1 Background
Medication errors are crucial causes of potential avoidable disability or deaths in hospital settings. According to World Health Organization (WHO), there were approximately one-tenth of patients experienced harmful medical treatment, and more than 3 million patients dead due to inappropriate medication treatments.
1 Medication errors can cause direct economic losses of 42 billion and indirect healthcare costs of 140 billion annually.
2 In China, in 2023, there were nearly 30,000 cases of medication errors reported by only 439 medical institutions in 26 provinces, highlighting the crucial safety problems in drug utilization practices.
3Several evidence suggested that intelligent pharmacy technologies show great potential in improving work efficiency,
4 enhancing medication safety,
5 and reducing risk of manual error.
6 Intelligent pharmacy typically refers to the integration of advanced digital technologies, such as artificial intelligence (AI), big data, cloud computing, the internet of things (IoT), and automated devices, into hospital pharmacy practice.
7–9 It is valuable in optimizing the drug utilization workflows and providing opportunities for clinical pharmacists to transform from manual, repetitive tasks toward more clinical activities closely related to patient health.
10The development of intelligent pharmacy requires continuous investment in advanced products, including automated and information systems.
11 Such investments must be strategically allocated based on comprehensive assessments of hospital-specific conditions, including current infrastructure maturity and operational needs. Currently, there are few studies conducting large-scale investigations on the overall picture of the implementation of intelligent pharmaceutical products in Chinese hospitals. Therefore, we conducted a nationwide research aiming to explore the current status of implementation of intelligent pharmacy in China and provide evidence for the balanced, sustainable and healthy development of intelligent pharmacy.
2 Methods
2.1 Study design
Our study was a cross-sectional survey of hospital pharmacy departments across China from May 29 to June 10, 2025. We used convenience sampling to recruit pharmacy directors from hospitals in 31 provincial administrative regions (except of Macao, Hong Kong, and Taiwan) of China.
12 We developed the questionnaire via an online survey platform in China and distributed the questionnaire link via social platform (Wechat) or email to pharmacy directors recruited.
2.2 Questionnaire
The questionnaire was developed following multiple stages. (1) policy file collection and review: (2) expert discussion; (3) pilot research. We searched on the official websites of national and local administrative departments for policy files related to the guidance, development plans, specific development actions, management measures, or evaluation measures for intelligent pharmacy in hospitals. The websites searched for policy files included the national and local Health Commission, Administration of Traditional Chinese Medicine, Health Security Administration, and Medical Product Administration. We used the following search terms: Intelligent pharmacy, automated pharmacy and intelligent pharmaceutical science. We also used combinations of these terms. The date of the last search was September 10, 2025.
The drafted items went through two rounds of delphi methods with senior hospital pharmacists and artificial intelligence specialists. After that, an investigation table was generated and a pilot test was then conducted in 10 hospitals to test question clarity and estimate completion time. Based on pilot feedback, the questionaire is appropriate and the estimated completion time was around 20–45 min. The final questionnaire comprised 48 questions covering three domains: (1) hospital characteristics (level, type, ownership, location, number of beds, pharmacy staff size); (2) current utilization of intelligent pharmacy products (IPDs); and (3) Supplement the IPDs not mentioned in the questionnaire and describe their usage. The whole investigation questionnaire is provided in the supplementary material.
2.3 Data collection
The online platform automatically recorded timestamps and prevented multiple submissions from the same account. After survey closure, data were exported to Microsoft Excel for cleaning and coding. We checked for missing values, logical inconsistencies, and outliers; records failing quality checks were excluded as described above. Key derived variables (region, hospital bed category, pharmacy staffing category) were generated prior to analysis. Exclusion criteria: (1) implausible or extreme completion times based on pilot estimates (responses completed in < 10 min or > 60 min); (2) missing key data (e.g., hospital bed count or region); (3) duplicate responses from the same hospital; (4) internally inconsistent or clearly erroneous data (for example, classification mismatches such as a primary hospital reporting ≥ 1,000 beds). Duplicate and inconsistent responses were resolved by checking reported hospital identifiers and contacting respondents for verification.
2.4 Statistical analysis
Hospitals were classified by number of beds into three categories: ≥ 1,000, 500–999, and < 500. The number of pharmacists per 100 beds (PPB) were calculated by formula as follows: PPB = (N_pharmacists/N_beds) × 100. Where N_pharmacists is the number of full - time pharmacists in the hospital pharmacy department in 2024, and N_beds is the number of open beds during the same period. The comparisons of the rates were performed using Pearson's chi-squared test. Statistical significance was set at two-sided p < 0.05. Confidence intervals were calculated for key proportions where appropriate. Statistical analyses were performed using SPSS version 26.0. Data visualization was produced in R (version 4.4.3) using the ggplot2, hchinamap, and fmsb packages.
3 Results
In total, 1,653 questionnaires were collected from 1,653 hospitals across 31 provincial administrative regions in China. After excluding 538 questionnaires, 1,126 questionnaires were finally included for analysis. The basic characteristics of the hospitals were shown in Table 1. Approximately, 54.88% of the hospitals were tertiary hospitals, 83.13% were general hospitals, and 93.25% were public hospitals.
3.1 Identifications and classifications of IPDs
Through literature research and Delphi method, we identified that 36 kinds of IPDs are currently utilized in China. The IPDs covered six stages of pharmaceutical practices in hospitals: intelligent drug supply (IDS), intelligent prescription review (IPR), intelligent drug preparation (IDP), intelligent drug dispensing (IDD), intelligent medication consultation (IMC), and intelligent medication monitoring (IMM). The names and classifications of the IPDs were shown in Fig. 1. The explanations of the IPDs were shown in Supplementary Table.
3.2 Utilization rates of IPDs
Among the 1,126 hospitals, there were 745 (66.16%), 935 (83.04%), 698 (61.99%), 982 (87.21%), 500 (44.40%), and 232 (20.60%) hospitals were equipped with IDS, IPR, IDP, IDD, IMC and IMM systems respectively (Table 1). The most used IPDs were the barcode management system (81.26%), antibiotic monitoring and management system (67.7%), and drug inventory system (52.4%) (Table 2). The least used were patient medication compliance monitoring equipment (2.8%), AI pharmacist intelligent agent (3.0%) and TDM monitoring result automatic interpretation system (3.6%) (Table 2).
Among the IPDs for IDS, the pharmaceutical inventory system (PIS) was the most used (52.4%), while the supplier management and evaluation system (SMES) was the least used (7.9%) (Table 2, Fig. 2). The most used IPD for IPR was the antibiotic monitoring and management system (AMMS) (67.7%), while the pharmaceutical risk alert system (PRAS) (13.9%) was the least used. Among the IDP systems, the automatic decoction and concentration machine (ADCM) (39.3%) had the highest utilization rate, while the automatic intravenous drug preparation machine (AIDPM) had the lowest utilization rate (5.8%). Among the IPDs for IDD, the barcode medication administration system (BCMA) had the highest utilization rate (81.26%), while the automatic injection dispensing machine (AIDM) had the lowest (3.6%). In the IMC stage, the Internet patient consultation system (IPCS) (26.3%) had the highest usage rate while the AI pharmacist agent (AIPA) had the lowest (3.0%). Among the IPDs for drug monitoring, the adverse drug reaction reporting system (ADRRS) had the highest utilization rate (11.9%) while the medication adherence monitoring equipment had the lowest (2.8%).
3.3 Discrepancies in utilization of IPDs
After classified by workflow stages, the utilization rates of IPD for all workflow stages in tertiary hospitals were significantly higher than those in secondary and primary hospitals (Table 1). Moreover, the utilization rates of hospitals with ≥ 1000 beds were significantly higher than those of hospitals with 500–1000 beds, and higher than those of hospitals with < 500 beds. The usage rates of IPR, IMC, and IMM systems in hospitals with ≥ 7 PPBs were significantly lower than those in hospitals with 5–7 and ≤ 5 PPBs (Table 1).
Specifically regarding the products, the usage rates of almost all IPDs were significantly related with the level of the hospitals except for the PIS and PEPMS. The usage rates of IPDs in tertiary hospitals were significantly higher than those in secondary or primary hospitals (Table 2).
Fig. 2 illustrated the differences in implementation patterns of IPDs in hospitals with different number of beds. The usage patterns were basically the same for IDS, IPR, IMC, and IMM systems. In the IDP stage, the gap in the ADCM usage rate among hospitals of different sizes is relatively small, but there are significant differences in APM. APM has the highest usage rate in hospitals with ≥ 1000 beds, while it is a weak item with a low usage rate in hospitals with < 500 beds. In the IDD stage, ADM is a relatively popular product in hospitals with ≥ 1000 beds, with a usage rate of over half, but its usage rate is less than a quarter in hospitals with < 500 beds. In contrast, the promotion rate of BCMA is relatively similar among hospitals of different sizes.
In terms of the regional distribution, we observed higher rates of implementation of IPDs for drug supply in the southeast regions of China, especially in Zhejiang (93.94%) (Table 3). On the contrary, the implementation rates in the central region were relatively low, especially in Guizhou (41.67%). The implementation of IPR systems were well carried out in most parts of the country, with the highest in Shandong (100%) and slight deficiencies in some parts of the southwest and northeast like Xizang Autonomous Region (60%). The provinces with highest rates for implementation of IDP systems were Zhejiang (90.91%) and Chongqing (89.47%), while the provinces with poor construction include Xizang Autonomous Region (20%) and Guizhou (29.17%). The areas with higher implementation rates of IDD systems were mainly located in the eastern coastal areas, and the best were Tianjin, Jiangsu and Guangdong (100%). Similarly, the implementation situation of IMC systems in the eastern coastal areas were better, like Shandong (80.77%) and Zhejiang (78.79%) while the construction situation in the southwest and northeast regions was relatively poor. In terms of IMM systems, the implementation rates in the southeastern areas is still a little better, and the implementation rate in Hunan is the best (50%), while the construction situation in the southwest and northeast regions is poor. Overall, the development of intelligent pharmacy is relatively better in the southeast of China than in the northwest and northeast.
4 Discussion
This nationwide survey including 1,126 hospitals from 31 provincial administration regions across China investigated the current status of the implementation of intelligent pharmacy technologies in hospitals. Our results highlight an uneven development pattern of intelligent pharmacy in China, suggesting that regional resources, institutional capacities, and workforce distribution may play critical roles in shaping implementation of the intelligent pharmacy technologies. Importantly, this study provides empirical evidence to inform policymakers in designing targeted strategies to promote more balanced and equitable development of intelligent pharmacy in the future.
The development of intelligent pharmacy was uneven and the level of the hospital and the number of beds are important factors affecting the implementation of intelligent pharmacy. The utilization rates of IPDs in tertiary hospitals were significantly higher than those in secondary and primary hospitals, because tertiary hospitals typically had more financial investment, scientific research funding, information investment and technology procurement capabilities to purchase complex intelligent pharmacy systems. Similarly, hospitals with ≥ 1000 beds exhibited significantly higher implementation rates than those with fewer beds, particularly for complex systems such as APM and ADM. Interestingly, however, the usage rates of IPR, IMC, and IMM systems were relatively lower in hospitals with a larger number of pharmacists (≥ 7 PPBs). These systems were all for medication medication processes which need more professional knowledge. The hospitals with more pharmacists may have developed a habitual work pattern, thus reducing their acceptance of innovative technologies. On the contrary, hospitals with lesser pharmacists have a heavier workload, and pharmacists lack the time and energy to engage in patient health-related work. Therefore, they are more eager to use technological innovation to streamline work processes.
Hospitals in the eastern coastal regions, particularly Zhejiang, Jiangsu, and Guangdong, demonstrated markedly higher adoption rates in multiple types of IPDs. The development of intelligent pharmacy is closely related to the economic development.
13 Disparities in the advancement of intelligent pharmacy across regions can often be attributed to varying economic conditions.
14 Additionally, policy initiatives play a crucial role in driving the disparities of intelligent pharmacy implementation. For instance, provinces including Jiangsu, Zhejiang, Shanghai, and Anhui have introduced an evaluation criteria for intelligent pharmacy systems.
15 This focus underscores their commitment to advancing intelligent pharmacy infrastructure, thereby accelerating its development in these regions.
Notably, the BCMA system within IDS systems was the most commonly used product (81.26%), followed closely behind by AMMS with a high implementation rate of 67.7%. These phenomena can be primarily attributed to recent national policies. The National Health Commission of China encouraged that hospitals of all levels should improve the clinical management of antimicrobial agents and promote rational utilization of antibiotics through systematic monitoring. Moreover, the National Healthcare Security Administration the nationwide regulatory application of drug traceability codes started in March 2025 to ensure the safety and authority of the drugs, and mandates that all medical institutions fully collect and upload drug traceability codes by January 1, 2026.
16–18 The results of this survey, concluded by June 2025, demonstrate that BCMA systems have already been widely implemented in most Chinese hospitals, aligning with these national directives.
In comparison with the most recent survey by the American Society of Health-System Pharmacists (ASHP) in 2023,
19 this study reveals distinct national patterns in the implementation of intelligent pharmacy technologies. In China, more than 60% of large hospitals equipped with ADM in central pharmacies, while the reported utilization rate of ADC in the wards was only 5.5%. By contrast, the ASHP 2023 survey showed that 86.1% of U.S. hospitals employed ADC in clinical units, while only 2.0% relied primarily on centralized ADM. Chinese hospitals face extremely high outpatient volumes and central pharmacy dispensing pressure,
20 which makes ADM a strategic choice for improving bulk dispensing efficiency. In the U.S., however, the hospital system places greater emphasis on point-of-care safety and timeliness for inpatients, leading to widespread adoption of ADC. This demonstrates that the integration of intelligent pharmacy equipment is not merely a matter of technology transfer but rather a context-specific decision influenced by workflow demands, cost-effectiveness, and patient service models.
With regard to other technologies, the adoption of AIDPM was relatively similar (5.8% in China vs. 3.7% in the U.S.). Moreover, BCMA has already become routine in both countries (81.3% in China vs. 73.6% in the U.S.), reflecting a shared recognition of its value for medication safety. Taken together, these findings suggest that while certain technologies demonstrate global consensus, others evolve along divergent pathways shaped by national healthcare environments. Future development of intelligent pharmacy should therefore consider not only technological advancement but also alignment with local clinical needs, patient care models, and system-level resource allocation.
There are several limitations that cannot be ignored. First, the present cross-sectional design only reflects the status of intelligent pharmacy technology utilization at one time point instead of temporal trends or causal relationships. Future researches can use longitudinal or repeated survey designs to track dynamic trends and policy impacts. Second, convenience sampling was adopted to recruit respondents, primarily through professional academic WeChat groups. This approach may limit the representativeness of the findings. However, the sample size of this study was large (> 1000) and the characteristics of the samples were clearly illustrated suggesting that the findings retain substantial reference value.
21 Finally, although the questionnaire was developed through two rounds of Delphi methods and a pilot testing, it may not cover all the types of intelligent pharmacy technologies. Future researches regarding of International experiences may be needed to further inform strategies to optimize intelligent pharmacy development in hospitals.
5 Conclusion
This study reveals the national patterns of the implementation of intelligent pharmacy technologies in China, reflecting an uneven landscape with clear variations by region, hospital levels, size and pharmacist numbers. BCMA and ADM systems have already been widely implemented in most Chinese hospitals while little hospitals have implemented ADC systems in wards. the integration of intelligent pharmacy equipment is not merely a matter of technology transfer but rather a context-specific decision influenced by workflow demands, and patient service models.
2026 The Authors. Publishing services by Elsevier B.V. on behalf of Higher Education Press and KeAi Communications Co. Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).