The Impact of a Remote Patient Care Program on Health Care Costs and Utilization Among Medicare Patients With Chronic Disease

Abstract

Remote patient monitoring coupled with technology-enabled, guideline-directed clinical care—or remote patient care (RPC)—has consistently led to improved outcomes for Medicare patients with chronic diseases. However, the ability for RPC to drive reductions in total cost of care and health care utilization is limited. We sought to determine whether an RPC program can reduce health care costs and utilization. Using patient-level Medicare claims data, a difference in difference analysis was conducted to assess the impact of an RPC program compared with a propensity score–matched control group on total health care costs and resource utilization over a 12-month period following program activation. The retrospective analysis included patients enrolled into an RPC program from July 1, 2022 to October 31, 2023 from primary care and cardiology clinics across 15 states. The RPC program included a group of clinicians who monitored and triaged vitals and conducted clinical visits using standardized clinical protocols to facilitate guideline-directed clinical interventions. We compared 5872 patients enrolled in an RPC program to 11,449 eligible propensity score–matched control patients. RPC resulted in a statistically significant reduction in total cost of care (−$1302 per patient per year; P<.01), which was driven primarily by a reduction in inpatient costs (−$1428 per patient per year; P<.01). Patients enrolled in the RPC program also had a lower rate of hospitalizations (−23 vs +41/1000 patients/y; 27% reduction; P<.01). These data highlight the potential for a nationwide RPC program to lead to significant cost savings and a reduction in health care utilization among Medicare patients at scale.

Health care cost related to cardiovascular disease (CVD), including conditions such as hypertension and diabetes, are expected to triple between 2020 and 2050.¹ Among adults 65 to 79 and 80+ years with CVD, health care cost and productivity losses are projected to increase by 140% (∼550 billion vs ∼$230 billion) and 371% (∼$600 billion vs ∼$125 billion), respectively.¹ In the United States, the annual cost for a patient with heart failure (HF) is ∼$30,000, with total costs exceeding $160 billion by 2030, primarily driven by hospitalizations and readmissions.² To address rising costs and unnecessary health care utilization associated with uncontrolled chronic CVD conditions, solutions aimed at optimizing control are needed.

Suboptimal control of blood pressure and blood glucose in individuals with hypertension and diabetes is costly at an estimated total of $2500 and $3761 per patient per year, respectively.3, 4, 5, 6 Meanwhile, optimization of guideline-directed medical therapy for HF is proven to be cost-effective.⁷ Remote patient monitoring (RPM) coupled with technology-enabled, guideline-directed clinical care, or remote patient care (RPC) is a scalable solution aimed at improving clinical outcomes while curbing health care costs and utilization.⁸ Prior analyses regarding the cost savings associated with RPM have been mixed, in part owing to large heterogeneity in the study design and type of RPM program.9, 10, 11, 12 To address ongoing uncertainty, this cohort study using patient-level Medicare claims data assessed the impact of a nationwide RPC program in collaboration with patients’ longitudinal clinicians vs usual care on health care costs and utilization.

Methods

The RPC program included a multidisciplinary group of clinicians, namely, nurse practitioners, registered nurses, and medical assistants, who leveraged RPM to deliver RPC to Medicare patients with chronic diseases, including hypertension, diabetes, and HF (according to International Classification of Disease-10 codes), alongside health care institutions and patients’ longitudinal providers. Further details about the program have been previously published.¹³

Briefly, the patient experience is initiated after an order is placed for RPM in the electronic health record based on an existing diagnosis by the longitudinal clinician. Patients then complete a virtual enrollment, including patient consent. Once the cellular-enabled devices are received by patients, they are instructed to take their program specific vital, including blood pressure and heart rate for hypertension, blood pressure, heart rate and weight for HF, and blood glucose for diabetes, daily. A remote clinical team monitored and triaged self-measured vitals and conducted virtual clinical visits using standardized clinical protocols to facilitate guideline-directed clinical interventions, including symptom, vital and lifestyle, and medication optimization, with the goal of achieving program specific clinical goals (ie, monthly average home-based blood pressure <130/80 mm Hg for hypertension patients). Technological support is provided by leveraging guideline-based artificial intelligence-assisted algorithms designed to prioritize evaluation and treatment of patients experiencing high-acuity clinical alerts and symptoms, as well as nudge clinicians when interventions could be considered to help patients achieve their clinical goals.

To be included in the treatment group, enrolled patients needed at least 6 months of Medicare eligibility before taking their first program vital (ie, activation), have at least 6 months of exposure time post activation, and prior to May 2024, match on various factors to patients in the Medicare claims database Virtual Research Data Center, have taken their first vital before November 2023, and have a matched control member with a propensity score within 0.4 SDs. A control group included non enrolled, RPC program-eligible patients who were propensity matched according to demographic, clinical, cost, and utilization data. A logistic regression model was used to calculate propensity scores, incorporating variables shown in Table 1.

Table 1.

Baseline Characteristics

	Total population			Rural/underserved population
	Treatment group, n=5872	Control group, n=11,449	P	Treatment group, n=2882	Control group, n=5599	P
Age (y)	74	74	.32	74	74	.80
Sex, n (%)
Female	3406 (58)	6526 (57)	.43	1700 (59)	3191 (57)	.06
Race, n (%)
White	5167 (88)	10,190 (89)	.19	2499 (87)	4847 (87)	.41
Remote patient care program enrolled/eligible condition, n (%)
Diabetes	447 (8)	876 (8)	>.99	236 (8)	457 (8)	>.99
Heart failure	897 (15)	1745 (15)		398 (14)	785 (14)
Hypertension	3936 (67)	7671 (67)		1945 (67)	3766 (67)
Hypertension + diabetes	592 (10)	1157 (10)		303 (11)	591 (11)
Risk score, mean	1.1	1.1	.97	1.1	1.1	.69
Comorbidities, n (%)
Heart disease	2173 (37)	4179 (37)	.42	1038 (36)	2016 (36)	.74
Cerebrovascular disease	528 (9)	1030 (9)	.84	259 (9)	560 (10)	.22
Atrial fibrillation	1116 (19)	2175 (19)	.65	548 (19)	1064 (19)	.65
Pulmonary disease	1820 (31)	3549 (31)	.44	922 (32)	1736 (31)	.66
Diabetes	294 (5)	572 (5)	.95	144 (5)	280 (5)	.49
Cancer	763 (13)	1488 (13)	.60	375 (13)	784 (14)	.45
Autoimmune disease	2819 (48)	5381 (47)	.54	1383 (48)	2576 (46)	.13
Behavioral health	1233 (21)	2290 (20)	.11	634 (22)	1176 (21)	.15
CKD	1703 (29)	3320 (29)	.27	836 (29)	1568 (28)	.55
Liver disease	117 (2)	229 (2)	.62	58 (2)	112 (2)	.46
Substance use	176 (3)	343 (3)	.76	86 (3)	168 (3)	.99
Health care costs ($; per patient per mo)
Total costs	838	847	.76	828	803	.58
Inpatients costs	264	249	.44	280	276	.89
ED costs	23	23	.96	24	25	.59
Outpatient costs	205	218	.29	189	199	.54
Professional service costs	339	323	.17	327	278	<.01
Skilled nursing facility costs	6	33	<.01	8	25	<.01
Health care utilization (×/1000 patients/y)
Inpatient admits	235	238	.80	240	257	.39
ED visits	450	462	.61	443	467	.39
Outpatient visits	4990	5083	.37	5098	5091	.96
Professional service visits	22,574	19,370	<.01	22,085	17,729	<.01
Skilled nursing facility days	134	831	<.01	172	607	.02

Values are n (%) unless specified.

CKD, chronic kidney disease; ED, emergency department.

Medicare claims data were collected between January 2022 and April 2024. The primary study outcomes were the change (difference in difference) in per patient per month health care costs and hospital admission frequency per 1000 patients per year over a 6- and 12-month period after program activation. The pre-period for the treatment group was defined as the 6 months prior to the program activation date and the post-period was defined as the 6 and 12 months afterward. The control group was then measured over the same period as the treatment group, using the 6 months prior to the matched treatment group activation date as the pre-period and after activation as the post-period. If a patient did not have a full 12 months of exposure, their per month averages up until the end of their exposure or data availability were used. Significance for the statistical analysis was defined as a P value of <.05. Rural and underserved areas were defined per the Health Resources and Services Administration and Federal Office of Rural Health Policy and Federal Housing Finance Agency standards, respectively.

Results

A total of 5872 patients enrolled in 1 of 3 available RPC programs were compared with a control group of 11,449 individuals. The average time of available data after activation was 10 (SD, 2) months for both the intervention and control groups. Baseline data are highlighted in Table 1, demonstrating similar demographic factors, risk scores, health status, and health care cost and utilization across the intervention and control groups.

After 12 months, patients enrolled in the RPC program had a reduction in the total cost of care of $108.50 per patient per month (PPPM) compared with the control group, inclusive of the cost of RPM (−$34.60 vs +$74.00; P<.01) (Supplemental Tables 1 and 2, available online at http://www.mcpiqojournal.org). Most cost savings were driven by a reduction in inpatient spend, which was $119.00 less PPPM for individuals enrolled in the RPC program compared with the control group over a 12-month period (−$30.30 vs +$88.70; P<.01). The RPC program also had a 27% relative reduction in inpatient admissions over a 12-month period vs the control group (−23/1000 vs +41/1000; P<.01). Improvements in cost and utilization related to enrollment in the RPC program were similar among enrolled patients from rural or underserved areas (Table 2).

Table 2.

Health Care Costs and Utilization

	Total population						Rural/underserved population
	Baseline	After 6 mo	After 12 mo	Difference (12 mo)	DiD	P	Baseline	After 6 mo	After 12 mo	Difference (12 mo)	DiD	P
Health care costs
Total costs ($), per patient per mo
Treatment group	838.00 (1834)	761.60 (1641)	803.50 (1583)	−34.60	−108.50	.003	827.70 (1978)	745.60 (1644)	785.00 (1603)	−42.70	−103.20	.03
Control group	846.60 (1951)	938.10 (2567)	920.60 (2295)	74.00			802.50 (1927)	865.30 (2307)	863.10 (2260)	60.50
Inpatient costs ($), per patient per mo
Treatment group	264.10	209.90	233.80	−30.30	−119.00	<.01	280.10	207.50	233.80	−46.40	−95.60	<.01
Control group	249.10	337.50	337.80	88.70			275.70	316.00	324.90	49.20
Emergency department costs ($), per patient per mo
Treatment group	23.20	23.40	24.20	1.10	0.90	.82	23.80	23.70	25.00	1.20	3.60	.37
Control group	23.20	23.00	23.40	0.20			24.90	23.40	22.50	−2.40
Outpatient costs ($), per patient per mo
Treatment group	205.40	164.40	177.30	−28.10	4.80	.92	189.20	152.30	163.80	−25.40	−8.10	.39
Control group	218.00	195.20	185.10	−32.90			198.50	186.10	181.30	−17.30
Professional service costs ($), per patient per mo
Treatment group	339.40	335.40	335.20	−4.20	−3.80	.57	326.90	322.80	318.00	−9.00	−7.90	.51
Control group	323.10	335.40	322.70	−0.40			278.20	284.00	277.20	−1.10
Skilled nursing facility costs ($), per patient per mo
Treatment group	5.90	28.50	33.00	27.00	8.60	.43	7.60	39.40	44.40	36.80	4.80	.91
Control group	33.30	47.00	51.70	18.40			25.20	55.90	57.30	32.10
Health care utilization
Inpatient admits/1000
Treatment group	235	199	212	−23	−64	<.01	240	196	203	−37	−50	.02
Control group	238	278	280	41			257	275	271	13
Emergency department visits/1000
Treatment group	450	445	457	6	39	.25	443	475	483	41	96	.05
Control group	462	434	429	−33			467	434	412	−55
Outpatient visits/1000
Treatment group	4990	5322	5416	426	887	<.01	5098	5684	5827	730	1306	<.01
Control group	5083	4695	4622	−461			5091	4609	4515	−576
Professional service visits/1000
Treatment group	22,574	29,542	28,812	6238	7008	<.01	22,085	28,104	27,404	5319	6440	<.01
Control group	19,370	18,916	18,600	−769			17,729	17,093	16,608	1121
Skilled nursing facility d/1000
Treatment group	134	621	681	546	158	<.01	172	832	860	688	14	<.01
Control group	831	1153	1219	389			607	1376	1280	674

DiD, difference in difference.

The reductions in inpatient costs and admissions were primarily driven by a decrease in hospitalizations related to HF (−$10.02 PPPM; 64% decrease in hospitalizations), sepsis/infection (−$21.37 PPPM; 57% decrease in hospitalizations), cardiac arrhythmias (−$11.54 PPPM; 27% decrease in hospitalizations), and stroke (−$15.56 PPPM; 71% decrease in hospitalizations). However, when admitted, individuals enrolled in the RPC program had a shorter length of stay in the hospital (Table 3).

Table 3.

Inpatient Admission Cost and Utilization Savings

	Treatment group			Control group			DiD
	Baseline	After 12 mo	Difference	Baseline	After 12 mo	Difference
Inpatient per patient per month ($)
Heart failure	13.35	11.11	−2.24	11.16	18.94	7.78	−10.02
Cardiac arrhythmias	14.92	8.31	−6.61	10.00	14.93	4.93	−11.54
Stroke and related complications	17.06	9.41	−7.65	8.06	15.97	7.91	−15.56
Infection/sepsis	27.31	20.61	−6.70	23.96	38.63	14.67	−21.37
Other	191.51	184.39	−7.12	195.91	249.33	53.42	−60.54
Average length of stay (d)
Heart failure	5.9	5.8	−0.1	4.8	5.9	1.1	−1.2
Cardiac arrhythmias	4.4	3.8	−0.6	3.5	4.3	0.8	−1.4
Stroke and related complications	33.4	28.8	−4.6	26.2	30.3	4.1	−8.7
Infection/sepsis	11.7	9.2	−2.5	12.2	13.3	1.1	−3.6
Other	13.4	20	6.6	35.2	66.3	31.1	−24.5
Inpatient admissions (×/1000 patients/y)
Heart failure	16.7	12.6	−4.1	15.0	21.7	6.7	−10.8
Cardiac arrhythmias	14.0	10.0	−3.9	11.4	11.1	−0.2	−3.7
Stroke and related complications	14.0	6.8	−7.1	7.3	10.1	2.8	−9.9
Infection/sepsis	21.5	18.4	−3.1	18.4	27.6	9.2	−12.3
Other	168.8	164.1	−4.7	186.2	209.2	23.0	−27.6

DiD, difference in difference.

Discussion

In this retrospective cohort study, when compared with RPC program-eligible, nonenrolled, propensity score–matched controls, a patient enrolled in the RPC program had an average total savings per year, inclusive of RPM costs, of $1302. The RPC program also resulted in a 27% reduction in hospital admissions at 12 months for common conditions frequently affecting Medicare patients with HF, hypertension, and diabetes. The cost and utilization impact of this RPC program extended to rural and underserved communities where access and affordability negatively affect quality of care.

Despite multiple studies attempting to demonstrate the ability for RPM to drive reductions in health care costs and utilization, the results have varied. One of the largest systematic reviews to date considered 34 RPM studies across 12 countries.¹⁴ Although investigators attempted to show the economic benefit of RPM across multiple chronic conditions, they ultimately were able to conclude that RPM was cost-effective for all patients with hypertension and varied according to disease severity for patients with HF. There was limited evidence for patients with diabetes.

A major analytical limitation noted by the authors was that the cost-effectiveness of RPM programs varies according to multiple factors, most importantly, the variability of the RPM program offered, which impacts the quality of clinical care as well as clinical outcomes, health care costs, and utilization. Although the current analysis involved a diverse patient population across 15 states, including a significant proportion from rural or underserved areas, it ensured a standardized RPC program for all patients according to their chronic disease. Such standardization helped enable a consistent effect on the primary outcomes—both clinical and financial—for all patients exposed to the intervention.

In 2023, a separate analysis sought to estimate RPM’s effect on hypertension care and spending.¹⁵ Tang et al¹⁵ reported that patients exposed to high- vs low-RPM practices had improved hypertension care outcomes, including fewer hypertension-related acute care encounters, at the expense of increased spending (+$274). The authors ultimately concluded that RPM should be considered for a limited group of patients and reimbursed only so long as clinical value is provided.

There are several important factors to consider before drawing conclusions from the analysis by Tang et al¹⁵ regarding the impact of RPM on health care costs and utilization including incomplete capture of cost savings, variable intensity of RPM services, and secular trends in pandemic-related telehealth. By limiting the cost analysis to hypertension-related spending, Tang et al¹⁵ may have underestimated potential cost reduction. Our study demonstrated reductions in cost outside of cardiovascular- or diabetes-related hospitalizations. For example, costs related to sepsis significantly decreased with support of a 24/7 registered nurse team that monitored regular vital readings and likely initiated earlier intervention as blood pressure and heart rate abnormalities occurred in the setting of infection. Intensity of RPM services may also contribute to the findings of Tang et al.¹⁵ Because a high-RPM practice was defined as any practice where 25% or more of patients received RPM, up to 75% of the high-RPM group may not have even received clinical care as part of an RPM program. Third, a significant percentage of the reported increase in hypertension-related costs are due to RPM costs, which increased by $355 during the COVID-19 pandemic (2019-2021) when individuals were more likely to seek remote clinical care regardless of the quality of care provided.

There are several limitations of the current analysis that must be considered. Although an eligible propensity score–matched control group was used to minimize confounding, we were unable to control for all differences that exist between individuals who chose to enroll vs not enroll in an RPC program. Second, a propensity matching threshold of 0.4 was used to not only maximize both the population studied but also minimize the baseline differences between treatment and control groups. This higher threshold may have resulted in the significant difference in baseline costs and utilization of professional services and skilled nursing facilities between treatment and control groups, although the difference in difference was not significant for either with regards to health care costs. Third, although only 6 months of pre-RPM claims were available to determine a baseline, cost savings were similar when using 6- or 12-month post-RPM claims data. Lastly, health care costs and utilization outcomes were not stratified by RPC program engagement and retention.

Conclusion

In the United States, the number of Medicare patients is growing at an unprecedented rate due to an aging population, which poses significant clinical and economic risk for our health system. Solutions aimed at improving control of chronic disease and curbing health care costs and utilization must be effectively implemented. In this retrospective cohort study, an RPC program enrolling Medicare patients with hypertension, HF and diabetes resulted in lower health care costs, owing to reductions in hospital admissions and associated inpatients costs. The impact also extends to those rural and underserved areas, where social determinants of health are most profound and negatively affect access, quality, and affordability of care.

Potential Competing Interests

Drs Feldman and Fudim are advisers at Cadence; Mr Reynolds, and Drs Babikian, Cunningham, Feldman and Curnow are employees at Cadence; Ms Zheng and Mr Budhiraja are employees at humbi.ai. The other authors report no competing interests.