Impacts of national volume-based procurement policy on insulin price, volume and expenditure in China: a quasi-experimental study

Abstract

Introduction

To alleviate the diabetes economic burden, the Chinese government implemented the national volume-based procurement (NVBP) policy exclusively for insulin in May 2022. However, there is limited evidence on the impact of NVBP policy on insulin procurement prices, volumes and expenditures.

Methods

We conducted an interrupted time series analysis (ITSA) to evaluate the policy’s impacts on insulin procurement prices, volumes and expenditure. Quarterly data for 15 insulins were extracted from the China Medicine Economic Information database, covering 761 public hospitals in 30 provinces from Q3 2020 to Q4 2023. The intervention time of the ITSA model is Q2 2022. Insulin prices are measured using the defined daily dose (DDD) cost, while the volumes are measured using DDDs.

Results

Following the NVBP implementation, all NVBP bid-winning insulin prices decreased by 51.61%, volume increased by 16.73% and expenditure decreased by 43.52%. All NVBP bid-winning insulin volumes increased by 9152.73 thousand DDDs and showed an upward trend (p<0.05). The expenditure decreased by 176.65 million Chinese yuan (CNY) (p<0.05). The prices of all six NVBP insulin groups decreased significantly, and the prices of the three NVBP insulin groups showed a downward trend (p<0.05). The volumes of all six NVBP insulin groups showed significant increases in either level or trend. Except for premixed human insulin analogues, expenditure significantly changed in level or trend in the remaining five groups (p<0.05). Individually, all 15 insulins exhibited significant price reductions, while 10 of them also showed significant increases in both volumes and expenditures.

Conclusions

China’s NVBP policy exclusively for insulin significantly reduced prices, increased overall volumes and lowered expenditures. These changes indicate that this policy enhances insulin utilisation and affordability, providing evidence for centralised procurement’s effectiveness in mitigating diabetes-related economic burden.

WHAT IS ALREADY KNOWN ON THIS TOPIC

• Diabetes imposes a heavy health and economic burden globally, particularly in China, where cases surged to 140.9 million adults in 2021, with healthcare costs projected to reach US$460.4 billion by 2030. And such an escalating burden has raised concern about medication utilisation and affordability.

• Insulin serves as a foundational therapy essential for managing both type 1 and type 2 diabetes, and domestic and international guidelines increasingly prioritise it as an initial therapy. However, its high cost in China creates substantial financial barriers for patients.

• China implemented the national volume-based procurement (NVBP) policy exclusively for insulin in 2022 to address insulin utilisation and affordability by categorising 15 insulins into six groups based on action time. However, the impacts of this targeted policy on insulin prices, volumes and expenditures, especially at a group or individual level, remained scant.

WHAT THIS STUDY ADDS

• Our analysis is the first to apply interrupted time series analysis to quantify the level and trend impacts of the NVBP policy exclusively for insulin on procurement prices, volumes and expenditures with data from Q3 2020 to Q4 2023.

• Our analysis provides a more comprehensive assessment of the NVBP policy’s impact on insulin by analysing its effects at both the group-insulin and individual-insulin levels.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• The study demonstrates the NVBP policy’s effectiveness in enhancing insulin utilisation and affordability, supporting its expansion to other high-burden chronic disease medications. To further promote equity, policymakers should implement tailored adjustments, such as expenditure increases for certain individual insulins.

• The analysis reveals how centralised procurement altered the competitive landscape across insulin groups and individual insulins while also explaining differential changes in procurement prices and volumes among insulin categories.

Introduction

The global prevalence of diabetes mellitus has reached epidemic proportions, driven by ageing populations, urbanisation and a concurrent rise in obesity and physical inactivity. By 2030, diabetes is expected to affect 4.4% of the global population, totalling approximately 366 million individuals.¹ With such a tremendous disease burden, diabetes has become a significant public health challenge, being associated with various morbidities and mortalities, including acute metabolic complications and microvascular and cardiovascular diseases.² As a densely populated country, China experiences a relatively high disease and economic burden resulting from diabetes mellitus.³ China now has the world’s largest population of adults (20–79 years) suffering from diabetes—approximately 140.9 million in 2021—with prevalence in this age group projected to rise from 8.2% in 2020 to 9.7% by 2030.^{4 5} Despite this growing burden, national surveillance found that only 36.5% of people with diabetes were aware of their condition, 32.2% received treatment and just 49.2% achieved adequate glycaemic control.⁶ Concurrently, the financial impact of diabetes in China is expected to surge from US$250.2 billion to US$460.4 billion within the same period, with an annual growth rate of approximately 6.32%, outpacing the nation’s economic growth due to the increasing proportion of healthcare expenditure relative to China’s gross domestic product from 1.58% to 1.69%.

To mitigate the economic burden of medication costs associated with diabetes, the Chinese government has implemented the national volume-based procurement (NVBP) policy.⁷ This policy leverages centralised purchasing power to pool procurement and secure a significant market share for winning bidders at lower prices, thereby reducing the financial burden on patients and successfully saving over 260 billion Chinese yuan (CNY) by the end of 2021. Insulin, a cornerstone in the treatment of both type 1 and type 2 diabetes, has been consistently recommended as an initial therapy by domestic and international diabetes guidelines. Patients with type 1 diabetes require lifelong insulin replacement to survive, as they are unable to produce endogenous insulin. In type 2 diabetes, insulin becomes necessary when oral antihyperglycaemic agents fail to achieve adequate glycaemic control or are contraindicated.⁸ However, high insulin costs have long been a significant barrier to treatment access, with evidence showing that insulin treatment costs could account for a significant proportion of monthly income, rendering some insulin products unaffordable for low-income populations.⁹ In response, a novel NVBP exclusively for insulin was launched in November 2021 and officially implemented in May 2022.^{10 11} The programme categorised second-generation and third-generation insulins into six groups based on their action time: basic human insulin, basic human insulin analogues, mealtime insulin, mealtime insulin analogues, premixed human insulin and premixed human insulin analogues. This policy established a maximum valid bidding price per 300 international units for each category, with bids considered successful only if they are below 60% of the preset maximum bid price. Winning insulin products are classified into three categories—A, B and C—based on their bidding prices, while qualifying but unsubmitted or unsuccessful products are designated as category D. Within each procurement group, medical institutions redistribute 30% of category C’s initial allocation and 80% of category D’s initial allocation to categories A and B, respectively.¹²

Previous studies have extensively investigated the impact of the NVBP on drug procurement prices, volumes and expenditure, particularly for antiviral medications for the hepatitis B virus and anticancer drugs, revealing significant price reductions, decreased total expenditure and increased drug volumes.13,17 However, the impact of the sixth batch of NVBP, specifically targeting insulin procurement, on utilisation and affordability remains unclear.

This study employs an interrupted time series analysis (ITSA) to quantitatively evaluate the impact of the targeted NVBP on insulin prices, procurement volumes and expenditure, aiming to provide valuable insights into the effectiveness of this health policy intervention.

Methods

Insulin selection and data source

A total of 16 insulin products were identified based on the sixth NVBP batch list released by the Chinese government and then converted to their corresponding generic drug names. During this process, products with the same generic drug names were merged together, resulting in the final list of 15 insulins, including human insulin, insulin aspart 30, insulin aspart 50, mixed protamine zinc recombinant human insulin lispro (25R), mixed protamine zinc recombinant human insulin lispro (50R), insulin degludec, insulin detemir, insulin glargine, insulin aspart, insulin glulisine, insulin lispro, mixed protamine human insulin (30R), mixed protamine human insulin (40R), mixed protamine human insulin (50R) and protamine human insulin.

Hospital procurement data for these insulins from Q3 2020 through Q4 2023 across 30 provincial-level administrative regions in mainland China (excluding Tibet, Taiwan, Hong Kong and Macau owing to data unavailability) were extracted from the Chinese Medicine Economic Information (CMEI) database. CMEI is one of the largest hospital procurement data platforms in China, collecting pharmaceutical purchase records from designated public hospitals in each province and providing detailed information on generic drug name, anatomical therapeutic chemical classification system, NVBP bidding status, unit purchase price, purchase volume, total expenditure and manufacturers. The data extracted for this analysis spanned continuously from 2020 to 2023, encompassing 761 public medical institutions across China.

Outcome variables

In this study, three main outcome measures were selected to assess the impact of the NVBP policy on insulins: procurement price, procurement volume and spending. Our study applied the defined daily dose (DDD) cost (DDDc) and DDDs to represent price and volume. The DDD, recommended by the WHO, is the commonly used measurement unit in drug utilisation research.¹⁸ Using drug volume data, both DDDc and DDDs can be calculated. The DDDs represent the ratio of the total drug dosage and DDD to reflect overall drug consumption. Higher DDDs indicate a greater frequency of drug use, increased treatment intensity and higher clinical preference. If the number of patients does not increase significantly after the policy is implemented, then the increase in DDDs can indicate an improvement in utilisation. DDDc is the ratio of procurement spending to total DDDs, serving as a surrogate measure of the actual insulin price. All prices in this study are reported in CNY. To assist international readers in understanding the price levels, the annual average exchange rates in 2021 were approximately 6.45 CNY per US$ and 7.62 CNY per Euro, based on data from the State Administration of Foreign Exchange.

Statistical analysis

Due to a substantial lack of data for non-winning insulins, identifying a suitable control group for controlled ITSA or difference-in-differences was not feasible. Therefore, our analysis performed ITSA, a robust quasi-experimental design widely used in policy evaluations.19,23 Insulins were stratified into their NVBP policy groups to quantify the impact of NVBP policy on insulins, procurement prices, procurement volumes and expenditure. The regression model assumes the following form:

$$Y_t={\beta }_0+{\beta }_1{T}_t+{\beta }_2{X}_t+{\beta }_3{X}_t{T}_t+{\epsilon }_t$$

$Y_t$ is the outcome variable measured at each quarterly time point t, including DDDc, DDDs and procurement expenditure. T_t represents the time since the start of the study, $X_t$ is a dummy variable representing the intervention (preintervention periods are 0, otherwise 1), and $X_t{T}_t$ is an interaction term of the time and intervention. ${\epsilon }_t$ is the error term, representing random errors’ variation not explained by variables in the model. ${\beta }_0$ represents the initial level of the outcome variable. ${\beta }_1$ estimates the trend or slope of the outcome variable before introducing policy. ${\beta }_2$ estimates the immediate level change of the outcome variable upon policy implementation, representing the intervention effect in the implementation quarter. ${\beta }_3$ estimates the net effect of the outcome variable before and after the policy implementation, representing the intervention effect over time.²⁴

When using monthly data for policy evaluation, study designs typically account for implementation lag by delaying the intervention time by 1 month in the models.²⁵ However, despite potential lag effects, the quarterly data used in this study indicate that the NVBP intervention occurred in the second quarter of 2022. Accordingly, the intervention point in the ITSA model was explicitly set as Q2 2022, aligning with the timing of NVBP policy implementation. The observation period of the model spanned from the third quarter of 2020 to the fourth quarter of 2023 (seven quarters before and after the policy implementation, ensuring that the ITS model has statistical validity).²⁶ All raw data processing was performed in Microsoft Excel 2021, and the construction, analysis and robustness test of the ITS model were performed in Stata 17 (V.1.4.1717). We applied the Newey–West estimator to our generalised linear model regression to correct for potential autocorrelation and heteroscedasticity in the parameter estimates.²⁴ In addition, this study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline and Consolidated Health Economic Evaluation Reporting Standards 2022 (CHEERS2022).

Results

Descriptive analysis

Table 1 displays insulin prices, volumes and expenditure before and after NVBP implementation, along with their respective growth rates (GR). Following the introduction of the policy, the prices of all NVBP bid-winning insulin products decreased significantly, with an average reduction of 51.61%. Premixed human insulin analogues demonstrated the largest price decline (GR=–53.87%), while at the individual insulin level, mixed protamine zinc recombinant human insulin lispro (50R) experienced the greatest reduction (GR=–69.68%). Although the total insulin volume increased overall, volumes decreased for certain NVBP insulin groups (eg, basic human insulin) and specific individual insulins (eg, insulin detemir). With the exception of insulin aspart 30, procurement expenditure for all other insulin products declined significantly.

Table 1. Comparison of insulin price (in DDDc), volume and expenditure before and after NVBP implementation.

NVBP groups	Price (CNY per DDD)			Volume (thousand DDDs)			Expenditure (million CNY)
	Pre-NVBP	Post-NVBP	GR	Pre-NVBP	Post-NVBP	GR	Pre-NVBP	Post-NVBP	GR
Basic human insulin	7.50	4.87	−35.07%	3398.28	1956.51	−42.43%	25.50	9.52	−62.67%
Protamine human insulin	7.50	4.87	−35.07%	3398.28	1956.51	−42.43%	25.50	9.52	−62.67%
Basic human insulin analogues	16.81	7.77	−53.78%	139 731.66	158 711.58	13.58%	2349.25	1233.25	−47.50%
Insulin degludec	25.62	11.76	−54.10%	13 688.85	23 494.49	71.63%	350.65	276.31	−21.20%
Insulin detemir	6.26	2.71	−56.71%	51 435.99	43 183.62	−16.04%	322.19	117.22	−63.62%
Insulin glargine	22.47	9.12	−59.41%	74 606.82	92 033.47	23.36%	1676.41	839.72	−49.91%
Mealtime insulin	6.98	4.16	−40.40%	10 787.05	15 980.25	48.14%	75.33	66.48	−11.75%
Human insulin	6.98	4.16	−40.40%	10 787.05	15 980.25	48.14%	75.33	66.48	−11.75%
Mealtime insulin analogues	9.57	4.96	−48.21%	64 717.18	77 011.06	19.00%	619.46	381.76	−38.37%
Insulin aspart	9.70	5.58	−42.47%	45 637.69	48 735.96	6.79%	442.52	272.03	−38.53%
Insulin glulisine	10.13	6.39	−36.92%	4200.99	4432.60	5.51%	42.54	28.31	−33.45%
Insulin lispro	9.03	3.41	−62.24%	14 878.50	23 842.50	60.25%	134.40	81.42	−39.42%
Premixed human insulin	7.25	4.64	−36.00%	37 468.47	28 040.04	−25.16%	271.64	130.11	−52.10%
Mixed protamine human insulin (30R)	7.27	4.70	−35.35%	33 180.22	24 030.06	−27.58%	241.15	112.97	−53.15%
Mixed protamine human insulin (40R)	6.97	4.05	−41.89%	2635.87	2144.56	−18.64%	18.37	8.68	−52.75%
Mixed protamine human insulin (50R)	7.33	4.54	−38.06%	1652.38	1865.42	12.89%	12.12	8.46	−30.20%
Premixed human insulin analogues	9.39	4.33	−53.87%	34 732.87	57 779.12	66.35%	326.17	250.30	−23.26%
Insulin aspart 30	9.77	5.09	−47.90%	10 752.83	34 788.01	223.52%	105.04	177.19	68.69%
Insulin aspart 50	9.34	6.00	−35.76%	6519.25	1826.24	−71.99%	60.86	10.96	−81.99%
Mixed protamine zinc recombinant human insulin lispro (25R)	8.69	3.28	−62.26%	5502.60	4435.42	−19.39%	47.81	14.53	−69.61%
Mixed protamine zinc recombinant human insulin lispro (50R)	9.40	2.85	−69.68%	11 958.19	16 729.45	39.90%	112.46	47.62	−57.66%
Total	12.61	6.10	−51.61%	290 835.51	339 478.56	16.73%	3667.35	2071.42	−43.52%

CNY, Chinese Yuan; DDD, defined daily doseDDDc, defined daily dose cost; GR, growth rate; NVBP, national volume-based procurement.

Changes in the price, volume and expenditure of all NVBP bid-winning insulins

During the NVBP implementation quarter, the price (in DDDc) of all NVBP bid-winning insulins decreased sharply by 6.34 CNY per DDD (${\beta }_2$=–6.34, 95% CI –7.92 to –4.75, p<0.001). Following the policy, all NVBP bid-winning insulin prices demonstrated a significant decline compared with the preintervention period (${\beta }_3$=–0.52, 95% CI –0.90 to –0.15, p<0.01). Insulin volumes increased by 9152.73 thousand DDDs (${\beta }_2$=9152.73, 95% CI 5404.75 to 12 900.71, p<0.001), and the previous downward trend (${\beta }_1$=−1095.49) reversed to an upward trend (${\beta }_3$=1821.56, 95% CI 785.21 to 2857.91, p<0.01). The level of total insulin procurement expenditure decreased by 176.65 million CNY (${\beta }_2$=–176.65, 95% CI –240.66 to –112.63, p<0.001), with no significant change observed in the postintervention trend (p>0.05) (online supplemental figure S1 and table S1).

Changes in the price, volume and expenditure of the NVBP insulin group

All six NVBP insulin groups exhibited significant declines in price. Over the longer term, basic human insulin analogues, mealtime insulin analogues and premixed human insulin analogues experienced statistically significant downward trend shifts (p<0.05). Although the trend changes for basic human insulin, mealtime insulin and premixed human insulin did not reach statistical significance, they nonetheless displayed a clear downward pattern visually (table 2, figure 1 and online supplemental table S2).

Table 2. Interrupted time series statistical results of NVBP insulin group price, volumes and expenditure.

Outcome variables for NVBP insulin groups	Coefficient estimates (95% CI)
	Level change (${\beta }_2$)	Trend change (${\beta }_3$)
Basic human insulin
Price (CNY per DDD)	−2.09 (−2.6 to −1.58)***	−0.01 (−0.16 to 0.13)
Volume (thousand DDDs)	175.74 (73.88 to 277.6)**	47.72 (28.67 to 66.76)***
Expenditure (million CNY)	0.58 (−0.15 to 1.3)	0.44 (0.26 to 0.61)***
Basic human insulin analogues
Price (CNY per DDD)	−8.96 (−11.34 to −6.58)***	−0.67 (−1.19 to −0.15)*
Volume (thousand DDDs)	3051.88 (1224.8 to 4878.96)**	358.48 (−28.85 to 745.82)
Expenditure (million CNY)	−146.75 (−191.49 to −102.02)***	−9.42 (−21.56 to 2.72)
Mealtime insulin
Price (CNY per DDD)	−2.08 (−2.62 to −1.55)***	−0.13 (−0.26 to 0.01)
Volume (thousand DDDs)	390.12 (207.29 to 572.95)***	159.92 (108.74 to 211.11)***
Expenditure (million CNY)	−1.24 (−2.21 to −0.27)*	0.44 (0.24 to 0.65)***
Mealtime insulin analogues
Price (CNY per DDD)	−3.69 (−4.81 to −2.57)***	−0.3 (−0.57 to −0.03)*
Volume (thousand DDDs)	448.34 (−424.26 to 1320.94)	265.9 (34.47 to 497.33)*
Expenditure (million CNY)	−32.39 (−47.06 to −17.71)***	−2.2 (−5.42 to 1.03)
Premixed human insulin
Price (CNY per DDD)	−2.01 (−2.57 to −1.46)***	−0.06 (−0.23 to 0.1)
Volume (thousand DDDs)	1485.46 (998.03 to 1972.89)***	276.58 (162.65 to 390.51)***
Expenditure (million CNY)	1.29 (−2.76 to 5.35)	2.45 (1.25 to 3.64)***
Premixed human insulin analogues
Price (CNY per DDD)	−4.36 (−5.26 to −3.47)***	−0.22 (−0.43 to 0)*
Volume (thousand DDDs)	3601.19 (1975.55 to 5226.82)***	712.96 (235.75 to 1190.17)**
Expenditure (million CNY)	1.86 (−8.8 to 12.52)	3.37 (−0.47 to 7.21)

*p<0.05; **p<0.01; ***p<0.001.

CNY, Chinese Yuan; DDD, defined daily dose; NVBP, national volume-based procurement.

Figure 1. Price changes for the national volume-based procurement insulin group.

Except for mealtime insulin analogues, the volumes of the other five NVBP insulin groups increased significantly. The trend of volumes of most NVBP insulin groups (5/6) increased significantly. (table 2, figure 2 and online supplemental table S2).

Figure 2. Volume changes for the national volume-based procurement insulin group.

Apart from premixed human insulin analogues, the procurement expenditure of the other five NVBP insulin groups has all changed significantly in level or trend. Procurement expenditure levels for basic human insulin analogues, mealtime insulin and mealtime insulin analogues all declined, while the downward pattern of procurement expenditure for basic human insulin and premixed human insulin significantly decelerated. Mealtime insulin procurement expenditure reversed direction, shifting from a pre-policy decline to a post-policy increase (table 2, figure 3, online supplemental table S2).

Figure 3. Expenditure changes for the national volume-based procurement insulin group.

Changes in the price, volume and expenditure of individual insulin

Following NVBP implementation, the prices of all 15 selected insulins decreased significantly (p<0.05). 10 insulin products, including human insulin, insulin aspart 30 and 50, insulin detemir, insulin glargine, insulin lispro, mixed protamine human insulin (30R and 50R), mixed protamine zinc recombinant human insulin lispro (50R) and protamine human insulin, showed significant increases in volume. Seven insulin products, including human insulin, insulin aspart 30, insulin glargine, insulin lispro, mixed protamine human insulin (30R), mixed protamine zinc recombinant human insulin lispro (50R) and protamine human insulin, exhibited significant upward usage trends. Procurement expenditure levels for 10 insulin products, including human insulin, declined significantly. However, only human insulin, mixed protamine human insulin (30r), and protamine human insulin demonstrated significant upward expenditure trends during the policy period. Notably, mixed protamine zinc recombinant human insulin lispro (25R) exhibited no significant changes in volume or expenditure levels or trends, except for its price (online supplemental tables S3–S17 and figures S2–S16).

Discussion

Principal findings

The results indicate that the price for all NVBP bid-winning insulins declined sharply in the quarter when the NVBP policy was implemented and continued to decline thereafter. Total DDDs (volume) reversed their prior downward trend and increased markedly, while overall expenditure declined immediately and then gradually stabilised. NVBP group-level analysis revealed significant level declines in price across all six insulin groups, with basic human insulin analogues, mealtime insulin analogues and premixed human insulin analogues exhibiting a steeper subsequent decline. Most insulin groups experienced significant volume increases, although some insulin analogues lagged, and expenditure shifted across groups. At the individual level, all 15 insulins experienced a significant price reduction level. Additionally, all other insulins, except for mixed protamine zinc recombinant human insulin lispro (25R), were observed to have significant changes in volume or expenditure, either in level or in trend. The total volume of insulin procured under the NVBP increased by 16.73%, whereas the study predicted that the prevalence of diabetes in China would increase by only 1.5% from 2020 to 2030. The increase in insulin volume was substantially higher than the projected growth in the diabetic patient population. Therefore, the NVBP policy had a significant impact on the volume of insulin.

In this study, a significant decrease of 51.61% was observed in the price of all NVBP bid-winning insulin, consistent with previous findings reporting a 42.08% price reduction for insulin under the NVBP policy. Both studies demonstrate that insulin analogues (basic human insulin analogues, mealtime insulin analogues and premixed human insulin analogues) experienced a more considerable decline in price.¹⁰ Additionally, our results are consistent with the outcomes of previous studies in terms of expenditure, as both indicate a reduction in insulin-related expenditure during the first year of policy implementation, resulting in significant cost savings for patients. Most importantly, compared with previous studies, our study further investigated the impact of the NVBP through a more in-depth analysis of various aspects. A study by Yuan et al merely evaluated the level changes in the year of policy implementation, whereas our study adopted an ITSA to provide a more comprehensive assessment of the policy’s impact. This approach allows us to capture both the immediate effects and long-term trends and changes in insulin prices, volumes, and expenditure following the NVBP policy implementation. Moreover, our study examined changes in insulin use at the NVBP group and individual levels, aiming to analyse the specific effects of the policy on all six groups and 15 individual insulin products.

The more substantial price decreases in basic human insulin analogues, mealtime insulin analogues and premixed human insulin analogues at the group level or individual levels may be partially explained by differences in baseline price levels. Prior to centralised procurement, the overall prices of insulin analogues (basic human insulin analogues, mealtime insulin analogues and premixed human insulin analogues) were substantially higher than those of basic human insulin, mealtime insulin and premixed human insulin, since their higher market costs of active pharmaceutical ingredients and the limited availability of biosimilars or other competitors in China before 2021 conferred strong market power to the originator pharmaceutical company.^{27 28} Such differences in initial price may have contributed to the greater price reductions observed for insulin analogues (basic human insulin analogues, mealtime insulin analogues and premixed human insulin analogues) compared with basic human insulin, mealtime insulin and premixed human insulin during the subsequent procurement period, as the higher price elasticity of the former made substantial markdowns more feasible to secure larger volumes under the NVBP policy.

The observed reduction in procurement volume of insulin aspart 50 may be related to redistribution mechanisms within procurement groups under the NVBP policy. Differing from the NVBP policy for chemical drugs, this insulin-specific policy categorises insulins with different generic names but similar therapeutic effects to compete within the same category. Within each procurement group, procurement volumes are partially reallocated based on the ranking of their bidding prices, with a proportion of volumes from lower-ranked categories (C and D) reassigned to higher-ranked categories (A and B). Under such arrangements, products with higher bidding prices may receive a smaller share of the allocated volume, while higher-ranked products may gain additional volume. As insulin aspart 50 had relatively higher bidding prices compared with other products in its procurement group, it may have been subject to such volume reallocation, potentially contributing to its post-policy volume decline.

The increasing economic burden of insulin treatments has constantly raised a worldwide concern among patients, prescribers, payers and policymakers. A survey across 43 countries found that insulin prices were generally high, and treatment was unaffordable for individuals with low incomes. Evidence from multiple low- and middle-income settings, including countries in Africa, Asia and Latin America, has shown that insulin remains insufficiently available in public-sector hospitals and may require several days’ wages for purchase, indicating substantial financial barriers to its use.²⁹ In response, countries have implemented a range of policy measures to improve affordability. These efforts have centred on pooled procurement and price negotiations to leverage buyer power and secure insulin price reductions. These are often complemented by tiered pricing strategies and public coverage mechanisms (eg, universal health coverage) aimed at lowering out-of-pocket costs. In addition, some settings have strengthened supply chains and adopted integrated delivery models to enhance both the availability and affordability of insulin.^{30 31} Following a series of healthcare reforms aimed at reducing prices in China,^{15 32} the NVBP policy exclusively for insulin was initiated in 2021. Our findings suggest that the NVBP policy has been effective in reducing insulin procurement prices while increasing overall procurement volumes. The observed reduction in total expenditure reflects improved efficiency in pharmaceutical purchasing, and such cost savings could provide fiscal space for national payers to redistribute financial resources to other high-value therapeutic areas.¹⁰ However, the impacts of such pricing policies on pharmaceutical innovation remain uncertain. On the one hand, firms may respond by pursuing cost advantages through process innovation or economies of scale; on the other hand, reduced profit margins could constrain financial resources available for research and development. Future studies are needed to examine the longer-term effects of centralised procurement on pharmaceutical innovation.

Limitations

This study has several limitations related to its data sources and study design. First, the study is based on hospital procurement data at the aggregated level and thus cannot evaluate patient access with regard to the numbers of patients treated or affordability based on out-of-pocket spending. Second, the ITS was employed due to significant data gaps for non-winning insulin. Even considered as a robust quasi-experimental design for policy evaluation, ITS is still incapable of fully attributing observed changes in insulin prices, volumes and expenditure to the NVBP policy, as other confounding factors or a lack of controls may influence the results. For instance, the increase in the number of diabetes patients may affect the dosage of insulin. In addition, our study relied on quarterly data to evaluate the effect of the target policy, which may obscure short-term fluctuations of the policy. Therefore, monthly data is preferred for future studies, as it can capture detailed changes in trend and level, offer greater statistical power and allow for a more detailed examination of the policy’s impact. Finally, future studies with subnational data could provide more insights into heterogeneous policy effects across regions.

Conclusion

Our results suggest that China’s national volume-based procurement policy exclusively for insulins significantly changed expenditure on the use of selected insulins. The increase in procurement volume, alongside the decreased total expenditure on overall insulin, suggests that the NVBP policy has the potential to enhance patients’ utilisation and improve the affordability of insulin.

Data availability statement

Data are available upon reasonable request.