Assessment of Changes in Diversity in Dermatology Clinical Trials Between 2010-2015 and 2015-2020

Key Points

Question

Has representation of racial and ethnic minority groups and women in randomized clinical trials for common dermatologic conditions changed from the 2010-2015 period to the 2015-2020 period?

Findings

In this systematic review of 392 randomized clinical trials conducted from 2015 to 2020, the reporting rate for race and ethnicity in US studies was higher in 2015-2020 than in 2010-2015. However, the proportion of articles including greater than or equal to 20% non-White representation remained unchanged, with psoriasis studies having the least diversity.

Meaning

The results of this study suggest that transparency in reporting of race and ethnicity has improved in US studies since 2010-2015, although inclusion of representative patient populations may be considered inadequate, particularly in psoriasis studies.

Abstract

Importance

Although there have been increased efforts in dermatologic research to improve representation of patient sex, race, and ethnicity, there are limited data evaluating resulting changes.

Objective

To characterize the diversity of participants in dermatologic clinical trials conducted in the US published from 2015 to 2020 pertaining to common dermatologic conditions affecting all patient demographic categories compared with findings from 2010-2015.

Evidence Review

A systematic literature review through the PubMed database was conducted for randomized clinical trials published between July 1, 2015, and July 1, 2020, using keywords alopecia areata, acne, atopic dermatitis, lichen planus, psoriasis, seborrheic dermatitis, and vitiligo. Data collected included distribution of participant demographic characteristics, funding source, and journal type. Reflecting US Census data, studies were defined as unrepresentative of race and ethnicity if they included less than 20% ethnically or racially diverse participants or unrepresentative of sex if they included less than 45% women. Python was used for statistical analysis by χ² tests or Fisher exact tests.

Findings

A total of 392 randomized clinical trials were included. In comparison with the period from 2010-2015, the reporting rate for race and ethnicity in US studies has increased from 59.8% to 71.9% (P = .05). However, the proportion of reporting articles including at least 20% non-White representation remains unchanged at 38.1% with 37 of 97 reporting randomized clinical trials in 2010-2015 and 53 of 139 reporting randomized clinical trials in 2015-2020 (P = .99). Psoriasis studies included the least diversity, with 12.1% of studies recording at least 20% non-White participants and 29.5% of studies recording at least 45% female participants.

Conclusions and Relevance

The findings of this systematic review suggest that reporting racial and ethnic data since 2010-2015 has become more transparent. However, inclusion of representative patient populations may still be considered inadequate, particularly in psoriasis studies. Diversity in clinical trials is important for representation of the affected patient populations, and additional efforts are warranted in support of this endeavor.

This systematic review reassesses representation since 2010-2015 of racial minority groups and women in randomized clinical trials published from 2015-2020 pertaining to dermatologic conditions that are well-studied, common, and affect patients within varied demographic categories.

Introduction

There is a call to improve disparities in diversity and equity in patient recruitment for dermatologic research. Charrow et al¹ characterized diversity in dermatologic randomized clinical trials (RCTs) published from 2010 to 2015 and found a low level of reporting for race and ethnicity data as well as low Hispanic representation.¹ Notable efforts to support diversity in RCTs have been made by the US Food and Drug Administration including an action plan in 2014 to support industry efforts at improving diversity in clinical trials and publishing concise information about who participated in clinical trials.² However, limited data are available to assess whether these efforts have been associated with increased diversity in dermatology research cohorts with respect to sex and race and ethnicity. To follow up on potential outcomes of these diversity efforts, we conducted a systematic review to reassess representation of racial and ethnic minority groups and women in RCTs published from 2015-2020 pertaining to dermatologic conditions that are well-studied, common, and affect patients within varied demographic categories.

Methods

Literature Search Strategy and Selection Criteria

A systematic literature review was conducted through a PubMed database search for all peer-reviewed English-language RCTs published between July 1, 2015, and July 1, 2020, using the keywords alopecia areata, acne, atopic dermatitis, lichen planus, psoriasis, seborrheic dermatitis, and vitiligo as used in Charrow et al.¹ A total of 1157 articles were identified after initial screening of titles and abstracts. Duplicates, non-English language articles, non-RCTs, and studies involving non-human participants or unrelated conditions were excluded. The flow diagram is available as the eFigure in the Supplement. Secondary screening with review of full text excluded studies conducted completely outside the US. Institutional review board review was waived because all data were obtained from publicly available sources. A meta-analysis was not performed for consistency with the Charrow et al¹ study and because this systematic review did not pool patient representation in a weighted manner from the included studies.

Data Extraction

Data were collected via full-text review of included articles, and any uncertainties were resolved through discussion until reviewers reached a consensus. Article screening and data collection were conducted by 2 independent reviewers (V.C. and S.A.). To ensure accuracy and consistency, data from the top 20% of articles based on participant cohort size (63% of total patients in the analysis) were validated by 2 of us (V.C. and S.A.). Study data were managed through a spreadsheet (Microsoft Excel), following data extraction categories from Charrow et al.¹

Categorization of race and ethnicity was consistent with the original study.¹ Individuals who indicated any race other than White, including those with multiracial background, were considered non-White for bivariate analysis. Studies were defined as unrepresentative of race and ethnicity if they included less than 20% ethnically or racially diverse participants (non-White race or Hispanic ethnicity) and as unrepresentative of sex if they included less than 45% female participants. Cutoff values reflect US census data during the period when these RCTs were published (2015-2020) and correspond with the Charrow et al¹ parameters of assessing representation level amongst sex and race and ethnicity.

Statistical Analysis

Representations of race and ethnicity as well as sex were presented as counts and percentages for each disease type, funding source, and journal type. Statistical significance of differences in representation was determined by χ² tests or Fisher exact tests as appropriate with statistical significance set at P ≤ .05. Analyses were performed using Python version 3.6.10 (Python Software Foundation).

Results

A total of 1157 unique articles were identified. After eliminating articles that did not meet inclusion criteria, 392 were included for data analysis. Of these, 139 studies were conducted exclusively in the US and 253 studies were conducted partially in the US. Conclusions on representation for vitiligo and seborrheic dermatitis trials were limited due to small sample size.

Reporting and Distribution of Race and Ethnicity

Race and ethnicity demographic characteristics for at least 1 category were reported in 295 of 392 studies (75.2%) conducted either exclusively or partially in the US, and in 100 of 139 studies (71.9%) conducted exclusively in the US. Funding source was associated with reporting racial and ethnic data (Table 1). Studies funded by government (n = 5) and government/industry mix (n = 5) more likely to report race and ethnicity (4 of 5 [80.0%]) than those funded by other organizations/institutions including industry (77.5%), foundation/nonprofit (54.5%), no funding (20.0%), or unspecified funding (73.3%) (P < .001).

Table 1. Publication Characteristics by Race Reporting for Dermatology RCTs.

Publication characteristic	No. of all RCTs	RCT reported race and ethnicity, No. (%)	P valuea	No. of exclusively US RCTs	RCT reported race and ethnicity, No. (%)	P valuea	≥20% Non-White representation, No. (%)b	P valuea
Overall	392	295 (75.3)		139	100 (71.9)		53 (38.1)
Funding
Government	5	4 (80.0)	.001	4	3 (75.0)	.13	3 (75.0)	.38
Industry	346	268 (77.5)		102	78 (76.5)		36 (35.3)
Government/industry mix	5	4 (80.0)		4	3 (75.0)		2 (50.0)
Foundation/nonprofit	11	6 (54.5)		10	6 (60.0)		6 (60.0)
No funding	10	2 (20.0)		7	2 (28.6)		2 (28.6)
Funding not specified	15	11 (73.3)		12	8 (66.7)		4 (33.3)
Disease type
Psoriasis	268	202 (75.4)	.14	66	49 (74.2)	.73	8 (12.1)	<.001
Eczema	64	50 (78.1)		38	28 (73.7)		27 (71.1)
Acne	53	38 (71.7)		32	20 (62.5)		17 (53.1)
Vitiligo	2	2 (100)		0	0		0
Alopecia areata	3	3 (100)		3	3 (100)		1 (33.3)
Seborrheic dermatitis	2	0		0	0		0
Journal type
Dermatology	258	200 (77.5)	.06	111	82 (73.9)	.52	41 (36.9)	.18
General medicine	31	25 (80.6)		7	4 (57.1)		2 (28.6)
Dermatology pharmacology	12	6 (50.0)		3	1 (33.3)		0
General pharmacology	9	9 (100)		3	3 (100)		3 (100)
Allergy	16	12 (75.0)		5	3 (60.0)		2 (40.0)
Other/undetermined	13	8 (61.5)		9	6 (66.7)		5 (55.6)
Rheumatology	53	35 (66.0)		1	1 (100)		0

Abbreviation: RCT, randomized clinical trial.

^a P values from χ² or Fisher exact tests as appropriate.

^b The percentage is calculated from exclusively US-based studies that report race.

Of the 100 studies conducted exclusively in the US that reported race or ethnicity, eczema studies and acne studies were more likely to include at least 20% racially or ethnically diverse patients than psoriasis studies (eczema studies, 27 [71.1%] and acne studies, 17 [53.1%] compared with psoriasis studies, 8 [12.1%]; P < .001) (Table 1). US-based psoriasis studies were least diverse, with only 12.1% of studies (8 of 66) recording at least 20% non-White representation and 85.2% of total reported participants (11277 of 13238) being White.

Reporting and Distribution of Sex

Distribution of sex was reported in 359 of 392 articles (91.6%). Articles about psoriasis and eczema were most likely to report sex distribution (psoriasis, 247 [92.2%] and eczema, 62 [96.9%] compared with acne, 44 [83.0]; P = .03) (Table 2).

Table 2. Publication Characteristics by Sex Reporting for Dermatology RCTs.

Publication characteristic	No. of all RCTs	RCT reported sex, No. (%)	P valuea	≥45% Women representation, No. (%)b	P valuea
Overall	392	359		162
Funding
Government	5	4 (80.0)	.64	3 (60.0)	.10
Industry	346	319 (92.2)		137 (39.6)
Government/industry mix	5	5 (100)		1 (20.0)
Foundation/nonprofit	11	9 (81.8)		6 (54.5)
No funding	10	9 (90.0)		8 (80.0)
Funding not specified	15	13 (86.7)		7 (46.7)
Disease type
Psoriasis	268	247 (92.2)	.03	79 (29.5)	<.001
Eczema	64	62 (96.9)		38 (59.4)
Acne	53	44 (83.0)		40 (75.5)
Vitiligo	2	1 (50.0)		0
Alopecia areata	3	3 (100)		3 (100)
Seborrheic dermatitis	2	2 (100)		2 (100)
Journal type
Dermatology	258	236 (91.5)	.87	89 (34.5)	<.001
General medicine	31	29 (93.5)		14 (45.2)
Dermatology pharmacology	12	10 (83.3)		2 (16.7)
General pharmacology	9	9 (100)		1 (1.11)
Allergy	16	14 (87.5)		7 (43.8)
Other/undetermined	13	12 (92.3)		5 (38.5)
Rheumatology	53	49 (92.5)		44 (83.0)

Abbreviation: RCT, randomized clinical trial.

^a P values from χ² or Fisher exact tests as appropriate.

^b The percentage is calculated from studies that report sex.

Of the studies analyzed, 41.3% of articles (162 of 392) included at least 45% female representation. Journal and disease type were associated with the representation of female participants in a given study (Table 2). For instance, rheumatology journal articles were more likely to have at least 45% female representation than dermatology or general medicine articles (rheumatology journal articles, 44 [83.0%] vs dermatology, 38 [59.4%]) and general medicine, 14 [45.2%]; P < .001), and acne clinical trials were more likely to have at least 45% female representation than psoriasis trials (acne clinical trials, 40 [75.5%] vs psoriasis trials, 79 [29.5%]; P < .001).

Comparison Between 2010-2015 Data and 2015-2020 Data

In comparison with the Charrow et al¹ study from 2010-2015, we found that reporting for race and ethnicity in US studies increased from 59.8% to 71.9% (P = .05). However, the proportion of reporting articles including at least 20% non-White representation remained unchanged at 38.1%, with 37 of 97 reporting RCTs in 2010-2015 and 53 of 139 reporting RCTs in 2015-2020 (P = .99).

Discussion

While our findings suggest that racial and ethnic reporting since 2010-2015 is more transparent, there have not been significant changes in representative inclusion for racial and ethnic minority groups. Prior analysis found that RCTs from 2010-2015 reflected lack of reporting for race and ethnicity and Hispanic representation.¹ In the 2015-2020 period, the reporting rate for racial and ethnic demographic characteristics among studies within the US increased by 12.1%. Charrow et al¹ found that articles with industry funding and dermatology pharmacology journals were more likely to report racial and ethnic data, but we found no such association with journal type. Albeit representing a small number of studies, government funding was associated with reporting of race and ethnicity, which may reflect efforts by government organizations such as the US Food and Drug Administration to support diversity in clinical trials.² The inclusion in Charrow et al¹ of non-US trials may be a factor in this discrepancy, or the discrepancy may reflect recent recommendations on the reporting of race and ethnicity in medical journals to encourage transparency in study diversity.³

The level of racial representation (≥20% non-White participants) has not changed from the initial analysis of Charrow et al¹ (Table 3). Different characterizations of Hispanic identity as race or ethnicity may limit our analysis of Hispanic representation. Similar to the prior study, US-based eczema and acne study populations had the greatest racial and ethnic diversity and psoriasis studies were predominantly White (85.2%). While this finding may reflect the slight predominance of White participants reported by several epidemiologic studies on psoriasis, prevalence among other racial groups may be underestimated and should be represented in studies.^4,5,6 Notably, female representation among psoriasis studies decreased from 87.2% of reporting studies having at least 45% women in the 2010-2015 period to 29.5% in the 2015-2020 period. The inclusion in Charrow et al¹ of exclusively non-US studies for analysis of sex representation may be a factor in this difference. Inclusion of racial and ethnic diversity in RCTs should reflect the diversity of the patient populations.

Table 3. Comparison of Representation Between 2010-2015 and 2015-2020.

Source	Exclusively US RCTs, No.	RCT reported race and ethnicity No. (%)	P valuea	≥20% Non-White representation, No. (%)	P valuea
Current study (2015-2020)	139	100 (71.9)	.05	53 (38.1)	.99
Charrow et al1 (2010-2015)	97	58 (59.8)		37 (38.1)

Abbreviation: RCT, randomized clinical trial.

^a P values from χ² or Fisher exact test as appropriate.

Limitations

There are various limitations to this review. First, PubMed was utilized as the sole database to identify eligible studies. Secondly, our analysis of Hispanic representation was limited by articles’ differing characterization of ethnically Hispanic study participants as ethnically Hispanic or racially Hispanic. Additionally, as our analysis was limited to outcomes of patient representation, we did not explore extraneous factors including patient recruitment methods and local demographics of participating regions. Finally, the scope of our study was limited to the dermatological conditions included in the original study, to facilitate direct comparison between time periods. However, dermatologic research has expanded significantly within the last 5 years in other areas including hidradenitis suppurativa, prurigo nodularis, and frontal fibrosing alopecia. As such, future studies on clinical trial patient representation should similarly extend to encompass other dermatologic conditions.

Conclusions

The results of this systematic review suggest that while reporting of racial and ethnic data has become more transparent from the 2010-2015 period to the 2015-2020 period, inclusion of representative patient populations (particularly in psoriasis studies) has not statistically significantly changed.

Supplement.

eFigure. Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) Flow Diagram