Effect of unifaceted and multifaceted interventions on antibiotic prescription control for respiratory diseases: A systematic review of randomized controlled trials

Background: The global health system is improperly using antibiotics, particularly in the treatment of respiratory diseases. We aimed to examine the effectiveness of implementing a unifaceted and multifaceted intervention for unreasonable antibiotic prescriptions. Methods: Relevant literature published in the databases of Pubmed, Embase, Science Direct, Cochrane Central Register of Controlled Trials, China National Knowledge Infrastructure and Wanfang was searched. Data were independently filtered and extracted by 2 reviewers based on a pre-designed inclusion and exclusion criteria. The Cochrane collaborative bias risk tool was used to evaluate the quality of the included randomized controlled trials studies. Results: A total of 1390 studies were obtained of which 23 studies the outcome variables were antibiotic prescription rates with the number of prescriptions and intervention details were included in the systematic review. Twenty-two of the studies involved educational interventions for doctors, including: online training using email, web pages and webinar, antibiotic guidelines for information dissemination measures by email, postal or telephone reminder, training doctors in communication skills, short-term interactive educational seminars, and short-term field training sessions. Seventeen studies of interventions for health care workers also included: regular or irregular assessment/audit of antibiotic prescriptions, prescription recommendations from experts and peers delivered at a meeting or online, publicly reporting on doctors’ antibiotic usage to patients, hospital administrators, and health authorities, monitoring/feedback prescribing behavior to general practices by email or poster, and studies involving patients and their families (n = 8). Twenty-one randomized controlled trials were rated as having a low risk of bias while 2 randomized controlled trials were rated as having a high risk of bias. Six studies contained negative results. Conclusion: The combination of education, prescription audit, prescription recommendations from experts, public reporting, prescription feedback and patient or family member multifaceted interventions can effectively reduce antibiotic prescription rates in health care institutions. Moreover, adding multifaceted interventions to educational interventions can control antibiotic prescription rates and may be a more reasonable method. Registrations: This systematic review was registered in PROSPERO, registration number: CRD42020192560.


Introduction
The discovery of antibacterial drugs is an important milestone in the history of human medical science. The emergence of antibiotics has changed the outcome of infections, thereby extending people's life expectancy. [1] However, in the past decade, 50% of the world's antibiotic prescriptions have been misused to treat coughs and colds, and many of these prescriptions have no indications for antibiotic use. [2] According to our previous research, [3] most unreasonable antibiotic prescriptions are used to treat these uncomplicated respiratory infections caused by viruses, which is common in many countries of the world. [4][5][6][7][8][9] As a result, globally, 4.95 million people died in 2019 from bacterial antimicrobial resistance caused by overuse and misuse of antibiotics. If unchecked, this figure could rise to 10 million by 2050, surpassing cancer as the leading cause of death. [10,11] Previous studies have shown that implementing unifaceted or multifaceted interventions for medical staff, patients and caregivers can effectively reduce antibiotic misuse and thus curb antibiotic resistance. [12,13] There are 2 Cochrane Systematic Reviews on interventions to improve antibiotic prescribing; one is in the ambulatory care setting [14] and the other is among hospital inpatients. [15] The ambulatory care review has not been updated since 2005 but more recent systematic reviews have been published about interventions in primary care, especially for respiratory diseases [12,13,16] and care homes. [17] They focused specifically on primary care institutions and physicians, or systematic reviews of educational interventions. There is strong evidence that educational interventions improve antibiotic prescribing but more evidence is required about the effectiveness of supplementing education with additional intervention elements and on the sustainability of interventions in a wider range of studies and study subjects. The antibiotic prescription rate (APR) is the main outcome indicator of interventions to control antibiotic prescriptions in those studies. [15,16] To further confirm the effectiveness of various interventions on antibiotic prescription misuse and overuse in respiratory diseases, we used the Cochrane systematic review and meta-analysis methods to evaluate published results of relevant randomized controlled trials (RCTs) to provide a reference for relevant decision-makers.

Protocol and registration
This systematic review was conducted using the PRISMA reporting guidelines (Appendix 1, Supplemental Digital Content, http://links.lww.com/MD/H434) [18] and was registered in PROSPERO with registration number CRD42020192560. The study was approved by the Human Trial Ethics (Appendix 2, Supplemental Digital Content, http://links.lww.com/MD/H435) Committee of Guizhou Medical University (Certificate No.: 2019 (148)) in December 27, 2019.

Inclusion criteria
The inclusion criteria of this study were based on the full-text information available in the English and Chinese literature databases and also included the following: (1) The study objective focused on respiratory diseases; (2) RCTs of intervention and control groups with measurements collected both before and after the intervention; (3) Intervention targets were medical staff (general practitioners, physicians, nurses), patients and caregivers; (4) The interventions were clearly described.

Selection strategy and information sources
A systematic literature search was conducted in PubMed, Embase, Science Direct, Cochrane Central Register of Controlled Trials, China National Knowledge Infrastructure and Wanfang databases. The search period was from the time of construction of the database to February 28, 2022. We collected published studies in English or Chinese evaluating the effectiveness of antibiotic prescription interventions. Keywords and search terms used included ("Antibiotic prescription" or "Antimicrobial prescription") and ("Intervention") and ("Respiratory"). Appendix 3 (Supplemental Digital Content, http://links.lww.com/MD/ H560) contains an example of the search strategy. Data were independently filtered and extracted based on the pre-designed inclusion and exclusion criteria.

Study setting
Referring to Vodicka et al [19] and Roque et al, [12] study settings included: primary care; hospital care; health care center; and nursing homes.

Study design
Study designs included: cluster randomized controlled trial and RCT.

Primary outcomes and findings
The primary outcomes and findings of studies were adapted from Roque et al. [12,20] The primary outcome was APR. The primary findings included: (+) positive findings (±), negative findings (-), both positive and negative (±).

Summary of APR for respiratory diseases
In the RCT studies we included, we performed a more detailed analysis if the outcome variable was the APR (defined as antibiotic prescriptions/total prescriptions × 100%) and there was a detailed prescription quantity report or it could be inferred indirectly from the literature.

Risk of bias assessment
The first 2 authors (YC and ZC) read the abstracts of all uncertain studies and the full-text of all studies that were still uncertain. A final agreement to include or exclude the studies was done after discussion of the discrepancies. The risk of bias in the included studies was independently assessed according to the "Cochrane systematic reviewer's manual." [21] There are 7 items in the criteria: random sequence generation (selection bias), allocation concealment (selection bias), blinding of participants and personnel (performance bias), blinding of outcome assessment (detection bias), incomplete outcome bias (attrition bias), selective reporting (reporting bias) and other bias. The bias risk assessment scoring criteria for each study were: "low risk" when all items were rated "low" or one or two of them were "unclear." If one or more items were rated "high" and more than one "unclear," it was rated as "high risk." RevMan 5.3 software was used to show the results.

Meta-analysis
We conducted a meta-analysis of all the studies selected for the systematic review, and the study objects were the medical records of the intervention group and the control group after the intervention. The risk difference (RD) of APR of the 2 groups was combined, and the combined effect value and 95% confidence interval (95% CI) were calculated. The T test and chi-square test statistics were used to analyze the heterogeneity between the results (a = 0.05), a fixed effect model (P > .05) or a random effect model (P ≤ .05) will be determined. If I 2 < 50%, a fixed effect model was used; if I 2 ≥ 50%, a random effect model was used. If there was still significant heterogeneity in the research results, a subgroup analysis or a descriptive analysis is required. RevMan 5.3 software was employed for meta-analysis.

Results
The search and selection process are shown in Figure 1. A total of 1390 studies were obtained, of which 384 were excluded after reading the title and abstract. The remaining 1006 studies were further screened by reading the full-text, of which 916 were excluded: 190 because they were treatment or drug intervention trials, 25 because they were antibiotic cost-benefit analysis studies, 170 because they were systematic reviews or cohort studies such as non-RCTs, 390 because they were observational studies (cross-sectional studies, and case-control studies), 125 because they were duplicate studies, and 15 studies because the full-text could not be accessed. Finally, of the 91 eligible studies, a total of 23 studies were included in the systematic review and meta-analysis after excluding 68 studies that did not have APR as the outcome variable.

Interventions of included APR studies
Twenty-three studies of interventions for health care workers also included: (6) regular or irregular assessment/audit of antibiotic prescriptions [22,25,27,[32][33][34]38,40] and (7) prescription recommendations from experts and peers delivered at a meeting or online. [23][24][25][26][30][31][32]34] Vervloet et al [34] implemented a prescription recommendation in an Electronic Prescribing System. When a family physician tried to prescribe an antibiotic to a patient with respiratory tract infections, the Electronic Prescribing System immediately prompted an alert with the message "no prescription," and if the doctor still wanted a prescription, a pop-up window containing the message "delayed prescription" would appear. After acknowledging these 2 electronic alerts, the doctor could write a prescription. (8) Two studies [29,37] reported publicly on doctors' antibiotic usage to patients, hospital administrators, and health authorities. The report contained APR, injection for APR, cost, and peer ranking. (9) Eleven studies reported monitoring/feedback prescribing behavior to GPs by email or poster (the prescribing behavior of individual physicians for 6 months or 1 year). [22,24,25,[27][28][29]32,34,37,40,41] Among the studies of interventions, 8 [22,24,26,29,32,36,38,39] involved patients and their families, including the distribution of leaflets and brochures on the rational use of antibiotics, the installation of multimedia education systems or poster/video in waiting rooms, and the use of flyers and posters. One [36] of the studies involved only patients and family members.

Risk of bias
As shown in Figure 2, 21 studies [22][23][24][26][27][28][29][30][31][32][34][35][36][37][38][39][40][41][42][43][44] were rated as having a low risk of bias while 2 [25,33] were rated as having a high risk of bias. Figure 2 shows the risk of bias assessment for each criterion. Among the 23 studies, 1 [33] in the "Random Sequence Generation" section were determined to have a high risk of bias as the one did not describe the sampling method. Although the study stated itself as using a randomized approach, the process described by it is more likely to be grouped according to the availability of interventions. One study [25] was identified as having a high risk of bias in the "Blinding of outcome assessment" section due to a lack of random sampling and unblinded Table 2 Basic characteristics of the 23 APR studies.

Author
Year Country Sample size included in the study (antibiotic prescriptions/total prescriptions)

Evaluation of publication bias
According to the included studies, the inverted funnel chart was produced, which has poor symmetry and may have publication bias. See Figure 3 for details.

Discussion
A total of 1390 studies were obtained of which 23 studies the outcome variables were APR with the number of prescriptions and intervention details were included in the study. In the 23  studies that were eventually included, GPs and physicians were the main subjects of the intervention. Fourteen studies (58%) did not have a specific population with the diseases and 13 studies (57%) had intervention periods of less than 1 year. Patient-only, long feedback intervals and non-blind nature of the intervention may have led to negative results and a high risk of bias. Of these studies, 17 (74%) had positive results and 6 (26%) contained negative results. A further analysis of 23 studies found that all but 2 (YC and MCG) were multifaceted interventions. Educational methods were the most common interventions. Among them, the distribution of antibiotic guidelines to doctors, short training sessions   (2). CI = confidence interval. www.md-journal.com of one to several days and training in doctors' communication skills were the most common educational interventions, and they were often used in combination. In addition, interventions for health care workers included monitoring and feedback of doctors' prescribing performance to them, auditing or evaluating the rationality of prescriptions, and prescribing recommendations from experts or via an electronic prescribing system. On the basis of the above interventions, some were added to the intervention of patients or caregivers. However, it is important to note that patients cannot be given an intervention alone, or the intervention may face ineffective results.
These 23 RCTs studies investigated a wide range of interventions targeted at both clinicians and patients (education, guidelines, prescriber feedback, patient pamphlets, etc.). This makes it very difficult to meta-analysis and to interpret the results as its unclear exactly which intervention targeted at which groups and in what setting is having the impact on prescribing. It also explains why there are such high rates of heterogeneity.
We identified the final 23 RCTs using a risk of bias tool. [21] Most (21) of the studies were low risk, but 19 of them had an "unclear risk of bias" due to insufficient information. Therefore, we hope that future studies will follow standard RCT procedures (e.g., SPIRIT 2013 Checklist) [45] to conduct trials and write manuscripts.
In summary, multifaceted feedback interventions were used in most of the included studies. Therefore, the education and training of doctors in prescribing antibiotics should be strengthened, organized medical staff should delve deep into the rules and regulations of antibiotics, and make full use of pharmacology, pharmacokinetics, pharmacodynamics, and other relevant knowledge to issue prescriptions. [46,47] On this basis, various feedback interventions can be added, such as communication between experts and peers, prescription audits, and ranking of doctors in the same department. In addition, interventions can improve the awareness of patients and their families toward antibiotics, such as providing them with brochures and leaflets, displaying posters in the waiting rooms, installing a multimedia education system in waiting areas, and encouraging patients to communicate more with their physicians about the use of antibiotics. Dekker et al [39] and Metlay et al [24] adopted certain intervention measures for patients and their families based on educational intervention measures for medical staff. Altiner et al [26] studied patients and their families based on using feedback intervention for medical staff, and the degree of reduction of antibiotic utilization was significantly higher than in other studies. Therefore, according to education and training interventions, feedback interventions were used to influence the prescribing behavior of doctors and improve the cognition of patients and their families about antibiotics. This multifaceted behavioral feedback intervention might be a more rational approach to antibiotic prescription control. In terms of policymaking, health administration departments should introduce regulations and relevant policies on the administration of antibiotics to strictly control the use of antibiotics. These departments can take strong administrative interventions against the unreasonable use of antibiotics, for example, patients or consumers could only obtain antibiotics from the pharmacies based on prescriptions, and doctors can prescribe antibiotics in a hierarchical manner.
Of the 24 studies, 6 contained negative results. Four [28,30,36,44] of them were negative and positive (±): Hürlimann et al [28] advocated use of an intervention that included providing guidelines for the treatment of the diseases and providing sustained, regular feedback (twice a year) on individual physicians' antibiotic prescribing behavior over a 2-year period, which did not reduce the APR, but increased the use of recommended antibiotics. Routine guidelines and long personal feedback intervals (twice yearly) may be the reason why the intervention was not so effective. Zahlanie et al [44] developed an Pediatric educational intervention. After the intervention, antibiotic prescribing rates were higher in the intervention group (83.1% vs 77.7%; P = .024), the treatment course was shorter than that of the control group. The author believed that the problem of small sample size and multifaceted intervention methods need to be improved. Chen et al [30] conducted a 2-month intervention in which educational messages were sent to the intervention group 3 times a week. Antibiotic prescriptions did not change, and prescriptions for other drugs declined. The study said that text messages was more convenient than looking at literature for health workers, but more research is needed to confirm the effectiveness. Lee et al [36] gave patients in the intervention group 9 days of intervention in the form of pamphlets and oral education. The study did not reduce antibiotic prescribing, except among Indian patients. This suggests that intervention in patients alone has no significant effect on reducing antibiotic prescriptions. Two [23,42] of them were negative (−): Coenen's et al [23] interventions in 2002 included a clinical practice guideline for acute cough, an educational outreach visit and a postal reminder. They concluded that the intervention was effective by comparing OR values before and after the intervention [OR adj (95% CI) 5 0.56 (0.36-0.87)]. However, RD or relative risk more interpretable for clinical trials or prospective studies. When RD values of this study were compared, there was no difference between the intervention group and the control group (see Fig. 4 forest plot). Mann et al [42] implemented a clinical decision support tool to guide evidence-based evaluation and treatment of streptococcal pharyngitis and pneumonia. According to the authors, the negative results may be due to the very low adoption and utilization rates of the tool at all clinical sites.
Our study has certain limitations. First, like most systematic reviews, there is a possibility of publication bias as the majority of the studies (17/23) reported positive results. Furthermore, some studies were not included due to incomplete data and non-randomization. This limitation may have reduced the objectivity of the results to a certain extent. Second, there were different degrees of quality differences in the design of the included studies, which may have affected our results. Third, these studies were conducted in different countries. The policies and management systems of antibiotic use differ by country, thus there was a risk of information bias. Fourthly, we only focused on the APR, but did not pay attention to the rational evaluation of antibiotics.
This systematic review found that combination of education, prescription audit, prescription recommendations from experts, public reporting, prescription feedback and patient or family member interventions and other multifaceted feedback interventions can effectively reduce the rate of antibiotic prescriptions and promote the rational use of antibiotics. However, due to the above limitations, we can only conclude that adding multifaceted feedback interventions to education interventions may be a more reasonable control method. In the future, more studies need to be included to obtain more accurate information.