Essays on the Health Economic Evaluation of Remote Monitoring in Healthcare

Abstract

In recent years, the Belgian healthcare system has faced numerous challenges, including an aging population and growing patient expectations for more active involvement in their care. These challenges have placed substantial pressure on healthcare budgets and increased the workload for healthcare professionals. Concurrently, technological advancements have created new opportunities to address these challenges. One such innovation is remote monitoring (RM), defined as the use of audio, video, and other telecommunication and electronic information processing technologies to monitor patients’ status at a distance. Despite its broad applicability, implementation in clinical practice remains limited due to a lack of conclusive evidence on clinical and economic outcomes. This dissertation contributes to addressing this gap by examining the economic added value of RM for two patient groups: pregnant women at risk of gestational hypertensive disorders (GHD) and patients with heart failure (HF). Both GHD and HF are cardiovascular conditions characterized by abnormalities in blood pressure regulation, though they differ in onset, chronicity, and patient population. GHDs represent an acute, pregnancy-specific condition that typically resolves after delivery, whereas HF is a chronic, progressive disease requiring lifelong management. By examining these two contrasting contexts, one acute and time-limited and the other chronic and ongoing, this dissertation illustrates the wide-ranging potential of RM to support cardiovascular care across different stages and patient needs. To comprehensively assess the economic value of RM, this dissertation applies an integrated health economic evaluation framework combining three complementary perspectives: (1) monetary valuation, captured through willingness to pay (WTP) analyses; (2) experience value, measured via patient-reported outcome and experience measures (PROMs and PREMs); and (3) health value, quantified through quality-adjusted life years (QALYs) in a cost-effectiveness analysis. GHDs represent a significant global health concern, affecting approximately 5 to 10% of all pregnancies worldwide and ranking among the leading causes of maternal and perinatal morbidity and mortality. Pregnant women at risk of GHD therefore require a more intensive follow-up, in which regular blood pressure monitoring plays a central role. Traditionally, this follow-up is conducted in a hospital setting. However, RM offers a promising alternative by enabling continuous assessment of patients’ health status from home. Another potential alternative is patient self-monitoring (PSM), a form of RM in which data are not automatically transmitted to or reviewed by healthcare providers. Instead, patients monitor their own blood pressure and are instructed to contact their care team if readings fall outside prespecified target ranges. Existing research on RM and PSM for pregnant women at risk of GHD remains limited and has focused mainly on clinical outcomes. However, beyond traditional health outcomes, these home blood pressure monitoring approaches may offer additional benefits that patients highly value. To capture the complete set of benefits perceived by patients, we applied the contingent valuation (CV) method to elicit patients’ WTP for RM and PSM. Based on insights from the literature and a pilot study, a CV survey combining a payment card and an open-ended follow-up question was developed and administered to the participants of the Pregnancy Remote Monitoring (PREMOM) II study at two time points: between 11 and 20 weeks of gestation and at six weeks postpartum. PREMOM II was a multicenter randomized controlled trial (RCT) involving pregnant women at increased risk of GHD. The trial compared two intervention groups, RM and PSM, with a control group receiving conventional care. The results indicate that patients value both RM and PSM more highly than conventional care, with an average WTP of approximately € 120 for RM and € 80 for PSM, in addition to the standard pregnancy invoice. These findings suggest that RM and PSM are perceived as meaningful alternatives to conventional care. Given that patients value both RM and PSM more highly than conventional care, an important question is whether there is a difference in patients’ valuation between these two home blood pressure monitoring approaches. The only distinguishing feature is the midwife’s follow-up of blood pressure readings in RM, compared with the self-management approach in PSM, so any observed difference in valuation can be attributed to the presence of this follow-up. Using data from the CV survey administered to the participants of the PREMOM II study, we applied two-part models that combined a logistic regression and a generalized linear model (GLM). The analysis shows that, at six weeks postpartum, the average WTP in the RM group was approximately € 50 higher than in the PSM group. This difference reflects the added value of the midwife’s follow-up in RM compared with the self-management approach in PSM. By including additional control variables in our two-part model, we were able to assess whether the added value of the midwife’s follow-up is driven solely by implicit follow-up (i.e. reassurance that the midwife is reviewing the patient’s data), or whether explicit follow-up (i.e. direct interactions and active interventions by the midwife) also plays a role. This strategy also allowed us to examine the mechanisms through which the midwife’s follow-up contributes to the perceived added value. First, we found that both implicit and explicit follow-up contribute to the added value of the midwife’s follow-up in RM. Second, the results show that the midwife’s follow-up enhances perceived added value through multiple mechanisms, including improved technology acceptance, a greater sense of safety, enhanced insight into their own health status, a stronger feeling of being closely monitored, higher adherence, and improved maternal Health-Related Quality of Life (HRQoL). In addition to patients’ monetary valuation of RM and PSM, it is important to understand how pregnant women at risk of GHD actually experience these home blood pressure monitoring approaches. As the primary users of these technologies, patients’ experiences are crucial for their successful implementation. However, quantitative research on patients’ perceptions regarding RM and PSM for pregnant women at risk of GHD is scarce and limited to single-group, single-time-point descriptive studies. To address this gap in the literature, we developed a survey focusing on PREMs and PROMs which was administered to the participants of the PREMOM II study at four time-points: (1) study inclusion between 11 and 14 weeks of gestation, (2) 20 weeks of gestation, (3) 26 weeks of gestation, and (4) six weeks postpartum. PREM questions were based on a modified version of the Telemedicine Perception Questionnaire, while PROM questions were selected from the patient-centered outcome measures set for Pregnancy and Childbirth developed by the International Consortium for Health Outcomes Measurement. Regarding PREMs, both the RM and PSM group reported overall satisfaction with the technology they used at six weeks postpartum. However, t-tests revealed that women in the PSM group were less likely than those in the RM group to perceive their technology as time-saving, to feel that the midwife understood their concerns well over the phone, or to believe that their health status could be adequately monitored remotely. Additionally, a greater proportion of women in the PSM group viewed the absence of in-person contact as problematic. To evaluate if RM and PSM create value for patients, dichotomized PROMs were compared between women who experienced RM or PSM and those receiving conventional care using linear probability models. At six weeks postpartum, women in the RM group were more likely to be satisfied with the results of their care, to feel confident in their role as a mother, and to report a positive birth experience compared to women in the control group. They were also less likely to express concerns about their own health. No robust evidence was found of significant differences in PROMs between women in the PSM group and those in the control group. As healthcare systems face growing financial constraints, it is also important to assess whether new technologies offer good value for money, ensuring that investments yield the greatest possible health benefits. To date, however, no studies have evaluated the cost-effectiveness of RM or PSM for pregnant women at risk of GHD compared with conventional care. To advance knowledge in this area, we conducted a cost-effectiveness analysis from a healthcare payer perspective embedded within the PREMOM II study. A repeated decision tree model was developed to capture both maternal and infant costs and effects over a time horizon spanning 40 weeks prior to delivery and 30 weeks postpartum. The EQ-5D-5L questionnaire was used to assess mothers’ HRQoL at inclusion, at 20 and 26 weeks of gestation, and at six weeks postpartum. At six weeks postpartum, mothers were also asked to complete the EQ-VAS on behalf of their infant. Patient-level cost data were obtained from the InterMutualistic Agency (IMA). The analysis reveals no significant differences in either costs or QALYs between RM or PSM and conventional care. These findings remained unchanged when the analyses were repeated separately for mothers and infants. As a consequence, based on our cost-effectiveness analysis from a healthcare payer perspective, there is insufficient evidence to conclude that RM or PSM is more or less cost-effective than conventional care. However, as demonstrated by the WTP analysis and the evaluation of PROMs and PREMs, patients reported substantial non-health-related benefits associated with these technologies. Using data from the PREMOM II study, we also contributed to an ongoing methodological debate in the CV literature: Which respondent group is most appropriate for completing the valuation task for a given treatment, and how does this choice influence WTP estimates? First, we investigated whether the WTP for RM and PSM differs between patients with experience using these technologies and those without, using data from the CV surveys administered to the PREMOM II study participants. Two-part models, combining a logistic regression and a GLM, were applied at each measurement point. Our analyses provide robust evidence that patients who experienced RM during their pregnancy reported a significantly higher WTP for RM at six weeks postpartum compared with patients in the control group. In contrast, no robust significant effect of experience was found for the WTP for PSM, possibly because using PSM is similar to operating a standard blood pressure monitor. These findings suggest that experience can influence patients’ WTP, but primarily for technologies whose benefits are difficult to fully understand without firsthand experience. Second, we assessed the impact of patient status on the WTP for RM and PSM for pregnant women at risk of GHD. For the patient sample, data were drawn from the CV survey administered to the PREMOM II study participants at six weeks postpartum, while a slightly modified version of this survey was used for participants in the non-patient sample. Propensity score matching was employed to reduce potential differences between the two samples. For RM, results indicate that patients reported a higher WTP than non-patients. Subgroup analyses were performed to determine whether the observed difference was attributable to patient status alone (i.e. presence of the medical condition in patients) or also to prior treatment experience, revealing that it was primarily driven by the latter. In contrast, no significant difference in the WTP for PSM was observed between patients and non-patients, regardless of treatment experience. A possible explanation is that differences in WTP between patients and non-patients are mainly driven by firsthand experience rather than by patient status alone. As noted previously, experience appears to influence the WTP primarily for technologies whose benefits are difficult to understand without direct use, which does not apply to PSM. In sum, we can conclude that being a patient alone does not impact the WTP for a treatment. However, patients with prior treatment experience may report a higher WTP than non-patients, particularly for technologies such as RM, where the benefits are more difficult to imagine without firsthand experience. Besides GHD, this dissertation also focused on HF, a chronic condition affecting approximately 2% of the global population and associated with high rehospitalization rates and a reduced HRQoL. Effective long-term management is crucial for improving the outcomes of these patients. However, conventional in-person disease management programs are resource-intensive and can place a heavy burden on patients. RM has emerged as a promising alternative, and similar to RM for pregnant women at risk of GHD, RM for patients with HF may provide additional benefits beyond clinical outcomes. Despite the wide range of potential advantages, research on patients’ WTP for RM for HF is extremely limited, with only one study conducted over a decade ago. To update this evidence, we developed a CV survey that combined a payment card with an open-ended follow-up question. This survey was administered at study inclusion to the participants of the Digitally Expanded Heart Failure Pathway (DEFY-HF) study, a multicenter non-randomized controlled trial including patients with HF, with one group receiving RM and the other receiving conventional care. Results indicate that patients were willing to pay an average of € 36 per month for RM, although responses varied widely, with a small subset assigning a high value and the majority reporting lower amounts. To inform the future development of RM for patients with HF, we explored which factors are correlated with patients’ WTP for RM using two-part models. Our analysis reveals that patients who needed technological assistance but did not have access to support, reported a lower WTP compared with those without such barriers. This difference was primarily due to a reduced likelihood of indicating a WTP greater than zero. These findings underscore the need to address digital health inequities when introducing RM, for example through patient training programs and the establishment of supportive care networks. From a methodological perspective, this dissertation benefits from several strengths. The PREMOM II study was designed as an RCT, ensuring a high internal validity. Additionally, the cost-effectiveness analysis utilized patient-level data from the IMA, Belgium’s most comprehensive healthcare cost database, which captures both inpatient and outpatient expenditures. Nonetheless, several limitations should be acknowledged. First, participation in both PREMOM II and DEFY-HF was voluntary, leading to potential self-selection bias. While only about 4% of eligible participants declined participation in PREMOM II, approximately 80% refused in the DEFY-HF study, limiting its external validity and representativeness for the broader Flemish HF population. Second, response rates also varied across studies, with less than 30% in PREMOM II and around 80% in DEFY-HF. However, comparisons between respondents and non-respondents indicated that attrition did not threaten external validity, and balance checks confirmed that randomization was still valid in PREMOM II. Third, in the cost-effectiveness analysis, a repeated decision tree model was used to estimate maternal and infant costs and effects over time. However, results are subject to uncertainty in model parameters, which may affect the precision of the estimated cost-effectiveness. Fifth, as RM in the DEFY-HF study was not randomized, causal inferences cannot be drawn from its results. Finally, data limitations constrained certain analyses. While several mechanisms underlying the added value of midwife follow-up were examined, the results suggest that additional, unmeasured mechanisms contributing to this added value exist. Potential mechanisms explaining the positive effects of RM on PROMs could not be explored due to a lack of available data, and in the WTP comparison between patients and non-patients, unobserved factors such as baseline attitudes toward IT may have introduced residual bias despite propensity score matching. In DEFY-HF, over 20% of respondents had missing income data, preventing inclusion of this variable without substantial sample loss.