No relevant economic studies on VibraTip device were identified by the sponsor or the External Assessment Centre. The sponsor submitted a de novo cost analysis using a decision tree model representing the flow of the entire UK diabetic population through the patient pathway for diabetic foot inspections over a 3 year time horizon. Full details of all cost evidence considered by the Committee are available in the assessment report overview.
The decision tree model submitted had 2 arms: a current practice and an intervention arm, in which a proportion of patients were tested with VibraTip and current practice was used for the remainder. The current practice arm used either the 10 g monofilament or the tuning fork. In the base case it was assumed that 40% of the patients in the intervention arm were tested with VibraTip. Over the 3 year time horizon, patients in the pathway had a risk of ulceration, repeat ulceration and amputation. The parameters and transition probabilities which impacted on the patient pathway included the probability of developing diabetic peripheral neuropathy, risk of foot ulceration and risk of amputation. These values did not influence the base case results because the costs and transition probabilities were identical for both arms. The difference in cost between the arms depended on differences in the cost between the devices per patient examination.
The sponsor estimated the per‑examination cost for each device based on the initial cost of the device, its estimated useful life and clinic throughput. The device costs in the sponsor's model were £9.95 for VibraTip, £15.20 for the 10 g monofilament and £28.80 for the tuning fork (all excluding VAT). The estimated per‑examination cost was £0.01 for the 10 g monofilament, £0.008 for the tuning fork and £0.002 for VibraTip. The value of VibraTip was based on a useable lifetime of 5000 activations. The External Assessment Centre did not agree with the sponsor's estimations of the pre‑examination costs and recalculated these values.
For the base case in the sponsor's model, with a 40% adoption of VibraTip in the intervention arm, the cumulative costs for monitoring 2.9 million patients over 3 years were £1467.86 million for VibraTip, £1467.91 million for the 10 g monofilament and £1467.90 million for the 128 Hz tuning fork. Therefore the base case results with diabetes suggested a saving over 3 years of £50,000 for VibraTip compared with the 10 g monofilament and £40,000 for VibraTip compared with the tuning fork. This is equivalent to a saving of 1.7 pence per patient over 3 years compared with 10 g monofilament; and a saving of 1.4 pence per patient over 3 years compared with the tuning fork.
The sponsor presented a sensitivity analysis in which the adoption rate of VibraTip in the intervention arm was varied. The results showed that the savings were proportionate to the adoption rate. The sponsor also conducted a 2‑way sensitivity analysis based on an assumption that VibraTip use was associated with a 1% relative risk reduction in ulcer formation compared with comparator devices; this was combined with a range of VibraTip adoption values. The results from this analysis showed a large increase in savings which reflected the high cost of treating ulcers and resultant amputation rates in the model. The External Assessment Centre noted that the assumption of a 1% reduction in ulcer rates was not based on the clinical evidence presented.
The External Assessment Centre considered that a weakness of the decision tree model chosen is that the number of states increases geometrically for each time step and that a longer time period would have been more appropriate, as diabetic peripheral neuropathy often occurs years or decades after diabetes is diagnosed. The External Assessment Centre also highlighted that there is no link between the clinical evidence, which describes differences in diagnostic accuracy between VibraTip and its comparators, and the assumptions used in the economic model. Thus, the model assumed the devices were clinically equivalent and any differences in costs were caused by differences in the technology costs per patient examination rather than any difference in diagnostic performance.
The External Assessment Centre highlighted that some of the assumptions and parameter values used in the model could not be located in the references cited and the rationale for using some parameters was unclear.
The External Assessment Centre considered that the unit costs presented by the sponsor were largely accurate, but that the sponsor's estimation of the per‑examination costs was incorrect. The External Assessment Centre recalculated the per‑examination costs for VibraTip based on the battery life evidence in Horsfield and Levy (2013) and obtained a range of values from £0.02 to £0.0995 depending on the number of sites (between 1 [1 foot] and 10 [5 per foot]) used per examination. For the Bailey's 10 g monofilament (a brand widely used in the NHS), the External Assessment Centre recalculated the per‑examination costs as £0.076 based on 10 sites per examination, from monofilament useable lifetime data in a published technical study (Lavery et al. 2012). The External Assessment Centre was unable to calculate a per‑examination cost for the tuning fork because of its unlimited useful life but considered that it would be very low.
The External Assessment Centre also highlighted a potentially important issue with both VibraTip and the 10 g monofilament, in that the operator may be unaware the devices are losing functionality (through battery discharge and reduced plasticity respectively). In both cases, clinical use beyond the devices' effective useful life would result in a reduced sensory force being applied to the patient and could result in increased false positives and associated increased costs.
In summary, the External Assessment Centre considered that the results of the de novo economic model did not provide comprehensive information with respect to the decision problem. If a cost minimisation analysis were adopted, including the per‑examination device costs only, VibraTip might be more expensive than either comparator under heavy usage, with a plausible per‑examination cost of between £0.0398 (for 4 sites per examination) and £0.0995 (for 10 sites per examination). This compares with a per‑examination cost of £0.076 for the 10 g monofilament assuming 10 sites per examination. The External Assessment Centre concluded that the economic case for the adoption of VibraTip had not been demonstrated robustly.
The Committee agreed with the External Assessment Centre's opinion that the economic model submitted did not fully address the costs and resources associated with the adoption of VibraTip. It recognised that the model did not capture any aspects of potential savings for changes resulting from differences in diagnostic accuracy between current practice and VibraTip. The Committee agreed with the External Assessment Centre that the introduction of any improvement in the diagnostic accuracy between the tests used to detect diabetic peripheral neuropathy would potentially have a substantial impact on overall clinical costs to the NHS. It noted that the small cost differences demonstrated in the current model were dependent only on the relative cost of the devices and the duration of their reusable lives.
The Committee concluded that further modelling would be needed of the economic case for adopting VibraTip, when this guidance is reviewed in the light of further research. This will need to include comparisons against the 10 g monofilament and the calibrated tuning fork. It will also need to include the impact of changes in diagnostic accuracy on long-term clinical outcomes, such as ulcer formation and amputation, which would manifest over at least a 5‑10 year period.