eTable. Systematic Review of the Demographics, Comorbidities, and Clinical Features of Necrobiotic Xanthogranuloma
Customize your JAMA Network experience by selecting one or more topics from the list below.
Identify all potential conflicts of interest that might be relevant to your comment.
Conflicts of interest comprise financial interests, activities, and relationships within the past 3 years including but not limited to employment, affiliation, grants or funding, consultancies, honoraria or payment, speaker's bureaus, stock ownership or options, expert testimony, royalties, donation of medical equipment, or patents planned, pending, or issued.
Err on the side of full disclosure.
If you have no conflicts of interest, check "No potential conflicts of interest" in the box below. The information will be posted with your response.
Not all submitted comments are published. Please see our commenting policy for details.
Nelson CA, Zhong CS, Hashemi DA, et al. A Multicenter Cross-Sectional Study and Systematic Review of Necrobiotic Xanthogranuloma With Proposed Diagnostic Criteria. JAMA Dermatol. 2020;156(3):270–279. doi:10.1001/jamadermatol.2019.4221
What are the characteristics of necrobiotic xanthogranuloma, and what criteria should establish the diagnosis?
This multicenter cross-sectional study and systematic review included 235 patients with necrobiotic xanthogranuloma. Informed by the results, 8 board-certified dermatologists participated in a consensus exercise; in the absence of foreign body, infection, or other identifiable cause, both major criteria (clinical and histopathological features) and at least 1 minor criterion (paraproteinemia, plasma-cell dyscrasia, and/or other associated lymphoproliferative disorder, or a periorbital distribution) were proposed to diagnose necrobiotic xanthogranuloma.
These findings and criteria have potential to advance clinical research and should be validated by further studies.
Necrobiotic xanthogranuloma (NXG) is a non–Langerhans cell histiocytosis classically associated with paraproteinemia attributable to plasma-cell dyscrasias or lymphoproliferative disorders. Despite the morbidity of NXG, the literature is limited to case reports and small studies, and diagnostic criteria are lacking.
To evaluate the characteristics of NXG and propose diagnostic criteria.
Design, Setting, and Participants
This multicenter cross-sectional study was conducted at tertiary academic referral centers and followed by a systematic review and a consensus exercise. The multicenter cohort included patients with NXG diagnosed at the Brigham and Women’s and Massachusetts General Hospitals (2000-2018), the University of Iowa Hospitals and Clinics (2000-2018), and the University of Pennsylvania Health System (2008-2018). The systematic review was conducted in 2018 and included patients with NXG identified in the Cochrane, Ovid EMBASE, PubMed, and Web of Science databases. The consensus exercise was conducted by 8 board-certified dermatologists to identify diagnostic criteria.
Main Outcomes and Measures
Demographic factors, comorbidities, clinical features, and treatment response.
Of 235 included patients with NXG (34 from the multicenter cohort and 201 from the systematic review results), the mean (SD) age at presentation was 61.6 (14.2) years; 147 (62.6%) were female. Paraproteinemia was detected in 193 patients (82.1%), most often IgG-κ (117 patients [50.0%]). A malignant condition was detected in 59 patients (25.1%), most often multiple myeloma (33 patients [14.0%]). The overall rate of paraproteinemia and/or a malignant condition was 83.8% (197 patients). In the multicenter cohort, evolution of paraproteinemia into multiple myeloma was observed up to 5.7 years (median [range], 2.4 [0.1-5.7] years) after NXG presentation. Cutaneous lesions consisted of papules, plaques, and/or nodules, typically yellow or orange in color (113 of 187 [60.4%]) with a periorbital distribution (130 of 219 [59.3%]). The eye was the leading site of extracutaneous involvement (34 of 235 [14.5%]). In the multicenter cohort, intravenous immunoglobulin had the best treatment response rate (9 of 9 patients [100%]), followed by antimalarial drugs (4 of 5 patients [80%]), intralesional triamcinolone (6 of 8 patients [75%]), surgery (3 of 4 patients [75%]), chemotherapy (8 of 12 patients [67%]), and lenalidomide or thalidomide (5 of 8 patients [63%]). The consensus exercise yielded 2 major criteria, which were (1) clinical and (2) histopathological features consistent with NXG, and 2 minor criteria, consisting of (1) paraproteinemia, plasma-cell dyscrasia, and/or other associated lymphoproliferative disorder and (2) periorbital distribution of cutaneous lesions. In the absence of foreign body, infection, or another identifiable cause, fulfillment of both major and at least 1 minor criterion were proposed to establish the diagnosis of NXG.
Conclusions and Relevance
Necrobiotic xanthogranuloma is a multisystem disorder associated with paraproteinemia and malignant conditions. The proposed diagnostic criteria may advance clinical research and should be validated.
Necrobiotic xanthogranuloma (NXG) is a non–Langerhans cell histiocytosis first described by Kossard and Winkelmann1 in 1980. It is typically associated with paraproteinemia attributable to plasma-cell dyscrasias or lymphoproliferative disorders.2 In addition to serving as a skin sign of systemic disease, NXG may result in cosmetic disfigurement. Extracutaneous involvement including the eye, heart, gastrointestinal tract, liver, and lung may result in organ dysfunction and death.1,3-5 Clinically, yellow-to-orange papules, plaques, and/or nodules in a periorbital distribution are classic; however, other morphological characteristics and sites have been described.6 Histopathologically, NXG demonstrates palisading granulomas with lymphoplasmacytic infiltrate and zones of necrobiosis. Cholesterol clefts and Touton or large, bizarre foreign body giant cells are other classical features.6 The pathogenesis of NXG remains unknown; however, a role for paraprotein-lipoprotein interaction has been hypothesized.7,8 Evaluation of 12 skin-biopsy specimens by Wood et al9 in 2009 found no evidence to support a monoclonal plasma-cell proliferation within the inflammatory infiltrate.
The literature on NXG consists of case reports and small studies, and diagnostic criteria are currently unavailable. The first objective of this study was to evaluate the demographic features, comorbidities, clinical features, and treatment response of a large NXG cohort. The second objective was to propose diagnostic criteria for NXG.
The institutional review boards of Partners Healthcare, the University of Iowa, and the University of Pennsylvania approved this study. The requirement for informed consent was waived because of the retrospective nature of the study.
For the multicenter cohort, patients were identified from both the electronic health records and the dermatopathology databases at the Brigham and Women's Hospital, Massachusetts General Hospital, the University of Iowa Hospitals and Clinics, and the University of Pennsylvania Health System using International Classification of Diseases, Ninth Revision and Tenth Revision codes (273.2 and D76.3) and a free-text search for the terms necrobiotic xanthogranuloma and NXG. Patients were identified from 2000 through 2018 at the Brigham and Women's Hospital, Massachusetts General Hospital, and the University of Iowa Hospitals and Clinics and from 2008 through 2018 at the University of Pennsylvania Health System, with years selected based on availability of readily searchable electronic health records. The diagnosis was verified by consensus of the principal investigators (K.A.W., M.R., and A.M.) based on clinical and histopathological features. Exclusion criteria were an age younger than 18 years or a lack of histopathological features interpreted by a dermatopathologist as being consistent with NXG.
For the systematic review, records were identified with the support of a medical librarian in the Cochrane, Ovid EMBASE, PubMed, and Web of Science databases on November 20, 2018, using the medical subject headings (MeSH) terms “necrobiotic disorders” AND “histiocytosis, non-Langerhans-cell” OR “necrobiotic xanthogranuloma” in PubMed and a free text search for “necrobiotic xanthogranuloma” OR “NXG” OR “periocular xanthogranuloma” OR “periorbital xanthogranuloma” in all databases. Titles and abstracts were independently reviewed by 2 authors (C.A.N. and C.S.Z.). Full texts of records deemed relevant were independently reviewed by both authors, and a consensus was reached to include or exclude each article. Exclusion criteria were a patient age younger than 18 years, a study with aggregated data, an alternative diagnosis, a patient duplicated in another study, a language other than English, a lack of histopathological features interpreted by a dermatopathologist as being consistent with NXG, or an excluded study design (a conference abstract or a case report with insufficient data).
Using a standardized data collection instrument, we abstracted patient demographics, comorbidities, and clinical features at the time of NXG presentation. For the multicenter cohort, we also abstracted body mass index (calculated as weight in kilograms divided by height in meters squared), timing of paraproteinemia and malignant conditions relative to NXG, and antilipemic therapy, characteristics inconsistently reported in the literature. In addition, we classified treatment response as a complete response, a partial response, no response, or worsening based on the assessment of the patient’s dermatologist. Response rates were calculated as the percentage of patients with a complete or partial response. We did not evaluate response rates in the literature out of a concern for positive outcome bias.
After reviewing the results, 8 board-certified dermatologists (A.M., C.A.N., K.A.W., M.H.N., M.R., R.A.V., R.G.M., and S.I.) participated in a consensus exercise to propose diagnostic criteria for NXG. In the first online survey, participants identified all possible characteristics of NXG that could be considered for inclusion. In the second round, participants ranked the importance of each characteristic on a Likert scale ranging from 1 (not at all important) to 5 (very important), along with a free-response commentary. We used a mean importance ranking of 3.5 of 5 or greater as the numerical cutoff score for inclusion. In the third round, the participants reviewed the results of the previous survey and provided a free-response commentary on how to structure the criteria. In the final round, the participants reviewed draft criteria, indicated the level of agreement with each criterion using a Likert scale ranging from 1 (signaling strongly disagreement) to 5 (strong agreement), along with a free-response commentary, and indicated approval or disapproval. We required a mean agreement ranking of 3.5 of 5 or greater for each criterion and unanimous approval to establish consensus.
Continuous variables were summarized with means and SDs. Categorical variables were reported as proportions and percentages. All statistical analyses were performed in Excel version 14.7.1 (Microsoft).
For the multicenter cohort, the initial search yielded 71 patients at the Brigham and Women's Hospital and Massachusetts General Hospital (combined), 41 patients at the University of Iowa Hospitals and Clinics, and 61 patients at the University of Pennsylvania Health System. Fifteen patients (21%), 6 patients (15%), and 13 patients (21%) from these 3 centers, respectively, were included after review and verification, for a total of 34 patients.
In the systematic review, the initial search yielded 1199 records, 151 of which were included3,4,7,8,10-156 (Figure). These reports described 201 patients with NXG, with demographics, comorbidities, and clinical features (detailed in the eTable in the Supplement).
Overall, 235 patients with NXG were included in the study. The demographic features and comorbidities of patients with NXG are detailed in Table 1. The mean (SD) age at presentation was 61.6 (14.2) years; 147 patients (62.6%) were female. Most patients were white (78 of 90 for whom these data were available [87%]). Of the 235 patients, 193 (82.1%) had paraproteinemia. The most common subtype was IgG-κ, in 117 of 235 patients (50.0%); however, IgG-λ, IgM, and IgA subtypes were also observed (in 49 patients [20.8%], 5 patients [2.1%], and 2 patients [0.9%], respectively). Fifty-nine of 235 patients (25.1%) had a malignant condition. Multiple myeloma, in 33 patients (14.0%), was the most common type; however, lymphoma, leukemia, and Waldenstrom macroglobulinemia were also observed (in 1 patient [0.4%] each). Solid tumors were rare (2 of 235 [0.9%]) and observed exclusively in patients with coexistent paraproteinemia. The overall rate of paraproteinemia and/or a malignant conditions was 83.8% (affecting 197 patients). In the multicenter cohort, the timing of paraproteinemia relative to NXG was antecedent in 13 of 24 patients (54%), concurrent in 6 of 24 patients (25%), and subsequent in 4 of 24 patients (16.7%). Evolution of paraproteinemia into multiple myeloma was observed up to 5.7 years after NXG presentation (median [range], 2.4 [0.1-5.7] years); in contrast, lymphoma and leukemia preceded NXG presentation (by a median [range] of 5.25 [2.2-6.8] years).
The clinical features of NXG are detailed in Table 2 and the eTable in the Supplement. Primary morphological characteristics were papules and plaques (163 of 212 patients [76.7%]) and/or nodules (52 of 212 [24.5%]). Ulceration (40 of 211 patients [19.0%]) was the most common secondary feature, followed by telangiectasia (30 of 211 [14.2%]), atrophy (29 of 211 [13.7%]), and induration (15 of 211 [7.1%]). Diverse colors were observed, from yellow or orange (113 of 187 patients [60.4%]) to red or pink (45 of 187 [24.1%]), brown (28 of 187 [14.9%]), and purple (21 of 187 [11.2%]). Pain and pruritus were reported by 27 of 88 patients (30.7%) and 15 of 88 patients (17.0%), respectively. Of 218 patients, 197 (90.4%) presented with multiple cutaneous lesions. The most common distribution was periorbital (130 of 219 [59.3%]); however, lesions were distributed elsewhere on the face (50 of 219 [22.8%]), trunk (116 of 219 [53.0%]), extremities (123 of 219 [56.1%]), and within scars (12 of 219 [5.5%]). The eye was the leading site of extracutaneous involvement. In the multicenter cohort, the rate of eye involvement in patients evaluated by ophthalmology was 27% (4 of 15 patients); however, this denominator was not reliably reported in the literature. Other sites were the brain, gastrointestinal tract, heart, liver, lung, lymphoreticular system, muscle, and parotid gland.
Patients with NXG in the multicenter cohort had an overweight mean (SD) body mass index (27.0 [5.6]). Overall, 14 of 59 patients (24%) had a low high-density lipoprotein–cholesterol level; however, this rate in the multicenter cohort was higher (7 of 17 patients [41%]), even though 4 of 7 patients (38%) tested were taking 1 or more medications known to elevate high-density lipoprotein cholesterol levels. Thirty-three of 54 patients (61%) and 9 of 39 patients (23%) tested had low complement C4 levels and positive test results for cryoglobulins, respectively. Four of 11 patients (36%) tested had low 25-hydroxyvitamin D; however, 1,25-dihydroxyvitamin D was normal or high in half of these patients (2 of 4 [50%]).
Treatment response rates in the multicenter cohort are presented in Table 3. Of treatments with response reported in more than 3 patients (10%), intravenous immunoglobulin had the best response rate (9 of 9 patients [100%]), followed by antimalarials (4 of 5 patients [80%]), intralesional triamcinolone (6 of 8 patients [75%]), surgery (3 of 4 patients [75%]), chemotherapy (8 of 12 patients [67%]), and lenalidomide or thalidomide (5 of 8 patients [63%]).
The proposed diagnostic criteria for NXG are presented in the Box. The response rate for each round of the consensus exercise was 100%. Participants identified 13 characteristics in the first round and ranked their importance in the second round. In the third round, participants expressed preference for proposing major and minor criteria instead of a point-scoring system and requiring the exclusion of other causes, such as a foreign body and infection. Four characteristics that received mean importance rankings of 3.5 of 5 points or greater were drafted into major and minor criteria. In the final round, the major criteria received mean agreement rankings of 4.6 of 5 points and 4.5 of 5 points, respectively; the minor criteria received mean agreement rankings of 4.6 of 5 points and 3.9 of 5 points, respectively; and all participants approved the criteria, establishing consensus.
Cutaneous papules, plaques, and/or nodules, most often yellow or orange in color.
Histopathological features demonstrating palisading granulomas with lymphoplasmacytic infiltrate and zones of necrobiosis. Characteristic features that are variably present include cholesterol clefts and/or giant cells (Touton or foreign body).
Paraproteinemia, most often IgG-κ, plasma-cell dyscrasia, and/or other associated lymphoproliferative disorder.
Periorbital distribution of cutaneous lesions.
a Both major criteria and at least 1 minor criterion are required for diagnosis, applicable only in the absence of foreign body, infection, or other identifiable cause.
To our knowledge, this study provides the largest review to date of the demographics, comorbidities, clinical features, and treatment response of NXG. In addition, it presents the outcome of a consensus exercise to propose the first diagnostic criteria for NXG.
The mean age of 61.6 years in the cohort of 235 patients with NXG falls just above the range of 54 to 60 years reported in previous cohorts that included 10 or more patients.5,7,157,158 The female predominance in our cohort (63%) equates to an elevated female-to-male ratio of 1.7. While 1 study reported an elevated female-to-male ratio of 2.7,7 other studies reported an equal sex distribution157,158 or an elevated male-to-female ratio of 1.4.5 The rate of paraproteinemia in this cohort (82.1%) falls within the range of 77% to 84% reported by studies that did not require paraproteinemia as an inclusion criterion,5,158 and the rate of the IgG-κ subtype (50.0%) falls within the previously reported range of 50% to 60%.5,157 The rate of malignant conditions in this cohort (25.1%) is just less than the previously reported range of 26% to 45%, while the rate of multiple myeloma specifically (14%) falls within the previously reported range of 5% to 36%.5,7,157,158 The overall rate of paraproteinemia and/or malignant condition in this cohort was 83.8%. While paraproteinemia in the multicenter cohort most often preceded the presentation of NXG, evolution into multiple myeloma was observed up to 5.7 years afterwards, highlighting the importance of long-term hematologic surveillance.
The primary morphology of papules and plaques (76.7%) and/or nodules (24.5%) in the cohort was consistent with the literature5,7; however, the ulceration rate of 19.0% was less than the rate of 42% reported in a prior study.5 In keeping with the classic description by Kossard and Winkelmann,1 the most frequently observed colors in the cohort were yellow or orange (60.4%). Interestingly, a subset of patients reported pain (30.7%) and/or pruritus (17.0%), symptoms of NXG that have not received attention in prior studies. The most common distribution of cutaneous lesions was periorbital (59.3%), which falls within the previously reported range of 35% to 78%,5,7,157,158 and the eye was the leading site of extracutaneous involvement. Based on the 27% rate of eye involvement in patients evaluated by ophthalmology in this multicenter cohort, we suggest that an ophthalmology referral be considered for all patients with NXG. In addition, these data highlight the potential for multiorgan involvement, including the brain, gastrointestinal tract, heart, liver, lung, lymphoreticular system, muscle, and parotid gland.
Body mass index has been inconsistently reported in the literature; therefore, it is difficult to determine whether the overweight mean body mass index of 27.0 in the multicenter cohort is a reliable feature of NXG. Diagnosis with NXG has been associated with a low high-density lipoprotein–cholesterol phenotype,8 which was found in 24% of patients tested. In the multicenter cohort, 41% of patients tested had low high-density lipoprotein–cholesterol level, even though 38% were taking 1 or more medications known to elevate high-density lipoprotein–cholesterol levels. By comparison, approximately 18% of adults in the United States have low high-density lipoprotein–cholesterol levels.159 Interactions between the paraprotein and lipoproteins are hypothesized to impair macrophage-lipid homeostasis, promote systemic inflammation, and generate immune complexes. Necrobiotic xanthogranuloma has been associated with acquired early complement deficiencies and cryoglobulinemia.7,8 Of the patients tested in our cohort, 61% had low C4 levels, and 23% had cryoglobulinemia. Finally, we identified a pattern of low 25-hydroxyvitamin D with normal or high 1,25-dihydroxyvitamin D. Granuloma-mediated production of 1α-hydroxylase, a well-documented phenomenon in patients with sarcoidosis, may lead to increased synthesis of 1,25-dihydroxyvitamin D and subsequent hypercalcemia.10,160 Given small sample sizes, the associations between these abnormal results of laboratory tests and NXG merit further investigations.
In the multicenter cohort, intravenous immunoglobulin had the best response rate (100%), followed by antimalarial drugs (80%), intralesional triamcinolone (75%), surgery (75%), chemotherapy (67%), and lenalidomide or thalidomide (63%). A diversity of treatments for NXG has been reported; however, the literature consists of case reports, case series, and retrospective studies.161 Prospective, controlled clinical studies are essential to guide practice.
The participants in the consensus exercise to propose diagnostic criteria for NXG strongly agreed (mean ranking, 4.6 of 5 points) that cutaneous papules, plaques, and/or nodules should be a major criterion. Given that 60% of NXG lesions in the cohort were yellow or orange in color, this was included with the qualifier most often to indicate a helpful but not mandatory clinical feature. Whether histopathological testing should be required for diagnosis emerged as a key point of discussion. To differentiate NXG from clinical mimics, participants strongly agreed (mean ranking, 4.5 of 5 points) that palisading granulomas with lymphoplasmacytic infiltrate and zones of necrobiosis should be a major criterion. Cholesterol clefts and/or giant cells (Touton or foreign body) were included as characteristic but variably present histopathological features. A second key point of discussion was whether paraproteinemia, most often IgG-κ, plasma-cell dyscrasia, and/or other associated lymphoproliferative disorder should be required for diagnosis. Noting the absence of these associations in 16% of patients with NXG in our cohort, participants strongly agreed (mean ranking, 4.6 of 5 points) that this should be a minor criterion. Participants agreed (mean ranking, 3.9 of 5 points) that periorbital distribution of cutaneous lesions should be a minor criterion. Finally, whether the major criteria should be sufficient to establish the diagnosis of NXG emerged as a third key point of discussion. Participants expressed concern that necrobiosis lipoidica, given shared clinicopathological features, could be misdiagnosed as NXG. Therefore, in the absence of foreign body, infection, or other identifiable cause, both major criteria and at least 1 minor criterion were required for diagnosis. Ongoing research is necessary to validate these criteria so that they may guide clinical research efforts, such as selecting patients for clinical trials.
These results should be interpreted in the context of the study design. The retrospective nature of the multicenter cross-sectional study and systematic review may lead to information bias, and the tertiary care setting may limit generalizability to other study populations. A standardized data-collection tool was used to minimize variability; however, data were collected from physician notes and the literature, creating the potential for inconsistent reporting. Moreover, data collected at initial presentation to a tertiary academic referral center may differ from that collected at NXG onset. Response rates are limited by an inability to determine individual response for concurrent treatments. Finally, a potential limitation of the consensus-exercise method was the lack of face-to-face discussion. However, we opted to maintain anonymity to enable participants to formulate and express opinions independently.
This study of 235 patients describes the largest NXG cohort in the published literature, to our knowledge. The results provide insight into the demographics, comorbidities, clinical features, and treatment response of NXG. Eight board-certified dermatologists informed by these results participated in a consensus exercise to propose the first diagnostic criteria for NXG. While these criteria require validation, they represent a step forward in standardizing the diagnosis of this rare disorder for use in clinical research.
Accepted for Publication: November 7, 2019.
Corresponding Author: Arash Mostaghimi, MD, MPA, MPH, Department of Dermatology, Brigham and Women’s Hospital, Harvard Medical School, 221 Longwood Ave, Boston, MA 02115 (firstname.lastname@example.org).
Published Online: January 15, 2020. doi:10.1001/jamadermatol.2019.4221
Author Contributions: Drs Mostaghimi and Nelson 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 Wanat, Rosenbach, and Mostaghimi contributed equally and share senior authorship.
Concept and design: Nelson, Hashemi, Noe, Vleugels, Rosenbach, Mostaghimi.
Acquisition, analysis, or interpretation of data: Nelson, Zhong, Hashemi, Ashchyan, Brown-Joel, Imadojemu, Micheletti, Wanat, Rosenbach, Mostaghimi.
Drafting of the manuscript: Nelson, Vleugels.
Critical revision of the manuscript for important intellectual content: Zhong, Hashemi, Ashchyan, Brown-Joel, Noe, Imadojemu, Micheletti, Vleugels, Wanat, Rosenbach, Mostaghimi.
Statistical analysis: Nelson, Hashemi, Ashchyan.
Administrative, technical, or material support: Ashchyan, Brown-Joel, Micheletti, Wanat, Rosenbach, Mostaghimi.
Supervision: Vleugels, Wanat, Rosenbach, Mostaghimi.
Conflict of Interest Disclosures: Dr Brown-Joel reports grants from the National Center for Advancing Translational Sciences of the National Institutes of Health during the conduct of the study. Dr Rosenbach reports personal fees from Merck, grants and personal fees from Processa Pharma, and personal fees from aTyr Pharma outside the submitted work. Dr Vleugels reports research support from Pfizer outside the submitted work. Dr Mostaghimi reports personal fees from Pfizer and 3Derm, personal fees from and equity in Hims, equity in Lucid, and participation in clinical trials with Incyte, Aclaris, Concert, and Lilly, outside the submitted work. No other disclosures were reported.
Funding/Support: For the use of electronic medical records data collection service at University of Iowa Hospitals and Clinics, research reported in this article was supported by the National Center for Advancing Translational Sciences of the National Institutes of Health (grant UL1TR002537).
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.
Disclaimer: Dr Rosenbach is Deputy Editor and Dr Mostaghimi is Associate Editor of JAMA Dermatology, but they were not involved in any of the decisions regarding review of the manuscript or its acceptance. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
Additional Contributions: We thank medical librarian Paul Bain, PhD, MLIS, Harvard Medical School, for his assistance in developing our search strategy for the systematic review. He was not compensated for his contributions.
Create a personal account or sign in to: