Development and validation of mortality prediction models based on the social determinants of health

BACKGROUND

There is no standardised approach to screening adults for social risk factors. The goal of this study was to develop mortality risk prediction models based on the social determinants of health (SDoH) for clinical risk stratification.

METHODS

Data were used from REasons for Geographic And Racial Differences in Stroke (REGARDS) study, a national, population-based, longitudinal cohort of black and white Americans aged ≥45 recruited between 2003 and 2007. Analysis was limited to participants with available SDoH and mortality data (n=20 843). All-cause mortality, available through 31 December 2018, was modelled using Cox proportional hazards with baseline individual, area-level and business-level SDoH as predictors. The area-level Social Vulnerability Index (SVI) was included for comparison. All models were adjusted for age, sex and sampling region and underwent internal split-sample validation.

RESULTS

The baseline prediction model including only age, sex and REGARDS sampling region had a c-statistic of 0.699. An individual-level SDoH model (Model 1) had a higher c-statistic than the SVI (0.723 vs 0.708, p<0.001) in the testing set. Sequentially adding area-level SDoH (c-statistic 0.723) and business-level SDoH (c-statistics 0.723) to Model 1 had minimal improvement in model discrimination. Structural racism variables were associated with all-cause mortality for black participants but did not improve model discrimination compared with Model 1 (p=0.175).

CONCLUSION

In conclusion, SDoH can improve mortality prediction over 10 years relative to a baseline model and have the potential to identify high-risk patients for further evaluation or intervention if validated externally.

Having a Say in Patient Care: Factors Associated with High and Low Voice among Home Care Workers

OBJECTIVES

To identify factors associated with high and low "voice"-or level of input in patient care decisions-among home care workers (HCWs), an often marginalized workforce that provides care in the home to older adults and those with chronic conditions.

DESIGN

We conducted a secondary analysis of data from a cross-sectional survey assessing experiences of HCWs in caring for adults with heart failure. The survey measured HCWs' voice using a validated, 5-item instrument.

SETTING AND PARTICIPANTS

The survey was conducted virtually from June 2020 to July 2021 in partnership with the 1199 Service Employees International Union (1199SEIU) Training and Employment Funds, a union labor management fund. English- or Spanish-speaking HCWs employed by a certified or licensed home care agency in New York, NY, were eligible.

METHODS

HCW voice was the main outcome of interest, which we assessed by tertiles (low, medium, and high, with medium as the referent group). Using multinominal logistic regression, we calculated odds ratios (ORs) and 95% CIs for the relationship between participant characteristics and low and high levels of voice.

RESULTS

The 261 HCWs had a mean age of 48.4 years (SD 11.9), 96.6% were female, and 44.2% identified as Hispanic. A total of 38.7% had low voice, 37.9% had medium voice, and 23.4% had high voice. In the adjusted model, factors associated with low voice included Spanish as a primary language (OR 3.71, P = .001), depersonalization-related burnout (OR 1.14, P = .04), and knowing which doctor to call (OR 0.19, P < .001). Factors associated with high voice included Spanish as a primary language (OR 2.61, P = .04) and job satisfaction (OR 1.22, P = .001).

CONCLUSIONS AND IMPLICATIONS

Organizational factors such as team communication practices-including among non-English speakers-may play an important role in HCW voice. Improving HCW voice may help retain HCWs in the workforce, but future research is needed to evaluate this.

Myocardial infarction and physical function: the REasons for Geographic And Racial Differences in Stroke prospective cohort study

Objective

To examine associations between myocardial infarction (MI) and multiple physical function metrics.

Methods

Among participants aged ≥45 years in the REasons for Geographic And Racial Differences in Stroke prospective cohort study, instrumental activities of daily living (IADL), activities of daily living (ADL), gait speed, chair stands, and Short Form-12 physical component summary (PCS) were assessed after approximately 10 years of follow-up. We examined associations between MI and physical function (no MI [n = 9,472], adjudicated MI during follow-up [n = 288, median 4.7 years prior to function assessment], history of MI at baseline [n = 745], history of MI at baseline and adjudicated MI during follow-up [n = 70, median of 6.7 years prior to function assessment]). Models were adjusted for sociodemographic characteristics, health behaviours, depressive symptoms, cognitive impairment, body mass index, diabetes, hypertension, and urinary albumin to creatinine ratio. We examined subgroups defined by age, gender, and race.

Results

The average age at baseline was 62 years old, 56% were women, and 35% Black. MI was significantly associated with worse IADL and ADL scores, IADL dependency, chair stands, and PCS, but not ADL dependency or gait speed. For example, compared to participants without MI, IADL scores (possible range 0-14, higher score represents worse function) were greater for participants with MI during follow-up (difference: 0.37 [95% CI 0.16, 0.59]), MI at baseline (0.26 [95% CI 0.12, 0.41]), and MI at baseline and follow-up (0.71 [95% CI 0.15, 1.26]), p < 0.001. Associations tended to be greater in magnitude among participants who were women and particularly Black women.

Conclusion

MI was associated with various measures of physical function. These decrements in function associated with MI may be preventable or treatable.

Treatment Intensity, Prescribing Patterns, and Blood Pressure Control in Rural Black Patients with Uncontrolled Hypertension

BACKGROUND/OBJECTIVE

Because racial disparities in hypertension treatment persist, the objective of the present study was to examine patient vs. practice characteristics that influence antihypertensive selection and treatment intensity for non-Hispanic Black (hereafter "Black") patients with uncontrolled hypertension in the rural southeastern USA.

METHODS

We enrolled 25 Black patients from each of 69 rural practices in Alabama and North Carolina with uncontrolled hypertension (systolic blood pressure (BP) ≥ 140 mm Hg) in a 4-arm cluster randomized trial of BP control interventions. Patients' antihypertensive medications were abstracted from medical records and reconciled at the baseline visit. Treatment intensity was computed using the defined daily dose (DDD) method of the World Health Organization. Correlates of greater antihypertensive medication intensity were assessed by linear regression modeling, and antihypertensive medication classes were compared by baseline systolic BP (SBP) level.

RESULTS

A total of 1431 patients were enrolled and had complete baseline data. Antihypertensive treatment intensity averaged 3.7 ± 2.6 equivalent medications at usual dosages and was significantly related to higher baseline systolic BP, older age, male sex, insurance availability, higher BMI, and concurrent diabetes, but not to practice type or medication barriers in regression models. Renin-angiotensin system inhibitors were the most commonly used medications, followed by diuretics and calcium channel blockers.

CONCLUSION/RELEVANCE

Antihypertensive treatment intensity for Black patients in the rural southeastern USA with a history of uncontrolled hypertension averaged the equivalent of almost four medications at usual dosages and was significantly associated with baseline SBP levels and other patient characteristics, but not clinic type.

TRIAL REGISTRATION

ClinicalTrials.gov NCT02866669.

Having a Say Matters: The Association Between Home Health Aides' Voice and Job Satisfaction

Purpose

Despite a rapidly growing need for home health aides (HHAs), turnover rates are high. While this is driven in large part by the demanding nature of their work and low wages, another factor may be that HHAs are often not considered part of the medical team which can leave them feeling unheard by other healthcare professionals. We sought to determine whether this concept, or HHAs' perceived voice, was associated with job satisfaction.

Methods and Design

This cross-sectional survey of English- and Spanish-speaking HHAs caring for adults with heart failure (HF) was conducted from June 2020 to July 2021 in New York, NY in partnership with a labor management fund of a large healthcare union that provides benefits and training to HHAs. Voice was assessed with a validated 5-item scale (total score range 5 to 25). Job Satisfaction was assessed with the 5-item Work Domain Satisfaction Scale (total score range 5 to 35). Multivariable linear regression analysis was used to examine the association between voice and job satisfaction.

Results

A total of 413 HHAs employed by 56 unique home care agencies completed the survey; they had a mean age of 48 years, 97.6% were female, 60.2% were Hispanic, and they worked as HHAs for a median of 10 years (IQR, 5, 17). They had a median Voice score of 18 (IQR 15-20) and mean job satisfaction score of 26.4 (SD 5.6). Higher levels of voice (1.75 [0.46-3.04]) were associated with greater job satisfaction (p=0.008). When adjusting for Race/Ethnicity, HF training, and HF knowledge, the association between Voice and job satisfaction remained significant ((1.77 [0.40-3.13]).

Conclusion

HHAs with a voice in the care of their patients experienced greater job satisfaction. Voice may be an important target for interventions aiming to improve HHAs' retention in the field.

Home Health Care Workers' Interactions with Medical Providers, Home Care Agencies, and Family Members for Patients with Heart Failure

BACKGROUND

Despite providing frequent care to heart failure (HF) patients, home health care workers (HHWs) are generally considered neither part of the health care team nor the family, and their clinical observations are often overlooked. To better understand this workforce's involvement in care, we quantified HHWs' scope of interactions with clinicians, health systems, and family caregivers.

METHODS

Community-partnered cross-sectional survey of English- and Spanish-speaking HHWs who cared for a HF patient in the last year. The survey included 6 open-ended questions about aspects of care coordination, alongside demographic and employment characteristics. Descriptive statistics were performed.

RESULTS

Three hundred ninety-one HHWs employed by 56 unique home care agencies completed the survey. HHWs took HF patients to a median of 3 doctor appointments in the last year with 21.9% of them taking patients to ≥ 7 doctor appointments. Nearly a quarter of HHWs reported that these appointments were in ≥ 3 different health systems. A third of HHWs organized care for their HF patient with ≥ 2 family caregivers.

CONCLUSIONS

HHWs' scope of health-related interactions is large, indicating that there may be novel opportunities to leverage HHWs' experiences to improve health care delivery and patient care in HF.

Better Preparation and Training Determine Home Care Workers' Self-Efficacy in Contributing to Heart Failure Self-Care

OBJECTIVE

Identify determinants of home care workers' (HCW) self-efficacy in contributing to heart failure (HF) self-care.

METHODS

Secondary analysis of a survey (n = 328) examining characteristics of HCWs caring for adults with HF in New York. Self-efficacy assessed using Caregiver Self-Efficacy in Contributing to Self-Care Scale. Standardized scores range 0-100; ≥ 70 points indicate adequate self-efficacy. Characteristics determined by self-efficacy (low vs. adequate). Prevalence ratios with 95% confidence intervals (PR [95% CI]) were estimated using multivariable Poisson regression with robust standard errors.

RESULTS

Home care workers with adequate self-efficacy had at least some prior HF training (55% vs. 17%, p < .001) and greater job satisfaction (90% vs. 77%, p = .003). Significant determinants for adequate self-efficacy were employment length (1.02 [1.00-1.03], p = .027), preparation for caregiving (3.10 [2.42-3.96], p < .001), and HF training (1.48 [1.20-1.84], p < .001).

CONCLUSION

Home care agencies and policy-makers can target caregiving preparation and HF training to improve HCWs' confidence in caring for adult HF patients.

Implications of Under-Reporting Medication Side Effects: Beta-Blockers in Heart Failure as a Case Example

INTRODUCTION

Perceiving medication side effects but not reporting them to a clinician is common. Patterns of "under-reporting" and their implications are not well described. We aimed to address this gap by examining patterns of under-reporting perceived side effects of beta-blockers among patients with heart failure.

METHODS

In 2016, a survey that evaluated medication-taking behavior was administered to 1114 participants (46.5% response rate) from The Reasons for Geographic and Racial Differences in Stroke (REGARDS) cohort with prior adjudicated heart failure hospitalization or a heart failure Medicare claim. We examined the results of survey respondents who reported taking a beta-blocker to understand patterns of under-reporting perceived beta-blocker side effects. We defined an under-reporter as a participant who perceived experiencing a side effect from their beta-blocker but did not share it with their clinician (according to survey responses). We conducted a multivariable logistic regression analysis to identify determinants of being an under-reporter. Co-variates included age, sex, race, income, level of education, geographical location, and pill burden. We also examined whether under-reporters differed in self-reported medication adherence and willingness to take additional medication to prevent a future healthcare encounter compared to participants who reported perceived side effects to their clinicians and those who did not experience side effects.

RESULTS

Among 310 respondents, 28% (n = 87) were under-reporters. Black race (odds ratio 2.11, confidence interval 1.21-3.67) and education less than college (odds ratio 2.00, confidence interval 1.09-3.67) were associated with being an under-reporter. Self-reported medication adherence was similar between groups (under-reporters: 46.3%; those who reported perceived side effects: 49.4%; those who did not experience side effects: 45.0%); under-reporters were more frequently unwilling to take additional medication to prevent a doctor's visit (18.9% vs 12.1% vs 10.8%), emergency room visit (21.6% vs 13.3% vs 9.9%), and hospitalization (17.6% vs 10.8% vs 9.0%) compared with the other groups.

CONCLUSION

We conclude that under-reporting perceived side effects of beta-blockers among adults with heart failure is common, is associated with Black race and low education, and may contribute to patient willingness to take additional medication to prevent future medical encounters.

Rural-urban differences in family and paid caregiving utilization in the United States: Findings from the Cornell National Social Survey

PURPOSE

While rates of family caregiving and paid caregiving are increasing, how often they occur together ("shared care") and whether utilization varies geographically are unknown. We examined differences in family and paid caregiving utilization by rurality and region in the United States.

METHODS

The 2020 Cornell National Social Survey is an annual cross-sectional telephone-based survey of a random sample of 1,000 US adults. Participants were asked if they have been a family caregiver, including if they provided care alongside a paid caregiver. Rural-Urban Commuting Area Codes and Census areas classified rurality and region. The association between residence and the prevalence of caregiving was determined with multivariable Poisson regression.

FINDINGS

Among 857 participants with geographic and caregiving data, 11.8% (n = 101) were rural dwellers and 34.2% were family caregivers. Rural residence (vs urban) was associated with a higher prevalence of family caregiving (PR: 1.59 [1.22, 2.06]), and Western residence (vs Northeast) was associated with a lower prevalence of family caregiving (PR: 0.63 [0.46, 0.87], P = .01). Forty percent of family caregivers shared care with a paid caregiver. There was no significant difference in shared care by rural residence in unadjusted (31.8% rural vs 43.1% urban, P = .22) or adjusted models (PR: 0.85 [0.51, 1.41], P = .53).

CONCLUSIONS

Although family caregiving was more prevalent in rural areas and certain regions, shared care did not differ by rurality or region. Studies are needed to understand why rural residents do more family caregiving without additional support from paid caregivers, and what the implications are for caregivers and care recipients.

Depressive symptoms, cognitive impairment, and all-cause mortality among REGARDS participants with heart failure

Aims

To ascertain whether depressive symptoms and cognitive impairment (CI) are associated with mortality among patients with heart failure (HF), adjusting for sociodemographic, comorbidities, and biomarkers.

Methods and results

We utilized Medicare-linked data from the Reasons for Geographic and Racial Differences in Stroke (REGARDS) Study, a biracial prospective ongoing cohort of 30 239 US community-dwelling adults, recruited in 2003-07. HF diagnosis was ascertained in claims analysis. Depressive symptoms were defined as a score ≥4 on the four-item Center for Epidemiological Studies-Depression scale. Cognitive impairment was defined as a score of ≤4 on the six-item screener that assessed three-item recall and orientation to year, month, and day of the week. Sequentially adjusted Cox proportional hazard models were used to estimate the risk of death. We analyzed 1059 REGARDS participants (mean age 73, 48%-African American) with HF; of those 146 (14%) reported depressive symptoms, 136 (13%) had CI and 31 (3%) had both. Over the median follow-up of 6.8 years (interquartile range, 3.4-10.3), 785 (74%) died. In the socio-demographics-adjusted model, CI was significantly associated with increased mortality, hazard ratio 1.24 (95% confidence interval 1.01-1.52), compared with persons with neither depressive symptoms nor CI, but this association was attenuated after further adjustment. Neither depressive symptoms alone nor their comorbidity with CI was associated with mortality. Risk factors of all-cause mortality included: low income, comorbidities, smoking, physical inactivity, and severity of HF.

Conclusion

Depressive symptoms, CI, or their comorbidity was not associated with mortality in HF in this study. Treatment of HF in elderly needs to be tailored to cognitive status and includes focus on medical comorbidities.

Individual and Neighborhood Influences on the Relationship Between Waist Circumference and Coronary Heart Disease in the REasons for Geographic and Racial Differences in Stroke Study

INTRODUCTION

The objective of this study was to describe how the relationship between waist circumference and incident coronary heart disease (CHD) is influenced by individual and neighborhood factors in the REasons for Geographic and Racial Differences in Stroke (REGARDS) Study.

METHODS

REGARDS is a cohort study of 30,239 US adults. The primary exposure was sex-specific quartiles of waist circumference. Individual covariates included sociodemographic characteristics, health status, health behavior, and usual source of care. Neighborhood (ie, zip code-level) covariates included access to primary care, poverty, rurality, and racial segregation. The main outcome was incident CHD from baseline (2003) through 2017. We used descriptive statistics, Kaplan-Meier curves, and Cox proportional hazard models to analyze the overall sample and race-sex subgroups.

RESULTS

During the study period, 23,042 study participants had 1,499 CHD events. We found a higher risk of incident CHD in the upper quartile of waist circumference compared with the first quartile in all 4 race-sex subgroups except African American men, among whom we found no relationship between waist circumference and incident CHD. Covariates did not attenuate these relationships.

CONCLUSION

In all groups except African American men, waist circumference in the highest quartile was associated with increased risk of incident CHD. Individual and neighborhood factors did not influence the relationship between waist circumference and development of CHD but differentially influenced incident CHD among race-sex subgroups.

Greater Social Functioning Associated With Lower Depressive Symptomatology Among Black Belt African Americans Enrolled in the Southeastern Collaboration to Improve Blood Pressure Control Study

Objective: In many populations, higher social functioning is associated with lower depressive symptomatology, which in turn is associated with improved cardiovascular health. This study aimed to establish an association between social functioning and depressive symptomatology, which has not yet been demonstrated in the African American Black Belt. This would be an important finding in a region with high cardiovascular morbidity.

Methods: This observational study used baseline data from 1,225 African American Black Belt residents with uncontrolled hypertension in the Southeastern Collaboration to Improve Blood Pressure Control (SEC) trial. Three Patient Reported Outcomes Measurement Information System (PROMIS) questionnaires-the emotional support, instrumental support, and social isolation questionnaires-and marital status assessed social functioning. The 8-item Patient Health Questionnaire assessed depressive symptomatology. Multivariable logistic regression models examined associations between social functioning and depressive symptomatology separately and then simultaneously. Data were collected from May 2017 to April 2021.

Results: Social functioning was higher than US-reported averages, and the prevalence of moderate to severe depressive symptomatology was low (20.8%) among primary care populations. In a separate model, lower emotional support, lower instrumental support, and increased social isolation were significantly associated with greater depressive symptomatology (odds ratio [OR] = 1.56, 95% CI, 1.20-2.02; OR = 1.33, 95% CI, 1.01-1.77; and OR = 2.39, 95% CI, 1.81-3.16, respectively). In a simultaneous model, only increased perceived social isolation remained significantly associated with greater depressive symptomatology (OR = 2.24, 95% CI, 1.67-3.00).

Conclusions: Greater social functioning is associated with lower depressive symptom burden in the Black Belt region. Future research into the directionality of this association could assist in the development of interventions to improve regional mental and cardiovascular health.

Trial Registration: ClinicalTrials.gov identifier: NCT02866669.

Utilization, Contributions, and Perceptions of Paid Home Care Workers among Households in New York State

Background and Objectives

While family caregivers have traditionally provided care for older adults with chronic conditions and disabilities, the demand for paid home care workers has increased in the last decade. Although typically thought to assist with personal care, emerging data suggest that paid home care workers assist with a wider scope of care. However, the extent and quality of the care they deliver remains poorly understood.

Research Design and Methods

Using the Empire State Poll, a telephone-based cross-sectional survey of 800 adults in New York State, we characterized the types of care that paid home care workers provided and the perceived value of that care.

Results

Of 800 participants surveyed, 274 reported that they or an immediate family member received care from a paid home care worker (34.3%). Of these, the majority (73.9%) reported that paid home care workers provided emotional and/or medical care, in addition to personal care. In adjusted models, providing emotional and medical care (compared to personal care alone) was associated with nearly a twofold greater perception of importance and experience by the care recipients.

Discussion and Implications

Our findings provide additional data on how paid home care workers contribute to patient care, from the perspective of the care recipient(s). The type of care provided is associated with varying magnitudes of perceived quality. Although limited to New York, these findings have implications for paid home care workers’ training and compensation. Future studies are warranted to investigate the specific factors that mediate the association between types of care provided and their perceived value.

Social Determinants of Health and 30-Day Readmissions Among Adults Hospitalized for Heart Failure in the REGARDS Study

BACKGROUND

It is not known which social determinants of health (SDOH) impact 30-day readmission after a heart failure (HF) hospitalization among older adults. We examined the association of 9 individual SDOH with 30-day readmission after an HF hospitalization.

METHODS AND RESULTS

Using the REGARDS study (Reasons for Geographic and Racial Differences in Stroke), we included Medicare beneficiaries who were discharged alive after an HF hospitalization between 2003 and 2014. We assessed 9 SDOH based on the Healthy People 2030 Framework: race, education, income, social isolation, social network, residential poverty, Health Professional Shortage Area, rural residence, and state public health infrastructure. The primary outcome was 30-day all-cause readmission. For each SDOH, we calculated incidence per 1000 person-years and multivariable-adjusted hazard ratios of readmission. Among 690 participants, the median age was 76 years at hospitalization (interquartile range, 71-82), 44.3% were women, 35.5% were Black, 23.5% had low educational attainment, 63.0% had low income, 21.0% had zip code-level poverty, 43.5% resided in Health Professional Shortage Areas, 39.3% lived in states with poor public health infrastructure, 13.1% were socially isolated, 13.3% had poor social networks, and 10.2% lived in rural areas. The 30-day readmission rate was 22.4%. In an unadjusted analysis, only Health Professional Shortage Area was significantly associated with 30-day readmission; in a fully adjusted analysis, none of the 9 SDOH were individually associated with 30-day readmission.

CONCLUSIONS

In this modestly sized national cohort, although prevalent, none of the SDOH were associated with 30-day readmission after an HF hospitalization. Policies or interventions that only target individual SDOH to reduce readmissions after HF hospitalizations may not be sufficient to prevent readmission among older adults.

Multiple Social Vulnerabilities to Health Disparities and Hypertension and Death in the REGARDS Study

Social vulnerabilities increase the risk of developing hypertension and lower life expectancy, but the effect of an individual's overall vulnerability burden is unknown. Our objective was to determine the association of social vulnerability count and the risk of developing hypertension or dying over 10 years and whether these associations vary by race. We used the REGARDS study (Reasons for Geographic and Racial Differences in Stroke) and included participants without baseline hypertension. The primary exposure was the count of social vulnerabilities defined across economic, education, health and health care, neighborhood and built environment, and social and community context domains. Among 5425 participants of mean age 64±10 SD years of which 24% were Black participants, 1468 (31%) had 1 vulnerability and 717 (15%) had ≥2 vulnerabilities. Compared with participants without vulnerabilities, the adjusted relative risk ratio for developing hypertension was 1.16 (95% CI, 0.99-1.36) and 1.49 (95% CI, 1.20-1.85) for individuals with 1 and ≥2 vulnerabilities, respectively. The adjusted relative risk ratio for death was 1.55 (95% CI, 1.24-1.93) and 2.30 (95% CI, 1.75-3.04) for individuals with 1 and ≥2 vulnerabilities, respectively. A greater proportion of Black participants developed hypertension and died than did White participants (hypertension, 38% versus 31%; death, 25% versus 20%). The vulnerability count association was strongest in White participants (P value for vulnerability count×race interaction: hypertension=0.046, death=0.015). Overall, a greater number of socially determined vulnerabilities was associated with progressively higher risk of developing hypertension, and an even higher risk of dying over 10 years.

Prevalence and Predictors of Home Health Care Workers' General, Physical, and Mental Health: Findings From the 2014‒2018 Behavioral Risk Factor Surveillance System

Objectives. To determine the prevalence and predictors of US home health care workers' (HHWs') self-reported general, physical, and mental health. Methods. Using the 2014-2018 Behavioral Risk Factor Surveillance System, we analyzed the characteristics and health of 2987 HHWs (weighted n = 659 000) compared with 2 similar low-wage worker groups (health care aides and health care support workers, not working in the home). We conducted multivariable logistic regression to determine which characteristics predicted HHWs' health. Results. Overall, 26.6% of HHWs had fair or poor general health, 14.1% had poor physical health, and 20.9% had poor mental health; the prevalence of each outcome was significantly higher than that of the comparison groups. Among HHWs, certain factors, such as low household income, an inability to see a doctor because of cost, and a history of depression, were associated with all 3 aspects of suboptimal health. Conclusions. HHWs had worse general, physical, and mental health compared with low-wage workers not in home health. Public Health Implications. Increased attention to the health of HHWs by public health experts and policymakers is warranted. In addition, targeted interventions appropriate to their specific health needs may be required. (Am J Public Health. 2021;111(12):2239-2250. https://doi.org/10.2105/AJPH.2021.306512).

Ambulatory Care Fragmentation and Subsequent Hospitalization: Evidence From the REGARDS Study

BACKGROUND

Previous studies have suggested that highly fragmented ambulatory care increases the risk of subsequent hospitalization, but those studies used claims only and were not able to adjust for many clinical potential confounders.

OBJECTIVE

The objective of this study was to determine the association between fragmented ambulatory care and subsequent hospitalization, adjusting for demographics, medical conditions, medications, health behaviors, psychosocial variables, and physiological variables.

DESIGN

Longitudinal analysis of data (2003-2016) from the nationwide REasons for Geographic And Racial Differences in Stroke (REGARDS) study, linked to Medicare fee-for-service claims.

SUBJECTS

A total of 12,693 Medicare beneficiaries 65 years and older from the REGARDS study who had at least 4 ambulatory visits in the first year of observation and did not have a hospitalization in the prior year.

MEASURES

We defined high fragmentation as a reversed Bice-Boxerman score above the 75th percentile. We used Cox proportional hazards models to determine the association between fragmentation as a time-varying exposure and incident hospitalization in the 3 months following each exposure period.

RESULTS

The mean age was 70.4 years; 54% were women, and 33% were African American. During the first year of observation, participants with high fragmentation had a median of 8 ambulatory visits with 6 providers, whereas participants with low fragmentation had a median of 7 visits with 3 providers. Over 11.8 years of follow-up, 6947 participants (55%) had a hospitalization. High fragmentation was associated with an increased hazard of hospitalization (adjusted hazard ratio=1.18; 95% confidence interval: 1.12, 1.24).

CONCLUSION

Highly fragmented ambulatory care is an independent risk factor for hospitalization.

Differences in ambulatory care fragmentation by race

BACKGROUND

More fragmented ambulatory care (i.e., care spread across many providers without a dominant provider) has been associated with more subsequent healthcare utilization (such as more tests, procedures, emergency department visits, and hospitalizations) than less fragmented ambulatory care. It is not known if race and socioeconomic status are associated with fragmented ambulatory care.

METHODS

We conducted a longitudinal analysis of data from the REasons for Geographic and Racial Differences in Stroke (REGARDS) study, using the REGARDS baseline visit plus the first year of follow-up. We included participants ≥65 years old, who had linked fee-for-service Medicare claims, and ≥ 4 ambulatory visits in the first year of follow-up. We used Tobit regression to determine the associations between race, annual household income, and educational attainment at baseline and fragmentation score in the subsequent year (as measured with the reversed Bice-Boxerman Index). Covariates included other demographic characteristics, medical conditions, medication use, health behaviors, and psychosocial variables. Additional analyses categorized visits by the type of provider (primary care vs. specialist).

RESULTS

The study participants (N = 6799) had an average age of 73.0 years, 53% were female, and 30% were black. Nearly half had low annual household income (<$35,000) and 41% had a high school education or less. Overall, participants had a median of 10 ambulatory visits to 4 providers in the 12 months following their baseline study visit. Participants in the highest quintile of fragmentation scores had a median of 11 visits to 7 providers. Black race was associated with an absolute adjusted 3% lower fragmentation score compared to white race (95% confidence interval (2% lower to 4% lower; p < 0.001). This difference was explained by blacks seeing fewer specialists than whites. Income and education were not independent predictors of fragmentation scores.

CONCLUSIONS

Among Medicare beneficiaries, blacks had less fragmented ambulatory care than whites, due to lower utilization of specialty care. Future research is needed to determine the effect of fragmented care on health outcomes for blacks and whites.

Association Between Caregiver Strain and Self-Care Among Caregivers With Diabetes

IMPORTANCE

Caregiver strain has been shown to be associated with adverse effects on caregivers' health, particularly among those with cardiovascular disease. Less is known about the association of caregiver strain with health behaviors among caregivers with diabetes, a disease that requires a high degree of self-care.

OBJECTIVE

To examine the association between caregiver strain and diabetes self-care among caregivers with diabetes.

DESIGN, SETTING, AND PARTICIPANTS

This cohort study was conducted between July 13, 2018, and June 25, 2020, using data on 795 US caregivers aged 45 years or older with self-reported diabetes from the Reasons for Geographic and Racial Differences in Stroke (REGARDS) study, which comprised 30 239 Black and White adults 45 years or older throughout the US enrolled from January 2003 to October 2007.

EXPOSURES

Caregiver strain, assessed by self-report in response to the question, "How much of a mental or emotional strain is it to provide this care?" Response options were no strain, some strain, or a lot of (high) strain.

MAIN OUTCOMES AND MEASURES

Diabetes self-care, which was assessed across 4 domains (Mediterranean diet adherence, physical activity, smoking status, and medication adherence), and a composite self-care score summing performance across these domains. The association between caregiver strain and diabetes self-care was examined with multivariable Poisson regression adjusting for demographic, clinical, physical and mental functioning, and caregiving covariates.

RESULTS

Among the 795 caregivers with diabetes included in the study, the mean (SD) age was 63.7 (8.6) years, 469 (59.0%) were women, and 452 (56.9%) were Black individuals. Overall, 146 caregivers (18.4%) reported high caregiver strain. In unadjusted models, high caregiver strain was associated with less physical activity (prevalence ratio [PR], 0.66; 95% CI, 0.45-0.97), low medication adherence (PR, 0.80; 95% CI, 0.68-0.94), and worse self-care (PR, 0.65; 95% CI, 0.44-0.98). In adjusted models, the association between some and high caregiving strain with low medication adherence remained significant (adjusted PR: some strain, 0.88 [95% CI, 0.78-0.99]; high strain, 0.83 [95% CI, 0.69-0.99]).

CONCLUSIONS AND RELEVANCE

In this cohort study of US adult caregivers with diabetes, a high level of strain was associated with low medication adherence. Increased awareness of the prevalence of caregiver strain and potential ramifications on caregivers' self-care appears to be warranted among health care professionals and caregivers.

Healthcare Fragmentation and Incident Acute Coronary Heart Disease Events: a Cohort Study

BACKGROUND

Highly fragmented ambulatory care (i.e., care spread across many providers without a dominant provider) has been associated with excess tests, procedures, emergency department visits, and hospitalizations. Whether fragmented care is associated with worse health outcomes, or whether any association varies with health status, is unclear.

OBJECTIVE

To determine whether fragmented care is associated with the risk of incident coronary heart disease (CHD) events, overall and stratified by self-rated general health.

DESIGN AND PARTICIPANTS

We conducted a secondary analysis of the nationwide prospective Reasons for Geographic and Racial Differences in Stroke (REGARDS) cohort study (2003-2016). We included participants who were ≥ 65 years old, had linked Medicare fee-for-service claims, and had no history of CHD (N = 10,556).

MAIN MEASURES

We measured fragmentation with the reversed Bice-Boxerman Index. We used Cox proportional hazards models to determine the association between fragmentation as a time-varying exposure and adjudicated incident CHD events in the 3 months following each exposure period.

KEY RESULTS

The mean age was 70 years; 57% were women, and 34% were African-American. Over 11.8 years of follow-up, 569 participants had CHD events. Overall, the adjusted hazard ratio (HR) for the association between high fragmentation and incident CHD events was 1.14 (95% confidence interval (CI) 0.92, 1.39). Among those with very good or good self-rated health, high fragmentation was associated with an increased hazard of CHD events (adjusted HR 1.35; 95% CI 1.06, 1.73; p = 0.01). Among those with fair or poor self-rated health, high fragmentation was associated with a trend toward a decreased hazard of CHD events (adjusted HR 0.54; 95% CI 0.29, 1.01; p = 0.052). There was no association among those with excellent self-rated health.

CONCLUSION

High fragmentation was associated with an increased independent risk of incident CHD events among those with very good or good self-rated health.

Heart Failure Training and Job Satisfaction: A Survey of Home Care Workers Caring for Adults with Heart Failure in New York City

Background

Home care workers (HCWs), who include home health aides and personal care attendants, frequently care for adults with heart failure (HF). Despite substantial involvement in HF care, prior qualitative studies have found that HCWs lack training and confidence, which creates challenges for this workforce and potentially for patient care. Herein, we quantified the prevalence of HF training among HCWs and determined its association with job satisfaction.

Methods

We conducted a cross-sectional survey of agency-employed HCWs caring for HF patients across New York, NY from 2018-2019. HF training was assessed with, "Have you received prior HF training?" Job satisfaction was assessed with, "How satisfied are you with your job?" The association between HF training and job satisfaction was determined with robust poisson regression.

Results

323 HCWs from 23 agencies participated; their median age was 50 years (IQR: 37,58), 94% were women, 44% were non-Hispanic Black, 23% were Hispanic, 78% completed ≥ high school education, and 72% were foreign-born. They had been caregiving for a median of 8.5 years (IQR: 4,15) and 73% had cared for 1-5 HF patients. Two-thirds received none/a little HF training and 82% felt satisfied with their job. In a fully adjusted model, HCWs with some/a lot of HF training had 14% higher job satisfaction than those with none/a little HF training (aPR 1.14; 95% CI 1.03-1.27).

Conclusions

The majority of HCWs have not received HF training. HF training was associated with higher job satisfaction, suggesting that HF training programs may improve HCWs' experience caring for this patient population.

Social Determinants of Health and 90-Day Mortality After Hospitalization for Heart Failure in the REGARDS Study

Background

Outcomes following heart failure (HF) hospitalizations are poor, with 90‐day mortality rates of 15% to 20%. Although prior studies found associations between individual social determinants of health (SDOH) and post‐discharge mortality, less is known about how an individuals’ total burden of SDOH affects 90‐day mortality.

Methods and Results

We included participants of the REGARDS (Reasons for Geographic and Racial Differences in Stroke) Study who were Medicare beneficiaries aged ≥65 years discharged alive after an adjudicated HF hospitalization. Guided by the Healthy People 2020 Framework, we examined 9 SDOH. First, we examined age‐adjusted associations between each SDOH and 90‐day mortality; those associated with 90‐day mortality were used to create an SDOH count. Next, we determined the hazard of 90‐day mortality by the SDOH count, adjusting for confounders. Over 10 years, 690 participants were hospitalized for HF at 440 unique hospitals in the United States; there were a total of 79 deaths within 90 days. Overall, 28% of participants had 0 SDOH, 39% had 1, and 32% had ≥2. Compared with those with 0, the age‐adjusted hazard ratio for 90‐day mortality among those with 1 SDOH was 2.89 (95% CI, 1.46–5.72) and was 3.06 (1.51–6.19) among those with ≥2 SDOH. The adjusted hazard ratio was 2.78 (1.37–5.62) and 2.57 (1.19–5.54) for participants with 1 SDOH and ≥2, respectively.

Conclusions

While having any of the SDOH studied here markedly increased risk of 90‐day mortality after an HF hospitalization, a greater burden of SDOH was not associated with significantly greater risk in our population.

Impact of gaps in care for malnourished patients on length of stay and hospital readmission

BACKGROUND

Few published articles have focused on identifying the gaps in care that follow a malnutrition diagnosis and their effects on length of stay (LOS) and 90-day readmission. We hypothesized that length of stay and readmission were associated with these gaps in care.

METHODS

Two registered dietitians retrospectively reviewed charts of 229 adult malnourished patients admitted to a medicine unit to determine their system level gap in care: communication, test delay, or discharge planning. In this secondary analysis, both readmission and length of stay were regressed on each gap in care.

RESULTS

Any system level gap was associated with a greater length of stay (β: 1.48, 95% CI: 1.15-1.91) and specifically the gap related to procedure/testing (β: 2.01, 95% CI: 1.62-2.47) resulted in a two-fold increase in length of stay. There was no association between 90-day readmission and any of the gaps in care.

CONCLUSIONS

There was a strong association between those who had any gap in their care and increased length of stay. Mitigating gaps in care may decrease length of stay and, in turn, result in less risk of infection and could potentially lead to reduced healthcare costs.