Long-term cost-utility analysis of remote monitoring of older patients with pacemakers: the PONIENTE study

The PONIENTE trial was conducted on older patients (81 years old on average) recruited in the Poniente Hospital (Almeria–Spain) implanted with commercially available pacemakers equipped with the Medtronic CareLink® Network. As previously reported [7, 30], 83 patients were initially included in the cost-utility study. Other relevant information about the PM technology, as well as its functioning mode, was mentioned in a previous paper [31]. Effectiveness results (functional capacity, event detection, and workload) were also depicted in an earlier paper [7]. Based on the study protocol, patients were followed-up for a period of 5 years after the implant date through scheduled in-hospital visits or remote monitoring. One month after the implant date, the physician scheduled an in-office visit with the patients, explained the characteristics of both types of monitoring and tried to determine which type of monitoring (remote or conventional) was better suited to the patient characteristics. Patients included in both groups were managed according to the usual practices of the hospital, abiding by the international guidelines with follow-up visits scheduled under physician’s criteria and regulation from the Spanish Health Authority. The number of visits to the hospital (CM = 7.49 and RM = 4.38 on average in the total 5-years of follow up period) and/or transmissions from home (RM = 6.62 on average in the total 5-years of follow up period) for each patient and year depended on the type of pacemaker implanted and patient characteristics.

Patients who opted for the RM were provided with a device (Medtronic Carelink Home Monitor) for data transmissions through a standard phone line from home to a central database, where data were processed and made accessible to physicians and healthcare team. It does not provide any additional therapeutic capabilities to the devices. The corresponding PM parameters were programmed, the use of the device as well as the protocol for sending data was explained and, finally, the service from the supplier company was requested by the physicians. Following the norms of the international consensus on the monitoring of cardiovascular implantable electronic devices [1], PM specifications, and physicians’ criteria, patients under RM were asked to submit data at different time points. No further in-hospital visits were scheduled for patients in the RM group. On the contrary, they were connected over the telephone and referred for an in-office visit, if data transmissions detected any device dysfunction or cardiovascular event. Data collected from all remote transmissions were scrutinized daily by the administrative staff and reviewed by the physicians through password-protected remote access.

Costs analysis was studied from two different perspectives (Table 1): firstly, costs incurred by the hospital, funded by the Spanish PHS; secondly, social costs afforded by the patients and their caregivers.

From the PHS perspective, according to the labor agreement of Poniente Hospital, labor costs were divided taking into consideration the two staff categories, physicians and administrative staffs, involved in the study. Only direct electricity supply costs were included as the consultation-room costs; other indirect costs were ignored. As previous research [28], ambulance direct costs were also taken into consideration. Pacemaker recipient costs were divided equally for both the groups. The costs for the Medtronic CareLink® Network (RM group) were contributed by Medtronic© and thus free for the PHS and patients.

Transportation costs (taxi [32], bus [33] or private transport [34]), as well as the employment income loss for the time spent by patients and (in case) caregivers in every consultation visit [35], were considered from the patient’s perspective.

The model compares intervention costs to clinical outcomes for each patient in terms of QALYs. QALY is a generic measure of disease burden, including both the quality and the quantity of life lived, used to assess the value of medical interventions. One QALY equates to 1 year in perfect health. In our study, the QALYs values were calculated by aggregating the utility scores of the EuroQol–5D (EQ–5D) Spanish validated version [36, 37] submitted by each patient at the end of the twelfth month (by 1 year) and fifth year (by 4 years), after applying the selected discount rate. However, due to temporary comparative reasons, the utility scores collected at the implantation date and after 6 months were ignored in this study. The health-related quality of life (HRQoL) in terms of utilities was measured by EQ–5D (from − 1 [the poorest imaginable health state] to 1 [perfect health]). The cost-utility ratio was evaluated by comparing differential QALYs and differential costs between CM and RM groups, after 5 years. The incremental cost-effectiveness ratio (ICER) was determined by calculating costs per QALYs gained for RM as compared to the CM group in the present study. According to the recommendations of the US Panel on Cost-Effectiveness in Health and Medicine [38, 39], future costs and QALYs were discounted by 3%.

The robustness of base-case results was evaluated by conducting a probabilistic sensitivity analysis (PSA) using Monte-Carlo simulation with 5000 iterations per group, as shown by existing research [40]. In PSA, uncertainty was evaluated by simultaneously varying stochastic variables in the Cost-per-patient model (Table 1) through random sampling from their assumed distributions (Table 2).

Statistical analysis was conducted using SPSS (Statistical Package for Social Sciences) statistical software v.18.0.0 (SPSS Institute, Inc., Chicago, IL, USA). Differences between groups were compared using Mann-Whitney U-test for continuous independent and non-normally distributed variables. Results were presented, including the corresponding 95% confidence intervals (95% CI). Logistic regression was used to assess the influence of age, sex, and the method of monitoring attrition.

After the five-year follow-up period, 55 patients finished the PONIENTE trial (CM: 34; RM: 21) with a mean age of 81 ± 7 years. The female population of the patients comprised of 31%. Significant differences with regards to the assessed clinical characteristics were not observed between the study groups. As shown in Table 3, syncope was the most important disease manifestation (60%), and 47.27% of the patients experienced paroxystic Atrial Fibrillation episodes. The atrioventricular block was the most common pacing indication (71%), followed by sick sinus syndrome (20%).

Owing to non-cardiovascular causes, 22 patients (RM: 8; CM: 14) died, whereas 6 patients (RM: 1; CM: 5) withdrew from the study. The loss of subjects at 5 years did not correlate with the sex (p = 0.910), the age of the participants (p = 0.880), or the study criteria variable, type of monitoring conducted (p = 0.110).

After the five-year follow-up period results in Table 4 revealed that total costs per patient were 23.02% lower for the RM group than the control group (CM), corresponding to a saving of €82.10 per patient (RM: €274.52 ± 128.45; CM: €356.62 ± 144.12; p = 0.033).

As shown in Table 5, statistically nonsignificant differences were revealed between groups alongside the five-year follow-up period.

At the time of enrolment in the study, the overall mean EQ–5D questionnaire of health-related quality of life scores in utilities showed better for patients in the CM group (EQ5DM00 = 0.72) in comparison to RM group (EQ5DM00 = 0.70). After the first 12-months follow-up period, figures became better in the RM group (EQ5DM12 = 0.90) than in the CM group (EQ5DM12 = 0.82). However, at the end of the study, after 5 years months of follow-up, the reduction was evident in the overall mean scores for the RM group (EQ5DM60 = 0.68) than the CM group (EQ5DM60 = 0.77). The functional capacity scores were discreetly maintained or slightly declined at 5 years after implantation. Evolution of EQ–5D utilities along the study was depicted in Fig. 1.

Costs/QALYs obtained at 5 years follow up by the RM group (€225.69) were higher than the CM group (€177.08), although this data was nonsignificance statistically (p = 0.773). Patients belonging to the CM group manifested a slightly better QALY at the end of the study (3.58) than the RM group (3.31). Regarding that, the RM group reported an €82.10 cost-saving per patient (p = 0.033), the ICER of CM in comparison to RM became positive (€301.16) as a measure of cost-effectiveness of RM in contrast to CM (Table 5).

The results of the robustness test in Table 6 were in accordance with the data obtained in the previous sections. Statistical analysis confirmed the overall costs per patient in RM group was significantly lower than the CM group, corresponding to a €53.68 cost saving per patient (RM: €384.92 ± 187.35; CM: €438.60 ± 324.29; p < 0.001). Significant statistical difference was, however, lacking in the labor costs between RM and CM groups (RM: €221.70 ± 99.03; CM: €222.70 ± 171.67; p = 0.720) as a consequence of the effect of physician labor costs (RM: €204.92 ± 91.68; CM: €207.96 ± 160.31; p = 0.244) and a contradictory effect of administrative staff costs (RM: €16.78 ± 7.47; CM: €14.74 ± 11.36; p < 0.001). Transportation by ambulance, as well as consultation room costs, was lower in the RM group. Therefore, the RM group saved a cost of 4.50% per patient from the PHS perspective (RM: €288.36 ± 134.59; CM: €301.95 ± 201.34; p < 0.001). Moreover, from patients’ perspective also a significant cost saving of RM group was confirmed by the robustness analysis (RM: €96.56 ± 86.20; CM: €136.64 ± 153.12; p < 0.001), as a consequence of a significant difference in travel cost in RM group as well as costs-saving in the RM group in comparison to CM group with respect to income loss of older adults patients and caregivers in every in-office visit.

A positive ICER of CM in comparison to RM (€501.48) affirmed by the cost-utility robustness analysis (Table 7) opined that RM remained a cost-utility alternative in comparison to CM. As in thse primary model, slightly better QALY figures for patients in the CM group (CM: 3.48 ± 2.48; RM: 3.38 ± 2.19; p = 0.022) were compensated by a cost per patient reduction in the RM group. Moreover, unlike the primary model, Costs/QALYs obtained by the RM group (€192.43) were significantly (p < 0.001) lower than the CM group (€238.73), as computed by statistical analysis.

The mean number of scheduled and unscheduled in-hospital visits per patient was reduced by 41.52% for the RM group, in line with existing research [14, 27, 41, 42]. These outcomes reinforced the results obtained in the 12-months PONIENTE study [30] that showed a reduction of in-hospital visits by 26.89%. Moreover, more recent research [11] confirmed this finding after evaluating a reduction of 79.20% in the number of in-office visits for 445 patients with PMs and Implantable Cardioverter Defibrillator (unlike our study that only focuses on patients with PMs) followed-up during more than 24 months through RM plus remote interrogations every 6 months. This reduction incurred a significant cost saving of 44.88% from the patients’ perspective, diminishing travel cost for the older adults as well as the level of income loss of patients and caregivers in the RM group.

However, after a five-year follow-up period, this reduction of in-hospital visits failed to display a significant impact on the costs from the PHS perspective, with a cost saving of 16.13% per patient (€215.48 vs. €256.92; p = 0.112). Although a significant reduction in costs from the PHS perspective of 57.64% in the RM group (p < 0.001) was reported in our 12-months PONIENTE study [29]. This finding was in accordance with the previous studies [5, 10, 25, 27, 43‐45]. After 5 years, the management costs of patients under RM seemed to be unbalanced by the incremental burdens of following up automatic wireless remote notifications by physicians and administrative staff. More recent pacemakers technology is presented as opportunities to allow constant patient “big data” collection, to optimize the outcomes of patients. Henceforth, for future research on remote monitoring, this should be taken into consideration. Thus, as recently suggested, including higher survival ratios in comparison to periodic in-hospital evaluations may be achieved through more frequent (even daily) RM transmissions [46]. However, higher efforts must still be made by the clinicians to enable a more efficient practice. With the aim to prioritize clinical decisions and interventions as well as to reduce practitioners’ burdens automatized RM data triage through artificial intelligence methods [46] may be implemented.

Although the 12-months PONIENTE study showed an increment in QALYs (+ 0.09) among older patients with PMs under RM as compared to those under CM, our results recorded a higher QALYs decrease (− 0.20) for the RM group after the five-year follow-up period. On the contrary, a negative effect on the patients’ quality of life was not published by other studies, although for shorter follow-up periods (maximum 24 months) [25, 47, 48] even by using other questionnaires (i.e., SF–36 questionnaire) [10]. These results may indicate a different evolution of utilities in patients under RM for a longer period. Published reports in the literature revealed long-term undesirable effects on the quality of life of patients with pacemakers implantation under conventional monitoring [49]. This previous study observed a moderate increase in the SF–36 general scores for patients with bradycardia pacemaker implantation for the first 24 months post-implantation, which was followed by a gradual decline over time. Physical reasons were the causes of this gradual decline, unlike mental scores, as a consequence of non-cardiovascular comorbid diseases usual among geriatric patients. Thus, persistent increase in mental scores drawn from emotional problems and social dysfunctions reflected improved well-being after PM implantation [49]. According to that, some of the major RM benefits from patients’ perspectives drawn from the reduction of their in-hospital visits may not be appreciated in the long term in comparison to CM. Moreover, this indicated the need for better patient education regarding the purpose and benefits of RM, its usability, and safety [16], contrary to the findings in short-terms remote monitoring studies [25]. This leads us to the wide debate around social acceptance of telehealth. For shorter follow-up periods, some research does state that a well-structured tele-medical centre available 24 h a day to act promptly according to the individual patient’s risk profile in addition to educational activities [50], as well as continuous ‘remote’ contact with patients sending clinical reports [11], contribute to balance patients’ acceptance. Other studies have found that patients may prefer an in-hospital consultation to establish a diagnosis or to plan their treatment [19]. In the long term this situation may be assimilated to the follow-up of patients with PM. However, to our knowledge, this is the first study where the EQ–5D questionnaire has been used to assess the HRQoL in older patients with PMs under RM for a long-term follow-up period of 5 years. Thus, QALYs obtained could not be compared with those in previous papers on the RM of users with PMs.

Regarding both, PHS and older patients’ perspective, cost-utility analysis confirmed that after a 5-years follow-up period, slightly better QALY figures for patients in the CM group may be compensated by a cost per patient reduction in the RM group, assessing in €301.16 the ICER of CM in comparison to RM.

Although no statistically significant differences were found between both groups in baseline variables assessed (Table 3), the first methodological limitation lies in the fact that it is a non-randomized study where the type of monitoring was recommended by older patients together with the physicians, according to the patients’ characteristics. However, no effect of other essential variables on the outcomes was found by the propensity score matching, conducted to explore the variables influencing the election of the type of monitoring [6, 30]. In any case, greater external validity seems to be achieved as the assignment form used fits the daily practice, and probably the outcomes will be closer to those expected in the real world [51, 52]. Furthermore, the basal characteristics studied were very similar to those reported in the Spanish Pacemaker Registry [53], which supported its generalization. Secondly, it is an open trial where the type of monitoring for every patient was known by both the patients and physicians. This might indulge in biasness in the patients’ and physicians’ behavior. Third, other decives more than pacemakers, like implantable cardioverter defibrillators (ICD) or cardiac resynchronization therapy devices (CRT), have not been inserted in this study. Then, our results can not be extended to complete devices patients. Fourth, differences of sample size among groups, as well as the limited population size drawn from a single center with a limited number of implants per year, might contribute to the reduction in the statistical power of the study. The similarity of the results with existing research [3, 5, 8, 14, 15] does not suggest the influence of differences in the size of both groups on our findings. The level of attrition along the five-year follow-up period may also have affected the results. However, the analysis of attrition demonstrated the similarity between participants who left and those who remained in the program for both groups of monitoring. In any case, as far as the loss of subjects is concerned, the methodology followed in this study prevented generalization of the results among non-survivors. In the same way, generalization among other European Health Public Systems should be cautious as the sample was concentrated in a single Spanish hospital. Finally, the final cost difference between both groups may be affected by the fact that the CareLink Monitor associated cost was afforded by the supplier (Medtronic©), so it was not considered within either the PHS or the patients’ costs.