T2D Management Among a Community Clinic Patient Sample After California’s May 2022 Medicaid Expansion Authors: Annie E. Ro, Jung Min Choi, Celina Morales, Nicole Tavares Kuhn, Cecilia Wu, Luohua Jiang Published: 2026 ABSTRACT Aims:In May 2022, California expanded its Medicaid program (‘Medi-Cal’) to all low-income adults over 50 years, regardless of immigration status. We compare haemoglobin A1c (HbA1c) and body mass index (BMI) among undocumented patients with Type 2 diabetes mellitus (T2D) in a primary care setting before and after the expansion.Materials and Methods:The study focused on encounters from undocumented and documented patients from two community clinics in Los Angeles County between May 1, 2021 to June 29, 2023 (n = 14 028). The primary outcomes were HbA1c and BMI. We included documented patients to compare to secular T2D trends. We used generalized linear mixed models to estimate the changes in HbA1c and BMI pre/post May 2022 and whether there were differential changes by immigration status. We also stratified by treatment status, as Medi-Cal offered undocumented patients access to newer classes of T2D therapeutics.Results:Patients’ mean age was 61.1 years, 65.1% were female and 80.9% were Latino. Both undocumented and documented patients saw significant declines after the policy. HbA1c declined across all encounters by 0.22 points (95% CI = −0.27, −0.18) and BMI declined by 0.15 points (95% CI = −0.21, −0.10); there were no significant differences by immigration status. Patients on the new medications experienced bigger drops in HbA1c and BMI compared to those who were not on the new medications after the policy was enacted. Again, this did not vary by immigration status.Conclusions:T2D management improved among both undocumented and documented patients after Medi-Cal expansion. These improvements may reflect access to new classes of T2D medication for all patients. Introduction In May 2022, California expanded full-scope Medicaid (Medi-Cal) to all low-income Californians 50 years of age or over, irrespective of immigration status. This was a significant policy change; over 349 000 older undocumented immigrants enrolled in the program, nearly 100 000 more than initial projections [1]. ‘Undocumented’ refers to immigrants who are not citizens, nor do they have a current valid visa, green card, or other legal immigration status to reside in the United States long-term. Prior to expansion, nearly two-thirds of California undocumented immigrants under 65 years were uninsured [2]. Other states, such as Illinois, Colorado, and New York, have similarly expanded their Medicaid programs to certain segments of the undocumented population [3]. Expanding access is a promising step towards addressing the medical needs of the rapidly aging undocumented immigrant population [4]. These policies remain heavily debated; however, empirical evidence is needed to understand their possible health impacts. While increased health care access has improved specific kinds of utilization among some undocumented populations (e.g., pregnant women and prenatal care) [5], it remains to be seen whether large-scale expansions in access can improve outcomes for cardiometabolic conditions that are highly prevalent among Latinos. One such condition is Type 2 diabetes mellitus (T2D); estimates of T2D prevalence among undocumented immigrants range between 13%–15% [6, 7]. In contrast, 7.4% of non-Hispanic Whites have T2D [8]. T2D is a major driver of undocumented immigrants’ health care utilization and they seem to fare poorer in controlling their T2D than documented patients. Undocumented immigrant patients in the Los Angeles County health system had 9% higher odds of having an emergency visit related to uncontrolled T2D than documented patients [9]. The little research on expanding access to care among undocumented immigrants has shown mixed results related to T2D. On the one hand, Latino patients in the Los Angeles County safety net system who were insured had lower Haemoglobin A1c (HbA1c) than those who were uninsured [10]. However, research on MyHealth LA (MHLA), a local health care coverage plan in Los Angeles County for undocumented residents that pre-dated Medi-Cal expansion, has shown mixed effects of the program on T2D management. Compared to documented patients, undocumented patients enrolled in the program had higher odds of visiting the ED for short-term T2D-related conditions (i.e., T2D with hypoglycemia) but not for long-term related complications (i.e., T2D with diabetic chronic kidney disease) [9]. Any changes in T2D management among older undocumented adults after Medi-Cal expansion remains therefore hard to anticipate. This present study compares measures of T2D management among undocumented patients before and after the older adult Medi-Cal expansion. We examine HbA1c, a reliable indicator of long-term diabetes control, and body mass index (BMI), as obesity is a major risk factor for T2D and weight loss is a common byproduct of newer T2D therapeutics. We focus on patients in two Los Angeles County Federally Qualified Health Centers (FQHCs) in the year before and after the May 2022 Medi-Cal expansion. This is a unique patient population study to any potential changes, as they were enrolled in the MHLA program prior to expansion, which offered a primary care home at FQHCs and specialty, inpatient, and ED care through the county health system [11]. As this population already had a primary care home prior to Medi-Cal expansion, we expect our estimates for the pre/post differences in T2D control to be conservative. According to Andersen’s Model of health care utilization [12], Medi-Cal expansion could be an enabling factor that provides more access to medical care. While our undocumented patient population was already receiving primary care through the local program, Medi-Cal expansion offered new health care services that could improve T2D management, such as mental health care and access to new specialists. Further, qualitative work suggests that the mental security of having Medi-Cal made patients more likely to seek medical services [13]. Medi-Cal expansion also facilitated undocumented immigrants’ access to newer, highly effective T2D therapeutic drugs, such as glucagon-like peptide-1 receptor agonists (GLP-1 RAs) and sodium-glucose cotransporter-2 (SGLT2) inhibitors [14–17]. Despite becoming recommended courses of treatment for many T2D patients [15], undocumented patients did not readily have access to these therapeutics prior to expansion. We stratify by treatment status to distinguish between increased access to medication from general improvement in health care access. Finally, we include a comparison group of documented patients to account for secular changes. We cannot use documented immigrants as a control group in a quasi-analysis with causal implications because of the overlapping timelines of prescription patterns and Medi-Cal expansion. Nonetheless, we can evaluate if disparities in T2D management by immigration status grew or remained after Medi-Cal expansion. This descriptive comparison can be an important baseline to test with more formal policy evaluations. Medi-Cal expansion remains an actively debated issue; to our knowledge, this is the first empirical study of California’s Medi-Cal expansion on outcomes related to a highly prevalent chronic condition. Methods Data and Participants We used electronic medical records (EMR) from two Federally Qualified Health Centers (FQHCs) in Los Angeles County from May 1, 2021 to June 29, 2023. Clinic 1 serves over 40000 patients annually, while Clinic 2 cares for more than 24 000 patients. At Clinic 1, 82% of patients are Hispanic or Latino, and 97% live below 200% of the federal poverty level. At Clinic 2, 78% of the population is Hispanic or Latino, and 95% of the population is at or below the same income level. The {UC Irvine} Institutional Review Board (UCI-IRB-#2393) reviewed and approved the protocol. We limited to encounters from patients with T2D, which we determined by ICD-10 codes E11.x. We further limited to encounters from patients who were 49 years or older as of May 1, 2022. This approach maximized the number of patients in the older undocumented sample (less than 2% of patients were 49 years old). BMI and HbA1c were often not concurrently measured; as a result, we created one sample contained BMI measures and covariates (‘BMI sample’ patient n = 4414, encounter n = 25 498) and one with HbA1c measures and covariates (‘A1c sample’ patient n = 4081, encounter n = 14 028). Measures Policy Exposure We used a binary variable to designate encounters before (May 1, 2021 to April 30, 2022) and after (May 1, 2022 to June 29, 2023) Medi-Cal expansion. If an encounter occurred before the policy was enacted on May 1, 2022, it was coded a zero. If the encounter occurred after the policy, it was coded a 1. Patient Group Encounters were categorized as being from documented or undocumented patients based on payer status for encounters prior to May 1, 2022. We identified documented patients as those who ever had an encounter covered by a Medicaid managed care plan prior to May 1, 2022. Given the federal restrictions for the program, we assumed that a patient covered by Medicaid prior to the state expansion was documented; this approach has been used in previous research [16]. We identified undocumented patients as those who had an encounter where the primary payer was the county health coverage program, MyHealth LA. MHLA offered health coverage for low-income county residents who did not meet the immigration requirements for Medi-Cal. While the program did not specifically ask about immigration status, we are confident that MHLA coverage could serve as a close proxy. Patients had to enroll in MHLA at a community clinic, where they were first screened for Medi-Cal eligibility. If they met the Medi-Cal income requirement but still did not qualify, they were instead enrolled in MHLA. Outcomes HbA1c was measured from standard venipuncture samples to assess recent patient glycemic control. BMI was calculated from height and weight measurements taken at each outpatient visit and converted to BMI using the standard formula. Observations with a BMI value greater than 84 or HbA1c values greater than 14% were excluded. Both measures were treated as continuous. Treatment Status An indicator for new medications was used as a covariate in the main analysis and as an effect modifier in secondary analyses. If a patient was prescribed one of the new classes of drugs (GLP-1, SGLT2) at least 1 month before the HbA1c or BMI measure, the encounter was coded as a ‘new medication’ encounter. Covariates The covariates were biological sex (female, male), ethnicity (Hispanic, non-Hispanic), age in years, and health center (Clinic 1, Clinic 2). Analysis Linear mixed-effects regression models were used to examine the relationship between our outcomes and covariates, accounting for repeated measurements within patients. Our first set of models included main effects for patient group and an indicator for the policy change, controlling for covariates. We then added an interaction between patient group and the policy indicator to compare differential changes in HbA1c and BMI by immigration status. Finally, we included the new medication indicator to examine whether differential changes in A1c and BMI by patient group were affected by prescriptions for new medications. We then conducted another set of linear mixed-effects regression models stratified by patient group to estimate the group-specific effect of new medications after the policy change. Here, we included main effects for the policy exposure, an indicator for the new medication, and an interaction between the new medication and policy change. We conducted several sensitivity checks. We first limited our encounters to patients who had at least one encounter before and after the policy change to ensure that our results were based on the same cohort of patients. As another sensitivity check, we dichotomized HbA1c into scores that were less than 9 or greater than or equal to 9 to examine changes in the odds of highly uncontrolled T2D. Data processing was performed using SAS software version 9.4 and all models were developed in R software version 4.1.3. Results Descriptive Statistics Table 1 provides the descriptive statistics for patients in the HbA1c sample (the BMI values are from the BMI sample and full descriptives for this sample are provided in Table S1). For documented patients, the average age for encounters before the policy change was 61.3, which was very similar to the average age of 61.2 for undocumented patients. Post policy, the average age was 60.8 for documented patients and 60.9 for undocumented patients. Around 77% of documented and 85% of undocumented patients were Hispanic/Latino. The HbA1c measures indicated poor glycemic control; the average HbA1c was 8.2%. The average HbA1c for pre-policy encounters from documented patients was 8.2%, which declined to 8.1% after the policy. For undocumented patients, the average HbA1c was 8.5% pre-policy, which dropped to 8.3% post-policy. BMI was also high in the patient population (BMI sample, n = 25 498). The average BMI was 31.1, which would be categorized as obese. The average BMI across encounters prior to the policy change for documented patients was 31.4, which declined to 31.2 after the policy. For undocumented patients, BMI declined from 31.1 to 30.9. The percent on new medication classes rose from 24.8% to 41.3% for documented patients and 13.4%–42.4% for undocumented patients after Medi-Cal expansion. Changes in HbA1c and BMI After Medi-Cal Expansion Table 2 presents the findings for the first set of models for changes in HbA1c. Following Medi-Cal expansion, HbA1c significantly declined across all encounters by 0.22 points (95% CI = −0.27, −0.18). Undocumented patients exhibited higher HbA1c levels overall compared to the documented patients (ß = 0.29, 95% CI = 0.21, 0.43). Older age was also associated with lower A1c (ß = −0.03, 95% CI = −0.04, −0.03) while being Hispanic/Latino ethnicity was linked to higher A1c (ß = 0.54, 95% CI = 0.40, 0.68). In Model 2, the interaction term between older undocumented status and post-policy change was not statistically significant (ß = 0.03, 95% CI = −0.05, 0.40), indicating no differential improvement in HbA1c among undocumented patients compared to the reference. In Model 3, after adjusting for treatment, new medication class use was associated with higher HbA1c before the policy (ß = 0.41, 95% CI = 0.33, 0.48), indicating that patients with more severe or poorly controlled T2D were more likely to be prescribed these new medication classes. The interaction term between patient group and policy change did not change after including the new medication class indicator, suggesting that patterns in new medication class use did not substantially affect the change in HbA1c by immigration status, or lack thereof. Table 3 shows the results for BMI. Across all encounters, BMI significantly declined after Medi-Cal expansion (ß = −0.29, 95% CI = −0.33, −0.25). Undocumented patients, on average, had lower BMI than documented patients (ß = −0.48, 95% CI = −0.85, −0.11). The interaction between patient group and policy change was not significant, indicating that the drop in BMI after Medi-Cal expansion was comparable between documented and undocumented patients. Finally, in Model 3, receiving the new medication classes was not significantly associated with BMI. Changes in HbA1c and BMI After Medi-Cal Expansion by Treatment Status and Immigration Status Tables S2 and S3 present the results for models estimating the differential impact of the new medication classes after the policy change, stratified by patient group. The change in predicted HbA1c and BMI before and after the policy by new medication classes by patient group and treatment status is illustrated in Figure 1. For undocumented patients, those who were not on the new medication classes saw a 0.20 point decline in HbA1c after the policy (95% CI = −0.27, −0.12). The change was larger for those who were on the new medication classes after the policy change; for these patients, their HbA1c dropped significantly more than those without the new meds (ß = −0.42, 95% CI = −0.59, −0.25), decreasing by a total of 0.62 points. Similarly, documented patients not on new medication classes saw a significant drop of 0.23 points after the policy was enacted (95% CI = −0.30, −0.16), while those on the new medications saw an additional decline of 0.29 points (95% CI = −0.42, −0.15), bringing the total decline to 0.51 points. The results for BMI are presented in Figure 2. Undocumented patients not on medications had a 0.19 drop in BMI score after the policy (95% CI = −0.27, −0.12) while those on the new medication saw an additional 0.51 drop in BMI (95% CI = −0.67, −0.35). The trends were similar for documented patients; they saw a 0.11 drop in BMI (95% CI = −0.20, −0.03) and the decline was 0.51 points larger for those on the new medication (95% CI = −0.65, −0.36). Sensitivity Checks We limited encounters from patients who had at least one visit before and after May 2022 to ensure overlap in the patient population (Tables S4 and S5). While our sample size decreased (HbA1c n = 2675 patients, 11 747 encounters; BMI n = 2712 patients, 21 019 encounters), we found substantively similar results to our main analyses. We also recoded HbA1c into a binary variable at a value > = 9 to examine possible differences in highly uncontrolled diabetes (Tables S6 and S7). We found similar results as the main analysis. The odds of having very uncontrolled diabetes were significantly lower after the policy change and there was no significant interaction between the policy and immigration status. Discussion HbA1c and BMI significantly declined among undocumented patients in two Los Angeles community clinics after California’s Medi-Cal expansion to older undocumented adults. We cannot conclude that this change was solely due to the policy; however, as documented patients displayed nearly identical trends. The policy period overlapped with the documented rise in GLP-1/SGLT2 prescriptions occurring statewide [17]. While we cannot make causal claims about the policy’s impact, we suspect that the rapidly changing treatment landscape improved disease management for all T2D patients and that undocumented immigrants were able to access this changing treatment environment through Medi-Cal expansion. By the end of our analysis period, the percent of undocumented T2D patients who were prescribed the therapeutics was comparable to the documented patients. We also observed that before Medi-Cal expansion, undocumented patients had higher HbA1c compared to documented patients. This persisted after Medi-Cal expansion, suggesting that Medi-Cal expansion did not reduce disparities in glycemic control by patient immigration status. Our analysis stratified by treatment status further underscores the medication access pathway. Both undocumented and documented patients who were on new medication saw larger drops in HbA1c and BMI than those who were not. However, patients who were not prescribed the new medications also saw a modest drop in HbA1c and BMI after Medi-Cal expansion. If this trend was unique to undocumented immigrants, we might conclude that Medi-Cal expansion improved T2D management through pathways other than access to new therapeutics, such as different specialists or mental health care access. Because this finding was also present for documented immigrants, however, the decline could be explained by compositional changes in who was being prescribed the new drugs. Prior to Medi-Cal expansion, it was rare for undocumented patients to have the new therapeutics unless they had highly uncontrolled diabetes and received prior authorization after a lengthy administrative process. There were fewer barriers after expansion, which could have led to more patients with less severe T2D cases being prescribed these new drug classes. Our paper has limitations. First, the undocumented patient population was already receiving care at the community clinics prior to Medi-Cal expansion, so any changes in T2D management are conservative. While we are confident that our marker of immigration status accurately identifies undocumented patients, we acknowledge that some undocumented patients could have been classified as documented. This would have also biased our pre/post changes to the null. We also acknowledge that the treatment data came from prescriptions and we do not know if the patient was actively taking the drugs. This is a common issue in EHR, as prescriptions are often collected off-site and makes our interpretations akin to an intent-to-treat analysis. If patients were not actually taking the new medications, the estimated changes in HbA1c and BMI in our study are a conservative estimate. Finally, our analysis is at the encounter level; it is also possible that the patient population changed over time. However, our sensitivity analysis that limiting patients to those with encounters both before and after the policy change showed similar results as our main analysis. To our knowledge, this is the first empirical examination of California’s Medi-Cal expansion on T2D. We found that both undocumented and documented patients saw similar declines in HbA1c and BMI after May 2022, and the gap in these measures between these patient groups persisted after Medi-Cal expansion. We cannot make causal claims about the policy impact, but the parallel trends between these two patient groups suggest that these trends are largely due to the changing landscape of medical treatment of T2D. Medi-Cal expansion offered access to newer, more effective drugs that were also on the rise among documented patients. This access to new medication appears to be the primary way that undocumented immigrants with poor T2D control experienced improvements over time. Without Medi-Cal expansion, it is entirely possible that disparities between documented and undocumented patients could have widened; future research could conduct quasi-experimental analyses to strengthen causal inference. Researchers could also examine longer-term trends as undocumented immigrants face more immigration enforcement threats and their access to Medi-Cal becomes more restrictive. Supplementary Material Supporting Information Additional supporting information can be found online in the Supporting Information section. Table S1: Pre and post-policy change of encounters with BMI encounters. Table S2: Mixed effect models estimating HbA1c between May 2021 and June 2023, stratified by immigration status (N = 14 028). Table S3: Mixed effect model estimating BMI between May 2021 and June 2023, stratified by immigration status (N = 25 498). Table S4: Mixed effect model estimating HbA1c between May 2021 and June 2023, limited to patients with at least one visit before and after May 2022 (N = 11 747). Table S5: Mixed effect model estimating BMI between May 2021 and June 2023, limited to patients with at least one visit before and after May 2022 (N = 21 019). Table S6: Mixed effect logistic regression model predicting HbA1c > =9 (N = 14 028). Table S7: Mixed effect logistic regression model predicting HbA1c > =9, stratified by immigration status (N = 14 028).