Chronic Care Management and Mortality Among Diabetic Patients with Multiple Chronic Conditions

Abstract

In 2015, the Centers for Medicare and Medicaid Services started reimbursing chronic care management (CCM) services for patients with multiple chronic conditions. This study used 2015–2020 Medicare claims data from Illinois, Iowa, Minnesota, and Wisconsin and conducted a retrospective cohort study of 885,132 beneficiaries with an evaluation and management visit, following a diabetes diagnosis with other co-occurring chronic conditions. A competing-risk model was estimated to analyze factors associated with patients’ receipt of their first CCM services and a Cox proportional hazard model was estimated to assess the risk of death post-CCM initiation. Diabetic patients with multiple chronic conditions had mean age of 70 years (SD = 10.3), 50.7% were female, and 81.3% were white. 1.0% (9,075 beneficiaries) had CCM claims. Excluding chronic conditions, variables associated with a higher likelihood of CCM initiation included age (sub-distribution hazard ratios [SHR] = 1.003 for each additional year, 95% CI:1.00–1.01), female (SHR = 1.10, 95%CI:1.05–1.15), Black (SHR = 1.27, 95% CI:1.19–1.36) or Hispanic (SHR = 1.40, 95% CI:1.23–1.58), receiving care at home (SHR = 5.00, 95% CI:4.55–5.51) or skilled nursing facilities (SHR = 1.60, 95% CI:1.48–1.73), being a non-Iowa resident, and getting a diabetes diagnosis post-2015. However, patients in non-urban areas were less likely to receive such services. No statistical difference was found in the likelihood of mortality with CCM initiation vs. non-CCM. After accounting for CCM initiation, variables associated with a higher likelihood of death included age, American Indian/Alaska Native, residing in non-urban areas, getting a diabetes diagnosis in 2020, and receiving care in non-outpatient settings. CCM remains largely underutilized among Medicare beneficiaries. Addressing barriers, including improving access in non-urban areas and managing chronic condition earlier, may enhance adoption and decrease the risk of death for patients with multimorbidity.

Keywords

chronic care management Medicare diabetes chronic conditions

Diabetes affects over 38 million adults in the United States¹ and is the most expensive chronic condition with an estimated total annual cost of $413 billion in 2022.² Disparities in diabetes outcomes have been documented in vulnerable populations (racial and ethnic minorities and rural populations)^3–7 with poorer disease control and management.^8,9 The Centers for Medicare and Medicaid Services (CMS) starting reimbursement for chronic care management (CCM) in primary care since 2015 offers an opportunity for improved diabetes outcomes and disparities.

CCM was introduced to improve patient health and care, and provides monthly compensation for non-face-to-face clinical time to enhance care coordination and continuity of care for patients with multiple chronic conditions^8,10,11 which are expected to last for at least 12 months or until death. To ensure patient engagement and awareness about their potential cost-sharing responsibilities, patients are required to provide verbal or written consent for clinicians to bill for the provided CCM services.¹² CCM services include, but are not limited to, the structured recording of patient health information, keeping comprehensive electronic care plans, managing care transitions, promptly coordinating and sharing patient health information with providers within and outside the practice, and providing 24/7 access to care continuity.¹⁰ However, earlier studies showed that CCM adoption has been low.^13–15 In 2015 and 2016, 1.2%−2.3% of eligible Medicare beneficiaries received CCM services,¹⁴ and take-up increased to 3.4% in 2019.¹⁶ Certain factors (eg, familiarity with new billing codes) influence the likelihood of practices adopting CCM as opposed to others.¹⁵ While CCM reimbursement increases revenue for practices, the complex billing process for these services, patient’s copay, and the practice’s operating system can potentially slow down take up.^12,17–19 Nonetheless, telemedicine has become a valuable tool in the expansion of CCM.^20,21

Non-face-to-face CCM was found to enhance outcomes including hemoglobin A1c, body mass index, and blood pressure for patients with diabetes.^22,23 However, it is not clear how the take-up of CCM has changed among diabetes patients, what factors influence CCM service receipts, and whether the use of CCM influences mortality in diabetes patients with multiple chronic conditions. The current study closes these gaps by analyzing the time to a first CCM service following a diabetes diagnosis, and the potential association of CCM receipt and mortality for Medicare beneficiaries with multiple chronic conditions. This study hypothesized that the likelihood of CCM initiation will differ across patient factors and that CCM receipt will reduce the likelihood of death among patients.

Study Data and Methods

This retrospective cohort study used administrative Medicare claims data from 2015 through 2020 for patients with diabetes residing in four Midwest states: Wisconsin and Illinois (2015–2020), Iowa (2016–2020), and Minnesota (2019–2020). The varying time periods across states were due to data availability. Patients’ Physician Services/Carrier (Part B) and Medicare Provider Analysis and Review (MedPAR) claims were used to obtain information on care patterns, diagnoses, utilization, and provider specialty. The Master Beneficiary Summary File provided information on patient demographics, ZIP code, and date of death. This study was approved by the University of Wisconsin-Madison Institutional Review Board and followed the STROBE reporting guidelines.²⁴

Patient population

The study included Medicare fee-for-service beneficiaries with Parts A & B coverage, without managed care, and residing in Wisconsin, Illinois, Iowa, and Minnesota in 2015–2020. Additionally, the study only included patients with evaluation and management (E&M) visits, a first diabetes diagnosis and at least one other chronic condition diagnosis during the study period. Multimorbidity makes these patients eligible for CCM services. A diagnosis for diabetes was identified in claims using both International Classification of Diseases, 9th Revision (ICD-9) (primarily 250.**) and 10^th Revision (ICD-10) (primarily E08.* to E13.*) codes (Table A1). Beside diabetes, other chronic conditions studied included coronary artery disease, cerebral hemorrhage/stroke, cancer, congestive heart failure (CHF), connective tissue disorders, chronic obstructive pulmonary disease (COPD), dementia, hematological/thrombotic disease, HIV/AIDS, immune disease, liver disease, Parkinson’s/Huntington’s disease, paralysis, renal disease, peripheral vascular disease, severe mental illness (SMI), substance use disorder (SUD), asthma, atrial fibrillation, autism spectrum disorders, hyperlipidemia, hypertension, and osteoporosis (Table A1). Each patient contributed one sample observation within the study period.

Variable Descriptions

Outcome variables

The study's outcome measures included the time to the first CCM service and time to death after a diagnosis of diabetes with multiple chronic conditions. CCM services, based on billed and approved claims, were identified in claims with Current Procedural Terminology (CPT) codes 99487, 99489, 99490, 99491, G0506, and G2058. Complex CCM was defined by CPT 99487, 99489 and non-complex CCM was identified with CPT 99490, 99491. Indicator variables were defined for the presence of each CCM code, for receiving any CCM services and for complex and non-complex CCM. The time to the first CCM service was calculated as the period starting from a diabetes diagnosis to the first CCM service reported or the study end date for patients not receiving any CCM. Time to death was calculated as the time from a patient’s first diabetes diagnosis to the time of death or the study end date for surviving patients.

Covariates

The analysis included patients’ demographic characteristics including age at first diabetes diagnosis (years), sex [male (reference), female], race/ethnicity [White (reference), Black, Native Hawaiian/Pacific Islander, Asian, Hispanic, American Indian/Alaska Native and Other/unknown], residence in urban or rural areas, state of residence [Iowa (reference), Illinois, Minnesota and Wisconsin], and chronic condition diagnoses. The patient’s ZIP code of residence served to identify urbanicity of residence using Rural-Urban Commuting Area codes [urban (reference), suburban, large town and rural].²⁵ Also included was the patients’ most common place of service [outpatient professional services (reference), home, hospice, and skilled nursing facility]. Comorbidities were measured with the chronic conditions mentioned above.

Statistical Analysis

The study employed survival analysis to assess (1) the correlates of a first CCM service (CCM initiation) and (2) the association between CCM service receipt and mortality following a diabetes diagnosis for patients with multiple chronic conditions. A Fine-Gray sub-distribution hazard competing risk regression model²⁶ was estimated to account for the competing risk of death to the likelihood of a first CCM visit. Unlike the Cox proportional hazard model,²⁷ this model treats only the event of interest (CCM service) as failure and all other events (eg, death) as competing risks and therefore as censored observations. The Fine-Gray model treats the competing risks as informative.²⁸ Because death is the primary competing risk in this study, a cause-specific model and Fine-Gray model provided similar results (results not shown). The Fine-Gray model sub-distribution hazard ratios (SHRs) reflect the risk (cumulative incidence) of patients in the study. The regression model included demographic characteristics, place of service, state of residence, and chronic condition diagnoses defined earlier. Regressions also included the year of first diabetes diagnosis to account for different patients’ observation time during the study period. Heteroskedasticity robust standard errors were estimated. Because no other event would preclude the occurrence of death, when modeling the likelihood of mortality (not differentiating between causes of death), a Cox proportional hazard model was estimated, including an indicator for any CCM receipt during the study period and adjusting for the propensity to receive CCM services. The same covariates were included in this model as in the regression modeling the likelihood of CCM initiation.

In sensitivity analyses, the Fine-Gray model for time to first CCM and the Cox proportional hazard model for time to death were re-estimated including patients who received care in the four Midwestern states, considered in the main analysis, but who reported residing outside of these four states. Non-residents who travel for medical care (i.e., domestic medical tourism) may differ in unobservable ways (eg, limited health care resources in their place of residence, needs for specialized care, shorter wait time, and higher quality of care) from those getting care in their state of residence. Then, a series of analyses were conducted, focusing on CCM receipt. Further analyses also considered the number of CCM visits as an outcome variable. Treating this variable as continuous, an ordinary least squares (OLS) regression was estimated to assess the correlates of the number of CCMs. The previous OLS regression was also re-estimated, focusing on patients who received at least one CCM visit during our study period, adjusting for sampling weight based on the probability of CCM receipt. Finally, the number of CCMs was treated as a count variable, and a Poisson regression was estimated. Covariates in all of these additional analyses were the same as in the main regression model.

All statistical tests were two-tailed, with P < 0.05 considered statistically significant and 0.1 ≤ P ≥ 0.05 considered marginally significant. Analyses were performed using STATA, version 18.0/SE (StataCorp).

Results

The analysis included 5,465,138 patient-visit-year observations from 885,132 Medicare beneficiaries with a first diagnosis of diabetes and at least one other chronic condition and with E&M visits during 2015–2020. Descriptive characteristics were presented in Table 1. Patients had an average age at first diabetes diagnosis of 70.0 years (SD = 10.3), and 50.7% were female. Most patients were White (81.3%), and 63.1% lived in urban areas. Over nine in ten patients (92.5%) received their care at outpatient professional services, while 1.5% received services at home and 6.0% received services at skilled nursing facilities. More than half of the observations were from patients residing in Illinois (57.9%), 21.2% of observations were from Wisconsin, 14.0% from Iowa, and the remaining 7.0% from Minnesota. The most common chronic condition diagnoses for diabetes patients were hypertension (69.5%), hyperlipidemia (49.7%), renal disease (14.6%), and other conditions were reported in 0.1% to 6.1% observations.

Table 1.

Descriptive Characteristics of Patients with Diabetes and Multiple Chronic Conditions with an Evaluation and Management Visit (2015–2020)

Variable	Proportion
Demographic characteristics
Age, mean (SD)	70.00 (10.26)
Female, %	50.73
Race/Ethnicity, %
White (Reference)	81.26
Black	10.83
Hispanic	2.32
Asian	1.92
Native Hawaiian/Pacific Islander/Multi	1.69
American Indian/Alaska Native	0.46
Other or unknown	1.53
Urbanicity, %
Urban (Reference)	63.07
Suburban	16.31
Large town	11.49
Rural	9.13
Place of service group^a
Outpatient Professional Services (Reference)	92.52
Home	1.46
Hospice	0.01
Skilled nursing facility	6.01
State code
Iowa (Reference)	14.01
Illinois	57.85
Minnesota	6.97
Wisconsin	21.17
Diagnoses
Coronary artery disease	2.61
Cerebral hemorrhage/stroke	1.56
Cancer	2.95
Congestive heart failure	5.36
Connective tissue disorders	1.45
Chronic obstructive pulmonary disease	5.83
Dementia	2.62
Hematologic/thrombotic disease	1.02
HIV/AIDS	0.13
Immune disease	0.21
Liver disease	0.76
Parkinson’s/Huntington’s	0.58
Paralysis	0.24
Renal disease	14.57
Peripheral vascular disease	4.70
Severe mental illness	4.84
Substance use disorder	0.38
Asthma	5.60
Atrial fibrillation	6.11
Autism spectrum disorders	0.05
Hyperlipidemia	49.66
Hypertension	69.49
Osteoporosis	1.36
Observations (patients)	885,132
Observations (patient-visit-years)	5,465,138

Place of service groups identify the most frequent place of service identified for the patient during the study period.

Table 2 shows the distribution of first CCM services by CTP codes among patients in this study. CCM services were low among diabetes patients with multiple chronic conditions: 0.86% (n = 7,587) of patients received non-complex CCM as their first CCM services, 0.04% (n = 340) received complex CCM services, and 1.03% (n = 9,075) of patients received a CCM service defined during the study period. In the sample, 23.25% (n = 205,820) of patients had a reported death date.

Table 2.

Non-Complex and Complex Chronic Condition Management First Use and Mortality in Patients with Diabetes and Another Chronic Condition with an Evaluation and Management Visit (2015–2020)

HCPCS	Description	Frequency	Percentage
99487	99487 complex CCM is a 60-minute timed service provided by clinical staff to substantially revise or establish comprehensive care plan that involves moderate- to high-complexity medical decision making.	340	0.04%
99489	CPT code 99489 is each additional 30 minutes of clinical staff time spent providing complex CCM directed by a physician or other qualified health care professional (reported in conjunction with CPT code 99487; cannot be billed with CPT code 99490)	130	0.01%
99490	CPT code 99490-non-complex CCM is a 20-minute timed service provided by clinical staff to coordinate care across providers and support patient accountability.	7,518	0.85%
99491	CPT code 99491-CCM services provided personally by a physician or other qualified health care professional for at least 30 minutes.	69	0.01%
G0506	Comprehensive assessment and care planning for patients requiring chronic care management services (billed separately from monthly care management services).	1,203	0.14%
G2058	Chronic care management services, each additional 20 minutes of clinical staff time directed by a physician or other qualified health care professional, per calendar month. G2058 can be billed alongside 99487, 99490, and 99491.	100	0.01%
Complex CCM (99487, 99489, G2058 with either 99487 or 99489)	Includes complex CCM services	340	0.04%
Non-Complex CCM (99490, 99491, G2058 with 99490 or 99491)	Includes non-complex CCM services	7,587	0.86%
Any CCM Service Received (Yes)		9,075	1.03%
No CCM Service Received (No)		876,057	98.97%
Death (Yes)	Recorded death within the study period	205,820	23.25%

Medicare beneficiaries contributed to one observation in the percentage calculations. Because billing for CPT code 99489 is conditional on having CPT code 99487, the frequency for complex CCM flag is the same as the frequency for CPT code 99487. CPT codes 99490 and 99491 cannot be billed at the same time for the same patient in a given calendar month.

HCPCS, Healthcare Common Procedure Coding System; CCM, chronic care management; CPT, Current Procedural Terminology.

Regression analyses

Time to first CCM

The Fine-Gray competing risk model results showed that factors associated with a higher hazard of receiving a first CCM visit following a diabetes diagnosis included age (SHR = 1.003 for each additional year, 95% CI: 1.00–1.01), female relative to male (SHR = 1.10, 95% CI: 1.05–1.15), being Black (SHR = 1.27, 95% CI: 1.19–1.36) or Hispanic (SHR = 1.40, 95% CI: 1.23–1.58) versus White, receiving CCM services at home (SHR = 5.00, 95% CI: 4.45–5.51) or in a skilled nursing facility (SHR = 1.60, 95% CI: 1.48–1.73) versus in an outpatient setting, residing in Illinois (SHR = 2.16, 95% CI: 1.95–2.39), Minnesota (SHR = 1.65, 95% CI: 1.44–1.90), or Wisconsin (SHR = 1.64, 95% CI: 1.47–1.83) compared to Iowa, or getting a diabetes diagnosis after 2015 and with multimorbidity (Table 3). Nonetheless, factors associated with a lower hazard of receiving CCM service included residing in non-urban areas (suburban SHR = 0.70, 95% CI: 0.65–0.75; large town SHR = 0.38, 95% CI: 0.34–0.42, and rural SHR = 0.48, 95%CI: 0.43–0.54), and receiving care in a hospice (SHR = 1.00e-06, 95%CI: 6.98e-07- 1.45e-06). Chronic conditions associated with a higher hazard of receiving a CCM service included cerebral hemorrhage/stroke, CHF, connective tissue disorders, COPD, dementia, renal disease, SMI, asthma, atrial fibrillation, hypertension, and osteoporosis.

Table 3.

Predictors of First Chronic Care Management Service and Death among Diabetes Patients with Multiple Chronic Conditions

Variable	Time to first CCM	Time to death
Variable	Fine-Gray model (SHR [95% CI])	Cox model (HR [95% CI])
Any CCM services	—	0.9440 (0.8798, 1.0128)
Demographic Characteristics
Age	1.0028 (1.0003, 1.0052)**	1.0506 (1.0496, 1.0516)****
Female	1.0959 (1.0453, 1.1489)****	0.7882 (0.7762, 0.8005)****
Race/Ethnicity
White (Reference)	—	—
Black	1.2724 (1.1904, 1.3600)****	0.9311 (0.9081, 0.9546)****
Native Hawaiian/Pacific Islander	0.8959 (0.7439, 1.0790)	0.7976 (0.7482, 0.8503)****
Asian	1.0660 (0.9143, 1.2429)	0.7430 (0.6976, 0.7915)****
Hispanic	1.3957 (1.2325, 1.5805)****	0.8199 (0.7743, 0.8682)****
American Indian/Alaska Native	0.7126 (0.4416, 1.1497)	1.1154 (0.9803, 1.2692)*
Other or unknown	1.0212 (0.8411, 1.2398)	0.7507 (0.6955, 0.8103)****
Urbanicity
Urban (Reference)	—	—
Suburban	0.6952 (0.6465, 0.7476)****	1.0198 (1.0001, 1.0399)**
Large Town	0.3758 (0.3363, 0.4200)****	1.0351 (1.0109, 1.0598)***
Rural	0.4820 (0.4291, 0.5415)****	1.0193 (0.9936, 1.0456)
Place of Service Group
Outpatient Professional Services (Reference)	—	—
Home	5.0047 (4.5479, 5.5075)****	1.8942 (1.8265, 1.9643)****
Hospice	1.00e-06 (6.98e-07, 1.45e-06)****	9.4475 (4.5359, 19.6776)****
Skilled Nursing Facility	1.5988 (1.4776, 1.7301)****	3.4268 (3.3634, 3.4914)****
Diagnoses
Coronary artery disease	0.9477 (0.8231, 1.0910)	0.9998 (0.9487, 1.0537)
Cerebral hemorrhage/stroke	1.2151 (1.0520, 1.4034)***	1.2935 (1.2322, 1.3578)****
Cancer	1.0156 (0.8912, 1.1573)	2.3463 (2.2519, 2.4447)****
Congestive Heart Failure	1.6510 (1.5254, 1.7870)****	1.5992 (1.5580, 1.6414)****
Connective tissue disorders	1.1630 (0.9778, 1.3832)*	1.0045 (0.9399, 1.0734)
Chronic obstructive pulmonary disease	1.3841 (1.2548, 1.5267)****	1.5484 (1.5022, 1.5961)****
Dementia	2.3872 (2.1718, 2.6241)****	1.4840 (1.4406, 1.5286)****
Hematological/thrombotic disease	0.8286 (0.6398, 1.0730)	1.2049 (1.1292, 1.2856)****
HIV/AIDS	0.7605 (0.4087, 1.4151)	1.3208 (1.0404, 1.6768)**
Immune disease	0.6337 (0.3405, 1.1790)	1.3704 (1.1944, 1.5724)****
Liver disease	0.9300 (0.7134, 1.2124)	2.1485 (1.9960, 2.3126)****
Parkinson’s/Huntington’s	1.0788 (0.8315, 1.3998)	1.3736 (1.2836, 1.4700)****
Paralysis	0.5904 (0.3415, 1.0204)*	1.6860 (1.4924, 1.9047)****
Renal disease	1.1869 (1.1121, 1.2668)****	1.4511 (1.4217, 1.4812)****
Peripheral vascular disease	0.6522 (0.5732, 0.7422)****	1.1910 (1.1532, 1.2302)****
Severe mental illness	1.6744 (1.5380, 1.8230)****	0.9774 (0.9388, 1.0177)
Substance use disorder	0.6348 (0.4244, 0.9497)**	1.4524 (1.2843, 1.6425)****
Asthma	1.6370 (1.4872, 1.8020)****	0.9535 (0.9188, 0.9896)**
Atrial fibrillation	1.1101 (1.0124, 1.2172)**	1.2605 (1.2268, 1.2950)****
Autism spectrum disorders	0.5357 (0.1328, 2.1598)	0.8472 (0.4663, 1.5391)
Hyperlipidemia	0.7826 (0.7434, 0.8239)****	0.6913 (0.6813, 0.7015)****
Hypertension	1.0543 (1.0014, 1.1100)**	0.8490 (0.8349, 0.8635)****
Osteoporosis	1.4555 (1.2337, 1.7172)****	0.8995 (0.8471, 0.9550)***
State code
Iowa (Reference)		—
Illinois	2.1567 (1.9473, 2.3886)****	0.9263 (0.9012, 0.9521)****
Minnesota	1.6519 (1.4398, 1.8953)****	0.7795 (0.7353, 0.8264)****
Wisconsin	1.6372 (1.4650, 1.8298)****	0.9891 (0.9603, 1.0189)
Year of study entry
2015 (Reference)		—
2016	1.4352 (1.3230, 1.5569)****	0.9453 (0.9227, 0.9684)****
2017	2.3507 (2.1716, 2.5446)****	0.9205 (0.8907, 0.9514)****
2018	2.4163 (2.2278, 2.6207)****	0.9319 (0.8968, 0.9683)****
2019	2.3676 (2.1819, 2.5691)****	1.0438 (0.9976, 1.0921)*
2020	3.1872 (2.8987, 3.5044)****	1.3871 (1.2954, 1.4853)****
Observations	883,311	884,989

The different sample sizes were due to observations being dropped due to failure upon entry into the sample. When modeling time to first CCM, 1,821 observations were dropped because the date of diabetes diagnosis was the same as the reported date, they received their first CCM service (time to first CCM = 0). Similarly, when modeling the time-to-death model, 143 observations were dropped because the date of diabetes diagnosis was the reported death date (time to death = 0). The regression modeling the time to death also included weights based on the probability of CCM receipt.

Inference: *P value < 0.10, **P value < 0.05, ***P value < 0.01,****P value < 0.001.

HR, hazard ratio; SHR, sub-distribution hazard ratio.

Time to death

Unadjusted analyses showed that a greater percentage of patients who died during the study period received CCM services relative to their counterparts who remained alive by the end of the study (Table A2). These results suggest that patients for whom CCM services were initiated may be relatively unhealthy and therefore more likely to have poorer outcomes than their counterparts for whom these services were not initiated. To address this selection into CCM, regression models included weights based on the probability of CCM receipt when modeling the hazard of death: CCM initiation was not found to be associated with a significant change in the likelihood of death (HR = 0.94, 95%CI: 0.88–1.01, Table 3). Moreover, factors associated with an increase in the hazard included age (HR = 1.051 for an additional year, 95% CI: 1.050–1.052), American Indian/Alaska Native (HR = 1.12, 95% CI: 0.98–1.27) versus White, residing in non-urban areas (suburban: HR = 1.02, 95% CI: 1.00–1.04; large town: HR = 1.04, 95% CI: 1.01–1.06; rural: HR = 1.02, 95% CI: 0.99–1.05), receiving care in non-outpatient settings (home: HR = 1.89, 95% CI: 1.83–1.96; hospice: HR = 9.45, 95% CI: 4.54–19.68; skilled nursing facility: HR = 3.43, 95% CI: 3.36–3.49), a diabetes diagnosis in 2020 (HR = 1.39, 95% CI: 1.30–1.49) versus in earlier years, and chronic conditions for diabetes patients with multimorbidity, except for asthma, hyperlipidemia, hypertension, and osteoporosis.

Sensitivity analyses

When adding patients residing outside the 4 states considered in the main analyses, results were robust for covariates across both outcome variables with some exceptions, whose estimated relationships gained statistical significance in general, when modeling time to first CCM (Table A3).

Focusing on CCM receipt as measured with the number of CCM visits and the sample of patients residing in Iowa, Illinois, Minnesota and Wisconsin (same as in main analysis), results were qualitatively similar, in general, to the main results and across models (Table A4).

Discussion

With the growing number of people with diabetes and multimorbidity in the US and reimbursement for CCM services since 2015, we evaluated individual factors correlated with a first CCM and the likelihood of death among diabetes patients with multimorbidity in four Midwestern states between 2015 and 2020. Like findings about overall use of CCM services in the first 2 years of CCM implementation,¹⁴ CCM take-up was relatively low in this study, with 1.03% patients having CCM services initiated, mostly driven by non-complex CCM services. Duration/survival analyses showed that besides clinical condition diagnoses, patient demographic characteristics, location of residence, care setting, and diagnoses were all associated with initiation of CCM services and mortality. This study found variation across patient non-medical factors with older patients, females, Black and Hispanic, and urban patients being more likely to have CCM service initiated during the study period. Sensitivity analyses corroborated the main results.

Beside the low CCM utilization, the lower likelihood of CCM service initiation for rural and other non-urban patients found in this study may be due to lack of adequate resources (eg, health care professional shortage), higher rates of uninsurance and underinsurance, lower health literacy levels that constitute important barriers to health care access and CCM provision (providers) or receipt (patients) in non-urban areas.²⁹ The lack or lower take-up of CCM in non-urban areas may also be contributing to reinforcing existing inequities and health disparities across urban and rural areas.^4,30,31 Investment in resources and clinical staff for increased support and CCM provision is essential because CCM services were found to increase patient and provider satisfaction in addition to saving costs.³² CCM in these underserved areas remains an important strategy for improving health care for individuals with multimorbidity and may serve to address the unique challenges facing these communities.

CCM initiation may also be delayed or not started if patients have not given their consent before CCM services are billed for them, potentially due to the possible cost-sharing responsibilities involved. Additionally, variation in CCM initiation across states suggests potential variation of such services across states beyond the four states studied.

The unadjusted analyses showing a greater percentage of decedents also receiving CCM services compared to patients who are alive at the end of the study, may signal that patients receiving such services are at a higher risk of poorer outcomes, as evidenced by the higher associations estimated between several chronic condition diagnoses and CCM initiation as well as between chronic conditions and death. Notwithstanding the negative relationship estimated between CCM receipt and the hazard of death, statistical insignificance of the estimate in the main analysis may suggest that non-complex CCM, which represents most of CCM services provided, may not be enough for patients at higher risk of death. More accessible complex CCM services may be needed and critical for meaningfully improving outcomes for patients with multiple chronic conditions.

Furthermore, besides higher compensation for complex CCM services ($92.4 for CPT 99487 versus $42.2 for non-complex CCM CPT 99490 using 2020 Medicare physician fee schedule and conversion factor),³³ more incentives should be provided to eligible clinical staff and physicians to substantially increase the use of such services to contribute to further reducing the risk of death in this population. The sizable increase in compensation rates of over 40% between 2020 and 2022 for both complex and non-complex CCM and qualifying more eligible Medicare beneficiaries for these services could contribute to enhancing the adoption of CCM. Nonetheless, CCM is subject to Medicare beneficiary’s annual deductible ($198 for Part B in 2020)³⁴ and 20% coinsurance, which may cost patients about $10 monthly after meeting their deductible for the year,³⁵ which could be a deterrent to patients’ consenting to initiate CCM. Robust discussion between providers and patients about the long-term financial and health benefits of CCM for the patients could help address this barrier.

These findings also highlight the need for coordinated care earlier during the patients’ care, particularly in people with one chronic condition, to target and better manage complexities of a single condition but also help to prevent or reduce the risk for other diagnoses and worse outcomes. The recently established principal care management (PCM) model of care (in 2020),³⁶ which reimburses providers for coordinating care and providing comprehensive CCM for patients with one high-risk chronic condition, may close the gap in coding and payment (and therefore care) created by the CCM requirement that people have multiple chronic conditions. PCM services include the development and implementation of disease-specific care plans or the adjustment of patients’ medications regularly. To further enhance the value of chronic condition management among patients with multimorbidity, PCM and CCM services could be provided to these patients by different providers managing different conditions in both programs. Future studies should explore the utilization and effects of PCM services in combination with CCM services for higher-risk patients with multimorbidity. Studies should also assess the potential of PCM services in delaying patients’ transition from one to multiple chronic conditions.

This study is subject to limitations. The authors focused on patients with a new diabetes diagnosis and multiple chronic conditions, and therefore, results may not be generalized to patients with a different combination of chronic conditions. Nonetheless, diabetes is highly prevalent in the United States, is among the costliest chronic conditions, and one of the major causes of death and disability in the country.³⁷ Diabetes is also a risk factor for other chronic conditions.³⁸ Although the likelihood of CCM initiation was studied in the primary analyses, the current study has also evaluated the correlates of the number of CCM services a patient received during the study period. However, the details and intensity of the CCM services provided beyond the CPT codes could not be assessed further. Some services depend on prior events or health care utilization, while others are based on general need support and services. For example, management of care transitions/referral is provided after discharge, emergency department visit, or referral, whereas comprehensive care management includes needs assessment, receipt of preventive services, medication reconciliation, management, and oversight of self-management.³⁶ Furthermore, overall CCM uptake over the first 6 years of implementation was studied. This study has not evaluated the utilization of the individual CCM codes over time, though they have been introduced at different points in time during the study period. For example, non-complex CCM CPT code 99490 was introduced in 2015, while CPT code 99491 was introduced much later, in 2019. Similarly, the complex CCM codes 99487 and 99489 were introduced in 2017. In 2021, additional add-on CCM CPT codes were introduced (eg, 99439 for CPT 99490 and 99437 for CPT 99491)³⁹ to allow for additional reimbursement for additional time spent providing non-complex CCM, which may have increased utilization of such services. The study has also not evaluated the differing impact from different providers (physician versus clinical staff) rendering CCM services, which is beyond the scope of this study.

Conclusion

CCM initiation was low among diabetes patients with multiple chronic conditions, and analyses showed differences in initiation across patient characteristics, clinical conditions, place of service, and geography. These inequalities in CCM utilization could be reduced through process simplification for billing, greater provider financial incentives, lower patient cost-sharing and telehealth service expansion that could boost CCM service uptake and potentially improve patient outcomes. The higher mortality risk associated with patients receiving CCM services calls for greater resources in addressing these chronic conditions early, before they become too complex to manage. The recent PCM codes allowing reimbursement for targeted management of single high-risk chronic conditions may help address some of these challenges earlier.

Authors’ Contributions

M.H.O.: Writing—review & editing, writing—original draft, supervision, methodology, investigation, formal analysis, conceptualization, visualization. Y.Z.: Writing—original draft, investigation, formal analysis, data curation, visualization.

Footnotes

Author Disclosure Statement

No competing financial interests exist.

Funding Information

No funding was received for this article.

Appendix

Table A4.

Association between Patient Factors and the Number of CCM Visits among Diabetes Patients with Multiple Chronic Conditions (2015–2020)

	Number of CCM visits
	Linear model		Linear model, adjusting for probability of CCM receipt weight-CCM visits > 0		Poisson model
Variable	Coefficient	Std. Err.	Coefficient	Std. Err.	IRR	Std. Err.
Demographic Characteristics
Age	0.0001**	(0.00005)	−0.002	(0.0034)	1.003*	(0.0015)
Female	0.004***	(0.0009)	0.119*	(0.0673)	1.117****	(0.0268)
Race/Ethnicity
White (Reference)	—	—	—	—	—	—
Black	0.009***	(0.0018)	−0.114	(0.0827)	1.212****	(0.0383)
Native Hawaiian/Pacific Islander	–0.004	(0.0034)	−0.310*	(0.1837)	0.890	(0.1065)
Asian	–0.006*	(0.0030)	−0.310	(0.2115)	0.870	(0.0874)
Hispanic	0.002	(0.0029)	−0.313**	(0.1356)	1.076	(0.0705)
American Indian/Alaska Native	–0.014***	(0.0035)	−0.053	(0.5566)	0.486***	(0.2537)
Other or unknown	–0.008***	(0.0030)	−0.538**	(0.2254)	0.776**	(0.1117)
Urbanicity
Urban (Reference)	—	—	—	—	—	—
Suburban	–0.011***	(0.0012)	−0.234**	(0.0987)	0.720****	(0.0399)
Large Town	–0.020***	(0.0012)	0.302	(0.2082)	0.482****	(0.0603)
Rural	–0.017***	(0.0015)	0.352*	(0.2057)	0.549****	(0.0657)
Place of Service Group
Outpatient Professional Services (Reference)	—	—	—	—	—	—
Home	0.093***	(0.0063)	−0.747****	(0.1001)	3.398****	(0.0553)
Hospice	–0.038**	(0.0150)	—	—	0.298	(1.0082)
Skilled Nursing Facility	0.023***	(0.0026)	−0.236**	(0.1022)	1.602****	(0.0459)
Diagnoses
Coronary artery disease	–0.005	(0.0031)	0.064	(0.2263)	0.888	(0.0806)
Cerebral hemorrhage/stroke	0.007	(0.0045)	0.223	(0.1840)	1.145	(0.0863)
Cancer	–0.004*	(0.0026)	0.043	(0.2225)	0.885*	(0.0736)
Congestive Heart Failure	0.021***	(0.0027)	−0.038	(0.1171)	1.512****	(0.0467)
Connective tissue disorders	0.003	(0.0038)	−0.266	(0.1659)	1.077	(0.0963)
Chronic obstructive pulmonary disease	0.018***	(0.0028)	−0.101	(0.1507)	1.492****	(0.0637)
Dementia	0.058***	(0.0049)	0.041	(0.1167)	2.391****	(0.0551)
Hematological/thrombotic disease	–0.002	(0.0043)	0.297	(0.3487)	0.919	(0.1359)
HIV/AIDS	–0.019**	(0.0073)	0.370	(0.5100)	0.536**	(0.3147)
Immune disease	–0.016***	(0.0056)	−1.014**	(0.4104)	0.566**	(0.2673)
Liver disease	–0.006	(0.0050)	−0.164	(0.3510)	0.830	(0.1607)
Parkinson’s/Huntington’s	0.004	(0.0071)	−0.112	(0.3505)	1.098	(0.1461)
Paralysis	–0.009	(0.0093)	0.007	(0.3755)	0.797	(0.2697)
Renal disease	0.003**	(0.0014)	0.021	(0.0990)	1.074*	(0.0371)
Peripheral vascular disease	–0.009***	(0.0020)	−0.089	(0.1663)	0.746****	(0.0683)
Severe mental illness	0.022***	(0.0029)	−0.042	(0.1303)	1.544****	(0.0481)
Substance use disorder	–0.019***	(0.0067)	−0.535	(0.5749)	0.594**	(0.2378)
Asthma	0.019***	(0.0030)	0.068	(0.1406)	1.338****	(0.0630)
Atrial fibrillation	0.004*	(0.0021)	0.112	(0.1418)	1.097*	(0.0530)
Autism spectrum disorders	–0.026***	(0.0091)	−0.266	(0.7121)	0.316	(0.7231)
Hyperlipidemia	–0.008***	(0.0010)	0.094	(0.0739)	0.768****	(0.0292)
Hypertension	0.004***	(0.0010)	0.093	(0.0771)	1.082***	(0.0291)
Osteoporosis	0.012***	(0.0042)	−0.398**	(0.1732)	1.339***	(0.0913)
State Code
Iowa (Reference)	—	—	—	—	—	—
Illinois	0.020***	(0.0015)	−1.165****	(0.1706)	1.869****	(0.0569)
Minnesota	0.020***	(0.0032)	−0.403*	(0.2090)	1.814****	(0.0736)
Wisconsin	0.011***	(0.0015)	−0.613***	(0.1974)	1.381****	(0.0620)
Year of Study Entry
2015 (Reference)	—	—	—	—	—	—
2016	0.011***	(0.0013)	0.116	(0.0993)	1.329****	(0.0474)
2017	0.029***	(0.0019)	−0.025	(0.1008)	2.318****	(0.0452)
2018	0.041***	(0.0022)	0.408****	(0.1120)	2.915****	(0.0444)
2019	0.037***	(0.0021)	0.227**	(0.1072)	2.719****	(0.0447)
2020	0.040***	(0.0024)	0.434***	(0.1254)	2.907****	(0.0477)
Observations	885,132		9,075		885,132

The year of study entry is the year of first diabetes diagnosis during the study period.

Inference: *P value < 0.10, **P value < 0.05, ***P value < 0.01, ****P value < 0.001.

References

Center for Disease Prevention and Control. Diabetes fast facts. 2023. Available from: https://www.cdc.gov/diabetes/basics/quick-facts.html#print [Accessed April 9, 2023].

Center for Disease Control and Prevention. Prevalence of both diagnosed and undiagnosed diabetes. 2022. Available from: https://www.cdc.gov/diabetes/data/statistics-report/diagnosed-undiagnosed-diabetes.html [Accessed September 6, 2023].

Dugani

, Wood-Wentz

, Mielke

, Bailey

, Vella

. Assessment of disparities in diabetes mortality in adults in US rural vs nonrural counties, 1999–2018. JAMA Netw Open, 2022; 5(9):e2232318.

Foss

, Fischer

, Lampman

, et al. Disparities in diabetes care: Differences between rural and urban patients within a large health system. Ann Fam Med, 2023; 21(3):234–239.

Chiou

, Tsugawa

, Goldman

, Myerson

, Kahn

, Romley

. Trends in racial and ethnic disparities in diabetes-related complications, 1997–2017. J Gen Intern MED, 2020; 35(3):950–951.

Mikhail

, Wali

, Brown

. Ethnic disparities in diabetes. Endocrinol Metab Clin North Am, 2021; 50(3):475–490.

Haw

, Shah

, Turbow

, Egeolu

, Umpierrez

. Diabetes complications in racial and ethnic minority populations in the USA. Curr Diab Rep, 2021; 21(1):2–8.

Peek

, Cargill

, Huang

. Diabetes health disparities. Med Care Res Rev, 2007; 64(5_suppl):101S–156S.

Zakaria

, Tehranifar

, Laferrère

, Albrecht

. Racial and ethnic disparities in glycemic control among insured US adults. JAMA Netw Open, 2023; 6(10):e2336307.

10.

Centers for Medicare and Medicaid Services. Chronic care management services. 2025. Available from: https://www.cms.gov/outreach-and-education/medicare-learning-network-mln/mlnproducts/downloads/chroniccaremanagement.pdf [Accessed July 30, 2025].

11.

Edwards

, Landon

. Medicare’s chronic care management payment—Payment reform for primary care. N Engl J Med, 2014; 371(22):2049–2051.

12.

Traynor

. Medicare mulls changes to boost participation in chronic care initiative. Am J Health-Syst Pharm, 2016; 73(22):1808–1810.

13.

Bindman

, Cox

. Changes in health care costs and mortality associated with transitional care management services after a discharge among Medicare beneficiaries. JAMA Intern Med, 2018; 178(9):1165–1171.

14.

Agarwal

, Barnett

, Souza

, Landon

. Adoption of Medicare’s transitional care management and chronic care management codes in primary care. JAMA, 2018; 320(24):2596–2597.

15.

Agarwal

, Barnett

, Souza

, Landon

. Medicare’s care management codes might not support primary care as expected: An analysis of rates of adoption for Medicare’s new billing codes for transitional care management and chronic care management. Health Aff, 2020; 39(5):828–836.

16.

Jang

, McCarthy

, Olivieri‐Mui

, et al. Use of chronic care management service among Medicare beneficiaries in 2015–2019. J Am Geriatr Soc, 2024; 72(9):2730–2737.

17.

Peters

, Bunkers

. Chronic care coordination. Chest, 2015; 148(4):1115–1119.

18.

Bazzano

, Wharton

, Monnette

, et al. Barriers and facilitators in implementing non-face-to-face chronic care management in an elderly population with diabetes: A qualitative study of physician and health system perspectives. J Clin Med, 2018; 7(11):451–412.

19.

Pannill

FC.

III ., Medicare’s chronic care management program: Will it help primary care survive? Ann Intern Med, 2015; 163(8):640–641.

20.

Badowski

, Wright

, Bainbridge

, et al. Implementation and evaluation of comprehensive medication management in telehealth practices. J Am Coll Clin Pharm, 2020; 3(2):520–531.

21.

Taylor

, Bingham

, Schussel

, et al. Integrating innovative telehealth solutions into an interprofessional team-delivered chronic care management pilot program. J Manag Care Spec Pharm, 2018; 24(8):813–818.

22.

Shao

, Stoecker

, Hong

, et al. The impact of reimbursement for non–face-to-face chronic care management on comprehensive metabolic biomarkers among multimorbid patients with type 2 diabetes. Med Care, 2023; 61(3):157–164.

23.

Shao

, Stoecker

, Hong

, et al. The impact of reimbursement for non-face-to-face chronic care management on health utilization among patients with type 2 diabetes in Louisiana. Value Health, 2023; 26(5):676–684.

24.

Equator Network. The strengthening the reporting of observational studies in epidemiology (Strobe) statement: Guidelines for reporting observational studies. 2023. Available from: https://www.equator-network.org/reporting-guidelines/strobe/ [Accessed July 30, 2025].

25.

U.S. Department of Agriculture Economic Research Service. Rural-urban commuting area (Ruca) codes. 2025. Available from: https://www.ers.usda.gov/data-products/rural-urban-commuting-area-codes [Accessed August 13, 2025].

26.

Fine

, Gray

. A proportional hazards model for the subdistribution of a competing risk. J Am Stat Assoc, 1999; 94(446):496–509.

27.

Cox

. Regression models and life‐tables. J R Stat Soc Series B Methodol), 1972; 34(2):187–202.

28.

Zhang

. Survival analysis in the presence of competing risks. Ann Transl Med, 2017; 5(3):47.

29.

Rural Health Information Hub. Barriers and facilitators for chronic disease management in rural areas. 2025. Available from: https://www.ruralhealthinfo.org/toolkits/chronic-disease/1/barriers-and-facilitators [Accessed August 5, 2025].

30.

Cross

, Califf

, Warraich

. Rural-urban disparity in mortality in the US from 1999 to 2019. JAMA, 2021; 325(22):2312–2314.

31.

Weeks

, Chang

, Pagán

, et al. Rural-urban disparities in health outcomes, clinical care, health behaviors, and social determinants of health and an action-oriented, dynamic tool for visualizing them. PLOS Glob Public Health, 2023; 3(10):e0002420.

32.

Centers for Medicare and Medicaid Services. Evaluation of the diffusion and impact of the chronic care management (Ccm) services: Final report. 2017. Available from: https://www.cms.gov/priorities/innovation/Files/reports/chronic-care-mngmt-finalevalrpt.pdf

33.

Centers for Medicare and Medicaid Services. Pfs relative value files. 2025. Available from: https://www.cms.gov/medicare/payment/fee-schedules/physician/pfs-relative-value-files [Accessed August 6, 2025].

34.

Centers for Medicare and Medicaid Services. 2020 Medicare parts A & B premiums and deductibles. 2019. Available from: https://www.cms.gov/newsroom/fact-sheets/2020-medicare-parts-b-premiums-and-deductibles [Accessed August 6, 2025].

35.

Lanmboley

. Chronic care management patient costs: Justifying their investment. 2023. Available from: https://blog.prevounce.com/chronic-care-management-patient-costs [Accessed August 6, 2025].

36.

Federal Register. Medicare program; CY 2020 revisions to payment policies under the physician fee schedule and other changes to part B payment policies; Medicare shared savings program requirements; Medicaid promoting interoperability program requirements for eligible professionals; establishment of an ambulance data collection system; updates to the quality payment program; Medicare enrollment of opioid treatment programs and enhancements to provider enrollment regulations concerning improper prescribing and patient harm; and amendments to physician self-referral law advisory opinion regulations final rule; and coding and payment for evaluation and management, observation and provision of self-administered Esketamine interim final rule. 2019.

37.

Watson

, Wiltz

, Nhim

, Kaufmann

, Thomas

, Greenlund

. Trends in multiple chronic conditions among US adults, by life stage, behavioral risk factor surveillance system, 2013–2023. Prev Chronic Dis, 2025; 22:E15.

38.

Bodke

, Wagh

, Kakar

, Wagh

. Diabetes mellitus and prevalence of other comorbid conditions: A systematic review. Cureus, 2023; 15(11):e49374.

39.

Centers for Medicare and Medicaid Services. Frequently asked questions about billing Medicare for chronic care management services. 2016.

40.

Bynum

, Chang

, Austin

, Carmichael

, Meara

. Outcomes in older adults with multimorbidity associated with predominant provider of care specialty. J Am Geriatr Soc, 2017; 65(9):1916–1923.

41.

National Bureau of Economic Research. International Classification of Diseases (ICD) to hierarchical condition categories crosswalk. 2018. Available from: https://www-nber-org.ezproxy.library.wisc.edu/research/data/international-classifcation-diseases-icd-hierarchical-condition-categories-crosswalk [Accessed October 1, 2023].