Currently, several anatomical approaches and intervention sites can be used to perform transcatheter aortic valve implantations (TAVIs), often with no clinical indications for choosing one or another. While these choices can have an impact on resource consumption, no costing study is available in the European context to provide information on resource use and assist decision-making.
To provide comparative data on the cost of the TAVI procedure, depending on anatomical approach and intervention site used, from a hospital perspective, and to analyze factors associated with cost of hospital stay.
Multicentre national registry data were collected in 16 centres between January and October 2009. For 287 patients, a descriptive costing study and a multivariable analysis of hospital stay cost were performed.
The mean cost of the TAVI procedure was € 22,876 and the mean initial hospital stay cost was € 35,164. The procedure cost, excluding valve cost, did not differ between anatomical approaches and was highest in the hybrid room and lowest in the catheterization laboratory. Factors associated with higher hospital stay cost were transapical approach, Society of Thoracic Surgeons score > 10%, warfarin use at inclusion, complications during procedure and pacemaker implantation following valve implantation.
If clinical considerations do not interfere, hospital staff may find it economically favorable to opt for the catheterization laboratory and against the hybrid room. The mean hospital stay cost is higher than the tariff paid in 2011, a difference that has grown since the change in tariff in 2012, representing an economic disincentive for the uptake of TAVI in France.
Il n’existe souvent aucune indication clinique permettant de faire le choix entre les différents types de salle et de voies d’abord disponibles pour effectuer une implantation de valve aortique par voie percutanée (TAVI). Bien que ces choix aient un impact en termes de consommation de ressources, aucune étude ne permet de le quantifier et d’apporter une aide à la décision dans le contexte européen.
L’objectif était de fournir des données comparatives de coût de la procédure TAVI du point de vue de l’hôpital en fonction de la voie d’abord et du type de salle utilisés ainsi que de déterminer les facteurs associés au coût de séjour.
Les données utilisées proviennent d’un registre national multicentrique correspondant à 16 centres et à la période allant de janvier à octobre 2009. Une étude descriptive de coût ainsi qu’une analyse multivariée du coût de séjour ont été effectuées sur une population de 287 patients.
Les coûts moyens de la procédure TAVI et de l’hospitalisation initiale étaient respectivement de 22 876 € et 35 164 €. Les coûts de la procédure ne différaient pas en fonction de la voie d’abord et étaient les plus élevés en salle hybride et les moins élevés en salle de cardiologie interventionnelle. Les facteurs associés à une augmentation de coût de séjour sont la voie transapicale, un score STS supérieur à 10 %, l’utilisation d’antivitamines K à l’inclusion, la présence de complications lors de la procédure et la pose d’un pacemaker à la suite de l’intervention.
Si les conditions cliniques du patient le permettent, le personnel hospitalier pourrait être amené à trouver l’utilisation de la salle de cardiologie interventionnelle économiquement plus avantageuse que celle de la salle hybride. Le coût moyen de l’hospitalisation est supérieur au tarif remboursé en 2011. Cet écart s’est amplifié depuis le changement de règle tarifaire survenue en août 2012 représentant un frein au développement de l’utilisation de TAVI en France.
In 2002, the first transcatheter aortic valve implantation (TAVI) was performed in a patient with severe aortic stenosis and cardiogenic shock, as a therapeutic alternative to cardiac surgery for patients at high surgical risk. Since then, over 50,000 patients are estimated to have been treated by TAVI worldwide .
The safety and efficacy of TAVI have been repeatedly demonstrated and 1-year as well as 2-year survival rates were recently reported to be similar to those for patients undergoing surgical valve replacement . The quality of life of patients has been shown to improve after TAVI and studies based on a randomized trial in the USA using only the Edwards SAPIEN valve suggest that the cost-effectiveness and cost-utility of TAVI are favourable in inoperable patients . However, no costing study is available in the European context to provide information on resource utilization and assist decision-making at the national and hospital level, especially in countries where payment is based on activity. Indeed, physicians and hospital managers face the choice of several TAVI devices, anatomical approach types (arterial or transapical) and intervention sites (catheterization laboratory, operating room or hybrid room) but lack comparative cost data in order to make informed decisions when the choice is not clinically driven.
The purpose of this study is therefore to provide these comparative data for the cost of the TAVI procedure and the hospital stay excluding the procedure cost, from a hospital perspective, and to analyse factors associated with higher cost.
Study design and population
Our study is based on the FRANCE registry, which has been described in detail in previous publications . In short, it involved symptomatic adults requiring TAVI due to severe degenerative valvular aortic stenosis and who were very-high-risk patients for aortic valve replacement at the time of enrolment. The registry is exhaustive because TAVI may only be performed in France in authorized participating centres.
The FRANCE registry case report form (CRF) was designed to collect detailed information on the resource requirements for the TAVI procedure and initial hospital stay, in addition to the collection of clinical and quality of life data. The subsequent FRANCE2 registry currently collects long-term data on follow-up and adverse events . The present study presents the findings of the FRANCE registry, allowing sound cost analysis; it reports on a total of 295 patients included in 16 French centres between January and October 2009, of whom 287 patients underwent TAVI implantation and were included. For the costing study, 11 patients were excluded due to missing data (anatomical approach type, device type, length of stay). For the analysis of hospital stay cost predictors, 31 patients were excluded because death occurred during the hospital stay and nine patients were excluded due to missing data (anatomical approach type, device type, length of stay).
Devices, procedure and intervention sites
Patients received one of two models of prosthesis: the Edwards SAPIEN valve (Edwards Lifesciences, Irvine, CA, USA) or the CoreValve (Medtronic, Minneapolis, MN, USA). The Edwards SAPIEN valve (the SAPIEN XT valve was not available at the time of the study) is a bioprosthetic valve made of bovine pericardial tissue mounted into a balloon-expandable stainless steel open-cell stent. The valve can be implanted using the transfemoral (TF) or transapical (TA) approach, depending on the peripheral access. The CoreValve consists of a trileaflet bioprosthetic porcine pericardial tissue valve mounted into a self-expandable nitinol frame. It is generally implanted using the TF arterial approach or, less frequently at the time of data collection, the subclavian approach (SC). There were no prespecified recommendations regarding the TF, SC or TA anatomical approaches or for intervention sites used to implant either valve, which included the catheterization laboratory, the operating room and the hybrid room (a multiusage site equipped with both catheterization and anaesthesia devices).
After informed consent, data were recorded using a standardized e-CRF and sent to a central database via the internet. Recorded data included demographic and clinical patient data, data on the devices and other materials used for the TAVI procedure, information about which anatomical approach and intervention site had been used, the duration and staffing required for the procedure, the length of hospital stay and the homogenous patient group code ( groupe homogène de malades ; GHM), which is the accounting system code used for depicting the level of hospital resource consumption for a patient stay.
A hospital perspective was taken for costing. Unit costs have been applied to the materials and staffing resources used in order to estimate the cost of the TAVI procedure, which also included overhead costs depending on the intervention site. As these costs are not available on a national basis, Paris hospital consortium average unit costs were used. The cost of the hospital stay was estimated based on the cost of the procedure and on direct and overhead cost items related to the patient’s GHM as identified in the national cost scale for the medicine, surgery and obstetrics sector (ENC MCO, which features a detailed accounting system that evaluates the total cost of each hospital stay). As only GHM data for patients in hospitals with accounting systems were available (47% of the study population), we calculated the mean daily hospital costs excluding procedure cost for these patients. To determine the cost of each hospital stay, this daily amount was then multiplied by the patient length of stay. Subgroup analysis distinguished patients by arterial (TF and SC) or transapical (TA) approaches and by intervention site. Costs are expressed in 2009 Euros (€) and are displayed with and without the valve price included.
Factors potentially associated with hospital stay cost
What was of interest was understanding the factors associated with higher hospital stay cost. As the cost of the valve represented a high fixed-cost component (close to € 20,000 which is over 62% of the tariff offer to hospitals for TAVI hospital stays in 2011), it was not integrated and the outcome studied was the hospital stay cost on top of the cost of the valve.
Predictors: factors potentially associated with higher hospital stay cost
Variables analysed include sociodemographic and clinical patient characteristics and treatment at inclusion, as described previously . Variables related to the procedure included the type of anaesthesia (general/local), device type (Edwards/Medtronic), anatomical approach chosen (arterial/TA), intervention site (catheterization laboratory/operating room/hybrid room) and occurrence of complications during the procedure (yes/no). As clinical scores assessing overall severity seemed likely to be overall cost predictors, we constructed our analysis around two mortality risk scores widely used in TAVI: the Society of Thoracic Surgeons (STS) score and the logistic Euroscore (LES). An STS score > 10% and an LES > 20% are commonly used in the literature to define high-risk patients in valvular surgery so we chose these values as thresholds for categorical analysis .
Because of the skewed distribution of the cost, we used a gamma distribution with a log link in a generalized linear model. As the data have a hierarchical structure, the unconditional means model was first fitted to test for centre effects, which were not detected.
Because classifications of patients in high-risk groups using the STS score and the LES are highly correlated, two different models were then tested with gamma regression: the STS score and all other variables except those used for its calculation; and the LES and all other variables except those used for its calculation. Variables were selected using the backward elimination method. Gamma coefficients generated by the regression models were exponentiated to retransform them into rate ratio estimates.
All statistical tests were two-sided. Statistical significance was accepted at P < 0.05. All data were analysed using SAS software version 9.3 (SAS Institute Inc., Cary, NC, USA). Results are expressed as means ± standard deviations while dichotomous variables are presented as frequencies and percentages.
A one-way sensitivity analysis was performed on the results of the costing study to estimate the effect of potential variation in main variables. Hypotheses tested included a decrease in the valve price that may follow an increase in uptake of the TAVI, a decrease in the duration of the procedure as a result of a learning effect after the diffusion of TAVI and, similarly, a decrease in the length of hospital stay (−10% and −20%, respectively). For the study of cost predictors, the sensitivity analysis tested the effect of including the patients who died during the hospital stay. A model was also performed that excluded the severity scores LES and STS, replacing them with variables available in the e-CRF (described elsewhere ).
Population baseline characteristics
Detailed baseline characteristics of the FRANCE registry population have been described elsewhere . In short, patients undergoing TAVI by the TA approach did not differ significantly from those undergoing the arterial approach in terms of age, sex, hypertension, diabetes, bypass surgery, peripheral artery surgery, New York Heart Association class, creatinine, warfarin treatment, LES, STS score and ejection fraction. However, they exhibited an aortic abdominal aneurysm more frequently ( Table 1 ).
|Approach||Transapical ( n = 78)||Arterial ( n = 209)||P||Total ( n = 287)|
|Age (years)||82.18 ± 7.1||82.24 ± 7.12||0.959||82.22 ± 7.1|
|Female||29 (37.2)||103 (49.3)||0.067||132 (46)|
|Hypertension||58 (74.4)||146 (70.2)||0.488||204 (71.3)|
|Diabetes||23 (29.5)||60 (28.8)||0.915||83 (29)|
|Previous bypass surgery||24 (30.8)||50 (24)||0.247||74 (25.9)|
|Previous PAD surgery||9 (11.5)||13 (6.3)||0.135||22 (7.7)|
|Aortic abdominal aneurysm||8 (10.3)||8 (3.8)||0.045||16 (5.6)|
|NYHA III or more||60 (76.9)||155 (75.2)||0.768||215 (75.7)|
|Creatinine (μmol/L)||139 ± 95||117 ± 53||0.145||123 ± 68|
|Warfarin treatment||26 (33.3)||59 (28.4)||0.413||85 (29.7)|
|Logistic Euroscore a > 20%||57 (76)||131 (66.2)||0.117||188 (68.9)|
|STS score b > 10%||37 (50)||101 (51.5)||0.822||138 (51.1)|
|Ejection fraction (%)||54 ± 12||53 ± 13||0.506||53 ± 13|
a The Logistic European System for Cardiac Operative Risk Evaluation (LES) score, which measures patient risk at the time of cardiovascular surgery, is calculated by logistic regression; scores range from 0 to 100%, with higher scores indicating greater risk; an LES of > 20% indicates very high surgical risk.
The arterial approach was performed more often than the TA approach (73% vs. 27%) and an Edwards valve was more frequently used than a CoreValve (66% vs. 34%). The combination of an Edwards valve and the arterial approach accounted for 38% of patients. The catheterization laboratory was the most frequently used intervention site (67%), regardless of the valve and approach type. The operating room was most frequently used to perform the TA approach, while the hybrid room was predominantly chosen to carry out TAVI using an Edwards valve by arterial approach.
Staffing and duration of the procedure
The TAVI procedure required, on average, 10 hospital staff. The duration, with a mean of 182 ± 85 min (median, 167 min), decreased during the study period of 10 months from a median of 177 min to a median of 150 min. The mean duration did not differ significantly when the arterial approach was chosen (185 ± 87 min) compared with the TA approach (175 ± 80 min; P = 0.40). The mean duration of the arterial procedure was significantly longer for patients with an Edwards valve (193 ± 77 min) compared with those with a CoreValve (174 ± 96 min; P = 0.009), which was not linked to a difference in the choice of intervention site. The mean duration for patients treated in the catheterization laboratory (165 ± 58 min) was significantly lower than the mean duration for patients treated in the operating room (229 ± 128 min; P < 0.0001) or the hybrid room (220 ± 84 min; P < 0.0001). The mean duration of the procedure did not differ significantly between the hybrid room and the operating room ( P = 0.9206).
Length of hospital stay
The average length of stay was 14.83 ± 11.04 days (median, 12 days) and decreased from a median of 13 days for the first third of patients (by order of implantation date) to a median of 11 days for the last third. Mean length of stay did not significantly differ between the subgroups distinguishing approach and intervention site.
Cost of the procedure
The total average cost of the TAVI procedure was € 22,876, of which 85% was attributable to the valve price, 7% to other material, 7% to staffing and 1% to logistics ( Table 2 ).
|Unit cost||Mean cost per patient ± SD||Median cost per patient|
|Devices and consumables a|
|Temporary bipolar electrode||123.91||124||124|
|Percutaneous vascular suture system||557.86||536||558|
|Packed red blood cells or platelets||106.56||70||0|
|Other consumables c||58||56|
|Total cost of devices and consumables||21,103 ± 2674||20,990|
|Staff d , e|
|Cardiac or vascular surgeon, interventional cardiologist, anesthesiologist, radiologist||1.17||1022||916|
|Pump oxygenator nurse||0.83||40||0|
|Total staff cost||1548 ± 786||1382|
|Intervention site d|
|Total room and ancillary cost||225 ± 150||174|
|Total cost of procedure||22,876 ± 2972||22,435|
|Hospital stay excluding procedure|
|Daily medical and overhead cost in addition to procedure cost||801.3||11,767||9616|
|Total cost of hospital stay excluding procedure||12,288 ± 8917||9616|
|Total cost of hospital stay including procedure||35,164 ± 9450||32,577|