Private pharmacies in Hanoi, Vietnam: a randomized trial of a 2-year multi-component intervention on knowledge and stated practice regarding ARI, STD and antibiotic/steroid requests


  • J. Chalker,

    1. 1Management Sciences for Health, Arlington, VA, USA 2Hanoi Medical University, Hanoi Vietnam 3Department of Public Health Sciences, Karolinska Institutet, Stockholm, Sweden
    Search for more papers by this author
  • 1 N. T. K. Chuc,

    1. 1Management Sciences for Health, Arlington, VA, USA 2Hanoi Medical University, Hanoi Vietnam 3Department of Public Health Sciences, Karolinska Institutet, Stockholm, Sweden
    Search for more papers by this author
  • 2 T. Falkenberg,

    1. 1Management Sciences for Health, Arlington, VA, USA 2Hanoi Medical University, Hanoi Vietnam 3Department of Public Health Sciences, Karolinska Institutet, Stockholm, Sweden
    Search for more papers by this author
  • and 3 G. Tomson 3

    1. 1Management Sciences for Health, Arlington, VA, USA 2Hanoi Medical University, Hanoi Vietnam 3Department of Public Health Sciences, Karolinska Institutet, Stockholm, Sweden
    Search for more papers by this author

John Chalker, Management Sciences for Health, Suite 400, 4301 N Fairfax Drive, Arlington, VA 22203-1627, USA. Fax: +1-703-248-1635; E-mail:


Objectives  To assess the effectiveness of a multi-component intervention on knowledge and reported practice amongst staff working in private pharmacies in Hanoi regarding four conditions: urethral discharge [sexually transmitted diseases (STD)], acute respiratory infection (ARI), and non-prescription requests for antibiotics and steroids.

Method  Randomized controlled trial with staff working in 22 matched pair intervention and control private pharmacies who were administered a semistructured questionnaire on the four conditions before and 4 months after the interventions. The interventions focused on the four conditions and were in sequence (i) regulations enforcement; (ii) face-to-face education and (iii) peer influence. Outcome measures were knowledge and reported change in practice for correct management of tracer conditions.

Results  The intervention/control-pairs (22 after drop-outs) were analysed pre- and post-intervention using the Wilcoxon signed rank test. STD: More drug sellers stated they would ask about the health of the partner (P = 0.03) and more said they would advise condom use (P = 0.01) and partner notification (P = 0.04). ARI: More drug sellers stated they would ask questions regarding fever (P = 0.01), fewer would give antibiotics (P = 0.02) and more would give traditional medicines (P = 0.03). Antibiotics request: Fewer said they would sell a few capsules of cefalexin without a prescription (P = 0.02). Steroid requests: No statistical difference was seen in the numbers who said they would sell steroids without a prescription as numbers declined in both intervention and control groups (P = 0.12).

Conclusion  The three interventions in series over 17 months were effective in changing the knowledge and reported practice of drug sellers in Hanoi.


With health sector reform drug sellers are often the first and only contact with health delivery services (Logan 1983; Cederlof & Tomson 1995; Goel et al. 1996). In Vietnam, more than 80% of people go directly to a drug seller when they become ill (Tangcharoensathien 1992; World Bank 1995). The Good Pharmacy Practice (GPP) guidelines issued by WHO in 1996 attest to the importance of private pharmacies in primary health care. However, the quality of the treatment from private pharmacies is often very poor (Thamlikiktul 1988; Igun 1994; Chalker et al. 2000; Stenson et al. 2001a). This may be because the licensed pharmacist is rarely present, and an unqualified person works behind the counter (Kamat & Nichter 1998), or the pharmacist has a low level of knowledge, especially of rational drug use (Ross-Degnan et al. 1996). Cultural, social and economic factors all play a part (Madden et al. 1997).

Vietnam legalized the private health sector in 1989 (Doi Moi). A study in 1994 showed that violations of pharmaceutical regulations are common and enforcement of regulations is weak (Falkenberg et al. 2000). Antibiotics are often sold without a prescription (Chuc & Tomson 1999; Duong et al. 1997) and both knowledge and practice for the management of sexually transmitted diseases (STD) (Chalker et al. 2000) and respiratory infections (Chuc et al. 2001) are very poor.

Interventions to promote better care with private practitioners in high income countries have shown that multi-faceted strategies which increase provider knowledge have had some success in improving service quality (Brugha & Zwi 1998). Few studies exist in low and middle income countries and there is a need to evaluate combinations of strategies balancing incentives, controls and the education of providers, patients and communities (Brugha & Zwi 1998). Finding ways of achieving better pharmacy practice is essential. Semi-structured questionnaires presented by trained interviewers have been shown to assess knowledge and attitudes including reported practice more reliably than structured questionnaires (Stuart & Wiles 1997). Although stated practice tends to be better than actual practice, an improvement in stated practice shows an increased level of awareness of correct procedures (Adams et al. 1999).

The aim of this study was to see if a multi-component intervention could change the knowledge and reported practice of the staff working in private pharmacies in Hanoi. The study is part of the ‘Towards Good Pharmacy Practice in Thailand and Vietnam’ Project.


Tracer conditions

The four chosen tracer conditions of critical public health importance were (a) management of a STD (urethral discharge) in an adult man; (b) management of a simple upper respiratory tract infection (ARI) in a child < 5 years old with a mild cough; (c) a request for two to five capsules of an antibiotic (cefalexin) without a prescription; and (d) a request for steroids (prednisolone) without a prescription. The correct questioning, advice and treatment were agreed for each condition (Fig. 3).

Figure 3.

Correct management.

Selection of pharmacies

There were 789 private pharmacies registered in the urban area of Hanoi, 641 outside a hospital and not mainly wholesalers. Thirty-four pairs of pharmacies were selected randomly from the 641. The pairs were formed according to the following matching criteria: turnover: high, medium or low according to district inspectors; whether the pharmacist was the license holder or not; and whether they were situated close to a hospital or not. The pharmacies in the pairs were randomly allocated into the intervention or control group (Fig. 1).

Figure 1.

Flow chart of selection process and criteria for intervention and control pharmacies.


A semi-structured questionnaire (available upon request from JC) was developed to elicit the respondents' knowledge and reported practice by describing what they would have done given a client complaining of such symptoms. The questionnaire contained open questions such as ‘How would you deal with someone who came to you with a small child who is coughing?’ If there was no response, the questioner would prompt with ‘would you ask any questions?’ or ‘would you give any advice?’ and record the answer as prompted.

The questionnaire was piloted in other pharmacies and then administered to each person working in the intervention and control pharmacies on duty at the time of the visit. Interviews were conducted out of earshot of other staff working in a pharmacy. Four months after the last intervention we returned to both intervention and control pharmacies to repeat the questionnaire (Fig. 2).

Figure 2.

Experimental design with timing and sequence of interventions and interviews.

The interventions

The three interventions were implemented sequentially over a 17-month period from May 1998 to September 1999. Each intervention lasted 3 months, with a gap of 4 months before the next intervention, and was designed in cooperation with pharmacists who have worked in private pharmacies as well as with the Health Authorities and the Pharmacy Association in Hanoi.

Regulatory enforcement

The first intervention focused on regulation of prescription only drugs, particularly the drugs related to tracer conditions in the study. Four inspectors of the Hanoi Health office were trained to cover these areas of inspection. In pairs they visited the intervention pharmacies twice a month apart. In addition to normal inspection procedures, they delivered the regulation about selling prescription-only drugs and then explained the regulation in more detail. A normal inspection would take place twice a year and not concentrate on these topics.


Core research members with a local pharmacologist and two clinicians developed the pharmacy treatment guidelines for the four tracer conditions (Fig. 3). Each intervention pharmacy was visited twice by two people for face-to-face education sessions with all staff present. Each session lasted about 45 min and included both written and verbal information. If customers came the session was interrupted until they had been served. The sessions focused on the importance of GPP and asking the right questions, giving the right advice and the right treatment in relation to the four tracer conditions.

Peer influence

The intervention pharmacies were geographically divided into five groups. The research group appointed one person in each group to be the leader, based on the experience gained during the educational intervention. A 1-day seminar on the importance of GPP including rational dispensing practice, targeting the four tracer conditions was held with the five group leaders and representatives of the Hanoi Pharmacy Association. The importance of peer influence to improve practice among private pharmacy staff was emphasized. Following this seminar, the five group leaders held meetings with the staff of the intervention pharmacies on the same issues. After the meetings they took notes of every case of patients with the above conditions that visited them while working in their shops. They then held monthly meetings (three times) to review what had been done in relation to these conditions by each pharmacy.

Data analysis

The results were coded and entered into a computer using EPI INFO version 6. The data was analysed on the pharmacy level, so that if there were two interviews in a pharmacy, if both interviewees answered yes or both answered no the answer was coded as 1 for yes or 0 for no. If one answered yes and one answered no then the answer was coded as 0.5. Positive answers after a prompt were recorded as yes. A matching pair was only analysed if therewere answers for the matching pairs in the intervention and control groups before and after the interventions. The method of summary statistics comparing the differences between pre- and post- intervention and the control was used to assess whether the interventions had had a significant effect (Diggle et al. 1994). Assuming approximately continuous and symmetric data, the Wilcoxon signed rank test was used (software SPSS version 11).

Ethical approval

The study had ethical approval from the Ministry of Health in Vietnam and the Karolinska Institute. It was conducted in collaboration with the Hanoi Provincial Health Bureau and the Pharmacy Association.


In the course of the study, four pharmacies closed and one refused to take part in the third intervention. No post-intervention interviews were taken from four intervention and four control pharmacies as they were closed early before the Vietnamese New Year (TET) holidays. These gave 22 matched pairs (Fig. 1). The results, presented by pharmacy, are summarized in Table 1.

Table 1.  Knowledge and reported management
 BaselinePost-interventionsWilcoxon signed rank test*P’ value (for 22 matched pairs)**
Intervention pharmaciesControl pharmaciesIntervention pharmaciesControl pharmacies
  • The numbers here are by pharmacy. If more than one interview was conducted in a pharmacy, for each question if both said yes it was recorded as yes (1), if both said no it was recorded as no (0), if they disagreed it was recorded as 0.5.

  • The values in parenthesis are in percentages.

  • The test is a Wilcoxon signed rank test (matched pairs difference of differences between pre- and post- between intervention and control).

  • **

     ‘P’ value (exact two sided) difference between matched pairs.

Number pharmacies22222222 
(A) Urethral discharge
 Sexual activity 8.5 (39) 4 (18)14.5 (66) 9.5 (43)0.80
 Health of partner 1 (5) 2 (9) 7.5 (34) 2 (9)0.03
 Condom use 6 (27) 3 (14)13.5 (61) 0 (0)0.01
 Partner notification 2 (9) 0 (0) 9.5 (43) 1 (5)0.04
 Sell drugs 7.5 (34) 5.5 (25)15.5 (70)12.5 (57)0.73
 Correct treatment 0 (0) 0 (0) 8 (36) 0 (0)
(B) Simple upper respiratory infection
 About breathing11 (50)12 (55)16 (73) 8.5 (39)0.10
 About fever14 (64)16.5 (75)16.5 (75) 9.5 (43)0.01
 Antibiotics 3.5 (16) 2.5 (11) 2 (9) 8 (36)0.02
 Traditional medicine 1 (5) 3 (14)12.5 (57) 5 (23)0.03
(C) Sell cefalexin with no prescription
Sell antibiotics12.5 (57)10 (45) 4.5 (20)13.5 (61)0.02
(D) Sell steroids with no prescription
Sell steroids10.5 (48) 7 (32) 0 (0) 3 (14)0.12

Sexually transmitted disease

After the three interventions significantly more drug sellers would ask about the health of the partner (P = 0.03) and advise on condom use (P = 0.01) and partner notification (P = 0.04). (Table 1). There is no statistical difference between the number in the intervention group who say they would sell drugs after the intervention package and controls (P = 0.73). But eight (36%) of the intervention group say they would give a treatment which is syndromically correct post-intervention vs. none in the controls. Because of the small number of matched pairs who both treated before and after this is not significant.

Acute respiratory infection

After the interventions significantly more drug sellers in the intervention group say they would ask questions about fever (P = 0.01) (Table 1). However, the increasein asking about the quality of the breathing isnot significant (P = 0.10). Significantly fewer of thepost-intervention group would give antibiotics (P = 0.02) and more would give traditional medicines (P = 0.03).

Antibiotic requests without prescription

Significantly fewer of the intervention group would sell antibiotics without a prescription (P = 0.02). More intervention interviewees gave as reasons: ‘the capsules would not be effective’, 13 (59%) compared with five (23%) controls; ‘this practice would help to cause the development of resistance’, 12 (55%) compared with two (9%) controls; and ‘the client has no prescription’ 10 (45%) compared with three (14%) controls. However these only comprised a few matched pairs and the changes were not significant.

Steroid requests without a prescription

There is no significant change between intervention and control in selling steroids without a prescription (P = 0.12), partly because there was a marked reduction in both intervention and control groups. The most common reason for not selling them in the post-intervention group was because of the regulations [18.5 (84%) compared with 8.5 (39%)] in the controls which was not significant when matched pairs were analysed.


This study, to our knowledge, is the first multi-intervention experiment in the private pharmacy sector in a low income country and reports important changes in staff knowledge and stated practices. We are aware that there is an important potential difference between stated and actual practice. Significantly more drug sellers stated that they would ask questions about partner health and give advice on condom use and partner notification, pre-requisites for STD control. For an uncomplicated upper respiratory tract infection in a child, more interviewees learnt to ask questions regarding fever and to treat with non-antibiotics. Fewer drug sellers said they would sell a few antibiotic capsules without prescription, which could contribute to containing the spread of antimicrobial resistance (Larsson et al. 2000; Chalker 2001) and decreasing the side-effects from unnecessary drugs.

Failure to interview people from four intervention and four control pharmacies after the interventions may have caused a bias in results, but this is unlikely because of the equivalent numbers between intervention and control. The unit for this analysis and change shown is by pharmacy. In the second round of questionnaires only 54% of the intervention group interviewees and 36% of the control group participants had been interviewed before. This may have been because of the staff turnover or different people working on different days of the week. As for the public sector health facilities we hypothesized that drug use would be more similar within facilities (WHO/DAP 1993) and that the interventions would diffuse through the pharmacy.

This study was a controlled trial over a 17-month period. The stringency in the design means that confidence can be placed on the differences found. The three intervention design here can be compared with the single face-to-face educational intervention of 2 h once in Kenya and twice in Indonesia which proved effective at least in the short term (Ross-Degnan et al. 1996). One-time training with little supervision was found to be inadequate in Niger (Wouters 1995). Enforcing regulations has led to better pharmacy practice in Laos (Stenson et al. 2001b). In Ghana with an education-only intervention, correct syndromic drug provision for urethral discharge also improved in the short-term but remained relatively low (Adu-Sarkodie et al. 2000). In this trial the innovative interventions combining regulatory enforcement, face-to-face education and strengthening peer influence is in line with recommendations (Brugha & Zwi 1998).

The questionnaire was open-format with questions such as ‘What would you do in this situation?’ to explore the drug sellers thinking. This methodology is less leading than structured questionnaires or multiple-choice questions, and therefore probably reflects practice more closely (Stuart & Wiles 1997). Some studies which interviewed drug sellers in a more structured manner and also observed their behaviour by using the surrogate client method (SCM) and then compared the results (Ross-Degnan et al. 1996) found that the actual behaviour is worse than that claimed by the drug seller during interview. The format of our participatory education and the peer influence interventions meant that drug-sellers actually performed improved practices during these interventions. A study in Sweden (Diwan et al. 1997) hypothesizes that such participatory improvement in performance lead to an improvement of knowledge and attitudes which in turn leads to improved practice when not observed. In this research only change of reported behaviour has been recorded so that the reality of actual behaviour change is speculative. But positive changes in reported practice reflect improved knowledge which is in itself an important determinant of later behaviour change.

Major criteria of GPP are to put the client's interest first and to promote rational use of drugs (WHO 1996). The private pharmacy is a business dependent on profit (Cederlof & Tomson 1995). In the STD group, although illegal without a prescription, more stated they would sell the correct treatment (this was not statistically significant due to the small number of matched pairs selling drugs). STD treatment is relatively expensive, indicating that where there is a positive financial incentive the effects could be expected to be greatest. For ARI there was a stated improvement in not selling antibiotics, matched by a reported increase in selling traditional medicine, maintaining profitability. During the peer influence meetings, some pharmacists stated that initially they had not wanted to ask questions and give advice to their clients. However, after the interventions they understood that asking and advising and even sometimes advising not to use drugs built up their reputation to the public, thereby increasing their status and turnover. These observations show that simple profit is not the only driving force.

The interventions were apposite because regulatory enforcement is often weak in low income countries and had previously been found to be so in Vietnam (Falkenberg et al. 2000). In the absence of effective regulatory mechanisms in the private sector, an increased role of professional societies and peer influence is one of the few options available. Such an approach is feasibly sustainable because in Vietnam, Provincial Health Authorities and the Pharmacy Association have the capacity to promote and carry out these interventions. The cost for the three interventions was approximately 5700 USD for 30 pharmacies which is less than 200 USD/pharmacy. This is an achievable target for most professional societies or donors.

We have shown that the intervention package was effective in improving knowledge and reported practice. Now the most effective intervention needs to be explored on actual practice. If the drug sellers are going to follow GPP and effectively adopt their de facto role as primary care providers, they need to take better care of their clients.


We dedicate this article to the memory of Prof. N.T. Do from the College of Pharmacy in Hanoi, who co-ordinated the activity in Hanoi, and played an active part and in the design and implementation of the research. The project was financially supported by a European Union grant (number ERB3514PL950674) and the WHO Drug Action Programme. Collaborating institutions were the Karolinska Institute, the London School of Hygiene and Tropical Medicine, The College of Pharmacy, the Centre for Social Science in Health in Hanoi and Dr Sauwakon Ratanawijitrasin, the national coordinators at the Health System Research Institute in Bangkok. We also acknowledge Dr D.L. Huong for her major role in the development of the questionnaire and Max Petzold for statistical advice. We thank all participating pharmacies as well as the Health Bureau of Hanoi and the Hanoi Pharmacy Association.