Cost‐effectiveness of home blood pressure telemonitoring and case management in the secondary prevention of cerebrovascular disease in Canada

Abstract

Home blood pressure (BP) telemonitoring and pharmacist case management reduce BP, but cost‐effectiveness assessments are mixed. We examined the incremental cost‐effectiveness of this intervention vs usual care in Canadians with cerebrovascular disease. A Markov decision model cost‐utility analysis examining community‐residing, high‐risk patients with a recent nondisabling cerebrovascular event was created. A lifetime time horizon and health care payer perspective were used. Achieved BP, future cardiovascular risks, and attendant consequences on quality‐adjusted life years and Canadian dollar costs were modeled. BP telemonitoring was assumed to occur for 3 months, then quarterly. Life tables were used to determine overall mortality, adjusted by cardiovascular disease mortality. Relative efficacies of intervention‐associated BP lowering, resource use, and costs were obtained from Canadian published literature. Reduction in systolic BP of 9.7 mmHg was used in the base case; subsequently, robust sensitivity analyses were conducted. The results showed that, over the lifetime horizon, telemonitoring with case management led to net health care savings of $1929 Canadian and increased per‐patient QALYs by 0.83. These findings were robust to sensitivity analysis, with the intervention remaining dominant or highly cost‐effective. Increasing telemonitoring costs by 50% still resulted in the intervention being dominant; if the costs of telemonitoring plus case management were 2‐3 times base case cost, incremental cost‐effectiveness was $1200‐$4700 per quality‐adjusted life year gained. In conclusion, home BP telemonitoring and pharmacist case management poststroke lowered costs and improved QALYs. Strategies and funding for broad implementation of this dominant strategy should be implemented.

1. INTRODUCTION

Globally, cerebrovascular disease (CVD) is a leading cause of chronic disability, dementia and, death.1 Individuals who suffer a cerebrovascular event have a high recurrence risk;2, 3 uncontrolled blood pressure (BP) is the leading cause of stroke recurrence.4 Treatment of elevated blood pressure postcerebrovascular event reduces the risk of subsequent strokes, stroke deaths, and cardiovascular deaths.5 A 5 mmHg BP reduction in stroke survivors is estimated to reduce recurrent stroke by 10% and cardiovascular death by 25%.5

Contemporary hypertension guidelines strongly endorse the use of home BP monitoring, based on its superior prognostic performance relative to office BP measurements and its benefits with respect to improving patient activation and self‐monitoring.6, 7, 8 Home BP telemonitoring, which consists of tele‐transmitting BP measurements to a web portal for review and use by health care providers, is the best way to perform home BP monitoring because it automates the telemonitoring process and eliminates biases inherent in self‐reporting of readings.9, 10 Telemonitoring can also be combined with clinical decision support, self‐management tools, algorithmic care, and/or case management (usually performed by a nurse or pharmacist) to improve the management of high BP.9, 10 The use of home BP telemonitoring, especially when combined with case management, leads to clinically important BP reductions.11, 12 In a meta‐analysis of 46 randomized controlled trials (13 875 subjects), home BP telemonitoring resulted in clinically important BP reductions (4.9 mmHg systolic in the subgroup of patients with hypertension), with greater BP reductions reported in studies that combined BP telemonitoring with case management (ie, counseling, education, medication management, protocolized medication titration).11 The benefits of home BP telemonitoring and case management primarily result from increased self‐monitoring and medication intensification.13, 14 Despite these benefits, there has been little adoption of this intervention in contemporary Canadian clinical practice.9

A potential barrier to adoption of home BP telemonitoring and case management is uncertainty about the cost and cost‐effectiveness. Economic assessments evaluating this intervention have demonstrated mixed results, ranging from cost savings to per‐patient costs of nearly $10 000 United States (US) dollars.15 Major limitations of existing studies are that they assessed cost‐effectiveness over a short time frame, without incorporating the long‐term benefits of BP control on cardiovascular disease prevention and subsequent effect on quality of life; further, they did not focus on high‐risk patient populations. A recent US analysis demonstrated that achieving BP control in high‐risk individuals, including those with prior cardiovascular disease, is actually a cost‐saving intervention, an uncommon occurrence in contemporary health care.16 Although this analysis was not focused specifically on examining BP telemonitoring and case management, it supports the premise that this intervention could be cost‐effective in high‐risk patients. Nevertheless, costs and cost‐effectiveness in the Canadian health care setting are uncertain.9

With increasing emphasis being placed on implementation of cost‐saving health care interventions that simultaneously improve care quality and reduce health care expenditures,17 we sought to examine the long‐term cost‐effectiveness of home BP telemonitoring and pharmacist case management in Canadian stroke survivors. Given the high‐risk nature of this patient population and the relatively low cost of implementing telemonitoring and case management in this country, we hypothesized that this intervention could be a dominant strategy when modeled over a patient's lifetime.

2. METHODS

A cost‐utility analysis using a Markov decision model was created based on a previously published model, in terms of both model structure and inputs, with modification to the poststroke and Canadian context.18 Patients started in the model with their baseline health state reflecting a recent minor stroke or TIA, and residing in their own residence (ie, not requiring nursing home level of care). The baseline characteristics of the cohort were based on the Canadian Preventing Recurrent Vascular Events and Neurological Worsening through Intensive Organized Case Management (PREVENTION) randomized controlled trial enrolling 279 patients with a recent minor cerebrovascular event (mean age 67.6 years, 58% men).19 Patients remained in their health state if they did not have a subsequent event or die. Suffering an event caused a patient to move into one of four possible health states: new stroke, myocardial infarction (MI), unstable angina (UA), or dead (Figure 1). Patients suffering an event other than death remained in these health states, and additional events were not modeled. Probabilities of suffering each event on an annual basis (Table 1) and health utilities, which measure the strength of an individual's preference for a given health state (Table 2), were determined from the published literature. Canadian life tables were used to determine overall mortality dependent on age and sex, adjusted by CVD mortality.

Figure 1.

Model overview. UA, unstable angina; TIA, transient ischemic attack; MI, myocardial infarction

Table 1.

Event probabilities, utilities, and efficacy inputs used in model

Parameter	Value	Probability distribution	Source
Annual event probabilities (incidence)
Stroke
60‐69 y old	0.0348		PROGRESS [26]; NICE [40]
70‐79 y old	0.0589
80‐89 y old	0.0713
Myocardial infarction and unstable angina
60‐69 y old	0.0139		PROGRESS [26]; NICE [40]
70‐79 y old	0.0232
80‐89 y old	0.0232
Probability of death from an event
Fatal stroke and MI	0.23	Beta	Bamford [41]; ONS [42]
Utility values for health states
Initial health state (recent cerebrovascular event)	0.84	Beta	Prevention [19]
Recurrent stroke	0.420		Wang [22]
Unstable angina	0.709		Sullivan [23]
Myocardial infarction	0.725		Sullivan [23]
Relative efficacy model inputs
Base case
Efficacy of home BP telemonitoring and pharmacist case management vs usual care	Incremental systolic BP reduction of 9.7 mmHg	Normal	Margolis [20]; Law [21]
Age‐related relative risk of stroke at 12 mo with telemonitoring and case management	0.67
Age‐related relative risk of myocardial infarction and unstable angina at 12 mo with telemonitoring and case management	0.80
Sensitivity analysis—lower efficacy assumptions
Efficacy of home BP telemonitoring and pharmacist case management vs usual care	Incremental systolic BP reduction of 4.9 mmHg	Normal	Duan [11]; Law [21]
Age‐related relative risk of stroke at 12 mo with telemonitoring and case management	0.82
Age‐related relative risk of myocardial infarction and unstable angina at 12 mo with telemonitoring and case management	0.89
Sensitivity analysis—high efficacy assumptions
Efficacy of home BP telemonitoring and pharmacist case management vs usual care)	Incremental systolic BP reduction of 15 mmHg	Normal	Expert opinion; Law [21]
Age‐related relative risk of stroke at 12 mo with telemonitoring and case management	0.54
Age‐related relative risk of myocardial infarction and unstable angina at 12 mo with telemonitoring and case management	0.72

Table 2.

Model cost inputs

Parameter	Cost explanation	Cost	Probability distribution	Source
Telemonitoring initial cost per patient—first 3 mo
Total cost		$305	Triangular (±50%)	Expert opinion, Alberta Health Services pharmacist rates; Alberta Health Care Insurance Plan Medical Price List (Jan 2017)
a. Pharmacist: 1 in person initial consultation (1 h), 3 remote follow‐up visits (30 min each)	Pharmacist wage: $56.37 per hour total cost (2.5 h) = $141
b. Physician: 1 multidisciplinary discussion to review case with pharmacist	Alberta health billing code 03.01NM = $17
c. BP device cost	$350 amortized over 3 y = $117 per y
d. Data cost	$10/mo = $30 (3 mo)
Telemonitoring subsequent cost per patient—annual
Total cost		$327	Triangular (±50%)	Expert opinion, Alberta Health Services pharmacist rates; Alberta Health Care Insurance Plan Medical Price List (Jan 2017)
a. Remote pharmacist follow‐up comprised of 4 quarterly remote visits = 1 h total time	$56.37 per h = $56.37
b. Pharmacist physician conference to review patient every 6 mo	03.01NM billing code every 6 mo = $34
c. Data cost	$120 per y
d. Replacement cycle for BP device (every 3 y)	$117 per y
Usual care initial cost per patient—first 3 mo
Total cost		$186	Triangular: 2‐4 GP F/U visits
a. Initial visit to GP (30 min)	Alberta Health billing codes 03.03A + CMGP02 = $73.89
b. Three follow‐up visits	Alberta Health code 03.03A × 3 = $112
Usual care ‐ subsequent annual cost per patient
2 GP visits, 30 min each	Alberta Health billing code (03.03A + CMGP02) × 2 = $148	$148	Triangular: 1‐9 GP visits
Additional annual drug costs
Incremental use of drugs in telemonitoring arm	0.4 more medications based on an annual cost of $38.53	15		Duan [11]; Klarenbach [43]; Alberta Drug Formulary
Costs of acute disease
Stroke	79 925 (2015$)	82 457	Gamma	Marra [44]
MI	11 511 (2015$)	11 876
UA	3764 (2015$)	3883
Costs of long‐term disease
Stroke	12 126 (2015$)	12 510	Gamma	Marra [44]
MI	3367 (2015$)	3474
UA	3764 (2015$)	3883

All costs not in $2017 were inflated using the all‐items Consumer Price Index.

Home BP telemonitoring with pharmacist case management was compared to usual care. Home BP telemonitoring was assumed to occur monthly for the first 3 months (four times daily for 1 week each month), then quarterly (four times daily for 1 week each quarter) thereafter, in accordance with Canadian Hypertension Clinical Practice Guideline recommendations, for the duration of the 20‐year time horizon.6 The time required for a pharmacist case manager to review each week of BP measurements was tabulated. Usual care was assumed to consist of in‐office BP measurements and follow‐up by a family physician.

In the base case analysis, published data from a recently conducted RCT of telemonitoring and pharmacist case management were used to derive an expected systolic BP reduction relative to usual care—estimated at 9.7 mmHg over 1 year (Table 3).20 Notably, this degree of systolic BP reduction is consistent with the 8.4 mmHg reduction found with pharmacist case management in the PREVENTION trial, which was performed in an Albertan setting.19 The baseline characteristics of the PREVENTION and Margolis trials were relatively similar in terms of demographics, but different in terms of comorbidities—poststroke patients were enrolled in PREVENTION whereas, in the Margolis trial, only 10% of subjects had prior cardiovascular disease. In sensitivity analysis, BP reductions of 4.9 mmHg and 15 mmHg were used (Table 3).11 Published meta‐analytic data summarizing the cardiovascular benefits of BP lowering on clinically important cardiovascular end points and mortality were then used to estimate the expected cardiovascular benefits corresponding to the BP reductions resulting from telemonitoring and case management (Table 3).21 The duration of difference in achieved BP between arms was also tested in sensitivity analyses.

Table 3.

Cost‐effectiveness results

Parameters	Incremental cost ($)	Incremental effectiveness (QALY)	Cost per QALY ($/QALY)
Base case (systolic BP reduction of 9.7 mmHg)	(1929)	0.826	Dominant
SBP reduction of 4.89 mmHg	(237)	0.424	Dominant
SBP reduction of 15 mmHg	(3656)	1.218	Dominant
Add hub costs to 20% of patients	(1879)	0.826	Dominant
Two additional visits for uncontrolled patients (15%) and no subsequent visits for others	(2369)	0.826	Dominant
Increase telemonitoring cost by 50%	(467)	0.826	Dominant
Double telemonitoring cost	994	0.826	1204
Triple telemonitoring cost	3917	0.826	4744
Increase usual care cost by 50%	(2509)	0.826	Dominant
Double usual care cost	(3089)	0.826	Dominant
Subsequent GP visits every 6 wk in usual care arm	(5657)	0.826	Dominant
Postrecurrent stroke health‐related quality of life increased to 0.71 [45] (instead of 0.42)	(1929)	0.752	Dominant
Baseline risk decreased by 50%	(1013)	0.623	Dominant
Baseline risk increased by 50%	(1905)	0.861	Dominant
No difference in BP between strategies after 5 y	346	0.393	879
No difference in BP between strategies after 10 y	(1090)	0.649	Dominant

The incremental costs and QALYs of the alternate treatment strategies were determined from the perspective of the Canadian health care payer and expressed in Canadian dollars. A lifetime horizon (20 years) with annual cycles was used. Discounting for both costs and outcomes was performed at 1.5% per year, with alternate values of 0% and 5% tested in sensitivity analyses.

The health utilities of various model health states were obtained from the published literature (Table 1).19, 22, 23 The baseline utility in the cohort of 0.84, taken from the PREVENTION trial, remained stable until an event occurred (Table 1).19 The utility of recurrent stroke was 0.42;22 in sensitivity analysis, 0.71 was used.24

Resource use and cost estimates were obtained from Alberta Health Services data, Alberta Health data, published data, and expert opinion (Table 2).25 In the base case analysis, telemonitoring costs totaled $305 for the first 3 months and $327 dollars as a subsequent annual cost for the next 20 years (Table 2). Corresponding expenditures for usual care were $186 and $148. Included in these totals were the costs related to care provision, which, in the telemonitoring arm, consisted of prescribing pharmacist case manager time to provide a one‐hour initial assessment, three 30‐minute monthly remote visits until BP control was achieved, and 15‐minute remote follow‐up visits each quarter thereafter. We also included two review discussions with a physician per year to account for primary care physician oversight in the telemonitoring and case management arm. We assumed that the intervention leads to 0.4 additional antihypertensive medications prescribed annually. In the usual care arm, costs incorporated an initial 30‐minute poststroke primary care physician assessment, three monthly physician visits to achieve BP control, and then semiannual follow‐up visits thereafter (for 20 years), a frequency consistent with prior published data.26 Specialist visits (eg, to a stroke neurologist) were assumed to take place equally between groups and were therefore not included in the model.

A sensitivity analysis was performed to assess the effect of variation in the efficacy of BP reduction with telemonitoring and case management, telemonitoring and case management costs, and baseline risk. A probabilistic analysis of 1000 simulations was also undertaken where input parameters were selected for each iteration from distributions listed in Tables 1 and 2.

3. RESULTS

3.1. Model validation

The model predicted a 30% reduction in recurrent stroke at 4 years based on an annual SBP reduction of 9.7 mmHg. These results are similar to those reported in the PROGRESS trial, in which a 28% overall relative risk reduction in recurrent stroke during a mean follow‐up period of 3.9 years was found.27

3.2. Base case results

In the base case analysis, per‐patient telemonitoring and pharmacist case management resulted in total costs of $21 640 and 8.83 QALYs; corresponding values for usual care were $23 020 and 8.00. Telemonitoring and pharmacist case management resulted in an incremental 0.83 QALYs and cost savings of $1929 compared to usual care. Therefore, the intervention was dominant, achieving improved health at a reduced cost.

3.3. Sensitivity analysis

Results were robust to a wide range of sensitivity analysis (Table 3). Even after the systolic BP lowering efficacy of home BP telemonitoring and case management was reduced to 4.9 mmHg from a base case of 9.7 mmHg, model results still indicated intervention dominance. Similarly, the intervention was dominant after increasing telemonitoring costs by 50%. Doubling or tripling telemonitoring cost led to results of $1204 to 4744 per QALY gained, which was the least attractive result identified through sensitivity analysis. Assuming a higher utility for recurrent stroke (0.71 vs 0.42) still led to a dominant result. If no difference in BP was observed after 5 years, the incremental cost‐utility ratio (ICUR) for telemonitoring was $879 per QALY gained; telemonitoring was still dominant when no difference in BP was assumed after 10 years.

3.4. Probabilistic analysis

The probabilistic sensitivity analysis results are shown as an incremental cost‐effectiveness acceptability curve (Figure 2). When comparing home telemonitoring to usual care, home telemonitoring was the preferred treatment in 73% of simulations at a willingness to pay (WTP) of $0 per QALY, and 99% of the ICURs fall below a WTP threshold of $10 000 per QALY.

Figure 2.

Incremental cost‐effectiveness acceptability curve. CE, cost‐effectiveness

4. DISCUSSION

In summary, our findings indicate that BP telemonitoring and pharmacist case management are a dominant strategy when implemented in community‐dwelling hypertensive patients who have experienced a recent minor cerebrovascular event and when modeled over their lifetime, in the Canadian context. Modeling results were robust with respect to a broad sensitivity analysis, demonstrating that the intervention was either dominant (ie, resulting in cost savings and health improvements) or highly cost‐effective.

Previously published economic assessments of home BP telemonitoring have reported varied results.12 In a cost‐effectiveness analysis based on a 2169‐patient RCT conducted in New York State, home BP telemonitoring and case management resulted in high incremental costs of nearly $10 000 in the intervention group because total intervention costs were high.15 In contrast, other studies that used much cheaper telemonitoring/case management interventions reported net costs close to zero.12, 28, 29, 30, 31, 32, 33 A limitation of these analyses is that they examined short‐term costs only; longer‐term economic modeling to incorporate the effect of expected reductions in cardiovascular disease was not performed, contrary to stated best practices of economic evaluations.34, 35 Cost‐effectiveness analyses of home BP telemonitoring and patient medication self‐titration, a similar but not identical intervention to telemonitoring and case management, indicate that this strategy is dominant (ie, improves health and is cost‐saving) in high‐risk patients, defined as those with diabetes, chronic kidney disease, and/or cardiovascular disease.18

In our analysis, a primary driver of cost‐effectiveness is the cost of telemonitoring and case management. It is important to keep these costs low to ensure cost‐effectiveness.12 In part, this reflects a parsimonious telemonitoring method, in which BP measurements are taken according to recommended home BP monitoring protocols (1 week of readings each month until the mean of these measurements indicates that control has been achieved, then quarterly thereafter).6 This telemonitoring method aligns with treatment adjustment protocols, in which medication titration is performed monthly, once sufficient time has passed to allow an antihypertensive agent to reach maximal effect.6 It minimizes provider workload and costs and avoids the need for frequent measurements and recurrent review of readings. Further streamlining can be performed by automating review of measurements such that only mean BPs that require provider attention are marked for review. Our analysis suggests that this parsimonious approach leads to net health care savings; however, even if the resources required are significantly greater, it would remain attractive using commonly cited thresholds of cost‐effectiveness.36, 37

A number of challenges and barriers exist to implementing home BP telemonitoring and case management into routine clinical practice. These have been reviewed in detail elsewhere.9 A major barrier is the lack of availability of specific telemonitoring reimbursement codes for physicians and funded case manager positions. In general, Canadian provincial physician reimbursement plans have been slow to adapt to support virtual/remote care models and still largely require for face‐to‐face visits as a prerequisite for payment. In particular, concerns regarding the potential costs of unconstrained billing of remote services explain the reticence to adopt virtual care codes.9 This underscores the importance of performing economic assessments such as the present study to understand clearly the costs and cost‐effectiveness of virtual or remote care interventions. For example, if physician billing code payments than the cost of the case management and physician oversight modeled in the present study, cost‐effectiveness would be reduced.

Ideally, to maximize cost‐effectiveness and efficacy, the case manager should have authority to independently prescribe and adjust medications, and should be given responsibility for remotely managing a roster of patients, with physician input provided only when necessary.38 Where the case manager should be situated—embedded within primary care vs specialty care—is an important and unresolved question. If the intervention is focused on high‐risk patients, embedding the case manager within a specialty program would optimize targeting of the intervention to a high‐risk population. Conversely, embedding case managers in primary care has the advantage of ensuring that overall care is properly coordinated. Another important question is how case managers should be remunerated, although a small trial reported no difference in achieved BP reduction between pay for performance vs. fee for service payment pharmacist case manager reimbursement models.39 An additional barrier to the adoption of blood pressure telemonitoring is the lack of market availability of telemonitoring systems that offer end‐to‐end telemonitoring without the need to program connectivity between devices, hubs/smartphones, and clouds. Integration into existing EMRs is also challenging, given the reticence of EMR purveyors to allow third‐party connectivity.9

The primary limitation of the present economic model is that Alberta costing data were used. Thus, the results are applicable specifically to the Alberta setting and, more generally, to the Canadian context—a publically funded health care system with universal access and no user fees at points of contact. Second, the assumption of long‐term BP reduction was made and the results are predicated upon this assumption; however, sensitivity analysis indicates that if incremental effectiveness ends after 5 years, telemonitoring remains a cost‐effective strategy. Third, in patients who suffered an event, additional events were not modeled. Fourth, only noninstitutionalized patients were modeled (ie, not those residing in nursing homes) and we assumed that case managers had independent prescribing capabilities. Extrapolation beyond these assumptions should be made with caution.

The model results are conservative in the sense that indirect costs were not incorporated and the societal perspective was not examined. Given that remote monitoring and case management would save time and travel expense, even greater cost savings would be expected if this perspective was taken. Moreover, our model results underestimate potential benefits of case management as we examined only the impact of changes in SBP but we have previously demonstrated in a randomized trial that case managers can effectively manage multiple risk factors concurrently, with an additive benefit in reduction of future cardiovascular events.40 This raises the potential for the use of telemonitoring and case management in other high‐risk patients, such as those with coronary disease, chronic kidney disease, or diabetes.

In conclusion, our findings strongly support implementation of telemonitoring and case management in hypertensive Canadians who have had a cerebrovascular event. Future work should focus on identifying the optimal ways to implement this intervention, in conjunction with addressing the challenges and barriers to implementation. In addition, similar economic assessments in other high‐risk groups where telemonitoring and case management is likely to improve health and reduce costs should be performed.

CONFLICT OF INTEREST

RP, FAM, TJ, MS, and SWK are supported by an alternative funding plan from the Government of Alberta and the University of Alberta. RP and PWW are Directors of a blood pressure measurement start‐up company, mmHg Inc. FAM is supported by the University of Alberta Chair in Cardiovascular Outcomes Research. JAS receives partial salary support from Alberta Health Services. S.K. is supported by the Kidney Health Research Chair and the Division of Nephrology at the University of Alberta. An abstract of this paper has been presented to the American Heart Association Council on Hypertension meeting (September 6‐9, 2018).

Padwal RS, So H, Wood PW, et al. Cost‐effectiveness of home blood pressure telemonitoring and case management in the secondary prevention of cerebrovascular disease in Canada. J Clin Hypertens. 2019;21:159–168. 10.1111/jch.13459