Changes in Low-Dose Aspirin Use After Updated Guidance on Sociodemographic Risk Factors for Preeclampsia

Authors: Taylor S. Freret, Allison S. Bryant, Kaitlyn E. James, Anjali J. Kaimal, Alexander Melamed, Mark A. Clapp

Abstract

Updated guidelines from the U.S. Preventive Services Task Force and the American College of Obstetricians and Gynecologists were associated with increased low-dose aspirin use among Black patients but not those with Medicaid insurance, although gaps still remain between recommended and actual use.

Rates of hypertensive disorders of pregnancy, including preeclampsia, have nearly tripled, from 2.8% in 1989 to 8.2% in 2020.^1^ Preeclampsia is a major contributor to severe maternal and perinatal morbidity; moreover, the estimated cost of preeclampsia within the first 12 months of delivery is more than $2 billion.^2^ Significant racial and ethnic disparities exist in preeclampsia incidence, morbidity, and mortality.^3–6^

There is one intervention that has been shown to reduce the rate of low-dose aspirin. Low-dose aspirin may reduce preeclampsia risk by 15–40% and may reduce the most severe forms by up to 50%.^7–10^ In 2014, the U.S. Preventive Services Task Force (USPSTF) updated its guidance to recommend low-dose aspirin for patients at risk of preeclampsia, categorizing them as high or moderate risk based on specific risk factors.^11^ In the 2014 guidance, “Sociodemographic characteristics (African American race, low socioeconomic status)” was considered a moderate-risk factor, and the USPSTF recommended that clinicians “consider” low-dose aspirin with several moderate risk factors.^11^ The American College of Obstetricians and Gynecologists (ACOG) published similar guidance for low-dose aspirin.^12^ Despite these national recommendations for low-dose aspirin use in patients at risk of preeclampsia, documented use remained well below 50%, both due to clinicians not recommending low-dose aspirin and patients not regularly adhering to the medication.^13–16^

In September 2021, building on new evidence that continued to support reduced risks of preeclampsia and perinatal mortality with low-dose aspirin and the sociodemographic disparities present in risk factors, the prevalence, and mortality from preeclampsia, the USPSTF updated its guidance. “Black persons (due to social, rather than biological, factors)” as well as a “lower income” status were listed as individual moderate-risk factors, noting that these factors are “associated with health inequities shaping health exposures, not biological propensities.” The USPSTF further recommended low-dose aspirin with two or more moderate-risk factors.^17^ The American College of Obstetricians and Gynecologists updated its own guidance in 2021, explicitly stating that Black racial identity as a risk factor was a “proxy for underlying racism.”^18^

We sought to assess whether modifications to the risk factors in the USPSTF and ACOG guidelines that emphasized the sociodemographic disparities associated with preeclampsia were associated with increased low-dose aspirin use in these groups.

METHODS

Patients were included if they were nulliparous and delivered at 24 weeks of gestation or later at one of four hospitals with at least 1,000 births per year within a single health care system in the Northeast between January 1, 2017, and December 31, 2022. Two of the hospitals are academic medical centers representing approximately 75% of the delivery volume, and two are community-based institutions. All hospitals were using the same electronic health record by the start of the study period. Multiparous patients were excluded because prior pregnancy history (eg, history of preeclampsia) could not be reliably obtained for all patients, particularly for those who delivered a prior pregnancy outside of the study institution's health system. Patients were also excluded if they did not establish care within the health care system before 14 completed weeks of gestation (because low-dose aspirin is ideally started by this gestational age) and if they transferred obstetric care after the first trimester, when clinicians may be less likely to initiate low-dose aspirin.

Risk factors for preeclampsia, as categorized by the USPSTF, were identified and abstracted from the electronic health record. Nulliparity, prepregnancy body mass index (BMI, calculated as weight in kilograms divided by height in meters squared), self-identified race, maternal age, payer for delivery admission, and multiple gestation were identified from structured data elements in the electronic health record. Race was included in our study because we were evaluating the effect of a guideline change that specifically emphasizes race. Chronic hypertension, pregestational diabetes, kidney disease, autoimmune disease, and in vitro fertilization (IVF) were identified from the International Classification of Diseases, Tenth Revision diagnosis codes if they were included in the delivery encounter or any prenatal encounter; a list of diagnosis codes is included in the Appendix 1, available online at http://links.lww.com/AOG/E33. Because only nulliparous patients were included, prior pregnancy risk factors such as history of preeclampsia, prior small-for-gestational-age birth weight, or adverse pregnancy outcomes were not applicable. Family history of preeclampsia was not reliably identifiable from the medical record and, therefore, not included as a risk factor in this analysis. Risk factors were categorized as moderate risk or high risk based on the USPSTF guidelines. The number of total moderate risk and high risk factors was also identified. In vitro fertilization did not contribute to the total number of moderate risk factors because it was not included as a risk factor in the 2014 USPSTF guidelines. Patients with missing data regarding risk factors were excluded.

The primary outcome was low-dose aspirin use (less than 325 mg), as documented on the patient's medication list and abstracted from the electronic health record from ambulatory encounters within the health care system, the medication reconciliation performed at time of hospital admission, and the text of the hospital discharge summary. Importantly, low-dose aspirin documentation on the medication list does not necessarily constitute adherence by the patient but should indicate that its use was recommended. Thus, “use” in this study is intended to represent the clinician recommendation and not the patient's adherence. The medication list includes prescriptions electronically written by clinician and those medications that are entered into the electronic health record by a health care professional (ie, patient-reported or not prescribed). The presence of low-dose aspirin listed from 6 months before conception or at any time during the pregnancy was considered indicative of low-dose aspirin use, because some patients on low-dose aspirin before conception may continue use during pregnancy without an electronic update to their medication list in the intervening period. Additionally, given the patient population, the use of low-dose aspirin for primary or secondary prevention (eg, non–pregnancy-related indications) was anticipated to be low.

To analyze the 2021 updated recommendation on the rates of low-dose aspirin use among Black people and those of lower income, we performed two independent patient-level difference-in-difference analyses. Difference-in-difference approaches use observational data to compare the change before and after an intervention between two groups with similar preintervention trends; the intervention is expected to affect only one group. The difference in the outcome in the periods before and after the intervention in the control group serves as a counterfactual in the treatment group. Any change beyond this counterfactual is attributed to the intervention.

First, we created the two cohorts of interest, excluding patients who did not have at least one high risk factor or two moderate risk factors (not including Black racial identity or Medicaid insurance, respectively) during the study period. Thus, all patients remaining in the analysis had an indication for low-dose aspirin use according to the 2014 USPSTF guidelines (either a consideration or recommendation) that did not depend on the cohort's parameter of interest. To account for known racial and socioeconomic disparities in the prevalence of risk factors for preeclampsia at baseline, we performed exact matching on binary preeclampsia risk factors (presence or absence of chronic hypertension, pregestational diabetes, chronic kidney disease, autoimmune disease, multiple gestation, maternal age 35 years or older, prepregnancy BMI 30 or higher, IVF, and either Black racial identity or Medicaid enrollment, as appropriate), year of delivery, and delivery site, stratified by racial identity (Black or White race) or Medicaid or private insurance as the primary payer for the delivery admission. For example, a patient with chronic hypertension and obesity who self-identified as Black and delivered in 2018 at Hospital A was matched to a patient with chronic hypertension and obesity who self-identified as White and delivered in 2018 at Hospital A. Patients who reported multiple races were excluded from the analysis. White race was chosen as the reference group both because the baseline rate of low-dose aspirin use, when controlling for preeclampsia risk factors, was higher than in any other group and because individuals identifying as White race made up the majority of the population. Patients were excluded if they did not have an exact match.

A patient-level difference-in-difference analysis among matched patients in each cohort was performed using linear regression. Updated USPSTF guidance was published in September 2021. Because patients who started low-dose aspirin in the first trimester at this time were unlikely to deliver before 2022, the post-period did not start until January 1, 2022. The pre-period included all patients who delivered from 2017 to 2021. To ensure conclusions drawn from difference-in-difference analysis are related to the intervention being studied, the parallel trends assumption must not be violated. This assumption states that the trends in the outcome of interest over time are parallel in both groups in the pre-period. Therefore, the group not targeted by the intervention can reasonably serve as a control for the intervention group in the post-period. We evaluated this assumption in the pre-period visually and with linear regression.

Analyses were performed using Stata MP 17.0. A two-sided P<.05 was considered statistically significant. This study was approved by the Mass General Brigham IRB (Protocol #2022P000382). STROBE reporting guidelines were followed.

RESULTS

Overall, 31,555 eligible pregnant patients delivered during the study period; after excluding those who were not eligible for low-dose aspirin before the 2021 update, there were 11,612 participants in the cohort that examined the risk factor of Black racial identity (cohort 1) and 14,208 in the cohort that examined the risk factor of having a lower income (cohort 2). Inclusion and exclusion criteria for the patients included in each analysis is shown in Figure 1.

Patients who self-identified as Black were more likely to also self-report their ethnicity as Hispanic, to have prepregnancy diabetes or obesity, to have Medicaid insurance, and to have more high-risk factors compared with patients who self-identified as White (Table 1). They were less likely to be of advanced maternal age, have a pregnancy resulting from IVF, or have a multiple gestation. After matching, 2,614 (22.5%) patients were included; each matched pair differed only by race (Black or White, Table 2).

The average rate of low-dose aspirin use was increasing by about 1.6 percentage points (pp) per year in the pre-period (95% CI, 1.1–2.0 percentage points). At the start of the study, Black patients had higher rates of low-dose aspirin use (2.1 pp, 95% CI, 0.52–3.7 percentage points). There was no difference in the rate of change of low-dose aspirin use among Black patients compared with White patients in the pre-period (0.48 percentage points, 95% CI, −0.16 to 1.1 percentage points). In addition to visual inspection, there was no concern that the parallel trends assumption was violated for this comparison.

The rate of low-dose aspirin use among White patients increased from 8.0% in the pre-period to 18.3% in the post-period (Fig. 2A). Low-dose aspirin use among Black patients increased from 11.1% in the pre-period to 30.2% in the post-period. This represents a difference-in-difference of 8.8 percentage points (95% CI, 0.76–16.9 percentage points, P=.03). Although 100% of patients who delivered in 2022 had a USPSTF recommendation for low-dose aspirin, the low-dose aspirin rate in the matched cohort was 24.3%.

Patients with Medicaid insurance were more likely to self-identify as having Hispanic ethnicity and Black racial identity, to have prepregnancy diabetes or obesity, and to have more moderate risk factors compared with a population with private insurance (Table 1). They were less likely to be of advanced maternal age or have a pregnancy resulting from IVF or multiple gestation. After matching, 3,206 (22.6%) patients were included; each matched pair differed only by Medicaid as the insurer (Table 2).

The average rate of low-dose aspirin use was increasing by 1.9 percentage points per year in the pre-period (95% CI, 0.48–3.2 percentage points). There was no difference in baseline rates of low-dose aspirin use between the two groups or in the rate of change of low-dose aspirin use in the pre-period (0.66 percentage points, 95% CI, −1.3 to 2.6 percentage points). In addition to visual inspection, there was no concern that the parallel trends assumption was violated for this comparison.

The rate of low-dose aspirin use among patients with private insurance increased from 10.4% in the pre-period to 21.6% in the post-period (Fig. 2B). Low-dose aspirin use among patients with Medicaid insurance increased from 10.5% to 20.6%. This represents a difference-in-difference of −1.1 percentage points (95% CI, −9.3 to 7.2 percentage points, P=.80). Although all patients who delivered in 2022 had a USPSTF recommendation for low-dose aspirin, the low-dose aspirin rate in the matched cohort was only 21.1%.

DISCUSSION

In this study of nulliparous patients, we demonstrate that updated guidelines from the USPSTF and ACOG regarding low-dose aspirin use for the prevention of preeclampsia were associated with a preferential increase in the rates of low-dose aspirin use among Black patients compared with White patients, but not among those with Medicaid insurance (a proxy for lower income status) compared with those with private insurance. However, overall rates of low-dose aspirin use in both cohorts remained far below what is recommended, particularly given that the USPSTF recommended low-dose aspirin use for 100% of the included patients who delivered in 2022, and less than one-quarter received it.

Given the sociodemographic inequities associated with preeclampsia and the significant evidence supporting the use of low-dose aspirin to reduce preeclampsia, it is essential that low-dose aspirin is appropriately and equitably recommended by clinicians.^3–10^ Although USPSTF and ACOG updated their guidance in 2021 to make both Black racial identity and lower income status independent and explicit moderate-risk factors for preeclampsia (previously, they were condensed under “sociodemographic characteristics”), low-dose aspirin use increased in our population for those who self-identified as Black but not for those with Medicaid insurance.^2,12,17^ We hypothesize that this difference could be due to a variety of reasons. One, a health system focus on better capturing self-identified race over time may have led to better ascertainment in the post-period relative to insurance type. Of note, however, the percentage of the population self-identifying as Black remained stable over time in the cohort. Additionally, lower income status is not specifically defined, and clinicians may find it more challenging to identify patients with this risk factor. Medicaid insurance is solely a proxy for and not clearly indicative of lower income status; furthermore, it is likely that clinicians are unaware of patient insurance status at the time of a clinical encounter when low-dose aspirin should be recommended. Finally, the guideline's specific emphasis on racism may have had a preferential effect on increasing clinician awareness and low-dose aspirin use for patients who identify as Black relative to those of lower income status or with Medicaid insurance.

Few studies have evaluated whether there are sociodemographic inequities in the use of low-dose aspirin itself. One study showed that among patients with chronic hypertension, there was no difference in the rates of clinician counseling for low-dose aspirin among Black and non-Black patients.^19^ Notably, this study was published before the 2021 guidelines were updated, did not consider rates of low-dose aspirin use among patients without chronic hypertension (a high risk factor), and compared Black patients with non-Black patients (which may include other minoritized patients and bias toward the null). Our results show that before guidance changed to emphasize sociodemographic risk factors, Black patients had a higher rate of low-dose aspirin use compared with their matched White counterparts. However, this should not be used as evidence of a lack of disparity, because every Black patient had an additional risk factor for preeclampsia and both the USPSTF and ACOG more strongly recommend low-dose aspirin among patients with multiple risk factors. There was no difference in baseline rates among patients with public insurance; again, these patients had an additional moderate-risk factor for preeclampsia.

We also found that even among patients with a recommendation for low-dose aspirin after guidance was updated, rates of use remained below expected, with up to three-quarters of patients not receiving low-dose aspirin. Our findings are consistent with that of multiple prior studies showing insufficient rates of low-dose aspirin use or adherence.^13–16^ Given the poor rates of low-dose aspirin use, in 2020, the Society for Maternal-Fetal Medicine published a checklist to improve risk factor identification and low-dose aspirin recommendations, which included a racial identity of “African ancestry (based on patient self-report)” and “Low socioeconomic status.”^20^ However, additional ways of identifying patients and ensuring clinicians adhere to recommendations equitably are needed. Moreover, although more strongly highlighting Black racial identity as a risk factor for preeclampsia was associated with increased recommendations for low-dose aspirin use in our study, it remains essential to acknowledge that race (and socioeconomic status) is being used as a proxy for disparities brought about by social, environmental, and historical injustices. Future work that seeks to reduce and eliminate baseline inequities without relying on proxy measures such as race or socioeconomic status remains essential.

Strengths of our study include its quasi-experimental nature and the use of exact matching to create a cohort for analysis that was identical with respect to risk factors for preeclampsia, except for the parameter of interest (Black race or Medicaid insurance). Limitations also include the use of exact matching, as the size of the cohort is significantly reduced; however, even with the smaller cohort we had sufficient power to detect a difference in low-dose aspirin rates. Other limitations include that the cohort is from a single health care system in the Northeast, although notably both community and academic hospitals were included. Our study excluded multiparous patients, because history of preeclampsia in a prior pregnancy could not be reliably ascertained from the data set; however, because updated guidelines were not targeted specifically toward nulliparous patients, we hypothesize that similar changes would be seen in multiparous patients. Aspirin use was determined from the electronic health record. Because aspirin is available over-the-counter, medication reconciliations are imperfect, and some patients may receive care or are prescribed the medication outside of the electronic health record, the absolute rates of low-dose aspirin use in our study may underestimate the true rate of use. However, we would not expect the reconciliation to differ by race among matched patients and therefore would expect the relative changes to be accurate. Additionally, at least one prior study has shown that recommendations for low-dose aspirin are made equally in patients who are Black compared with all other races.^19^ It is possible that low-dose aspirin use is better documented among patients with Medicaid insurance compared with those who have private insurance, because Medicaid does not require a co-pay for prescriptions for low-dose aspirin in our state (prescriptions are automatically entered into the medication list and do not require additional reconciliation). However, because this coverage did not change over time, we would not expect it to meaningfully change the difference-in-difference result. Importantly, our findings do not consider adherence to low-dose aspirin but only its documentation in the medical record. Additional limitations include that Medicaid enrollment is an imperfect proxy for income status and the use of International Classification of Diseases, Tenth Revision codes to identify risk factors; however, the use of both delivery and prenatal encounter codes makes our findings more robust.

Finally, there are limitations inherent to the use of time series methodologies such as difference-in-difference approaches. Most importantly is the requirement that no competing interventions occurred during the same time period that would influence the outcome of interest. Given the global increase in awareness of and belief in the effects of structural racism on inequities in preeclampsia outcomes, it is possible that clinicians were becoming increasingly more sensitized to the already-existing language around race from the 2014 guidance without being particularly aware of the 2021 USPSTF or ACOG practice updates. Alternatively, if an intervention improved electronic health record-based documentation of over-the-counter medications specifically in Black patients, this could have biased our results. All hospitals were using the same electronic health record by the start of the study period; although general ascertainment of over-the-counter medication use may have improved over the course of the study, we are unaware of any interventions that would have led to improved low-dose aspirin documentation specifically in Black patients but not in White patients.

In summary, updated guidelines from the USPSTF and ACOG that explicitly identified Black racial identity and lower income status as risk factors for preeclampsia (not bundled as “sociodemographic factors”) were associated with improved low-dose aspirin use preferentially among nulliparous Black patients but were not associated with improvement in rates of use among nulliparous lower-income patients as defined by Medicaid enrollment. Perhaps most importantly, our study further emphasizes the need to improve low-dose aspirin use among all patients at risk for preeclampsia, which remains far below the recommended use.