5.1 The company conducted a search of the health economics literature on HeartFlow FFRCT and the comparators specified in the decision problem. They identified a total of 174 studies, 24 of which it considered relevant to the decision problem.
5.2 The external assessment centre (EAC) reviewed this search, and considered that most of the studies included neither an appropriate patient population nor a treatment pathway. Only 1 published study, Rajani et al. (2015), was considered by the EAC to be relevant to the decision problem. It conducted a further review of the literature up to February 2016 and identified an additional relevant published study, Hlatky et al. (2015).
5.3 Rajani et al. (2015) was a single-centre retrospective cost analysis of 410 patients referred to a rapid-access chest pain clinic in Guy's and St Thomas' Hospital, London, from April 2012 to March 2013. Patients were grouped into pre-test likelihood categories and diagnostic imaging was done based on standardised protocols as recommended in the previous NICE guideline on chest pain. A standardised unit cost for each test and procedure was taken from the NHS National Tariff 2013/14. A decision-tree economic model was used to evaluate the cost of 1,000 patients passing through the current treatment pathway compared with the same 1,000 patients after incorporating HeartFlow FFRCT. The authors found that introducing HeartFlow FFRCT to the pathway resulted in cost savings of £200 per patient. The EAC noted that although the derivation of costs in the study is explicit, details of the decision model structure are unclear.
5.4 Hlatky et al. (2015) investigated the quality-of-life and economic outcomes of fractional flow reserve CT (FFRCT) in the PLATFORM study (see section 3.17). Cumulative medical costs were measured over 90 days for each patient by multiplying a standardised cost weight for each medical resource by the number of resources used by the patient. Medicare reimbursement rates (the national average of technical and professional fees in the US) from 2015 were applied because cost weights and online pharmacy costs were used for drugs. Patients were prospectively assigned to either functional imaging (usual care, n=287) or coronary CT angiography (CCTA)/HeartFlow FFRCT (n=297). In the planed invasive group, mean costs were $7,343 among the CCTA/FFRCT patients and $10,734 among functional imaging patients (p<0.0001). In the planned non-invasive group, mean costs were not significantly different (p=0.26) between the CCTA/FFRCT patients ($2,679) and the functional imaging patients ($2,137). Overall, each quality-of-life (EQ‑5D) score improved at 90 days compared with baseline in the study population (p<0.0001), and scores improved more in CCTA/FFRCT patients than in functional imaging patients. In the invasive group, quality-of-life improvements were similar in both arms.
5.5 Douglas et al. (2016) published data on the 1‑year economic outcomes of FFRCT in the PLATFORM study. Costs were calculated in the same manner as the 90‑day results in Hlatky et al. (2015). In the planned invasive arm, the mean per-patient cost was $8,127 in FFRCT patients and $12,145 for usual care patients (p<0.0001). The cost savings at 1 year increased by 1.5% from the cost savings at 90 days. In the non-invasive arm, mean costs were not significantly different (p=0.82) between the FFRCT patients ($3,049) and the usual care patients ($2,579).
5.6 The company presented a decision-tree model based on integrating HeartFlow FFRCT into the existing diagnostic pathway at the time of its submission. A theoretical population of 1,000 patients with suspected coronary artery disease was allocated to either the current treatment pathway (based on the previous NICE guideline on chest pain) or the company's revised pathway, which included HeartFlow FFRCT. The cost consequences of the treatment pathways were compared based on the mix of diagnostic technologies used in each. The model had a 1‑year time horizon, included the impact of different testing strategies, and relevant clinical outcomes.
5.7 The proportion of patients eligible for CCTA as a first-line test and their probability of having coronary artery disease were taken from Rajani et al. (2015). In the model, 10% of patients were assumed to be ineligible for invasive coronary angiography (ICA), have an inconclusive CCTA result and have an uncertain single-photon emission CT (SPECT) result.
5.8 The diagnostic accuracy of HeartFlow FFRCT and its comparators in the company's model were based on per-patient level results reported in selected papers, as follows:
HeartFlow FFRCT: sensitivity 86%, specificity 79% (Nørgaard et al. 2014)
SPECT: sensitivity 76%, specificity 38% (Melikian et al. 2010)
CCTA: sensitivity 94%, specificity 48% (Meijboom et al. 2008)
ICA: sensitivity 69%, specificity 67% (Meijboom et al. 2008).
The cost of HeartFlow FFRCT (£888) was based on the company's original list price. Costs for comparator tests were based on 2014/15 hospital resource group (HRG) tariffs, as follows:
SPECT: £220 (HRG code RA37Z, nuclear medicine category 3)
CCTA: £136 (HRG code RA14Z, CT scan, more than 3 areas)
Calcium scoring: £77 (HRG code RA08Z, CT scan, 1 area, no contrast)
ICA: £1,241 (HRG code EA36A, catheter 19 years and over)
Percutaneous coronary intervention (PCI): £2,832 (weighted average of 2 tariffs, assuming that 25% of patients needing PCI will need more than 2 stents. HRG codes EA31Z [£2,704] and EA49Z [£3,216]).
5.9 The company's base-case results reported an average per-patient cost of £2,239 using the current pathway and £2,080 using the adapted pathway with HeartFlow FFRCT, representing an average saving of £159 per patient.
5.10 The company conducted 1‑way sensitivity analyses on the sensitivity and specificity of HeartFlow FFRCT and the comparator tests, as well as the costs of HeartFlow FFRCT. The analyses showed that HeartFlow FFRCT continued to be cost saving until its price reached £1,126. With regard to changes in the sensitivity and specificity, HeartFlow FFRCT remained cost saving for nearly all the values tested when considered in the context of the entire patient population.
5.11 The EAC incorporated the changes to the updated guideline on chest pain in the company economic model. In this context, the EAC assumed that HeartFlow FFRCT would be used following an initial CCTA, and that non-invasive functional imaging tests would subsequently be used only if the CCTA result were uncertain or non-diagnostic. The EAC reviewed the parameters and costs used in the company's model. It revised the company's sensitivity and specificity parameters for the comparator diagnostic tests, based on its own analyses of diagnostic accuracy (see table 1).
5.12 The EAC used the company's revised list price of £700 for HeartFlow FFRCT, instead of £888 as used in the company's model.
5.13 The EAC used the updated NICE guideline on chest pain to determine the costs of all comparator tests except MRI, to ensure that they were consistent with 2014/15 reference costs. The cost of MRI was taken from the Payment by Results tariff, because the chest pain guideline committee determined this to be more representative of the true cost. These costs were as follows:
SPECT: £367 (RN21Z, myocardial perfusion scan, stress only )
CCTA: £122 (RD28Z, complex CT scan)
ECHO: £271 (EY50Z, complex echocardiogram)
ICA: £1,685 (EY43A to EY43F, standard cardiac catheterisation)
MRI: £515 (RA67Z, cardiac MRI scan, pre and post contrast)
PCI: £2,865 (weighted average of 2 tariffs, assuming that 25% of patients needing PCI will need more than 2 stents. HRG codes EA31Z [£2,704] and EA49Z [£3,216]). Includes an estimated annual cost of £33 for medication following a PCI [aspirin and clopidogrel, British national formulary (2015)].
5.14 The EAC noted that the company's model did not include costs of drug therapy for patients having PCI. It consulted the NICE guideline on stable angina and estimated an annual drug treatment cost for these patients of £33 based on British national formulary (2015) prescription costs for aspirin and clopidogrel, and used a cost of £2,865 (PCI tariff with drug costs) in its revised model.
5.15 The EAC included a cost for optimal medical therapy. It obtained expert advice that optimal medical therapy usually consists of aspirin, statins, nitrates and beta blockers. Based on this information it estimated an annual cost of £84 (aspirin, simvastatin, glyceryl trinitrate and propranolol hydrochloride) from the British national formulary (2015) and used it in the revised model.
5.16 Using these updated assumptions, the EAC found a base-case cost saving of £214 per patient for HeartFlow FFRCT compared with the current treatment pathway for all functional imaging tests (SPECT, MRI and ECHO).
5.17 The EAC ran a number of sensitivity analyses, varying: the price of HeartFlow FFRCT; the diagnostic accuracy of the functional imaging tests, HeartFlow FFRCT, ICA and CCTA; and the proportion of uncertain CCTA and functional imaging tests. It also used estimates of diagnostic accuracy for CCTA and ICA from the updated NICE guideline on chest pain. In all instances, HeartFlow FFRCT remained cost saving.
5.18 The committee considered the cost modelling done by the EAC to be both appropriate and plausible. The committee heard from experts that percutaneous or surgical revascularisation is only offered to patients following ICA, and sometimes a confirmatory invasive FFR. The availability of data from HeartFlow FFRCT may help to plan treatment in individual vessels and patients.