The present study is a full economic evaluation based on decision-analytic model which compared two strategies of CBA and RFA in Iran. Accordingly, the cost utility analysis method was used. Various stages of economic evaluation were performed based on reference guideline of national institute for health and care excellence (NICE) to perform economic evaluation of health projects (13).
Modeling
Designation of economic evaluation model was performed after literature review based on natural history of disease, process of performing ablation methods in patients with AF, clinical outcomes, probabilities of occurrence of outcomes and incidence of expenses.
In order to design the model, specialized panels were formed with the presence of a team of clinical specialists and economic team. After trial and error of various models, the final model was extracted by consensus and considering the most important clinical and economical outcomes based on natural history. The designed economic evaluation model is observed in Fig. 1.
The model structure is designed so that each strategy is based on a life time Markov structure with a one-year cycle length. In each Markov structure, five health states including AF before ablation, AF post intervention, AF post re-intervention, post stroke, normal sinus rythem (NSR) and death were considered. Individuals at both comparable groups were at the AF before ablation health state in zero cycle. Markov process of following cycles is so that individuals in each group and health states remained at the end of each cycle or went through other health states or died. Also patients can place at re-ablation state once, and due to lack of sufficient evidences, third ablations and higher were not considered in the model, and considering of AF post re-intervention health state was for this purpose.
In order to extract evidences regarding mortality rate and other evidences, mean age of 50 years (due to mean age of patients undergoing ablation based on accessible hospital information) for patients at initiation of Markov models were considered.
Extracting Parameters and Analyzes
This study was performed from the perspective of Iran health system and as mentioned, time horizon was considered as life time and costs and outcomes were estimated accordingly.
Since the current study was a full economic evaluation, QALY index was considered as outcome measure and its value at each health state is computed through estimating patients’ utility at each state. Also, in this section, the amount of disutility caused by the side effects of ablation in health states was considered.
Status of cost effectiveness of each strategy was finally assessed based on cost per each units of QALY. Evidences related to patients’ utility at each health state were extracted from international studies.
Regarding costs, due to study perspective, direct medical costs were only considered.
Costs of each strategy were estimated divided by costs of various of health states based on cost unit used in interventions. Costs of performing procedure of CBA and RFA including costs of CBA and RFA, costs related to management and supervision of receiving services in hospital, costs of hospitalization, supportive, therapeutic and pharmaceutical cares, and costs related to side effects were considered. Data collection for costing was from medical records of 47 patients of Shahid Rajaie Cardiovascular Medical and Research Center in Tehran in 2019. Since all the medical records related to 2019 were accessible, sampling was not done for this purpose, and hospital bills of all patients undergoing both ablation methods were assessed. Accordingly, 27 and 20 medical records related to CBA and RFA were assessed, respectively. Cost of other medical attempts required at health states independent from cost of procedure was determined based on therapeutic protocols. Accordingly, type of anti-coagulant and anti-arrhythmic medications and their doses until definite time, cost of hiring holter monitor device and cost of electrical cardioversion were determined. All the mentioned cases were different in various patients and therefore, by consultation with clinical consultant, a moderate amount of costs was considered. Cost of side effects was estimated based on evidences of previous studies and re-costing based on tariffs and domestic currency.
All the stages of costing was calculated by holding a specialized panel with clinical team and based on governmental tariffs of the Iran’s Ministry of Health.
Other parameters and variables related to transition probabilities among health states of two strategies, risk of mortality at each health states, mortality risk caused by ablation methods, efficacy of interventions, and other parameters including risk of side effects of each ablation methods were extracted from international evidences. In this regard, based on each parameter, a distinct literature review was performed in scientific databases, and studies with appropriate evidences were classified and finally the best evidences were extracted.
Values of parameters and variables of the model and their references are given in Table 1.
Table 1
Model Inputs and Parameters
Statistic variable
|
Base case
|
SD/(CI)
|
Distribution
|
Source
|
Annual discount rate
|
0.05
|
(0-0.1)
|
Beta
|
|
Time Horizon(year)
|
Life Time
|
|
|
|
Probability of transition(CBA group)
|
|
|
|
|
AF recurrence after ablation, first year
|
0.269
|
± 0.0538
|
Beta
|
(14)
|
AF recurrence after ablation, >first year
|
0.0938
|
± 0.0235
|
Beta
|
(15)
|
pericardial effusion or cardiac tamponade
|
0.0084
|
|
|
(14)
|
permanent phrenic nerve palsy
|
0.032
|
|
|
(14)
|
vascular complications
|
0.0156
|
|
|
(14)
|
Stroke rate per year
|
0.05
|
|
|
(16)
|
Re-intervention with RFA
|
0.5516
|
± 0.11032
|
Beta
|
(15)
|
Re-intervention with CBA
|
0..0951
|
± 0.01902
|
Beta
|
(15)
|
Probability of transition(RFA group)
|
|
|
|
|
AF recurrence after ablation, first year
|
0.3326
|
± 0.0665
|
|
(14)
|
AF recurrence after ablation, >first year
|
0.1055
|
± 0.0264
|
|
(15)
|
pericardial effusion or cardiac tamponade
|
0.0231
|
|
|
(14)
|
permanent phrenic nerve palsy
|
0.0005
|
|
|
(14)
|
vascular complications
|
0.023
|
|
|
(14)
|
Stroke rate per year
|
0.05
|
|
|
(16,17)
|
Re-intervention with RFA
|
0.5685
|
± 0.1137
|
|
(15)
|
Re-intervention with CBA
|
0.0587
|
± 0.01174
|
|
(15)
|
Probability of death
|
|
|
|
|
Annually probability of death (First year)
|
0.0057
|
|
|
Life Table
|
Probability of operative death
|
0.000487
|
|
|
(18)
|
Stroke-specific mortality
|
0.3536
|
|
|
(19)
|
Costs($)
|
|
|
|
|
AF average annual costs
|
372.81
|
± 55.92
|
Gamma
|
Survey and Calibration
|
NSR average annual costs
|
273.32
|
± 40.99
|
Gamma
|
Survey and Calibration
|
CBA
|
7751.88
|
± 516.72
|
Gamma
|
Survey and Calibration
|
RFA
|
5027.10
|
± 1530.66
|
Gamma
|
Survey and Calibration
|
Stroke costs, first year
|
1804.49
|
± 180.44
|
Gamma
|
Survey and Calibration
|
Post stroke costs, >first year
|
541.34
|
± 54.13
|
Gamma
|
Survey and Calibration
|
pericardial effusion or cardiac tamponade
|
1060.19
|
± 106.01
|
Gamma
|
(10), Calibration
|
permanent phrenic nerve palsy
|
11.09
|
± 106.01
|
|
(10), Calibration
|
vascular complications
|
60.23
|
|
|
(10), Calibration
|
Utilities
|
|
|
|
|
NSR
|
0.8
|
± 0.00577
|
Beta
|
(20)
|
AF
|
0.6
|
± 0.0721
|
Beta
|
(20)
|
Post stroke
|
0.46
|
± 0.0577
|
Beta
|
(21)
|
Disutility due to complications
|
-0.0314
|
|
Beta
|
(22)
|
Disutility due to stroke, first year
|
-0.296
|
|
Beta
|
(23)
|
Cost effectiveness Analysis
In order to perform analysis and determine the most cost-effective strategy, due to costs and effectiveness of each strategy, incremental cost-effectiveness ratio (ICER) was calculated.
The equation for this index was as below:
ICER = C 1 -C 2 /E 1 -E 2
In which, C1, 2 represents for cost of CBA and RFA, and E1, 2 represents for their effectiveness.
Amount of ICER was compared with the amount of Iran’s cost effectiveness threshold (WTP), and the most cost-effective strategy was determined.
In this study, cost effectiveness threshold was considered to be one times of GDP per capita equal to $7142 due to recommendation of WHO for developing countries.
In order to perform all stages of modeling and analysis of results, TreeAge 2011 software was used.
Sensitivity Analysis
Given uncertainty regarding some parameters used in the model, Deterministic and Probabilistic Sensitivity Analysis of the results of the model was performed.
In order to perform Deterministic Sensitivity Analysis (DSA), one way sensitivity analysis and Tornado diagram and two-way sensitivity analysis were used.
PSA was performed considering probability distribution of uncertain variables using Monte Carlo simulation. The range used for uncertainty in point estimation of each variable and statistical distributions used in PSA are presented in Table 1. In cases which no evidences regarding variance of the variable were found, 10–20% of mean values of variable was considered as standard deviation, and appropriate distribution was selected due to the type of variable.