Study design and sample

To estimate gross costs of alcohol consumption in Thailand, 2021, a prevalence-based cost-of-illness methodology was employed. Cost-of-illness, also known as economic cost-of-illness or economic cost [14], is a summary of the costs of a particular disease or condition to society [15]. Cost-of-illness aimed to measure the economic burden of the disease or condition imposed on society [14, 16]. The terms economic burden and economic costs are frequently used interchangeably.

A societal perspective was adopted for the analysis. Both direct and indirect costs were estimated. All costs were presented in Thai baht, 2021 (36.14 baht = 1 US$). Then, the estimates were displayed as a percentage of the country’s GDP to facilitate the cross-country comparison without the intention to measure the impact of drinking on the growth of the economy.

In this study, 25 diseases/conditions attributable to alcohol identified by the World Health Organization (WHO) Global Burden of Disease 2020 [17, 18] were included in the analysis. To estimate the number of patients and deaths attributable to alcohol consumption, alcohol-attributable fractions (AAFs) were calculated for each disease, using the following formula [19]:

$$AAF_ \left( } \right) = \frac^ P_ *(RR_ - 1)}}^ P_ * \left( - 1} \right) + 1}}*100$$

where Pj = Prevalence of alcohol consumption at drinking level j, RRij = Relative risk (RR) of developing alcohol-related disease i at drinking level j compared to non-drinker.

In this analysis, the prevalence of alcohol consumption by drinking categories was adopted from the previous study [13], which was estimated from the Smoking and Drinking Behavior Survey 2017 [20]. The drinking categories were classified based on consumption of pure alcohol, measured in grams per day into (1) light alcohol drinking (female > 0 −  < 20 g/day, male > 0 −  < 40 g/day), (2) moderate alcohol drinking (female 20 – 40 g/day, male 40 – 60 g/day), (3) heavy alcohol drinking (female > 40.0 g/day, male > 60.0 g/day) [21]. Information on the RRs of diseases caused by alcohol consumption was derived from the relevant meta-analyses [11, 17, 22,23,24,25]. The AAFs estimated in this study are displayed in Table 1.

Table 1 The Alcohol-attributable Fraction (AAF) used in the studyCost estimatesDirect costs

In this study, direct care costs included healthcare costs, costs of law enforcement, and cost of property damage due to road traffic accidents.

Healthcare costs

Healthcare costs consisted of the cost incurred in the out-patient department (OPD) and the cost incurred in the in-patient department (IPD). Healthcare cost attributable to alcohol was estimated as the product of the number of alcohol-attributable patients and the disease-specific unit cost of treatment. To determine the number of patients attributable to alcohol in 2021, the number of patients with each disease in 2021 was multiplied by the corresponding alcohol-attributable fraction (AAF). The total number of patients for a particular disease and the disease-specific unit cost of treatment in 2021 were derived from the database of National Health Security Office (NHSO), according to the ICD-10 codes. This database covered information on patients under the Universal Health Coverage (UHC) scheme in Thailand. The assumption that the total number of patients covered under the UHC program represented 70% of the overall patient population in the country was employed to estimate the total number of patients in Thailand.

Costs of law enforcement

In this analysis, the costs of law enforcement attributable to alcohol consumption included justice system costs incurred at police stations, court of justice, office of the attorney general, and prison. The costs were calculated by multiplying the number of crimes and offenses attributable to alcohol consumption with the unit cost per case of each state agency.

To estimate the unit cost per case, the annual budget related to the justice work allocated to each agency by the Budget Bureau in 2021 [26,27,28,29,30] was divided by the number of crimes and offenses prosecuted by the agency in the same year. The number of prosecuted crimes and offenses in 2021 was obtained from the annual report of the following agencies: the Royal Thai Police, the Court of Justice, the Office of the Attorney General, and the Department of Corrections, Ministry of Justice [31,32,33,34]. On the other hand, the proportion of crimes and offenses attributable to alcohol consumption was derived from the previous study in Thailand [35], as shown in Table 1.

Cost of property damage due to road traffic accidents

In this study, the cost of property damage due to road traffic accidents attributable to alcohol consumption was estimated by multiplying the total cost of property damage due to road traffic accidents in 2021, which was reported by the Royal Thai Police [36] with the proportion of road traffic accidents attributable to alcohol during New Year and Songkran festival in Thailand, 2021, which reported by the Office of the National Economic and Social Development Council [37, 38].

Indirect costs

Indirect costs included in the analysis were cost of premature mortality and absenteeism from out-patient hospital visits and hospitalization.

Cost of premature mortality

In this analysis, premature mortality was defined as death that occurs before the expected life expectancy of the Thai population. Human capital approach was adopted to estimate the cost of productivity loss due to premature mortality. To estimate the number of alcohol-attributable deaths for each disease, the total number of deaths by age and gender for each disease was multiplied by the corresponding disease-specific AAF. Information on the number of deaths by age and gender was derived from the Strategy and Planning Division, Office of the Permanent Secretary, Ministry of Public Health [39] and was adjusted by the proportion of unidentified causes of death based on the recent study in Thailand [40]. The cost of premature mortality was, then, estimated by multiplying the number of alcohol-attributable deaths, by age and gender with the present value of lost earnings by age and gender. As recommended by the Guideline for Health Technology Assessment in Thailand, Gross National Income (GNI) per capita in 2021 (232,176 baht) [41] was used. The growth rate of 2.94% per year, which was the average growth rate between 2000 and 2020 was applied [42]. The number of years lived beyond each certain age up to the life expectancy was obtained from the WHO life table for Thailand [43]. To convert the future cost into the present value, the discount rate of 3% was adopted [44].

Costs of hospital-related absenteeism

Costs of absenteeism from out-patient hospital visits and hospitalization attributable to alcohol consumption were estimated by multiplying the number of patients attributable to alcohol by the average number of absent days from out-patient visits and hospitalization per patient by the daily wage. The information on the annual number of disease-specific out-patient visits and in-patient length of stay per patient in 2021 was derived from the database of NHSO, which contained ICD-10 codes for each hospital visit. For our study, one out-patient visit was set to a half day loss. Meanwhile, the number of absent days due to hospitalization was assumed to be equal to the length of stay.

Sensitivity analysis

One-way sensitivity analyses were performed to examine the extent to which the results are affected by the choice of parameters and method used in the estimations. In the analyses, discount rates of 0% and 6% were used, as recommended by the Thai Guideline for Health Technology Assessment [45]. The alteration of the prevalence of alcohol consumption (± 10%, ± 20%) and the use of net estimation approach was also examined in the analyses. In addition, the alternative assumptions used in evaluating productivity loss were also explored. At present, Thailand’s official retirement age is 60 years old. Nevertheless, data from Thailand’s Labor Force Survey 2022 [46] shows that slightly less than 50% of Thais aged 60 years and over are in the labor market. In the sensitivity analyses, we, therefore, assumed a 50% drop in income after aged 60 years old. In the analyses, we also assumed that there was no productivity loss after the age of 65 years old, as there is a proposal to extend the retirement age in the country from 60 to 65 years.