The Association of Patient and Clinician Demographics and Concordance with Medicaid Sterilization Consent Form Validity

Introduction

The Medicaid Sterilization Consent policy was enacted in the 1970s and mandated a specific informed consent form and 30-day waiting period prior to a sterilizing procedure as a protective mechanism against coercive sterilization practices.^1–3 There is a lack of data demonstrating whether the Medicaid sterilization policy has been successful at reducing or eliminating coercive sterilization practices. However, the Medicaid sterilization policy has been identified as a barrier to desired care, as patients with Medicaid are less likely to receive desired permanent contraception after delivery than their privately insured counterparts.^1,4,5 Racial and ethnic disparities in the impact of the Medicaid sterilization policy have also been demonstrated.^6,7 Due to structural racism and discrimination, Black and Hispanic women are more likely to have Medicaid insurance and therefore more likely to face this federal policy as a barrier to desired care.⁸

Diversity in healthcare organizations has been linked to improved patient satisfaction and improved health outcomes.^9–14 Patients from historically marginalized groups also have improved health outcomes when cared for by healthcare clinicians who are more similar to them.^15,16 However, there is a lack of data regarding the association of race- and ethnicity-concordance between patient and clinician for contraceptive health outcomes. This is especially important given the unique role of the clinician in ensuring the Medicaid sterilization consent form is signed at least thirty days before delivery for patients desiring inpatient postpartum permanent contraception. Therefore, we sought to analyze whether Medicaid sterilization form validity varied by race and ethnicity of the patient and the association between clinician demographics and form validity. We hypothesized that patients of marginalized racial and ethnic identity were less likely to have valid Medicaid sterilization consent forms, clinicians of a minoritized race and ethnicity were more likely to have patients with valid forms, and that patients cared for by a race- and ethnicity- concordant clinician would be more likely to have a valid form.

1. Materials and Methods

This was a planned secondary analysis of a multi-site retrospective cohort study of patients who delivered at or beyond 20 weeks of gestation between January 1, 2018 and December 31, 2019. Full methodological details have been previously published.⁵ Patients were included in the cohort if permanent contraception was documented during the delivery hospitalization as the planned method of postpartum contraception. Delivery sites included University of California San Francisco (UCSF) in San Francisco, California; Northwestern Memorial Hospital (NMH) in Chicago, Illinois; MetroHealth Medical System (MHS) in Cleveland, Ohio; and University of Alabama at Birmingham (UAB) in Birmingham, Alabama. However, for this analysis, data from NMH were excluded since outpatient clinician names were unavailable. Data were abstracted from the linked inpatient and outpatient electronic medical records.

Our primary outcome of interest was whether the Medicaid sterilization consent form was valid at delivery (binary; yes or no). A Medicaid sterilization consent form was considered valid if: 1) at least 30 days and no more than 180 days had elapsed between when the form was signed and the delivery date for those with a term birth or 2) at least 72 hours had elapsed for those with a preterm birth or emergency abdominal surgery. Patients under 21 were excluded from this analysis as the Medicaid sterilization consent form stipulates a minimum age for surgery.

Patient demographic data (age, race, ethnicity, marital status, and college education) were extracted from the electronic medical record. Race and ethnicity were analyzed independently and conceptualized as proxy measures for experienced interpersonal, systemic, and structural racism. For this analysis, the clinician who provided the highest number of prenatal visits for the patient was recorded.

We received self-reported clinician demographics from UAB, MetroHealth, and UCSF. Clinician demographics and training variables reported included age range (10-year increments), gender identity, race, ethnicity, training, specialty, scope, religion, and comfort with counseling in permanent contraception. Race and ethnicity were analyzed independently. We excluded clinicians who did not report all 3 of the following: age range, race, gender identity. We also received peer-reported clinician demographics from UAB, MetroHealth, and UCSF. Due to higher counts of clinician peer-reported demographics, compared to self-reported, we chose to use peer-reported demographics for our primary analysis and use self-reported demographics in a secondary analysis. As a check, we compared the concordance among clinicians who were included in both the peer-reported records and self-reported records.

First, we examined the relationship between patient demographic characteristics and Medicaid sterilization consent form validity. We treated patient-level observations as independent and used unadjusted logistic regression models to assess these associations. Second, we investigated the relationship between clinician demographic characteristics and form validity. Because clinicians potentially cared for multiple patients in the cohort, observations were not independent. To account for repeated observations, we specified mixed-effects models with clinician demographics as fixed effects and clinician identity as a random effect. Third, we analyzed the association between patient-clinician demographic concordance and form validity by specifying a mixed-effects model to treat clinician identity as a random effect. To increase statistical power, racial identity was treated as binary (white vs. non-white), and age concordance was categorized to (clinician older vs. same age vs. clinician younger). For all three analyses above, chi-square p-values were derived from comparing each full model to a null intercept-only model.

Next, we also performed sensitivity analyses to determine whether accounting for specific hospital site and/or clinician as a fixed effect would change the estimated association between patient demographics and form validity. We constructed 3 logistic models for each patient demographic variable: 1) a baseline model with only the demographic, 2) a model including the demographic with hospital as a fixed effect, and 3) a model including the demographic with prenatal clinician as a fixed effect. We report the odds ratios for each demographic class category.

Finally, we calculated odds ratios and 95% confidence intervals for both clinician demographics and clinician-patient demographic concordance predicting form validity. We utilized multivariable generalized linear mixed-effects models to examine associations between clinician-patient demographic concordance and form validity and adjusted for clinician racial identity, age, and gender as covariates. All variables with more than two categories were treated as binary for simplicity (e.g., white vs. non-white, age ≥40 vs. <40). The confidence intervals for odds ratios were calculated using the Wald method.

2. Theory

Clinician implicit bias and structural racism impact counseling and fulfillment of permanent contraception.^6,17–18 Some patients are counseled away from permanent contraception (downward coercion) based on characteristics such as racial identity, age, and parity, whereas other patients are counseled towards permanent contraception (upward coercion) based on factors such as racial identity, ethnic identity, and socioeconomic status.^19–21 Demographic characteristics of the clinician may also impact contraceptive recommendations.²² Such patient-level factors and clinician-level barriers interplay with structural barriers to desired permanent contraception.²³

3. Results

There were 1,644 deliveries performed by 124 clinicians included in this analysis. 840 (51%) of delivering patients had a valid Medicaid sterilization form while 804 (49%) did not (Table 1). All invalid forms were due to the form being signed too late in prenatal care rather than too early. There were multiple statistically significant associations between patient demographics and form validity: age, race, marital status, and education level (Table 1). Among black patients, fifty-nine percent had valid forms compared to a 46% valid rate for white patients and a 42% for Asian patients (p<0.001). Patients not married had a valid rate of 57%, compared to 39% among married patients (p<0.001). Patients who did not attend college had a valid rate of 61%, compared to 52% among college educated patients (p<0.001).

We received peer-reported demographic information for 124 clinicians compared to only 57 clinicians who provided self-reported demographics. When we compared concordance between overlapping clinicians, we saw high concordance except for religion (age range: 87%, race: 93%, gender: 100%, ethnicity: 100%, training: 97%, specialty: 96%, scope: 96%, religion: 34%). We saw low levels of missingness except for peer-reported religion (81% missing). Thus, due to high missingness and low concordance, we excluded peer-reported religion from our analysis. However, the remaining high concordance confirmed our decision to select peer-reported clinician demographics as our primary analysis, and self-reported clinician demographics as our secondary analysis.

There was no significant association of peer-reported clinician demographics and Medicaid sterilization form validity (Table 2). Clinician and patient demographic concordance and form validity associations are shown in Supplemental Table 1. There were no statistically significant associations between concordance of ethnicity or age and form validity. However, patients who were of a different race than their clinicians were more likely to have a valid form (54% vs 47%, p=0.017) compared to patients of the same race as clinician.

Multivariable odds ratios for both models (clinician demographics and form validity as well as demographic concordance and form validity) are provided in Table 3. Patients who were of a different race than their clinicians were more likely to have a valid form (aOR 1.32, 95% confidence interval (CI) 1.05–1.66) compared to patients of same race as clinician.

Table 4 shows our sensitivity analyses to determine whether accounting for site and/or clinician as fixed effects would change the estimated association between patient demographics and form validity. The odds ratios look similar to our baseline odds ratios.

When we tested the peer-reported demographics model for robustness, comparing to the self-reported demographics model, we observed similar associations (Supplemental tables 2–4). Clinician religion showed a significant association (p=0.04, Table S2). Again, patients who were of a different race than their clinicians were more likely to have a valid form (aOR 1.98, 95% CI 1.40–2.80) compared to patients of same race as clinician. Additionally, we found that in the self-reported cohort, patients primarily seen by an attending physicians were less likely to have a valid report compared to patients primarily seen by a trainee physician (aOR 0.13, 95% CI 0.04–0.44).

4. Discussion

Using a multicenter cohort, we investigated the association of patient and clinician demographics, patient-clinician demographic concordance, and Medicaid sterilization consent form validity. Patients who identified as Black, unmarried, or not college educated were more likely to have valid sterilization Medicaid sterilization consent forms. Patients were also more likely to have a valid form if they were race-discordant from their clinician. As most clinicians in our study were white, their non-white patients were more likely to have valid Medicaid sterilization consent forms whereas their white patients were less likely. Importantly, these findings persisted in sensitivity analyses.

The differential form validity rate of Medicaid sterilization consent forms based on race, marital status, and education level is deserving of further study. While patients receiving their autonomously desired method of contraception is the goal, it is important for systems to continually evaluate their processes to ensure that counseling and provision practices are standardized. For example, given patient characteristics have been demonstrated to impact contraceptive counseling recommendations, it is important to consider whether such bias may also have impacted prioritization of Medicaid sterilization consent form completion during prenatal care.^19–21 This is especially important and troublesome given our historical experience with coerced sterilization and continued experience of stratified reproduction, in which some people’s fertility is devalued compared to other people’s.^24,25

Racial concordance has been demonstrated to improve outcomes outside of reproductive healthcare and hypothesized as a component of a multi-level solution to maternal health disparities.^26–28 However, our data demonstrate that patients with Medicaid who desired permanent contraception were more likely to have a valid Medicaid sterilization consent form if they were cared for by race-discordant clinicians. It is unclear whether this is a result of bias as described above – or advocacy towards reducing barriers for patient-desired contraceptive care. It is important to note that studies have shown that minoritized women often perceive race-based discrimination when receiving family planning services and also rate their family planning experiences less positively compared to white women.^29,30 Thus, understanding the complexity of associations between racial concordance, patient satisfaction and trust in the medical system, and healthcare outcomes is critical.

While this study had strengths such as the use of a large cohort across a geographically diverse sample, there were also several limitations inherent to the retrospective design. The use of the Medicaid sterilization consent form as the outcome was purposeful given the direct role the clinician plays in removing this known barrier to desired permanent contraception postpartum. Thus, having a valid form is a desired outcome for a patient desiring permanent contraception. However, we are unable to ascertain from medical record data regarding bias in counseling. Second, we relied on peer-reported demographic characteristics for our primary analysis given missingness issues with self-reported demographics, However, our findings are stable across both peer- and self-reported findings. Further, we did not measure patient perceptions of clinician demographics, which may be more important to the therapeutic relationship than self- or peer-reported clinician demographics.

5. Implications

Further research is needed to explore the complex social, cultural, and political forces that impact each step of this process. The current Medicaid sterilization consent policy plays a paradoxical role in serving as a barrier to care for the very population it was intended to protect. The possibility that clinicians can help mitigate or exacerbate this barrier either from a place of bias or patient advocacy is deserving of further reflection on the part of individual clinicians as well as ongoing research to inform potential Medicaid policy revision. Such research is necessary to ensure that the Medicaid sterilization policy’s goal of reducing coercion is being met. Investigation, clinical uptake, and incorporation into policy of measures to promote informed decision-making regarding permanent contraception, including protection from bias and coercion, remain imperative to ensure that neither upward nor downward coercion are continuing to impact permanent contraception care.

Supplementary Material