An observational descriptive study with cross-sectional design was conducted between the 1^st July 2016 and 15^th July 2016 among the adult population (≥ 18 years) of a selected area of the state of West Bengal. A GP, lowest tier of the 3 tier Panchayati Raj System for local administration in India, namely Guchati under Daspur II block of Ghatal Sub-Division at Paschim Medinipur district in the state of West Bengal was selected using multistage sampling technique with the help of function “RANDBETWEEN” in Microsoft Excel [Table/Fig-1].

[Table/Fig-1]:

Multistage Sampling for AUDIT.

From the Guchati GP, Study population were selected using Lot Quality Assurance Sampling (LQAS) technique [18]. There are 11 villages under the Guchati GP [Table/Fig-2]. According to LQAS technique, each village under that GP was considered as a Lot. As per 2011 census total adult population of the Guchati GP was 11596. Considering desired level of confidence at 95% and desired level of accuracy at± 10%, total sample size, as per LQAS, was 96. As the sampling fraction was 0.83% i.e., less than 10%, so 96 individuals were considered to be the final sample size. So, minimum Lot sample size was calculated as 9 individuals. Nine sampling points from every village were located using Arc GIS, a Geographic Information System (GIS) software, random point selection technique. During data collection each point was identified using a GPS (Global Positioning System) enabled android smart phone. One adult individual from each sampling point or nearest to the sampling point was randomly selected for the study purpose using “RAND” function in Epi Info (android smart phone application) [19]. Then He/she was interviewed after taking informed consent.

[Table/Fig-2]:

Guchati Gram Panchayat (GP) under Daspur II block.

Those who are the permanent resident of that area were included in the study and those who never consumed alcohol in their life time were excluded from the study. Ethical approval was obtained from the Institutional Ethics Committee of RG Kar Medical College and Hospital, Kolkata, West Bengal, India.

Total 99 adult persons were interviewed using a pre designed and pretested, Epi Infobased digital schedule. That digital schedule was developed using Epi Info^TM7 software [19] in computer. The schedule had 2 parts, first part consisted of socio demographic and spatial information i.e., village, age, gender, education, employment status, per capita monthly income, and latitude, longitude and second part was AUDIT [17] questions. During data collection real time synchronisation between Epi Info application in smart phone and Epi Info^TM7 software in Windows platform based personal computer, was made using cloud computing (Microsoft Azure) technology. Collected data were recorded automatically into the computer Epi Info^TM7 database in real time manner.

Then recorded data were analysed by the R software and some relevant information were represented using ArcGIS maps [20]. During analysis age was grouped into “less than 40 years” and “More than or equal to 40 years”, education status was grouped into “Illiterate” and “Literate”, employment status was grouped into “Unemployed” and “Employed”. Socio economic status was derived, as per Modified BG Prasad’s social classification adjusted for May 2016 Consumer Price Index, from per capita monthly income (Indian Rupees). Then social class was grouped into “Lower and Lower middle” and “Upper and Upper middle and Middle”. Outcome (dependent) variable i.e., AUDIT scores were grouped based on risk level according to WHO guideline. Total scores of 8 or more are recommended as indicators of hazardous and harmful alcohol use, as well as possible alcohol dependence [17]. So, AUDIT score between 0 to 7 i.e., Zone-I (Low risk drinking or abstinence) risk level was considered as one group and other group was having score between 8 to 40 i.e., Zone-II (8-15) (Alcohol use in excess of low-risk), Zone-III (16-19) (Harmful and hazardous drinking) and Zone-IV (20-40) (Alcohol Dependence) risk level. Appropriate statistical tests like Intraclass correlation, chi-square test, logistic regression and Bootstrapping were applied whenever necessary. Alfa (α) was set at 5% level.

Among 99 adult respondents, there were 54 (54.50%) men and 45 (45.50%) women. Mean (±Standard Deviation) age of the study population was 38.62 (±11.05) (95% CI 36.44 to 40.76) year. Mean (±SD) per capita family income of the study population was 1544 (±476) (95% CI 1454.39 to 1635.13). Among them 79.9% (95% CI 69.9% to 89.9%) were literate, 86.7% (95% CI 76.7% to 96.7%) were employed and 60.3% (95% CI 50.3% to 70.3%) belongs to “lower” and “lower middle” socio economic class as Modified BG Prasad’s social classification adjusted for May 2016 Consumer Price Index.

Mean (±SD) AUDIT score of the study population was 7.11 (±8.11) (95% CI 5.55 to 8.74). AUDIT score were consistent, as evident by Intraclass Correlation Coefficient (ICC) 0.93 (95% CI 0.84 to 0.98), between different villages.

Prevalence of hazardous and harmful use of alcohol among the study population was calculated as per Lot Quality Coverage Survey technique [Table/Fig-3].

Chi-square test for association between socio demographic factors and risk level:

Association between risk level of hazardous and harmful use of alcohol with various socio demographic factors show that there is statistically significant association (χ² = 32.675, p<0.05, α= 5%, df-1) only with the gender of the study population [Table/Fig-4].

After testing the assumptions, binary logistic regression was performed to ascertain the effects of age, gender (women/men), employment status (employed/unemployed), education (literate/illiterate) and per capita family income on the likelihood that study population have in the High Risk level (AUDIT Score 8-40) i.e., in Zone-II, III, and IV risk level of hazardous and harmful use of alcohol. The logistic regression model was statistically significant, χ²(5) = 68.982, p< .000. The model explained 66.9% (Nagelkerke R²) of the variance in pattern of alcohol consumption and correctly classified 87.9% of instances. The Wald test (“Wald” column) is used to determine statistical significance for each of the independent variables. The statistical significance of the test is found in the “Sig” column. From these results it has been noticed that only “gender” (p = 0.00) and “employment status” (p = 0.01) added significantly to the model with adjusted odds ratio of 82.27 (95% CI 18.17-372.58) and 0.13 (0.03 – 0.66) [Table/Fig-5].

A map was developed using the ArcGIS software, showing the distribution of the villages according to risk level group. Mean AUDIT score between 0 to 7 i.e., risk level Zone-I were in the village Chaksultan, Guchati, Juakhali, and Rana. Rest of the villages had mean AUDIT score between 8 to 15 i.e., risk level Zone-II [Table/Fig-6].

[Table/Fig-6]:

Classification of villages according to mean AUDIT score.

From this study, it was estimated that the prevalence of low risk drinking or abstinence (Zone I) was 65.5% (95% CI 55.5% to 75.5%) and the prevalence of alcohol use in excess of low-risk (Zone II) was 17.6% (95% CI 7.6% to 27.6%), and the prevalence of harmful and hazardous drinking (Zone III) was 8.5% (95% CI 0% to 18.5%) and the prevalence alcohol dependence (Zone-IV) was 8.4% (0% to 18.4%). So, the prevalence of hazardous and harmful use of alcohol including alcohol dependence in the study population was 34.5% (95% CI 24.5% to 44.5%). Study conducted by SD Rathod et al., in 2013-14, found that prevalence of hazardous and harmful use of alcohol among drinkers were 42% at Sehore district of Madhya Pradesh in India [6]. In a study, conducted by Pillai A et al., in northern Goa, they found that the prevalence of drinking habit was 74.3% of adult population aged between 30 to 33 years [5]. Current study also revealed that habit and pattern of drinking in high risk level is much higher among men, which is consistent with the study conducted by Rathod et al., in Madhya Pradesh [5,6].

Individual association between various socio-demographic factors and risk level of hazardous and harmful use of alcohol shows that only gender have the statistically significant association (p<0.05). Some other studies found significant association between educational status, employment status and socio economic condition with risk level of hazardous and harmful use of alcohol also [13]. Binary logistic regression model shows that “gender” (p = 0.00) and “employment status” (p = 0.01) added significantly to the model with adjusted odds ratio of 82.27 (95% CI 18.17-372.58) and 0.13 (0.03 – 0.66). This model also reveals that those who were employed, more prone to become high risk level drinker. To compare the results of this study, significant observations of similar studies have been tabulated in [Table/Fig-7] [6,21-24].

There may be some amount of bias in the regression analysis due to confounding variables like stress factor and peer pressure. Moreover, larger cross-sectional study, involving other Gram Panchayats in India with larger sample size would have been closer to generalization.

Keeping the study results in mind, it can be concluded that there is a need for a comprehensive screening and treatment programme to deal with the problems of AUDs among adults to achieve good health and well beings to meet the Sustainable Development Goals (SDG).