Racial Differences in Contraception Encounters and Dispensing Among Female Medicaid Beneficiaries with Systemic Lupus Erythematosus

Abstract

Objective

Black and Hispanic women with SLE have the highest rates of potentially avoidable pregnancy complications, yet racial disparities in family planning among reproductive-age women with SLE have not been well-studied. We examined whether there are racial differences in contraception encounters and dispensing among U.S. Medicaid-insured women with SLE.

Methods

Using Medicaid claims data from 2000–2010, we identified women aged 18–50 with SLE. We examined contraception encounters and uptake over 24 months. We used multivariable logistic regression to estimate the odds (OR, 95% CI) by race/ethnicity of contraception encounters, any contraception dispensing, and highly effective contraception (HEC) use, adjusted for age, region, year, SLE severity, and contraindication to estrogen. We also compared contraception encounters and dispensing among women with SLE to the general population and women with diabetes.

Results

We identified 24,693 reproductive-age females with SLE; 43% were Black, 35% White, 15% Hispanic, 4% Asian, 2% other race, and 1% American Indian/Alaska Native. Nine percent had a contraception visit, 10% received any contraception, and 2% received HEC. Compared to White women, Black and Asian women had lower odds of contraception dispensing, and Black women had lower odds of HEC use. Women with SLE were more likely to receive HEC than the general population and women with diabetes.

Conclusion

In this study of reproductive-age women with SLE, Black and Asian women had lower odds of contraception dispensing and Black women had lower odds of HEC use. Further study is needed to understand the factors driving these racial disparities among this population.

Introduction

Systemic lupus erythematosus (SLE) is an autoimmune condition that is most common in women of reproductive age (15–50 years).¹ Many medications used to treat SLE and its complications are teratogenic and require concurrent use of effective contraception for women at risk for pregnancy. Active SLE, including lupus nephritis (LN), is associated with adverse pregnancy outcomes that are potentially avoidable if patients are able to time pregnancy for when their SLE is well-controlled.^2–4 Furthermore, there are known racial/ethnic disparities in pregnancy outcomes among women with SLE. As compared to White women with SLE, Black and Hispanic women with SLE have higher rates of pregnancy complications, including preterm labor, preeclampsia, and fetal growth restriction, even after controlling for baseline medical conditions and insurance status.⁵ These complications may result in acute care use and poorer outcomes. Thus, comprehensive contraceptive counseling and care in the outpatient setting for women with SLE, in particular Black and Hispanic women, is important to minimize pregnancy complications.

Highly effective contraception (HEC) includes long-acting reversible contraception (LARC) methods (intrauterine devices (IUDs) and implants), as well as sterilization. LARC methods are safe for women with SLE^6–8 and have a <1% failure rate for preventing pregnancy.⁹ LARC methods also do not contain estrogen, which is contraindicated in women with SLE who have positive antiphospholipid antibodies or high disease activity.¹⁰ Thus, LARC is an ideal form of contraception for many women with SLE. Additionally, sterilization is another HEC method for women who desire permanent contraception. Prior studies have revealed variable rates of contraceptive use among reproductive-age women with SLE^11–21, and rates of HEC use among this population have been shown to be low, in the range of 6–9%, though slightly higher than the 5% HEC use reported in the general population of reproductive-age females.^19,22

Despite evidence that contraceptive care for reproductive-age women with SLE is suboptimal and Black and Hispanic women with SLE have worse pregnancy outcomes, racial disparities in contraception care among reproductive-age women with SLE have not been well-studied. However, prior qualitative studies have found that mistrust is more likely to complicate the relationships of Black and Hispanic female patients with their contraception providers, which may ultimately lead to disparities in contraception care. For example, a thematic analysis of coded interview data from 27 Black and Hispanic reproductive-age women found that the majority felt implicit pressure to accept their providers’ preferred method of birth control, despite concerns that risks were not appropriately emphasized. Consequences of this included discontinuation of recommended contraceptive methods and disengagement from the health care system.²³ A separate study of 135 women in New York found that in survey data, Black and Hispanic women were much more likely to report birth-control related mistrust of the medical system and government than white women (p<0.001).²⁴ Using United States (U.S.) Medicaid data, we examined racial/ethnic differences in outpatient contraceptive encounters, contraception dispensing, and specifically HEC use, among reproductive-age women with SLE, and among subpopulations of women with SLE using teratogens and those with LN. We also compared contraceptive encounters and dispensing in the SLE population with the general and diabetes mellitus (DM) populations. We chose to examine DM is this is another chronic medical condition that may adversely impact pregnancy outcomes, and patients with DM may struggle with similar factors impacting contraceptive use among women with SLE such as pill burden and competing priorities. We hypothesized that contraceptive encounters, contraceptive dispensing, and HEC use would be more likely in patients who are at the highest risk for potentially avoidable adverse pregnancy outcomes, including Black patients, Hispanic patients, patients with LN, and teratogen users.

Patients and Methods

Patient Population

We used Medicaid claims data (Medicaid Analytic eXtract, MAX) from 47 U.S. states between 2000–2006 and the 29 most populated U.S. states between 2007–2010 to identify reproductive-age women (18–50 years) with prevalent SLE (≥3 International Classification of Diseases, Ninth Revision (ICD-9) codes 710.0 separated by ≥30 days, index date = date of the 3^rd SLE code).²⁵ Only 29 states were analyzed during the period of 2007–2010 because our data use agreement was limited to these areas, which included the majority of the U.S. and SLE population during this time period. MAX includes demographics information, healthcare encounters, and drug prescription and dispensing information. Additionally, we examined the subpopulation who had ≥1 teratogenic medication prescription during the baseline period or on the index date (methotrexate, mycophenolate mofetil, cyclophosphamide, leflunomide, angiotensin-converting enzyme inhibitor, or warfarin), as well as the subpopulation with LN (defined as ≥2 ICD-9 codes for nephritis, proteinuria, and/or renal failure on or after the SLE diagnosis date and ≥30 days apart, index date = date of the 2^nd LN-related code).²⁶ Lastly, we age-matched female Medicaid patients in the general population (4:1) and DM population (2:1) to SLE patients enrolled between 2007 and 2010. General population patients did not have any codes for SLE and had hospital discharge diagnoses or physician visit claims on the same index date as each prevalent SLE patient. DM patients did not have any codes for SLE and had ≥3 ICD-9 codes for DM from hospital discharge diagnoses or physician visit claims separated by ≥30 days on the same index date as each prevalent SLE patient (see Appendix for DM codes).

For the non-matched analyses, we required ≥6 months of continuous enrollment prior to the index date (baseline period) and ≥24 months of continuous enrollment after (follow-up period). For the matched analyses, we required ≥6 months of continuous enrollment prior to the index date (baseline period) and ≥12 months of continuous enrollment after (follow-up period); the matched analyses had a shorter follow-up period as data were available for the general and DM populations from 2007–2010 only. Patients were excluded if they were ineligible for new contraceptive use (baseline period codes for HEC, hormone replacement therapy, hysterectomy, premature ovarian failure, or menopause (see Appendix)). We also excluded patients with baseline or follow-up period codes for pregnancy (see Appendix). Patients were censored at death, end of study database, or end of Medicaid enrollment (whichever came first). This study was approved by the Institutional Review Board at Brigham and Women’s Hospital, which provided a waiver of written informed consent. This study was conducted in compliance with the Helsinki Declaration.

Exposure and Outcomes

The exposure of interest was race/ethnicity, which was determined by patient self-report. Possible categories included White Non-Hispanic, Black Non-Hispanic or African-American, Hispanic, Asian or Pacific Islander, American Indian or Alaska Native, or other race. Outcomes of interest included outpatient encounters for contraception management, receipt of any form of contraception (including combined hormonal pills, progestin-only pills, IUDs, implants, estrogen-based intravaginal rings, estrogen-based transdermal patches, medroxyprogesterone injections, or sterilization), receipt of HEC (IUDs, implants, or sterilization), and receipt of LARC (IUDs or implants). These four outcomes were assessed using encounter, pharmacy, and procedure codes (see Appendix) during the follow-up period, as well as drug dispensing data. We were not able to assess the use of barrier methods in this dataset.

Baseline Covariates

In the multivariable logistic regression models, we adjusted for age group at the index date (18–24, 25–31, 32–38, 39–45, 46–50), calendar year of the index date, geographic region (Northeast, Midwest, South, West), the Ward SLE risk adjustment index (proxy for SLE severity), and a composite variable for contraindication to estrogen (baseline period codes for ischemic heart disease, hypertension, stroke, venous thromboembolism, smoking, migraine with aura, or breast cancer; see Appendix). We also examined zip code median household income, but this covariate was removed from the model as it did not contribute significantly. We did not have data on sexual activity, practice setting (public vs. private), provider specialty, or census tract, and thus these were not included as covariates.

Statistical Analyses

For the non-matched analyses, we employed multivariable logistic regression models adjusted for the aforementioned covariates to estimate the odds ratio (OR) and 95% confidence interval (CI) by race/ethnicity of: A) any outpatient encounter for contraceptive management (vs. none), B) any contraception (vs. none), C) HEC (vs. no HEC), D) LARC (vs. no LARC), E) HEC (vs. no HEC) in teratogen user subgroup analyses, and F) HEC (vs. no HEC) in LN subgroup analyses. For HEC use, we tested for interaction between age and calendar year of index date, as well as age and race, and the terms were not statistically significant. For the matched analyses, we employed conditional logistic regression models adjusted for the aforementioned covariates (except for calendar year of index date given shortened study period of 2007–2010), to estimate the OR (95% CI) of A) any outpatient encounter for contraceptive management (vs. none), B) any contraception (vs. none), C) HEC (vs. no HEC), and D) LARC (vs. no LARC). In these models, we also examined the OR (95% CI) by race/ethnicity of A) any outpatient encounter for contraceptive management (vs. none), B) any contraception (vs. none), C) HEC (vs. no HEC), and D) LARC (vs. no LARC).

Additionally, we estimated multivariable logistic regression models adjusted for age group, race/ethnicity, and state of residence (reference = Massachusetts), to examine state-level trends in A) any outpatient encounter for contraceptive management (vs. none), B) any contraception (vs. none), and C) HEC (vs. no HEC), among reproductive-age women with SLE in the 29 most populated U.S. states between 2007–2010.

We set α=0.05 to determine statistical significance, and all p-values were two-sided. Data were analyzed using SAS 9.4 (Cary, NC). Medicaid data were obtained through a Data Use Agreement with the Centers for Medicare and Medicaid Services and are presented in accordance with their policies (cell sizes <11 are suppressed).

Sensitivity Analyses

We conducted sensitivity analyses including women with pregnancy codes, as well as examining incident SLE (≥3 ICD-9 codes 710.0 separated by ≥30 days, with 24 months without prior SLE codes).²³

Results

We identified 24,693 female Medicaid beneficiaries with SLE between 2000–2010 (Table 1). The mean age was 38 (SD 9); 43% were Black, 35% White, 15% Hispanic, 4% Asian, 2% other race, and 1% American Indian/Alaska Native. Thirty-nine percent of patients resided in the South, 21% in the Northeast, 21% in the West, and 20% in the Midwest. Nine percent had an encounter for contraceptive management, 10% received any form of contraception, and 2% received HEC. Among the subpopulation of 5,754 women with SLE using a teratogenic medication and the subpopulation of 5,229 women with prevalent LN, there were no differences in frequency of contraception visits, contraception dispensing, or HEC use as compared to the overall SLE population.

Table 1.

Baseline Characteristics of Female Medicaid Beneficiaries with Systemic Lupus Erythematosus (SLE) Aged 18–50 Years, 2000–2010.

Characteristics	SLE (N=24,693)	Teratogen Use (N=5,754)	Lupus Nephritis (N=5,229)
Age Category
18–24	2,685 (11)	780 (14)	1,042 (20)
25–31	4,153 (17)	921 (16)	1,199 (23)
32–38	5,918 (24)	1,243 (22)	1,201 (23)
39–45	7,122 (29)	1,643 (29)	1,104 (21)
46–50	4,815 (20)	1,167 (20)	683 (13)
Race/Ethnicity
White	8,585 (35)	1,630 (28)	1,085 (21)
Black or African-American	10,687 (43)	2,760 (48)	2,805 (54)
Hispanic	3,648 (15)	894 (16)	884 (17)
Asian	945 (4)	260 (5)	308 (6)
American Indian/Alaska Native	272 (1)	56 (1)	49 (1)
Other	556 (2)	154 (3)	98 (2)
Geographic Region
Northeast	5,102 (21)	1,190 (21)	980 (19)
Midwest	5,004 (20)	1,227 (21)	1,074 (21)
South	9,512 (39)	2,168 (38)	2,115 (41)
West	5,075 (21)	1,169 (20)	1,060 (20)
Contraindication to Estrogen	9,553 (39)	3,200 (56)	2,490 (48)
Ward Index ≥1	8,870 (36)	2,918 (51)	3,581 (69)
Calendar Year of Index Date
2000–2004	13,125 (53)	3,355 (58)	2,514 (48)
2005–2006	3,742 (15)	900 (16)	793 (15)
2007–2008	7,826 (32)	1,499 (26)	1,922 (37)

All values are N (%).

SLE Cohort

The results of the multivariable logistic regression analyses are displayed in Table 2. Compared to White women with SLE, Black women with SLE had 1.40 (95% CI 1.25–1.55) times higher odds of a contraceptive visit but 0.89 (95% CI 0.81–0.98) times lower odds of any contraception dispensing and 0.71 (95% CI 0.56–0.89) times lower odds of HEC use. Compared to White women with SLE, Hispanic women with SLE had 1.38 (95% CI 1.20–1.59) times higher odds of a contraceptive visit. Compared to White women with SLE, Asian women with SLE had significantly lower odds of both a contraceptive visit (OR 0.66, 95% CI 0.50–0.88) and any contraception dispensing (OR 0.72, 95% CI 0.57–0.92).

Table 2.

Multivariable Logistic Regression Analyses Examining Factors Associated with Contraception Encounters and Dispensing Among Female Medicaid Beneficiaries with Systemic Lupus Erythematosus (SLE) Aged 18–50 Years, 2000–2010 (N=24,693).

Characteristic	Encounter for Contraception Management OR (95% CI)	Any Contraception Dispensing* OR (95% CI)	HEC Use OR (95% CI)	LARC Use OR (95% CI)
Age Category
46–50	1.00 (Ref)	1.00 (Ref)	1.00 (Ref)	1.00 (Ref)
18–24	18.03 (14.25–22.82)	21.72 (17.06–27.66)	12.36 (7.12–21.44)	11.91 (6.25–22.69)
25–31	10.61 (8.41–13.40)	13.74 (10.83–17.44)	11.65 (6.81–19.91)	8.15 (4.31–15.43)
32–38	5.73 (4.53–7.24)	6.78 (5.34–8.62)	7.14 (4.17–12.24)	5.46 (2.88–10.35)
39–45	2.53 (1.98–3.24)	3.29 (2.57–4.21)	3.62 (2.08–6.31)	2.41 (1.23–4.74)
Race
White	1.00 (Ref)	1.00 (Ref)	1.00 (Ref)	1.00 (Ref)
Black	1.40 (1.25–1.55)	0.89 (0.81–0.98)	0.71 (0.56–0.89)	0.98 (0.72–1.32)
Hispanic	1.38 (1.20–1.59)	0.89 (0.77–1.01)	1.12 (0.85–1.47)	1.36 (0.94–1.98)
Asian	0.66 (0.50–0.88)	0.72 (0.57–0.92)	0.69 (0.40–1.22)	1.34 (0.73–2.47)
AI/AN	1.45 (0.95–2.23)	0.94 (0.61–1.46)	1.63 (0.79–3.38)	2.16 (0.86–5.44)
Other	1.24 (0.89–1.75)	1.00 (0.73–1.36)	1.27 (0.68–2.37)	1.57 (0.67–3.67)
Region
Northeast	1.00 (Ref)	1.00 (Ref)	1.00 (Ref)	1.00 (Ref)
South	1.14 (1.01–1.30)	1.00 (0.88–1.12)	0.89 (0.69–1.16)	0.87 (0.61–1.25)
West	1.31 (1.13–1.51)	1.06 (0.93–1.22)	0.98 (0.74–1.31)	1.18 (0.81–1.74)
Midwest	1.14 (0.99–1.32)	1.10 (0.96–1.26)	0.97 (0.72–1.30)	1.25 (0.85–1.84)
Ward SLE Risk Adjustment Index	0.88 (0.80–0.98)	0.71 (0.64–0.78)	0.94 (0.76–1.16)	1.07 (0.81–1.41)
Contraindication to Estrogen	0.88 (0.80–0.98)	0.86 (0.78–0.95)	0.99 (0.80–1.23)	0.95 (0.72–1.27)
Calendar Year
2000–2004	1.00 (Ref)	1.00 (Ref)	1.00 (Ref)	1.00 (Ref)
2005–2006	1.12 (0.99–1.27)	0.90 (0.80–1.01)	1.37 (1.04–1.80)	1.50 (1.01–2.22)
2007–2008	0.90 (0.81–1.00)	0.46 (0.41–0.51)	1.91 (1.55–2.35)	3.12 (2.36–4.12)

Regression model adjusted for age group, geographic region, calendar year of the index date, the SLE risk adjustment index, and composite variable for contraindication to estrogen

^*Any contraception dispensing includes sterilization, intrauterine devices, implants, oral contraceptives, estrogen patch, estrogen ring, medroxyprogesterone injections. Data on use of barrier methods was not available.

HEC=highly effective contraception, LARC=long-acting reversible contraception, AI/AN=American Indian/Alaska Native

Younger age (<46 years) was strongly associated with contraceptive visits, contraceptive dispensing, HEC use, and LARC use. Living in the South (OR 1.14, 95% CI 1.01–1.30) and living in the West (OR 1.31, 95% CI 1.13–1.51) was associated with increased odds of a contraceptive visit. More severe SLE was associated with lower odds of a contraceptive visit (OR 0.88, 95% CI 0.80–0.98) and lower odds of any contraception dispensing (OR 0.71, 95% CI 0.64–0.78). Having a contraindication to estrogen was also associated with lower odds of a contraceptive visit (OR 0.88, 95% CI 0.80–0.98) and lower odds of any contraception dispensing (OR 0.86, 95% CI 0.78–0.95). Having an index date during the later years of 2005–2008 was associated with increased odds of HEC use (OR 1.37, 95% CI 1.04–1.80 for 2005–2006; OR 1.91, 95% CI 1.55–2.35 for 2007–2008) and LARC use (OR 1.50, 95% CI 1.01–2.22 for 2005–2006; OR 3.12, 95% CI 2.36–4.12 for 2007–2008).

SLE Subpopulation Using Teratogens

Among the 5,754 women with SLE using teratogens, there were no significant racial differences in HEC use (Table 3). Younger age was strongly associated with HEC use among this subpopulation, as was index date year of 2007–2008 (OR 2.39, 95% CI 1.57–3.64).

Table 3.

Multivariable Logistic Regression Analyses Examining Factors Associated with Highly Effective Contraception (HEC) Use Among Female Medicaid Beneficiaries with Systemic Lupus Erythematosus (SLE) and Teratogen Use (N=5,754) and Lupus Nephritis (N=5,229) Aged 18–50 Years, 2000–2010.

Characteristic	HEC Use on Teratogen* (N=5,754) OR (95% CI)	HEC Use with Lupus Nephritis (N=5,229) OR (95% CI)
Age Category
46–50	1.00 (Ref)	1.00 (Ref)
18–24	5.36 (2.29–12.53)	6.18 (2.89–13.20)
25–31	5.57 (2.42–12.82)	3.83 (1.75–8.41)
32–38	3.44 (1.48–8.03)	3.01 (1.33–6.82)
39–45	2.37 (1.01–5.59)	1.00 (Ref)**
Race
White	1.00 (Ref)	1.00 (Ref)
Black	0.84 (0.52–1.36)	0.82 (0.45–1.48)
Hispanic	1.20 (0.67–2.17)	0.98 (0.48–2.02)
Asian	0.75 (0.25–2.21)	0.84 (0.30–2.40)
AI/AN	1.77 (0.40–7.74)	7.95 (2.48–25.50)
Other	2.08 (0.78–5.55)	<0.001 (<0.001–>999.99)
Region
Northeast	1.00 (Ref)	1.00 (Ref)
South	1.37 (0.79–2.39)	0.98 (0.50–1.95)
West	1.22 (0.65–2.27)	1.33 (0.64–2.77)
Midwest	1.20 (0.63–2.26)	1.29 (0.63–2.67)
Ward SLE Risk Adjustment Index	1.04 (0.69–1.56)	2.42 (1.26–4.66)
Contraindication to Estrogen	0.76 (0.51–1.14)	0.93 (0.57–1.51)
Calendar Year
2000–2004	1.00 (Ref)	1.00 (Ref)
2005–2006	1.56 (0.90–2.69)	1.15 (0.57–2.30)
2007–2008	2.39 (1.57–3.64)	1.67 (1.03–2.72)

Regression model adjusted for age group, geographic region, calendar year of the index date, the SLE risk adjustment index, and composite variable for contraindication to estrogen

^*Teratogens included methotrexate, mycophenolate mofetil, mycophenolic acid, cyclophosphamide, leflunomide, angiotensin-converting enzyme inhibitor, or warfarin

^**Age 39–50 was used as the reference group as there were no patients with HEC use in the 46–50 age group

AI/AN=American Indian/Alaska Native

SLE Subpopulation with LN

Among the subpopulation of patients with LN (n=5,229), American Indian/Alaska Native race was associated with increased odds of HEC use (OR 7.95, 95% CI 2.48–25.50; Table 3). There were no other significant racial differences in HEC use among this subpopulation. Younger age, SLE severity (OR 2.42, 95% CI 1.26–4.66), and index date year of 2007–2008 (OR 1.67, 95% CI 1.03–2.72) were strongly associated with HEC use among this subpopulation.

We also examined contraception use by age group for the overall SLE population and for the LN subpopulation. In the 18–24 years age group, 29% of SLE patients and 19% of LN patients received any contraception. In the 25–31 years age group, 21% of SLE patients and 10% of LN patients received any contraception. In the 32–38 years age group, 11% of SLE patients received any contraception vs. 7% of LN patients, and in the oldest 2 age categories (39–50 years, combined given smaller size of LN subpopulation), 8% of SLE patients received any contraception vs. 3% of LN patients.

State-Level Analyses

We identified 7,826 reproductive-age women with SLE residing in the 29 most populated U.S. states between 2007–2010 for the state-level analyses. Alabama residence (OR 0.14; 95% CI 0.03–0.66) and Ohio residence (OR 0.29, 95% CI 0.10–0.85) were associated with lower odds of any contraception dispensing. Texas residence (OR 2.08, 95% CI 1.03–4.18) and Wisconsin residence (OR 2.33, 95% CI 1.09–4.99) were associated with higher odds of a contraceptive management visit. There were no significant state-level differences in HEC use. These results did not change when race was excluded from the model.

Age-Matched SLE, Diabetes and General Populations

For the age-matched results, we identified 6,946 women with SLE, 13,718 women with DM, and 29,698 women in the general population (Table 4). There were more Black patients in the SLE population (46%) as compared to the DM (31%) and general populations (22%). There were more Hispanic patients in the general population (27%) as compared to the SLE (13%) and DM populations (15%). Table 5 displays results of the conditional logistic regression analyses comparing contraception encounters and dispensing among the age-matched SLE, DM, and general populations. As compared to the DM population, the SLE population had significantly lower odds of contraceptive visits (OR 0.82, 95% CI 0.73–0.92) and any contraception dispensing (OR 0.42, 95% CI 0.37–0.47) but significantly higher odds of HEC use (OR 1.39, 95% CI 1.12–1.72) and LARC use (OR 1.36, 95% CI 1.05–1.75). As compared to the general population, the SLE population had significantly lower odds of contraceptive visits (OR 0.64, 95% CI 0.55–0.75) and any contraception dispensing (OR 0.45, 95% CI 0.38–0.52 but significantly higher odds of HEC use (OR 1.39, 95% CI 1.03–1.89) and LARC use (OR 1.52, 95% CI 1.06–2.19). Across all 3 populations, Black women had higher odds of a contraceptive visit but lower odds of HEC use (Table 6). Among the general population, American Indian/Alaska Native race and other race were associated with higher odds of a contraceptive visit, and Hispanic race was associated with lower odds of a contraceptive visit but higher odds of any contraception dispensing, HEC use, and LARC use.

Table 4.

Baseline Characteristics of Age-Matched Female Medicaid Beneficiaries in the Systemic Lupus Erythematosus (SLE) population, Diabetes Mellitus (DM) Population, and General Population (GP), 2007–2010.

Characteristic	SLE	DM	GP
Total N Count	6,946	13,718	29,698
Mean Age (SD)	38 (9)	38 (9)	37 (9)
Race/Ethnicity
White	2,472 (36)	6,121 (45)	12,466 (42)
Black or African-American	3,187 (46)	4,254 (31)	6,385 (22)
Hispanic	917 (13)	2,079 (15)	8,147 (27)
Asian	213 (3)	321 (2)	790 (3)
American Indian/Alaska Native	67 (1)	192 (1)	312 (1)
Other	90 (1)	751 (6)	1,598 (5)
Geographic Region
Northeast	1,352 (20)	2,693 (20)	5,605 (19)
Midwest	1,555 (22)	3,112 (23)	5,939 (20)
South	2,856 (41)	5,380 (39)	8,135 (27)
West	1,183 (17)	2,533 (19)	10,019 (34)

Unless indicated otherwise, all values are N (%).

SD=standard deviation

Table 5.

Conditional Logistic Regression Results Examining Contraception Encounters and Dispensing Among Female Medicaid Beneficiaries in the Systemic Lupus Erythematosus (SLE) Population, as Compared to the Age-Matched Diabetes Mellitus (DM) Population and General Population (GP), 2007–2010.

Comparison Group	Encounter for Contraception Management OR (95% CI)	Any Contraception Dispensing* OR (95% CI)	HEC Use OR (95% CI)	LARC Use OR (95% CI)
SLE vs DM	0.82 (0.73–0.92)	0.42 (0.37–0.47)	1.39 (1.12–1.72)	1.36 (1.05–1.75)
SLE vs GP	0.64 (0.55–0.75)	0.45 (0.38–0.52)	1.39 (1.03–1.89)	1.52 (1.06–2.19)

^*Any contraception dispensing includes sterilization, intrauterine devices, implants, oral contraceptives, estrogen patch, estrogen ring, medroxyprogesterone injections. Data on use of barrier methods was not available.

Table 6.

Conditional Logistic Regression Results Examining Contraception Encounters and Dispensing by Race/Ethnicity Among Age-Matched Female Medicaid Beneficiaries in the Systemic Lupus Erythematosus (SLE) Population, Diabetes Mellitus (DM) Population, and General Population (GP), 2007–2010.

Group/Race and Ethnicity	Encounter for Contraception Management OR (95% CI)	Any Contraception Dispensing* OR (95% CI)	HEC Use OR (95% CI)	LARC Use OR (95% CI)
SLE
White	1.00 (Ref)	1.00 (Ref)	1.00 (Ref)	1.00 (Ref)
Black	1.27 (1.03–1.57)	0.89 (0.71–1.12)	0.68 (0.48–0.98)	1.00 (0.65–1.53)
Hispanic	1.29 (0.95–1.73)	0.93 (0.67–1.28)	1.07 (0.67–1.71)	1.18 (0.66–2.11)
Asian	0.98 (0.55–1.73)	0.72 (0.38–1.34)	1.29 (0.60–2.77)	1.88 (0.81–4.36)
AI/AN	1.15 (0.44–3.02)	0.93 (0.32–2.68)	0.57 (0.08–4.20)	0.96 (0.13–7.21)
Other	1.55 (0.76–3.17)	1.44 (0.69–3.03)	2.00 (0.77–5.18)	2.68 (0.92–7.82)
DM
White	1.00 (Ref)	1.00 (Ref)	1.00 (Ref)	1.00 (Ref)
Black	1.20 (1.04–1.38)	0.91 (0.80–1.02)	0.66 (0.48–0.92)	0.88 (0.61–1.27)
Hispanic	1.13 (0.94–1.37)	0.92 (0.78–1.08)	1.04 (0.72–1.51)	0.95 (0.60–1.49)
Asian	1.06 (0.68–1.66)	1.24 (0.88–1.75)	1.53 (0.75–3.09)	1.93 (0.90–4.12)
AI/AN	0.95 (0.54–1.66)	1.05 (0.68–1.63)	0.81 (0.25–2.59)	0.83 (0.20–3.44)
Other	0.99 (0.74–1.32)	0.89 (0.69–1.13)	0.67 (0.35–1.29)	0.97 (0.49–1.90)
GP
White	1.00 (Ref)	1.00 (Ref)	1.00 (Ref)	1.00 (Ref)
Black	1.16 (1.05–1.27)	1.06 (0.97–1.15)	0.77 (0.63–0.95)	0.87 (0.69–1.10)
Hispanic	0.82 (0.72–0.92)	1.77 (1.63–1.92)	1.55 (1.31–1.84)	1.65 (1.37–2.00)
Asian	0.76 (0.56–1.02)	1.10 (0.91–1.32)	0.89 (0.59–1.35)	1.01 (0.66–1.55)
AI/AN	1.80 (1.28–2.54)	0.96 (0.71–1.30)	0.97 (0.51–1.85)	1.09 (0.55–2.16)
Other	1.29 (1.09–1.52)	0.91 (0.78–1.06)	0.90 (0.64–1.27)	0.94 (0.64–1.37)

^*Any contraception dispensing includes sterilization, intrauterine devices, implants, oral contraceptives, estrogen patch, estrogen ring, medroxyprogesterone injections. Data on use of barrier methods was not available.

HEC=highly effective contraception, LARC=long-acting reversible contraception, AI/AN=American Indian/Alaska Native

Sensitivity Analyses

For our sensitivity analyses including women with pregnancy codes, we identified 29,206 reproductive-age women with SLE between 2000–2010. Fifteen percent had an encounter for contraceptive management, 15% received any form of contraception, and 5% received HEC. Similar to our primary results, we found that as compared to White women, despite 1.24 (95% CI 1.15–1.35) times higher odds of a contraceptive visit, Black women had 0.87 (95% CI 0.80–0.94) times lower odds of any contraception dispensing and 0.71 (95% CI 0.62–0.81) times lower odds of HEC use. Asian women had significantly lower odds of contraceptive visits (OR 0.62, 95% CI 0.49–0.77), any contraception dispensing (OR 0.59, 95% CI 0.48–0.73), and HEC use (OR 0.47, 95% CI 0.32–0.70).

For our sensitivity analyses examining women with incident SLE, we identified 9,521 reproductive-age women with incident SLE between 2000–2010. 10% had an encounter for contraceptive management, 11% received any form of contraception, and 2% received HEC. We found that Black (OR 1.46; 95% CI 1.24–1.73) and Hispanic patients (OR 1.30; 95% CI 1.04–1.62) with incident SLE were more likely to have a visit for contraception management, but there were no significant racial differences in receipt of contraception dispensing or HEC use.

Discussion

In this nationwide study of reproductive-age women with SLE, we found that contraception dispensing and HEC uptake was very low. This remained true among women at highest risk for potentially avoidable adverse pregnancy outcomes: teratogen users or women with LN. Despite the LN subpopulation being younger, we did not find that they were more likely than the overall SLE population to receive contraception, even when stratified by age group. We also identified racial disparities in contraception encounters and dispensing. Despite more contraceptive visits compared to White women, Black women had lower odds of receiving contraception and HEC, while Asian women had fewer contraceptive visits and lower odds of receiving contraception. As compared to the DM and general populations, reproductive-age women with SLE were less likely to receive any form of contraception but more likely to receive HEC or LARC. Across the SLE, DM, and general populations, Black women were more likely to have a contraceptive visit but less likely to receive HEC. We also found notable state-based differences in contraception encounters and dispensing in exploratory analyses. Compared to Massachusetts residence, Alabama and Ohio residence were associated with decreased odds of any contraception dispensing, and Texas and Wisconsin residence were associated with increased odds of a contraception visit.

A table displaying the studies to date examining contraception care among women with SLE is included in the Appendix.^{8,11–22, 27–32} These prior studies have revealed variable rates of contraceptive use among reproductive-age women with SLE (22–85%)^11–21, and rates of HEC use among this population are even lower, in the range of 6–9%.^19,22 Additionally, studies examining contraceptive counseling in the outpatient setting among women with SLE have shown that 28–59% of patients at risk for pregnancy are not receiving counseling^{13,18–20,27}, including up to 46% of women using teratogens.²⁸ Unplanned pregnancies and lack of contraception use have been found to be significantly more common among women with SLE as compared to women with rheumatoid arthritis (RA) or healthy controls.¹⁷ Moreover, combined hormonal contraceptive use has been found to be common in women with SLE who have contraindications to its use, including the presence of antiphospholipid antibodies.²⁹ Despite known racial disparities in pregnancy outcomes among women with SLE, only one of these studies examined racial differences in contraception care among this population,²⁷ and the majority did not provide any racial data. Although racial differences in contraception care among women with SLE have not been well-studied, prior qualitative studies have found that Black and Hispanic women perceive that their providers are pressuring them about their contraception choices²³ and report higher levels of birth control-related mistrust as compared to White women,²⁴ which may adversely affect their reproductive choices and lead to racial disparities.

In this study, we found that as compared to patients with DM or the general population, patients with SLE were less likely to receive contraception in general but more likely to receive HEC or LARC. This could reflect enhanced concerns about estrogen exacerbating disease activity or leading to thromboembolic disease. These findings also suggest that the presence of a chronic disease is not the only factor leading to decreased contraception dispensing among women with SLE. As far as our state of residence findings, state-level variation in contraceptive policies and access for Medicaid-insured women could contribute to these differences. For example, a prior study of state-level differences in postpartum contraception use found that women in Alabama and Ohio had very low rates of LARC use (2%).³³ We also found that during the study period, Texas had a state-funded program to cover family planning services for adult women living at ≤185% of the federal poverty level (FPL) and Wisconsin has a state plan amendment to provide family planning services for those living at ≤306% of the FPL, which is the highest threshold among all states.³⁴

Strengths of our study include use of a large, nationwide, and racially diverse study population with detailed medication data. We had detailed information on a range of contraceptive services including encounters for contraception management, and pharmacy and medical claims for combined oral contraceptive pills, progestin-only pills, IUDs, implants, estrogen-based intravaginal rings, estrogen-based transdermal patches, medroxyprogesterone injections, and sterilization.

The limitations of this study include lack of data on use of non-prescription methods of contraception or contraceptives that were not reimbursed by Medicaid. We were also not able to ascertain practice setting (public vs. private), provider specialty, patient census tract, male partner sterilization, whether patients desired pregnancy, whether patients declined contraception, and whether patients were sexually active with men during the study period. We were limited in the number of covariates we could include in our models given the small number of outcomes, and thus there was a possibility of residual confounding. Lastly, due to variability in enrollment in Medicaid, we were unable to identify LARC insertions prior to the baseline period for our population. However only 0.6% of women in our cohort had LARC removals during the follow-up period without prior claims for insertion during the baseline and follow-up periods. Given this small percentage, we feel it is unlikely that we are missing a significant number of LARC users who had placement prior to the baseline period.

In conclusion, racial disparities in contraceptive encounters and dispensing among Black and Asian reproductive-age women with SLE in the U.S. Medicaid population is likely multifactorial. Contributing factors may include patient preference, cultural factors, ineffective communication and/or mistrust between healthcare providers and patients, racial bias of providers (implicit or explicit), state-level policies affecting reproductive care, and the legacy of historical injustices (for example, the history of forced sterilization of Black women in the U.S. may account for lower odds of HEC use but not LARC use among Black women in our study). In particular, ineffective communication and mistrust in the patient-provider relationship may account for our observation of increased contraception management visits but decreased receipt of contraception or HEC among Black women with SLE. With known disparities in pregnancy outcomes by race/ethnicity among women with SLE, qualitative studies are needed to further understand the factors that are driving these racial differences. We feel that every effort should be made to mitigate these racial disparities and to ensure that patients who are at the highest risk for adverse pregnancy outcomes are receiving appropriate reproductive care.

Significance and Innovation.

• Within a large population of reproductive-age women with SLE, rates of contraception visits, contraception dispensing, and highly effective contraception (HEC) use were low, even among women with lupus nephritis (LN) or using teratogens, who are at higher risk of potentially avoidable adverse pregnancy outcomes.

• Despite more contraception visits compared to White women, Black women had lower odds of receiving contraception and HEC. Asian women had fewer contraception visits and lower odds of receiving contraception.

• Compared to the general population, reproductive-age women with SLE were less likely to receive any form of contraception. Those SLE patients who received contraception were more likely to receive HEC or long-acting reversible contraception (LARC), potentially due to avoidance of estrogen-based methods in women with high lupus disease activity or positive antiphospholipid antibodies. Across the SLE and general populations, Black women were more likely to have a contraception visit but less likely to receive HEC.

• These racial disparities in contraception care identified among reproductive-age women with SLE and lupus nephritis are likely multifactorial in origin. Further research is necessary to ensure that patients who are at the highest risk for avoidable, adverse pregnancy outcomes are receiving appropriate reproductive care.