Management of perioperative iron deficiency anemia as part of patient blood management in France: A budget impact model-based analysis based on real world data

1 INTRODUCTION

Patient Blood Management (PBM) is an evidence-based, patient-focused, multidisciplinary approach to improve patient outcomes.¹ This strategy allows optimal conservation of the patient's own blood to improve the quality of care in terms of morbidity and mortality.^1-3 PBM lies on three pillars^{1, 2, 4}: (i) detection and management of anemia and iron deficiency (ID), (ii) minimization of blood loss and (iii) optimization of patient's tolerance to anemia. The first pillar is of crucial importance, preoperative anemia being an independent predictive factor for postoperative mortality and morbidity.^{5, 6} It is associated with poorer outcomes, increased length of hospital stay and subsequent higher hospital costs. The presence of iron deficiency and/or of anemia, is also associated with increased mortality, increased incidence of serious adverse events, major adverse cardiac events, transfusion, and prolonged hospital stay.^{3, 7-11}

Despite the evidence and many national and international recommendations,^{1, 12} it is recognized that implementation of PBM could be wider.¹³ One of the barriers to PBM implementation is the estimated cost associated with preoperative anemia treatment and the belief that PBM is not cost-effective.¹⁴ Cost-effectiveness of PBM programs have been reported from local or regional experiments. In the state of Western Australia (>600,000 patients) implementation of PBM achieved product cost savings of AU$18 M with a 41% decrease in the number of blood units transfused over 5 years.¹⁵ Implementation of Pillar 3 (i.e., monitoring and feedback program as part of the implementation of restrictive transfusion triggers) in a PBM program at University Hospital of Zürich, Switzerland, allowed a total saving of €11.1 M over 4 years (2014–2017). The introduction of this program reduced allogeneic blood product transfusions by 35%, from 825 units per 1000 hospital discharges in 2012 to 536 units in 2017.¹⁶ To date, no study in France had evaluated the potential economic impact of implementing a national PBM program. Hence, this budget impact analysis aimed to model the economic impact of implementing preoperative hematopoiesis optimization in France, from both National Health Insurance and hospital perspectives, using real-life data collected in French hospitals.

2 METHODS

3 RESULTS

A total of 980,125 patients from the “Scan Santé” database were modeled for all four surgical areas, including 339,117 patients in orthopedic surgery, 33,087 patients in cardiac & cardiovascular surgery, 209,567 patients in vascular and thoracic surgery and 398,354 patients in urologic and visceral surgery. Table 1 displays the main parameters used for modeling both scenarios. All the parameters, including population size and data sources for each therapeutic area, are shown in Table S1 in the Appendix.

Based on real-life data, the proportion of patients preoperatively treated with ferric carboxymaltose in the scenario without PBM was 10% in orthopedic surgery, 3% in urologic & visceral surgery, and 0% in cardiac & cardiovascular and vascular & thoracic surgeries. In the PBM scenario, we considered that all patients diagnosed with ID (i.e., 75% in orthopedic surgery, cardiac & cardiovascular surgery, urologic & visceral surgery according to Munoz et al (in their cohort of 2884 patients, 2163 (75%) had some ID)¹⁸ and 21% in vascular & thoracic surgery¹⁰) were preoperatively treated with ferric carboxymaltose.

Regarding the preoperative use of ESA in orthopedic surgery, hospital data showed that 5% of patients received it in current practice. In the scenario with PBM, Rineau et al. data were used (i.e., 10%) (Table 1).²²

Average LOS gathered in the nationwide database and used for the current practice scenario ranged from 4.8 days (vascular & thoracic surgery) to 12.8 days (cardiac & cardiovascular surgery). According to the literature,² the expected decrease in the average LOS due to PBM varied from 7% for orthopedic surgery to 44% for urologic and visceral surgery (Table 1).

Proportions of transfused patients for the current practice scenario varied between 7% (urologic & visceral surgery) and 48% (cardiac & cardiovascular surgery) and the mean number of PRBC units administered per transfused patient ranged from 2.4 (cardiac & cardiovascular surgery) to 2.9 (urologic & visceral surgery). The decrease in transfusion rate in the PBM scenario was based on literature data^{2, 19, 20} and varied between 8% and 60%, depending on the therapeutic area. The mean decrease in number of transfused PRBC units per patient due to PBM oscillated between −0.2 and 0.9 (Table 1).

For the 980,125 patients, total annual savings from National Health Insurance perspective would be €1079 M (Table 2, Figure 1). Depending on the surgical area, the estimated savings for National Health Insurance ranged from €48 M (cardiac & cardiovascular surgery) to €814 M (urologic & visceral surgery). Hospital stays were reduced by 1,472,153 days, which corresponds to a 25% decrease in all surgical areas combined (orthopedic surgery, −7%; urologic & visceral surgery, −44%).

From a hospital perspective, these savings would be €1018 M for the study population and would range from €45 M (orthopedic surgery) to €733 M (urologic & visceral surgery) (Table 2, Figure 1). The 52% decrease in the number of PRBC units transfused after implementation of the PBM program resulted in savings of €37 M compared with the scenario without PBM. Based on the model, a PBM program implementation would spare 181,451 units of transfused PRBC units, with 74,674 patients transfused with a total of 166,989 PRBC units compared to 127,757 patients transfused with a total of 348,440 (Figure 1).

The deterministic sensitivity analysis showed that savings are generated for both perspectives for most of the parameters tested. The length of stay after the PBM implementation and the average cost per hospitalization day are the most important driving factors for the budget impact model. A ± 20% variation of these parameters in the National Health Insurance perspective would correspond to a budget impact of −€824 M € and −€1334 M (Table 3). In the hospital perspective, the budget impact after this variation would be −€750 M and −€1285 M (Table 4). The other parameters had a lower impact on the model results (Tables 3 and 4). To complete this sensitivity analysis, the combined effect of a ± 20% variation of all three parameters together after implementation of PBM (i.e., average LOS, proportion of patients transfused, and number of RBC units transfused) was performed and leads to a saving of €309 M from the health insurance perspective and €122 M from the hospital perspective. Compared with the main analysis, the implementation of PBM would spare 115,068 RBC units and avoid transfusion for 38,148 patients (vs. 181,451 RBC units spared and transfusion avoided for 53,053 patients).

4 DISCUSSION

To our knowledge, this is the first published analysis estimating the budget impact of implementing hematopoiesis optimization as part of a PBM program in France. According to our model, and for the analyzed population of 980,125 patients, 181,451 RBC units would be spared (7.8% of total RBC units transfused in France in 2021²⁴) with an estimated annual savings of €1079 M from the National Health Insurance perspective and €1018 M from the hospital perspective. This analysis plaids for PBM, as its implementation would allow important savings for both National Health Insurance and hospital, and it would help reducing national transfusion dependency.^{13, 25}

From the National Health Insurance perspective, the savings are mostly based on the reduction of the duration of hospital stay related to the implementation of the PBM program. The hospital stay was reduced by 1,472,153 days, corresponding to a 25% decrease in all surgical areas combined for the entire population. This leads to a reduced hospital costs and offsets additional costs associated with screening and treatment of anemia/ID. This analysis did not consider the possible reduced rate of hospital readmission (for a surgical complication) observed in anemic patients treated with iron compared to placebo and all the other related costs.^{15, 26}

From the hospital perspective, the cost-savings in all surgical areas are 4 times higher than the cost increased from implementing hematopoiesis optimization. As in the case of the National Health Insurance perspective, savings from the hospital perspective originated from the decrease in the average LOS, corresponding to savings of €1338 M, but also to the 52% reduction in transfusions (€37 M). This could also be even more important, as the activity-based costs of transfusion could be estimated 3–5 times more than the actual product cost.²⁷

It is important to highlight that we modeled the cost of implementing preoperative anemia treatment, but not the less costly cost of implementing the second and third pillars of PBM. Nevertheless, our results are in accordance with other studies reporting budget benefits related to the implementation of PBM programs in other countries. The savings induced by the decreased number of transfusions were evaluated, in these studies, to be US$2 M (approximately €1.8 M) for 101,794 patients, or also AUS$18 M (approximately €12 M) for 605,046 patients.^{15, 16} In our study with 980,125 patients the generated saving (by decreased blood use) is €37 M is very similar to these reports.

Although using only a selection of hospitals may have led to a partial representation of the practices observed in the studied surgeries, the use of real-life data from 10 French public/private hospitals that had not established a PBM program had the advantage of having transfusion rates comparable to those observed in current practice in most French hospitals, as opposed to others available in medico-administrative databases which tend to be underestimated, notably because of under-coding post-operative transfusions. However, it is possible that transfusion rates are lower in current practice in France.

One of the limitations of the observational study is that the proportion of ID patients is underestimated compared to what is observed in the literature.^{18, 28} This may be since diagnosis of iron deficiency was not routinely done in the study hospitals. It is likely that the proportion of ID patients diagnosed would be higher with the implementation of a PBM program protocolizing a systematic diagnosis of this condition. To overcome this underestimation, data from the literature were used to estimate the proportion of ID patients in France, considering a very conservative hypothesis (i.e., 75% of ID),¹⁸ we could have based our hypothesis on more recent data.²⁸ In addition, we do not know whether (and if, to what degree) a change in the ID-prevalence and patient treatment ratio would be associated with a change in savings. We were neither able to differentiate the savings according to the cause of anemia (pure ID or functional ID).

The impact of PBM is likely to be influenced by patients' comorbidities. French data lacking, effects of a PBM program according to DRG severity levels could not be modeled in the current analysis.

In this analysis, the budget impact of the PBM implementation was only modeled according to the first pillar of this program, which consists of hematopoiesis optimization through the management of ID and anemia. Clinical data not being available, the impact of the other two pillars of PBM programs could not be modeled, but their costs are very low.

The main limitation of our analysis is the uncertain impact of PBM and optimizing preoperative hematopoiesis (which specifically includes the treatment of ID) on the average LOS, the main source of savings observed in the model. Indeed, it is likely that other factors influence the decrease of LOS, particularly with the encouraged development of day-case surgery. Similarly, LOS and transfusion rates mainly depend on the medical practice of each institution. Based on the medico-administrative database PMSI (Programme de médicalisation des systèmes d'information), the average LOS evolution over the 2018–2019 period (2020–2021 not used as reference due to potential impact of the COVID-19) shows a decrease in the mean LOS of 9.3% in orthopedic surgery for the DRG 08C48 (Hip replacements for conditions other than recent trauma) and 9.2% in digestive surgery for the DRG 06C04 (Major small bowel and colon procedures). For vascular and cardiac procedures, a slight increase in LOS was observed over the same period (between 0.1% and 1%). The potential decrease in the mean LOS used in the model, derived from the literature, could therefore be related not only to the implementation of hematopoiesis optimization in the framework of PBM, but also to other parameters. The benefit of treating anemia according to the different type of anemia is not known and have not been considered. Nevertheless, the deterministic sensitivity analysis showed that the PBM generates savings for both perspectives. With the lower value tested for the mean LOS and mean cost for hospitalization in the PBM scenario in all surgical areas combined, savings are still generated for both National Health Insurance (€824 M) and hospitals (€750 M). It is also worth mentioning that increasing the preoperative dose of IV iron generates additional costs for hospitals.

We used published data to estimate the reduction in transfusion rates after PBM implementation^{2, 19, 20} and to determine transfusion rates for the “current practice” scenario (in hospitals that did not have a well-established PBM protocol, but were motivated to implement it).¹⁷ It is possible that transfusion practices are more restrictive and/or that PBM implementation is less effective in other institutions in France. Nevertheless, the results of the sensitivity analysis show that varying the proportion of patients transfused does not change the budgetary impact. Data show that preoperative anemia and iron deficiency are associated with poorer outcomes, increased hospital length of stay, mortality, incidence of serious adverse events, major adverse cardiac events, transfusion, and prolonged hospital stay, whereas PBM programs are associated with a reduction in adverse outcomes. However, adverse events were not considered in the analysis, because French baseline data were not attainable. In addition, there is considerable uncertainty about the proportion of patients requiring medical transportation for hospitalization and preoperative iron injections. These transportation costs were not included in this analysis.

5 CONCLUSION

The results of this budget impact modeling confirm that the implementation of a PBM program, and especially the hematopoiesis optimization, could lead to significant savings for both the National Health Insurance and hospitals.

FUNDING INFORMATION

This work was founded by Vifor France.

CONFLICT OF INTEREST STATEMENT

Sigismond Lasocki received consulting fees, educational fees from VIFOR PHARMA, educational fees from MASIMO, consulting fees, educational fees from PFIZER. David Delahaye received during the past 3 years, lecture, and consulting fees from VIFOR PHARMA and PFIZER. Axel Hofmann received during the past 3-year honoraria or consulting fees and travel support from BMS/Celgene, PBMe SOLUTIONS, VIFOR PHARMA, WERFEN. He also received travel support from IFPBM. David Fuks, Pierre-Henri Savoie, Claude Dussart and Pascal Paubel declare no conflicts of interests.