Abstract
The Cape Verdean diaspora in the United States is growing, yet Cape Verdean adults remain underrepresented in health services research. We retrospectively analyzed 583 hospitalized Cape Verdean adults admitted to a tertiary care center from 2017 to 2023, and compared their post-discharge outcomes and healthcare utilization with other racial and ethnic groups. Sequential nested regression models using inverse-probability weighting assessed how demographic, clinical, and community-level characteristics using the Environmental Justice Index (EJI) mediated the observed differences. We found that Cape Verdean adults had uniquely high primary care engagement, a difference which was magnified after EJI adjustment. They also had lower rates of specialist visits and higher emergency department use, but these associations attenuated or reversed after adjustment for clinical factors and EJI. These findings suggest that care continuity is strong within this population, potentially due to robust Community Health Center engagement, but socioeconomic and structural factors may hinder specialist access and drive emergency care use.
Keywords
Introduction
Cape Verde, an island nation off the coast of West Africa, is home to a population with a unique Afro-Portuguese linguistic and cultural identity. 1 More people with Cape Verdean ancestry live abroad than in Cape Verde, with the majority of this diaspora residing in the United States. 1 Of the estimated 110,000 Cape Verdean individuals in the US, more than 60% reside in Massachusetts and nearly 20% in Rhode Island, with additional communities in Connecticut, California, and Florida.2 -4
Despite being one of the largest African immigrant groups in New England for over a century, Cape Verdean individuals remain underrepresented in medical research; when included in epidemiological or public health assessments, they are often inappropriately grouped into broader racial or ethnic categories.3,5 Prior work suggests that Cape Verdean communities face a distinct combination of barriers to healthcare engagement, including limited English proficiency, inadequate Kriolu translation services, and low health literacy, as well as cultural norms regarding medical misinformation and mistrust which may influence health-seeking behaviors.1,5,6 These factors underscore the critical need for disaggregated analyses to identify and address disparities in healthcare access and utilization for this population.
To date, no studies have examined post-hospitalization healthcare utilization or outcomes among Cape Verdean adults. The transition from inpatient to outpatient care following hospitalization is a particularly vulnerable period where linguistic, cultural, and social determinants of health may compound the risk of adverse events. Therefore, in this study, we describe a cohort of hospitalized Cape Verdean adults and evaluate how social determinants of health mediate differences in post-discharge care utilization and outcomes – including readmission, emergency department, primary care, and specialist visits, and mortality – compared with other racial and ethnic groups.
Methods
Study Design and Study Population
For this retrospective, observational cohort study, we used electronic health record (EHR) data to identify adults age ≥18 years hospitalized at a 740-bed tertiary academic medical center serving a diverse urban population in Boston, MA, between January 2017 and September 2023. Inpatient status was defined based on billing designation. Only patients with primary care providers (PCPs) within the system’s EHR were included in order to allow for outcome ascertainment. Patients with observation status, a pregnancy-related indication for admission, or a primary care provider (PCP) outside the EHR system were excluded, as we could not ascertain healthcare utilization outside of this EHR.
Outcomes
Time-to-event outcomes at 90 days included PCP visits, specialist visits, and emergency department (ED) visits, readmissions as ascertained from the EHR billing data, as well as mortality obtained from the Massachusetts Registry of Vital Statistics. It is standard institutional practice to schedule a PCP visit prior to discharge or to provide instructions for scheduling a PCP visit at the time of discharge following an inpatient admission.
Statistical Analysis
Participants were first grouped by self-reported race and ethnicity (Cape Verdean, Hispanic, or Non-Hispanic Asian, Black, or White) as documented in dedicated fields within the EHR. Baseline characteristics between Cape Verdean and the other racial and ethnic groups were compared using chi-square tests.
Next, time-to-event outcomes at 90 days (PCP visits, specialist visits, ED visits, readmissions, and mortality) were analyzed using Fine-Grey sub-distribution hazard models to account for competing mortality (for non-mortality endpoints) and adjusted using inverse probability of treatment weighting, with White adults as the reference.
We then used a sequential nested modeling strategy to estimate the extent to which sociodemographic, clinical, and community-level factors mediated the differences in outcomes between groups: Model 1 adjusted for age and sex, Model 2 added sociodemographic characteristics (marital status, community health center PCP, non-English language preference, and insurance type), Model 3 added hospitalization and clinical characteristics (admission service, Elixhauser comorbidities, intensive care unit stay, discharge disposition), and Model 4 added the Environmental Justice Index (EJI) at the Zip Code level as a proxy for social determinants of health. 7 This approach allowed us to assess the incremental contribution of individual, clinical, and neighborhood-level factors to racial and ethnic differences in post-discharge outcomes.
As a sensitivity analysis, in an attempt to restrict to specialist visits which were likely to be related to the index admission (as opposed to specialist visits for reasons unrelated to hospitalization), we repeated the analysis above, but only included specialist visits which were concordant with the diagnosis category of the index admission, using a crosswalk between admission diagnosis category and specialist visit provider type, as outlined in Supplemental Table 1.
A 2 tailed P-value <.05 defined statistical significance. Analyses were performed in Stata 18.
Ethical Approval
Ethical approval was obtained from the institutional review board at Beth Israel Deaconess Medical Center (2023P000659), which waived the requirement for written informed consent.
Results
We identified 583 hospitalized Cape Verdean adults (median age 67 years; 57.6% female). More Cape Verdean adults had a community health center (CHC) PCP (82.7%), preferred a non-English language (75.8%), and were enrolled in Medicaid (30.2%) or Medicare Advantage (26.2%) than any other racial or ethnic group. There were significant differences in admission service and diagnoses between the Cape Verdean and non-Cape Verdean groups – despite cardiovascular disease being the most common reason for hospitalization, the Cape Verdean cohort had the lowest rates of admission to a cardiology service. While there were no significant differences in length of stay, disposition location varied between the Cape Verdean and non-Cape Verdean groups (Table 1).
Baseline Characteristics.
Abbreviations: EHR, electronic medical record; ICU, Intensive care unit; PCP, Primary care provider.
Categorical variables are presented as n(%). Continuous variables are presented as median [interquartile range]. Baseline characteristics were obtained from the EHR. Race and ethnicity categories are mutually exclusive and were assigned as follows: First, Cape Verdean adults were defined as those with self-reported race, ethnicity, or preferred language of “Cape Verdean” within coded fields in the EHR. Then, Hispanic adults were defined as remaining participants with a self-reported ethnicity of “Hispanic” within the EHR. Remaining participants with a non-Hispanic ethnicity were then categorized into non-Hispanic Asian, non-Hispanic Black, non-Hispanic Other, or non-Hispanic White based on their self-reported race within the EHR. Non-Cape Verdean patients who had a missing or non-Hispanic ethnicity and also had missing a Race category in the EHR were defined as missing. Missingness rates were 3.2%; a complete case analysis was performed.
This P-value compares the Cape Verdean group with the aggregated values of the other non-Cape Verdean racial and ethnic groups.
While all Elixhauser comorbidities were included in our models, selected comorbidities are presented in this table for brevity.
Primary Care Provider Visits
When adjusting for age and sex (Model 1), Cape Verdean adults had higher rates of post-discharge PCP visits compared to White adults (HR 1.40 [95% CI: 1.28-1.54]; P < .001). Similar findings were observed for the Black (HR 1.32 [95% CI: 1.25-1.38]; P < .001) and Hispanic groups (HR 1.32 [95% CI: 1.25-1.40]; P < .001). These differences persisted after further adjustment for sociodemographic factors (Model 2). However, after accounting for hospitalization and clinical characteristics (Model 3), this association was attenuated for Black adults, but not for Cape Verdean or Hispanic adults. Finally, in the fully adjusted models including social determinants of health (Model 4), only the Cape Verdean group was significantly more likely to have a PCP visit within 90 days compared to the White reference (HR 2.07 [95% CI: 1.48-2.90; P < .001) (Table 2).
Post-Discharge Outcomes at 90 Days.
Adjusted for age and sex.
In addition to covariates in Model 1, adjusted for marital status, community health center primary care provider, non-English language preference, and insurance.
In addition to covariates in Model 2, adjusted for Elixhauser comorbidities, intensive care unit stay, discharge to facility, and admission service.
In addition to Model 3, adjusted for percentile ranks of the Environmental Justice Index, which is equal to the summation of the Health Vulnerability Module, Environmental Burden Module, and Social Vulnerability Module percentile ranks.
Bolded text indicates p < 0.05.
Specialist Visits
In the age- and sex-adjusted models (Model 1), Cape Verdean adults were less likely to see a specialist compared to the white reference (HR 0.87 [95% CI: 0.79-0.95]; P = .002). Similar findings were observed for the Asian, Black, and Hispanic groups. After adjustment for sociodemographic factors (Model 2), these differences were attenuated for the Cape Verdean, Asian, and Hispanic groups, but not for Black adults. However, after further adjustment for comorbidities and hospitalization characteristics (Model 3) as well as social determinants of health (Model 4), there were no longer any significant differences in specialist visits for any group compared to the White reference (Table 2).
In the sensitivity analysis evaluating admission diagnosis-concordant specialist visits, results were similar for the Cape Verdean group, however Black adults had consistently lower rates of specialist visits across all models (Table 3).
Diagnosis-Concordant Specialist Visits at 90 Days.
This table shows the adjusted hazard ratio of specialist visits at 90 days which were concordant with the diagnosis-related group code of the index admission, based on the crosswalk provided in Supplemental Table 1.
Adjusted for age and sex.
In addition to covariates in Model 1, adjusted for marital status, community health center primary care provider, non-English language preference, and insurance.
In addition to covariates in Model 2, adjusted for Elixhauser comorbidities, intensive care unit stay, discharge to facility, and admission service.
In addition to Model 3, adjusted for percentile ranks of the Environmental Justice Index, which is equal to the summation of the Health Vulnerability Module, Environmental Burden Module, and Social Vulnerability Module percentile ranks.
Bolded text indicates p < 0.05.
ED Visits
Cape Verdean adults were more likely to have an ED visit after hospital discharge compared to the White reference in both the age- and sex-adjusted Model 1 (HR 1.74 [95% CI: 1.43-2.12]; P < .001) and after adjustment for sociodemographic factors in Model 2 (HR 1.40 [95% CI: 1.05-1.88]; P = .02). However, this difference was attenuated after adjustment for clinical and hospitalization characteristics in Model 3 (HR 0.77 [95% CI: 0.48-1.24]; P = 0.3) and reversed after further adjusting for social determinants of health in Model 4 (HR 0.40 [95% CI: 0.23-0.72]; P = .002). A similar pattern was observed among Hispanic adults. In contrast, Black adults had a persistently higher risk of ED visits across Models 1-3 that was attenuated, but not reversed, after adjustment for social determinants of health in Model 4. Asian adults, however, had a lower risk of readmissions than the White reference after adjustment for clinical and hospitalization characteristics (Model 3) as well as social determinants of health (Model 4) (Table 2).
Readmissions
There were no significant differences in readmissions between the Cape Verdean adults and the White reference across all models. While Black adults had a higher likelihood of readmission in the age- and sex- adjusted models (Model 1 HR 1.15 [95% CI: 1.04-1.27] P = .007), these differences were attenuated after further adjustment for sociodemographics (Model 2). In contrast, the Hispanic group had lower readmission risk compared to the White reference group after adjustments for demographics (Model 2), a finding which persisted after further adjustment for clinical and hospitalization characteristics (Model 3), but was attenuated after adjusting for social determinants of health (Model 4). Overall, there were no significant differences in 90-day readmissions in the fully adjusted models across any of the groups (Model 4) (Table 2).
Mortality
The Cape Verdean group had uniquely lower mortality at 90 days compared to the White reference in the age- and sex-adjusted Model 1 (HR 0.68 [95% CI 0.53-0.87]; P = .002) and after further adjustments for sociodemographics (Model 2). However, this difference was attenuated after adjusting for clinical and hospitalization characteristics (Model 3). In the fully adjusted models (Model 4), there were no significant differences in mortality between any of the racial and ethnic groups compared to the White reference (Table 2).
Discussion
This is the first description of a modern, hospital-based cohort of Cape Verdean adults. We found that this population had uniquely high engagement with primary care after discharge – a finding which persisted after adjustment for sociodemographic factors, clinical and hospitalization characteristics, and community-level social determinants of health. As the vast majority of Cape Verdean adults had a community health center PCP, this may be due to effective outreach programs, lower barriers to access, the strength of the patient-provider relationship, or other cultural factors that promote PCP utilization, such as patient-provider language concordance.
At the same time, our results suggest lower rates of specialist appointments for Cape Verdean adults after discharge. The attenuation of this association after adjustment of sociodemographic factors suggest that language preference, provider referral patterns, and/or insurance coverage are potential contributors to this finding and warrant further investigation. In addition, the lower rates of diagnosis-concordant post-discharge specialist visits among Black adults, which persisted even in the fully adjusted models, require additional investigation.
While Cape Verdean adults had high rates of post-discharge ED visits despite robust PCP access, their risk of ED utilization was the same as the White reference group when accounting for clinical factors and lower than the White reference group after further adjusting for social determinants of health. These findings suggest that clinical, socioeconomic, and structural factors – rather than interruptions in care continuity – are likely driving these disparities, and should be the focus of future interventions.
Finally, the disproportionate Medicare Advantage enrollment among Cape Verdean patients might be associated with the high rates of non-English language preference. Prior work demonstrates these plans specifically target low English proficiency immigrant populations, 8 though the implications for care delivery and outcomes among Cape Verdean beneficiaries are unknown and requires further research. Limitations of this study include single-center inclusion, potential race/ethnicity misclassification, inability to distinguish between route of admission (e.g., elective, direct, or through the emergency department), the requirement for patients to have an PCP within a single EHR system, and the inability to account for non-mortality outcomes external to the EHR. We also could not account for post-discharge visits that were scheduled but not attended in our analyses.
Supplemental Material
sj-docx-1-jpc-10.1177_21501319251383427 – Supplemental material for Primary Care Engagement and Post-Hospitalization Outcomes Among Cape Verdean Adults
Supplemental material, sj-docx-1-jpc-10.1177_21501319251383427 for Primary Care Engagement and Post-Hospitalization Outcomes Among Cape Verdean Adults by Lucas X. Marinacci, Leny Dias-Cunha, Sara Booth, Ashley L. O’Donoghue, Daniele D. Olveczky, Talya Salant, Doreen DeFaria Yeh, Robert W. Yeh, Julio Teixeira, Jean-Jacques Alves, Djanira Fernandes, Azariah Boyd, Rishi K. Wadhera and Jennifer Stevens in Journal of Primary Care & Community Health
Footnotes
Acknowledgements
None.
Abbreviations
CHC: Community Health Center; EJI: Environmental Justice Index.
Author’s Note
The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Ethical Considerations
The Institutional Review Board at Beth Israel Deaconess Medical center provided ethical oversight for this study.
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Dr. Marinacci received research support from grant T32-HL160522 from the National Institutes of Health.
Declaration of Conflicting Interests
The authors declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: Dr. Wadhera receives research support from the National Heart, Lung, and Blood Institute, American Heart Association Established Investigator Award, and the Donaghue Foundation, and serves as a consultant for Abbott, CVS Health, and ChamberCardio, outside the submitted work. Dr. Yeh serves as a consultant and has research grants from Abbott Vascular, Boston Scientific, and Medtronic, outside of the submitted work. Dr. Teixeira serves as a consultant for Intuitive Surgical, outside of the submitted work. All other authors report no conflicts of interest.
Data Availability Statement
Data cannot be publicly shared to protect patient privacy.
Supplemental Material
Supplemental material for this article is available online.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
