For both treatment strategies, our model calculated quality adjusted life expectancy (quality-adjusted life-years [QALY]) and cost in 1-year time horizon. We compared the performance of the two treatment strategies through the incremental cost-effectiveness ratio (ICER), defined as marginal cost divided by the marginal effectiveness, and incremental net monetary benefit (INMB), defined as difference in dollar value of healthcare outcomes. We conducted one-way sensitivity analyses for every variable in our model to assess the influences of them within a clinically plausible range on cost-effectiveness or net monetary benefit. We then plotted the tornado diagrams for every parameter. Additionally, we conducted probabilistic sensitivity analysis (PSA) through Monte Carlo simulation. The Monte Carlo simulation was run for 1,000 samples, where parameter values were randomly selected from assigned distributions. We assumed probabilities and QALY followed a Beta distribution and the cost of interventions followed a Log-normal distribution. A willingness-to-pay threshold of $100,000/QALY was chosen. This threshold is a commonly considered reasonable value used in other research.22 Parameters estimates
The base case estimates and ranges of all parameters, quality adjusted life-year, and costs used in the model are summarized in Table 1.
To estimate the parameters used in our model, we pooled data from the published literature with emphasis on randomized clinical trials and large population-based studies.15,23-26 In Dumas’ stduy15, 714 patients with OHCA were included and immediate coronary angiogram were introduced for 435 patients of them with no obvious extra-cardiac reason of cardiac arrest. 134 had ST-segment elevation and among the 134 patients with ST-segment elevation, 128 has ≥ 1 significant coronary lesions. 301 patients who did not have ST-segment elevation and among the 301 patients without ST-segment elevation, 176 has ≥ 1 significant coronary lesions. The probabilities of having successful PCI in the immediate angiography strategy was 58% and the probability of having successful PCI in the standard care strategy given patients having ECG ST-segment elevation was 74%. Because the trial was targeting the patients with out-of-hospital cardiac arrest, we used 50% occlusion in the vessel as the threshold to assign patients into different intervention groups and defined the presence of a culprit lesion as an angiographic acute-appearing coronary lesion. We derived the sensitivity of ECG ST-Elevation for diagnosing occlusions from Menown’s paper24 after comparison among different studies. Since the specificity of ECG ST-segment elevation for diagnosing artery occlusions is stable, we derived it from Dumas’ study.15
Quality Adjusted Life-Years
Quality-of-life weight is also called utility score, which has a range from 0 to 1, in which 0 denotes death and 1 denotes perfect health. The quality adjusted life year (QALY) can be obtained through multiplying the utility score and life expectancy. We estimated the QALY based on the treatments that patients received and the health conditions they are in. We assumed the condition of OHCA patients with cardiac causes is similar to those with acute myocardial infarction. In the model, the patients who were likely to have severe stenosis could receive invasive treatments, for example, percutaneous coronary intervention (PCI) and coronary artery bypass graft surgery (CABG). Patients who were likely to have mild or no stenosis would receive conservative medical treatment only. Thus, the estimates of one-year quality-of-life for the model were based on the research of unstable angina myocardial infarction27-29. The QALY after late PCI or CABG procedure for patients with severe stenosis were obtained from Kim et al.27 And the QALY after immediate PCI or CABG procedure for patients with severe stenosis were obtained from the study of Cohen et al.29 QALY after conservative treatment for patients with mild or severe stenosis were derived from Josephine et al30 and Mark et al31 respectively.
We derived cost estimates from published literature. We estimated the cost of base conservative treatment, base invasive treatment, ECG test, angiography procedure, PCI procedure and CABG procedure from Aasa’s paper.32 Recalculation was conducted to estimate the cost of patients based on the treatment they received and health conditions they are in. More details of cost are in Table 1. All costs were adjusted to 2015 US dollars using the medical care component of the Consumer Price Index.33 We included the cost of initial hospitalization and cost of 1-year follow-up into our model. In the initial hospitalization cost, the procedure cost, hospital stay and physician fees were incorporated. In the 1-year follow-up cost, we estimated the cost of re-hospitalization, out-patient care, medications and physician fees.
Table 1. Base case estimates and ranges
Base Case Estimate
Sensitivity of ECG ST-Elevation for diagnosis of artery occlusions
α= 9.02; β=7.09
Specificity of ECG ST-Elevation for diagnosis of artery occlusions
α=5.30 ; β=0.28
Prevalence of having >50% stenosis among general population
1-yr QALY for who underwent late PCI with Severe Stenosis
1-yr QALY for who underwent late CABG with Severe Stenosis
1-yr QALY for who underwent conservative treatment with Mild Stenosis
α =9.79; β =3.16
1-yr QALY for who underwent conservative treatment with Severe Stenosis
The base case value means the best estimate for each variable. Unless otherwise noted, ranges are defined by 95% confidence intervals.
* The range are estimated from 80% to 120%
Base case analysis
Immediate angiography strategy was more effective (+0.03 QALY) and less expensive (-$1,281/year). Thus, the Immediate Angiography strategy is dominant in base case analysis in the condition of $100,000/QALY willingness-to-pay. (Table 2)
Table 2. Incremental net monetary benefit
Note: Incr Cost: incremental cost; Eff: effectiveness; Incr eff: incremental effectiveness; NMB: net monetary benefit; INMB: incremental net monetary benefit.
One-way sensitivity analysis showed that no parameters in the model had significant impact on the result. We plotted the Tornado diagram using incremental net monetary benefit (Figure 3, immediate angiography vs. standard care). Net monetary benefit is equals to the difference between cost and the product of willingness-to-pay and effect (NMB = λ * Effect - Cost). The positive results of INMB favors immediate angiography. The “Sensitivity of ECG ST-segment elevation for diagnosis of acute myocardial infarction” has the largest effect on choosing favorable strategy. The one-way sensitivity analysis showed that the result of the base case analysis is robust within the range of all parameters.
We tested every parameter through probabilistic sensitivity analysis and the result showed that the strategy of immediate angiography has 91% probability, of being more cost effective than the standard care strategy with a $100,000/QALY threshold. There was around 97% space probability that the strategy of immediate angiography is more cost effective than the standard care in the condition of $50,000/QALY threshold. (Appendix B)
Figure 2. One-way sensitivity analysis (incremental net monetary benefit)
This cost-effectiveness analysis compares immediate angiography to standard care. The findings indicate that the immediate angiography strategy may improve both survival and neurological outcomes of the patients resuscitated from OHCA, but cost less than the standard care strategy. This is consistent with Dumas et al15 found in a randomized clinical trial in Paris. Because few studies have demonstrated whether the immediate angiography is cost effective in the United States, there is uncertainty about the cost estimates. Previous studies have used markov models.35 However, we chose a decision-tree model with 1-year time horizon because we believe that for patients with OHCA, most of cost and effectiveness could be included in a short-term period because of the low discharge survival rate.1 In the base case analysis, we found out immediate angiography is favorable with a $47,513 NMB in a condition of $100,000/QALY threshold. In the one-way sensitivity analysis, immediate angiography was favored within the ranges of all variables. It indicates that the result is robust. Currently, National Institute for Health and Care Excellence (NICE) recommends further drug treatment for the patients with unstable angina and NSTEMI first (Non-ST-Elevation Myocardial Infraction) (Appendix C), then decide when to offer coronary angiography. However, those assessments are subjective and finally it recommends revascularization strategies (PCI or CABG) based on the angiography results. The decision process and treatment procedures used in our model are similar to NICE’s, while we are planning to do invasive interventions for those with larger than 50% culprit lesions immediately. Both algorithms in NICE and our model demonstrated the significance of implementing coronary angiography in myocardial infarction with either ST-segment elevation or non-ST-segment elevation.
Prior research has reported the comparisons between non-invasive diagnostic strategies, for example, exercise echocardiography and exercise ECG.35 The OHCA patients who were resuscitated, could be in coma, under this circumstance, ECG and other non-invasive strategies could not be appropriate to diagnose the etiology. Ollendorf et al36 found that angiography could provide better diagnosis than the ECG test, especially in emergent setting. Furthermore, angiography has been regarded as gold standard in diagnosing arterial lesions and because of the high prevalence of coronary occlusions and difficulties in interpreting the ECG results in the OHCA patients, immediate angiography should be considered as the first choice for diagnostic and treatment strategy.37 Abraham et al38 found that aggressive catheterization was performed in patients with non-cardiac causes of ST-segment elevation on ECG. He emphasized the inessentiality of pushing every cardiac arrest patient to the catheterization laboratory without appropriate evaluations.
However, Abraham did not define what were the appropriate evaluations. Abraham did state an issue that the result of ECG alone cannot be used to determine the arterial lesions. In the base case results of our analyses, immediate angiography strategy was dominant, which means it costs less but benefits more for the OHCA patients than the standard care strategy. In the acceptability curve, as the willingness-to-pay (WTP) increases, the probability of immediate angiography being cost effective decreases. This could be caused by the distribution properties in the Monte Carlo simulation. As the WTP increases, the increase rate of becoming cost-effective is lower than the decrease rate of becoming non cost-effective.
There are several limitations in our research. For the model structure, we built a 1-year time horizon decision tree model, but Markov model could involve more details of the disease pattern and healthcare outcomes. For example, the recurrence in the long-term time horizon could affect the quality of life (QoL) and then affect the quality adjusted life-years (QALY). As what we stated above, short-term time horizon could involve major details of both strategies. Even though long-term time horizon model, like Markov model will reflect more information, the essentiality of it in our research needs to be discussed.
In the analysis, we used secondary data from extant published literature and checked the reliability with other professionals, like the physicians and other modelers. The results of the sensitivity analysis of costs showed that none of them had a significant impact on our conclusions. Nonetheless, the property of the secondary data could affect the credibility and reliability of the model. In the future, we will conduct micro-costing analysis for obtaining more accurate data on cost. When conducting the literature review, we assumed that the condition of OHCA patients due to cardiac causes are like the myocardial infarction (MI) patients since MI is the major cause of cardiac arrest.4,5 Since other non-cardiac causes are easily excluded, like, drowning, asphyxia and falls, and brain-related causes can be diagnosed in a relative short time based on the disease history of the patient, we thought this assumption is reasonable and acceptable.
From the research, we found out that applying the new immediate angiography has the potential of lowering the cost of treatment of OHCA. In 2010, the annual direct and indirect cost of treatment of cardiovascular disease (CVD) in the U.S. was 107.2 billion dollars.1 Considering the annual number of individuals suffering from the out-of-hospital cardiac arrest, the new immediate angiography strategy could help control lower the cost of treatment of cardiovascular disease. Additionally, based on previous research and our study, immediate angiography can improve healthcare outcomes, like survival and neurological functions. Among the developed countries, the healthcare system in the United States did not provide the benefit it should have provided for the public.39 Since 1974, the mortality rate decreased less in the US than it did in Australia, however, the healthcare cost increased faster. We believe that one of the major reasons is the unreasonable distribution of healthcare resources. We should apply new technology, strategy and management to help reallocate the limited healthcare resources.
In the future, we are going to build a discrete states model, like Markov model, to include more details and assess whether the conclusion in this study is robust. Since the new immediate angiography strategy is cost effective, it is reasonable to conduct a budget impact analysis (BIA) to determine how the new immediate strategy influences the budgets in the local, state and federal level of US government.
This study provides an evidence for modifying the current treatment strategy of OHCA. In conclusion, based on our modeling and simulation results, the strategy of immediate angiography is more effective and less costly than standard care for the diagnosis and treatment of out-of-hospital cardiac arrest patients.
Our results suggest that immediate angiography is more cost effective than the standard care for OHCA patients from a societal perspective because the ICER is well below the upper limit of the threshold that is generally considered to be cost-effective by many health-care agencies. Our cost-effectiveness study is going to help advance the treatment strategy of OHCA so that the healthcare resource can be utilized. Public health resource is limited in the world, the saved resource in the cardiovascular disease area could be used in other areas, like HIV/AIDS, purification water or poverty. From this point of view, the research we did can provide evidence and support for allocating the limited public health resource.
APPENDIX A: DISTRIBUTION PARAMETERS CALCULATION
For the Beta distribution, we calculated parameters through the equations of mean and standard deviation (mean = α/(α+β); variance = αβ/((α+β)2 * (α+β+1))). And we followed the similar procedures but with equations of mean = e(μ+σ^2/2) and variance = (e(σ^2 )-1) * e(2μ+σ^2) for Log-normal distribution. The calculation was perfumed through WolframAlpha website. we regarded the base case estimation as the mean and the range of every variable has ±2 standard deviations. For those not providing range of the values, we made estimation from 80% to 120% of variables.
APPENDIX C: MANAGEMENT STRATEGIES FROM NATIONAL INSTITUTE FOR HEALTH AND CARE EXCELLENCE (NICE)
Early management of unstable angina and NSTEMI: https://pathways.nice.org.uk/pathways/chest-pain/early-management-of-unstable-angina-and-nstemi#content=view-node%3Anodes-intermediate-high-or-highest-risk