Real-world versus randomised controlled trial data: a case study on the cost-effectiveness of laparoscopic surgery for chronic reflux

TY - GEN

T1 - Real-world versus randomised controlled trial data

T2 - ISPOR 17th Annual Congress

AU - Faria, Rita

AU - Liu, S

AU - Epstein, D

AU - Manca, A

PY - 2014

Y1 - 2014

N2 - OBJECTIVES Real-world (observational) data has the potential to address the limitations of randomised controlled trials (RCTs) but presents its own challenges given the increased risk of bias. We compared the costs and quality-adjusted life years (QALYs) of patients following random vs preference-based allocation and assessed the performance of different methods to address these challenges. METHODS The REFLUX study was a pragmatic trial in patients with chronic reflux comparing laparoscopic fundoplication (surgery) with medical management (MM) over 5 years of follow-up. The trial included randomised and non-randomised preference-based allocation. We compared the cost-effectiveness of surgery in the RCT vs preference cohorts as unadjusted raw differences, applying methods to handle biases from selection and confounding (regression adjustment, propensity score matching and instrument variable analysis) and explored the impact of receiving the preferred treatment on the results. RESULTS The preference surgery group accrued greater costs and QALYs than the randomised surgery group (£3,524 vs £2,852; 3.723 vs 3.612 QALYs). The preference MM group had lower costs but slightly better QALYs than the randomised MM group (£861 vs £1,415; 3.541 vs 3.411 QALYs). The incremental cost-effectiveness ratio (ICER) for the preference cohorts was similar to that obtained in the RCT using the different methods at around £8,000 per QALY gained (vs £7,149 in the RCT and £14,632 unadjusted raw differences). Receiving the preferred treatment was significantly associated with lower costs and better QALYs after adjusting for prognostic variables. CONCLUSIONS Real-word data can be used in cost-effectiveness analysis to complement RCT evidence. However, more research is needed on how to choose the most appropriate method to adjust for selection bias and how to account for patient preferences when making recommendations on value for money.

AB - OBJECTIVES Real-world (observational) data has the potential to address the limitations of randomised controlled trials (RCTs) but presents its own challenges given the increased risk of bias. We compared the costs and quality-adjusted life years (QALYs) of patients following random vs preference-based allocation and assessed the performance of different methods to address these challenges. METHODS The REFLUX study was a pragmatic trial in patients with chronic reflux comparing laparoscopic fundoplication (surgery) with medical management (MM) over 5 years of follow-up. The trial included randomised and non-randomised preference-based allocation. We compared the cost-effectiveness of surgery in the RCT vs preference cohorts as unadjusted raw differences, applying methods to handle biases from selection and confounding (regression adjustment, propensity score matching and instrument variable analysis) and explored the impact of receiving the preferred treatment on the results. RESULTS The preference surgery group accrued greater costs and QALYs than the randomised surgery group (£3,524 vs £2,852; 3.723 vs 3.612 QALYs). The preference MM group had lower costs but slightly better QALYs than the randomised MM group (£861 vs £1,415; 3.541 vs 3.411 QALYs). The incremental cost-effectiveness ratio (ICER) for the preference cohorts was similar to that obtained in the RCT using the different methods at around £8,000 per QALY gained (vs £7,149 in the RCT and £14,632 unadjusted raw differences). Receiving the preferred treatment was significantly associated with lower costs and better QALYs after adjusting for prognostic variables. CONCLUSIONS Real-word data can be used in cost-effectiveness analysis to complement RCT evidence. However, more research is needed on how to choose the most appropriate method to adjust for selection bias and how to account for patient preferences when making recommendations on value for money.

M3 - Conference contribution

BT - The Ispor Scientific Presentations Database

Y2 - 8 November 2014 through 12 November 2014

ER -