Document resource
WHAT IS ALREADY KNOWN ON THIS TOPIC Prior research has found disparities in diabetes monitoring by social determinants of health, but limited research has focused on reproductive-aged women, who may experience unique social and structural barriers to care.WHAT THIS STUDY ADDS This study provides first-of-its-kind information about the association between social determinants of health and recommended A1C testing among postpartum women with new-onset diabetes. We find disparities by race-ethnicity, insurance, and parity, and highlight that few women meet recommendations for timing and frequency of A1C testing.HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY Our findings call attention to the need to better engage all reproductive-aged women with diabetes in routine A1C monitoring, and the potential for social interventions to improve diabetes monitoring in this population.A growing proportion of reproductive-aged women have diabetes.1 Pregnancy may play an important role in this trend as a “stress test” for future cardiometabolic health.2 Gestational diabetes mellitus (GDM) increases the risk of later diabetes 12-fold over 9 years post partum, but even in women without GDM, mild glucose intolerance during pregnancy is associated with reduced beta cell function in the postpartum period and transitions to pre-diabetes and diabetes within 5 years post partum.3–5The emergence of diabetes in earlier life stages is also associated with earlier and increased severity of complications.6 As type 2 diabetes increasingly impacts women at younger ages, particularly during their reproductive years,7 disease management in this early-life to mid-life period will become an important public health priority to prevent later morbidity and mortality.A core component of diabetes management is monitoring glycemic control, which is strongly predictive of complications from diabetes.8 9 Hemoglobin A1C (A1C) is a gold-standard measure of 3-month diabetes management, and engagement in monitoring presents an opportunity to increase—and maintain—diabetes control.10 The American Diabetes Association (ADA) recommends monitoring blood glucose at least biannually among patients with type 2 diabetes and controlled glycemia.9Understanding the degree to which postpartum women newly diagnosed with diabetes receive recommended routine A1C testing may provide important information to improve long-term health trajectories.Limited data suggest that social determinants of health (SDOH) contribute to diabetes incidence and control,11 but few studies have assessed adherence to testing guidelines. SDOH are defined as the “conditions in which people are born, grow, work, live, and age”, and forces shaping access to power, money, and resources.12 Advantage is unequally distributed based on social status, including race-ethnicity and other individual-level features.12 Prior research has found disparities in diabetes prevalence and in risk of microvascular complications by race-ethnicity, education level, and income.11 13 A substantial body of literature has found racial-ethnic disparities in access to postpartum glucose screening to detect type 2 diabetes mellitus onset.14 15 However, though some studies have noted a lack of diabetes screening in premenopausal women,7 to our knowledge no study has examined the social determinants of access to testing in the postpartum period. During this critical period of biological and social transition, metabolic changes may begin or accelerate, women may lose access to care gained during pregnancy due to loss of pregnancy Medicaid or missed transitions from obstetrics to primary care in the postpartum period, and may experience unique social and structural barriers to care, such as the burden of caring for young children.16–18 Moreover, much of the research surrounding engagement in diabetes care has been conducted at a single clinic or small group of facilities.19 Prior findings may be influenced by individual clinic characteristics (eg, quality of care) and may not be representative of engagement in glucose monitoring at the population level.Taken together, although the prevalence of diabetes in reproductive-aged women is increasing, and pregnancy plays an important role in diabetes risk even in non-GDM pregnancies, the SDOH that drive access to A1C testing in postpartum women with diabetes have been understudied. To address these gaps, we analyzed data from the A1C in Pregnancy and Postpartum Linkage for Equity (APPLE) cohort, a population-based dataset of postpartum women in New York City (NYC).5 20 Our main objectives were to explore the extent to which postpartum women with diabetes receive recommended A1C testing and identify the association between markers of SDOH and receipt of recommended A1C monitoring.Research design and methods Data source The APPLE cohort 5 20 was constructed by linking birth certificate and hospital discharge data for all births in New York City between 2009–2016 with NYC A1C Registry data from 2009 to 2019, which is reported mandatorily.21For the purposes of this study, we defined “recommended” A1C monitoring as at least biannual A1C testing.22 We defined postpartum diabetes diagnosis as two or more tests ≥6.5%/48 mmol/mol recorded after birth in A1C records, the standard method for NYC A1C Registry data.23 It is impossible to discern whether diabetes identified based on A1C values represent type 1 or type 2 diabetes mellitus. However, it is expected that almost all cases identified (approximately 95%) represent type 2 diabetes mellitus.24 We thus use the term “diabetes” throughout.Study sample We excluded women with pre-pregnancy diabetes identified in the Statewide Planning and Research Cooperative System (SPARCS) record or the birth certificate ( online supplemental figure 1). A total of 6,986 women who delivered between 2009–2016 met the threshold for postpartum diabetes diagnosis. We included women who delivered between 2009–2016 to allow for adequate (up to 3 years) follow-up. Of women drawn from the birthing population, 6.6% of women with GDM (n=3,233) and 0.6% of women without GDM (n=2,357) were included in our final sample. We excluded women whose delivery insurance was listed as “self-pay” due to small numbers (n=73 after complete case analysis). We conducted a complete case analysis due to a small percent of missing data (2.1%). For our first outcome, time to first A1C test after onset, we focused on women who had at least one A1C test after diabetes onset because we cannot differentiate between outmigration from NYC and people who never had a follow-up test (n=5,590).SP110.1136/bmjdrc-2025-005745.supp1Supplementary dataFor our analysis of the rate of recommended monitoring, to ensure that all included women had the opportunity to experience the outcome, we additionally excluded women with less than 3 years of data after diabetes onset in the A1C registry (n=2,638) (online supplemental figure 1). We identified women with inadequate follow-up as those whose last interaction with the A1C registry during the study period was less than 3 years after diabetes onset. As a subanalysis, we assessed differences in SDOH and maternal characteristics by the presence of ≥1 post-diagnosis A1C test and by those with versus without 3 or more years of post-diagnosis data in the A1C registry (online supplemental table 1). We did not have information on gender identity. We use the term “women” throughout for consistency with existing literature, recognizing that not all people who give birth identify as women.SP210.1136/bmjdrc-2025-005745.supp2Supplementary dataOutcome measures To proxy the date of diabetes diagnosis, we used the date of the second A1C test with a value of ≥6.5%/48 mmol/mol in the A1C registry. The diagnosis date was considered the starting point of follow-up time. For our first outcome, we defined time to first post-diagnosis follow-up test as the difference in months between diagnosis and subsequent A1C test.For the second outcome (rate of A1C testing over 3 years), we defined 6-month intervals for each person following their diagnosis date. We calculated the number of 6-month intervals within which an individual received a test, resulting in a scale of 0 to a maximum of six tests in the recommended time frame over 3 years. Women with 0 tests in the recommended time frame over 3 years (n=109) were those whose first post-diagnosis test was 3 or more years after diabetes diagnosis.Covariates We obtained individual-level measures of social status as proxies for some SDOH from the birth certificate. These included: race-ethnicity, nativity (foreign-born vs US-born), insurance status at delivery (Medicaid vs private/other government), enrollment in the income-based Special Supplemental Nutrition Program for Women, Infants, and Children (WIC), education (less than high school diploma, high school diploma/high school equivalency (GED), some college or higher), and parity. Parity at the index pregnancy was categorized as 0, 1, 2 or 3+ prior births and used as a proxy for family size, a social context associated with time and financial burden, which has previously been found to be associated with attendance in postpartum care. 25 Nativity was included as a marker for social status due to structural inequalities and migration-related stressors.26 We conceptualized race as a social construct27 of racialization that shapes the distribution of resources, opportunities, and information within society, the ways in which people interact with health systems, and generates and upholds injustices in access to care.28 29 Race-ethnicity was self-reported. Due to small sample sizes in those who identified as Hispanic and another race, we were not able to fully disaggregate race-ethnicity. We combined race and Hispanic ethnicity into the following categories: non-Hispanic (NH) Black, Black Hispanic, other Hispanic, NH White, NH South Asian, NH Central/East Asian, and NH other/unknown. Previous research has shown that NH white reproductive-aged women have the lowest risk of having diabetes, and are significantly more likely to have controlled diabetes in comparison with Hispanic and NH Black women.30 Thus, we used NH White as the reference category. We further adjusted for maternal characteristics (pre-pregnancy body mass index (BMI) in kg/m2, age at diagnosis (10–19, 20–29, 30–39, 40+), and GDM status) as we hypothesized that these factors may drive A1C testing and have common causes with SDOH. We examined age as a categorical variable because we hypothesized that access to preventive care would not change linearly with age. We conducted a sensitivity analysis with continuous age and results were unchanged. We obtained BMI from the birth certificate and used a classification based on the National Heart, Lung, and Blood Institute (underweight: <18.5, normal weight: 18.5 to <25.0, overweight: 25.0 to <30.0, obesity: ≥30.0).31 We calculated age at diagnosis using the date of diagnosis/maternal age at delivery. We defined a woman’s index pregnancy as complicated by GDM if indicated either on the birth certificate or on the hospital discharge record (International Classification of Diseases, 9th Revision (ICD-9) codes: 648.01–648.04).Statistical analysis For our analysis of time to first follow-up A1C test after diagnosis, we estimated the median and IQR of time to first post-diagnostic test and A1C level at the first follow-up test as measures of central tendency because they are more robust to outliers than the mean. Using survival analysis, we assessed associations between SDOH and time to A1C follow-up test. We generated Kernel-smoothed hazard function plots to graphically examine time to test for each SDOH ( online supplemental figure 2) and estimated associations between SDOH and time to test using Cox proportional hazards models. We tested the proportional hazards assumptions by examining Schoenfeld residuals (online supplemental table 2) and visually examining Kernel-smoothed hazard function plots (online supplemental figure 2). We examined each SDOH individually in unadjusted models, mutually adjusting for all other SDOH, and via a third multivariable model that, in addition to mutually adjusting for SDOH, further controlled for maternal BMI, GDM, and age at diagnosis.For our second outcome (rate of biannual recommended testing), we computed the proportion of participants in each SDOH group who received at least one A1C monitoring test every 6 months for 3 years, meeting ADA recommendations for follow-up A1C testing. We constructed unadjusted Poisson regression models to estimate associations between each SDOH and the number of recommended tests completed within the first 3 years following diabetes diagnosis. As with outcome 1, we constructed a model mutually adjusting for all SDOH, and a third model further controlling for BMI, GDM status, and age at diagnosis. We conducted a supplementary analysis assessing rates of annual A1C testing, in addition to main analyses of biannual testing (online supplemental table 3). We assessed this lower threshold of A1C monitoring (annually vs biannually) to explore the extent of unmet need for A1C monitoring. We also present models adjusted for control covariates but not mutually adjusted for SDOH in online supplemental tables 4 and 5. Analyses were conducted using SAS Enterprise Guide, V.8.3 (SAS Institute, Clary, North Carolina, USA).Results Sample characteristics Characteristics of this NYC-based cohort of n=5,590 women with postpartum-onset diabetes who gave birth between 2009–2016 are presented in table 1. Most women were between 30–39 years old at diagnosis (52.0%) (median: 36, minimum: 17, maximum: 57), identified as other Hispanic (31.4%) or NH Black (30.8%), were born outside of the USA (64.2%), had less than a college education (60.2%), and used Medicaid at delivery (76.1%). A substantial proportion of women had pre-existing hypertension (10.7%), gestational hypertension (10.8%), or GDM (57.8%). More than three-quarters of women had pre-pregnancy overweight/obesity (80.4%).Table 1Characteristics of women in New York City who delivered between 2009–2016 who were diagnosed with diabetes post partum, n=5,590Socio-demographic/clinical characteristicTotaln=5,590Race-ethnicity Black Hispanic305 (5.5) NH Black1,722 (30.8) Other Hispanic1,753 (31.4) NH White440 (7.9) NH South or Southeast Asian723 (12.9) NH East or Central Asian542 (9.7) NH other/unknown105 (1.9)Nativity US-born2,004 (35.9) Foreign-born3,586 (64.2)Education Less than high school1,859 (33.3) High school diploma/GED1,501 (26.9) Some college or higher2,230 (39.9)Insurance Private1,196 (21.4) Medicaid4,256 (76.1) Other government138 (2.5)WIC enrollment No1,603 (28.7) Yes3,987 (71.3)Previous live births 01,945 (34.8) 11,584 (28.3) 21,068 (19.1) 3+993 (17.8)Gestational diabetes No2,357 (42.2) Yes3,233 (57.8)BMI Underweight44 (0.8) Normal weight1,048 (18.8) Overweight1,668 (29.8) Obesity2,830 (50.6)Age (years) at diagnosis 10–1929 (0.5) 20–29989 (17.7) 30–392,909 (52.0) 40+1,663 (29.8)BMI, body mass index; GED, High school equivalency ; NH, non-Hispanic; WIC, Women, Infants, and Children.Time to first follow-up A1C test after onset The median observed time to first follow-up A1C test after diagnosis was 5.2 months (IQR 3.1–10.0) and was relatively similar across race-ethnicity groups ( table 2). Besides women who identified as other/unknown race, other Hispanic women had the shortest median time to first follow-up test (4.8, IQR 3.0–9.5), while NH Black and Black Hispanic women experienced the longest median time to first follow-up test (5.7, IQR 3.0–11.2 and 5.6, IQR 3.1–10.5, respectively). Median A1C levels at the first follow-up A1C test after diagnosis were relatively similar across all race-ethnicity groups.Table 2Social determinants of time to follow-up A1C test (months) after diabetes diagnosis among postpartum women who gave birth in NYC from 2009 to 2016 using Cox proportional hazards model, n=5,590Median A1C % at test (IQR)Median time to test, months (IQR)HR (95% CI)aHR* (95% CI)aHR† (95% CI)Total6.8 (6.4–7.8)5.2 (3.1–10.0)Race-ethnicity NH White6.8 (6.4–8.0)4.9 (2.8–9.8) NH Black6.9 (6.4–7.8)5.7 (3.0–11.2)0.89 (0.81 to 0.99)0.88 (0.80 to 0.98)0.90 (0.80 to 1.00) Hispanic Black6.9 (6.4–8.1)5.6 (3.1–10.5)0.92 (0.79 to 1.06)0.89 (0.77 to 1.03)0.91 (0.79 to 1.06) Other Hispanic7.0 (6.4–8.5)4.8 (3.0–9.4)0.99 (0.89 to 1.10)0.96 (0.86 to 1.07)0.98 (0.88 to 1.09) NH South/Southeast Asian6.7 (6.4–7.2)5.3 (3.4–8.6)1.02 (0.91 to 1.15)0.97 (0.86 to 1.10)0.99 (0.87 to 1.12) NH Central or East Asian6.7 (6.4–7.3)5.2 (3.2–9.5)0.95 (0.84 to 1.08)0.93 (0.81 to 1.06)0.93 (0.81 to 1.06) Other/unknown6.8 (6.4–7.6)4.2 (2.7–8.0)1.11 (0.90 to 1.38)1.07 (0.87 to 1.33)1.09 (0.88 to 1.35)Nativity US-born6.9 (6.4–8.2)5.4 (3.0–11.0) Foreign-born6.8 (6.4–7.6)5.1 (3.2–9.5)1.07 (1.01 to 1.13)1.05 (0.99 to 1.11)1.01 (0.95 to 1.08)Education Less than high school6.9 (6.4–8.1)5.0 (3.0–8.9)1.05 (0.99 to 1.12)1.01 (0.94 to 1.08)1.01 (0.94 to 1.09) High school6.8 (6.4–7.7)5.4 (3.1–10.2)1.01 (0.95 to 1.08)0.99 (0.92 to 1.06)0.99 (0.92 to 1.06) Some college or higher6.8 (6.4–7.7)5.3 (3.0–10.7)Insurance Private/other6.8 (6.4–7.7)5.7 (3.2–11.7) Medicaid6.9 (6.4–7.9)5.1 (3.0–9.5)1.12 (1.05 to 1.19)1.12 (1.04 to 1.20)1.14 (1.06 to 1.22)WIC No6.9 (6.4–7.9)5.3 (3.1–10.6) Yes6.8 (6.4–7.8)5.2 (3.0–9.7)1.03 (0.97 to 1.09)0.99 (0.92 to 1.05)1.00 (0.94 to 1.07)Parity 06.9 (6.4–8.0)5.2 (3.0–10.1) 16.8 (6.4–7.6)5.3 (3.1–10.0)0.97 (0.90 to 1.03)0.96 (0.90 to 1.03)0.93 (0.87 to 1.00) 26.8 (6.4–7.7)5.3 (3.2–9.9)1.00 (0.93 to 1.08)0.99 (0.92 to 1.07)0.95 (0.87 to 1.02) 3+6.9 (6.5–8.0)5.2 (3.0–10.0)0.98 (0.91 to 1.06)0.96 (0.89 to 1.04)0.89 (0.82 to 0.97)*Mutually adjusts for social determinants of health: race-ethnicity, education, insurance, WIC enrollment, nativity, and family size.†Mutually adjusts for social determinants of health: race-ethnicity, education, insurance, WIC enrollment, nativity, and family size, and further adjusted for maternal characteristics: BMI, gestational diabetes, and age at diagnosis.aHR, adjusted HR; BMI, body mass index; NH, non-Hispanic; NYC, New York City; WIC, Women, Infants, and Children.NH Black women experienced a lower hazard of receiving their first follow-up A1C monitoring test after diabetes onset. Mutually adjusting for all SDOH, NH Black women experienced a 12% lower hazard of receiving a first follow-up A1C monitoring test after diabetes diagnosis (95% CI 0.80 to 0.98) compared with NH White women. This pattern was maintained after adjusting for GDM status, BMI, and age at diagnosis (adjusted HR (aHR): 0.90, 95% CI 0.80 to 1.00). Other Hispanic women experienced a slightly lower hazard of receiving a first test compared with NH White women (aHR: 0.98, 95% CI 0.88 to 1.09).Women insured by Medicaid at delivery experienced a shorter median time to first A1C follow-up test (5.1, IQR 3.0–9.5) compared with women who used private/other government insurance at delivery (5.7, IQR 3.2–11.7). Additionally, mutually adjusting for SDOH and further adjusting for maternal BMI, GDM, and age at diagnosis, on average, women who used Medicaid insurance experienced a 14% higher hazard of receiving a first follow-up A1C test after diagnosis than a woman insured by private/other government insurance (95% CI 1.06 to 1.22). Foreign-born women experienced a higher hazard of a first follow-up A1C test (HR: 1.07, 95% CI 1.01 to 1.13), but this finding did not persist after adjustment.Mutually adjusting for SDOH and further adjusting for BMI, GDM, and age, a subtle dose-response pattern emerged in which women with more children experience lower hazard of a follow-up test. Those with three or more previous children experienced an 11% lower hazard of receiving their first A1C test after diagnosis earlier than those with no prior children (HR: 0.89, 95% CI 0.82 to 0.97).Rate of biannual A1C monitoring Overall, few women met ADA guidelines for biannual A1C testing (13.0%) ( table 3). Approximately one-fifth (19.8%) received 5/6 recommended biannual tests. The highest proportion of women meeting biannual testing guidelines identified as NH Central or East Asian (18.8%), other Hispanic (16.2%), or NH South Asian (15.4%). Mutually adjusting for SDOH and clinical characteristics, NH Black women experienced an 8% lower rate of biannual testing compared with NH White women (adjusted rate ratio (aRR): 0.92, 95% CI 0.84 to 0.99). Trends among Black Hispanic women were similar but non-significant. Conversely, in fully adjusted models, no differences were observed between rates of testing among other Hispanic (aRR: 0.99, 95% CI 0.91 to 1.08), NH South Asian (aRR: 1.04, 95% CI 0.94 to 1.14), or NH Central/East Asian women (aRR: 1.02, 95% CI 0.92 to 1.12) compared with NH White women.Table 3Social determinants of biannual A1C testing over 3 years of follow-up using Poisson regression among postpartum women with diabetes who gave birth between 2009–2016 in NYC and who were not lost to follow-up within 3 years of diagnosis, n=2,638CategoriesProportion meeting testing guidelines (%)*RR (95% CI)aRR† (95% CI)aRR‡ (95% CI)Total13.0Race-ethnicity NH White12.1 NH Black8.10.92 (0.85 to 1.00)0.91 (0.84 to 0.99)0.92 (0.84 to 0.99) Hispanic Black7.90.92 (0.82 to 1.04)0.90 (0.80 to 1.01)0.92 (0.81 to 1.03) Other Hispanic16.21.01 (0.94 to 1.10)0.97 (0.90 to 1.06)0.99 (0.91 to 1.08) NH South Asian15.41.08 (0.99 to 1.18)1.02 (0.93 to 1.12)1.04 (0.94 to 1.14) NH Central or East Asian18.81.07 (0.97 to 1.18)1.03 (0.93 to 1.13)1.02 (0.92 to 1.12) NH other/unknown14.31.01 (0.86 to 1.18)0.97 (0.83 to 1.14)0.99 (0.85 to 1.16)Nativity US-born9.5 Foreign-born14.91.11 (1.06 to 1.16)1.07 (1.02 to 1.12)1.04 (0.99 to 1.09)Education Less than high school15.81.09 (1.04 to 1.14)1.05 (1.00 to 1.11)1.07 (1.01 to 1.12) High school12.11.04 (0.99 to 1.10)1.03 (0.98 to 1.09)1.04 (0.98 to 1.10) Some college or higher11.2Insurance Private/other insurance10.5 Medicaid insurance13.81.09 (1.04 to 1.15)1.07 (1.01 to 1.13)1.09 (1.03 to 1.16)WIC No12.9 Yes13.11.01 (0.97 to 1.06)0.98 (0.93 to 1.03)0.98 (0.93 to 1.03)Parity 013.6 112.91.01 (0.96 to 1.07)1.00 (0.95 to 1.05)0.97 (0.92 to 1.03) 211.80.99 (0.94 to 1.05)0.98 (0.92 to 1.04)0.94 (0.88 to 1.00) 3+13.61.00 (0.95 to 1.06)0.99 (0.93 to 1.05)0.93 (0.87 to 0.99)*Defined as having attended all six recommended A1c testing visits within 3 years of onset.†Mutually adjusts for social determinants of health: race-ethnicity, education, insurance, WIC enrollment, nativity, and family size.‡Mutually adjusts for social determinants of health: race-ethnicity, education, insurance, WIC enrollment, nativity, and family size and further adjusts for maternal characteristics: BMI, gestational diabetes, and age at diagnosis.aRR, adjusted rate ratio; BMI, body mass index; NH, non-Hispanic; NYC, New York City; RR, rate ratio; WIC, Women, Infants, and Children.Adjusting for SDOH and maternal characteristics, women insured by Medicaid at delivery had a higher rate of regular monitoring over 3 years (aRR: 1.09, 95% CI 1.03 to 1.16) than women who used private/other government insurance. A pattern emerged after adjustment for maternal characteristics in which women with more children experienced lower rates of biannual testing compared with women with no prior children. The lowest rate ratio was observed among women with three or more prior children compared with those with no prior children (aRR: 0.93, 95% CI 0.87 to 0.99).In the supplementary analysis assessing engagement in annual, rather than biannual, A1C testing over 3 years, the advantage among women who used Medicaid at delivery did not persist (online supplemental table 3). Approximately 56.1% of women in the sample received annual tests over the follow-up period.Discussion We explored the association between SDOH and recommended A1C follow-up testing in a diverse population of women in NYC diagnosed with diabetes post partum. NH Black women had a 10% lower hazard of receiving a first A1C follow-up test after diagnosis, and an 8% lower rate of A1C testing over time, compared with NH White women. Women who used Medicaid at delivery experienced an approximately 14% higher hazard of receiving an earlier first A1C follow-up test, and a higher rate of biannual testing over follow-up, compared with women who used private/other government insurance. Women with a greater number of children experienced a lower rate of testing over time. Overall, few women met ADA guidelines for A1C testing.Our findings build on a growing body of literature investigating SDOH and diabetes care, and are consistent with prior evidence that has found that Black adults with diabetes are less likely to receive ≥2 A1C tests per year compared with White adults with diabetes.32 Provider biases have resulted in poorer communication for Black patients, differences in provider adherence to standards of care, and reduced patient trust, all of which could reduce care utilization.33 Centuries of residential segregation and systematic disinvestment in Black neighborhoods have shaped disparities in care access, reducing individual-level resources to access care and the availability of skilled clinicians and high-quality facilities in these areas.34 While we were not able to test effects of structural racism, we observed disparities consistent with these processes among Black women.We did not find different rates of recommended follow-up A1C testing among non-Black Hispanic women. This is consistent with a review that did not find disparities in A1C testing rates among Hispanic compared with White adults.35 Despite similar routine A1C testing rates, prior research has sounded the alarm of stark ethnic disparities in diabetes outcomes, highlighting that Hispanic/Latino people with diabetes are less likely to maintain glycemic control than NH White people.36 Additional research is needed to understand this paradox, and the myriad of factors—such as access to follow-up care and prescription medications, and structural conditions—that may prevent Hispanic people with diabetes from maintaining glycemic control despite engagement in routine testing.36 37Notably, we found a higher hazard of a first follow-up A1C test among foreign-born women. This may reflect the success of programs implemented in NYC over this period, including ActionHealthNYC, which increased access to primary care among immigrants ineligible for insurance.38 Our findings did not persist after adjustment, consistent with prior research in NYC that found that other social status markers attenuated the effect of nativity on postpartum healthcare utilization.39 This suggests that other markers of SDOH, such as education and insurance, may also be important drivers of care access in this population.We surfaced differences in access to A1C testing by insurance status. The reduced access to A1C testing we observed among privately insured women highlights that interruptions in care from pregnancy to post partum occur across insurance types, and is consistent with prior findings of gaps in diabetes care after GDM among privately insured women.40 Conversely, our finding of an advantage in diabetes monitoring among women who used Medicaid at delivery is promising and may reflect successes in programs intended to engage people who used Medicaid in healthcare. For example, the Medicaid Incentives for the Prevention of Chronic Diseases program, which NY implemented, may increase later access to care after diabetes diagnosis.41 These findings also imply that we can expect a positive impact of current policies extending Medicaid to 12 months post partum, consistent with evaluations of Medicaid expansion and recommended glucose testing after GDM.42 43 The extended coverage period will provide a greater opportunity to connect women to care and initiate an A1C testing schedule. Future research might test if this is the case in our study population.For both outcomes, a pattern emerged in which women with more children received later and less frequent A1C testing, and women with ≥3 children experienced significantly lower hazards of a first test and lower rates of testing compared with women with no prior children. Prior research has surfaced childcare as an important social determinant of access to diabetes prevention, with marginalized parents disproportionately likely to delay or miss care due to childcare needs.44 Providing on-site childcare could be one potential leverage point to improve access to diabetes monitoring in this population.45Although we found little difference by SDOH in testing rates over the follow-up period, few women met ADA guidelines for A1C monitoring, suggesting a need to address conditions that hinder engagement of reproductive-aged women with diabetes in routine A1C monitoring. People with diabetes who are lost to care for a year or more are at heightened risk of diabetes-related complications and experience higher A1C when re-engaged in care.46 Patient navigators, reminder systems, and structural solutions such as paid sick leave may reduce missed and delayed primary and preventive care among reproductive-aged women.47 48Several limitations of this study, common to retrospective analyses of administrative data, should be acknowledged. Given the use of birth certificate data, other social and structural determinants of engagement in care could not be considered. Future research should incorporate measures of structural inequalities, such as distance to primary care and policies such as paid family leave, to surface upstream causes of disparities in diabetes care access. We could not assess changes in SDOH over time and assumed all SDOH were static, though some features, such as insurance status, may change over time. Some women may have had another child during the follow-up period, and we were unable to account for this. Further, we could not discern why some women had less than 3 years of data in the A1C registry or were missing any follow-up test after diagnosis. As we used A1C tests to proxy diagnosis, it is possible that some women meeting the threshold for diabetes never received a formal diagnosis and may not have known to receive follow-up testing. It is also possible that women relocated outside of NYC and received A1C follow-up testing elsewhere; if so, this analysis may underestimate receipt of regular A1C screenings. We minimized this risk by excluding women who had less than 3 years of data in the A1C registry.These findings are representative of NYC and may not be transportable to other contexts in the USA, particularly given the urbanicity and uniquely diverse population of NYC. Future research may surface context-specific social determinants of A1C testing in other geographies. Further, this study is generalizable to women who were initially connected with care and should not be generalized to women who were never connected with testing and who may differ from women who had initial connections to care on key SDOH.This study has several advantages. We used a large dataset of mandatorily reported A1C tests from clinical laboratories for all people living in NYC. Our results do not focus on a single clinical setting, which may be influenced by availability of staff and appointments and quality of care. Prior evidence has also found bias in physician ascertainment of engagement in care by race and insurance status, with physicians being more likely to classify Black patients or patients using Medicaid insurance as non-adherent.49 As our results used administrative records of care utilization, our results were not biased due to ascertainment. A key strength of this analysis was our follow-up of participants over 3 years, which provided a substantial period to assess engagement in A1C monitoring. ADA guidelines specify that A1C monitoring is a part of “continuous care” because exposure to hypoglycemia/hyperglycemia at any point in the disease course can be harmful.9Conclusion This study provides first-of-its-kind information regarding access to routine diabetes care among postpartum women, a population often excluded from chronic disease research, which can be used to target engagement efforts at the primary care level. The link between SDOH and access to routine diabetes care have not previously been explored among postpartum women, who may experience unique social and structural barriers and facilitators to care such as occupational commitments, the burden of childcare, and changes in care linkages over pregnancy and into the postpartum period. As diabetes is increasingly being diagnosed in reproductive-age women, maximizing utilization of routine diabetes monitoring will be an important priority to improve life-course cardiometabolic health trajectories and reduce disparities in diabetes-related morbidity and mortality.