Dexmedetomidine versus standard care sedation with propofol or midazolam in intensive care: an economic evaluation

Intensive care unit (ICU) costs can take up to 20% of hospital budgets [1]. The daily costs of mechanically ventilated patients may be 20% to 44% higher than those of non-ventilated patients [2-5]. Therefore, shortening the time to extubation and the duration of mechanical ventilation (MV) are among key factors in reducing total ICU resource utilization and the respective ICU costs.

Mechanically ventilated ICU patients are commonly sedated to facilitate tolerance to artificial airway and other interventions. Dexmedetomidine, a highly selective alfa-2-agonist, was approved in Europe in 2011 for light to moderate ICU sedation. Recent studies suggest that dexmedetomidine may reduce the time to extubation compared with standard sedation [6-8]. However, reductions in MV or length of stay in the ICU do not necessarily always lead to cost savings [9-11]. Therefore, using a cost-minimization approach [12], we aimed to evaluate the net effect of dexmedetomidine on total ICU costs and to compare it to sedation with propofol or midazolam (or per cohort, a 1:1 mix of both, annually) in ICU patients requiring prolonged MV in Europe.

This economic analysis suggests that the higher acquisition costs of dexmedetomidine compared with standard sedation may be compensated for by reductions in other ICU costs. In this study, dexmedetomidine resulted in lower ICU costs compared with standard care, primarily by reducing the time to extubation. In economic evaluations of ICU care, however, it is important to consider the net impact on the total duration of ICU and hospital stay [26]. In our analysis, dexmedetomidine reduced the duration of total ICU stay compared with standard care, without prolonging post-ICU ward stay. The finding of reduced time to extubation is supported by the results of the sensitivity analyses of the original studies [8] as well as similar findings in previous studies [6,7,27].

The estimated unit costs and the assumption of less expensive care after MV and ICU discharge are the key drivers for the potential cost savings from dexmedetomidine use. The unit cost assumptions were built on the mean daily ICU costs obtained from 18 trial centers, representing six out of nine study countries, and 34% of all patients included in this economic analysis. However, the intention of this study was not to calculate the precise costs for the specific ICUs who were involved in the trials but rather to provide a more general picture of the potential cost implications in different European settings, where daily ICU costs may vary quite substantially by country and center. In fact, the mean daily ICU cost of the 18 study centers appears to be well in line with the available cost references in the literature, indicating an approximate ICU cost range of €700 to €3,000 per day across Europe [5,18-25], a range that was well covered in the respective sensitivity analysis.

When an economic assessment of ICU care is performed, a single mean daily ICU cost should not be used; instead, the declining cost of ICU days over time needs to be taken into consideration [9-11]. In the present analysis, this was done in two different ways: first, using three decreasing unit cost levels tied to the decrease in MV use over time (base case analysis), and second, on a day-by-day basis, directly based on the daily TISS points of each patient, thus reflecting marginal costs (sensitivity analysis) [9-11]. Both of these costing approaches were driven by the daily decline in variable costs toward discharge and indicated similar results in favor of dexmedetomidine. The impact of marginal variable costs was given further consideration in another sensitivity analysis assessing only the estimated declining direct variable daily costs. Also, this analysis supported the same conclusions.

In terms of methodology, we considered a direct cost comparison based on consumed resources (cost-minimization) to be appropriate as dexmedetomidine has been shown to be non-inferior to standard care as measured by the proportion of time patients are maintained at the targeted, light to moderate sedation levels (Richmond Agitation and Sedation Scale: 0 to −3), without a difference in mortality [8]. The chosen approach should be considered to be conservative since it focuses purely on resource consumption and costs. We assumed that dexmedetomidine’s effectiveness, for which there is no simple definition, is as good as that of the standard sedatives. Any quality-of-life improvement that could potentially result from earlier extubation, patient’s improved ability to communicate with the nursing staff [6,8], or reduced neurocognitive adverse events [8] or delirium [6,28,29] were thus overlooked in this analysis.

An alternative methodological approach focusing on net costs could have been a cost-benefit analysis, in which not only the sedative consumption but also the established differences in the effectiveness of the treatment options are expressed in monetary terms and compared against each other resulting in ‘net monetary impact’ [12]. In this approach, resource utilization parameters, such as differences in time to extubation, duration of MV, and ICU length of stay, would have been considered to be constituents of benefit, to which eventually a monetary value would have been assigned. The result of such an analysis, ‘the net monetary benefit of dexmedetomidine’, would have been equal to the difference in the total ICU costs demonstrated in the present analysis, and thus the choice of analysis would not have impacted the conclusion.

We did not include any specific extra costs for adverse events. However, in a previous cost-minimization analysis of dexmedetomidine on the basis of the SEDCOM (Safety and Efficacy of Dexmedetomidine Compared to Midazolam) trial [6], such an approach was taken. In that analysis, even when the costs due to differences in adverse events were considered, dexmedetomidine was still associated with median total ICU cost savings of USD $9,679 compared with midazolam [30].

Since the scope of this analysis was economic, there are some distinct differences in the definitions and handling of the length of stay parameters, compared with the main clinical analysis [8]. In the present analysis, the intention was to reflect the total ICU costs per treatment group as realistically as possible, considering also the potential administrative delays, availability of a suitable ward bed, and so on, which finally determined when a patient actually got discharged. Therefore, the cost calculation was based on the observed actual length of ICU stay rather than the point ‘when medically fit for ICU discharge’ [8]. Also, for non-survivors, the length of ICU stay and the other relevant time points not yet reached were censored at time of death, or when moving to palliative care, as the accumulation of ICU costs ceased. For most patients who discontinued study medication prematurely, the actual times of all of the three relevant events had been captured as intended. In the relatively small number of cases (22 of a total of 990 patients) in which the subsequent data were lost to follow-up, the missing end points were worst ranked to the next actually recorded time point or (if none was available) to 45 days. The same imputation principle was applied on the two study withdrawals that took place early on, before MV or the ICU period had been completed. Thus, no significant bias should be expected from the way missing data were handled in this economic analysis. Nevertheless, since even non-significant imbalances between the treatment groups in censoring or imputation (or both) might introduce some degree of bias, we performed a sensitivity analysis including only those patients for whom observed data were available from all of the three ICU periods. Also, this analysis supported similar conclusions as the base case.

As a relatively large share of the total ICU costs may be of a fixed or at least semi-fixed nature [17,18], it may not be straightforward to realize the full savings potential indicated by our analysis, at least in the short term. However, even when only the direct variable ICU costs were included, the analysis still indicated savings with dexmedetomidine compared with standard care. Furthermore, from another perspective, even with the same staff costs, dexmedetomidine may potentially have implications for the cost-efficiency of the ICU, through freeing up beds faster. For example, if the ICU stay was reduced from 8.2 to 6.9 days (30 hours) as in the present case, then the same ICU staff could manage 19 additional patients per 100 patients sedated with dexmedetomidine, compared with standard sedatives. In ICU units with shorter underlying mean lengths of stay, the potential to increase the number of treated patients is even larger. Whether such increased patient turnover (and potentially a subsequently increased third-payer compensation to the hospital) is realistic depends on the organization and capacity of the referring wards and on the subsequent step-down units’ ability to admit the patients earlier.

The conclusion on the potential for cost savings with dexmedetomidine vs. standard care was robust in all sensitivity analyses performed. These included e.g. testing the impact of several variations of the ICU daily unit costs intended to represent different types of ICUs and economies (or countries), with higher or lower underlying costs vs. our base case, two scenarios considering the direct variable costs only, as well as a complementary cost analysis based on a well-established ICU resource use measure, the TISS points, collected daily in the PRODEX and MIDEX trials. In the analysis in which the share of direct variable costs was assumed to be 25% of the total ICU costs (which decline over time), the marginal direct cost estimate for the last ICU days was €347.50 (25% of €1,390 total cost of the Off MV days). This is in line with the $397 estimate reported by Kahn et al. [9] for the last ICU day and the $400 to $450 reported by Taheri et al. [11] for the marginal direct cost of the last hospital day. Even this conservative variable cost analysis still indicates that dexmedetomidine has the potential for leading to lower costs.

There was a statistically significant reduction in the duration of MV as well as in the cumulative sum of TISS points in comparison with midazolam or pooled standard care but not in comparison with propofol. Nevertheless, the likelihood of dexmedetomidine to result in lower total ICU costs exceeded 90% compared with pooled standard care and midazolam and was still relatively high (72%) even when compared with propofol. The conclusion of our analysis is also supported by other similar findings regarding lower resource use and ICU costs with dexmedetomidine versus both midazolam [30,31] and propofol [32].

These results reflect a rather heterogenic, general ICU patient population requiring prolonged sedation of over 24 hours. However, some relevant groups often requiring prolonged sedation, such as patients with acute severe neurological disorders, were excluded from the trials [8]. Furthermore, the reported cost savings were obtained within a rigorously controlled clinical trial implementing frequent sedation level control based on a validated sedation scale, daily sedation interruption, and spontaneous breathing trials. These interventions, recommended by recent guidelines, may help avoid unnecessarily deep sedation as such and are associated with improved outcomes, including reduced MV and ICU stay [33-36]. In practice, however, adherence to these best practices varies largely between hospitals [37-40]. Therefore, the cost consequences and impact of dexmedetomidine on ICU cost-efficiency in different patient groups in a real-life setting should be further assessed.

This analysis demonstrates that, when targeting light to moderate sedation, dexmedetomidine represents an economically sound option that may provide ICU cost savings compared with the standard sedatives, primarily through reducing the time to extubation. If this can be translated to a shorter ICU stay, more patients can be managed with the same staff and fixed costs.

For this specific secondary manuscript, no grants or payments have been made. The MIDEX and PRODEX studies were sponsored by Orion Corporation, Orion Pharma, through research contracts negotiated directly with the participating hospitals and through grants allocated to the specific intensive care departments. No author received any personal financial gain (for details, see [8]). The authors thank the ‘Dexmedetomidine for Long-term Sedation Investigators’ [8] for participating in the data collection.

Heidi Turunen, Email: heidi.turunen@orionpharma.com.

Stephan M Jakob, Email: stephan.jakob@insel.ch.

Esko Ruokonen, Email: esko.ruokonen@kuh.fi.

Kirsi-Maija Kaukonen, Email: maija.kaukonen@hus.fi.

Toni Sarapohja, Email: toni.sarapohja@orionpharma.com.

Marjo Apajasalo, Email: marjo.apajasalo@orionpharma.com.

Jukka Takala, Email: jukka.takala@insel.ch.