Obesity affects nearly half (40%) of adults in the United States [1,2] and contributes substantially to a pandemic of obesity-associated chronic conditions. Obesity commonly leads to multiple chronic conditions such as type 2 diabetes, hypertension, cardiovascular disease, cerebrovascular disease, and arthritis [3]. Our recent research shows that 34% of patients with obesity-associated chronic conditions in the midsouthern United States have diabetes. Projections indicated that 81 million adults will have multimorbidity by 2020, and obesity serves as an underrecognized root cause for this national pandemic of multimorbidity [4-7]. The pandemic of obesity-associated chronic conditions is particularly severe in low-income, medically underserved, and predominantly African-American areas in the southern United States.

Furthermore, the burdens of uncontrolled obesity-associated chronic conditions are rising nationwide, particularly in areas designated by the Human Resources and Services Administration as primary care health professional shortage areas of the southern United States [8-10]. The economic burden of adult obesity on the United States exceeds $215 billion [11]; the most common obesity-associated chronic conditions are known to increase health care resource utilization and costs [12-15]. Access to regular primary and specialty care is associated with improved chronic care management and reduces potentially preventable health care utilization attributed to obesity-associated chronic conditions [16,17]. Prior evidence from observational studies suggests that continuity of care is associated with higher medication adherence and lower health care resource utilization among older adults and patients with multiple chronic conditions [18-24]. But little is known regarding the impact of continuity of care for patients with type 2 diabetes or other obesity-associated chronic conditions in medically underserved areas of the southern United States.

There is a critical need to quantify the impact of geographic, income, and racial disparities on continuity of care and health care utilization and identify characteristics of health systems with higher continuity and lower health care utilization for patients with obesity-associated chronic conditions and type 2 diabetes. Effective population health improvement requires examining comprehensive data across health systems to look at variations in care. Such pragmatic research has the potential to translate best health systems practices across different communities in the southern United States.

This protocol aims to assess (1) whether outpatient continuity of care is associated with lower overall and potentially preventable emergency department and hospital utilization for patients with obesity-associated chronic conditions and with type 2 diabetes, respectively; (2) the effect of geographic, income, ethnic, and racial disparities on continuity of care and overall and potentially preventable emergency department and hospital utilization among patients with obesity-associated chronic conditions and with type 2 diabetes; (3) whether continuity of care particularly protects individuals with obesity-associated chronic conditions and with type 2 diabetes at risk for geographic, income, ethnic, and racial disparities from overall and potentially preventable hospital and emergency department utilization; and (4) the characteristics of health systems with higher continuity and lower health care utilization for vulnerable patients with obesity-associated chronic conditions and in the type 2 diabetes.

We hypothesize that continuity of care will protect vulnerable individuals with obesity-associated chronic conditions, and within that group, individuals with type 2 diabetes, from overall and potentially preventable emergency department and hospital utilization. Additionally, continuity of care will be associated with lower emergency department and hospital use for (1) African-American individuals compared with that of white individuals or other individuals, (2) patients residing in health professional shortage areas compared with that of patients not residing in health professional shortage areas, and (3) patients residing in low-income areas compared with that of patients not residing in low-income areas. We also hypothesize that health systems with (1) a higher proportion of primary care providers to total providers, (2) a higher proportion of ambulatory to total encounters, and (3) a greater geographic distribution of ambulatory care sites will experience lower levels of overall and potentially preventable emergency department and hospital use among patients receiving ambulatory care within the system.

We will conduct a retrospective cohort analysis and meta-analysis of all adult patients with obesity-associated chronic conditions—defined as patients who are obese (BMI≥30 kg/m²) and who have ≥1 additional diagnosed obesity-associated chronic condition (a list of the included conditions is available in Multimedia Appendix 1) [25,26]—who are seen in health systems and practice-based research networks participating in the Southern Obesity and Diabetes Coalition.

This study will be conducted separately at the following sites: (1) University of Tennessee Health Science Center, (2) University of Mississippi Medical Center, (3) Tulane Medical Center, and (4) Ochsner Health System. The data from all sites will be aggregated to conduct meta-analysis. Electronic medical records from between 2015 and 2018 including encounters from all adult emergency department, hospitals, and clinics will be utilized. The index visit will be defined as the first visit of any type occurring between January 1, 2016 and December 31, 2017. The baseline year will be defined as the first full year prior to the qualifying index visit date (as defined above) for which the full inclusion criteria are met. The baseline year will be used to identify adult patients with obesity-associated chronic conditions and type 2 diabetes seen 2 or more times for ambulatory care. Outcomes will be assessed in a 1-year period following the index visit. The study design is shown in Figure 1.

The study will include patients with obesity-associated chronic conditions and with type 2 diabetes who meet the following criteria: patients aged >18 years at their index visit, who are obese (defined as BMI>30 kg/m²) at their index visit (if BMI>100 kg/m², they will be excluded; if there are multiple valid BMI values associated with the index visit, the mean BMI values will be used), with ≥1 diagnosed obesity-associated chronic conditions at any time in any location in their baseline year (see Multimedia Appendix 1 for definition of obesity-associated chronic conditions and qualifying ICD-9-CM and ICD-10-CM diagnosis codes) [25,26], with > 2 ambulatory visits in the baseline year, and with >1 encounter of any type in the follow-up year. Of those, patients with diagnosed type 2 diabetes at any time in any location in their baseline year will be identified.

Outcome variables will include overall emergency department visits and overall hospitalizations. These are 2 of the 4 primary outcomes for outpatient continuity of care association with health care use; the effect of geographic, income, ethnic, and racial disparities; and continuity of care protective effects; and are secondary outcomes for characteristics of health care systems with high continuity of care and low utilization. Hospitalizations will include observations visits, and emergency department visits resulting in observation or inpatient visits on the same or following day will be counted as hospitalizations.

Preventable emergency department visits and preventable hospitalizations will be defined by The Agency for Healthcare Research and Quality Prevention Quality Indicators chronic composite measures and will be identified using the primary diagnosis for diabetes short-term complications, diabetes long-term complications, chronic obstructive pulmonary disease or asthma in older adults, hypertension, angina without procedure, uncontrolled diabetes, asthma in younger adults, and lower extremity amputation among patients with diabetes [27]. Hospitalizations will include observations visits. Emergency department visits resulting in observation or inpatient visits on the same or following day will be counted as hospitalizations.

Continuity of care will be assessed for each patient using a modified version of the Bice-Boxerman index to assess continuity of care at the clinic or facility level [18,19,28]. The continuity of care index is a measure of concentration of the patient-clinic visit patterns ranging from 0 (each visit involved a different practice) to 1 (all visits were made to a single practice). It reflects the relative share of all of a patient’s visits during the year that were made to distinct practices. Thus, the higher the index score, the better the modified continuity of care. In this study, a modified Bice-Boxerman continuity of care index score will be used to assess the concentration of the patient-clinic visits patterns during the 12-month baseline period. For this measure, a patient-clinic visit is defined as an encounter with a unique combination of admit date, facility ID (or clinic name), and an ambulatory encounter type. The measure will include both primary and specialty care visits and is calculated as:

The main analysis will be conducted among patients with obesity-associated chronic conditions. Multivariable negative binomial models will be used to model overall and preventable hospitalizations and emergency department visits. For continuity of care as an outcome, a fractional regression model will be used. Interaction terms between (1) continuity of care and patient residence health professional shortage area designation, (2) continuity of care and patient residence low income status, and (3) continuity of care and patient race will be included to examine whether the association between continuity and health care utilization varies by these factors. The same will be assessed in patients with type 2 diabetes.

Finally, we will conduct a distributed meta-analysis using direct aggregate data from each site following the method employed by Toh [31,32] and Zhou and colleagues [33] to integrate study findings from the 4 different study sites. This approach employs inverse variance-weighted meta-analysis using multivariable confounding adjustment across distributed data networks without sharing of patient-level data [32].

This study protocol has been approved by the Institutional Review Board of the 4 sponsoring institutions (University of Tennessee Health Science Center protocol 18-06394-XP, Tulane Medical Center protocol 2018-2306, University of Mississippi Medical Center protocol 2019-0005, Ochsner Health System protocol 2019.035). The findings of this study will be submitted as manuscripts to peer-reviewed journals to aid dissemination to policy makers and other researchers in the field. It will also be presented and discussed at Annual Delta Clinical and Translational Science Health Disparities Conference.