記住我
Hypertension and diabetes are major risk factors for cardiovascular diseases (CVD) associated with mortality and morbidity, imposing huge economic burdens [,]. Adequate blood pressure (BP) and glycemic control are important in health promotion and health care systems worldwide [,]. However, despite significant advances in evidence-based lifestyle modifications and pharmaceutical interventions, less than 40% of patients treated for hypertension or diabetes achieve the recommended target BP or blood glucose levels in Korea []. Therefore, novel interventions are needed to support patient efforts for behavior changes to promote healthy lifestyles and disease self-management [,]. Mobile health (mHealth) based on information and communication technology in primary care is an innovative approach to such interventions []. The widespread use of mobile phones combined with the ability to process and communicate data instantly enables real time individually tailored health care delivery and overcomes barriers of time and place [].
A national pilot project for chronic disease management was recently conducted under the supervision of the Ministry of Health and Welfare to evaluate the effectiveness of information and communications technology (ICT)–based monitoring service for tailored chronic disease management in primary care in Korea [,]. The ICT-based tailored management (TM), using automated advanced systems for patients with hypertension and diabetes, provides a continuous and integrated customized health management service by linking the mHealth service platform (smartphone app) and the provider’s (primary care physician and care coordinator) operation web combined with the patient’s health information (examination and drug prescription information data from the National Health Insurance Service). Several mHealth interventions (eg, mobile phone SMS text messages, wearable monitoring devices, and telemedicine) for managing hypertension and diabetes have demonstrated efficacy [], but evidence for interventions using smartphone apps is limited. Furthermore, their cost-effectiveness varied substantially based on the target disease and type of technology, and the results of economic evaluations have been inconsistent []. In particular, the cost-effectiveness of interventions using complex smartphone communications in older people cannot be judged due to a lack of information [].
To our knowledge, no studies have examined the long-term cost-effectiveness of tailored management interventions for patients with hypertension or diabetes using automated advanced mobile technology in primary care. Although a large pilot project on TM intervention based on the highest smartphone penetration rate globally and advanced technology was conducted [], it remains unclear whether the additional benefits from TM are sufficient to justify its use over and above usual care (UC) alone in terms of economic efficiency. This study aimed to estimate the cost-effectiveness of TM for adult patients older than 19 years of age with hypertension or diabetes in primary care compared with UC, using data collected in a pragmatic trial of ICT-based tailored chronic disease management (ICT-CM).
The ICT-CM was a pragmatic trial designed to test the real-world effectiveness of the ICT-based monitoring program among patients with chronic diseases in primary care []. Study participants were patients aged 19 years or older diagnosed with hypertension or diabetes, recruited from 8 clinics in Seoul and Gyeonggi-do, South Korea. Patients with myocardial infarction, stroke, end-stage renal disease, or liver failure within 1 year were excluded. Among the participants (n=1004) in the ICT-CM trial, those aged between 60 and <70 accounted for the highest proportion at 39% (392/1004), while those younger than 60 years accounted for 36.8% (369/1004), and those aged 70 and older accounted for 24.2% (243/1004). The average age was 62.4 years.
The study included a TM intervention group and a UC control group. Participants who received UC for hypertension and diabetes were treated at primary care institutions according to Korean clinical practice guidelines, and UC encompasses continuous monitoring and follow-up, appropriate medication management based on the patient’s condition, and lifestyle modifications to promote healthy habits [,]. Participants who received TM intervention were treated using automated advanced mHealth systems by physicians and care coordinators. A care coordinator trained participants who received TM on how to use the technology, downloaded the mobile application Carecrew (Huraypositive Inc) available for both Android (Google) and iOS (Apple Inc) onto their own smartphones, and were given Bluetooth-enabled devices (sphygmomanometer and glucometer). Physicians established personalized management goals for each patient (referred to as a tailored care plan) based on the lifestyle survey results and clinical examination at baseline. The care plan, including each participant’s target BP or blood glucose level, daily measurement frequency, and management priorities, was automatically sent to the Carecrew. The participants were asked to record their lifestyle such as diet and body weight and periodically check their blood glucose, BP, and medication use. Results measured by the devices were automatically uploaded to the provider’s operation web (CareCrew Web by Huraypositive Inc). The physicians and care coordinators had access to this system to monitor the patient’s condition constantly. Patients who missed self-recording were encouraged to self-measure and input lifestyle data through SMS text messages and phone calls. When the system reported out-of-range BP or glucose readings, a tailored mobile SMS text message was sent to the patient, and mobile-based feedback was performed for high-risk patients requiring intensive care. A customized examination voucher was issued for each patient and assessments were conducted at the clinic visit at 6 months. Source data from ICT-CM trial were verified through periodic on-site monitoring. Monitors identified any discrepancies between all case report form entries and the source data and issued queries. The queries were resolved by investigators, cooperating with monitors. The ICT-CM trial is described in detail in .
Ethical ConsiderationsThe study was approved by the Investigation Review Board of Kangbuk Samsung Hospital (IRB KBSMC 2020-07-026-012). Informed consent and the ability of participants to withdraw were provided in the ICT-CM trial used for source data in this study. Expenses, including intervention costs and examination fees incurred due to study participation, were covered by the study, but no financial compensation was paid. All data used in this study were anonymized and deidentified.
Model OverviewLong-term costs and health outcomes over the lifetime were compared for the TM intervention arm and UC arm. A target cohort population comprising Korean patients with hypertension or diabetes diagnosed in primary care, which was consistent with the populations examined in the ICT-CM trial. The Markov model was designed to mirror the natural progression of conditions in the population and the clinical pathway based on clinical guidelines. This model reflects the short-term outcomes of treatment (controlled-uncontrolled status of BP or blood glucose) in patients with hypertension or diabetes, their characteristics that were considered CVD risk factors, and the risk of long-term complications that patients may experience. The short-term outcomes were as observed in the ICT-CM trial at 6 months, and subsequent long-term outcomes were extrapolated from the trial data. The model was programmed in Microsoft Excel, supporting macro programming through Visual Basic for Applications. This study is reported according to the Consolidated Health Economic Evaluation Reporting Standards checklist [].
Model StructureThe structure of the Markov model is presented in . The model structure’s validity was checked by reviewing the literature and expert opinion [,]. A Markov cohort could experience 1 of several health states, including “Well” with no major complications (ie, CVD), “CVD,” “Death due to CVD,” or “Death due to other cause.” “CVD” was defined as a composite outcome of coronary heart disease (ie, myocardial infarction and angina), cerebrovascular diseases (ie, cerebral infarction, hemorrhagic stroke, and transient ischemic attack), heart failure, and peripheral artery disease. All patients in the TM or UC groups started the Markov process in the “Well” state. Patients could remain in the “Well” state or transfer to another state at every cycle length of 6 months, corresponding to the frequency of assessments in the ICT-CM study. From any health state, patients could transit to the “Death due to other cause” state, and from the “CVD” state, patients could transit to the “Death due to CVD” state. Because the age of the patient cohorts on entry into the model was 60 years, a treatment discontinuation rate of 5% was assumed, considering that TM intervention could be discontinued in the real-world setting. Patients who discontinued TM intervention continued with UC. Patients receiving the UC at the model start were assumed to remain on the treatment until the end of the time horizon or death, whichever occurred first.
Figure 1. Model structure. (A) Model framework. The short-term outcomes were as observed in the ICT-CM trial at 6 months, with subsequent long-term CVD events with extrapolation from the trial data. (B) State transition diagram. The CVD was defined as a composite outcome of coronary heart disease (ie, myocardial infarction and angina), cerebrovascular diseases (ie, cerebral infarction, hemorrhagic stroke, and transient ischemic attack), heart failure, and peripheral artery disease. BP: blood pressure; CVD: cardiovascular diseases; HbA1c: glycated hemoglobin A1c; HDL-C: high-density lipoprotein cholesterol; ICT: information and communications technology; ICT-CM: information and communications technology–based monitoring service for tailored chronic disease management. Clinical OutcomesThe outcomes derived from the ICT-CM trial used in the study are presented in . Through post hoc analysis using individual patient data from the ICT-CM trial, the CVD risk factor outcomes were measured for each treatment arm. Demographic information including age and medical history were examined in screening tests, and smoking history was assessed in health surveys using a complete self-health questionnaire at both baseline and 6 months. Total cholesterol, high-density lipoprotein-cholesterol, and glycated hemoglobin A1c (HbA1c) levels were assessed during clinical examinations at both baseline and 6 months, while BP was measured through body assessments at the same intervals. Details regarding the measurements and the outcomes assessed in the analysis are described in . Using the input values associated with CVD risk factors obtained from the trial, the 10-year cardiovascular risk for each patient was calculated using the Framingham CVD risk prediction model []. The average risk estimate for each treatment arm was then converted into an annual probability of CVD events. The risk of CVD increased with the age of the patient cohort based on the age-related relative risk of CVD obtained from Kim et al []. The base-case regarding the long-term effect persistence assumed that the 6-month difference in BP or blood glucose between TM and UC decreased by 10% after 5 years, reflecting that the effectiveness may decline with age.
Table 1. Information and communications technology–based tailored chronic disease management trial participants’ data used in the Framingham risk model.Risk factor outcomesAll patients with hypertension or diabetesPatients with hypertensionPatients with diabetesaTM: information and communications technology–based tailored management.
bUC: usual care.
cBP: blood pressure.
dHDL: high-density lipoprotein.
eHbA1c: glycated hemoglobin A1c.
All-cause and CVD-related mortality of the general population were based on cause-of-death statistics by age of the Korean Statistical Office []. The annual probability of death due to other-cause was calculated by excluding death due to CVD from all-cause death. The fatality rate for CVD by age was estimated as the number of deaths due to CVD among patients with CVD, using the cause-of-death statistics of the Korean Statistical Office [] and the Korean National Health Insurance statistics by the Health Insurance Review and Assessment Service []. presents a detailed description of the model inputs.
Table 2. Model inputs for patients with hypertension or diabetes.ParametersEstimate in base-caseDistributionaSourcesAnalysis settingaDistributions used in probabilistic sensitivity analysis.
bNot applicable.
cICT-CM: information and communications technology–based tailored chronic disease management.
dCVD: cardiovascular diseases.
eHIRA: Health Insurance Review and Assessment Service.
fIn base-case for all patients, the 10-year cardiovascular risk was calculated with Framingham CVD risk prediction models based on the CVD risk factors obtained from the ICT-CM trial. 1-year probability was calculated from the 10-year risk values.
gTM: information and communications technology–based tailored management.
hKNHANES: Korea National Health and Nutrition Examination Survey.
iAll costs are expressed in 2023 US $ using an exchange rate of US $1=1320 KRW.
jThrough the TM trial, the annual running cost was calculated by dividing the care coordinator’s labor cost by the number of managed patients (approximately 250 per year).
kNHIS: National Health Insurance Service.
Utility WeightsThe utility weights for the health states of patients with hypertension or diabetes were derived from the Korea National Health and Nutrition Examination Survey (KNHANES, 2007-2019) data conducted by the Korea Centers for Disease Control and Prevention. KNHANES is a nationwide survey conducted annually for the representative general population in Korea []. The survey includes individual-level information on health-related quality of life measured by the EQ-5D developed by the EuroQol Group. The 5 dimensions of EQ-5D-3L comprise mobility, self-care, usual activities, pain or discomfort, and anxiety or depression, and each dimension was answered with 3 levels. The health state index scores calculated based on the tariff of Lee et al [] range from 0 (where 0 is a health state equivalent to death) to 1 (perfect health), with higher scores indicating higher health utility. The utility weight of “Well” status was higher than that of “CVD” by 0.05 for patients in their sixties, and the utility weight decreased with age. However, the difference in utility weight between the 2 health states remained similar. The utility weights by age are presented in .
CostsAll costs were expressed in 2023 US $ using an exchange rate of US $1=1320 KRW and adjusted for inflation where applicable using the Consumer Price Index for health care []. Intervention costs of patients who received TM were obtained from the TM trial. Bluetooth-enabled device costs per patient were applied as one-off costs to the model’s initial health state (“Well”). Equipment costs included initial development, network use, and system update expenditure. The per capita total equipment costs were applied to the model, estimated based on the patient number of target cohorts reflecting smartphone penetration []. Running costs comprised costs associated with training and care coordinator. Through the TM trial, the number of manageable patients per care coordinator was investigated, and the annual running cost was calculated by dividing the care coordinator’s labor cost by the number of managed patients (approximately 250 per year). Running and equipment, annuitized based on a lifespan of 5 years, were included in intervention costs over the lifetime time horizon.
The cost of “Well” and “CVD” health states were derived from data analysis of the national health insurance claims data [], and were applied equally to the 2 arms. Each health state’s costs comprised expenditure for diagnostic test, medication, surgical treatments, and resources use associated with inpatient or outpatient services. The costs of cardiovascular events including initial acute care and long-term care were estimated. The costs of acute phase CVD were applied to the model in the first year after event, and in subsequent years, annual costs of chronic phase CVD were applied. All cost inputs are presented in .
Cost-Effectiveness AnalysisIn a base-case analysis of all patients with hypertension or diabetes, the cost-effectiveness analysis of TM compared with UC was conducted from the health care system perspective. A result was presented as an incremental cost-effectiveness ratio (ICER) calculated by dividing the incremental cost by the additional quality-adjusted life-year (QALY) gained. A 4.5% annual discount rate for both cost and QALY were performed with half-cycle corrections. Deterministic and probabilistic sensitivity analysis (PSA) was performed to address parameter uncertainty. The parameters, including probabilities, utility weights, costs, number of manageable patients per care coordinator, and discount rate, were varied through deterministic sensitivity analyses. PSA of 1000 Monte Carlo simulations was undertaken, in which each of the parameter estimates was sampled from its distribution listed in . PSA results were expressed as a cost-effectiveness plane and cost-effectiveness acceptability curve that shows graphically the probability of cost-effectiveness for all alternatives across a range of an ICER threshold of US $26,515 (KRW 35 million) per QALY []. In addition, several scenarios were explored to assess the structural uncertainty associated with model assumptions. The assumptions regarding the effect persistence between the 2 arms were tested by varying the period during which the 6-month effect difference lasts, from the 3 years beyond the trial period to the remainder of the lifetime. In consideration of treatment persistence, the scenario considering a treatment discontinuation rate of 0% or 10% in patients receiving TM was also evaluated. Finally, the model’s time horizon varied from a lifetime to between 10 and 20 years.
A subgroup analysis was performed for patients with hypertension and diabetes. The Framingham risk model used in the base-case was reported to overestimate the risk of CVD in Koreans [], and underestimate the risk of CVD in patients with diabetes []. Therefore, other CVD risk models were used in each subgroup. The Korean Hypertension cardiovascular (KH-CVD) risk model for patients with hypertension [] and the United Kingdom prospective diabetes study (UKPDS) model that estimates the risk of coronary heart disease and stroke [,] for patients with diabetes were used to predict long-term effects. Table S1 in shows model inputs, such as probability of CVD events, utility, and costs, estimated for each subgroup.
Compared with UC, the cost-effectiveness analysis indicated that TM was more effective (12.006 QALYs for TM vs 11.868 QALYs for UC) and more costly (US $23,157.4 for TM vs US $22,390.5 for UC). TM had incremental costs of US $766.9 and incremental QALYs of 0.138 compared with UC over the lifetime horizon for patients with hypertension or diabetes in South Korea. Through the base-case analysis, ICER was US $5556 per QALY gained ().
Table 3. Cost-effectiveness results in base-case and sensitivity analyses.aAll costs are expressed in 2023 US $ using an exchange rate of US $1=1320 KRW.
bQALYs: quality-adjusted life-years.
留言 (0)