Sex and racial differences in systemic lupus erythematosus among U.S. adults in the All of Us Research Program

Abstract

Objective:

Men with systemic lupus erythematosus (SLE) are an understudied demographic. This report characterized differences between males and females with SLE.

Methods:

We examined cross-sectionally participants with SLE in the All of Us Research Program, a U.S. cohort with participant survey at enrollment (May 2018-June 2022) and linked electronic health record (EHR) data. We described and compared characteristics of males and females with SLE encompassing disease manifestations and prescribed medications from EHR data and socioeconomic factors, including health literacy and healthcare access and utilization, from surveys. We reported racial variations stratified by sex.

Results:

Of 1,462 participants with SLE, 126 (9%) were male. Males reported lower educational attainment and less fatigue than females. Myocardial infarction was significantly more common in males. Males had significantly less confidence in completing medical forms than females, and a trend towards requiring more help in reading health-related materials. Barriers to healthcare access and utilization were common in both males and females (40% versus 47% reporting some reason for delay in care; P-value=0.35). Women of race other than Black or African American or White more often reported delaying care due to cultural differences between patient and provider.

Conclusion:

Our study demonstrated major clinical and health literacy differences in males and females with SLE. Socioeconomic factors were significant barriers to healthcare in both sexes. Our study suggests males have disproportionately poorer health literacy, which may exacerbate pre-existing disparities. Further large, prospective studies, focusing on recruiting men, are needed to better characterize racial differences in males with SLE.

Systemic lupus erythematosus (SLE) is approximately nine times more common in women than men in the United States.¹ Sex-specific differences in sex hormones, toll-like receptor expression, and microRNA profiles may play a role in the sex-dependent susceptibility to SLE.² SLE is still a substantial burden in men, with an estimated prevalence of 14.6 per 100,000 person-years¹ and perhaps a more aggressive clinical course.³ In men and women, SLE disproportionately affects certain racial and ethnic underserved populations.¹ Black or African American and Hispanic or Latino persons with SLE have higher disease activity, more disease-related complications, and excess mortality compared with non-Hispanic or Latino White persons.⁴

Several studies report sex differences in SLE disease manifestations, but findings are inconsistent across cohorts and sample sizes of men are often small.^5–10 Even less understood is the role of race in the clinical phenotype of SLE in men. Rare reports have identified racial differences in disease manifestations in men with SLE,^{5, 11–15} but differences are incompletely characterized. The limited data on how socioeconomic factors differ between men and women with SLE is conflicting, but education level and income may vary by sex.^{3, 5}

Healthcare cost and access disparities have been reported to disproportionately impact persons with SLE.¹⁶ Men with SLE have reported more perceived difficulty in accessing healthcare than women,³ and this access disparity is corroborated by literature demonstrating less outpatient clinic, including rheumatology subspecialty, visits in men.^{17, 18} Sex differences in healthcare utilization may be altered by race as Black or African American women with SLE were found to be significantly less likely to be referred to a rheumatologist compared to their White male counterparts.¹⁸ However, further data on sex differences in healthcare access and utilization by race is needed.

The purpose of this report was to characterize differences in SLE clinical manifestations, prescribed medications, and socioeconomic determinants of health, including health literacy and barriers to healthcare access and utilization, by sex and race among men with SLE. To this end, we utilized data from the All of Us Research Program, a U.S. national, de-identified data repository consisting of both patient survey and linked electronic health record (EHR) data.

Patients and Methods

Study cohort

The National Institutes of Health (NIH) All of Us Research Program is an ongoing longitudinal study aimed at recruiting 1 million volunteers representative of the U.S. population to contribute data to the All of Us data repository with the goal of accelerating biomedical research and improving health. All of Us study procedures have been previously described.¹⁹ In brief, adults age 18 years and older who reside within the U.S. or a U.S. territory are eligible to participate. All of Us initiated enrollment in May 2018 and participants enroll and consent to participate either via the All of Us website (https://joinallofus.org) or a smart-phone application. Volunteer participants are invited to complete several health surveys, composed of validated instruments or questions when appropriate.²⁰ Participants may opt to provide authorization to share EHR data, in which case survey data is linked to billing codes, medication history, laboratory results, and encounter records from EHR data from any of 60 health care provider organizations in All of Us’ network and also, in a subset, from other providers using Fast Healthcare Interoperability Resources (FHIR)-based connections. Each participant is also eligible to undergo an initial evaluation for physical measurements. The All of Us protocol was approved by the Institutional Review Board of the All of Us Research Program, which follows the regulations and guidance of the NIH Office for Human Research Protections.

Participants with SLE were identified from the All of Us database, version 6. This version was released in June 2022 and accessed in November 2022. We restricted our cohort to participants with linked EHR data, as the accuracy of self-reported SLE diagnosis has been demonstrated to be poor.²¹ We included a participant in the All of Us database as having SLE if they had ≥ 3 SLE diagnosis codes on separate occurrences and had ever been prescribed an antimalarial medication (including hydroxychloroquine, chloroquine, or quinacrine). SLE diagnosis codes accepted included: International Classification of Diseases (ICD), Ninth Revision (ICD-9) code 710.0; ICD, Tenth Revision (ICD-10) codes M32.1, M32.8 or M32.9; or Standardized Nomenclature of Medicine (SNOMED) code 55464009. We excluded participants with a medical billing diagnosis code for dermatomyositis (ICD-9: 710.3; ICD-10: M33.0, M33.1) or systemic sclerosis (ICD-9: 710.1; ICD-10: M34). This algorithm has been validated to have 88-91% positive predictive value in correctly identifying participants with SLE from EHR databases using ICD-9 billing codes.²²

Participant characteristics

Participants were categorized as male or female based on self-reported biological sex assigned at birth. Other sociodemographic data was self-reported in “The Basics” enrollment survey and included age at consent for study participation, race (Black or African American, Other race, or White), ethnicity (Hispanic or Latino versus any other response), highest level of educational attainment, health insurance provider, employment status, and annual household income category. Other race is a composite of responses: “Another Single Population,” “Asian,” “More Than One Population,” “None Indicated,” or “None of These.”

The enrollment “Lifestyle” questionnaire captured self-reported cigarette smoking status, current alcohol use, and current and ever marijuana use. Current cigarette smoking was defined by self-report of smoking (either “some days” or “every day”) of cigarettes. Former cigarette smokers reported having ever smoked at least 100 cigarettes in their lifetime, but now smoking cigarettes “not at all.” Current heavy alcohol use was defined as present if the participant self-reported drinking a drink containing alcohol two times per week or more and self-reported greater than or equal to three alcoholic drinks on a typical day when they drink. Current marijuana use describes any self-reported marijuana use (including cannabis, pot, grass, hash, weed, etc.) in the past three months.

An enrollment survey on “Overall Health” ascertained general health metrics of average pain level, general health perception, and fatigue level. Average pain level was defined by self-reported average pain over the past seven days on a scale of 0 (no pain) to 10 (worst pain imaginable). We categorized these scores as mild (0-3), moderate (4-7), and severe (8-10). Participants were asked, “In general, would you say your health is: poor, fair, good, very good, or excellent”? Participants rated their fatigue over the past seven days as none, mild, moderate, severe or very severe.

Physical measurements were obtained per standardized protocol by a trained program staff member following enrollment with patient consent or from linked EHR data.¹⁹ Physical metrics included height and weight, from which body mass index (BMI) was calculated.

SLE disease manifestations and prescribed medications

Organ-specific SLE disease involvement was identified for each participant from EHR data using ICD-9, ICD-10, SNOMED, and/or Current Procedural Terminology, 4^th Edition (CPT-4) codes as detailed in Supplementary Table S1. Cardiovascular disease (including myocardial infarction, coronary artery disease with or without angina, limb claudication, congestive heart failure, peripheral arterial disease, and stroke), lupus nephritis, end-stage renal disease (ESRD), antiphospholipid syndrome, lupus pericarditis, lupus lung disease, and Raynaud’s phenomenon were evaluated because sex-specific differences have previously been reported for these features.^{6, 23–28}

Medications commonly prescribed in the treatment of SLE were ascertained from prescription drug history in linked EHR data, including mycophenolate mofetil or mycophenolic acid, azathioprine, cyclophosphamide (oral or intravenous), methotrexate, tacrolimus, rituximab, and belimumab.²⁹ Participants were considered as having had a prescribed medication if it was currently or ever previously prescribed to that participant. There were no participants in All of Us, version 6, prescribed anifrolumab or voclosporin.

Health literacy

The “Overall Health” enrollment survey includes three questions modified from the Brief Health Literacy Screen:²⁰ (1) how confident are you filling out medical forms by yourself: extremely, quite a bit, somewhat, a little bit, not at all, prefer not to answer; (2) how often do you have someone help you read health-related materials: always, often, sometimes, occasionally, never, prefer not to answer; and (3) how often do you have problems learning about your medical condition because of difficulty understanding written information: always, often, sometimes, occasionally, never, prefer not to answer. Factor analysis showed these questions to be the three most strongly explanatory factors of health literacy among the survey items, with coefficients of (1) −0.586, (2) 0.755, and (3) 0.748, respectively.²⁰ We defined lack of confidence in completing medical forms as present if a participant responded somewhat, a little bit, or not at all to question (1). We defined requiring help reading health-related materials as present if the participant responded always, often or sometimes to question (2). We defined having difficulty understanding written health information as present if the participant responded always, often or sometimes to question (3).

Healthcare access and utilization

Domains of healthcare access and utilization were assessed on the “Health Care Access & Utilization” survey administered after enrollment. On this survey, participants were queried if they had delayed getting care in the past 12 months for any of the following reasons: (1) could not afford the copay, (2) insurance deductible was too high or could not afford the deductible, (3) they had to pay out of pocket for some or all of the procedure, (4) they could not get time off of work, (5) they could not get childcare, (6) they could not get elderly care, (7) they did not have transportation, (8) they live in a rural area where distance to the health care provider is too far, or (9) they were nervous to see a healthcare provider. We collapsed these delays into meaningful categories for analysis: delay due to affordability (1-3); delay due to time constraint (4-6); delay due to transportation (7-8); and (9) alone.

Any medication challenge due to a cost barrier was defined as presence of any of the following actions reported to save money in the past 12 months: skipped medication doses, took less medicine, delayed filling a prescription, bought prescription medications from another country, requested a lower cost medication from their doctor, or used alternative therapies. Participants were also asked “how often have you either delayed or not gone to see doctors or health care providers because they were different from you in any of these ways (race, ethnicity, religion, beliefs, or native language)? Always, most of the time, some of the time, none of the time, don’t know, did not answer.” We defined ever having delayed care because their healthcare provider was different from them if a participant responded always, most of the time, or some of the time on this item.

Statistical analyses

Descriptive statistics were calculated as n (%) by sex, and by race stratified by sex. According to the All of Us Research Program data and statistics dissemination policy, cell values and aggregate statistics that correspond to 1 to 20 participants are obscured to protect participant privacy (reported as ≤20 in tables). Inferential statistics compared sociodemographic, disease manifestations, prescribed medications, health literacy items, and barriers to healthcare access and utilization by Chi-squared test with Yates’ continuity correction. All analyses were performed using the All of Us Researcher Workbench (version 6) and R environment for statistical computing.³⁰ An alpha level of 0.05 was pre-specified. A Bonferroni adjusted alpha level was also considered and reported, adjusting for the 25 simultaneous Chi-squared tests (adjusted alpha level of 0.002). P values that, in combination with presented summary statistics, would allow the reader to deduce a participant count of 1 to 20 were reported as not applicable (N/A).

Results

We identified 1,462 participants with SLE, including 126 males (9% of the total cohort) (Figure 1). Baseline characteristics of the study cohort by sex are summarized in Table 1.

Figure 1:

Study flow chart

EHR: electronic health record; SLE: systemic lupus erythematosus; ICD-9: International Classification of Diseases, ninth revision; ICD-10: International Classification of Diseases, tenth revision

Dermatomyositis: ICD-9 code 710.3; ICD-10 codes M33.0, M33.1

Systemic sclerosis: ICD-9 code 710.1; ICD-10 code M34

Antimalarial includes hydroxychloroquine, chloroquine or quinacrine

Table 1.

Baseline characteristics of study cohort by sex, n (%)¹

Population Characteristics	Males (n=126)	Females (n=1336)	P values
Age (at consent for study participation)
18 to 44 years	41 (33)	512 (38)	0.08
45 to 64 years	57 (45)	625 (47)
65 or more years	28 (22)	199 (15)
Race			0.13
Black or African American	29 (24)	403 (31)
Other race2	44 (36)	371 (28)
White	50 (41)	538 (41)
Hispanic or Latino Ethnicity	35 (28)	315 (24)	0.34
Body mass index (BMI) (kg/m2) category			1.00
Underweight or Normal (BMI <25)	33 (27)	354 (27)
Overweight or Obese (BMI ≥25)	89 (73)	947 (73)
Highest educational attainment3			0.02§
At most high school completion	49 (39)	373 (28)
At least one year of college	77 (61)	941 (72)
Health insurance status4			0.73
Exclusively Government insurance	56 (62)	608 (60)
Exclusively Non-government insurance	34 (38)	410 (40)
Employment status			0.76
Employed for wages or self-employed	41 (35)	461 (37)
Not currently employed for wages	76 (65)	788 (63)
Annual household income category			0.09
Less than $35,000	44 (45)	566 (54)
More than $35,000	54 (55)	477 (46)
Cigarette smoking status5			0.02§
Current smoker	24 (20)	153 (12)
Former smoker	29 (24)	253 (20)
Never smoker	70 (57)	874 (68)
Current heavy alcohol use6	≤20 (≤16)	39 (29)	N/A
Ever marijuana use	64 (51)	560 (42)	0.07
Current marijuana use7	≤20 (≤16)	126 (9)	N/A
Pain level8			0.06
Mild	48 (41)	382 (30)
Moderate	46 (39)	596 (48)
Severe	23 (20)	276 (22)
General health perception9			0.96
Good to Excellent	53 (43)	555 (42)
Fair to Poor	71 (57)	765 (58)
Fatigue level10			0.01§
Mild or less	55 (45)	426 (32)
Moderate to very severe	68 (55)	895 (68)

N/A: not applicable

^§Designates statistical significance at alpha level 0.05; No P values remained statistically significant at Bonferroni adjusted alpha level of 0.002 (25 simultaneous tests)

¹Data summarized as n (%); according to the All of Us Research Program data and statistics dissemination policy, cell values and aggregate statistics that correspond to 1 to 20 participants are obscured to protect participant privacy. In these cases, P values are designated as N/A.

²Other race corresponds to participants who self-reported as Another Single Population, Asian, More Than One Population, None Indicated, or None Of These

³Highest educational attainment: At most high school completion includes less than a high school degree or equivalent and highest grade completed as twelve or GED; At least one year of college includes persons who completed one to three years of college or are a college graduate or have another advanced degree

⁴Government insurance includes Indian, Medicaid, Medicare, Military and Veterans Affairs health insurance categories; Non-government insurance includes Employer or union, purchased, or other health plan health insurance categories

⁵Current smoker defined by self-report of smoking (either “some days” or “every day”) of cigarettes; Former cigarette smoker defined by self-report of having ever smoked at least 100 cigarettes in their lifetime but now smoking cigarettes “not at all”

⁶Current heavy alcohol use defined as present if the participant self-reports drinking a drink containing alcohol 2 times per week or more AND self-reports ≥ 3 alcoholic drinks on a typical day when they drink; an alcoholic drink includes a bottle of beer, a glass of wine or a wine cooler, a shot of liquor, or a mixed drink with liquor in it

⁷Current use defined as self-report of marijuana use in the past 3 months

⁸Pain level defined by self-reported average pain over the past 7 days on a scale of 0 to 10 with 0 being no pain and 10 being worst pain imaginable; mild corresponds to a score of 0-3; moderate a score of 4-7; severe a score of 8-10

⁹General health perception defined by self-reported response to survey query “In general, would you say your health is:” rated as fair to poor (fair, poor) versus good to excellent (good, very good or excellent)

¹⁰Fatigue level defined by self-reported fatigue over the past 7 days rated as mild or less (mild, none) versus moderate to very severe (moderate, severe, or very severe)

Sex differences in sociodemographic and lifestyle characteristics

The age distribution of males and females in the SLE cohort was similar, with most participants of either sex less than 65 years old at consent for participation in the study. Among males, White race was the most commonly represented (41%), followed by Other race (36%) and then Black or African American race (24%). Twenty-eight percent of males identified as Hispanic or Latino. Among females, the ethnic and racial distribution was similar: 31% Black or African American, 28% Other race, 41% White, 24% Hispanic or Latino.

Most males (73%) and females (73%) were overweight or obese. Males showed lower educational attainment (P=0.02). The majority of participants of both sexes were not employed (63-65%) and on exclusively government-issued insurance (60-62%). Annual household income less than $35,000 was common in both males (45%) and females (54%). Males were more likely to be current or former cigarette smokers (P=0.02). Males were less likely to report moderate to severe fatigue (P=0.01).

Sex differences in SLE disease manifestations and prescribed medications

Organ-specific SLE disease manifestations and prescribed medications by sex are summarized in Table 2. Males were more likely than females to have any cardiovascular event, and in particular, myocardial infarction and coronary artery disease with or without angina. The difference in cross-sectional incidence of myocardial infarction in males (18%) compared to females (8%) remained statistically significant with Bonferroni correction of alpha significance level (P<0.002). There was a trend towards males having more lupus nephritis (25% versus 18%) and antiphospholipid syndrome (19% versus 13%) compared to females, but these differences did not reach statistical significance. Medication prescription rates of mycophenolate mofetil or mycophenolic acid and methotrexate did not differ by sex. The remainder of comparisons by sex for prescribed medications were limited due to the small sample size in males.

Table 2.

Systemic lupus erythematosus disease manifestations and prescribed medications by sex, n (%)¹

	Males (n=126)	Females (n=1336)	P values
Cardiovascular disease manifestations
Any cardiovascular event 2	58 (46)	442 (33)	0.005 §
Myocardial infarction	22 (18)	102 (8)	<0.001§,†
Coronary artery disease with or without angina	41 (33)	292 (22)	0.009§
Limb claudication	≤20 (≤16)	≤20 (≤1.5)	N/A
Congestive heart failure	25 (20)	188 (14)	0.10
Peripheral arterial disease	≤20 (≤16)	47 (4)	N/A
Stroke	≤20 (≤16)	77 (6)	N/A
Other organ-specific manifestations
Lupus nephritis	31 (25)	235 (18)	0.07
End-stage renal disease	≤20 (≤16)	70 (5)	N/A
Antiphospholipid syndrome	24 (19)	172 (13)	0.07
Lupus pericarditis	≤20 (≤16)	56 (4)	N/A
Lupus lung disease	≤20 (≤16)	47 (4)	N/A
Raynaud’s phenomenon	≤20 (≤16)	241 (18)	N/A
SLE medication: Ever prescribed 3
Mycophenolate mofetil or mycophenolic acid	40 (32)	328 (25)	0.09
Azathioprine	≤20 (≤16)	253 (19)	N/A
Cyclophosphamide (oral or intravenous)	≤20 (≤16)	45 (3)	N/A
Methotrexate	29 (23)	324 (24)	0.84
Tacrolimus	≤20 (≤16)	157 (12)	N/A
Rituximab	≤20 (≤16)	84 (6)	N/A
Belimumab	≤20 (≤16)	108 (8)	N/A

SLE: Systemic lupus erythematosus; N/A: not applicable

^§Designates statistical significance at alpha level 0.05

^†Designates statistical significance at Bonferroni adjusted alpha level of 0.002 (25 simultaneous tests)

¹Data summarized as n (%); according to the All of Us Research Program data and statistics dissemination policy, cell values and aggregate statistics that correspond to 1 to 20 participants are obscured to protect participant privacy. In these cases, P values are designated as N/A.

²Any cardiovascular event includes myocardial infarction, coronary artery disease with or without angina, limb claudication, congestive heart failure, peripheral arterial disease, stroke, undergoing percutaneous transluminal coronary angioplasty, or undergoing coronary artery bypass grafting

³Participant was considered to have ever been prescribed a specific medication or class if there was a past or current prescription documented in the electronic health record for that medication

Racial differences in SLE disease manifestations and prescribed medications

Organ-specific SLE disease manifestations and prescribed medications by race, stratified by sex, are summarized in Supplementary Table S2. White females reported more Raynaud’s phenomenon (11%) compared to Black or African American (4%) or Other race (5%). Otherwise, findings were similar across racial groups studied among females.

Sex differences in health literacy

Sex differences in the three domains of health literacy are shown in Table 3. A significantly larger proportion of males reported a lack of confidence completing medical forms than females (23% in males versus 12% in females; P=0.0017). This difference remained statistically significant with Bonferroni correction of alpha significance level (P<0.002). Males also more frequently reported requiring help reading health-related materials (30% of males versus 19% of females; P=0.009). The proportion of males and females having difficulty understanding written health information was similar.

Table 3.

Health literacy by sex and by race, stratified by sex, n (%)¹

	Males	Females	P values
Participants of All Races	n=125	n=1326
Lack of confidence in completing medical forms2	28 (23)	162 (12)	0.0017§,†
Requiring help reading health-related materials3	37 (30)	254 (19)	0.009§
Having difficulty understanding written health information4	28 (23)	231 (18)	0.22
Black or African American	n=29	n=401
Lack of confidence in completing medical forms2	≤20 (≤69)	64 (16)	N/A
Requiring help reading health-related materials3	≤20 (≤69)	86 (21)	N/A
Having difficulty understanding written health information4	≤20 (≤69)	85 (21)	N/A
Other Race 5	n=44	n=366
Lack of confidence in completing medical forms2	≤20 (≤45)	66 (18)	N/A
Requiring help reading health-related materials3	≤20 (≤45)	102 (28)	N/A
Having difficulty understanding written health information4	≤20 (≤45)	88 (24)	N/A
White	n=49	n=535
Lack of confidence in completing medical forms2	≤20 (≤41)	30 (6)	N/A
Requiring help reading health-related materials3	≤20 (≤41)	62 (12)	N/A
Having difficulty understanding written health information4	≤20 (≤41)	51 (10)	N/A

N/A: not applicable

^§Designates statistical significance at alpha level 0.05

^†Designates statistical significance at Bonferroni adjusted alpha level of 0.002 (25 simultaneous tests)

¹Data summarized as n (%); according to the All of Us Research Program data and statistics dissemination policy, cell values and aggregate statistics that correspond to 1 to 20 participants are obscured to protect participant privacy. In these cases, P values are designated as N/A.

²Present if participant answered the survey question “How confident are you filling out medical forms by yourself?” with somewhat, a little bit, or not at all

³Present if participant answered the survey question “How often do you have someone help you read health-related materials?” with always, often or sometimes

⁴Present if participant answered the survey question “How often do you have problems learning about your medical condition because of difficulty understanding written information?” with always, often or sometimes

⁵Other race corresponds to participants who self-reported as Another Single Population, Asian, More Than One Population, None Indicated, or None Of These

Racial differences in health literacy

Racial differences, stratified by sex, in the three domains of health literacy are shown in Table 3. In females, participants of Black or African American or Other race more frequently reported a lack of confidence in completing medical forms (16-18% versus 6%), requiring help reading health-related materials (21-28% versus 12%), and having difficulty understanding written health information (21-24% versus 10%) than participants of White race.

Sex differences in healthcare access and utilization

Only a minority of participants completed at least one item on the “Health Care Access & Utilization” survey (53 males; 634 females). Sex differences in the evaluated domains of healthcare access and utilization are shown in Table 4. Barriers to healthcare access leading to delays in care were common in both males and females, with 40% of males and 47% of females reporting at least some reason for delay in care in the past 12 months, and these rates did not differ by sex (P=0.35). A sizeable minority of females reported delays in care due to affordability (23%), time constraints (18%), transportation (17%), and nervousness to see a healthcare provider (17%).

Table 4.

Domains of healthcare access and utilization by sex, n (%)¹

	Males (n=53)	Females (n=634)	P values
Any reason for delay in care (past 12 months) 2	21 (40)	300 (47)	0.35
Delay due to affordability3	≤20 (≤38)	147 (23)	N/A
Delay due to time constraint4	≤20 (≤38)	112 (18)	N/A
Delay due to transporation5	≤20 (≤38)	108 (17)	N/A
Delay due to nervousness to see healthcare provider	≤20 (≤38)	110 (17)	N/A
Any medication challenge due to cost barrier (past 12 months)6	≤20 (≤38)	249 (39)	N/A
Ever delayed care because their healthcare provider was different from them in race, ethnicity, gender, religion, beliefs or native language7	≤20 (≤38)	106 (17)	N/A

N/A: not applicable

¹Data summarized as n (%); according to the All of Us Research Program data and statistics dissemination policy, cell values and aggregate statistics that correspond to 1 to 20 participants are obscured to protect participant privacy. In these cases, P values are designated as N/A.

²Present if participant reported any of a delay due to affordability, a delay due to time constrain, and delay due to transportation, or a delay due to nervousness to see healthcare provider in the past 12 months

³Present if participant reported they had delayed getting care in the past 12 months because they could not afford the copay, the insurance deductible was too high or could not afford the deductible, or they had to pay out of pocket for some or all of the procedure

⁴Present if participant reported they had delayed getting care in the past 12 months because they could not get time off work, could not get child care, or could not get elderly care

⁵Present if participant reported they had delayed getting care in the past 12 months because they did not have transportation or they live in a rural area where distance to the health care provider is too far

⁶Present if participant reported during the past 12 months they had skipped medication doses to save money, took less medicine to save money, delayed filling a prescription to save money, bought prescription medications from another country to save money, requested a lower cost medication from their doctor to save money or used alternative therapies to save money

⁷Present if participant self-reposed they always, most of the time or some of the time have delayed or not gone to see a health care provider because they were different from the participant in race, ethnicity, gender, religion, beliefs or native language

Racial differences in healthcare access and utilization

Racial differences, stratified by sex, in the evaluated domains of healthcare access and utilization are shown in Table 5. In females, delays in care in the past 12 months for any reason were reported more often by females of Other race (57%) than Black or African American (45%) or White (44%) race, however, delays were common across all races. Females of Other races reported more delays due to time constraints (24% versus 15-16%) and transportation (23% versus 16%) than females of Black or African American or White race. Females of Other race also more often reported ever delaying care because their healthcare provider was different from them in race, ethnicity, gender, religion, beliefs, or native language (25% versus 14-16%). In contrast, females of White race reported more medication challenges due to cost barriers (43%) compared with Black or African American (34%) or Other race (34%) females.

Table 5.

Domains of healthcare access and utilization by race, stratified by sex, n (%)¹

	Black or African American	Other Race2	White
Male Participants	n≤20	n≤20	n=29
Any reason for delay in care (past 12 months) 3	≤20 (≤100)	≤20 (≤100)	≤20 (≤69)
Any medication challenge due to cost barrier (past 12 months)4	0 (0)	≤20 (≤100)	≤20 (≤69)
Ever delayed care because their healthcare provider was different from them in race, ethnicity, gender, religion, beliefs or native language5	≤20 (≤100)	≤20 (≤100)	≤20 (≤69)
Female Participants	n=142	n=128	n=353
Any reason for delay in care (past 12 months) 3	64 (45)	73 (57)	157 (44)
Delay due to affordability6	34 (24)	28 (22)	85 (24)
Delay due to time constraint7	23 (16)	31 (24)	53 (15)
Delay due to transporation8	23 (16)	29 (23)	55 (16)
Delay due to nervousness to see healthcare provider	21 (15)	25 (20)	60 (17)
Any medication challenge due to cost barrier (past 12 months)4	48 (34)	43 (34)	151 (43)
Ever delayed care because their healthcare provider was different from them in race, ethnicity, gender, religion, beliefs or native language5	23 (16)	32 (25)	48 (14)

N/A: not applicable

¹Data summarized as n (%); according to the All of Us Research Program data and statistics dissemination policy, cell values and aggregate statistics that correspond to 1 to 20 participants are obscured to protect participant privacy. In these cases, P values are designated as N/A.

²Other race corresponds to participants who self-reported as Another Single Population, Asian, More Than One Population, None Indicated, or None Of These

³Present if participant reported any of a delay due to affordability, a delay due to time constrain, and delay due to transportation, or a delay due to nervousness to see healthcare provider in the past 12 months

⁴Present if participant reported during the past 12 months they had skipped medication doses to save money, took less medicine to save money, delayed filling a prescription to save money, bought prescription medications from another country to save money, requested a lower cost medication from their doctor to save money or used alternative therapies to save money

⁵Present if participant self-reposed they always, most of the time or some of the time have delayed or not gone to see a health care provider because they were different from the participant in race, ethnicity, gender, religion, beliefs or native language

⁶Present if participant reported they had delayed getting care in the past 12 months because they could not afford the copay, the insurance deductible was too high or could not afford the deductible, or they had to pay out of pocket for some or all of the procedure

⁷Present if participant reported they had delayed getting care in the past 12 months because they could not get time off work, could not get child care, or could not get elderly care

⁸Present if participant reported they had delayed getting care in the past 12 months because they did not have transportation or they live in a rural area where distance to the health care provider is too far

Discussion

Nine percent of participants with SLE in our cross-sectional study were male, consistent with the U.S. SLE male:female prevalence ratio.¹ Our study has one the of largest numbers of males among reports systematically examining sex differences in SLE.^5–10 We observed differences between males and females in a number of clinically relevant features, including sociodemographic characteristics, disease manifestations, and health literacy. Our findings contribute to the currently small body of literature reporting differences in socioeconomic factors and healthcare disparities between males and females in SLE.^{3, 17, 18}

Males with SLE in this U.S. national cohort had lower educational attainment than females. Lower education in men with SLE has been reported in other cohorts⁵ and may reflect national patterns of higher college enrollment rates in women, especially among Hispanic or Latina and Black or African American women.³¹ Lower educational attainment may contribute to the reported risk of more severe SLE in men.³ More years of education were protective from (1) SLE-associated organ damage in men and women³ and (2) SLE-associated death in White, but not Black or African American, men with SLE.³² In our study, most males and females were unemployed, congruent with reported low employment rates in SLE.³³ Males may be at particular risk for health-related work cessation due to higher disease activity and associated damage^{3, 5, 25, 34} and lower levels of educational attainment, both of which have been associated with work status.³³ In another large SLE cohort, disability was 70% more common in men.⁵ Despite lower educational attainment, there was no sex difference in annual household income in our study, consistent with one prior report.⁵ In contrast, Andrade et al. have reported that women with SLE are more likely to experience poverty.³

Males in our study were more likely to be current or former cigarette smokers than females. More tobacco smoking in males may contribute to increased SLE-related damage accrual in males compared to females.³⁵

Fatigue of at least moderate severity was reported in the majority of both males and females with SLE in our study, consistent with known high incidence of self-reported fatigue in SLE.³⁶ Fatigue in SLE is multidimensional and has been associated with SLE disease activity and cumulative disease damage.³⁷ Therefore, a trend towards less fatigue in males despite literature suggestive of a more aggressive SLE clinical course in males³ is surprising and merits further investigation. It is possible that males perceive and report fatigue differently than females,³⁸ as no difference in general health perception between males and females was observed.

In our study, males had more cross-sectional cardiovascular events including coronary artery disease with or without angina, and statistically significantly more myocardial infarctions, than females with SLE. These findings are in accord with previous reports,^{3, 5, 23, 28, 39} and may reflect the increased cardiovascular risk in men compared to women in general population.⁴⁰ Existing literature suggests that men are at increased risk for lupus nephritis¹⁴ and antiphospholipid syndrome,^{3, 23, 39} and our study corroborates these studies with a trend towards more lupus nephritis and anti-phospholipid syndrome in males than females in our cohort.

Prevalence estimates of impaired health literacy in SLE vary widely in the literature from 8.5% to 48%.⁴¹ Our study suggests males with SLE have less health literacy than females with SLE. Hearth-Holmes et al. showed that Black or African American race and education level were associated with lower health literacy, but sex was not significantly associated.⁴² Males in our cohort also had a trend towards lower educational attainment. The interplay of sex, education level, and health literacy in SLE merits further study. Lower health literacy, together with biological factors, may contribute to reported worse SLE prognosis in males, as low health literacy in SLE is associated with poorer patient-reported outcomes.⁴³ In contrast, health literacy did not have a significant association with hydroxychloroquine adherence in a predominately Hispanic SLE cohort.⁴⁴

In females in our study, participants of Black or African American or Other race more frequently reported items consistent with lower health literacy than participants of White race. Similar to Hearth-Holmes et al.,⁴² Maheswaranathan et al. also reported Black or African American race was associated with lower health literacy,⁴⁵ but to our knowledge none have reported similar trends for other races than Black or African American compared to White race. Given the lack of specificity within our “Other race” category, further studies are needed to better elucidate the relationship of health literacy in SLE by race in demographics other than Black or African American or White persons.

Socioeconomic barriers to healthcare access and utilization are common and numerous in both males and females in our SLE cohort. Close to half of both males and females had some delay in their health care in the past 12 months. Males with SLE have been reported to have both more perceived difficulty in accessing healthcare³ as well as objectively less clinical encounters for care including less outpatient clinic,¹⁷ and specifically rheumatology subspecialty,¹⁸ visits compared to females. The potential impact of lower outpatient healthcare utilization in males with SLE remains unclear.^{17, 32}

Reasons for delays in care in females with SLE were often multiple and could be attributed to a variety of causes, both directly related to affordability, but also only indirectly related to cost (time constraints or transportation issues). Further, nervousness to see a healthcare provider was equally commonly reported as a barrier to care in females with SLE as other categories of reasons. Medication challenges due to cost barriers were also common in females in our cohort. Prescription medication challenges have been associated with more emergency department visits in persons with SLE.¹⁸

Barriers to healthcare access and utilization may differ by race in females with SLE. Our study raises the possibility that females of race other than Black or African American or White may be disproportionately affected by delays in care. Yazdany et al. reported that racial and ethnic minorities in a predominately female SLE cohort are less likely to receive recommended healthcare for SLE.⁴⁶ In their study, all racial groups including Black or African American were compared to White race. Healthcare fragmentation has been shown to disproportionately affect Black or African American patients in another predominately female SLE cohort.⁴⁷

In our study, females of race other than Black or African American or White more commonly reported delayed care due to perceived differences between themselves and their provider with respect to race, ethnicity, gender, religion, beliefs, or native language. To our knowledge, this has not previously been reported. Vina et al. reported in a predominately female SLE cohort that African American individuals with SLE were less willing to receive cyclophosphamide if their SLE worsened compared to White patients, and this difference was mediated by less trust in the physician. However, this study did not include races other than African American or White.⁴⁸ Further study is needed to refine our understanding of barriers to healthcare access and utilization in persons with SLE of all races.

Our study has limitations. Low response rates and low sample sizes in males limited the strength of our conclusions that can be drawn for this study on racial differences in SLE in males, and this remains an area in need of further research. Participation in the All of Us Research Program is voluntary and participants may not be representative of the U.S. SLE population. Incomplete ascertainment of EHR data from consenting participants who receive care outside of the All of Us’ network health care provider organizations may bias our results. Given its cross-sectional design, our study cannot determine whether clinical manifestations (i.e. myocardial infarction) occurred following SLE diagnosis and the relative attribution of these complications to SLE itself. Data identifying provider type (i.e. rheumatologist, nephrologist, etc.) in medication prescribing patterns and to whom delays in care refer was not included in this study. We cannot provide any information on causal relationships in features explored. The number of participants identifying within our “Other” race category was large, and likely heterogeneous in racial backgrounds. Finally, in our data collection of barriers to access to care in SLE, barriers were non-weighted, so priority of those variables could not be established.

Our study demonstrated major clinical and health literacy differences between males and females with SLE. Socioeconomic factors were significant barriers to healthcare access and utilization in both sexes. However, males reported poorer health literacy in our study compared to females with SLE, which may exacerbate pre-existing socioeconomic disparities and barriers and lead to worse outcomes. Further large, prospective studies of SLE with an aim at recruiting men are needed to better understand racial differences in men with SLE in all domains affecting healthcare.

Significance and Innovations.

• Males with SLE in this U.S. national cohort had a trend towards lower educational attainment and significantly less confidence in completing medical forms than females, suggesting disproportionately poorer health literacy in males with SLE

• Socioeconomic barriers to healthcare access and utilization are common and numerous in both males and females with SLE in this U.S. national cohort

• Barriers to healthcare access and utilization may differ by race in females with SLE, with females of race other than Black or African American or White more frequently reporting delayed care due to cultural differences between patient and provider