Impact of the Pilot Volume-Based Drug Purchasing Policy in China: Interrupted Time-Series Analysis with Controls

Centralizing procurement for prescription drugs has the potential to reduce drug spending by creating economies of scale and by improving purchasing power. In March 2019, the Chinese government launched a volume-based purchasing (VBP) pilot program using a competitive bidding process to purchase accredited generic drugs for which branded drug substitutes were available. We performed an interrupted time-series design to estimate the change in monthly drug purchase quantity and spending comparing 14 months before and 7 months after the VBP pilot. We obtained monthly prescription drug purchase data for all purchases from public medical institutions in the three large pilot cities (Beijing, Shanghai and Xi’an) and two non-pilot cities (Changsha and Zhengzhou) between January 2018 to September 2019. We used negative binomial regression and log-linked Gamma Generalized Linear Model for purchase quantity and spending respectively. We evaluated heterogeneity of impact by pilot city, drug type (selected or non-selected drugs), and therapeutic class (cardiovascular disease, mental disorder and cancer) separately. The implementation of the pilot reform was associated with a 132% (95%-CI: 104–165%, p < 0.001) increase in the purchase quantity of selected drugs in pilot cities compared to an 17% decrease (95%-CI: 9–25%, p < 0.001) in control cities. In contrast, the purchase quantity of branded and other drugs in pilot cities decreased by 38% (95%-CI: 27–46%, p < 0.001) and 77% (95%-CI: 71–81%, p < 0.001), respectively; while in control cities, these remained at similar levels. Overall, in pilot cities, there was a 35% (95%-CI: 28–41%, p < 0.001) decrease in the purchase spending for all drugs in the first post-policy month, from 8.1 billion CNY estimated in the absence of VBP down to 5.3 billion CNY; in control cities, the change was negligible. The largest reduction in spending occurred for drugs for the treatment of cardiovascular diseases. The evidence suggests a positive impact of the VBP pilot in reducing overall drug spending and increasing the use of accredited generics in three pilot cities. This overall trend is not observed in two non-pilot cities. Assessments of long-term impact of the VBP policy on additional key outcomes including drug prescriptions, drug utilization, patients’ health outcomes and payments on drugs are needed.

. Observed average monthly drug purchase before and after the pilot program Table S3. Interrupted time-series regression model estimates (stratified by drug categories and cities) Table S4. Overall changes in purchase stratified by drug and cities, March-September 2019 Figure S1. Quantity of drug purchase stratified by cities and drug categories over time.

The RECORD Statement
This supplementary material has been provided by the authors to give readers additional information about their work.  actual purchase quantity (in thousand) by drug categories; (B) Expected versus actual purchase quantity (in 1,000 DDD) and spending (in 100,000 CNY) for all drugs. Specific point estimates for absolute change and the corresponding 95% confidence intervals are provided in Table 2.    where Yjt is the independent outcome variable (either monthly purchase quantity/spending) for drug type (D) j in city (i) at time t , T and T-T0 is the time (month) since the start of the study (January 2018) and the time since the implementation of pilot program (T0 : March 2019) respectively. Xt is an indicator variable where pre-intervention is coded as 0 and post-intervention is coded as 1. M is the indicator variable for calendar month used to account for seasonal variation.
The RECORD statement -checklist of items, extended from the STROBE statement, that should be reported in observational studies using routinely collected health data.

Item
No. Objectives 3 State specific objectives, including any prespecified hypotheses "Our study aims to evaluate the impact of the pilot on drug procurement using an interrupted time series analysis with controls in the three large pilot cities (i.e., Beijing, Shanghai, and Xi'an) with a total population of about 60 million. We also examine if the patterns of change associated with the reform differed by cities, drugs, and therapeutic categories. Since the majority of individuals seek care in public hospitals in China, and hospitals automatically convert prescriptions to selected generics, our analysis is generalizable to almost the entire population in these pilot cities.

STROBE items
"

Study Design 4
Present key elements of study design early in the paper "We examined changes in the pattern of drug procurement associated with the implementation of the pilot using interrupted time-series (ITS) design, a quasi-experimental design for strong causal inference in the evaluation of population- Participants 6 (a) Cohort study -Give the eligibility criteria, and the sources and methods of selection of participants.

Describe methods of follow-up
Case-control study -Give the eligibility criteria, and the sources and methods of case ascertainment and control selection. Give the rationale for the choice of cases and controls Cross-sectional study -Give the eligibility criteria, and the sources and methods of selection of participants

Results
Participants 13 (a) Report the numbers of individuals at each stage of the study (e.g., numbers potentially eligible, examined for eligibility, confirmed eligible, included in the study, completing follow-up, and analysed) (b) Give reasons for nonparticipation at each stage.
(c) Consider use of a flow diagram RECORD 13.1: Describe in detail the selection of the persons included in the study (i.e., study population selection) including filtering based on data quality, data availability and linkage. The selection of included persons can be described in the text and/or by means of the study flow diagram.
"In total, 20.5 billion CNY was spent on 481.5 billion drugs in pilot cities, and 2.6 billion CNY was spent on 41.3 million drugs in non-pilot cities from January 2018 to September 2019 (Table S2)." Descriptive data 14 (a) Give characteristics of study participants (e.g., demographic, clinical, social) and information on exposures and potential confounders