**4. Discussion**

#### *4.1. In Relation to the Study with 2016 Data*

The study concludes that, in the given context, telemedicine is an unequivocally preferable option to usual care from an economic point of view. The strength of this diagnosis is similar to that derived from the analysis performed with 2016 data for the specific case of teledermatology, i.e., the result of including other specialties (teleulcers, teleophthalmology and teleaudiometries), lengthening the time period (by using the complete sample available) and adding the indirect cost approach of the caregiver results in savings per visit 35% above the base case studied by Vidal Alaball et al. with 2016 data [12] (Table 6). We note that once caregivers' opportunity costs are introduced, the most important differential corresponds precisely to the calculation of the cost in terms of the time of users. The similarity of results between the different types of costs reflects the robustness of the methodology used.


**Table 6.** Differential costs per visit. Comparison between studies.

### *4.2. Sensitive Variables*

The magnitude of the result is highly sensitive to the parameter corresponding to the opportunity cost (lost productivity) of the user and this has been calculated homogeneously among the different beneficiary profiles (minors, of working-age and retirees); although an eventual differential calculation by profile would not change the results, it would far better approximate the representative total of the savings. It should be borne in mind that in contexts with higher labor productivity of both professionals and users, the results of the analysis would be much more favorable to telemedicine.

With regard to the extrapolation of these conclusions and with the "travel time" factor, it is worth keeping in mind that the study was performed in a mostly rural and semirural setting. The average distance per journey may be higher than in urban settings, although it is not clear if the journey time would be higher (as moving within a city is much slower). Whatever the case, the results show that both factors (i.e., travel cost and time lost) are sufficient to reach the same conclusion, namely, that even if telemedicine did not save on travel costs (being "zero kilometer"), it would be cost-effective, and even if it did not save anything in terms of time (for the user and the healthcare system), it would also be cost-effective.

As to the assumption that patients travel by car, it is reasonable to assume that some of them use public transport. If we consider this possibility, telemedicine savings would be even higher, since in rural settings, where the frequency of public transport is very low, the potential savings in terms of travel costs (using public transport instead of private transport) would clearly be far outweighed by more travel time (with and without waiting time). In the context involved in the study, which was almost devoid of a railway network (except in the south of the city of Manresa), it is unlikely that the bus is faster than private transport.

#### *4.3. Factors not Included in the Analysis*

While it is true that this assessment includes the differential essential elements between the two analyzed models, it does not include objective or easily monetizable intangible factors such as the users' and professionals' satisfaction with the service or the improved managemen<sup>t</sup> of cases in function of their clinical severity. This improvement in care managemen<sup>t</sup> could reduce waiting lists to the access of GPs, one of the biggest problems in the Catalan healthcare system. In this context, telemedicine allows for better allocation of care time according to the complexity of the case. Future lines of research ought to quantify these factors, which are complementary but key in order to evaluate the service's e ffectiveness.

In addition, the type of analysis performed assumes that clinical e ffectiveness is equivalent. Although a time period which includes aspects strictly related to managemen<sup>t</sup> seems su fficient to make a good diagnosis, as is the case, and despite the complexity of the information which would be needed, we ought to try to ensure the hypothesis of equivalence in health impact and add any significant and di fferential costs which go beyond and which can be calculated in a rigorous manner.

It needs to be borne in mind that as doctors are remunerated, their increased productivity does not imply a direct translation into the healthcare provider's income account; instead, the freer the practitioners, the fewer practitioners the healthcare provider will need to hire. In other words, savings might occur in the mid-term, as opposed to the short term.

It should also be considered that the increased ease with which referrals can be made might have incentivized GPs to use interconsultation as a second opinion tool to support the diagnosis of patients they would normally have treated. This might have increased the ratio of saved face-to-face visits.

Finally, it should be mentioned that the study also assumed that the di fferential cost of expenses such as cameras or clinical software is zero, since this was the case, but in the case of introducing this service from scratch in another context, these costs would have to be taken into account. In any case, the magnitude of the savings made by the service makes it unlikely that including them could significantly alter the results of the analysis.
