This study evaluated the long-term efficacy, tolerability, and quality of life of VNS therapy in clinical practice using both types of stimulator. Over the 10-year observation period, 32% of patients treated with iVNS had a sustained reduction in seizures and were identified as responders. For tVNS users, 50% reported a permanent seizure reduction at 5 years, with a responder rate of 30%.

Our results with tVNS are consistent with those of a recent randomized double-blind study showing a 27% responder rate [20, 23]. A pilot study found a gradual increase in responder rates over time from 28.57% to 53.85% [24].

A few studies described similar responder rates of 24.5–46.6% for iVNS treatment [17, 41, 42], but other studies found significantly higher responder rates of around 44.4–64.4% [16, 18, 19, 43].

Almost all of these previous studies of iVNS and tVNS described a reduction in seizure frequency over time, which could only be verified in our tVNS cohort.

This slight difference may be due to the limitations of systematic seizure frequency recording in our retrospective study design, as well as our severely affected patient population. In addition, due to our small sample size, patients with increasing seizure frequency had a greater impact on our results. This could also explain the findings of the survival analysis including tVNS, iVNS, and ASM patients combined, emphasizing the need for further large-scale investigations.

Seizure freedom was achieved in only a minority of patients treated with iVNS (5.9%), which is consistent with previous findings of approximately 6–9% [18, 19, 43].

By contrast, the seizure freedom rate in the tVNS cohort was relatively high (20%), but this may be due to the small sample size or concomitantly introduced ASM. Other studies have reported seizure freedom rates as low as 2.7% [23].

In line with previous studies, our results confirm the efficacy of VNS therapy in the treatment of refractory epilepsy.

No significant predictors of seizure frequency reduction were identified in the present study. In line with previous studies [8, 36, 43], we found no difference in treatment response with regard to the epilepsy type or seizure semiology, whereas a recent systematic review and examination of patients from the VNS Therapy Patient Outcome Registry on predictors of seizure freedom suggested that treatment response is more frequent in patients with generalized seizures [15].

It should be noted that the number of patients with generalized epilepsy in the present study was low, which may influence the results, but in conjunction with the controversial findings of previous studies, the type of epilepsy does not seem to have a major influence on seizure control.

Regarding stimulation settings, there was no improvement in seizure control with rapid cycling compared to standard cycling, which is consistent with other studies [24, 25, 36, 44, 45]. The disadvantage of higher battery consumption in rapid cycling suggests that the decision to switch to rapid cycling should be made on a case-by-case basis.

In our study, concomitant ASMs were changed and adjusted. Consistent with other studies [26, 44], the number of ASMs remained almost stable with a tendency to increase after 5 and 10 years of treatment. Although some previous studies have reported a similar treatment response without a change in medication [6, 43, 46], a synergistic effect of VNS and medical treatment can be assumed.

Adverse effects reported by our patients were consistent with previous studies [20, 36, 47]. On average, patients reported only a single side effect, indicating that all VNS therapies are well tolerated.

Patients reported a significant improvement in quality of life after VNS treatment. Positive effects on mood and concentration were also reported. These results are in line with recent studies [20, 28, 30, 36]. Some authors suggest that the improvement in well-being is independent of seizure control [17, 29, 37]. Our results support this assumption, as significantly more patients (59.5%) would choose VNS therapy again than benefited from it in terms of seizure frequency reduction (32%).

Our second aim was to compare invasive and transcutaneous stimulation.

No differences were found between iVNS and tVNS users in terms of response rates, quality of life, and effects on mood and concentration. The profile of adverse effects was different, but few events occurred with either type of stimulator. A high percentage of patients in both cohorts would choose VNS therapy again. It can be concluded that invasive and transcutaneous VNS seem equally effective, safe, and well tolerated. Our results should, however, be interpreted with caution as the sample of the tVNS group was small.

The marked differences, with iVNS patients being both younger and having a shorter duration of epilepsy, potentially limit the comparability between iVNS and tVNS; however, they may reflect a clinical reality in which VNS is more often prescribed to drug-resistant epilepsy patients who lack the cognitive resources or maturity to implement and adhere to tVNS treatment.

The limitations are mainly due to our retrospective study design, which made complete data collection difficult and resulted in a high heterogeneity of the patient population. However, it is worth noting that the present study is highly reflective of clinical reality, in which there was no artificial adjustment of ASMs, stimulation settings, or other influencing components.

Another limiting factor was our relatively small sample size, especially the tVNS group and at later follow-up time points, which may have led to a potential selection bias. In addition, in many cases we had to rely on retrospective information from patients, family members, or caregivers. Therefore, inaccuracies in reporting must be taken into account.

We did not use standardized scales of quality of life, well-being, and mood, as these would have been too complex to assess retrospectively by telephone interview; however, our results are in line with current literature.