In the current analysis, it was examined in a more granular fashion where and when the cost savings in caring for HF from remote monitoring might be realized. What we found is that there appears. to be no cost savings early on (at 12 months) for HF-specific or all-cause treatment in either high- or lower-intensity care settings (eg, hospital, home, and physician office) What likely strengthens (further supports) this finding of no real cost savings with remote care is that the costs for the disease management program in the largest trial Galbreath were not included in the analysis, which would increase the costs for this intervention.
Much of the remote monitoring examined in this review was relatively simple and similar in nature, eg, telemonitoring, patient measurements of weight, phone contact. All related to periodic contact with the patient via personal or impersonal means were not terribly sophisticated. Since patients were intervened upon only when overt symptoms appeared, it may have been too late to realize any costs savings. There are more advanced implantable remote monitoring systems (non-CHF and HF), which monitor and alert patients (intracardiac and pulmonary artery pressures) prior to symptoms developing and have demonstrated an ability to intervene earlier.[48, 49] These types of remote systems may aid in earlier intervention (such as the titration of medications based on physiologic data), reduce costs, and improve patient satisfaction.
Differences With Other Published Reports
We have demonstrated dissimilar findings to Klersy in that remote monitoring does not reduce the overall costs for inpatient care related to HF. Dissimilar with Klersy, we did not find that remote monitoring reduces the overall costs for inpatient care related to all-cause readmissions. In the Klersy article it was found, based on modeling at 1 year, that costs were lower (ranging from ∈300 to ∈1000 [$395 to $1315 US]), favoring remote monitoring. What we found, unlike findings from Klersy, is that costs may be shifted to other care settings, resulting in no actual cost savings to the system. Further, in another prior analysis, Seto examined published trials and also found cost savings but did not perform meta-analyses. In addition, it only examined Scopus and PubMed databases, which were searched on April 10, 2007. This current review examines additional sources of information that were not examined in Seto (ie, specific journals related to CHF as per the search methodology outlined above) and performs a meta-analysis of like trials, which Seto did not perform. Lastly, this review examined RCTs only, where direct costs for care were evaluated, whereas Seto examined randomized and nonrandomized trials and direct and indirect costs (such as decreased work productivity, absenteeism, unpaid care, lost leisure time). As it relates to prior analyses, a recent Cochrane systematic review and meta-analysis published by Inglis and colleagues found that the use of remote monitoring (examining structured telephone support and telemonitoring programs only) had a positive effect on reducing the risk of all-cause mortality and CHF-related hospitalizations in patients with CHF and that it improved the quality of life, reduced costs, and evidence-based prescribing. However, with the Inglis review, two studies were included that we did not: Barth and Wakefield and associates. We found that the study by Barth and colleagues was not an RCT and that the Wakefield and coworker study did not examine costs. Further, we examined several other studies that were not examined by Inglis and colleagues, even though they were RCTs. These studies included Benatar (nurse telemanagement), Dunagan (telephonic monitoring), Gregory (nurse manager), Jerant (telecare system), and Krumholz (nurse telemanagement)..
While we did not examine other study outcomes, remote monitoring appears to improve the quality of life of people with HF (eg, as defined by improvement in New York Heart Association functional class,[10, 12, 13, 16, 38] reduce hospitalizations (for HF, reduce mortality,[12, 17] and improve evidence-based prescribing.[12, 21] However, there were also other studies in which there were no differences seen in the outcomes of: cardiovascular mortality, all-cause mortality, functional status, treatment satisfaction, depressive symptoms, or quality of life (which was a combination of functional status, depressive symptoms, and treatment satisfaction).[18, 19] Lastly, there were no studies identified that showed patients were statistically worse off in any of the above non-cost outcomes under remote care. Thus, outcomes such as quality of life (if costs are equal in all care settings) may be more relevant to decision makers (than overall costs) in deploying resources in order to care for these types of patients.
One of the issues that is of importance when examining HF is its presence in other comorbid conditions such as chronic obstructive pulmonary disease, renal impairment/disease, cardiac dysrhythmias, ischemic heart disease, diabetes, hypertension, cardiac valve disorders, hyperlipidemia, and other circulatory diseases. Due to costs associated with these conditions, an analysis that incorporates all-cause care, in all care settings (such as performed here and as others have done) is likely appropriate and useful to understand.
As noted above, we did not include RCTs from outside of the United States. The reason they were excluded had to do with their different (from the United States) cost structures and resource use patterns. However, even if these trials were included in the analyses above, there still remained no statistical difference in overall costs, no matter the setting or patient's disease state (ie, HF only or HF plus comorbid conditions) (data on file).
Limitations of this analysis include the assumption that costs as calculated by cost accounting systems are the same as costs as reflected by Medicare reimbursement and costs as converted via Medicare C/C ratios. Further, based on American Hospital Association data, the Medicare reimbursement amounts provided for care have been averaging/tracking very closely to the estimated costs for care. Most US researchers use Medicare reimbursement and C/C ratios as means of calculating costs. Therefore, we believe that the overall estimates that were derived from these costing methods are a reasonable way for calculating costs. While there were 10 studies included in this analysis, with costs examined at various time points, longer-term costs should (>1 year) should also be undertaken and larger numbers of patients should also be evaluated. Lastly, as it relates to missing data, it was assumed that data were missing at random and thus no further analysis was undertaken (imputing missing data with replacement values). As shown in Figure 2, there were trials where data were missing or it was not clear why the data were missing. We have therefore assumed that the available data for evaluation were unbiased.