Evaluation design
We used a pre-and post-test community intervention design covering all alcohol retailers both on premise (bar, hotel, restaurant, karaoke bars) and off-premise (grocery shops). All establishments that had alcohol sales permits were selected from four district towns: Pema Gatshel town in Pema Gatshel district in the east, Damphu town in Tsirang district in the central and adjoining Khuruthang town in Punakha district and Bajo town in Wangdue district in the western region. The four districts were selected to represent the three geographical administrative regions commonly used as a reference in the country.
Proxy Purchasers
Outlets’ willingness to sell alcohol was assessed using proxy purchasers comprising underage-looking purchasers (<18 years), sober adults and adults acting alcohol intoxicated. Four underage-looking young adults (2 females and 2 males) and 4 adults (2 females and 2 males) were recruited in Damphu town. In Pema Gatshel town, six purchasers – two underage-looking young adults (2 females and 2 males) and 4 adults (2 females and 2 males) were recruited. In Bajo and Khuruthang towns, eight underage-looking purchasers (2 females and 6 males) and eight adult purchasers were recruited. Purchasers were recruited based on the acting skills. Under-age looking young adults were identified as below 18 years by a group of judges. Proxy purchasers were trained on survey instruments using purchasing scenarios. The duration of training varied from two days in Damphu town to three hours training in Pema Gatshel, and one day for Bajo and Khuruthang towns. The same purchasers completed the purchase surveys in Bajo and Khuruthang. Separate group of purchasers were recruited in pre- and -post intervention purchase surveys in intervention districts of Pema Gatshel town and Damphu town.
The proxy clients attempted purchase at illegal hours (before 1 pm during non-dry days); during dry day (Tuesday) and during legal hours (1-10 pm on a non-dry day) for underage looking and pseudo intoxicated adults. Legal visits between 1-10 p.m. by sober adults were included for comparison too. Every purchase attempt was made by a pair of proxy clients. After leaving the establishment, the purchasers were required to report to an enumerator who waited at a distance. The enumerator filled the survey tool as reported by the purchasers in their presence. Pre-intervention purchases were conducted in August 2019 and post-intervention purchase surveys in March 2020. Only pre-intervention purchase surveys were completed in Bajo and Khuruthang for baseline data acquisition.
Intervention
The intervention consisted of a briefing of outlet owners and sellers explaining the alcohol sales rules, and warning of the legal consequences of breach of alcohol rules by the government officers. The briefing was led by the Dzongdag (district governor), police officer, trade officer and district health officer. One representative from each retail outlet was invited from the study sites. The briefing was conducted on 29 October 2019 in Damphu and 10 October, 2019 in Pema Gatshel town. Briefing included existing alcohol sales rules, penalties and possible charges for violators, and the negative health effects of alcohol. The alcohol regulation toolkit containing information on sales rules and a copy of the laminated A3-size alcohol rules notification was handed over by the governor and officials and sellers were asked to display the notification in a prominent location within the premise. Retailers were informed that surprise checks will be conducted, and violators penalized as per the regulations that involve court hearings if required.
Ethics approval
Ethics approval was obtained from the Research and Ethics Board of the Ministry of Health, Royal Government of Bhutan for purchase surveys in Damphu and Pema Gatshel towns and local administrative approvals were obtained for Tsirang and Pema Gatshel district administrations. Wangdue and Punakha purchase attempts were done as a part of the routine monitoring activity of the chief medical officer at Bajo hospital and no ethical approval were sought.
Sampling
All establishments with alcohol sales licenses (i.e. bars, hotels, restaurants, grocery, karaoke bars) in Damphu and Pema Gatshel towns and Bajo and Khuruthang were included for the study. A total of 60 outlets in Damphu and 12 outlets in Pema Gatshel town were provided with interventions while 55 outlets in Bajo and Khuruthang were included in the purchase survey(Table1.1)
Table1.1 Study design - Number of purchases made by shopper type and time/day of purchase
|
|
Damphu town, Tsirang
|
Pemagatshel town, Pema Gatshel
|
Wangdue (Bajo) & Punakha (Khuruthang)
|
Intervention
|
|
Time/day of purchase attempt
|
Time/day of purchase attempt
|
Time/day of purchase attempt
|
Shopper type
|
before 1 p.m.
|
1-10 p.m.
|
Tuesday
|
Total
|
before 1 p.m.
|
1-10 p.m.
|
Tuesday
|
Total
|
before 1 p.m.
|
1-10 p.m.
|
Tuesday
|
Total
|
Pre-
|
Sober adults
|
53
|
0
|
40
|
93
|
5
|
7
|
10
|
22
|
-
|
3
|
103
|
106
|
|
Intoxicated adults
|
4
|
54
|
0
|
58
|
0
|
12
|
0
|
12
|
-
|
36
|
-
|
36
|
|
Underage
|
7
|
51
|
0
|
58
|
0
|
11
|
0
|
11
|
-
|
49
|
29
|
78
|
|
Total
|
64
|
105
|
0
|
209
|
5
|
30
|
0
|
45
|
-
|
88
|
132
|
220
|
Post-
|
Sober adults
|
24
|
24
|
49
|
97
|
24
|
24
|
49
|
97
|
-
|
-
|
-
|
-
|
|
intoxicated adults
|
1
|
52
|
1
|
54
|
1
|
52
|
1
|
54
|
-
|
-
|
-
|
-
|
|
Underage
|
0
|
24
|
15
|
39
|
0
|
24
|
15
|
39
|
-
|
-
|
-
|
-
|
|
Total
|
25
|
100
|
65
|
190
|
25
|
100
|
65
|
190
|
-
|
-
|
-
|
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Analysis
The data was entered and managed using Epidata Entry Software version 3.1. A double entry was made and validated. Data analysis was carried out using Stata 15 IC (StataCorp. 2017. Stata Statistical Software: Release 15. College Station, TX: StataCorp LLC). Analysis included all completed purchase attempts which included illegal purchasing of alcohol of 4 different scenarios: (1) before 1 p.m., (2) on Tuesdays, (3) to underage-appearing patrons (i.e. younger than 18 years old), (4) to patrons who appeared to be intoxicated. While alcohol purchasing after 11 p.m. would be illegal, these purchasing attempts were not carried out as none of the outlets were open till 11 p.m. Few purchase attempts of sober adults during legal hours were completed too for comparison.
Information on main characteristics of sampled establishments were summarized. Primary outcome was compliance defined as failed purchase of illegal alcohol sales. Effect of the intervention on the difference of compliance rate to legal restriction was estimated using linear probability regression model. Both crude effect and adjusted effect were estimated for the full sample in Tsirang and Pema Gatshel. Effects were estimated for each purchaser scenario and two-way cluster was accounted for in the standard error at the level of the shopper and the establishment using Stata command ivreg2. To account for overlapping subgroups, (i.e. underaged shopper on Tuesday), each shopper type and time/day of attempt were included as covariates for each sub-group analysis. Finally, we identified predictors of compliance with alcohol service laws using a multi-variable logistic model. Multiple models with various predictors were tested and the final model was selected based on the lowest AIC score.