Association Between Atopic Eczema and Cancer in England and Denmark

Key Points

Question  Is atopic eczema associated with increased cancer risk?

Findings  In 2 large cohort studies conducted in England (471 970 and 2 239 775 individuals with and without atopic eczema, respectively) and Denmark (44 945 and 445 673 individuals with and without atopic eczema, respectively), no evidence was found of an increased risk of most cancers among people with atopic eczema compared with those without eczema. However, atopic eczema was associated with an increased risk of lymphoma, particularly non-Hodgkin lymphoma, with risk increasing with greater eczema severity.

Meaning  The findings in this study did not support an association between atopic eczema and most cancers; however, there was evidence of higher lymphoma risk with increasing eczema severity.

Importance  Associations between atopic eczema and cancer are unclear, with competing theories that increased immune surveillance decreases cancer risk and that immune stimulation increases cancer risk. Establishing baseline cancer risk in people with atopic eczema is important before exploring the association between new biologic drugs for atopic eczema and cancer risk.

Objective  To investigate whether atopic eczema is associated with cancer.

Design, Setting, and Participants  Matched cohort studies were conducted from January 2, 1998, to March 31, 2016, in England and from January 1, 1982, to June 30, 2016, in Denmark. We conducted our analyses between July 2018 and July 2019. The setting was English primary care and nationwide Danish data. Participants with atopic eczema (adults only in England and any age in Denmark) were matched on age, sex, and calendar period (as well as primary care practice in England only) to those without atopic eczema.

Exposure  Atopic eczema.

Main Outcomes and Measures  Overall cancer risk and risk of specific cancers were compared in people with and without atopic eczema.

Results  In England, matched cohorts included 471 970 individuals with atopic eczema (median [IQR] age, 41.1 [24.9-60.7] years; 276 510 [58.6%] female) and 2 239 775 individuals without atopic eczema (median [IQR] age, 39.8 [25.9-58.4] years; 1 301 074 [58.1%] female). In Denmark, matched cohorts included 44 945 individuals with atopic eczema (median [IQR] age, 13.7 [1.7-21.1] years; 22 826 [50.8%] female) and 445 673 individuals without atopic eczema (median [IQR] age, 13.5 [1.7-20.8] years; 226 323 [50.8%] female). Little evidence was found of associations between atopic eczema and overall cancer (adjusted hazard ratio [HR], 1.04; 99% CI, 1.02-1.06 in England and 1.05; 99% CI, 0.95-1.16 in Denmark) or for most specific cancers. However, noncutaneous lymphoma risk was increased in people with atopic eczema in England (adjusted HR, 1.19; 99% CI, 1.07-1.34 for non-Hodgkin lymphoma [NHL] and 1.48; 99% CI, 1.07-2.04 for Hodgkin lymphoma). Lymphoma risk was increased in people with greater eczema severity vs those without atopic eczema (NHL adjusted HR, 1.06; 99% CI, 0.90-1.25 for mild eczema; 1.24; 99% CI, 1.04-1.48 for moderate eczema; and 2.08; 99% CI, 1.42-3.04 for severe eczema). Danish point estimates also showed increased lymphoma risk in people with moderate to severe eczema compared with those without atopic eczema (minimally adjusted HR, 1.31; 99% CI, 0.76-2.26 for NHL and 1.35; 99% CI, 0.65-2.82 for Hodgkin lymphoma), but the 99% CIs were wide.

Conclusions and Relevance  The findings from 2 large population-based studies performed in different settings do not support associations between atopic eczema and most cancers. However, an association was observed between atopic eczema and lymphoma, particularly NHL, that increased with eczema severity. This finding warrants further study as new immunomodulatory systemic therapeutics are brought to market that may alter cancer risk.

Atopic eczema is the most common inflammatory skin disease, affecting 2% to 10% of adults^1-3 and up to 20% of children.⁴ Eczema may be associated with cancer for many reasons, including immune dysregulation related to eczema, unhealthy lifestyle choices (eg, smoking and alcohol consumption as coping mechanisms for persistent itch), stress, low self-esteem, and sleep deprivation.^5,6 Although most eczema is managed with topical treatment, systemic treatments (eg, glucocorticoids, azathioprine, and cyclosporine) for severe disease could also affect cancer risk.⁷

Cancer is 1 of the leading causes of death globally.⁸ Skin disease is the most common reason for a new primary care practice consultation in England,^9,10 and eczema accounts for a substantial proportion of that burden.⁹ Hence, any association between atopic eczema and cancer is potentially relevant to public health. Furthermore, as new therapeutics (that may alter cancer risk) are introduced, it is important to establish baseline cancer risk in people with atopic eczema.¹¹

A recent systematic review and meta-analysis¹² highlighted that existing evidence of cancer risk in atopic eczema is conflicting, with studies limited by lack of consideration of eczema severity and treatment and by the inability to adjust for key confounders. Studies have reported reductions in specific brain cancers and increased lymphoma risk in people with eczema.^5,6,13-19 Two competing theories may explain the complex association between atopic eczema and cancer, namely, that increased immune surveillance decreases cancer risk and that immune stimulation increases cancer risk. Immunosuppressive therapy (including topical calcineurin inhibitors) and an impaired skin barrier may also increase risk, particularly of skin cancer, but recent findings are inconclusive.²⁰ Experimental mouse models support reduced skin cancer in individuals with atopic eczema,²¹ but population-based studies are conflicting.^6,22-25

We used health care data from matched cohort studies conducted in England and Denmark to address uncertainties identified by the recent systematic review and meta-analysis.¹² We explored cancer risk in persons with atopic eczema compared with those without atopic eczema, including variation in risk with eczema severity and activity.

This investigation was composed of 2 matched cohort studies conducted from January 2, 1998, to March 31, 2016, in England and from January 1, 1982, to June 30, 2016, in Denmark. We conducted our analyses between July 2018 and July 2019. Participants with atopic eczema (adults only in England and any age in Denmark) were matched on age, sex, and calendar period (as well as primary care practice in England only) to people without atopic eczema. We compared overall cancer risk and risk of specific cancers in those with and without atopic eczema. The English study was approved by the London School of Hygiene and Tropical Medicine Research Ethics Committee and by the United Kingdom Clinical Practice Research Datalink (CPRD) Independent Scientific Advisory Committee (protocol 17_108). The English trial protocol is available at Supplement 1. The Danish study was approved by the Danish Data Protection Agency (record 2015-57-0002; AU-2016-051-000001; 632). The Danish trial protocol is available at Supplement 2. These are deidentified data and do not require explicit consent; individual consent for use of electronic health record data for research is implied. In England, individuals are offered the right to opt out of the use of their anonymized data. Danish legislation does not require approval by an ethical review board or informed consent from participants in registry-based studies.

Two matched cohorts of people with and without atopic eczema in England and Denmark were identified (Table 1 and Figure 1). We did not include children in the English study (unlike the Danish study, in which we did not restrict age) because of limited power to study childhood cancers. Childhood cancer is rare, and the median follow-up in the CPRD is only 5 years.²⁶

In England, electronic health records of adults attending primary care practices contributing to the CPRD GOLD were used along with linked hospital admissions (Hospital Episode Statistics [HES]) and death registration (Office for National Statistics [ONS]) data. The CPRD includes data from participating primary care practices covering 7% of the UK population.^26-28 The HES covers all admissions for National Health Service–funded patients treated in England.²⁹ Most English primary care practices contributing to the CPRD (75%) are linked to the HES.²⁶ The ONS mortality data include cause of death for all deaths in the UK.

In Denmark, linked nationwide data from the Danish National Patient Registry (hospital admissions, outpatient clinic appointments, and emergency department contacts),^30,31 the Danish Cancer Registry (incident cancers),³² the Civil Registration System (demographics),³³ the Danish National Prescription Registry (prescriptions filled at community pharmacies),³⁴ and socioeconomic data were used. The socioeconomic data included highest educational level and gross personal income gathered by Statistics Denmark.^35,36

Morbidity code lists and additional details of all variable definitions are provided in eAppendix 1 and eTables 16, 17, and 18 in Supplement 3. Complete code lists for all variables used in the English study are available online.³⁷

All adults (aged ≥18 years) registered with primary care practices contributing data meeting CPRD quality control standards between 1998 and 2016 were eligible for inclusion (Figure 1 and eAppendix 2 in Supplement 3). Individuals needed at least 1 year of registration before cohort entry. Individuals with atopic eczema were identified based on a previously validated algorithm (eAppendix 1 in Supplement 3).³⁸ We randomly selected an individually matched cohort (without replacement) of up to 5 people for each individual with atopic eczema by age, sex, and primary care practice (eAppendix 2 in Supplement 3).

All individuals (any age) born in Denmark with a hospital-based (outpatient and inpatient records) diagnosis of atopic eczema (mostly moderate to severe eczema) between 1982 and 2016 were identified (Figure 1 and eAppendix 2 in Supplement 3). Also identified was a comparison cohort (with replacement) of up to 10 people without atopic eczema matched to each individual with atopic eczema by sex and exact birth year.³⁹

In England and Denmark, we excluded individuals with any history of cancer (except nonmelanoma skin cancer [NMSC] or keratinocyte cancer). In analyses focused specifically on skin cancer, we further excluded individuals with a previous NMSC record.

We investigated the association between atopic eczema and cancer, overall and the site-specific cancers (Figure 2), chosen a priori because they are common or were previously associated with atopic eczema or atopy.⁵ Overall cancer excluded NMSCs because of possible ascertainment bias (frequent skin examinations in atopic eczema) and possible poor capture because NMSCs may be treated topically (eg, photodynamic therapy) without histological confirmation.⁴⁰ We excluded cutaneous lymphomas from lymphoma definitions to reduce ascertainment bias because of frequent skin examinations or misdiagnosis of cutaneous lymphoma as eczema. In site-specific cancer analyses, we censored individuals at diagnosis of any cancer (excluding NMSC) and additionally at first NMSC diagnosis in skin cancer analyses.

We considered the following variables as potential confounders: age, sex, calendar period, and socioeconomic status. In England, we used the Index of Multiple Deprivation (IMD) as a measure of socioeconomic deprivation of residential neighborhood.^41-43 In Denmark, we used highest educational level, partner status, and gross personal income as measures of socioeconomic status.

Also considered were factors potentially on the pathway between atopic eczema and cancer (ie, potential mediators), specifically body mass index (BMI), smoking, and harmful alcohol use (in England). Data on these variables were not available in Denmark; we relied instead on diagnoses or treatments for diseases associated with lifestyle factors as proxies, including chronic obstructive pulmonary disease, hyperlipidemia, hypertension, alcohol-related conditions, ischemic heart disease, hospital-diagnosed obesity, and type 2 diabetes.

We considered descriptive characteristics for individuals with and without atopic eczema. We used Cox proportional hazards regression, stratified by matched set, to estimate hazard ratios (HRs) comparing overall cancer risk and risk of specific cancers in people with and without atopic eczema. In England, we used age as the underlying timescale. In Denmark, because those with atopic eczema entered the cohort at first hospital atopic eczema diagnosis, we used time since first atopic eczema diagnosis for individuals with atopic eczema and the corresponding cohort entry date for matched individuals.

The minimally adjusted models implicitly accounted for matching and timescale factors, including age, sex, cohort entry date, and (in England only) primary care practice. In England, we followed with sequential models adjusting for (1) confounders (IMD and calendar period) to estimate the potential association between atopic eczema and cancer and then (2) potential mediators that might be on the pathway between atopic eczema and cancer. Sequential models, adjusting first for confounders and then for potential mediators, allowed us to differentiate between possible direct (eg, inflammatory) and indirect (ie, via mediating variables [eg, lifestyle factors]) associations of atopic eczema with cancer.⁴⁴ In Denmark, we were only able to adjust for potential confounders (ie, socioeconomic status) in sensitivity analyses restricted to individuals 30 years or older at cohort entry (socioeconomic status may be unreliable in younger people, who are less likely to have attained their highest educational level and have partner status and gross personal income recorded).

We repeated our main analyses in a series of sensitivity analyses assessing the robustness of our findings. Details are listed in eTable 1 in Supplement 3.

In our secondary analyses, we investigated whether the association between atopic eczema and cancer was (1) more pronounced in individuals with severe or active atopic eczema (unfeasible in Denmark because of small event numbers) and (2) modified by age, sex, or presence of asthma. Details are available in eAppendix 3 in Supplement 3.

We assessed the proportional hazards assumption using Schoenfeld residual plots. In all analyses, as per our approved protocols (listed in the Methods section), we used 99% CIs to reduce the risk of type I error in the context of multiple analyses.⁴⁵ For data management and analyses, Stata, version 15 (StataCorp LLC), and SAS, version 9.4 (SAS Institute Inc), were used.

In England, 2 629 640 individuals remained in valid matched sets and were included in analyses (eFigure 1 and eTable 2 in Supplement 3). The matched cohorts in England included 471 970 individuals with atopic eczema (median [IQR] age, 41.1 [24.9-60.7] years; 276 510 [58.6%] female) and 2 239 775 individuals without atopic eczema (median [IQR] age, 39.8 [25.9-58.4] years; 1 301 074 [58.1%] female).

In Denmark, 490 618 individuals remained in valid matched sets and were included in the analyses (eFigure 2 in Supplement 3). The Danish matched cohorts included 44 945 individuals with atopic eczema (median [IQR] age, 13.7 [1.7-21.1] years; 22 826 [50.8%] female) and 445 673 individuals without atopic eczema (median [IQR] age, 13.5 [1.7-20.8] years; 226 323 [50.8%] female).

In England, the median follow-up was 4.5 years (IQR, 1.7-9.0 years). The cohorts with and without atopic eczema had broadly similar age, sex, and IMD (Table 2). Prevalence of smoking or overweight/obesity was higher in individuals with atopic eczema than in those without (46.7% vs 43.5% for smoking and 44.6% vs 40.7% for overweight/obesity). However, smoking and BMI data were less likely to be missing in people with atopic eczema compared with those without, and those with missing BMI or smoking status were more likely to be young and male (eTable 2 in Supplement 3).

In Denmark, the median follow-up was 14.3 years (IQR, 6.3-20.8 years). Distribution of age and sex was equal, and comorbidities at baseline were rare in both groups (Table 2). Among individuals aged 30 years at cohort entry, higher educational level and being single were more common in those with atopic eczema (eTable 3 in Supplement 3) than in those without atopic eczema.

In both countries, little evidence was found of any association between atopic eczema and overall cancer (Figure 2 and eTable 4 and eTable 5 in Supplement 3). Hazard ratios (HRs) comparing cancer risks in people with and without atopic eczema were close to 1, showing evidence of only a minor, clinically unimportant increase in minimally adjusted estimates (HR, 1.05; 99% CI, 0.95-1.16 in Denmark) and confounder-adjusted estimates (HR, 1.04; 99% CI, 1.02-1.06 in England). After further adjusting for potential mediators, estimates moved slightly closer to the null. For all specific cancer sites, only small differences in absolute cancer risk were seen between individuals with and without atopic eczema.

For most site-specific cancers, English and Danish analyses showed no strong evidence of an association with atopic eczema (Figure 2). For the following cancers, HRs comparing cancer risk in people with and without atopic eczema were close to the null: breast, prostate, pancreas, leukemia, multiple myeloma, and central nervous system cancers. There were decreases in melanoma risk and small increases in NMSC risk.

We found evidence of larger increases in lymphoma risk in people with atopic eczema compared with those without atopic eczema. In England, non-Hodgkin lymphoma (NHL) risk was 20% higher (HR, 1.20; 99% CI, 1.07-1.34) and Hodgkin lymphoma risk was 48% higher (HR, 1.48; 99% CI, 1.07-2.04) in people with atopic eczema compared with those without atopic eczema after adjusting for confounders (Figure 2). Although Danish point estimates for lymphoma were less precise than English estimates, they showed similar increased risk (minimally adjusted HR, 1.31; 99% CI, 0.76-2.26 for NHL and 1.35; 99% CI, 0.65-2.82 for Hodgkin lymphoma), but the 99% CIs were wide.

An initial small increased lung cancer risk was observed in people with atopic eczema vs those without (confounder-adjusted HR, 1.08; 99% CI, 1.01-1.16 in England and minimally adjusted HR, 1.45; 99% CI, 1.08-1.96 in Denmark) (Figure 2). However, risk attenuated after further adjusting for potential mediators (including smoking or smoking-related disease) (HR, 1.02; 99% CI, 0.95-1.10 in England and 1.21; 99% CI, 0.88-1.65 in Denmark) (eTable 5 in Supplement 3).

Our sensitivity analyses yielded similar results (including analyses restricted to individuals with a 3-year cancer-free window after eczema diagnosis to limit reverse causality as an explanation for our findings). These results are summarized in eTable 6 and eTable 7 in Supplement 3.

In England, we generally found no evidence of different overall cancer or site-specific cancer risk in severe eczema (Figure 3 and eTable 8 in Supplement 3). An exception was lymphoma, particularly NHL, for which the observed increased risk became more pronounced with increasing eczema severity. The NHL confounder-adjusted HRs compared with those without atopic eczema were 1.06 (99% CI, 0.90-1.25) for mild eczema, 1.24 (99% CI, 1.04-1.48) for moderate eczema, and 2.08 (99% CI, 1.42-3.04) for severe eczema (eTable 8 in Supplement 3).

In England, we saw limited evidence of an association between atopic eczema activity and most site-specific cancers (eFigure 3 and eTable 9 in Supplement 3). Notably, there were higher risks of NHL in people with more active atopic eczema (confounder-adjusted HR compared with those with no atopic eczema, 1.07; 99% CI, 0.83-1.38 for never active eczema; 1.08; 99% CI, 0.92-1.26 for moderately active eczema; and 1.58; 99% CI, 1.28-1.94 for very active eczema) and higher overall cancer risk in people with more active eczema (confounder-adjusted HR, 1.11; 99% CI, 1.07-1.15) (eTable 9 in Supplement 3).

Little evidence was found of effect modification by age, sex, or presence of asthma. These results are summarized in eTables 10, 11, 12, 13, 14, and 15 in Supplement 3.

In 2 matched cohort studies from England and Denmark, no evidence was found of an association between atopic eczema and most cancers. However, increased risk of lymphoma, particularly NHL, was seen among people with atopic eczema, specifically those with more severe or active disease. In England, we observed a 23% increased risk of lymphoma overall (similar to Danish estimates), with a 100% increase in NHL risk in people with severe eczema, although absolute risk differences were small.

A recent systematic review and meta-analysis¹² of cancer risk in people with atopic eczema highlighted the heterogeneity and limitations of existing data. Overall, our findings generally confirm those reported in prior smaller studies.^{12,14,16,19,23,46-54}

Our study is one of the few longitudinal studies^{12,14,16,23,46,47,52} to adjust for numerous potential confounders and mediators and to address the role of eczema severity and activity in 2 cohorts with complementary strengths and limitations. Our results suggest that the overall cancer risk among individuals with atopic eczema is small. This baseline risk is particularly important to establish as many new immunomodulatory systemic therapeutics are brought to market.

Previous studies assessing the association between atopic eczema and site-specific cancers have had conflicting results.¹² A discussion follows for brain cancer, solid-organ cancers, melanoma, and noncutaneous lymphomas.

A recent systematic review and meta-analysis⁴⁶ reported a 23% reduction in brain cancer risk in people with atopic eczema. However, this protective association was not observed with more specific atopic eczema diagnostic criteria, and there was substantial heterogeneity (I² > 75%) between studies. We could not reliably assess the risk of brain cancer subtypes because of low statistical power.

Some prior cohort and case-control studies^47-51 that assessed other solid-organ cancers found an increased risk of lung cancer associated with atopic diseases. Although we initially observed a small increased lung cancer risk in individuals with atopic eczema, this association disappeared after adjusting for possible mediators, suggesting that the increased risk may be attributable to smoking among people with atopic eczema.

The observation of a potentially protective association of atopic eczema with melanoma is of interest. Our findings are consistent with observations from mouse models²¹ and may reflect sun avoidance in people with atopic eczema or increased immune surveillance (further studies would be required to elucidate biologic mechanisms). Having atopic eczema might have led to more skin examinations, resulting in ascertainment bias and either an increased skin cancer risk (as seen for NMSC in this study and in other studies^23,47,51,52) or an apparently decreased risk because of better detection of melanoma in situ (with consequent lower invasive cancer risk). If ascertainment leads to increased melanoma detection, we would expect to see an even larger risk reduction in advanced melanoma; additional studies could examine this hypothesis.

The association between atopic eczema and lymphoma is consistent with previous findings in a systematic review and meta-analysis¹⁶ of an overall relative risk of lymphoma of 1.43, as well as the finding in a CPRD-based case-control study¹⁹ of an association between Hodgkin lymphoma and allergic disease (asthma, eczema, and allergic rhinitis). The association of eczema severity with lymphoma has previously been assessed in only 1 cohort study⁵⁵ and 2 case-control studies.^53,54 Although the cohort study⁵⁵ used data from Medicaid, a US administrative database, and focused on psoriasis, it demonstrated a 2-fold increased lymphoma risk among individuals with severe eczema in a limited secondary analysis. However, because NHL and Hodgkin lymphoma have different risk factor profiles,⁵⁶ our findings in this study might suggest detection bias (because of the use of chest radiographs in people with comorbid asthma) or residual confounding, perhaps because of immunosuppression⁵⁷ (although sensitivity analyses adjusting for immunosuppression in the present study were similar to our main analysis results).

Previous studies attempting to clarify whether topical calcineurin inhibitors (used in our study to identify moderate eczema severity) increase lymphoma risk in people with atopic eczema have failed to convincingly demonstrate an association, although most analyses were limited by low statistical power and confounding by severity.¹⁶ A low proportion of individuals with atopic eczema treated in primary care receive calcineurin inhibitors because these agents are recommended for specialist use only.⁵⁸

In this study, we found an increased lymphoma risk in individuals with atopic eczema, especially for those with severe eczema; however, absolute differences were small, and there was little evidence of an increased risk of other cancers. This finding may be because of reverse causation (ie, atopic eczema could be an early sign of lymphoma), but a sensitivity analysis extending the cancer-free lag period to 3 years did not show any evidence of this hypothesis (eTable 6 in Supplement 3). Future research could explore underlying mechanisms for the associations observed for lymphoma, particularly NHL, including studies assessing Epstein-Barr virus in people with atopic eczema, as well as addressing the therapeutic implications (associated with existing and novel immunosuppressive treatments) of our findings in individuals with severe eczema.

This study has strengths and limitations. A major strength of this study is the use of prospectively collected data from large representative population-based databases from 2 countries with free health care access.^26,30,59 Our results reflect real-world clinical practice and are likely to be generalizable to other settings. Although the Danish findings are unlikely to apply to individuals with mild eczema because the Danish data were restricted to hospital-based atopic eczema diagnoses, the Danish results are consistent with those for individuals with moderate to severe eczema in England. Therefore, combining the English and Danish population-based data lends credence to the study conclusions by virtue of triangulation.⁶⁰

It is possible that atopic eczema was misclassified in our study. Although we used a validated algorithm to identify atopic eczema in England and the validity of diagnoses in a dermatology department was found to be high in Denmark (positive predictive value, 99%⁶¹), we may have missed atopic eczema diagnoses (likely unrelated to cancer risk), which could have underestimated the magnitude of any association between atopic eczema and cancer.

The quality of diagnoses in the Danish Cancer Registry is excellent: reporting is mandatory, diagnoses stem from a range of sources, the proportion of histologically verified tumors is high (>90% for major cancers), and few diagnoses are based only on death certificates (0.1%).^32,62,63 In England, most nationally registered cancer cases (>90%) can be identified in the CPRD; by including the HES and ONS data, we further improved cancer identification.⁶⁴ Nevertheless, we cannot exclude ascertainment bias; cancers may be diagnosed more frequently in individuals with atopic eczema because of frequent skin examinations for skin cancer, regular blood test monitoring for systemic immunosuppressive drugs (in severe eczema only), or chest radiographs for concomitant asthma or before drug initiation in severe eczema (eg, cyclosporine).

Potential misclassification of covariates may have led to residual confounding, biasing the estimates of direct associations in mediation analyses. However, increasing adjustment made little change to our HR estimates. In Denmark, we were only able to adjust for potential confounders (ie, socioeconomic status) in sensitivity analyses, but the results of these sensitivity analyses were similar to those of the main analysis. However, matching with replacement in the Danish cohort may have resulted in mild misleadingly narrow 99% CIs.

A limitation that is common to many dermatological conditions lacking quantitative measures is reliance on treatments to classify disease severity and activity. In Denmark, sample sizes were insufficient for eczema severity and activity analyses; however, we note that, by definition, most individuals had moderate to severe eczema. Atopic eczema was identified in hospital records because we had no access to data from Danish primary care, where mild eczema is managed. The study also had limited power to examine certain cancers, specifically brain cancer subtypes.

In this study, no evidence was found that people with atopic eczema are at increased risk of most cancers. An exception is the observed association between atopic eczema and lymphoma, particularly NHL, that increased with eczema severity. This finding warrants further study as new immunomodulatory systemic therapeutics are brought to market that may alter cancer risk.

Accepted for Publication: April 28, 2020.

Published Online: June 24, 2020. doi:10.1001/jamadermatol.2020.1948

Open Access: This is an open access article distributed under the terms of the CC-BY License. © 2020 Mansfield KE et al. JAMA Dermatology.

Corresponding Author: Kathryn E. Mansfield, PhD, Department of Non-communicable Disease Epidemiology, London School of Hygiene and Tropical Medicine, Keppel Street, London WC1E 7HT, United Kingdom (kathryn.mansfield@lshtm.ac.uk).

Author Contributions: Drs Mansfield and Schmidt had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Drs Mansfield and Schmidt contributed equally as co–first authors.

Concept and design: Mansfield, Schmidt, Darvalics, Abuabara, Smeeth, Bhaskaran, Langan.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: Mansfield, Schmidt.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Mansfield, Darvalics, Mulick, Silverwood.

Obtained funding: Schmidt, Langan.

Administrative, technical, or material support: Mansfield, Sørensen, Bhaskaran.

Supervision: Schmidt, Abuabara, Sørensen, Smeeth, Bhaskaran, Langan.

Conflict of Interest Disclosures: Dr Mansfield and Ms Mulick reported being funded under a Wellcome Trust grant awarded to Dr Langan. Dr Abuabara reported receiving personal fees from TARGET-DERM and receiving grants from Pfizer. Dr Sørensen reported that the Department of Clinical Epidemiology, Aarhus University Hospital, receives funding for other studies from companies in the form of research grants to (and administered by) Aarhus University. Dr Smeeth reported receiving grants from the Wellcome Trust, Medical Research Council, National Institute for Health Research, GlaxoSmithKline, British Heart Foundation, and Diabetes UK and being a trustee of the British Heart Foundation. Dr Bhaskaran reported receiving grants from the Wellcome Trust, Royal Society, Medical Research Council, British Heart Foundation, and Diabetes UK. Dr Langan reported receiving grants from the Wellcome Trust and from IMI Horizon 2020 funding BIOMAP. No other disclosures were reported.

Funding/Support: Dr Bhaskaran was supported by a Sir Henry Dale Fellowship from the Wellcome Trust and the Royal Society (107731/Z/15/Z). Dr Langan was supported by a Wellcome Trust Senior Clinical Fellowship (205039/Z/16/Z). The Danish study was funded by the Dagmar Marshalls Fund and the Aase and Ejnar Danielsens Fund.

Role of the Funder/Sponsor: The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Additional Information: This article used data provided by patients and collected by the United Kingdom National Health Service as part of their care and support.