This is where a large graphic or chart can go.

1. Abstract Effectiveness of a Multi-Component Intervention on Dispensing Practices at Private Pharmacies in Vietnam and Thailand: A Randomized Controlled Trial Chalker J, Ratanawijitrasin S, Chuc NTK, Petzold M and Tomson G Karolinska Institutet, Sweden; Management Sciences for Health ( MSH); Chulalongkorn University, Thailand; Hanoi Medical University, Vietnam; Nordic School of Public Health, Sweden Problem Statement: Private pharmacies are the first line of health care in many communities, commonly selling antibiotics in too small doses and prescription-only drugs such as steroids without medical supervision. Objectives: To study the effectiveness of a multi-faceted intervention on the dispensing practices of drug sellers within Hanoi and Bangkok. Design: A randomized, controlled trial. Setting: Private pharmacies in Hanoi and Bangkok Study Population: 34 intervention private pharmacies in Hanoi and 39 in Bangkok and an equal number of controls were randomly selected and assigned. In Hanoi, 28 intervention and 27 control pharmacies completed the study, as did 34 intervention and 35 control pharmacies in Bangkok Intervention: Three three-month interventions were implemented sequentially with four months in between: enforcement of regulations with local inspectors visiting to emphasize the importance of prescription-only medicine legislation; education, performed face-to-face in Hanoi and by a large group in Bangkok; and peer review, voluntary in Bangkok and effectively compulsory in Hanoi. Outcome Measures: Behaviour was assessed by five simulated client visits per pharmacy per dispensing practice, at baseline and a month or more after each intervention. Whether the requested antibiotic or steroid was dispensed and whether relevant questions were asked and advice given were recorded. Results: In Hanoi, significant improvements compared to controls was shown after the educational intervention, reducing the dispensing of illegal steroids (29% v 62%) and low dose antibiotics (69% v 90%), sustained by means of the peer review (17% v 57% steroids and 71% v 95% antibiotics), and in fewer dispensers asking no questions and giving no advice (11% v 30% steroids and 51% v 81% antibiotics). The only significant improvement in Bangkok was the reduction in illegally dispensing steroids (25% v 44%) after the regulatory intervention which was not sustained. In Bangkok, fewer of those in the group who volunteered for the peer review asked no questions and gave no advice for low-dose antibiotics requests after the peer review (58% v 81%). Conclusions: A multi-component intervention can have a profound effect in changing dispensers’ behaviour, but the effect is dependent on the context and the method of implementation. Possible reasons for differences between the cities are discussed. This is where a large graphic or chart can go.

2. Background • Drug sellers are becoming the first line of PHC in many communities • Dispensing Practices are often both bad and illegal • Few attempts to change but evidence shows multifaceted interventions best • No comparison of effectiveness of multifaceted interventions in different environments Study Aim To study the effectiveness of a multi-faceted intervention on dispensing practices of drug sellers in Hanoi and Bangkok on two critical behaviors: : • Selling antibiotics in small doses without prescription (which is particularly dangerous for the spread of antimicrobial resistance) • Selling prescription only drugs (steroids) without prescription

3. Study Design: Randomized Controlled Trial Monitoring by Simulated Client Visits, 5 visits per pharmacy for each tracer condition. 4 times: Baseline & after each intervention. The Clients asked for: 1) A small dose of an antibioitic 2) Steroids for a bad back

4. Interventions 1: Enforcement of Regulations Focusing on dispensing of prescription only drugs Hanoi: 2 visits giving a summary of px only regulations with a letter from the Provincial Health Bureau Bangkok: Checked steroids & steroid pxs, Gave a warning of violation of the regulations and gave instruction to the seller on the respective regulations 2: Education Hanoi: Academic detailing: 2 visits, 45 mins each by 2 people one visit steroids, and one antibiotics • Questions, advice and treatment (QAT) stressed with written and verbal info. Bangkok: Owners and counter attendants invited to 2 day workshops; included steroids, and antibiotic requests. • 9 shops who did not attend were visited twice- for 2-hour academic detailing one steroids, and one antibiotics 3: Peer Influence Hanoi: Hanoi divided 5 area groups with 5-6 PPS. Built on QAT. 5 meetings per group. Collected and reported cases. All pharmacies attended Bangkok:All intervention PPs staff invited to a meeting "Techniques to increase the revenue of drugstore". Setting up of peer groups discussed. Sent out minutes and invited for peer groups. Groups set own agendas, with some guidance. 16/34 did NOT attend.

5. Bangkok Hanoi 4 area types 789 PPs 2 districts in each 641 fit criteria Int / cont 34 pairs 39 cont 34 int 34 cont 39 int 34 int 35 cont 28 int 27 cont Sampling Private Pharmacies (PP) for intervention (int) and control (cont) groups

6. Antibiotic Results

7. Steroid Results

8. Results • In Hanoi, significant improvement compared to control was seen for the dispensing of antibiotics and steroids as well as a reduction in those not asking relevant questions or giving advice • In Bangkok there were no significant changes by the end of the intervention package • The importance of individually tailoring interventions to the locations and societies in which they will be implemented through formative research has been shown.

9. Methodological considerations-1 • Caution is needed in interpreting the difference in effectiveness between cities (inter-city) . Contextual factors are as likely as the details of the implementation to explain the difference in effectiveness in Hanoi and Bangkok. • Caution is needed in interpreting longitudinal trends. as the consistency of simulated client reporting varies • This does not affect the validity of the intra-city interpretation between intervention and control groups • The interventions were performed one after the other, so there is no possibility of isolating the effect of each individual intervention.

10. Methodological considerations -2 The randomized controlled trial (RCT)is a robust design to judge intra city effect RCTs are the cornerstone of clinical medicine for assessing the efficacy of medication or clinical intervention because of the minimizing of bias. The RCT can also be used for assessing the effectiveness in specific real life situations There are problems with RCTs • RCTs are expensive, the results are rarely produced within two years. Therefore the results are not useful for steering the design of the intervention • To avoid bias it is necessary to standardize the content and format of the delivery of an intervention and with complex behavioural interventions it is difficult to reproduce them exactly

11. Methodological considerations 3 • We know Multi faceted interventions are most likely to be effective. It is the very nature of multi faceted interventions to be contextualized. This reduces their external validity • So in our work the contextualization, all compromise the external validity and may be the reason for the difference between the results in Hanoi and Bangkok • In Hanoi, the success of the multi-intervention package gives important evidence showing that these drug-seller practices are changeable • The interventions are adapted to the opportunities of the environment and the goal is therefore to find whether the intervention works in that setting

12. Conclusion and recommendations • This study show that improvements are possible to achieve in the private sectors. However even with improvements major problems remain. • Isolated multi faceted interventions are not generalisable, leave problems of scaling up and will not solve the problems of antimicrobial resistance. • The successes of interventions depend on place. The search for the interventions that will universally work is therefore illusory. The art and science of developing specific strategies relevant to specific locations is needed. • Both indicators and interventions have to focus on chosen key problems. • If information from monitoring such indicators is produced in a timely manner, the data can be used to iteratively develop the intervention.