Inpatient antibacterial use trends and patterns, China, 2013–2021

Abstract Objective To analyse trends and patterns in inpatient antibacterial use in China’s tertiary and secondary hospitals between 2013 and 2021. Methods The analysis involved quarterly data from hospitals covered by China’s Center for Antibacterial Surveillance. We obtained information on hospital characteristics (e.g. province, a de-identified hospital code, hospital level and inpatient days) and antibacterial characteristics (e.g. generic name, drug classification, dosage, administration route and usage volume). We quantified antibacterial use as the number of daily defined doses per 100 patient-days. The analysis took into account the World Health Organization’s (WHO’s) Access, Watch, Reserve classification of antibiotics. Findings Between 2013 and 2021, overall antibacterial use in inpatients decreased significantly from 48.8 to 38.0 daily defined doses per 100 patient-days (P < 0.001). In 2021, the variation between provinces was almost twofold: 29.1 daily defined doses per 100 patient-days in Qinghai versus 55.3 in Tibet. The most-used antibacterials in both tertiary and secondary hospitals throughout the study period were third-generation cephalosporins, which comprised around one third of total antibacterial use. Carbapenems entered the list of most-used antibacterial classifications in 2015. The most frequently used antibacterials in WHO’s classification belonged to the Watch group: usage increased significantly from 61.3% (29.9/48.8) in 2013 to 64.1% (24.4/38.0) in 2021 (P < 0.001). Conclusion Antibacterial use in inpatients decreased significantly during the study period. However, the rising proportion of last-resort antibacterials used is concerning, as is the large gap between the proportion of antibacterials used belonging to the Access group and WHO’s global target of no less than 60%.


Introduction
The use of antibacterial drugs is a key driver of antibiotic resistance. 1 China is one of the largest producers and consumers of antibacterials worldwide and, in recent decades, its government has unveiled a series of antibacterial stewardship regulations and clinical guidelines for appropriate antibacterial use. 2 Although these have had a profound effect, inappropriate antibacterial use and antibacterial resistance remain challenges for the country.
The World Health Organization (WHO) regards surveillance of antibacterial use as a cornerstone of antimicrobial stewardship but, until recently, surveillance has often been ignored and underresourced. 3 In 2005, the former Chinese health ministry established the Center for Antibacterial Surveillance to strengthen the clinical supervision and management of antibacterial drugs. Initially the surveillance network included only 109 hospitals. By 2021, it covered 2694 tertiary hospitals and 4100 secondary hospitals, which accounted for 90% and 40% of all these hospitals, respectively, in mainland China. [4][5][6] Most previous studies of antibacterial use in China reported regional data, [7][8][9][10][11] whereas national studies tended to use procurement data, [12][13][14] which may have been subject to bias given the discrepancy between procurement and usage. Since 2016, the Chinese National Health Commission has issued annual reports of antibacterial use based on data from the Center for Antibacterial Surveillance. However, these data were limited by small sample sizes and covered only 191 tertiary general hospitals. Moreover, the indicators used to evaluate antibacterial use were relatively limited: for example, inpatient indicators included only the total volume and frequency of antibacterial use without more detailed information on the pattern of use.
The aims of this study were to produce up-to-date data on the trends and patterns of inpatient antibacterial use in China's hospitals, to highlight the challenge of inappropriate antibacterial use, and to provide a benchmark for promoting appropriate antibacterial use and curbing antibacterial resistance.

Methods
Our study involved aggregated, quarterly data on antibacterial drug use by inpatients at hospitals covered by the Center for Antibacterial Surveillance between 1 January 2013 and 31 December 2021. Data in the nationwide surveillance database covered 31 provinces in China (excluding the Hong Kong Special Administrative Region, the Macao Special Administrative Region and Taiwan, China), as detailed in China's Health Statistics Yearbook. 15 All secondary and tertiary hospitals that reported antibacterial use during the study period were included; the number of hospitals varied from 1630 in 2013 to 3880 in 2021 (details are available in the online repository). 16 Participating hospitals uploaded quarterly data using a website-based system.
The data covered antibacterial drugs intended for systemic use, as defined by WHO's Anatomical Therapeutic and Antibacterial use trends, China Haishaerjiang Wushouer et al.
Chemical (ATC) classification J01. 17 Information on each hospital included: (i) the province; (ii) a de-identified hospital code; (iii) the hospital level; and (iv) the number of inpatient days. Information on each antibacterial drug included: (i) the generic name; (ii) the drug classification; (iii) the specified dosage and form of administration; and (iv) the usage volume (i.e. the number of doses administered per quarter). Antibacterial use was expressed as the number of defined daily doses per 100 patient-days. 17 For antibacterials without an ATC code or whose defined daily dose was not specified in WHO's ATC classification, the defined daily dose was imputed from the manufacturers' recommended daily dose, as approved by China's Food and Drug Administration. 13 Antibacterials were further disaggregated at the pharmacological (i.e. ATC level 3), chemical (i.e. ATC level 4) and chemical substance levels (i.e. ATC level 5).

Data analysis
Trends in, and patterns of, antibacterial use were analysed by ATC level-3 subgroup. In addition, total antibacterial use was also analysed by province and hospital level. The proportion of all antibiotics administered through different routes (i.e. parenteral or oral) and the proportion that belonged to subsections of WHO's Access, Watch, Reserve (AWaRe) classification were also determined. 18 The AWaRe classification reflects an antibiotic's strength and its potential impact on antimicrobial resistance. Antibiotics in the Access group are first-or second-line treatments for common infections and should be widely accessible. Antibiotics in the Watch group should be used only for a limited group of well-defined syndromes and should be closely monitored. Antibiotics in the Reserve group are primarily used as a last resort to treat infections caused by multi-or extensively-drug resistant bacteria. A fourth group -antibiotics not recommended by the AWaRe classification -comprised mainly inappropriate antibiotic combinations with a negative impact on antimicrobial resistance and patient safety. We also added the fifth group -not included antibiotics -that comprised antibiotics which were not in WHO's AWaRe classification but which were used in China (available in the online repository). 16 In addition, we identified the most frequently used antibacterial pharmacological classifications (i.e. ATC level 4) that together accounted for 90% of annual antibacterial use and the top 10 individual antibacterial drugs (i.e. ATC level 5) used in each year.
To derive a comparable metric of antibacterial use across time for different types of hospital, province and drug classification, we calculated a compound annual growth rate (CAGR) for antibacterial use: where C 2021 was the total inpatient antibacterial use in 2021 and C 2013 was the total inpatient antibacterial use in 2013. The percentage usage of different types and classifications of antibacterials (i.e. ATC level 3 and 4, respectively) and of individual antibacterials (i.e. ATC level 5) in any year was calculated by dividing the usage volume for each type, classification or individual antibacterial, as appropriate, by the total usage volume of all antibacterials in that year.

Statistical analysis
We report descriptive results. Trends in antibacterial use over time were assessed using linear regression, with antibacterial use as the dependent variable and time as the independent variable. 19 A Pvalue less than 0.05 was considered statistically significant. Data were managed and analysed in Excel 2019 (Microsoft Corporation, Redmond, United States of America), Stata v. 14.0 (StataCorp LLC, College Station, USA) and Orig-inPro 2020b (OriginLab Corporation, Northampton, USA).

National antibacterial use
National antibacterial use by inpatients in all study hospitals decreased continuously from 48.8 to 38.0 defined daily doses per 100 patient-days between 2013 and 2021 (Table 1 and online repository), 16 with a CAGR of −3.1% (P < 0.001). During the study period, significant decreasing trends were observed in both tertiary (from 47.9 to 37.1 defined daily doses per 100 patient-days; P-value: < 0.001) and secondary hospitals (from 54.9 to 41.3 defined daily doses per 100 patient-days; P-value: < 0.001), with the CAGR being larger in secondary than tertiary hospitals: −3.5% versus −3.1%, respectively. Antibacterial use was greater in secondary than tertiary hospitals throughout the study period, except in 2020. Of all antibacterial classifications at ATC level 3, only the J01A classification (tetracyclines) significantly increased in use, from 0.3 to 0.5 defined daily doses per 100 patient-days (CAGR: +5.4%; P-value: < 0.001).

Provincial antibacterial use
Antibacterial use was highest in the province of Tibet (55.3 defined daily doses per 100 patient-days), which was 1.9-times greater than in Qinghai, where use was lowest (29.1 defined daily doses per 100 patient-days). There was a downward trend in antibacterial use in 29 of 30 provinces (Tibet was excluded because only 5 years of data were available). However, there was a considerable variation in the trend between provinces: the largest decreasing CAGR was for Shanghai (−5.5%), whereas uniquely there was an increasing CAGR for Hebei (+0.7%). The inpatients antibacterial use in tertiary and secondary hospitals in 31 provinces in 2021 and the corresponding CAGRs between 2013 and 2021 are available in online repository. 16 Parenteral and oral antibacterial use Table 2 reports parenteral and oral antibacterial use by inpatients in tertiary and secondary hospitals and in all hospitals combined between 2013 and 2021. During the study period, the proportion of antibacterials administered parenterally increased significantly from 79.7% to 87.4% (CAGR for the proportion: +1.2%; P-value: < 0.001), while the proportion administered orally decreased significantly from 20.3% to 12.6% (CAGR: −5.8%; P-value: < 0.001). The proportion administered parenterally was higher in secondary than tertiary hospitals in every year. Table 3 reports the use of antibacterials in different AWaRe categories by inpatients in tertiary and secondary hospitals and in all hospitals combined between 2013 and 2021. During the study period, the proportion of antibacterials used that belonged to the Access group fluctuated between 15.1% and 18.4% (P-value: 0.099). Overall, most antibacterials used belonged to the Watch group and the proportion that belonged to   (. . .continued) this group increased significantly from 61.3 in 2013 to 64.1% in 2021 (CAGR: +0.6%; P-value: < 0.001). In addition, the proportion of antibacterials used that were not recommended increased significantly from 10.6% in 2013 to 11.8% in 2021 (CAGR: +1.3%; P-value: < 0.05). The proportion of antibacterials used that were not in the AWaRe classification fell from 8.6% in 2013 to 4.9% in 2021 (CAGR: −6.9%; P-value: < 0.001). The patterns of antibacterial use according to AWaRe classification were similar in secondary and tertiary hospitals, though the proportion of Access group antibacterials used was higher in secondary hospitals: 18.6 to 23.7% compared with 14.3 to 17.4% in tertiary hospitals.

Drugs accounting for 90% of use
The most frequently used antibacterial classifications (ATC level 4), which together accounted for 90% of use in each study year by inpatients in tertiary and secondary hospitals, are shown in Table 4. In 2021: (i) 30.9% of antibacterials used were third-generation cephalosporins (i.e. classification J01DD); (ii) 15.1% were second-generation cephalosporins (J01DC); (iii) 12.5% were fluoroquinolones (J01MA); (iv) 11.6% were penicillins (J01CR); (v) 6.0% were first-generation cephalosporins (J01DB); (vi) 4.2% were imidazole derivatives (J01XD); (vii) 3.8% were macrolides (J01FA); (viii) 3.6% were carbapenems (J01DH); and (ix) 2.9% were penicillins with extended spectrum (J01CA). Between 2015 and 2019, carbapenems (J01DH) replaced other aminoglycosides (J01GB). Neither other aminoglycosides (J01GB) nor carbapenems (J01DH) appeared on the list for secondary hospitals, in contrast to tertiary hospitals.  Table 3.              (Table 1 and the online repository). 16 Although the overall use of parenteral antibacterials decreased to 33.2 defined daily doses per 100 patient-days in 2021, the proportion of antibacterial use that was parenteral increased to 87.4% ( Table 2). The proportion of antibacterials that belonged to the Access group started to increase in 2020, to 16.1%, and rose to 17.8% in 2021, whereas the proportion of drugs that were not recommended decreased from 13.3% in 2020 to 11.8% in 2021. The pattern was the same in secondary and tertiary hospitals. There was no change in the antibacterial classifications that comprised 90% of total antibacterial use during the COVID-19 period. However, ceftizoxime and amoxicillin/clavulanic acid first entered the top 10 in tertiary hospitals in 2020 and 2021, respectively (Table 5).

Discussion
During the study period, inpatient antibacterial use decreased significantly in tertiary and secondary hospitals in China and was lower than has been reported in both low-and middle-income countries, such as Ethiopia (81.6 defined daily doses per 100 patient-days), 20 Eritrea (79.5 defined daily doses per 100 patient-days) and Nepal (113.7 defined daily doses per 100 patient-days), 21,22 and high-income countries, such as Australia (93.6 defined daily doses per 100 patient-days), 23 Israel (84.0 defined daily doses per 100 patient-days), 24 the Netherlands (85.7 defined daily doses per 100 patient-days), 25 Saudi Arabia (94.5 defined daily doses per 100 patient-days) and Sweden (65.7 defined daily doses per 100 patient-days). 26,27 This finding suggests that the antibacterial stewardship system probably reduced antibacterial overuse in China. 28 In 2018, the Chinese government stated that antibacterial stewardship should gradually move from being administratively led to a comprehensive, multidisciplinary collaboration and that it should focus on high-risk groups, such as paediatric patients, pregnant women and elderly people. 29 Our study found that interprovincial variability in antibacterial use was substantial, which means that surveillance at the provincial level is important for identifying the determinants of antibacterial use, for assessing the prevalence of antibacterial-resistant pathogens and for guiding appropriate interventions. Moreover, the potential impact of interprovincial variability should be tackled more comprehensively under the OneHealth approach to ensure the risk of antibacterial resistance is minimized. 30 Our study found that the proportion of antibacterials used by inpatients that belonged to the Access group in the AWaRe classification was under 20%. Although the Chinese government has promoted the appropriate use of narrow-spectrum antibacterials and has strived to increase their proportional use to international levels, 29 the proportion is still far below the target set by WHO (i.e. Access group antibacterials should account for no less than 60% of total antibacterial use) and is lower than in most countries that reported this figure. 18,31 In China, antibacterial formulary restriction lists were established at the provincial level in 2012. 5 Antibacterials were divided into three groups according to the prevalence of resistance, safety, drug efficacy and cost: (i) non-restricted; (ii) restricted; and (iii) highly-restricted. 5 The Chinese classification categorized many drugs in the AWaRe classification's Watch group (e.g. levofloxacin) as non-restricted, which may have contributed to the greater proportionate use of broad-spectrum antibacterials we found. We also found that ornidazole and cefathiamidine, which were not included in the AWaRe classification, were two of the top 10 antibacterials used in China. This finding highlights the importance of ensuring that the categorization of antibacterials is evidence-based and that antibacterials are used appropriately.
Although inpatient antibacterial use decreased more in secondary than tertiary hospitals, use per patient-day was consistently higher in secondary hospitals. This finding is concerning given that secondary hospitals had fewer inpatients with less severe disease than tertiary hospitals, according to China's Health Statistics Yearbook. 15 Moreover, the proportion of antibacterials administered parenterally was much higher than the proportion administered orally in both secondary and tertiary hospitals and was increasing. The proportion was higher than that found in previous studies based on procurement data, which included outpatients and primary care. 13,32 The higher proportion of parenteral administration among inpatients may have been due to the hierarchical nature of the health-care delivery system: 33 patients with mild or moderate disease may have been encouraged to seek treatment in primary care, with more severe patients attending higher-level hospitals. 34 In addition, the prohibition of drug infusion for outpatients (except for children's hospitals) was gradually introduced from 2016 onwards, 35 which may have further increased parenteral antibacterial use.
Cephalosporins, especially secondand third-generation cephalosporins, were consistently the most-used antibacterials in China. In Europe, New Zealand, Switzerland and the United Kingdom, in contrast, penicillins have been the most-used antibacterials. 19,31,[36][37][38] The difference between China and other countries might be explained by the setting: in China, most antibacterials are given to hospital inpatients, whereas in other countries most are given to ambulatory patients and to those treated in general practice or in the community. Moreover, cephalosporins are recommended by Chinese national guidelines for most perioperative prophylaxis indications, which may have increased their use. 39 We found that levofloxacin was the most-used antibacterial between 2013 and 2020. Ease of administration may have contributed to levofloxacin's use.
Penicillins, in contrast, require timeconsuming skin allergy testing before administration. Carbapenems (ATC classification: J01DH) became one of the most frequently used antibacterial drug classes between 2015 and 2021 despite being classified as highly restricted and requiring preauthorization before use. The National Health Commission now requires hospitals to establish specialized management for carbapenem. 40 The increase in carbapenem use we found (from 2.6% of all antibacterials in 2015 to 3.6% in 2021) may have been due to a rise in the prevalence of infections with extended-spectrum, β-lactamaseproducing, gram-negative bacteria. For example, carbapenem use was observed to increase in parallel with the increased incidence of carbapenem-resistant Klebsiella pneumoniae infection seen in tertiary hospitals. 33 Hence, it is important to have dynamic data on temporal trends and patterns in antibacterial use and antibacterial resistance to identify correlations between them.
Our finding that inpatient antibacterial use decreased between 2013 and 2021 contrasts with our previous, population-weighted, antibiotic use research, 13 which demonstrated a 38.2% increase in antibiotic consumption in hospitals between 2011 and 2018. There are several possible reasons for this discrepancy. First, the hospitals sampled in this and previous studies were not identical, which may have introduced sample bias. Second, the procurement data used in previous studies may not have reflected actual drug utilization. Third, the metric adopted in previous studies was the number of daily defined doses per 1000 inhabitant-days, which was calculated by dividing the number of daily defined antibiotic doses procured by the population size. A manufacturing statistics yearbook -the Report of pharmaceutical market development in China (2018) -showed that the pharmaceutical market for antiinfection drugs has grown continuously since 2011. 41 In parallel, the number of inpatients in Chinese tertiary hospitals almost tripled between 2010 and 2017, which implies that demand for antibiotics increased. Meanwhile, according to China's Health Statistics Yearbook, 15 the population grew only slightly from 1.35 billion in 2011 to 1.40 billion in 2018. As a result, antibiotic use per 1000 inhabitants would be expected to increase because antibiotic use (the numerator) grew much faster than the population (the denominator).
The sudden drop in inpatient antibacterial use we found in 2020 was probably due to China's COVID-19 prevention and control strategies. The successive series of isolation and protection measures introduced would have diminished the number of hospitalizations and, consequently, inpatient antibacterial use. 42 Researchers showed that, after adjusting for provincial COVID-19 cases, antibiotic use remained constant. The impact of the COVID-19 pandemic on population-level antibiotic consumption in China could mainly be attributed to the decline in medical services caused by epidemic prevention and control measures, therefore, rather than to the treatment of COVID-19 cases. 43 The rebound of antibacterial use in secondary hospitals in 2021, then, may reflect the adaptation of the medical system to these measures. However, more detailed studies are needed to improve understanding of the impact of the pandemic at the population level. Our study has several limitations. First, the lack of data on antibacterial use before stewardship initiatives were introduced makes it difficult to be sure the initiatives were responsible for the trends over time. Second, the representativeness of the hospitals included in the study may have been affected by selection bias because participation in the surveillance network was voluntary. However, according to China's Health Statistics Yearbook, 15 tertiary and secondary hospitals in the study accounted for over 63.5% and 16.5%, respectively, of the total number of these hospitals in mainland China. Third, the Center for Antibacterial Surveillance's database does not cover antibacterial use in primary health care or in the community, hence we could not assess overall antibacterial use. Our analysis focused on hospital inpatient antibacterial use so we did not have to consider import or manufacturing data, which use different measures. However, our results were robust within the scope of the study. Fourth, only aggregated data were available for analysis. Nonetheless, the Center for Antibacterial Surveillance has an internal quality control process and the National Health Commission has conducted training programmes to improve data collection, which should ensure data quality. Fifth, the use of aggregated data meant that no patient-level information on the clinical use of antibacterials was available and, therefore, we could not assess the appropriateness of antibacterial use in individuals.

Research
Antibacterial use trends, China Haishaerjiang Wushouer et al.