Maintenance Note

On Friday, August 31, 2018 at 12:00 pm Eastern Time, JMIR will be completing a server migration to improve site stability and user experience. We expect to be back online Friday, August 31, 2018 at 5:00 pm Eastern Time. Should any problems arise our technical team will be using the weekend to resolve them, and users will be able to access our site by Sunday, September 2, 2018 at 1:00pm Eastern Time.

Who will be affected?

Advertisement

Citing this Article

Right click to copy or hit: ctrl+c (cmd+c on mac)

Published on 12.04.18 in Vol 20, No 4 (2018): April

This paper is in the following e-collection/theme issue:

    Original Paper

    Prevalence of Multiple Chronic Conditions Among Older Adults in Florida and the United States: Comparative Analysis of the OneFlorida Data Trust and National Inpatient Sample

    1School of Information, Florida State University, Tallahassee, FL, United States

    2Department of Health Outcomes and Biomedical Informatics, University of Florida, Gainesville, FL, United States

    3Department of Behavioral Sciences and Social Medicine, Florida State University, Tallahassee, FL, United States

    4Department of Statistics, Florida State University, Tallahassee, FL, United States

    5Department of Psychology, Florida State University, Tallahassee, FL, United States

    *these authors contributed equally

    Corresponding Author:

    Zhe He, PhD

    School of Information

    Florida State University

    271 Shores Bldg

    142 Collegiate Loop

    Tallahassee, FL, 32306

    United States

    Phone: 1 8506445775

    Email:


    ABSTRACT

    Background: Older patients with multiple chronic conditions are often faced with increased health care needs and subsequent higher medical costs, posing significant financial burden to patients, their caregivers, and the health care system. The increasing adoption of electronic health record systems and the proliferation of clinical data offer new opportunities for prevalence studies and for population health assessment. The last few years have witnessed an increasing number of clinical research networks focused on building large collections of clinical data from electronic health records and claims to make it easier and less costly to conduct clinical research.

    Objective: The aim of this study was to compare the prevalence of common chronic conditions and multiple chronic conditions in older adults between Florida and the United States using data from the OneFlorida Clinical Research Consortium and the Healthcare Cost and Utilization Project (HCUP) National Inpatient Sample (NIS).

    Methods: We first analyzed the basic demographic characteristics of the older adults in 3 datasets—the 2013 OneFlorida data, the 2013 HCUP NIS data, and the combined 2012 to 2016 OneFlorida data. Then we analyzed the prevalence of each of the 25 chronic conditions in each of the 3 datasets. We stratified the analysis of older adults with hypertension, the most prevalent condition. Additionally, we examined trends (ie, overall trends and then by age, race, and gender) in the prevalence of discharge records representing multiple chronic conditions over time for the OneFlorida (2012-2016) and HCUP NIS cohorts (2003-2013).

    Results: The rankings of the top 10 prevalent conditions are the same across the OneFlorida and HCUP NIS datasets. The most prevalent multiple chronic conditions of 2 conditions among the 3 datasets were—hyperlipidemia and hypertension; hypertension and ischemic heart disease; diabetes and hypertension; chronic kidney disease and hypertension; anemia and hypertension; and hyperlipidemia and ischemic heart disease. We observed increasing trends in multiple chronic conditions in both data sources.

    Conclusions: The results showed that chronic conditions and multiple chronic conditions are prevalent in older adults across Florida and the United States. Even though slight differences were observed, the similar estimates of prevalence of chronic conditions and multiple chronic conditions across OneFlorida and HCUP NIS suggested that clinical research data networks such as OneFlorida, built from heterogeneous data sources, can provide rich data resources for conducting large-scale secondary data analyses.

    J Med Internet Res 2018;20(4):e137

    doi:10.2196/jmir.8961

    KEYWORDS



    Introduction

    Background

    Chronic conditions (CCs) affect nearly half of the adult population in the United States. The prevalence of some CCs such as hypertension, asthma, cancer, and diabetes has increased over the last a few years [1-3]. Older patients with multiple chronic conditions (MCCs) are often faced with increased health care needs and subsequent higher medical costs, posing significant financial burden to patients, their caregivers, and the health care system.

    Understanding the trends in the prevalence of MCC informs policy makers, health care providers, and payers about chronic disease management and prevention and helps to predict future health care needs [4]. The literature on MCC research mostly uses national claims data or national surveys to estimate the prevalence of MCCs [4-7]. Freid et al [4] presented the estimates of the population aged 45 and older with 2 or more self-reported CCs using the National Health Interview Survey (NHIS) data. They reported that the percentage of adults with MCCs increased in both 45 to 64 years and 65 and older age groups between 1999 and 2010. Ward and Schiller [5] analyzed the prevalence of MCCs among US adults also using the 2010 NHIS data and reported an increasing prevalence of MCCs from 2001 to 2010. Ashman and Beresovsky did an MCC analysis among US adults who visited physician offices, using the National Ambulatory Medical Care Survey data [6]. They found that hypertension was the most prevalent CC that appeared in the top 5 MCC dyads and triads. He et al [7] used the National Health and Nutrition Examination Survey data and a public clinical trial registry—ClinicalTrials.gov—to analyze the gap between the prevalence of MCCs and the clinical trials on the prevalent MCCs. They found that the current and past clinical trials rarely investigate the prevalent MCCs.

    Recent years have witnessed a wide adoption of electronic health record (EHR) systems driven by the Health Information Technology for Economic and Clinical Health (HITECH) Act of 2009 [8]. By 2015, over 90% of nonfederal acute care hospitals adopted a certified EHR [9]. By the end of 2017, about 90% of the office-based physicians have been using EHRs in the Unites States [10]. With public health reporting as part of the meaningful use criteria for hospitals to receive the incentive payments of the HITECH Act, EHRs have been recognized as an important data source for public health surveillance [11] (especially in chronic disease surveillance [12-14]), cohort identification for clinical studies [15], and disease-risk prediction [16]. The advantage of using EHRs over survey data is multifaceted. First, EHRs have fine-grained clinical data that are rarely collected and reported in the survey or claims data. Second, EHRs contain longitudinal patient data, whereas survey data mostly provide merely a snapshot of the health conditions for a person. However, as EHR data only contain patients who paid a visit to the health care facilities, they may not be as representative of the national population as the survey data. Therefore, it is necessary to investigate the extent to which EHR data can represent the broader population to inform researchers who are using EHRs for public health and chronic disease surveillance. Recently, Perlman et al created an EHR-based public health surveillance system in New York City [14]. They compared the CC estimates generated in this system with those from a population-based survey in New York and found that diabetes, hypertension, smoking, and obesity prevalence was close to the survey results, but depression and influenza vaccination estimates were substantially lower than the survey-based estimates [14].

    The last few years have witnessed an increasing number of clinical research networks focused on building large collections of clinical datasets from EHRs and claims to offer a collaborative environment for researchers across disparate organizations. It is anticipated that the analysis of such data will lead to advances in medical knowledge, progress in health care delivery, and improvements in population health [17-21]. One notable example is the National Patient-Centered Clinical Research Network (PCORnet) [17,22], funded by the Patient-Centered Outcomes Research Institute (PCORI). PCORnet comprises a coordinating center and 33 partner networks, including 13 Clinical Data Research Networks (CDRNs) and 20 Patient-Powered Research Networks. PCORnet is “designed to make it faster, easier, and less costly to conduct clinical research than is now possible by harnessing the power of large amounts of health data and patient partnerships” [22]. It is a national “network of networks” that routinely collects data from a variety of health care organizations, including hospitals, community clinics, health plans, and national data registries (eg, cancer registries and vital statistics).

    PCORnet empowers individuals and organizations to use this big dataset to answer practical questions that help patients, clinicians, and other stakeholders make informed health care decisions. For example, PCORnet provides an invaluable cohort discovery service that proves particularly useful for identifying cohorts of a variety of health conditions, especially for rare diseases. With such a large collection of electronic patient data, PCORnet can effectively support large-scale randomized clinical trials, comparative effectiveness research studies, and longitudinal observational studies. EHRs such as those warehoused in CDRNs have been widely used for comparative effectiveness analysis [23-26], cohort identification [27-29], and public health surveillance studies [25,30,31]. However, it is not yet known the extent to which the population in these CDRNs such as OneFloridais is representative of the national population. This is an important metric that needs to be examined to understand the comprehensiveness of the OneFlorida population now and to improve the interpretability and generalizability of the OneFlorida data and the reproducibility of the aforementioned studies.

    Florida has the largest elderly population in the United States. OneFlorida is one of the 13 CDRNs contributing to the national PCORnet [32]. The OneFlorida Data Trust is a secure centralized data repository that integrates various data sources from contributing organizations in the OneFlorida research consortium, including 22 hospitals and 914 community-based clinical practices that provide care to 48% of Floridians. As of June 2017, the Data Trust contains 10.9 million patient records including data from partners’ EHR systems, as well as claims data from Florida Medicaid. Ultimately, the Data Trust will include claims data for Florida Medicare beneficiaries, Florida Vital Statistics records, and Florida Cancer Data System records. The OneFlorida Data Trust employs the PCORNet Common Data Model (CDM) version 3.1 [33], which uses standard vocabularies to encode diagnoses (ie, International Classification of Diseases, ICD), procedures (ie, ICD procedure codes, Current Procedural Terminology, and Healthcare Common Procedure Coding System codes), laboratory observations (ie, Logical Observation Identifiers Names and Codes), and medications (ie, RxNorm and National Drug Code). The OneFlorida and PCORnet data only contains Health Insurance Portability and Accountability Act limited data, for which we obtained permission to use. Throughout this paper, OneFlorida refers to the inpatient data extracts used to conduct this analysis unless otherwise noted.

    Objective

    The purpose of this study is to estimate and compare the prevalence of common CCs and MCCs among older adults in Florida and United States from the OneFlorida Data Trust and a national data source—the National Inpatient Sample (NIS) from the Healthcare Cost and Utilization Project (HCUP) of Agency for Healthcare Research and Quality [34]. The NIS is a comprehensive source of inpatient hospital data in the United States. As NIS contains only the inpatient data, we also used the inpatient EHR records in the OneFlorida Data Trust to estimate Florida population. For this paper we define MCC as 2 or more CCs according to the Center for Medicare and Medicaid Services (CMS) algorithm [35].

    We formulated 2 research questions (RQs) in this study:

    RQ1: What is the prevalence of common CCs in hospital discharge records for older adults in the OneFlorida Data Trust inpatient data and how does it compare with unweighted national estimates from the HCUP NIS?
    RQ2: Are the 10 most common CCs and the prevalence of MCC in hospital discharge records for older adults in the OneFlorida Data Trust consistent with the unweighted HCUP NIS national population?

    Methods

    Data Collection and Preparation

    OneFlorida inpatient discharge records for 2012 to 2016 for 22 CCs were identified using the CMS Chronic Condition Warehouse (CCW) algorithm [35]. We included records with an admission source of home, another facility, or the emergency department. The 2013 discharge records were used for the cross-sectional analysis and the 2012 to 2016 records were used for a longitudinal comparison.

    NIS is the largest publicly available all-payer inpatient health care database in the United States. Unweighted, it contains 7 million hospital discharge records each year and the weighted sample represents 25 million discharges. Beginning within the 2012 data year, the NIS approximates a 20% stratified sample of all discharges from US community hospitals, excluding rehabilitation and long-term acute-care hospitals. The 2013 NIS file was used for our cross-sectional analysis, and the 2003 to 2013 data were used for the longitudinal analysis. NIS includes information on all patients, including individuals covered by Medicare, Medicaid, or private insurance, uninsured. Researchers and policy makers use NIS to make national estimates of inpatient health care utilization [36], access to care [37], inpatient charges [36,38,39], quality of hospital care [37], and outcomes [39,40].

    Figure 1 illustrates the process of data preparation and analysis. As the first step, we identified patients with CC using the CMS CCW algorithm [35]. The CMS CCW algorithm identifies cases for 27 condition categories using the criteria, such as (1) a validated list of ICD-9-CM and ICD-10 diagnosis codes, (2) the number of discharge record occurrences with diagnosis codes meeting the case definition within a year, (3) the number of consecutive years with confirming diagnoses in order to identify an individual case within a specific CC category in a given year to identify 27 conditions, and (4) the source type of service. We excluded 2 algorithm conditions that do not use inpatient records for case identification for cataracts or glaucoma, because those conditions are typically not associated with inpatient care. We modified the algorithm criteria for 7 other conditions, which were (1) rheumatoid arthritis and osteoarthritis, (2) chronic kidney disease, (3) heart failure, (4) diabetes, (5) Alzheimer disease, (6) Alzheimer disease and related conditions, and (7) ischemic heart disease. These 7 conditions require 2 or 3 consecutive years with the diagnosis to meet the case criteria or in the case of rheumatoid arthritis or osteoarthritis, 2 diagnoses within a year. Due to privacy concerns, the NIS does not assign unique patient identifiers that can be tracked across facilities or time. Therefore, we modified the criteria for those 7 conditions and identified cases based on a single inpatient discharge record. We limited the analysis to persons aged 65 years or older for the 25 remaining conditions defined by the CMS algorithm [35].

    We identified older adults as those who were above 65 years at the time of inpatient discharge in both data sources. We stratified our analysis by age group, namely, 65 to 74, 75 to 84, and 85 and above. Besides age, we also extracted the gender and race or ethnicity variables of the patients. For OneFlorida analysis, we generated 2 datasets, one for a cross-sectional analysis (2013) and the other for a longitudinal analysis with data from all the years currently available in the OneFlorida Data Trust (2012-2016). For HCUP NIS, we used the 2013 data for the cross-sectional analysis and 2003 to 2013 data for the longitudinal analysis. The decision of using different year range for OneFlorida Data Trust and HCUP NIS was made based on the availability of the data and the richness of the analysis.

    Figure 1. Workflow of data preparation and analysis. CRC: Clinical Research Consortium. CMS: Center for Medicare and Medicaid Services. NIS: National Inpatient Sample. MCC: multiple chronic conditions.
    View this figure

    Nevertheless, both OneFlorida Data Trust and HCUP NIS have the 2012 to 2013 data.

    OneFlorida Data Trust uses the PCORNet CDM version 3.1, which is a relational schema. The data are stored in a Microsoft SQL server hosted by the University of Florida Health Science Center. We included patients who had either direct inpatient admissions or emergency-to-inpatient admissions. The HCUP NIS data were released in the SAS format. We preprocessed the HCUP SAS datasets and loaded them into a Microsoft SQL server.

    Data Analysis

    The analysis included descriptive statistics for the 25 individual conditions and MCC in 3 analytic files, that is (1) the OneFlorida data for the year 2013 only (OneFlorida 2013), (2) the NIS data for the year 2013 only (NIS 2013), and (3) the OneFlorida data for 2012 to 2016 (OneFlorida 2012-2016). Chi-square tests were used to examine group differences in the proportions of interest across the 3 data files.

    We first analyzed the basic demographic characteristics of the older adults in the two 2013 datasets and the OneFlorida 2012 to 2016 data. Then we analyzed the prevalence of each of the 25 CCs for each of the 3 datasets. We did a deep dive, stratified the analysis, of the older adults with hypertension, the most prevalent condition. The prevalence of hypertension in the 24 age-gender-race-ethnicity strata was compared across the 3 datasets. We also examined the number of conditions per hospital record in each dataset for 2013 and further stratified the prevalence of patients with MCCs in 2013 by gender and race or ethnicity.

    Additionally, we examined trends (ie, overall trends and then by age, race, and gender) in the prevalence of discharge records representing MCCs across time for the OneFlorida and NIS cohorts. Pearson correlation coefficient was computed to compare the MCC trends stratified by age group, sex, and race or ethnicity.


    Results

    Basic Characteristics

    Table 1 shows the basic characteristics of the older adult populations in the OneFlorida 2013 data, OneFlorida 2012 to 2016 data, and the HCUP NIS 2013 datasets. The average age of older adults in the OneFlorida 2013 and 2012 to 2016 data was similar (2-tailed t test, degrees of freedom=63435.3226, P>.05). The older adults in OneFlorida 2013 were slightly younger than those in NIS 2013. There were more elderly female patients than elderly male patients across all 3 datasets. OneFlorida 2012 to 2016 had a statistically significantly higher percentage of Hispanics, non-Hispanic (NH) blacks, and a lower percentage of non-Hispanic whites and Asian or Pacific Islanders than the NIS 2013 (chi-square statistics 72587091891.83, P<.001).

    Prevalence of Chronic Conditions

    The rankings of the top 10 prevalent conditions were the same across the 3 datasets. These conditions were hypertension, hyperlipidemia, ischemic heart disease, diabetes, anemia, chronic kidney disease, atrial fibrillation, heart failure, chronic obstructive pulmonary disease, and RA. However, there were differences in the prevalence of each disease between the NIS and OneFlorida data. Comparing the NIS and OneFlorida data, one can observe that a higher percentage of older adults in OneFlorida had hypertension (80.97% vs 76.32%), hyperlipidemia (52.42% vs 45.94%), and diabetes (35.32% vs 33.93%) than in NIS; whereas a higher percentage of older adults in NIS had chronic kidney disease (33.22% vs 31.24%) and heart failure (25.36% vs 19.77%). The prevalence of arthritis was 43% in male and 54% in female respondents in a recent national survey of older adults (65 and older) with self-reported chronic medical conditions in 2013 to 2014 [3]. The numbers were nearly twice the prevalence of such a condition in the inpatient clinical data reported in Table 2. This likely reflects the fact that people with arthritis were mostly treated in outpatient settings and thus diagnosis of arthritis is irrelevant to most inpatient discharges.

    Prevalence of Hypertension by Gender, Age Group, and Race or Ethnicity

    Table 3 shows the prevalence of hypertension in older adults stratified by sex, age group, race and ethnicity in the NIS 2013, the OneFlorida 2013, and the pooled OneFlorida 2012 to 2016 data. Hypertension was chosen because it was the condition with the highest prevalence among the older persons we studied. The largest differences in the estimates between the 2013 files (OneFlorida and NIS) was about 3% for NH white females aged 85 years and older, and NH white males aged 65 to 74 years. We observed differences of more than 1% for females in the following 4 strata—NH black aged 65 to 74 years, NH white aged 65 to 74 years, NH white aged 75 to 84 years, and NH white aged 85 years and older. Among males, differences of greater than 1% were observed for the strata except for NH white aged 75 to 84 years. Estimates between OneFlorida 2013 and the pooled OneFlorida 2012 to 2016 data were largely similar with some increases in OneFlorida 2012 to 2016 data for hypertension prevalence, perhaps reflecting the increasing trends associated with obesity and sedentary life styles.

    Prevalence of Multiple Chronic Conditions

    Figure 2 illustrates the percentage of the population with one or more CCs, which is, MCCs in older adults in the HCUP NIS and OneFlorida for 2013. The 3 datasets exhibited similar characteristics. Out of the 25 CCs, more than 18% older adults had 4 conditions. More than 65% older adults had 4 or more conditions. Persons with MMCs were very prevalent among older Americans.

    Prevalence of Multiple Chronic Conditions by Gender

    Figure 3 illustrates the prevalence of MCC stratified by sex. With respect to the number of MCCs, male and female older adults did not exhibit notable difference in both the OneFlorida and NIS data. No statistical tests were performed to test the statistical difference among the groups. This contrasted with the population aged 18 to 64 years in which women had a higher prevalence of MCCs.

    Prevalence of Multiple Chronic Conditions by Race or Ethnicity

    Figure 4 illustrates the prevalence of MCCs by race or ethnicity. It appears that the distribution of records with one or more CCs were similar among race or ethnicity groups. Note that even though Hispanic was overrepresented and Asian was underrepresented in OneFlorida, their MCC distribution within each race or ethnicity was similar to the NIS.

    Table 1. Descriptive statistics for the National Inpatient Sample and OneFlorida patient datasets of older adults. HCUP NIS: Healthcare Cost and Utilization Project National Inpatient Sample.
    View this table
    Table 2. Prevalence of the 25 chronic conditions in the 2013 National Inpatient Sample, the 2013 OneFlorida, and the pooled 2012 to 2016 OneFlorida data. HCUP NIS: Healthcare Cost and Utilization Project National Inpatient Sample.
    View this table
    Table 3. Prevalence of hypertension stratified by gender-age-racial or ethnic groups in the 2013 Healthcare Cost and Utilization Project (HCUP) National Inpatient Sample (NIS), the 2013 OneFlorida, and the pooled 2012 to 2016 OneFlorida data.
    View this table
    Figure 2. Number of conditions in older adults in the Healthcare Cost and Utilization Project National Inpatient Sample (NIS) 2013 and OneFlorida2013 data.
    View this figure
    Figure 3. Prevalence of multiple chronic conditions in older adults by gender in Healthcare Cost and Utilization Project National Inpatient Sample (NIS) 2013 and OneFlorida 2013.
    View this figure
    Figure 4. Prevalence of multiple chronic conditions in older adults by race and ethnicity groups in Healthcare Cost and Utilization Project National Inpatient Sample (NIS) 2013 and OneFlorida 2013 data. NH: non-Hispanic.
    View this figure

    Prevalence of Multiple Chronic Conditions by Pairs of Conditions

    Table 4 shows the prevalence of the 10 most common pairs of co-occurring chronic conditions. Even though the prevalent MCCs of 2 conditions were the same in both OneFlorida and NIS cohorts, their rankings were slightly different. OneFlorida cohort had a higher percentage of patients with anemia and hypertension than the NIS cohort (32.17% vs 25.79%). OneFlorida cohort had a slightly higher percentage of older adults with atrial fibrillation and hypertension than the NIS cohort (22.72% vs 19.88%). The most prevalent MCCs of 2 conditions among the 3 datasets were—hyperlipidemia and hypertension, hypertension and ischemic heart disease, diabetes and hypertension, chronic kidney disease and hypertension, anemia and hypertension, and hyperlipidemia and ischemic heart disease.

    Trends in Multiple Chronic Conditions

    The following 4 figures (Figures 5-8) present a longitudinal examination of the number of discharges reflecting 2 or more conditions for the period 2012 to 2016 for the OneFlorida data and 2003 to 2013 for the NIS data. In Figure 5, the overall prevalence of 2 or more CCs raised steadily from approximately 66% in 2003 to approximately 83% in 2013 in the NIS data. The OneFlorida data began in 2012 at approximately 81% prevalence of MCC and rose to approximately 84% by 2016. Both slopes showed a monotonic increasing trend in the prevalence of MCCs.

    The slope of the MCC prevalence by gender in Figure 6 appeared to be very similar to the overall slope in Figure 5. The slopes for males and females in the NIS data were parallel with 1% to 2% difference for males and females and ultimately converged at approximately 84% by 2013. Pearson correlation coefficient showed a strong positive correlation between male and female older adults with an R value of .9966. The lines for OneFlorida data for males and females were nearly coincident and appeared to continue the slope of the NIS data.

    In Figure 7, the prevalence of MCC by age group is presented for NIS and OneFlorida data. The NIS slopes for the 3 age groups were parallel through 2013. Pearson correlation coefficient showed a strong positive correlation among the 3 age groups—the R value between NIS 65 to 74 years age group and 75 to 84 years age group was .9972; the R value between NIS 65 to 74 years age group and NIS over 85 years age group was .9961. Nevertheless, there was about an 8-percentage point difference between the youngest age group (65-74 years) and the middle age group (75-84 years). The oldest age group (over 85 years) appeared to be about 4 percentage points higher than the 75 to 84 years age group throughout the time range. Similar differences were seen between the parallel slopes for OneFlorida data, although the 85 years and over group was trending somewhat higher as compared with the same age group in the NIS.

    Finally, in Figure 8, we present the prevalence of MCCs by racial-ethnic groups. The general trend was the same as seen in Figures 5-7. The non-Hispanic black and non-Hispanic white groups ran parallel with the black group averaging about 2 percentage points higher. The Pearson correlation coefficient showed a strong positive correlation between non-Hispanic black and non-Hispanic white groups with an R value of .9959. The Hispanic group and the Asian and Pacific Islander group, both averaged a bit lower than the non-Hispanic white population, but there was more volatility probably due to smaller sample size. This was particularly true for the OneFlorida data.

    Table 4. Prevalence of the 10 most common pairs of co-occurring chronic conditions. HCUP NIS: Healthcare Cost and Utilization Project National Inpatient Sample.
    View this table
    Figure 5. Trends of multiple chronic conditions in older adults for the Healthcare Cost and Utilization Project (HCUP) National Inpatient Sample (NIS; 2003-2013) and OneFlorida (2012-2016). The denominator in the prevalence is the total number of older adults with at least one of the 25 conditions in each year.
    View this figure
    Figure 6. Trends of multiple chronic conditions in older adults by gender for the Healthcare Cost and Utilization Project (HCUP) National Inpatient Sample (NIS) (2003-2013) and OneFlorida (2012-2016). The denominator in the prevalence is the total number of older male or female with at least one of the 25 conditions in each year.
    View this figure
    Figure 7. Trends of multiple chronic conditions in older adults by age group for the Healthcare Cost and Utilization Project (HCUP) National Inpatient Sample (NIS) (2003-2013) and OneFlorida (2012-2016). The denominator in the prevalence is the total number of older adults in each age group with at least one of the 25 conditions in each year.
    View this figure
    Figure 8. Trends of multiple chronic conditions in older adults by racial-ethnic groups for the Healthcare Cost and Utilization Project (HCUP) National Inpatient Sample (NIS; 2003-2013) and OneFlorida (2012-2016). The denominator in the prevalence is the total number of older adults in each racial-ethnic group with at least one of the 25 conditions in each year.
    View this figure

    Discussion

    Principal Findings

    The main objective of our study was to compare the prevalence of common CCs and MCCs in older adults in the Florida and US national population using the OneFlorida Data Trust and the NIS of HCUP. The results showed that CCs and MCCs were prevalent in older adults, both nationally and in the Florida population. The most prevalent CCs were the same for older adults in the OneFlorida Data Trust and HCUP NIS. For hypertension, the largest differences in the estimates between the 2013 OneFlorida Data Trust and NIS were about merely 3% for non-Hispanic white females 85 years and older and males 65 to 74 years old. Regarding the number of MCCs, OneFlorida Data Trust and NIS did not exhibit any notable difference with respect to gender and race or ethnicity. The most prevalent MCCs of 2 CCs were also the same for OneFlorida 2013, NIS 2013, and OneFlorida 2012 to 2016. With regard to the MCC trends, the slopes of the increasing trend in the number of discharges reflecting 2 or more conditions appeared quite similar in both data sources. With respect to age group, the oldest age group (over 85 years of age) appeared to be about 4 percentage points higher than the 75 to 84 years age group and 12 percentage points higher than the 65 to 75 years age group throughout the time range. Even though slight differences were observed, similar estimates of prevalence of CCs and MCCs across OneFlorida Data Trust and NIS showed that large clinical research networks such as OneFlorida provide rich data resources for conducting large-scale secondary data analyses.

    Although the MCC prevalence presented in this study is generalizable to the older US adults in the noninstitutionalized national population, the use of OneFlorida Data Trust and the HCUP NIS has limitations. OneFlorida and NIS both only captured the conditions that were confirmed by a doctor or health professionals in inpatient settings, potentially leading to the underrepresentation of conditions that remain undiagnosed or were not recorded in the inpatient care (eg, arthritis [3]). Many uninsured adults would not get into these databases until 65 years of age when they become eligible for Medicare. Undocumented immigrants would never make it into Medicare. For example, the prevalence of arthritis reported in a self-reported national survey almost doubles the prevalence of arthritis in the inpatient clinical data reported in Table 2. Of the conditions captured, we only used the CCW algorithm from the Centers for Medicare and Medicaid Services (CMS) and considered a single occurrence of the diagnosis code of a particular condition when identifying patients who had such a condition. There might be false positive cases included in the analysis. Furthermore, although the OneFlorida Clinical Research Consortium [41] covers care for approximately 48% Floridians, the consortium is missing representations from a few of the key health care markets in Florida, such as Tampa, and cities in the Florida panhandle area. Moreover, the prevalence of CCs might be overestimated for Florida, as there might be duplicated patient records across the different health care organizations in the OneFlorida consortium. For example, EHRs from health care providers and claims data from payers can have records for the same patient. In addition, the same patient can seek care in different health care organizations in the network. Thus, linking related data and resolving duplicates in a clinical research network is a significant task in improving the quality of a dataset. In our recent effort, we have linked and deduplicated patient records across 2 of the data sources in the OneFlorida consortium—University of Florida Health system and Florida Medicaid. We eliminated 430,106 duplicate patient records across these 2 sources, which is approximately 6.4% of the Florida Medicaid population.

    Our study confirmed the previous literature [5] and showed the increasing trend in the prevalence of MCCs among the older US adults. We also showed that the characteristics of the patient population in these clinical research networks such as OneFlorida are comparable to national-level sample data. Furthermore, these clinical research networks have integrated fine-grained details of the patients (eg, encounters, procedures, diagnoses, medications, lab results, as well as patient-reported outcomes) from multiple health care organizations, which can provide a more complete picture of the patients’ health traits. Enabled by clinical research networks such as OneFlorida, large-scale secondary data analyses can be conducted to discover novel findings in biomedical research, such as sophisticated relationships among diseases, medications, vital signs, adverse events, and outcomes.

    Implication and Future Directions

    The OneFlorida Data Trust is the informatics infrastructure that supports pragmatic trials, comparative effectiveness research, implementation science, and other research in the OneFlorida Clinical Research Consortium. The most key research functions supported by OneFlorida and PCORnet include cohort discovery and participant enrollment, recognizing the barriers in identifying and recruiting research participants for clinical research studies, especially for rare diseases. Furthermore, the population representativeness of clinical research has long been a concern [42]. Particularly, older adults are widely reported to be underrepresented in clinical studies across major medical conditions such as cardiovascular diseases [43,44], cancer [45,46], dementia [47], and diabetes [48,49]. Due to the lack of evidence in the clinical practice guideline in treating older adults with MCCs, it is imperative to generate such evidence by involving older adults with normal age-related organ impairment and comorbid conditions that may not interact with the treatment under study. However, older adults are often unfairly excluded by restrictive eligibility criteria in clinical studies [46,50]. Meanwhile, MCCs are most prevalent in the Medicare population. Persons with MCCs are at an increased risk of mortality, morbidity, hospitalization, high medical costs, and adverse events [51]. In order to understand how older adults with MCCs are represented in clinical trials, it is important to understand the prevalence of MCCs in older adults. In future work, we will use laboratory test results and medications to enhance the sensitivity and specificity of case assignment for some conditions. We will also compare the outpatient data of OneFlorida Data Trust with the national outpatient databases such as the Nationwide Emergency Department Sample.

    Acknowledgments

    This work was supported by the Institute for Successful Longevity of the Florida State University. The work was also supported in part by the OneFlorida Clinical Data Network, funded by PCORI #CDRN-1501-26692, in part by the OneFlorida Cancer Control Alliance, funded by the Florida Department of Health’s James and Esther King Biomedical Research Program #4KB16, and in part by the University of Florida Clinical and Translational Science Institute, which is supported in part by the NIH National Center for Advancing Translational Sciences under award number UL1TR001427. The content is solely the responsibility of the authors and does not necessarily represent the official views of PCORI, its Board of Governors or Methodology, the OneFlorida Clinical Research Consortium, the University of Florida’s Clinical and Translational Science Institute, the Florida Department of Health, or the National Institutes of Health.

    Conflicts of Interest

    None declared.

    References

    1. Schneider KM, O'Donnell BE, Dean D. Prevalence of multiple chronic conditions in the United States' Medicare population. Health Qual Life Outcomes 2009 Sep 08;7:82 [FREE Full text] [CrossRef] [Medline]
    2. Vogeli C, Shields AE, Lee TA, Gibson TB, Marder WD, Weiss KB, et al. Multiple chronic conditions: prevalence, health consequences, and implications for quality, care management, and costs. J Gen Intern Med 2007 Dec;22(Suppl 3):391-395 [FREE Full text] [CrossRef] [Medline]
    3. Federal Interagency Forum on Aging-Related Statistics. Older Americans 2016 - Key Indicators of Well-Being Federal Interagency Forum on Aging-Related Statistics 2016   URL: https://agingstats.gov/docs/LatestReport/Older-Americans-2016-Key-Indicators-of-WellBeing.pdf [WebCite Cache]
    4. Freid VM, Bernstein AB, Bush MA. Multiple chronic conditions among adults aged 45 and over: trends over the past 10 years. NCHS Data Brief 2012 Jul(100):1-8 [FREE Full text] [Medline]
    5. Ward BW, Schiller JS. Prevalence of multiple chronic conditions among US adults: estimates from the National Health Interview Survey, 2010. Prev Chronic Dis 2013 Apr 25;10:E65 [FREE Full text] [CrossRef] [Medline]
    6. Ashman JJ, Beresovsky V. Multiple chronic conditions among US adults who visited physician offices: data from the National Ambulatory Medical Care Survey, 2009. Prev Chronic Dis 2013 Apr 25;10:E64 [FREE Full text] [CrossRef] [Medline]
    7. He Z, Charness N, Bian J, Hogan WR. Assessing the comorbidity gap between clinical studies and prevalence in elderly patient populations. IEEE EMBS Int Conf Biomed Health Inform 2016 Feb;2016:136-139 [FREE Full text] [CrossRef] [Medline]
    8. Adler-Milstein J, Jha AK. HITECH Act drove large gains in hospital electronic health record adoption. Health Aff (Millwood) 2017 Aug 01;36(8):1416-1422. [CrossRef] [Medline]
    9. Henry J, Pylypchuk Y, Searcy T, Patel V. The Office of the National Coordinator for Health Information Technology. 2016. Adoption of Electronic Health Record Systems among U.S. Non-Federal Acute Care Hospitals: 2008-2015   URL: https:/​/dashboard.​healthit.gov/​evaluations/​data-briefs/​non-federal-acute-care-hospital-ehr-adoption-2008-2015.​php [WebCite Cache]
    10. PracticeFusion. 2017. HR adoption rates: 20 must-see stats   URL: https://www.practicefusion.com/blog/ehr-adoption-rates/ [WebCite Cache]
    11. Paul MM, Greene CM, Newton-Dame R, Thorpe LE, Perlman SE, McVeigh KH, et al. The state of population health surveillance using electronic health records: a narrative review. Popul Health Manag 2015 Jun;18(3):209-216. [CrossRef] [Medline]
    12. Klompas M, Cocoros NM, Menchaca JT, Erani D, Hafer E, Herrick B, et al. State and local chronic disease surveillance using electronic health record systems. Am J Public Health 2017 Sep;107(9):1406-1412. [CrossRef] [Medline]
    13. Birkhead GS. Successes and continued challenges of electronic health records for chronic disease surveillance. Am J Public Health 2017 Sep;107(9):1365-1367. [CrossRef] [Medline]
    14. Perlman SE, McVeigh KH, Thorpe LE, Jacobson L, Greene CM, Gwynn RC. Innovations in population health surveillance: using electronic health records for chronic disease surveillance. Am J Public Health 2017 Jun;107(6):853-857. [CrossRef] [Medline]
    15. Weng C, Batres C, Borda T, Weiskopf NG, Wilcox AB, Bigger JT, et al. A real-time screening alert improves patient recruitment efficiency. AMIA Annu Symp Proc 2011;2011:1489-1498 [FREE Full text] [Medline]
    16. Boland MR, Shahn Z, Madigan D, Hripcsak G, Tatonetti N. Birth month affects lifetime disease risk: a phenome-wide method. J Am Med Inform Assoc 2015 Sep;22(5):1042-1053 [FREE Full text] [CrossRef] [Medline]
    17. Fleurence RL, Curtis LH, Califf RM, Platt R, Selby JV, Brown JS. Launching PCORnet, a national patient-centered clinical research network. J Am Med Inform Assoc 2014;21(4):578-582 [FREE Full text] [CrossRef] [Medline]
    18. Curtis LH, Brown J, Platt R. Four health data networks illustrate the potential for a shared national multipurpose big-data network. Health Aff (Millwood) 2014 Jul;33(7):1178-1186. [CrossRef] [Medline]
    19. Collins FS, Hudson KL, Briggs JP, Lauer MS. PCORnet: turning a dream into reality. J Am Med Inform Assoc 2014;21(4):576-577 [FREE Full text] [CrossRef] [Medline]
    20. Corley DA, Feigelson HS, Lieu TA, McGlynn EA. Building data infrastructure to evaluate and improve quality: PCORnet. J Oncol Pract 2015 May;11(3):204-206 [FREE Full text] [CrossRef] [Medline]
    21. Fleurence RL, Beal AC, Sheridan SE, Johnson LB, Selby JV. Patient-powered research networks aim to improve patient care and health research. Health Aff (Millwood) 2014 Jul;33(7):1212-1219. [CrossRef] [Medline]
    22. pcornet. PCORnet, the National Patient-Centered Clinical Research Network   URL: http://pcornet.org/ [accessed 2017-09-13] [WebCite Cache]
    23. Hernandez AF, Fleurence RL, Rothman RL. The ADAPTABLE Trial and PCORnet: shining light on a new research paradigm. Ann Intern Med 2015 Oct 20;163(8):635-636. [CrossRef] [Medline]
    24. Lin KJ, Singer DE, Glynn RJ, Murphy SN, Lii J, Schneeweiss S. Identifying patients with high data completeness to improve validity of comparative effectiveness research in electronic health records data. Clin Pharmacol Ther 2018:- (forthcoming)(forthcoming). [CrossRef] [Medline]
    25. Reams C, Powell M, Edwards R. State synergies and disease surveillance: creating an electronic health data communication model for cancer reporting and comparative effectiveness research in kentucky. EGEMS (Wash DC) 2014;2(2):1064 [FREE Full text] [CrossRef] [Medline]
    26. Fiks AG, Grundmeier RW, Steffes J, Adams WG, Kaelber DC, Pace WD, Comparative Effectiveness Research Through Collaborative Electronic Reporting (CER2) Consortium. Comparative effectiveness research through a collaborative electronic reporting consortium. Pediatrics 2015 Jul;136(1):e215-e224 [FREE Full text] [CrossRef] [Medline]
    27. Mo H, Thompson WK, Rasmussen LV, Pacheco JA, Jiang G, Kiefer R, et al. Desiderata for computable representations of electronic health records-driven phenotype algorithms. J Am Med Inform Assoc 2015 Nov;22(6):1220-1230 [FREE Full text] [CrossRef] [Medline]
    28. Vassy JL, Ho Y, Honerlaw J, Cho K, Gaziano JM, Wilson PW, et al. Yield and bias in defining a cohort study baseline from electronic health record data. J Biomed Inform 2018 Feb;78:54-59. [CrossRef] [Medline]
    29. Abrahão MT, Nobre MR, Gutierrez MA. A method for cohort selection of cardiovascular disease records from an electronic health record system. Int J Med Inform 2017 Dec;102:138-149. [CrossRef] [Medline]
    30. Sidebottom AC, Johnson PJ, VanWormer JJ, Sillah A, Winden TJ, Boucher JL. Exploring electronic health records as a population health surveillance tool of cardiovascular disease risk factors. Popul Health Manag 2015 Apr;18(2):79-85. [CrossRef] [Medline]
    31. McVeigh KH, Newton-Dame R, Chan PY, Thorpe LE, Schreibstein L, Tatem KS, et al. Can electronic health records be used for population health surveillance? Validating population health metrics against established survey data. EGEMS (Wash DC) 2016;4(1):1267 [FREE Full text] [CrossRef] [Medline]
    32. OneFlorida. 2016. OneFlorida: Clinical Research Consortium   URL: http://onefloridaconsortium.org [WebCite Cache]
    33. PCORnet. 2016. PCORnet Common Data Model (CDM)   URL: http://www.pcornet.org/pcornet-common-data-model [WebCite Cache]
    34. Agency for Healthcare Research and Quality. 2017. Overview of the National (Nationwide) Inpatient Sample (NIS)   URL: https://www.hcup-us.ahrq.gov/nisoverview.jsp [WebCite Cache]
    35. Centers for Medicare & Medicaid Services. 2016. CCW Chronic Condition Algorithms in the Chronic Conditions Data Warehouse   URL: https://www.ccwdata.org/web/guest/condition-categories [WebCite Cache]
    36. Ehrenpreis ED, Zhou Y. Hospital costs, length of stay and prevalence of hip and knee arthroplasty in patients with inflammatory bowel disease. World J Gastroenterol 2017 Jul 14;23(26):4752-4758 [FREE Full text] [CrossRef] [Medline]
    37. Wilson D, Jin DL, Wen T, Carmichael JD, Cen S, Mack WJ, et al. Demographic factors, outcomes, and patient access to transsphenoidal surgery for Cushing's disease: analysis of the Nationwide Inpatient Sample from 2002 to 2010. Neurosurg Focus 2015 Feb;38(2):E2. [CrossRef] [Medline]
    38. Khorgami Z, Aminian A, Shoar S, Andalib A, Saber AA, Schauer PR, et al. Cost of bariatric surgery and factors associated with increased cost: an analysis of national inpatient sample. Surg Obes Relat Dis 2017 Aug;13(8):1284-1289. [CrossRef] [Medline]
    39. Spector WD, Limcangco R, Furukawa MF, Encinosa WE. The marginal costs of adverse drug events associated with exposures to anticoagulants and hypoglycemic agents during hospitalization. Med Care 2017 Sep;55(9):856-863. [CrossRef] [Medline]
    40. Balla A, Batista RG, Corradetti S, Balagué C, Fernández-Ananín S, Targarona EM. Outcomes after bariatric surgery according to large databases: a systematic review. Langenbecks Arch Surg 2017 Sep 01;402(6):885-899. [CrossRef] [Medline]
    41. OneFlorida. 2017. Partners of OneFlorida Clinical Research Consortium   URL: https://www.ctsi.ufl.edu/research-initiatives/projects/oneflorida [WebCite Cache]
    42. Leaf C. The New York Times. 2013. Do Clinical Trials Work?   URL: http://www.nytimes.com/2013/07/14/opinion/sunday/do-clinical-trials-work.html [WebCite Cache]
    43. Lee PY, Alexander KP, Hammill BG, Pasquali SK, Peterson ED. Representation of elderly persons and women in published randomized trials of acute coronary syndromes. J Am Med Assoc 2001 Aug 08;286(6):708-713. [Medline]
    44. Sardar MR, Badri M, Prince CT, Seltzer J, Kowey PR. Underrepresentation of women, elderly patients, and racial minorities in the randomized trials used for cardiovascular guidelines. JAMA Intern Med 2014 Nov;174(11):1868-1870. [CrossRef] [Medline]
    45. Hutchins LF, Unger JM, Crowley JJ, Coltman CA, Albain KS. Underrepresentation of patients 65 years of age or older in cancer-treatment trials. N Engl J Med 1999 Dec 30;341(27):2061-2067. [CrossRef] [Medline]
    46. Lewis JH, Kilgore ML, Goldman DP, Trimble EL, Kaplan R, Montello MJ, et al. Participation of patients 65 years of age or older in cancer clinical trials. J Clin Oncol 2003 Apr 01;21(7):1383-1389. [CrossRef] [Medline]
    47. Schoenmaker N, Van Gool WA. The age gap between patients in clinical studies and in the general population: a pitfall for dementia research. Lancet Neurol 2004 Oct;3(10):627-630. [CrossRef] [Medline]
    48. Cigolle CT, Blaum CS, Halter JB. Diabetes and cardiovascular disease prevention in older adults. Clin Geriatr Med 2009 Nov;25(4):607-41, vii. [CrossRef] [Medline]
    49. He Z, Ryan P, Hoxha J, Wang S, Carini S, Sim I, et al. Multivariate analysis of the population representativeness of related clinical studies. J Biomed Inform 2016 Apr;60:66-76 [FREE Full text] [CrossRef] [Medline]
    50. Beers E, Moerkerken DC, Leufkens HG, Egberts TC, Jansen PA. Participation of older people in preauthorization trials of recently approved medicines. J Am Geriatr Soc 2014 Oct;62(10):1883-1890. [CrossRef] [Medline]
    51. Szlejf C, Farfel JM, Curiati JA, Couto Jr ED, Jacob-Filho W, Azevedo RS. Medical adverse events in elderly hospitalized patients: a prospective study. Clinics (Sao Paulo) 2012 Nov;67(11):1247-1252 [FREE Full text] [Medline]


    Abbreviations

    CC: chronic condition
    CCW: chronic condition warehouse
    CDM: common data model
    CDRN: Clinical Data Research Network
    CMS: Center for Medicare and Medicaid Services
    EHR: electronic health record
    HCUP: Healthcare Cost and Utilization Project
    HITECH: Health Information Technology for Economic and Clinical Health
    ICD: International Classification of Diseases
    MCC: multiple chronic conditions
    NH: non-Hispanic
    NIS: National Inpatient Sample
    PCORnet: Patient-Centered Clinical Research Network
    PCORI: Patient-Centered Outcomes Research Institute
    RQ: research question


    Edited by G Eysenbach; submitted 13.09.17; peer-reviewed by R Zhang, I de la Torre, S Wang, B Ward; comments to author 10.12.17; revised version received 20.01.18; accepted 15.02.18; published 12.04.18

    ©Zhe He, Jiang Bian, Henry J Carretta, Jiwon Lee, William R Hogan, Elizabeth Shenkman, Neil Charness. Originally published in the Journal of Medical Internet Research (http://www.jmir.org), 12.04.2018.

    This is an open-access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in the Journal of Medical Internet Research, is properly cited. The complete bibliographic information, a link to the original publication on http://www.jmir.org/, as well as this copyright and license information must be included.