Evaluation of the ventricular assist device programme in the UK.

OBJECTIVES: To summarise the relevant clinical effectiveness and cost-effectiveness literature, to collect data on survival, transplantation rates, health-related quality of life (HRQoL) and resource use for ventricular assist device (VAD) and non-VAD transplant candidates in the UK, and to construct cost-effectiveness and cost-utility models of VADs in a UK context. Also to investigate the factors that drive costs and survival.

DESIGN: A comprehensive systematic review was carried out. Data were collected from April 2002 to December 2004, with follow-up to March 2005. Cost-effectiveness and cost-utility models of VAD devices were developed based on UK activity and outcomes collected from April 2002 to March 2005.

SETTING: National Specialist Commissioning Advisory Group funded VAD implantation was carried out at the Freeman, Harefield and Papworth transplant centres in the UK.

PARTICIPANTS: Seventy patients were implanted with a VAD as a bridge to transplantation between April 2002 and December 2004. Non-VAD-supported transplant candidates (n = 250), listed at the three centres between April 2002 and December 2004, were divided into an inotrope-dependent group (n = 71) and a non-inotrope-dependent group (n = 179). Although patients in the inotrope-dependent group were closest to the VAD group they were less sick. The last group comprised a hypothetical worst case scenario, which assumed that all VAD patients would die in the intensive care unit (ICU) within 1 month without VAD technology.

INTERVENTIONS: Patients were included who were implanted with a VAD designed for circulatory support for more than 30 days, with intention to bridge to transplantation. A multistate model of VAD and transplant activity was constructed; this was populated by data from the UK.

MAIN OUTCOME MEASURES: Survival from VAD implant or from transplant listing for non-VAD patients to 31 March 2005. Serious adverse events and quality of life measures were used. Cognitive functioning was also assessed. Utility weights were derived from EuroQoL responses to estimate quality-adjusted life-years (QALYs). Incremental cost-effectiveness ratios (ICERs) were defined as the additional cost of VADs divided by additional QALYs. Time-horizons were 3 years, 10 years and the lifetime of the patients.

RESULTS: Of 70 VAD patients, 30 (43%) died pretransplant, 31 (44%) underwent transplantation, and four (6%) recovered and had the VAD removed. Five patients (7%) were still supported for median of 279 days at the end of March 2005. Successful bridge-to-transplantation/recovery rates were consistent with published rates. Survival from VAD implantation was 74% at 30 days and 52% at 12 months. There were 320 non-fatal adverse events in 62 patients during 300 months of VAD support, mostly in the first month after implantation. Commonly observed events were bleeding, infection and respiratory dysfunction. Twenty-nine (41%) patients were discharged from hospital with a VAD. The 1-year survival post-transplantation was 84%. For the inotrope-dependent and non-inotrope-dependent transplant candidates, death rates while listed were 10% and 8% and the median waiting times were 16 and 87 days, respectively. For transplant recipients, 1-year survival was 85% and 84%, respectively. Both VAD and non-VAD patients demonstrated similar significant improvements in their New York Heart Association class after transplantation. All patients had poor EQ-5D pretransplantation; after transplantation the groups had similar EQ-5D of 0.76 irrespective of time after surgery. HRQoL was poor in the first month for VAD patients but better for those who waited longer in all groups. VAD patients reported more problems with sleep and rest and with ambulation in the first month. Symptom scores were similar in all groups pretransplant. After transplantation all groups showed a marked and similar improvement in physical and psychosocial function. Mean VAD implant cost, including device, was pound 63,830, with costs of VAD support for survivors of pound 21,696 in month 1 and pound 11,312 in month 2. Main cost drivers were device itself, staffing, ICU stay, hospital stay and events such as bleeding, stroke and infection. For the base case, extrapolating over the lifetime of the patients the mean cost for a VAD patient was pound 173,841, with mean survival of 5.63 years and mean QALYs of 3.27. Corresponding costs for inotrope-dependent patients were pound 130,905, with mean survival 8.62 years and mean QALYs 4.99. Since inotrope-dependent patients had lower costs and higher QALYs than VAD patients, this group is said to be dominant. Non-inotrope-dependent transplant candidates had similar survival rates to those on inotropes but lower costs, also dominant. Compared with the worst case scenario the mean lifetime ICER for VADs was pound 49,384 per QALY. In a range of sensitivity analyses this ranged from pound 35,121 if the device cost was zero to pound 49,384. Since neither inotrope-dependent transplant candidates nor the worst case scenario were considered fair controls the assumption was investigated that, without VAD technology, there would be a mixture of these situations. For mixtures considered the ICER for VADs ranged from pound 79,212 per QALY to the non-VAD group being both cheaper and more effective.

CONCLUSIONS: There are insufficient data from either published studies or the current study to construct a fair comparison group for VADs. Overall survival of 52% is an excellent clinical achievement for those young patients with rapidly failing hearts. However, if the worst case scenario were plausible, and one could reliably extrapolate results to the lifetime of the patients, VADs would not be cost-effective at traditional thresholds. Further randomised controlled trials are required, using current second generation devices or subsequent devices and conducted in the UK.