Effects of Non-Face-to-Face Chronic Care Management on Service Utilization and Outcomes Among US Medicare Beneficiaries with Diabetes

Background Type 2 diabetes mellitus (T2DM) results in heavy economic and disease burdens in Louisiana. The Centers for Medicare and Medicaid Services has reimbursed non-face-to-face chronic care management (NFFCCM) for patients with two or more chronic conditions since 2015. Objective To assess the impacts of NFFCCM on healthcare utilization and health outcomes. Design, Setting, and Participants This retrospective cohort study included Medicare fee-for-service beneficiaries with T2DM and at least one additional chronic disease between 2014 and 2018. Exposures At least one record of NFFCCM Current Procedural Terminology codes. Main Measures The health outcomes in the study included major adverse cardiovascular events (MACE), all-cause mortality, and heart failure. The monthly service utilization and continuity of care index for primary care were also included. The propensity score method was used to balance the baseline differences between the two groups. Weighted multivariate regression models were developed using propensity score weights to assess the impacts of NFFCCM on outcomes. Key Results During the 5 years of study period, 8415 patients among the 118,643 Medicare beneficiaries received at least one NFFCCM. Patients receiving any NFFCCM had reduced healthcare utilization compared with patients not receiving NFFCCM, including 0.012 (95% CI − 0.014 to − 0.011; p < 0.001) fewer monthly hospital admissions, 0.017 (95% CI − 0.019 to − 0.016; p < 0.001) fewer monthly ED visits, and 0.399 (95% CI 0.375 to 0.423; p < 0.001) more monthly outpatient encounters. Patients receiving NFFCCM services had lower MACE event rates of 7.4% (95% CI 7.1 to 7.8%; p < 0.001), all-cause mortality rate of 7.8% (95% CI 7.4 to 8.1%; p < 0.001), and heart failure rate of 0.3% (95% CI 0.2 to 0.5%; p < 0.001), respectively. Conclusions and Relevance These findings suggest that reimbursement for NFFCCM was associated with the shifting high-cost utilization to lower-cost primary health care settings among patients with diabetes in Louisiana.


ABSTRACT
BACKGROUND: Type 2 diabetes mellitus (T2DM) results in heavy economic and disease burdens in Louisiana.The Centers for Medicare and Medicaid Services has reimbursed non-face-to-face chronic care management (NFFCCM) for patients with two or more chronic conditions since 2015.OBJECTIVE: To assess the impacts of NFFCCM on healthcare utilization and health outcomes.DESIGN, SETTING, AND PARTICIPANTS: This retrospective cohort study included Medicare fee-for-service beneficiaries with T2DM and at least one additional chronic disease between 2014 and 2018.EXPOSURES: At least one record of NFFCCM Current Procedural Terminology codes.

MAIN MEASURES:
The health outcomes in the study included major adverse cardiovascular events (MACE), all-cause mortality, and heart failure.The monthly service utilization and continuity of care index for primary care were also included.The propensity score method was used to balance the baseline differences between the two groups.Weighted multivariate regression models were developed using propensity score weights to assess the impacts of NFFCCM on outcomes.KEY RESULTS: During the 5 years of study period, 8415 patients among the 118,643 Medicare beneficiaries received at least one NFFCCM.Patients receiving any NFFCCM had reduced healthcare utilization compared with patients not receiving NFFCCM, including 0.012 (95% CI − 0.014 to − 0.011; p < 0.001) fewer monthly hospital admissions, 0.017 (95% CI − 0.019 to − 0.016; p < 0.001) fewer monthly ED visits, and 0.399 (95% CI 0.375 to 0.423; p < 0.001) more monthly outpatient encounters.Patients receiving NFFCCM services had lower MACE event rates of 7.4% (95% CI 7.1 to 7.8%; p < 0.001), all-cause mortality rate of 7.8% (95% CI 7.4 to 8.1%; p < 0.001), and heart failure rate of 0.3% (95% CI 0.2 to 0.5%; p < 0.001), respectively.

CONCLUSIONS AND RELEVANCE:
These findings suggest that reimbursement for NFFCCM was associated with the shifting high-cost utilization to lower-cost primary health care settings among patients with diabetes in Louisiana.

BACKGROUND
][12] CVD is the leading cause of morbidity and mortality and accounts for $37.3 billion in cardiovascular-related spending per year among patients with T2DM. 13According to the Behavioral Risk Factor Surveillance System from CDC, Louisiana has the 4th highest prevalence of diagnosed diabetes (14.1%), and the 10th highest rate of CVD (5.0%) and among all the states in 2020. 14lderly patients with diabetes covered by Medicare usually have multiple chronic conditions including diabetes. 15n order to improve chronic care management among elderly people with multiple chronic conditions, a new policy went into effect on January 1, 2015, under which non-face-toface chronic care management (NFFCCM) is reimbursable by the Centers for Medicare and Medicaid Services (CMS).The Current Procedural Terminology (CPT) code 99,490 for NFFCCM reimbursement went into effect in 2015.Codes 99,487 and 99,489 went into effect in 2017 to provide additional reimbursement for more time spent delivering NFF-CCM services. 16,175][26][27] Specifically, Schurrer et al. found that receipt of NFFCCM coded with 99,490 decreased the likelihood of admission for common chronic conditions and increased primary care utilization using Medicare claims. 27Thus, our study assessed the impacts of NFFCCM on healthcare utilization and health outcomes among Medicare beneficiaries with T2DM in Louisiana.

Data and Study Population
The study obtained the Medicare fee-for-service (FFS) claims for Louisiana beneficiaries from CMS, for 5 years between January 1, 2014, and December 31, 2018.The data included Medicare FFS claims submitted by professional providers (e.g., physicians, physician assistants, clinical social workers, nurse practitioners, institutional outpatient providers), and inpatient hospital providers.
We used the CMS Conditions Data Warehouse (CCW) condition algorithms to identify chronic conditions in our study. 28There are 27 chronic condition categories in the algorithms, including diabetes, heart failure, and stroke.
Type 2 diabetes was defined as at least one inpatient or two outpatient claims with the International Classification of Diseases, Ninth/Tenth Revision, Clinical Modification (ICD 9/10-CM) codes forT2DM.In our study, cardiovascular disease includes CHD, heart failure, and stroke, which were identified as at least one inpatient or outpatient claim with ICD 9/10-CM codes of the corresponding diagnoses.
Figure 1 presents a flow diagram of patient selection.Among 999,999 Medicare beneficiaries obtained from the Medicare fee-for-service claims data in Louisiana from 2014 to 2018, 376,929 beneficiaries were excluded since they had ever enrolled in Medicare advantage plan (Medicare Part C coverage) during the study period and their claims information may be incomplete since our data source is Medicare fee-for-service claims.Another 467,445 beneficiaries were excluded since they had type 1 diabetes or they did not have T2DM during the study period, and 36,982 patients were excluded since they had T2DM after the initiation date of receiving NFFCCM services.Therefore, the final analytical sample had 118,643 Medicare beneficiaries with T2DM and Medicare Part A and Part B coverage, including 8415 beneficiaries in the treatment group who received at least one NFFCCM service and 110,228 beneficiaries in the control group.

Study Design
The intervention group comprised beneficiaries who received NFFCCM under three billing codes (99,490, 99,487, and/or 99,489) in Medicare claims, and the dates of the first NFFCCM coded were the initiation dates.In the control group, initiation dates were randomly assigned for untreated beneficiaries based on the distribution of initiation dates in the treated population.The baseline period was defined as 12 months before the initiation date of NFFCCM.All covariates, including demographics, chronic conditions, and healthcare utilization, were identified during the baseline period.The follow-up period was defined as the time period from initiation date to beneficiary's death or the last date of Medicare enrollment or 12/31/2018.Outcomes were identified during the follow-up.
Healthcare Utilization.Healthcare utilization measures included the number of hospital admissions, emergency department (ED) visits, and outpatient (OP) visits per month.The number of hospital admissions was identified as the distinct admission and discharge date in inpatient claims.The ED claims in the Medicare FFS Outpatient and Inpatient files are identified via Revenue Center Code values of 0450-0459 (emergency room) or 0981 (professional fees-emergency room). 36The number of OP visits was identified as the distinct date of service in OP settings.Besides the frequency of healthcare utilization, we also assessed the continuity of care (COC) for primary care.][39][40] In our study, we counted primary care provider (PCP) visits, including general practitioners (CMS specialty code 01), family practitioners (CMS specialty code 08), internal medicine specialists without subspecialty training (CMS specialty code 11), physician assistants (CMS specialty code 97), and nurse practitioners (CMS specialty code 50).The COC index values range from zero (each visit made to a different provider) to one (all visits made to a single provider). 38

Statistical Analyses
The study compared outcomes in Medicare beneficiaries who received NFFCCM services and those who did not based on the presence or absence of reimbursement codes for NFFCCM.Propensity score weighting was used to balance baseline differences between the two groups.2][43][44] The indicators of trajectory groups were included as covariates in the propensity score weighting process to obtain a successful balance between treatment and control groups.The standardized mean difference (SMD) was used to measure the balance after propensity score weighting.
Multiple weighted linear or logistic regression models were developed using propensity score weights to assess the impact of NFFCCM on each outcome.To address the concerns about health systems selecting patients based on their likelihood of better survival, we conducted sensitivity analysis to assess the all-cause mortality in different time periods of follow-up.Multiple comparisons were adjusted by the conservative Bonferroni correction to the standard alpha-value of 0.05 by the number of comparisons. 45All analyses were performed using SAS 9.4 and Stata 15.1.A STATA plugin for package group-based trajectory balancing was used in the analyses. 46

RESULTS
Table 1 describes the baseline characteristics of the 118,643 Medicare beneficiaries.Group-based trajectory balancing and propensity score weighting methods were used to obtain comparable characteristics between the treatment (n = 8415) and control (n = 110,228) groups.All covariates were balanced between the treatment group and control group after propensity score weighting as the standardized mean differences between the two groups were all less than 0.1.For both groups, the mean age at first NFFCCM services was about 73 years old, 57.9% of both groups were females, nearly 28% were Black, and 2.1% were Hispanic.The most prevalent chronic condition among the study population was hypertension (98.2%).Other prevalent chronic conditions at baseline included arthritis and hyperlipidemia.In addition, 57.3% of the treatment group and 57.1% of the control group had CHD, while 46.8% of both treatment and control groups had MACE at baseline.The patients in the analytical sample had an average of 39.9 months of data before the initiation date of receiving NFFCCM services.The mean of monthly hospital admissions was 40 per 1000 patients, the mean of monthly ED visits was 80 per 1000 patients, and the mean of monthly outpatient visits was 255 per 1000 patients at baseline.The continuity of care index (COCI) for primary care providers was 0.46.
Table 2 presents that patients receiving any NFFCCM had reduced healthcare utilization compared with patients not receiving NFFCCM.The mean (SD) duration of followup was 1.46 (1.01) years.Compared with those without the NFFCCM services, patients receiving any NFFCCM had 0.012 (95% CI − 0.014 to − 0.011; p < 0.001) fewer monthly hospital admissions or 12 fewer hospital admissions out of 1000 patients, 0.017 (95% CI − 0.019 to − 0.016; p < 0.001) fewer monthly ED visits or 17 fewer monthly ED visits out of 1000 patients, and 0.399 (95% CI 0.375 to 0.423; p < 0.001) more monthly outpatient visits or 399 more outpatient visits out of 1000 patients.Receiving any NFFCCM was not associated with monthly outpatient utilization when NFFCCM encounters were excluded.In addition, receiving any NFFCCM was associated with a statistically significant higher COCI for primary care physicians of 0.094 (95%CI 0.091 to 0.097; p < 0.001).Furthermore, the four measures of health care utilization were significant at the p < 0.001 level, passing the adjusted Bonferroni correction for p-value of 0.01.
Table 3 presents the impacts of receiving any NFFCCM on five health outcomes with the adjusted Bonferroni correction of p-value of 0.01.Compared with those without the NFFCCM services, patients receiving the NFFCCM   the probability of all-cause mortality across all the time periods of follow-up.Specifically, there was a 2.4% (95% CI 2.3 to 2.6%; p < 0.001) decrease in all-cause mortality within 0-3 months after the index date of NFFCCM services.There was a 2.6% (95% CI 2.4 to 2.8%; p < 0.001) decrease in all-cause mortality 3-6 months after the index date.In the 6-12 months and 12 + months after initiating NFFCCM, there were 5.0% (95% CI 4.7 to 5.3%; p < 0.001) and 3.9% (95% CI 3.5 to 4.3%; p < 0.001) decreases, respectively, in all-cause mortality.

DISCUSSION
This study of the impact of reimbursement for NFFCCM services found more outpatient visits and fewer inpatient admissions and ED visits among patients with T2DM.Our core estimates indicate a decrease of 14.4 (0.012 × 12 × 100) inpatient admissions per 100 patients annually and a decrease of 20.4 (0.017 × 12 × 100) ED visits per 100 patients annually, which provides strong evidence to encourage the use of NFFCCM in diabetes care.Further, the increase in outpatient use in the NFFCCM group is due to the NFFCCM services, while all the other types of outpatient visits remain stable.
Our findings from Medicare claims showed a similar impact of NFFCCM on health care utilization as a CMSfunded study among Medicare beneficiaries. 27The study found reduced rates of ED visits and hospitalizations among the NFFCCM beneficiaries.But the previous findings were limited to only non-complex NFFCCM before the complex NFFCCM services were available for Medicare beneficiaries with at least two chronic conditions.Our findings further demonstrate the positive effects of NFFCCM, including both non-complex NFFCCM and complex NFFCCM, on decreasing hospitalization for diabetes patients with any other chronic diseases.
In addition, our results further demonstrate the impact of NFFCCM on health outcomes in diabetes.We found significantly lower cardiovascular event rates, while a risk reduction of 0.3% for both acute MI and heart failure may not be clinically meaningful, which were smaller than the treatment effects of newer diabetes medications, such as dapagliflozin observed in the clinical trial with a lower rate of 1.1%. 47The NFFCCM services include structured recording of patient health information, 24/7 access to providers, and comprehensive care management, which are free of side effects and may be more cost-effective and acceptable to patients. 17The reduction of MACE and mortality rates could be associated with the better control of blood glucose, blood pressure, and low-density lipoprotein outcomes. 48Further research may benefit from assessing cardiovascular death or including the reason of death.The combination of these findings could be used to support reimbursement of NFFCCM services, and we provide evidence for practitioners and patients to utilize NFFCCM in diabetes care.Policymakers could take into consideration covering more NFFCCM services or other remote services of CCM in the future.
There is an important issue of whether the NFFCCM services were additions to other outpatient care or substitutes because of its implications for cost and administrative complexity.We found there were no significant increases in all the other outpatient visits in Medicare claims analysis.Our finding based on claims records suggested that the NFFCCM services were additive to relatively constant outpatient visits, indicating that NFFCCM participation may be a marker of patient commitment to self-care.
Our study also found an improvement in continuity of primary care, which is often considered a core value of patient care in primary care medicine. 49The increased COCI suggests that NFFCCM beneficiaries may have been encouraged to see the same primary care physician over time, which may have improved health outcomes as a result of having one person managing their healthcare needs. 49his study has several limitations.The major limitation of this study is that we do not have random assignment of NFFCCM.Selection into treatment may be initiated by clinicians that think patients may benefit from NFFCCM either because of the severity of their disease or likelihood of complying to the intervention and thus potentially introduce confounding by indication.If this clinical judgment is based on unobservable characteristics in our study, we may have bias in our estimates.Our study might be influenced by selfselection bias, where patients already committed to self-care are more likely to participate, then the results would overestimate the impacts of NFFCCM.To mitigate potential bias from lack of randomization, we used trajectory balancing and propensity score weighting to balance the NFFCCM and the control groups on measured variables, while unmeasured confounders could still lead to biased results.The patient population created by the propensity score weighting was different from Medicare population, including having overall more chronic conditions, more women, fewer individuals over 85 years, and less Alzheimer's disease, which would yield the generalizability to the broader Medicare population.Additionally, we lack physician and organizational level factors and other socioeconomic factors, such as education and income, to address potential confounding.
Second, Medicare claims data does not have clinical biomarkers that reflect the impacts of NFFCCM on glycemic control and other important biomarkers in diabetes care.Thus, the direct effects of NFFCCM on some clinical outcomes cannot be analyzed in this study due to the data limitations.Future study could improve the data completeness by using a linked Medicare FFS claims to EHR.In our study's data, we did not have hospice information to better adjust for patients' general condition, which is a possible mechanism contributing to the high mortality reduction.We assessed the impacts of NFF-CCM on all-cause mortality at different follow-up times in order to address this issue, and found significant reductions in mortality in 6-12 months, as well as > 12 months after excluding those who died in 0-6 months, who might have used hospice services.We used logistic regression to assess mortality incidence as the relatively short follow-up duration might not be critical for a non-medication intervention; however, future studies would benefit by applying survival analyses with longer follow-up data.These factors limit the generalizability of our results, and stronger evidence of the impact of NFFCCM may be found in a larger database with more NFFCCM recipients and clinical measurements.

CONCLUSIONS
Our study is the first real-world study using Medicare claims data to present evidence that the reimbursement of NFFCCM, complex or non-complex, was associated with the shifting high-cost health utilization to low-cost primary health care setting among patients with diabetes in Louisiana.Policymakers could consider covering more NFFCCM services or other remote services of CCM in the future.Incorporating modern technology for disease management could be coupled with NFFCCM to provide significant clinical benefits.

Figure 1 Louisiana
Figure 1 Louisiana Medicare population with T2DM selection.

Table 1 Baseline Characteristics Before and After Propensity Score Weighting For Patients with Type 2 Diabetes SMD
, standardized mean difference; NFFCCM, non-face-to-face chronic care management; MI, myocardial infarction; MACE, major adverse cardiovascular events; ED, emergency department The four measures of health outcomes were significant at the p < 0.001 level, passing the adjusted Bonferroni correction for p-value of 0.01.Table4presents the impacts of receiving any NFFCCM on all-cause mortality in different time periods of follow-up as sensitivity analyses.Receiving any NFFCCM decreased

Table 2 Impacts of Receiving Any NFFCCM Services on Health Care Utilization *
p < 0.05, **p < 0.01, ***p < 0.001.The number of inpatient visits per month = total number of inpatient visits / the number of corresponding follow-up months.We controlled for (1) demographic characteristics, including age, sex, race, and ethnicity; (2) chronic disease status, including hypertension, Alzheimer's disease, arthritis, asthma, atrial fib, autism, cancer, COPD, ESRD, depression, heart failure, hyperlipidemia, chronic heart disease, osteoporosis, and stroke; (3) other outcomes before initiation date; and (4) outpatient trajectory balancing groups' score, including three outpatient trajectories.Applying the conservative Bonferroni correction, the significant p-values are below 0.05/5 = 0.01.Coefficients are followed by 95% CI and p-value.The mean (SD) duration of follow-up was 1.46 (1.01) years

Table 3 Impacts of Receiving Any NFFCCM Services on Health Outcomes
The number of inpatient visits per month = total number of inpatient visits / the number of corresponding follow-up months.We controlled for (1) demographic characteristics, including age, sex, race, and ethnicity; (2) chronic disease status, including hypertension, Alzheimer's disease, arthritis, asthma, atrial fib, autism, cancer, COPD, ESRD, depression, heart failure, hyperlipidemia, chronic heart disease, osteoporosis, and stroke; (3) other outcomes before initiation date; and (4) outpatient trajectory balancing groups' score, including three outpatient trajectories.Applying the conservative Bonferroni correction, the significant p-values are below 0.05/5 = 0.01.Coefficients are followed by 95% CI and p-value.The mean (SD) duration of follow-up was 1.46 (1.01) years *p < 0.05, **p < 0.01, ***p < 0.001.

Table 4 Impacts of Receiving Any NFFCCM on All-Cause Mortality in Different Time Periods of Follow-up
*p < 0.05, **p < 0.01, ***p < 0.001.The number of inpatient visits per month = total number of inpatient visits / the number of corresponding follow-up months.We controlled for (1) demographic characteristics, including age, sex, race, and ethnicity; (2) chronic disease status, including hypertension, Alzheimer's disease, arthritis, asthma, atrial fib, autism, cancer, COPD, ESRD, depression, heart failure, hyperlipidemia, chronic heart disease, osteoporosis, and stroke; (3) other outcomes before initiation date; and (4) outpatient trajectory balancing groups' score, including three outpatient trajectories.Coefficients are followed by 95% CI and p-value.The mean (SD) duration of follow-up was 1.46 (1.01) years