Using Machine Learning to Identify Patients at Risk of Acquiring HIV in an Urban Health System.

BACKGROUND: Effective measures exist to prevent the spread of HIV. However, the identification of patients who are candidates for these measures can be a challenge. A machine learning model to predict risk for HIV may enhance patient selection for proactive outreach. SETTING: Using data from the electronic health record at Parkland Health, 1 of the largest public healthcare systems in the country, a machine learning model is created to predict incident HIV cases. The study cohort includes any patient aged 16 or older from 2015 to 2019 (n = 458,893). METHODS: Implementing a 70:30 ratio random split of the data into training and validation sets with an incident rate <0.08% and stratified by incidence of HIV, the model is evaluated using a k-fold cross-validated (k = 5) area under the receiver operating characteristic curve leveraging Light Gradient Boosting Machine Algorithm, an ensemble classifier. RESULTS: The light gradient boosting machine produces the strongest predictive power to identify good candidates for HIV PrEP. A gradient boosting classifier produced the best result with an AUC of 0.88 (95% confidence interval: 0.86 to 0.89) on the training set and 0.85 (95% confidence interval: 0.81 to 0.89) on the validation set for a sensitivity of 77.8% and specificity of 75.1%. CONCLUSIONS: A gradient boosting model using electronic health record data can be used to identify patients at risk of acquiring HIV and implemented in the clinical setting to build outreach for preventative interventions.

Childcare needs as a barrier to healthcare among women in a safety-net health system.

BACKGROUND: Childcare needs are an understudied social determinant of health. The effect of childcare needs on access to healthcare must be understood to inform health system interventions and policy reform. This study sought to characterize childcare needs, access to childcare, and prior experience with navigating childcare needs in healthcare settings among women in a safety-net population. METHODS: We conducted a cross-sectional study of patient-reported survey data collected in-person between April and October 2019. Surveys were administered in waiting rooms of ambulatory services in a large, urban safety-net health system in Dallas, Texas. Survey respondents were derived from a random convenience sample of women waiting for outpatient appointments. Participants were screened for having children under the age of 13 and/or childcare responsibilities for inclusion in the sample. Outcomes of interest included self-reported delayed or missed care, reasons for delayed or missed care, perceived difficulty in accessing childcare, prior methods for managing childcare during healthcare appointments, and prior experience with childcare centers. RESULTS: Among the 336 respondents (96.7% response rate), 121 (36.0%) reported delaying or missing a mean 3.7 appointments/year. Among women with delayed or missed care, 54.5% reported childcare barriers as the primary reason for deferral of care, greater than transportation (33%) or insurance (25%) barriers. Respondents rated childcare access as more difficult than healthcare access. Delayed or missed care due to childcare was more common among White (68.8%) and Black (55.0%) women compared to Hispanic women (34.3%). Common methods of navigating childcare needs during scheduled appointments included bringing children to appointments (69.1%) and re-scheduling or missing the scheduled appointment (43.0%). 40.6% of patients reported leaving an appointment before completion due to childcare needs. CONCLUSIONS: Childcare needs are a leading barrier to healthcare among women accessing care in safety-net settings. Unmet childcare needs result in deferral of care, which may impact health outcomes. Childcare access is perceived as more challenging than healthcare access itself. Health system and policy interventions are needed to address childcare as a social determinant of health.

Missed Opportunities for HIV Prevention in a Large County Safety Net Health System.

INTRODUCTION: HIV pre-exposure prophylaxis (PrEP) is effective at reducing HIV transmission. However, PrEP uptake is low for racial and ethnic minorities and women, especially in the Southern US Health care clinicians should be prepared to identify all patients eligible for PrEP, provide counseling, and prescribe PrEP. METHODS: Retrospective analysis of persons newly diagnosed with HIV was conducted at a large public health system from January 2015 to June 2021. Interactions with the health system in the 5 years preceding HIV diagnosis were analyzed, and missed opportunities for HIV prevention interventions, including PrEP and condom use counseling, were identified. RESULTS: We identified 454 patients with a new HIV diagnosis with previous health system interactions. 166(36.6%) had at least 1 identifiable indication for PrEP: 42(9.3%) bacterial STI, 63(13.9%) inconsistent condom use, or 82(18%) injection drug use before HIV diagnosis. Only 7(1.5%) of patients were counseled on PrEP. Most patients (308; 67.8%) had no documented condom use history in the EHR before diagnosis, a surrogate marker for obtaining a sexual history. Patients who exclusively interacted with the emergency care setting did not receive PrEP education and were less likely to receive condom use counseling. CONCLUSION: Missed opportunities to offer HIV prevention before diagnosis were common among patients newly diagnosed with HIV. Most patients did not have sexual history documented in the chart before their HIV diagnosis. Educational interventions are needed to ensure that clinicians are prepared to identify those eligible and discuss the benefits of PrEP.

Intersecting social determinants of health among patients with childcare needs: a cross-sectional analysis of social vulnerability.

INTRODUCTION: Access to childcare is an understudied social determinant of health (SDOH). Our health system established a childcare facility for patients to address childcare barriers to healthcare. Recognizing that social risk factors often co-exist, we sought to understand intersecting social risk factors among patients with childcare needs who utilized and did not utilize the childcare facility and identify residual unmet social needs alongside childcare needs. METHODS: We conducted a cross-sectional analysis of patients who enrolled in the childcare facility from November 2020 to October 2022 to compare parameters of the Social Vulnerability Index (SVI) associated with the census tract extracted from electronic medical record (EMR) data among utilizers and non-utilizers of the facility. Overall SVI and segmentation into four themes of vulnerability (socioeconomic status, household characteristics, racial/ethnic minority status, and housing type/transportation) were compared across utilizers and utilizers. Number of 90th percentile indicators were also compared to assess extreme levels of vulnerability. A sample of utilizers additionally received a patient-reported social needs screening questionnaire administered at the childcare facility. RESULTS: Among 400 enrollees in the childcare facility, 70% utilized childcare services and 30% did not. Utilizers and non-utilizers were demographically similar, though utilizers were more likely to speak Spanish (34%) compared to non-utilizers (22%). Mean SVI was similar among utilizers and non-utilizers, but the mean number of 90th percentile indicators were higher for non-utilizers compared to utilizers (4.3 ± 2.7 vs 3.7 ± 2.7, p = 0.03), primarily driven by differences in the housing type/transportation theme (p = 0.01). Non-utilizers had a lower rate of healthcare utilization compared to utilizers (p = 0.02). Among utilizers who received patient-reported screening, 84% had one unmet social need identified, of whom 62% agreed for additional assistance. Among social work referrals, 44% were linked to social workers in their medical clinics, while 56% were supported by social work integrated in the childcare facility. CONCLUSIONS: This analysis of SDOH approximated by SVI showed actionable differences, potentially transportation barriers, among patients with childcare needs who utilized a health system-integrated childcare facility and patients who did not utilize services. Furthermore, residual unmet social needs among patients who utilized the facility demonstrate the multifactorial nature of social risk factors experienced by patients with childcare needs and opportunities to address intersecting social needs within an integrated intervention. Intersecting social needs require holistic examination and multifaceted interventions.