의약품 급여관리를 위한 실제임상근거(RWE) 플랫폼 마련 후향적 연구(청구자료 분석)

Abstract

Background It is more difficult to conduct randomized controlled trials (RCT) in R&D of pharmaceuticals for treating cancer or rare disease than other diseases, therefore leading to increasing the uncertainty in pricing and reimbursement. To resolve these uncertainties of decision-making, health technology assessment (HTA) agencies in many countries including Korea currently explore possibilities for using the real-world evidence (RWE).

Objectives This study aims to generate the Korea real-world evidence (K-RWE) drawn from the K-RWD (Korea real-world data). Therefore based on the results, this study aims to suggest model for constructing the infrastructure of collecting K-RWD and the possibility using K-RWE for decision-making in pharmaceutical policy (i.e. HTA reassessment).

Methods We implemented this study to compare effectiveness at the time of drug approval time and after the appproval for two selected anti-cancer medicines (ramucirumab for gastric cancer and trastuzumab emtansine for breast cancer). First, we reviewed the quality level on K-RWD available such as HIRA claims data, electronic medical records(EMR). Second, we decided the necessary information for pharmaceutical health technology assessment (HTA). Second, we decided the necessary information for pharmaceutical health technology assessement (HTA). Third, we made case report form (CRF) for two selected medicines each. Fourth, we collected K-RWD through retrospective studies on the effectiveness and safety of two selected medicines. Fifth, we analyzed the effectiveness and safety of two selected anti-cancer medicines using K-RWD (HIRA claims data and EMR in 68 hospitals). Lastly, we explored further steps for robust RWE generation and constructing infrastructure of collecting K-RWD.

Results HIRA claims data covered almost all Korean people because Korea achieved universal health coverage (UHC), but there is still lack of information such as disease stage, progress of disease, and so on.We surveyed on the quality of EMR data, and 55 of 68 hospitals participated. It displayed the coding rate of RECIST (44.4% of tertiary hospitals vs 73.7% of secondary hospitals 73.7%) and the NCI-CTCAE (58.3% of tertiary hospitals vs 73.7% of secondary hospitals). It needs coding on the ECOG, diagnosis date, the results of RECIST, disease progression, and death date. The difference for review period on IRB among participated hospitals were shown. We confirmed the consistence (gastric cancer team: 73.5±21.3%, breast cancer team: 71.9±16.9%) and confidence (0.89±0.15, 0.86±0.17 respectively) on major contents in CRF among researchers. The missing rate highly showed difference among hospitals. This study compared the results between RCT and RWE. Overall survival (OS) period on ramucirumab showed 9.6 months (95% CI 8.5-10.8) from RCT (n=330), and 10.0 months (95% CI 9.3-10.7) from EMR (n=1,063), and 7.9 months (95% CI 7.6-8.6) from HIRA claims data (n=1,419). Progression free survival (PFS) period on trastuzumab emtansine respectively displayed 9.6 months from EMILIA study (n=495), and 6.7 months (95% CI 6.2-7.4) from total patients in K-RWE (n=824). PFS in severe patients were 8.0 months (95% CI 6.8-9.4) from 2nd line use in K-RWE (n=329), 6.2 months from TH3RESA study (n=404, 95% CI 5.6-6.9). Shortly, the results from HIRA data were the lowest in gastric cancer and in breast cancer. However, these results required careful interpretation. HIRA data has not clinical information of patients and EMR data did not know whether patients die in others hospitals or nursing homes. Finally, this study found that K-RWD could be better planning to collect data in advance, developing common code on the major contents in individual hospital to integrate each information, and processing common IRB committee on government research.

Discussion This study confirmed the potential power on K-RWD even though some limitation. K-RWD will be powerful data when these limitations should be resolved and other data could be linked. HIRA also tried to collect the patient reported outcomes (utility, financial burden and improving the subjective symptoms), clinical and safety information. Ultimately, we try to the integrated RWD dataset per individual patient through the linkage between the claims data and EMR. Therefore, it is necessary for prospective collecting K-RWD to know the baseline information of patients for assessment on intervention.