In this cost-effectiveness evaluation of SCAD, a home-based interactive telemonitoring service for HF in France, we identified significant clinical benefits (one life-year gained in the ten years following initiation of the programme in the base case of the model) for a relatively modest cost (€4,245 over the lifetime of the model). The estimated ICER was €6,491/QALY. An intervention is considered to be cost-effective when the estimated ICER is below the willingness-to-pay threshold in the given country. While there is no formal willingness-to-pay threshold in France, a threshold of €150,000/QALY has been proposed . The ICER for the SCAD programme is well below this threshold and, in the probabilistic sensitivity analysis, would be below a willingness-to-pay threshold of €11,800 in 90% of simulations.
Concerning the benefits of the home telemonitoring programme, in the analysis of SCAD participants in the SNDS database, the rate of unplanned hospitalisations for a cardiovascular diagnosis was halved, the rate of unplanned hospitalisations for HF divided by three, and mortality reduced by around twenty percent . In the present Markov model, these benefits translate into one life-year gained and 0.65 QALYs gained over the time horizon of ten years. These gains can be considered to be clinically important.
With respect to costs, the most expensive component, both with the standard care strategy and with the SCAD strategy, was the management cost of patients with HF who were not hospitalised. This was also the cost item that was associated with the most variability, ranging from zero to >€80,000 for the standard care strategy. Due to this variability, the management cost of non-hospitalised patients was the most important determinant of the precision of the ICER estimate. Although the management cost of non-hospitalised patients was higher for the SCAD strategy than for the standard care strategy, this cost differential was partly offset by reduced costs of hospitalisation and palliative care.
The ICER for the SCAD programme varied over the lifetime of the model, increasing with the time horizon applied. This relationship can probably be explained by longer survival of patients included in the programme. These survivors continue to consume healthcare resources for longer, increasing total cost, whereas in standard care, more rapid attrition limits the cost of healthcare over the long term. In contrast, in the absence of data on the effectiveness of the SCAD programme for longer periods, efficacy inputs into the model were identical after five years, and for this reason, utility outcomes after this time would not be expected to differ between strategies. However, even if the time horizon of the model is extended to the patient’s entire life expectancy, the ICER remains well below €10,000/QALY. When the time-horizon of the model was limited to five years, which is the longest period for which observed data on the impact of the SCAD programme on benefits and costs are available, the SCAD strategy is actually dominant over the standard care strategy, being both more effective in terms of QALYs gained and cost-saving. In the model, a conservative approach was taken in which the benefits of the SCAD programme were limited to the five-year period for which data were available. It is possible that as more long-term data becomes available, estimations of the ICER for time horizons beyond five years may change.
The scenario analyses indicated that the ICER was lower by around ten percent for patients with preserved ejection fraction compared with those with reduced ejection fraction. As a result, SCAD programme appears to be more cost-effective in patients with preserved ejection fraction. This may be explained by the fact that patients with preserved ejection fraction are on average older and have had more previous hospitalizations than patients with reduced ejection fraction. With the availability of more long-term data, it may be possible to evaluate whether participation in the SCAD programme actually slows the decrease in ejection fraction over time in patients with HF.
The scenario analysis in which all patients entered the model in the ‘not hospitalised after one previous hospitalisation’ state corresponds to how the SCAD programme currently operates, where only patients who are hospitalised are offered participation in the programme. The base case represented a conservative hypothesis, where certain patients are never hospitalised are also included in the model, and who will contribute equally to both the standard care and the SCAD strategy. In the base case, this will have the effect of diluting incremental utility and cost differences between the two strategies. The lower ICER determined in the scenario analysis may for this reason reflect more accurately the cost-effectiveness of the SCAD programme as it is operated today.
Although telemonitoring programmes such as SCAD bear a cost, which is attributable both to the cost of the programme itself and to higher management costs related to longer survival of the patients, this cost is relatively small compared to the total cost to health services of HF. In France, there around 540,000 patients managed for chronic HF, who generated a total cost to national health insurance in 2013 of €1,186 million . Since SCAD programme is dominant in the short term (between 1 and 5 years), making it available to all patients at the current cost (€470 for 6 months) would generate short-term savings. Indeed, even though the cost of SCAD programme for all these patients would represent an additional cost of €250 million, the savings generated from the reduction in hospitalizations would reduce total health insurance expenditures. As shown in the final scenario analysis, even when paying for patients to continue in the programme for ten years (rather than for six months as in current practice, modelled in the base case), the estimated ICER remains below €10,000/QALY and the utility benefit is greater.
The current COVID-19 pandemic has illustrated the attractiveness of telemedicine interventions for managing patients with chronic diseases in their homes . Given their clinical efficacy and their cost-effectiveness, it may be expected that home telemonitoring programmes such as SCAD will be introduced in other diseases or extended to other hospitals or regions. It should however be noted that implementation of the SCAD programme bears a specific cost, which needs to be taken into account. In particular, functioning of the programme requires funding for the full-time involvement of a trained nurse dedicated to patient monitoring and alert management.
Comparison of the present findings with those of cost-effectiveness analyses performed in other countries is not straightforward, due to differences in the nature of the telemonitoring programme considered and in how they are financed. Nonetheless, recent studies of intense telemonitoring programmes for heart failure have been consistent in showing them to be cost-effective. For example, a Markov model similar to the present one has been used to evaluate cost-effectiveness of telehealth programs for congestive heart failure in the context of the United States health system . Inputs were derived from a meta-analysis of multiple home telemonitoring programmes . At the five-year time horizon, the authors found that enrolment in such a programme would result in cost savings of $4,456 with a gain of 0.50 life years. In the European context, data from the Trans-European Network–Home-Care Management System (TEN-HMS) study  were used in a Markov model involving transitions between different NHYA severity classes . At the twenty-year time horizon, the ICER was estimated to be €12,479/QALY. Most recently, Vesterggard et al. have reported on the TeleCare North HF study in Denmark . This is a telemedicine programme implemented by nurses with a therapeutic education component and remote monitoring of clinical data provided by the patient. This cost-effectiveness analysis was not a modelling study but used real data on utilities and costs collected from patients participating in the programme. This analysis found the telemonitoring strategy to be dominant over standard care, with a net monitory benefit of £5,164 (approximately €6,100) at a time-horizon of one year. Taken together, the findings of these different studies provide a strong argument that intensive telemedicine programmes are a cost-effective way to manage patients with HF in different healthcare systems.
The study presents certain limitations. The model inputs come from multiple published sources, principally the ODIN study  and the SCAD-SNDS cohort , and the different source populations may not be fully comparable. In addition, in the SCAD-SNDS cohort, which was used as the source of the relative risks of hospitalisation and mortality, there was no control group without home telemonitoring, and the event rates in the low adherence group were used to represent standard care. However, in the SCAD study, even low-level users had lower rehospitalisation rates compared to the period before joining the SCAD programme, suggesting that the low adherence group may gain some benefit from the programme compared to non-participants. For this reason, the incremental gain in QALYs compared to standard care may have been underestimated.