Association Between Health Care Utilization and Immigration Enforcement Events in San Francisco

This cohort study explores whether an inclusive, local health care system could serve as a buffer against adverse effects on access to health care due to actions related to immigration status in patients who are likely undocumented.


Introduction
Researchers have documented a negative association between immigration enforcement and health care utilization among immigrants 1 and expressed concern about decreased utilization after the 2016 US presidential election. 2, 3 We explored whether an inclusive, local health care system in San Francisco acts as a buffer against adverse utilization effects of enforcement and related political events among patients who likely have undocumented immigration status.

Methods
Data for this cohort study came from a single large, integrated health system that provides services to patient members of Healthy San Francisco (HSF), a health care program that provides access to a broad array of health care services to adults unable to access other public insurance options. 4 San Francisco Health Network includes primary and specialty clinics and a hospital and trauma center and serves as the medical home for most patient members of HSF. 4  After California's Medi-Cal expansion took effect, immigration status was the primary reason HSF members were ineligible for other types of insurance. 4 Individuals with undocumented immigration status are generally excluded from public health insurance programs such as Medi-Cal.
We used participation in HSF as a proxy for adults' immigration status. 5 For analyses of adults, the 2 groups we expected would be most affected were (1) all patients who had all encounters billed to HSF (HSF always) and (2) Hispanic patients who had at least 1 encounter billed to HSF between November 1, 2015, and March 1, 2018 (HSF ever, Hispanic). Groups we expected would be less affected or not affected were Hispanic patients and non-Hispanic patients who had encounters billed to Medi-Cal only (Medi-Cal always, Hispanic and Medi-Cal always, non-Hispanic). For analyses of pediatric patients, the group we expected to be more affected was Hispanic children and the group we expected to be less affected was non-Hispanic children.
We identified 6 periods in which actual or anticipated adverse immigration policy or enforcement events (eg, local Immigration and Customs Enforcement raids, immigration enforcement executive orders, the 2016 US presidential election) occurred at the federal or local level (Figure 1). The 3 primary outcomes were the log number of encounters in primary care clinics, urgent care, and emergency departments. We also examined preventive care visits in primary care clinics, emergency department encounters specific to ambulatory care-sensitive conditions, and pediatric patient visits across all health care settings.
Our analysis was at the week-group level covering 5 weeks before and after each event.
We also pooled all 6 events and groups to achieve statistical precision. We used a difference-indifferences design controlling for week and group fixed effects. Stata Statistical Software (release 15.1) was used for analysis. Significance was set at P < .05, and tests were 2-sided. See the eAppendix in the Supplement for additional methodological details.  A, Timeline of immigration-related enforcement and policy events analyzed in this article. B, Difference-indifferences estimates of the association between these events and health care utilization among patients hypothesized to be most likely affected by these events. ICE indicates Immigration and Customs Enforcement.

Results
Among the 168 975 encounters involving 22 525 patients, 2815 patients (12.5%) were included in the HSF always group; 4627 (20.5%) in the HSF ever, Hispanic group; 5001 (22.2%) in the Medi-Cal always, Hispanic group; and 10 082 (44.8%) in the Medi-Cal always, non-Hispanic group. Plots of pre-event health care utilization suggested parallel trends before each event across groups and settings (Figure 2). In pooled estimates that compared outcomes for groups likely to be most affected with outcomes for less affected groups across all events, there were no significant associations between immigration events and utilization of primary care (difference in differences estimate, −0.008; 95% CI, −0.07 to 0.05), urgent care (difference in differences estimate, −0.024; 95% CI, −0.17 to 0.12), or the emergency department (difference in differences estimate, 0.11; 95% CI, −0.08 to 0.30) (Figure 1).

Discussion
Prior research has documented an association between decreased health care utilization and immigration enforcement in Alabama and Arizona. 1,2 We did not find systematic evidence of an association between enforcement events and changes in utilization among patients with potentially