Introduction
The United States is currently experiencing demographic changes that reflect a growing proportion of older adults and greater racial/ethnic diversity in the population [
1]. At the same time, the prevalence of older adults living with multiple chronic conditions is increasing [
2,
3] for which access to medical care and supportive environments is essential to ensure health and quality of life [
4]. This development highlights the importance of identifying and ameliorating potential racial/ethnic disparities in healthcare utilization in the United States.
In 2008, one in ten hospitalizations in the United States were identified as potentially avoidable [
5]. Ambulatory care sensitive conditions are a group of conditions—such as asthma, poor glycemic control, or urinary tract infection—for which hospitalization would not have been necessary if timely outpatient treatment had been provided. Hospitalizations with these conditions have been widely used as an indicator of access and quality of primary healthcare in the community [
6‐
8]. A number of previous studies have described markedly greater rates of potentially avoidable hospitalizations among race/ethnic minority groups compared to non-Hispanic (NH) White adults in the US [
8‐
10]. Chronic conditions have furthermore been identified as important risk factors of potentially avoidable hospitalizations and hospitalization costs [
11‐
16]. Yet, whether and how the burden of chronic conditions impact the risk of potentially avoidable hospitalizations across racial groups still needs to be elucidated. To focus prevention and healthcare planning with the current demographic development in the United States, it is important to identify the most important chronic conditions for hospitalizations and potentially avoidable hospitalizations in diverse racial groups. This information can help guide efforts to connect patients to preventive primary care services and whether to focus on specific patient groups to tackle racial disparity in hospitalizations and potentially avoidable hospitalizations.
This study was undertaken to identify chronic conditions preceding hospitalization and potentially avoidable hospitalizations among NH Black and NH White Medicare beneficiaries.
Methods
Data source
We used 2014 Centers for Medicare and Medicaid Services (CMS) claims for participants of the Health and Retirement Study (HRS) linked with sociodemographic data from their most recent, preceding HRS interview. In brief, the HRS is a nationally representative survey of individuals aged ≥ 51 years. The survey has been conducted biennially since 1992 with refreshment samples added every 6 years [
17]. HRS survey data was linked to Medicare administrative data for age-eligible fee-for-service beneficiaries. The study protocol was approved by Oregon Health and Science University—Research Integrity Office Institutional Review Board (STUDY00017034).
Study samples
The selection of participants (
N = 4,993) for the study of hospitalizations was based on enrollment in fee-for-service Medicare Part A and B in 2014, age ≥ 65 years, being NH White or NH Black, and complete information on contributors including the Chronic Condition Data Warehouse (CCW) algorithms. At least three years of enrollment in the Medicare fee-for-service program was required to identify the conditions based on the CCW algorithms [
18,
19]. From this sample, a subset (
N = 1,120) of the study sample with at least one hospitalization was defined for the study of potentially avoidable hospitalizations. The details of the selection of study participants are available in Supplementary Figure
S1.
Variables
Chronic conditions up to the time of HRS interview were the main exposure variables: hypertension, hyperlipidemia, anemia, rheumatoid arthritis/osteoarthritis, ischemic heart disease, heart failure, chronic kidney disease, diabetes, depression, chronic obstructive pulmonary disease (COPD), fibromyalgia/chronic pain/fatigue, atrial fibrillation, acquired hypothyroidism, Alzheimer’s disease and related disorders/senile dementia (ADRD), stroke/transient ischemic attack, anxiety disorders, osteoporosis, cancer, obesity, asthma, pressure and chronic ulcers, acute myocardial infarction, mobility impairments, substance abuse (both drug and alcohol), multiple sclerosis, spinal injury, hip fracture, autism spectrum disorder, bipolar disorder, hepatitis, HIV and schizophrenia. The chronic conditions were identified from linked Medicare beneficiary files and administrative claims using CCW algorithms [
18,
19]. A description of the methodology to ascertain each chronic condition can be found at the CCW website [
18,
19].
Sociodemographic factors were identified in HRS data and include: sex (male as reference); race categorized as NH White and NH Black; education as a continuous variable of years in school; and wealth as a continuous variable with increments of $10,000. Wealth was truncated at its 95th percentile ($2,044,220) and any value of wealth above $2,044,220 was given that value. Age as a continuous variable was included from the CMS claim for participants’ first event in 2014.
The outcomes were hospitalizations and potentially avoidable hospitalizations occurring in 2014 CMS claims. Hospitalizations were identified as a binary indicator (no/yes) by all inpatient hospital claims in the Medicare claims data. Potentially avoidable hospitalization was constructed as a binary indicator (no/yes) based on the definition by Segal et al. 2014 specifically developed for Medicare-Medicaid Eligible Beneficiaries [
20] and widely used in studies of Medicare claims data [
21‐
24]. Our potentially avoidable hospitalization variable used the classification based on both institutional and non-institutional settings that included the following nine inpatient hospital claims with the 9
th revision of International Classification of Disease diagnosis categories: COPD, chronic bronchitis and asthma; congestive heart failure; constipation, fecal impaction, and obstipation; dehydration, volume depletion including acute renal failure and hyponatremia; hypertension and hypotension; poor glycemic control; seizures; urinary tract infection; and weight loss (failure to thrive) and nutritional deficiencies. The ICD-9 codes are available in Supplementary Table
S1. Conditions for institutional settings only were not included in this study.
Statistical models
Descriptive statistics were conducted using means with standard deviations or medians with interquartile ranges (IQR) for continuous variables and frequencies with percentages for categorical variables for each of the outcomes.
Statistical models were constructed in two steps. First, to identify chronic conditions with the highest variable importance in predicting the outcomes, conditional inference random forests were implemented using the R package ‘party’ [
25]. This non-parametric, machine learning method uses bootstrap aggregation to create multiple decision trees, each using a random sample of variables as split candidates, and collects their results. Recursive binary partitioning is conducted by the decision trees to explore the relationship between multiple explanatory variables and one outcome. In this process, a decision tree is constructed by testing the null hypothesis of independence between each variable and the outcome. If the hypothesis cannot be rejected, the algorithm is stopped. The variable with the greatest reduction of heterogeneity in the outcome is selected and a binary split of the variable is performed. Each forest was created using 1,500 trees and we repeated the analyses three times with different random seeds to confirm the robustness of results. The number of potential variables to try at each potential split were set to the default (square root of the number of predictors in the model). From these analyses, we obtained a ranking of variable importance in predicting the outcome. The conditional inference random forest differs from the random forest implemented in the R package ‘randomForest’ by 1) being unbiased when predictor variables are of different types and 2) include a conditional permutation importance measure that helps evaluate the importance of correlated predictor variables [
26]. Second, the top three chronic conditions identified in the ranking were included in multivariable logistic regression analyses adjusted for sociodemographic factors to identify risk estimates (adjusted odds ratios [aORs]) and 95% confidence intervals (CIs) and quantify the association for each of the two outcomes.
Discussion
This study showed differences in the importance of chronic disease preceding hospitalization and potentially avoidable hospitalization between NH White and NH Black Medicare beneficiaries in the United States. We suggest that both outcomes permit indirectly assessing the importance of chronic diseases to two different but related healthcare utilization measures. These outcomes reflect access to care for emergent problems in the case of hospitalization and access to continuity of care in the case of potentially avoidable hospitalization. Similar to previous studies, we identified greater rates of having at least one potentially avoidable hospitalization among NH Black older adults compared to NH White older adults [
8‐
10]. Minority groups may experience greater rates of potentially avoidable hospitalizations because they are subject to greater morbidity and mortality from various chronic conditions while at the same time receive lower quality of care and have lower usage of preventive health care services compared to non-minorities [
27,
29,
29‐
33].
To the best of our knowledge, this study is the first to identify whether different chronic conditions are associated with hospitalization and potentially avoidable hospitalizations among NH White and NH Black Medicare beneficiaries. Potentially avoidable hospitalization have been widely used as an indicator of access and quality of primary healthcare in the community, thus, individuals with the chronic conditions found to be associated with avoidable hospitalizations may not have received optimal treatment for a variety of reasons. Asthma and COPD were identified among the top 3 contributors of potentially avoidable hospitalizations for both NH White and NH Black beneficiaries in addition to heart failure in NH White and fibromyalgia /chronic pain/fatigue in NH Black beneficiaries. Yet, asthma was not statistically significantly associated with potentially avoidable hospitalization in either of the groups when adjusted for potential sociodemographic factors. This finding illustrates that, at least to a certain extent, the same chronic conditions are important for predicting potentially avoidable hospitalizations among NH White and NH Black Medicare beneficiaries. Further, our findings that COPD in NH Black and COPD and heart failure in NH White beneficiaries are risk factors of potentially avoidable hospitalizations are in line with previous findings. A study by Dantas et al. 2016 showed that especially chronic conditions of the circulatory and respiratory systems were risk factors for potentially avoidable hospitalizations in Canada. They furthermore found that the number of chronic conditions and the number of influenced body systems were important risk factors for potentially avoidable hospitalizations [
13]. An older study by Culler et al. 1998 of a representative sample of Medicare beneficiaries showed that fair/poor health, coronary heart disease, myocardial infarction, and diabetes were associated with increased odds of having an potentially avoidable hospitalization whereas hypertension, stroke, and cancer were not [
16]. Our findings of the importance of fibromyalgia /chronic pain/fatigue for potentially avoidable hospitalizations among NH Black beneficiaries may be explained by poor pain treatment among NH Black adults in the United States. A previous review documented extensive racial disparities in pain treatment where minority patients are less likely to have their pain assessed and treated [
34]. The results from our study and the previous review, thus, suggest a need for better addressing chronic pain in ambulatory and outpatient care settings for NH Black older adults.
On a final note, our findings highlight a need of increased focus on improving access to and coordinating care for especially NH Black older adults in primary care settings to prevent potentially avoidable hospitalizations. They further highlight that special focus should be placed on patients with COPD among NH Black and patients with COPD and heart failure patients among NH White to potentially prevent potentially avoidable hospitalizations in old age.
This study has several strengths. Through linkage between HRS and CMS, we were able to identify a comprehensive number of chronic conditions to be investigated as potential top contributors of hospitalization and potentially avoidable hospitalizations. We included information on chronic conditions and hospitalizations from CMS, which are not subject to recall bias. We were able to explore our research questions and the impact of these multiple chronic conditions in machine learning models instead of restricting our hypothesis to a priori knowledge. Most clinical risk prediction models are based on regression models, which are limited by only being able to handle a limited number of potential exposure variables in the same model [
35]. On the other side, conditional inference random forest is limited by not providing estimates of the relationship between an exposure and outcome variable. A strength of this study is sequencing conditional inference random forest to select the top 3 important chronic conditions then applying them in the logistic regression to estimate adjusted ORs of the associations. Finally, potentially avoidable hospitalizations were measured by a definition developed and widely used for Medicare-Medicaid Eligible Beneficiaries [
20‐
24]. The limitations of the study should also be mentioned. First, the observational nature of the study limits the possibility to draw causal conclusions and that unmeasured confounders may exist. Generalizability of the results to all older Americans may be hampered by restricting the study sample to Medicare beneficiaries with ≥ 3 years of enrollment in the Medicare fee-for-service program at baseline who have participated in the HRS. In this regard, we would like to highlight that individuals with Medicare Advantage are not included, which is critical given the proportion of race/ethnic minorities and other healthcare vulnerable groups who elect managed care over fee-for-service. Furthermore, Table
S2 shows sociodemographic characteristics of excluded respondents without 2014 CMS data. Excluded NH white respondents had age and educational level similar to the study sample, whereas they were less often female and had lower wealth. Excluded NH black respondents had similar age to the study sample, whereas they were less often female and had lower wealth and educational level. The selection of the study sample was necessary to identify chronic conditions by the CCW algorithms [
18] and sociodemographic factors available in the HRS (Figure
S1). The conditional random forest package does not yet allow for survey weights and analysis of a subset meeting inclusion criteria. Furtermore, due to small sample sizes, we were unable to analyze Hispanic participants. We were unable to assess other racial categories—such as Asian Americans—due to the inability to identify additional racial groups in the HRS data. The small number of 142 NH Black older adults with at least one hospitalization, of which 42 have a potentiallty avoidable hospitalization, has low power for estimating the associations. Thus, the findings are exploratory and should be confirmed in futute studies with larger sample before conclusions are drawn. We modeled all cause hospitalization and not causes as a direct results of pre-existing chronic conditions. A final concern is that the findings from the conditional inference random forest of hospitalization were not stable as acute myocardial infarction, COPD, and chronic kidney disease were identified as the top 3 contributors of hospitalizations among NH Black in the first two analyses with different seeds, whereas Alzheimer’s disease and related disorders/senile dementia (previously the 4
th ranked) was identified as 3
rd ranked with a different seed.
Open AccessThis article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit
http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (
http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.