditional therapies, such as topical and oral corticosteroids, have responded to dupilumab. The more common reports of this responsiveness are with fragrances; multiple case series described patients with ACD to fragrance mix I¹ and balsam of Peru^1,2 who improved on dupilumab when other treatments failed. There also are reports of response when ACD was secondary to nickel,^2,3p-phenylenediamine,¹ Compositae,⁴ and non–formaldehyde-releasing preservatives (non-FRPs).⁵ Therefore, not all ACD is propagated through the T_H1 cellular pathway.

As noted in these cases, ACD can be a response to an allergen whose pathogenesis involves the T_H2 pathway or when patient characteristics favor a T_H2 response. It has been suggested that AD patients are more susceptible to T_H2-mediated contact sensitization to less-potent allergens, such as fragrances.⁶

Patch Test Results

Positive patch test results for allergens have been reported while patients are on dupilumab therapy, including a few studies in which results prior to starting dupilumab were compared with those while patients were on dupilumab therapy. In a retrospective chart review of 48 patients on dupilumab for AD with persistent disease, 23 patients were patch tested before and during dupilumab therapy. In these patients, the majority of contact allergies were persistent and only 10% (13/125) of patch test–positive results resolved on dupilumab therapy.⁷ Contact allergies that resolved included those to emulsifiers (propylene glycol, Amerchol L101 [lanolin-containing products found in cosmetics and other goods], dimethylaminopropylamine), fragrances (fragrance mix I, balsam of Peru), sunscreens (sulisobenzone, phenylbenzimidazole-5-sulfonic acid), and metals (vanadium chloride, phenylmercuric acetate).⁷ The following results observed in individual cases demonstrated conflicting findings: persistence of allergy to non-FRPs (methylisothiazolinone [MI]) but resolution of allergy to formaldehyde⁸; persistence of allergy to corticosteroids (budesonide and alclometasone)⁹; persistence of allergy to an antibiotic (neomycin sulfate) but resolution of allergies to a different antibiotic (bacitracin), glues (ethyl acrylate), bleach, and glutaraldehyde⁹; persistence of nickel allergy but resolution of allergies to fragrances (cinnamic aldehyde, balsam of Peru) and non-FRPs (methylchloroisothiazolinone or MI)¹⁰; and persistence of allergies to non-FRPs (MI) and FRPs (bronopol) but resolution of allergies to nickel, fragrances (hydroperoxides of linalool), and Compositae.¹¹ Additional case reports of positive patch test results while on dupilumab but with no pretreatment results for comparison include allergies to rubber additives,^12-14 nickel,¹⁴ textile dyes,¹⁴ cosmetic and hair care additives,^12,14,15 corticosteroids,¹⁵ FRPs,¹⁵ fragrances,^15,16 emulsifiers,¹⁶ and non-FRPs.¹⁷

An evident theme in the dupilumab patch-testing literature has been that results are variable and case specific: a given patient with ACD to an allergen will respond to dupilumab treatment and have subsequent negative patch testing, while another patient will not respond to dupilumab treatment and have persistent positive patch testing. This is likely because, in certain individuals, the allergen-immune system combination shifts ACD pathogenesis from a purely T_H1 response to at least a partial T_H2 response, thus allowing for benefit from dupilumab therapy. T helper 1 cell–mediated ACD should not be affected by dupilumab; therefore, reliable results can be elucidated from patch testing despite the drug.

Final Thoughts

We propose that AD patients with residual disease after taking dupilumab undergo patch testing. Positive results indicate allergens that are not inhibited by the drug. Patients will need to follow strict allergen avoidance to resolve this component of their disease; failure to improve might suggest the result was a nonrelevant positive.

If patch testing is negative, an alternative cause for residual disease must be sought. We do not recommend stopping dupilumab prior to patch testing to avoid a disease flare from AD or possible T_H2-mediated ACD.

It has been said that “extraordinary claims require extraordinary evidence.”¹ In the pursuit of evidence-based medicine, we are encouraged to follow a similar standard, with an emphasis on waiting for multiple studies with good-quality data and high levels of agreement before changing any aspect of our clinical practice. The ostensible purpose is that studies can be flawed, conclusions can be incorrect, or biases can be overlooked. In such cases, acting on questionable results could imperil patients. It is for this reason that so many review articles sometimes frustratingly seem to conclude that further evidence is needed.²

Based on this standard, recently published addendum guidelines from the National Institute of Allergy and Infectious Diseases for prevention of peanut allergy in the United States³ are somewhat striking in that they make fairly bold recommendations based on results from the 2015 Learning Early about Peanut Allergy (LEAP) study,⁴ a randomized trial evaluating early peanut introduction as a preventive strategy for peanut allergy. Of note, this study was not placebo controlled, was conducted at only 1 site in the United Kingdom, and only included 640 children, though the number of participants was admittedly large for this type of study.⁴ Arguably, the LEAP study alone does not provide enough evidence upon which to base what essentially amounts to an about-face in the official recommendations for prevention of peanut and other food allergies, which emphasized delayed introduction of high-risk foods, especially in high-risk individuals.^5-7 To better understand this shift, we need to briefly explore the context of the addendum guidelines.

As many as one-third of pediatric patients with atopic dermatitis (AD) have food allergies, thus diet often is invoked by patients and providers alike as an underlying cause of the disease.⁸ Many patients in my practice are so focused on potential food allergies that actual treatment of the affected skin is marginalized and often dismissed as a stopgap that does not address the root of the problem. A 2004 study of 100 children with AD found that diet was manipulated by the parents in 75% of patients in an attempt to manage the disease.⁹

Patients are not the only ones who consider food allergies to be a driving force in AD. The medical literature indicates that this theory has existed for centuries; for instance, with regard to the relationship between diet and AD, the author of an article from 1830 quipped, “There is probably no subject in which more deeply rooted convictions have been held . . . than the connection between diet and disease, both as regards the causation and treatment of the latter . . .”¹⁰ More apropos perhaps is a statement from the 2010 National Institute of Allergy and Infectious Diseases guidelines on food allergy management, which noted that while the expert group “does not mean to imply that AD results from [food allergies], the role of [food allergies] in the pathogenesis and severity of this condition remains controversial.”¹¹

Prior to the LEAP study, food allergy recommendations for clinical practice in the United Kingdom in 1998¹² and the United States in 2000¹³ recommended excluding allergenic foods (eg, peanuts, tree nuts, soy, milk, eggs) from the diet in infants with a family history of atopy until 3 years of age. However, those recommendations did not seem to be working, when in fact just the opposite was happening. From 1997 until the LEAP study was conducted in 2015, the prevalence of peanut allergy more than quadrupled and became the leading cause of anaphylaxis and death related to food allergy.¹⁴ Additionally, study after study concluded that elimination diets did not seem to help most patients with AD.¹⁵ As is required in good scientific thinking, when a hypothesis is proven false, other approaches must be considered.

The idea arose that perhaps delaying introduction of allergenic foods was the opposite of the answer.⁴ The LEAP study tested the notion that peanut allergies are rare in countries where peanuts are introduced early and if telling families to delay introduction of peanuts in infants might actually be causing development of a peanut allergy, and the tests bore fruit. It was found that giving infants peanut-containing foods resulted in a more than 80% reduction in peanut allergy at 5 years of age (P<.001).⁴ What was perhaps even more interesting was the connection between AD and peanut allergy. An important idea articulated in the LEAP study is in some ways revolutionary: Rather than foods causing AD, it could be that “early environmental exposure (through the skin) to peanut may account for early sensitization, whereas early oral exposure may lead to immune tolerance.”⁴ This concept—that impaired eczematous skin may actually lead to the development of food allergies—turns the whole thing upside down.

What do these updated guidelines actually suggest? The first guideline focuses on infants with severe AD, egg allergy, or both, who therefore are thought to be at the highest risk for developing peanut allergy.³ Because of the higher baseline risk in this subgroup, measurement of the peanut-specific IgE (peanut sIgE) level, skin prick testing (SPT), or both is strongly recommended before introducing peanut protein into the diet. This testing can be performed by qualified providers as a screening measure, but if positive (≥0.35 kU_A/L for peanut sIgE or >2 mm on the peanut SPT), referral to an allergy specialist is warranted. If these studies are negative, it is thought the likelihood of peanut allergy is low, and it is recommended that caregivers introduce age-appropriate peanut-containing foods (eg, peanut butter snack puffs, diluted peanut butter) as early as 4 to 6 months of age. The second guideline recommends that peanut-containing foods should be introduced into the diets of infants with mild or moderate AD at approximately 6 months of age without the need for prior screening via peanut sIgE or SPT. Lastly, the third guideline recommends that caregivers freely introduce peanut-containing foods together with other solid foods in infants without AD or food allergies in accordance with family preference.³

The results of the LEAP study are certainly exciting, and although the theoretical basis makes good scientific sense and the updated guidelines truly address an important and growing problem, there are several issues with this update that are worth considering. Given the constraints of the LEAP study, it certainly seems possible that the results will not be applicable to all populations or foods. More research is needed to ensure that this robust finding applies to other children and to explore the introduction of other allergenic foods, which the LEAP study investigators also emphasized.⁴

In fairness, the updated guidelines clearly state the quality of evidence of their recommendations and make it clear that expert opinion is playing a large role.³ For the first guideline regarding recommendations for those with severe AD and/or egg allergy, the quality of evidence is deemed moderate, while the contribution of expert opinion is listed as significant. For the second and third guidelines regarding recommendations for mild to moderate AD and those without AD, respectively, the quality of evidence is low and expert opinion is again listed as significant.³

Importantly, delineating severe AD from moderate disease—which is necessary because only severe AD warrants evaluation with peanut sIgE and/or SPT—can be difficult, as the distinction relies on a degree of subjectivity that may vary between specialists. Indeed, 2 publications suggest extending the definition of severe AD to include infants with early-onset AD (<3 months of age) and those with moderate AD not responding to treatment.^16,17

Despite these reservations, the updated guidelines represent a breakthrough in understanding in an area truly in need of advancement. Although the evidence may not be exactly extraordinary, the context for these developments and our deeper understanding suggest that we do indeed live in extraordinary times. 

It has been said that “extraordinary claims require extraordinary evidence.”¹ In the pursuit of evidence-based medicine, we are encouraged to follow a similar standard, with an emphasis on waiting for multiple studies with good-quality data and high levels of agreement before changing any aspect of our clinical practice. The ostensible purpose is that studies can be flawed, conclusions can be incorrect, or biases can be overlooked. In such cases, acting on questionable results could imperil patients. It is for this reason that so many review articles sometimes frustratingly seem to conclude that further evidence is needed.²

Based on this standard, recently published addendum guidelines from the National Institute of Allergy and Infectious Diseases for prevention of peanut allergy in the United States³ are somewhat striking in that they make fairly bold recommendations based on results from the 2015 Learning Early about Peanut Allergy (LEAP) study,⁴ a randomized trial evaluating early peanut introduction as a preventive strategy for peanut allergy. Of note, this study was not placebo controlled, was conducted at only 1 site in the United Kingdom, and only included 640 children, though the number of participants was admittedly large for this type of study.⁴ Arguably, the LEAP study alone does not provide enough evidence upon which to base what essentially amounts to an about-face in the official recommendations for prevention of peanut and other food allergies, which emphasized delayed introduction of high-risk foods, especially in high-risk individuals.^5-7 To better understand this shift, we need to briefly explore the context of the addendum guidelines.

As many as one-third of pediatric patients with atopic dermatitis (AD) have food allergies, thus diet often is invoked by patients and providers alike as an underlying cause of the disease.⁸ Many patients in my practice are so focused on potential food allergies that actual treatment of the affected skin is marginalized and often dismissed as a stopgap that does not address the root of the problem. A 2004 study of 100 children with AD found that diet was manipulated by the parents in 75% of patients in an attempt to manage the disease.⁹

Patients are not the only ones who consider food allergies to be a driving force in AD. The medical literature indicates that this theory has existed for centuries; for instance, with regard to the relationship between diet and AD, the author of an article from 1830 quipped, “There is probably no subject in which more deeply rooted convictions have been held . . . than the connection between diet and disease, both as regards the causation and treatment of the latter . . .”¹⁰ More apropos perhaps is a statement from the 2010 National Institute of Allergy and Infectious Diseases guidelines on food allergy management, which noted that while the expert group “does not mean to imply that AD results from [food allergies], the role of [food allergies] in the pathogenesis and severity of this condition remains controversial.”¹¹

Prior to the LEAP study, food allergy recommendations for clinical practice in the United Kingdom in 1998¹² and the United States in 2000¹³ recommended excluding allergenic foods (eg, peanuts, tree nuts, soy, milk, eggs) from the diet in infants with a family history of atopy until 3 years of age. However, those recommendations did not seem to be working, when in fact just the opposite was happening. From 1997 until the LEAP study was conducted in 2015, the prevalence of peanut allergy more than quadrupled and became the leading cause of anaphylaxis and death related to food allergy.¹⁴ Additionally, study after study concluded that elimination diets did not seem to help most patients with AD.¹⁵ As is required in good scientific thinking, when a hypothesis is proven false, other approaches must be considered.

The idea arose that perhaps delaying introduction of allergenic foods was the opposite of the answer.⁴ The LEAP study tested the notion that peanut allergies are rare in countries where peanuts are introduced early and if telling families to delay introduction of peanuts in infants might actually be causing development of a peanut allergy, and the tests bore fruit. It was found that giving infants peanut-containing foods resulted in a more than 80% reduction in peanut allergy at 5 years of age (P<.001).⁴ What was perhaps even more interesting was the connection between AD and peanut allergy. An important idea articulated in the LEAP study is in some ways revolutionary: Rather than foods causing AD, it could be that “early environmental exposure (through the skin) to peanut may account for early sensitization, whereas early oral exposure may lead to immune tolerance.”⁴ This concept—that impaired eczematous skin may actually lead to the development of food allergies—turns the whole thing upside down.

What do these updated guidelines actually suggest? The first guideline focuses on infants with severe AD, egg allergy, or both, who therefore are thought to be at the highest risk for developing peanut allergy.³ Because of the higher baseline risk in this subgroup, measurement of the peanut-specific IgE (peanut sIgE) level, skin prick testing (SPT), or both is strongly recommended before introducing peanut protein into the diet. This testing can be performed by qualified providers as a screening measure, but if positive (≥0.35 kU_A/L for peanut sIgE or >2 mm on the peanut SPT), referral to an allergy specialist is warranted. If these studies are negative, it is thought the likelihood of peanut allergy is low, and it is recommended that caregivers introduce age-appropriate peanut-containing foods (eg, peanut butter snack puffs, diluted peanut butter) as early as 4 to 6 months of age. The second guideline recommends that peanut-containing foods should be introduced into the diets of infants with mild or moderate AD at approximately 6 months of age without the need for prior screening via peanut sIgE or SPT. Lastly, the third guideline recommends that caregivers freely introduce peanut-containing foods together with other solid foods in infants without AD or food allergies in accordance with family preference.³

The results of the LEAP study are certainly exciting, and although the theoretical basis makes good scientific sense and the updated guidelines truly address an important and growing problem, there are several issues with this update that are worth considering. Given the constraints of the LEAP study, it certainly seems possible that the results will not be applicable to all populations or foods. More research is needed to ensure that this robust finding applies to other children and to explore the introduction of other allergenic foods, which the LEAP study investigators also emphasized.⁴

In fairness, the updated guidelines clearly state the quality of evidence of their recommendations and make it clear that expert opinion is playing a large role.³ For the first guideline regarding recommendations for those with severe AD and/or egg allergy, the quality of evidence is deemed moderate, while the contribution of expert opinion is listed as significant. For the second and third guidelines regarding recommendations for mild to moderate AD and those without AD, respectively, the quality of evidence is low and expert opinion is again listed as significant.³

Importantly, delineating severe AD from moderate disease—which is necessary because only severe AD warrants evaluation with peanut sIgE and/or SPT—can be difficult, as the distinction relies on a degree of subjectivity that may vary between specialists. Indeed, 2 publications suggest extending the definition of severe AD to include infants with early-onset AD (<3 months of age) and those with moderate AD not responding to treatment.^16,17

Despite these reservations, the updated guidelines represent a breakthrough in understanding in an area truly in need of advancement. Although the evidence may not be exactly extraordinary, the context for these developments and our deeper understanding suggest that we do indeed live in extraordinary times. 

It has been said that “extraordinary claims require extraordinary evidence.”¹ In the pursuit of evidence-based medicine, we are encouraged to follow a similar standard, with an emphasis on waiting for multiple studies with good-quality data and high levels of agreement before changing any aspect of our clinical practice. The ostensible purpose is that studies can be flawed, conclusions can be incorrect, or biases can be overlooked. In such cases, acting on questionable results could imperil patients. It is for this reason that so many review articles sometimes frustratingly seem to conclude that further evidence is needed.²

Based on this standard, recently published addendum guidelines from the National Institute of Allergy and Infectious Diseases for prevention of peanut allergy in the United States³ are somewhat striking in that they make fairly bold recommendations based on results from the 2015 Learning Early about Peanut Allergy (LEAP) study,⁴ a randomized trial evaluating early peanut introduction as a preventive strategy for peanut allergy. Of note, this study was not placebo controlled, was conducted at only 1 site in the United Kingdom, and only included 640 children, though the number of participants was admittedly large for this type of study.⁴ Arguably, the LEAP study alone does not provide enough evidence upon which to base what essentially amounts to an about-face in the official recommendations for prevention of peanut and other food allergies, which emphasized delayed introduction of high-risk foods, especially in high-risk individuals.^5-7 To better understand this shift, we need to briefly explore the context of the addendum guidelines.

As many as one-third of pediatric patients with atopic dermatitis (AD) have food allergies, thus diet often is invoked by patients and providers alike as an underlying cause of the disease.⁸ Many patients in my practice are so focused on potential food allergies that actual treatment of the affected skin is marginalized and often dismissed as a stopgap that does not address the root of the problem. A 2004 study of 100 children with AD found that diet was manipulated by the parents in 75% of patients in an attempt to manage the disease.⁹

Patients are not the only ones who consider food allergies to be a driving force in AD. The medical literature indicates that this theory has existed for centuries; for instance, with regard to the relationship between diet and AD, the author of an article from 1830 quipped, “There is probably no subject in which more deeply rooted convictions have been held . . . than the connection between diet and disease, both as regards the causation and treatment of the latter . . .”¹⁰ More apropos perhaps is a statement from the 2010 National Institute of Allergy and Infectious Diseases guidelines on food allergy management, which noted that while the expert group “does not mean to imply that AD results from [food allergies], the role of [food allergies] in the pathogenesis and severity of this condition remains controversial.”¹¹

Prior to the LEAP study, food allergy recommendations for clinical practice in the United Kingdom in 1998¹² and the United States in 2000¹³ recommended excluding allergenic foods (eg, peanuts, tree nuts, soy, milk, eggs) from the diet in infants with a family history of atopy until 3 years of age. However, those recommendations did not seem to be working, when in fact just the opposite was happening. From 1997 until the LEAP study was conducted in 2015, the prevalence of peanut allergy more than quadrupled and became the leading cause of anaphylaxis and death related to food allergy.¹⁴ Additionally, study after study concluded that elimination diets did not seem to help most patients with AD.¹⁵ As is required in good scientific thinking, when a hypothesis is proven false, other approaches must be considered.

The idea arose that perhaps delaying introduction of allergenic foods was the opposite of the answer.⁴ The LEAP study tested the notion that peanut allergies are rare in countries where peanuts are introduced early and if telling families to delay introduction of peanuts in infants might actually be causing development of a peanut allergy, and the tests bore fruit. It was found that giving infants peanut-containing foods resulted in a more than 80% reduction in peanut allergy at 5 years of age (P<.001).⁴ What was perhaps even more interesting was the connection between AD and peanut allergy. An important idea articulated in the LEAP study is in some ways revolutionary: Rather than foods causing AD, it could be that “early environmental exposure (through the skin) to peanut may account for early sensitization, whereas early oral exposure may lead to immune tolerance.”⁴ This concept—that impaired eczematous skin may actually lead to the development of food allergies—turns the whole thing upside down.

What do these updated guidelines actually suggest? The first guideline focuses on infants with severe AD, egg allergy, or both, who therefore are thought to be at the highest risk for developing peanut allergy.³ Because of the higher baseline risk in this subgroup, measurement of the peanut-specific IgE (peanut sIgE) level, skin prick testing (SPT), or both is strongly recommended before introducing peanut protein into the diet. This testing can be performed by qualified providers as a screening measure, but if positive (≥0.35 kU_A/L for peanut sIgE or >2 mm on the peanut SPT), referral to an allergy specialist is warranted. If these studies are negative, it is thought the likelihood of peanut allergy is low, and it is recommended that caregivers introduce age-appropriate peanut-containing foods (eg, peanut butter snack puffs, diluted peanut butter) as early as 4 to 6 months of age. The second guideline recommends that peanut-containing foods should be introduced into the diets of infants with mild or moderate AD at approximately 6 months of age without the need for prior screening via peanut sIgE or SPT. Lastly, the third guideline recommends that caregivers freely introduce peanut-containing foods together with other solid foods in infants without AD or food allergies in accordance with family preference.³

The results of the LEAP study are certainly exciting, and although the theoretical basis makes good scientific sense and the updated guidelines truly address an important and growing problem, there are several issues with this update that are worth considering. Given the constraints of the LEAP study, it certainly seems possible that the results will not be applicable to all populations or foods. More research is needed to ensure that this robust finding applies to other children and to explore the introduction of other allergenic foods, which the LEAP study investigators also emphasized.⁴

In fairness, the updated guidelines clearly state the quality of evidence of their recommendations and make it clear that expert opinion is playing a large role.³ For the first guideline regarding recommendations for those with severe AD and/or egg allergy, the quality of evidence is deemed moderate, while the contribution of expert opinion is listed as significant. For the second and third guidelines regarding recommendations for mild to moderate AD and those without AD, respectively, the quality of evidence is low and expert opinion is again listed as significant.³

Importantly, delineating severe AD from moderate disease—which is necessary because only severe AD warrants evaluation with peanut sIgE and/or SPT—can be difficult, as the distinction relies on a degree of subjectivity that may vary between specialists. Indeed, 2 publications suggest extending the definition of severe AD to include infants with early-onset AD (<3 months of age) and those with moderate AD not responding to treatment.^16,17

Despite these reservations, the updated guidelines represent a breakthrough in understanding in an area truly in need of advancement. Although the evidence may not be exactly extraordinary, the context for these developments and our deeper understanding suggest that we do indeed live in extraordinary times.