Clinical and Sociobehavioral Prediction Model of 30-Day Hospital Readmissions Among People With HIV and Substance Use Disorder: Beyond Electronic Health Record Data

Model-based estimation of individual-level social determinants of health and its applications in All of Us

OBJECTIVES

We introduce a widely applicable model-based approach for estimating individual-level Social Determinants of Health (SDoH) and evaluate its effectiveness using the All of Us Research Program.

MATERIALS AND METHODS

Our approach utilizes aggregated SDoH datasets to estimate individual-level SDoH, demonstrated with examples of no high school diploma (NOHSDP) and no health insurance (UNINSUR) variables. Models are estimated using American Community Survey data and applied to derive individual-level estimates for All of Us participants. We assess concordance between model-based SDoH estimates and self-reported SDoHs in All of Us and examine associations with undiagnosed hypertension and diabetes.

RESULTS

Compared to self-reported SDoHs, the area under the curve for NOHSDP is 0.727 (95% CI, 0.724-0.730) and for UNINSUR is 0.730 (95% CI, 0.727-0.733) among the 329 074 All of Us participants, both significantly higher than aggregated SDoHs. The association between model-based NOHSDP and undiagnosed hypertension is concordant with those estimated using self-reported NOHSDP, with a correlation coefficient of 0.649. Similarly, the association between model-based NOHSDP and undiagnosed diabetes is concordant with those estimated using self-reported NOHSDP, with a correlation coefficient of 0.900.

DISCUSSION AND CONCLUSION

The model-based SDoH estimation method offers a scalable and easily standardized approach for estimating individual-level SDoHs. Using the All of Us dataset, we demonstrate reasonable concordance between model-based SDoH estimates and self-reported SDoHs, along with consistent associations with health outcomes. Our findings also underscore the critical role of geographic contexts in SDoH estimation and in evaluating the association between SDoHs and health outcomes.

Social determinants of health and sepsis: a case-control study

PURPOSE

We aimed to identify whether social determinants of health (SDoH) are associated with the development of sepsis and assess the differences between individuals living within systematically disadvantaged neighbourhoods compared with those living outside these neighbourhoods.

METHODS

We conducted a single-centre case-control study including 300 randomly selected adult patients (100 patients with sepsis and 200 patients without sepsis) admitted to the emergency department of a large academic tertiary care hospital in Hamilton, ON, Canada. We collected data on demographics and a limited set of SDoH variables, including neighbourhood household income, smoking history, social support, and history of alcohol disorder. We analyzed study data using multivariate logistic regression models.

RESULTS

The study included 100 patients with sepsis with a median [interquartile range (IQR)] age of 75 [58-84] yr and 200 patients without sepsis with a median [IQR] age of 72 [60-83] yr. Factors significantly associated with sepsis included arrival by ambulance, absence of a family physician, higher Hamilton Early Warning Score, and a recorded history of dyslipidemia. Important SDoH variables, such as individual or household income and race, were not available in the medical chart. In patients with SDoH available in their medical records, no SDoH was significantly associated with sepsis. Nevertheless, compared with their proportion of the Hamilton population, the rate of sepsis cases and sepsis deaths was approximately two times higher among patients living in systematically disadvantaged neighbourhoods.

CONCLUSIONS

This study revealed the lack of available SDoH data in electronic health records. Despite no association between the SDoH variables available and sepsis, we found a higher rate of sepsis cases and sepsis deaths among individuals living in systematically disadvantaged neighbourhoods. Including SDoH in electronic health records is crucial to study their effect on the risk of sepsis and to provide equitable care.

Clinic-level complexities prevent effective engagement of people living with HIV who are out-of-care

Approximately half of people living with HIV (PLWH) in the United States are not retained in HIV care. Although numerous studies have identified individual-level barriers to care (i.e., substance abuse, mental health, housing, transportation challenges), less is known about institutional-level barriers. We aimed to identify clinic-level barriers to HIV care and strategies to address them to better engage PLWH who have been out of care (PLWH-OOC). As part of a larger qualitative study in a Ryan White-funded HIV Clinic in Atlanta, which aimed to understand the acceptance and feasibility of community-based HIV care models to better reach PLWH-OOC, we explored barriers and facilitators of HIV care engagement. From October 2022-March 2023, 18 in-depth-interviews were conducted with HIV-care providers, administrators, social workers, and members of a Community Advisory Board (CAB) comprised of PLWH-OOC. Transcripts were coded by trained team members using a consensus approach. Several clinic-level barriers emerged: 1) the large burden placed on patients to provide proof of eligibility to receive Ryan White Program services, 2) inflexibility of provider clinic schedules, 3) inadequate processes to identify patients at risk of disengaging from care, 4) poorly-resourced hospital-to-clinic transitions, 5) inadequate systems to address primary care needs outside of HIV care, and 6) HIV stigma among medical professionals. Strategies to address these barriers included: 1) colocation of HIV and non-HIV services, 2) community-based care options that do not require patients to navigate complex transportation systems, 3) hospital and community-based peer navigation services, 4) dedicated staffing to identify and support PLWH-OOC, and 5) enhanced systems support to help patients collect the high burden of documentation required to receive subsidized HIV care. Several systems-level HIV care barriers exist and intersect with individual and community-level barriers to disproportionately affect HIV care engagement among PLWH-OOC. Findings suggest several strategies that should be considered to reach the remaining 50% of PLWH who remain out-of-care.

Characteristics of hospital and health system initiatives to address social determinants of health in the United States: a scoping review of the peer-reviewed literature

Background

Despite the incentives and provisions created for hospitals by the US Affordable Care Act related to value-based payment and community health needs assessments, concerns remain regarding the adequacy and distribution of hospital efforts to address SDOH. This scoping review of the peer-reviewed literature identifies the key characteristics of hospital/health system initiatives to address SDOH in the US, to gain insight into the progress and gaps.

Methods

PRISMA-ScR criteria were used to inform a scoping review of the literature. The article search was guided by an integrated framework of Healthy People SDOH domains and industry recommended SDOH types for hospitals. Three academic databases were searched for eligible articles from 1 January 2018 to 30 June 2023. Database searches yielded 3,027 articles, of which 70 peer-reviewed articles met the eligibility criteria for the review.

Results

Most articles (73%) were published during or after 2020 and 37% were based in Northeast US. More initiatives were undertaken by academic health centers (34%) compared to safety-net facilities (16%). Most (79%) were research initiatives, including clinical trials (40%). Only 34% of all initiatives used the EHR to collect SDOH data. Most initiatives (73%) addressed two or more types of SDOH, e.g., food and housing. A majority (74%) were downstream initiatives to address individual health-related social needs (HRSNs). Only 9% were upstream efforts to address community-level structural SDOH, e.g., housing investments. Most initiatives (74%) involved hot spotting to target HRSNs of high-risk patients, while 26% relied on screening and referral. Most initiatives (60%) relied on internal capacity vs. community partnerships (4%). Health disparities received limited attention (11%). Challenges included implementation issues and limited evidence on the systemic impact and cost savings from interventions.

Conclusion

Hospital/health system initiatives have predominantly taken the form of downstream initiatives to address HRSNs through hot-spotting or screening-and-referral. The emphasis on clinical trials coupled with lower use of EHR to collect SDOH data, limits transferability to safety-net facilities. Policymakers must create incentives for hospitals to invest in integrating SDOH data into EHR systems and harnessing community partnerships to address SDOH. Future research is needed on the systemic impact of hospital initiatives to address SDOH.

Question-answering system extracts information on injection drug use from clinical notes

BACKGROUND

Injection drug use (IDU) can increase mortality and morbidity. Therefore, identifying IDU early and initiating harm reduction interventions can benefit individuals at risk. However, extracting IDU behaviors from patients' electronic health records (EHR) is difficult because there is no other structured data available, such as International Classification of Disease (ICD) codes, and IDU is most often documented in unstructured free-text clinical notes. Although natural language processing can efficiently extract this information from unstructured data, there are no validated tools.

METHODS

To address this gap in clinical information, we design a question-answering (QA) framework to extract information on IDU from clinical notes for use in clinical operations. Our framework involves two main steps: (1) generating a gold-standard QA dataset and (2) developing and testing the QA model. We use 2323 clinical notes of 1145 patients curated from the US Department of Veterans Affairs (VA) Corporate Data Warehouse to construct the gold-standard dataset for developing and evaluating the QA model. We also demonstrate the QA model's ability to extract IDU-related information from temporally out-of-distribution data.

RESULTS

Here, we show that for a strict match between gold-standard and predicted answers, the QA model achieves a 51.65% F1 score. For a relaxed match between the gold-standard and predicted answers, the QA model obtains a 78.03% F1 score, along with 85.38% Precision and 79.02% Recall scores. Moreover, the QA model demonstrates consistent performance when subjected to temporally out-of-distribution data.

CONCLUSIONS

Our study introduces a QA framework designed to extract IDU information from clinical notes, aiming to enhance the accurate and efficient detection of people who inject drugs, extract relevant information, and ultimately facilitate informed patient care.

Realizing the Potential of Social Determinants Data: A Scoping Review of Approaches for Screening, Linkage, Extraction, Analysis and Interventions

Background

Social determinants of health (SDoH) like socioeconomics and neighborhoods strongly influence outcomes, yet standardized SDoH data is lacking in electronic health records (EHR), limiting research and care quality.

Methods

We searched PubMed using keywords "SDOH" and "EHR", underwent title/abstract and full-text screening. Included records were analyzed under five domains: 1) SDoH screening and assessment approaches, 2) SDoH data collection and documentation, 3) Use of natural language processing (NLP) for extracting SDoH, 4) SDoH data and health outcomes, and 5) SDoH-driven interventions.

Results

We identified 685 articles, of which 324 underwent full review. Key findings include tailored screening instruments implemented across settings, census and claims data linkage providing contextual SDoH profiles, rule-based and neural network systems extracting SDoH from notes using NLP, connections found between SDoH data and healthcare utilization/chronic disease control, and integrated care management programs executed. However, considerable variability persists across data sources, tools, and outcomes.

Discussion

Despite progress identifying patient social needs, further development of standards, predictive models, and coordinated interventions is critical to fulfill the potential of SDoH-EHR integration. Additional database searches could strengthen this scoping review. Ultimately widespread capture, analysis, and translation of multidimensional SDoH data into clinical care is essential for promoting health equity.

Evaluation of the Predictive Value of Routinely Collected Health-Related Social Needs Measures

The objective was to assess the value of routinely collected patient-reported health-related social needs (HRSNs) measures for predicting utilization and health outcomes. The authors identified Mayo Clinic patients with cancer, diabetes, or heart failure. The HRSN measures were collected as part of patient-reported screenings from June to December 2019 and outcomes (hospitalization, 30-day readmission, and death) were ascertained in 2020. For each outcome and disease combination, 4 models were used: gradient boosting machine (GBM), random forest (RF), generalized linear model (GLM), and elastic net (EN). Other predictors included clinical factors, demographics, and area-based HRSN measures-area deprivation index (ADI) and rurality. Predictive performance for models was evaluated with and without the routinely collected HRSN measures as change in area under the curve (AUC). Variable importance was also assessed. The differences in AUC were mixed. Significant improvements existed in 3 models of death for cancer (GBM: 0.0421, RF: 0.0496, EN: 0.0428), 3 models of hospitalization (GBM: 0.0372, RF: 0.0640, EN: 0.0441), and 1 of death (RF: 0.0754) for diabetes, and 1 model of readmissions (GBM: 0.1817), and 3 models of death (GBM: 0.0333, RF: 0.0519, GLM: 0.0489) for heart failure. Age, ADI, and the Charlson comorbidity index were the top 3 in variable importance and were consistently more important than routinely collected HRSN measures. The addition of routinely collected HRSN measures resulted in mixed improvement in the predictive performance of the models. These findings suggest that existing factors and the ADI are more important for prediction in these contexts. More work is needed to identify predictors that consistently improve model performance.

Hospital Readmissions Among Persons With Human Immunodeficiency Virus in the United States and Canada, 2005-2018: A Collaboration of Cohort Studies

BACKGROUND

Hospital readmission trends for persons with human immunodeficiency virus (PWH) in North America in the context of policy changes, improved antiretroviral therapy (ART), and aging are not well-known. We examined readmissions during 2005-2018 among adult PWH in NA-ACCORD.

METHODS

Linear risk regression estimated calendar trends in 30-day readmissions, adjusted for demographics, CD4 count, AIDS history, virologic suppression (<400 copies/mL), and cohort.

RESULTS

We examined 20 189 hospitalizations among 8823 PWH (73% cisgender men, 38% White, 38% Black). PWH hospitalized in 2018 versus 2005 had higher median age (54 vs 44 years), CD4 count (469 vs 274 cells/μL), and virologic suppression (83% vs 49%). Unadjusted 30-day readmissions decreased from 20.1% (95% confidence interval [CI], 17.9%-22.3%) in 2005 to 16.3% (95% CI, 14.1%-18.5%) in 2018. Absolute annual trends were -0.34% (95% CI, -.48% to -.19%) in unadjusted and -0.19% (95% CI, -.35% to -.02%) in adjusted analyses. By index hospitalization reason, there were significant adjusted decreases only for cardiovascular and psychiatric hospitalizations. Readmission reason was most frequently in the same diagnostic category as the index hospitalization.

CONCLUSIONS

Readmissions decreased over 2005-2018 but remained higher than the general population's. Significant decreases after adjusting for CD4 count and virologic suppression suggest that factors alongside improved ART contributed to lower readmissions. Efforts are needed to further prevent readmissions in PWH.

Screening for poverty and related social determinants to improve knowledge of and links to resources (SPARK): development and cognitive testing of a tool for primary care

BACKGROUND

Healthcare organizations are increasingly exploring ways to address the social determinants of health. Accurate data on social determinants is essential to identify opportunities for action to improve health outcomes, to identify patterns of inequity, and to help evaluate the impact of interventions. The objective of this study was to refine a standardized tool for the collection of social determinants data through cognitive testing.

METHODS

An initial set of questions on social determinants for use in healthcare settings was developed by a collaboration of hospitals and a local public health organization in Toronto, Canada during 2011-2012. Subsequent research on how patients interpreted the questions, and how they performed in primary care and other settings led to revisions. We administered these questions and conducted in-depth cognitive interviews with all the participants, who were from Saskatchewan, Manitoba, Ontario, and Newfoundland and Labrador. Cognitive interviewing was used, with participants invited to verbalize thoughts and feelings as they read the questions. Interview notes were grouped thematically, and high frequency themes were addressed.

RESULTS

Three hundred and seventy-five individuals responded to the study advertisements and 195 ultimately participated in the study. Although all interviews were conducted in English, participants were diverse. For many, the value of this information being collected in typical healthcare settings was unclear, and hence, we included descriptors for each question. In general, the questions were understood, but participants highlighted a number of ways the questions could be changed to be even clearer and more inclusive. For example, more response options were added to the question of sexual orientation and the "making ends meet" question was completely reworded in light of challenges to understand the informal phrasing cited by English as a Second Language (ESL) users of the tool.

CONCLUSION

In this work we have refined an initial set of 16 sociodemographic and social needs questions into a simple yet comprehensive 18-question tool. The changes were largely related to wording, rather than content. These questions require validation against accepted, standardized tools. Further work is required to enable community data governance, and to ensure implementation of the tool as well as the use of its data is successful in a range of organizations.

Leveraging natural language processing to augment structured social determinants of health data in the electronic health record

OBJECTIVE

Social determinants of health (SDOH) impact health outcomes and are documented in the electronic health record (EHR) through structured data and unstructured clinical notes. However, clinical notes often contain more comprehensive SDOH information, detailing aspects such as status, severity, and temporality. This work has two primary objectives: (1) develop a natural language processing information extraction model to capture detailed SDOH information and (2) evaluate the information gain achieved by applying the SDOH extractor to clinical narratives and combining the extracted representations with existing structured data.

MATERIALS AND METHODS

We developed a novel SDOH extractor using a deep learning entity and relation extraction architecture to characterize SDOH across various dimensions. In an EHR case study, we applied the SDOH extractor to a large clinical data set with 225 089 patients and 430 406 notes with social history sections and compared the extracted SDOH information with existing structured data.

RESULTS

The SDOH extractor achieved 0.86 F1 on a withheld test set. In the EHR case study, we found extracted SDOH information complements existing structured data with 32% of homeless patients, 19% of current tobacco users, and 10% of drug users only having these health risk factors documented in the clinical narrative.

CONCLUSIONS

Utilizing EHR data to identify SDOH health risk factors and social needs may improve patient care and outcomes. Semantic representations of text-encoded SDOH information can augment existing structured data, and this more comprehensive SDOH representation can assist health systems in identifying and addressing these social needs.

Risk factors associated with 30-day hospital readmissions among persons living with HIV in Nantong, China

OBJECTIVE

To estimate 30-day hospital readmission rates among persons living with HIV (PLWH) at the Nantong Infectious Disease Hospital in China and analyse the related risk factors.

METHODS

A single-centre retrospective cohort study was conducted. There were 894 PLWH records obtained from the electronic medical record (EMR) system at the Nantong Infectious Disease Hospital in China, from October 2013 to February 2018. The 30-day readmission rates were calculated, and the risk factors were analysed by generalised estimating equations (GEEs).

RESULTS

A total of 1153 hospitalizations from 894 patients were recorded between October 2013 and February 2018. The median time of 30-day readmissions was 13 days (interquartile range (IQR), 6-23). The reasons for all causes, acquired immunodeficiency syndrome (AIDS)-defining illnesses (ADIs), and non-AIDS-defining infections (non-ADIs) were 9.08, 13.52, and 7.91%, respectively. The results from the GEE analysis demonstrated that the risk factors associated with 30-days readmissions were as follows: no antiretroviral therapy (ART) prior to hospitalisations (odds ratio (OR) = 1.90, 95% confidence interval (CI): 1.21-3.00), low CD4 counts (OR = 2.17, 95% CI: 1.33-3.54), and multiple comorbidities (OR = 6.45, 95% CI: 1.62-25.73).

CONCLUSION

Early detection of HIV infection and early initiation of ART treatment are the keys to controlling 30-day readmissions.

Transitions in care between hospital and community settings for individuals with a substance use disorder: A systematic review

BACKGROUND AND AIMS

Individuals with a substance use disorder (SUD) have high rates of hospital service utilization including emergency department (ED) presentations and hospital admissions. Acute care settings offer a critical opportunity to engage individuals in addiction care and improve health outcomes especially given that the period of transition from hospital to community is challenging. This review summarizes literature on interventions for optimizing transitions in care from hospital to community for individuals with a SUD.

METHODS

The literature search focused on key terms associated with transitions in care and SUD. The search was conducted on three databases: MEDLINE, CINAHL, and PsychInfo. Eligible studies evaluated interventions acting prior to or during transitions in care from hospital to community and reported post-discharge engagement in specialized addiction care and/or return to hospital and were published since 2010.

RESULTS

Title and abstract screening were conducted for 2337 records. Overall, 31 studies met inclusion criteria, including 7 randomized controlled trials and 24 quasi-experimental designs which focused on opioid use (n = 8), alcohol use (n = 5), or polysubstance use (n = 18). Interventions included pharmacotherapy initiation (n = 7), addiction consult services (n = 9), protocol implementation (n = 3), screening, brief intervention, and referral to treatment (n = 2), patient navigation (n = 4), case management (n = 1), and recovery coaching (n = 3).

CONCLUSIONS

Both pharmacologic and psychosocial interventions implemented around transitions from acute to community care settings can improve engagement in care and reduce hospital readmission and ED presentations. Future research should focus on long-term health and social outcomes to improve quality of care for individuals with a SUD.

A Multicomponent Intervention to Reduce Readmissions Among People With HIV

BACKGROUND

Hospital readmissions are common, costly, and potentially preventable, including among people with HIV (PWH). We present the results of an evaluation of a multicomponent intervention aimed at reducing 30-day readmissions among PWH.

METHODS

Demographic, socioeconomic, and clinical variables were collected from the electronic health records of PWH or those with cellulitis (control group) hospitalized at an urban safety-net hospital before and after (from September 2012 to December 2016) the implementation of a multidisciplinary HIV transitional care team. After October 2014, hospitalized PWH could receive a medical HIV consultation ± a transitional care nurse intervention. The primary outcome was readmission to any hospital within 30 days of discharge. Multivariate logistic regression and propensity score analyses were conducted to compare readmissions before and after intervention implementation in PWH and people with cellulitis.

RESULTS

Overall, among PWH, 329 of the 2049 (16.1%) readmissions occurred before and 329 of the 2023 (16.3%) occurred after the transitional care team intervention. After including clinical and social predictors, the adjusted odds ratio of 30-day readmissions for postintervention for PWH was 0.81 (95% confidence interval: 0.66 to 0.99, P= 0.04), whereas little reduction was identified for those with cellulitis (adjusted odds ratio 0.91 (95% confidence interval: 0.81 to 1.02, P= 0.10). A dose-response effect was not observed for receipt of different HIV intervention components.

CONCLUSIONS

A multicomponent intervention reduced the adjusted risk of 30-day readmissions in PWH, although no dose-response effect was detected. Additional efforts are needed to reduce overall hospitalizations and readmissions among PWH including increasing HIV prevention, early diagnosis and engagement in care, and expanding the availability and spectrum of transitional care services.

CD4+ cell count and outcomes among HIV-infected compared with uninfected medical ICU survivors in a national cohort

BACKGROUND

People with HIV (PWH) with access to antiretroviral therapy (ART) experience excess morbidity and mortality compared with uninfected patients, particularly those with persistent viremia and without CD4+ cell recovery. We compared outcomes for medical intensive care unit (MICU) survivors with unsuppressed (>500 copies/ml) and suppressed (≤500 copies/ml) HIV-1 RNA and HIV-uninfected survivors, adjusting for CD4+ cell count.

SETTING

We studied 4537 PWH [unsuppressed = 38%; suppressed = 62%; 72% Veterans Affairs-based (VA) and 10 531 (64% VA) uninfected Veterans who survived MICU admission after entering the Veterans Aging Cohort Study (VACS) between fiscal years 2001 and 2015.

METHODS

Primary outcomes were all-cause 30-day and 6-month readmission and mortality, adjusted for demographics, CD4+ cell category (≥350 (reference); 200-349; 50-199; <50), comorbidity and prior healthcare utilization using proportional hazards models. We also adjusted for severity of illness using discharge VACS Index (VI) 2.0 among VA-based survivors.

RESULTS

In adjusted models, CD4+ categories <350 cells/μl were associated with increased risk for both outcomes up to 6 months, and risk increased with lower CD4+ categories (e.g. 6-month mortality CD4+ 200-349 hazard ratio [HR] = 1.35 [1.12-1.63]; CD4+ <50 HR = 2.14 [1.72-2.66]); unsuppressed status was not associated with outcomes. After adjusting for VI in models stratified by HIV, VI quintiles were strongly associated with both outcomes at both time points.

CONCLUSION

PWH who survive MICU admissions are at increased risk for worse outcomes compared with uninfected, especially those without CD4+ cell recovery. Severity of illness at discharge is the strongest predictor for outcomes regardless of HIV status. Strategies including intensive case management for HIV-specific and general organ dysfunction may improve outcomes for MICU survivors.

Man vs. Machine: Comparing Physician vs. Electronic Health Record-Based Model Predictions for 30-Day Hospital Readmissions

BACKGROUND

Electronic health record (EHR)-based readmission risk prediction models can be automated in real-time but have modest discrimination and may be missing important readmission risk factors. Clinician predictions of readmissions may incorporate information unavailable in the EHR, but the comparative usefulness is unknown. We sought to compare clinicians versus a validated EHR-based prediction model in predicting 30-day hospital readmissions.

METHODS

We conducted a prospective survey of internal medicine clinicians in an urban safety-net hospital. Clinicians prospectively predicted patients' 30-day readmission risk on 5-point Likert scales, subsequently dichotomized into low- vs. high-risk. We compared human with machine predictions using discrimination, net reclassification, and diagnostic test characteristics. Observed readmissions were ascertained from a regional hospitalization database. We also developed and assessed a "human-plus-machine" logistic regression model incorporating both human and machine predictions.

RESULTS

We included 1183 hospitalizations from 106 clinicians, with a readmission rate of 20.8%. Both clinicians and the EHR model had similar discrimination (C-statistic 0.66 vs. 0.66, p = 0.91). Clinicians had higher specificity (79.0% vs. 48.9%, p < 0.001) but lower sensitivity (43.9 vs. 75.2%, p < 0.001) than EHR model predictions. Compared with machine, human was better at reclassifying non-readmissions (non-event NRI + 30.1%) but worse at reclassifying readmissions (event NRI - 31.3%). A human-plus-machine approach best optimized discrimination (C-statistic 0.70, 95% CI 0.67-0.74), sensitivity (65.5%), and specificity (66.7%).

CONCLUSION

Clinicians had similar discrimination but higher specificity and lower sensitivity than EHR model predictions. Human-plus-machine was better than either alone. Readmission risk prediction strategies should incorporate clinician assessments to optimize the accuracy of readmission predictions.

MIMIC-SBDH: A Dataset for Social and Behavioral Determinants of Health

Social and Behavioral Determinants of Health (SBDHs) are environmental and behavioral factors that have a profound impact on health and related outcomes. Given their importance, physicians document SBDHs of their patients in Electronic Health Records (EHRs). However, SBDHs are mostly documented in unstructured EHR notes. Determining the status of the SBDHs requires manually reviewing the notes which can be a tedious process. Therefore, there is a need to automate identifying the patients' SBDH status in EHR notes. In this work, we created MIMIC-SBDH, the first publicly available dataset of EHR notes annotated for patients' SBDH status. Specifically, we annotated 7,025 discharge summary notes for the status of 7 SBDHs as well as marked SBDH-related keywords. Using this annotated data for training and evaluation, we evaluated the performance of three machine learning models (Random Forest, XGBoost, and Bio-ClinicalBERT) on the task of identifying SBDH status in EHR notes. The performance ranged from the lowest 0.69 F1 score for Drug Use to the highest 0.96 F1 score for Community-Present. In addition to standard evaluation metrics such as the F1 score, we evaluated four capabilities that a model must possess to perform well on the task using the CheckList tool (Ribeiro et al., 2020). The results revealed several shortcomings of the models. Our results highlighted the need to perform more capability-centric evaluations in addition to standard metric comparisons.

Racial, ethnic, and gender disparities in hospitalizations among persons with HIV in the United States and Canada, 2005-2015

OBJECTIVE

To examine recent trends and differences in all-cause and cause-specific hospitalization rates by race, ethnicity, and gender among persons with HIV (PWH) in the United States and Canada.

DESIGN

HIV clinical cohort consortium.

METHODS

We followed PWH at least 18 years old in care 2005-2015 in six clinical cohorts. We used modified Clinical Classifications Software to categorize hospital discharge diagnoses. Incidence rate ratios (IRR) were estimated using Poisson regression with robust variances to compare racial and ethnic groups, stratified by gender, adjusted for cohort, calendar year, injection drug use history, and annually updated age, CD4+, and HIV viral load.

RESULTS

Among 27 085 patients (122 566 person-years), 80% were cisgender men, 1% transgender, 43% White, 33% Black, 17% Hispanic of any race, and 1% Indigenous. Unadjusted all-cause hospitalization rates were higher for Black [IRR 1.46, 95% confidence interval (CI) 1.32-1.61] and Indigenous (1.99, 1.44-2.74) versus White cisgender men, and for Indigenous versus White cisgender women (2.55, 1.68-3.89). Unadjusted AIDS-related hospitalization rates were also higher for Black, Hispanic, and Indigenous versus White cisgender men (all P < 0.05). Transgender patients had 1.50 times (1.05-2.14) and cisgender women 1.37 times (1.26-1.48) the unadjusted hospitalization rate of cisgender men. In adjusted analyses, among both cisgender men and women, Black patients had higher rates of cardiovascular and renal/genitourinary hospitalizations compared to Whites (all P < 0.05).

CONCLUSION

Black, Hispanic, Indigenous, women, and transgender PWH in the United States and Canada experienced substantially higher hospitalization rates than White patients and cisgender men, respectively. Disparities likely have several causes, including differences in virologic suppression and chronic conditions such as diabetes and renal disease.

Brief Report: Hospitalization Rates Among Persons With HIV Who Gained Medicaid or Private Insurance After the Affordable Care Act in 2014

BACKGROUND

It is unknown whether gaining inpatient health care coverage had an effect on hospitalization rates among persons with HIV (PWH) after implementation of the Affordable Care Act in 2014.

METHODS

Hospitalization data from 2015 were obtained for 1634 adults receiving longitudinal HIV care at 3 US HIV clinics within the HIV Research Network. All patients were engaged in care and previously uninsured and supported by the Ryan White HIV/AIDS Program in 2013. We evaluated whether PWH who transitioned to either Medicaid or private insurance in 2014 tended to have a change in hospitalization rate compared with PWH who remained uncovered and Ryan White HIV/AIDS Program supported. Analyses were performed by negative binomial regression with robust standard errors, adjusting for gender, race/ethnicity, age, HIV risk factor, CD4 count, viral load, clinic site, and 2013 hospitalization rate.

RESULTS

Among PWH without inpatient health care coverage in 2013, transitioning to Medicaid [adjusted incidence rate ratio 1.26, (0.71, 2.23)] or to private insurance [0.48 (0.18, 1.28)] in 2014 was not associated with 2015 hospitalization rates, after accounting for demographics, HIV characteristics, and prior hospitalization rates. The factors significantly associated with higher hospitalization rates include age 55-64, CD4 <200 cells/µL, viral load >400 copies/mL, and 2013 hospitalization rate.

CONCLUSIONS

Acquiring inpatient coverage was not associated with a change in hospitalization rates. These results provide some evidence to allay the concern that acquiring inpatient coverage would lead to increased inpatient utilization.

Data Science Methods for Nursing-Relevant Patient Outcomes and Clinical Processes: The 2019 Literature Year in Review

Data science continues to be recognized and used within healthcare due to the increased availability of large data sets and advanced analytics. It can be challenging for nurse leaders to remain apprised of this rapidly changing landscape. In this article, we describe our findings from a scoping literature review of papers published in 2019 that use data science to explore, explain, and/or predict 15 phenomena of interest to nurses. Fourteen of the 15 phenomena were associated with at least one paper published in 2019. We identified the use of many contemporary data science methods (eg, natural language processing, neural networks) for many of the outcomes. We found many studies exploring Readmissions and Pressure Injuries. The topics of Artificial Intelligence/Machine Learning Acceptance, Burnout, Patient Safety, and Unit Culture were poorly represented. We hope that the studies described in this article help readers: (1) understand the breadth and depth of data science's ability to improve clinical processes and patient outcomes that are relevant to nurses and (2) identify gaps in the literature that are in need of exploration.

Hospitalization Rates and Outcomes Among Persons Living With Human Immunodeficiency Virus in the Southeastern United States, 1996-2016

BACKGROUND

Antiretroviral therapy (ART) advances, aging, and comorbidities impact hospitalizations in human immunodeficiency virus (HIV)-positive populations. We examined temporal trends and patient characteristics associated with hospitalization rates and outcomes.

METHODS

Among patients in the University of North Carolina Center for AIDS Research HIV Clinical Cohort receiving care during 1996-2016, we estimated annual hospitalization rates, time to inpatient mortality or live discharge, and 30-day readmission risk using bivariable Poisson, Fine-Gray, and log-binomial regression models.

RESULTS

The 4323 included patients (29% women, 60% African American) contributed 30 007 person-years. Overall, the hospitalization rate per 100 person-years was 34.3 (95% confidence interval [CI], 32.4-36.4) with a mean annual change of -3% (95% CI, -4% to -2%). Patients who were black (vs white), older, had HIV RNA >400 copies/mL, or had CD4 count <200 cells/μL had higher hospitalization rates (all P < .05). Thirty-day readmission risk was 18.9% (95% CI, 17.7%-20.2%), stable over time (P > .05 for both 2010-2016 and 2003-2009 vs 1996-2002), and higher among black patients, those with detectable HIV RNA, and those with lower CD4 cell counts (all P < .05). Higher inpatient mortality was associated with older age and lower CD4 cell count (both P < .05).

CONCLUSIONS

Hospitalization rates decreased from 1996 to 2016, but high readmissions persisted. Older patients, those of minority race/ethnicity, and those with uncontrolled HIV experienced higher rates and worse hospitalization outcomes. These findings underscore the importance of early ART and care engagement, particularly at hospital discharge.

Social determinants of health in electronic health records and their impact on analysis and risk prediction: A systematic review

OBJECTIVE

This integrative review identifies and analyzes the extant literature to examine the integration of social determinants of health (SDoH) domains into electronic health records (EHRs), their impact on risk prediction, and the specific outcomes and SDoH domains that have been tracked.

MATERIALS AND METHODS

In accordance with the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines, we conducted a literature search in the PubMed, CINAHL, Cochrane, EMBASE, and PsycINFO databases for English language studies published until March 2020 that examined SDoH domains in the context of EHRs.

RESULTS

Our search strategy identified 71 unique studies that are directly related to the research questions. 75% of the included studies were published since 2017, and 68% were U.S.-based. 79% of the reviewed articles integrated SDoH information from external data sources into EHRs, and the rest of them extracted SDoH information from unstructured clinical notes in the EHRs. We found that all but 1 study using external area-level SDoH data reported minimum contribution to performance improvement in the predictive models. In contrast, studies that incorporated individual-level SDoH data reported improved predictive performance of various outcomes such as service referrals, medication adherence, and risk of 30-day readmission. We also found little consensus on the SDoH measures used in the literature and current screening tools.

CONCLUSIONS

The literature provides early and rapidly growing evidence that integrating individual-level SDoH into EHRs can assist in risk assessment and predicting healthcare utilization and health outcomes, which further motivates efforts to collect and standardize patient-level SDoH information.

Depression and its association with adverse childhood experiences in people with substance use disorders and comorbid medical illness recruited during medical hospitalization

AIMS

People who have experienced adverse childhood experiences (ACEs) are more susceptible to substance use disorder (SUD) and depression. The present study examined depression prevalence in hospitalized patients with SUD and examined the association of individual ACEs with major depression. Depression rates 3 months after discharge were also examined.

METHODS

Medical inpatients with SUD were recruited from Temple University Hospital. Depression was assessed using the Patient Health Questionnaire-9 (PHQ-9) at baseline and 3 months post-discharge. Participants were also assessed using an ACE scale at baseline.

RESULTS

Of 79 baseline participants, 48% (38) had moderate to severe major depressive disorder (MDD) with PHQ-9 scores ≥15. Among those with baseline MDD, 38% (9/24) continued to have MDD 3 months post discharge, and 42.9% (12/28) of those without MDD at baseline met criteria at 3 months. Sixty-three percent (50/79) of the participants reported 4+ ACEs at baseline. Two ACEs, Household Incarceration and Household Mental Illness, were significantly associated with having MDD at baseline and 3 months (adjusted mean PHQ-9 total score increase (SE) and p-value: 2.97 (1.35), p < .05; 5.32 (1.37), p < .005, respectively).

CONCLUSIONS

In this exploratory study, nearly half of medical inpatients with substance use disorder had moderate to severe major depression, with a similar percentage of participants having MDD as outpatients at 3 months. Approximately two thirds of participants reported four or more adverse childhood experiences at baseline. Inpatient medical hospitalization should be utilized as an opportunity to engage people with SUD in multidisciplinary treatment including psychiatric, trauma informed care, and substance abuse treatment.