1-The Model overview and structure
This study used the Preventable Risk Integrated Model (PRIME) to simulate effects of different scenarios on age-specific and sex-specific mortalities from type 2 diabetes in Iran. The PRIME model is a scenario-based model that links 12 behavioral risk factors such as diet, physical inactivity, alcohol consumption and tobacco consumption with NCD mortality(10). This model have 24 health outcomes including cardiovascular diseases, cancers, kidney disease, liver disease, chronic obstructive pulmonary disease, as well as type 2 diabetes. This framework is parameterized using the best available evidence from meta-analyses of epidemiological studies: prospective cohort studies for links dietary risk factor with NCD mortality(11, 12).
PRIME requires data on prevalence of modifiable risk factors, mortality rates, and population estimates by 5-year age groups and sex, for both baseline and counterfactual scenarios. For this study, PRIME was used to estimate the number and death rate of type2 diabetes that would be delayed or averted in the studied population. The baseline scenario (no policy option) used the current levels of modifiable risk factors in those living in areas, where the counterfactual scenario involved applying the level of risk factors.
2-The model inputs
2-1-Population data
This study used the data from the Prospective Epidemiological Research Studies in Iran (Persian cohort) study results of Azar Cohort (azarcohort.tbzmed.ac.ir) in Shabestar, East Azerbaijan province (North West of Iran) in Iran(13). The AZAR Cohort Study is a prospective study assessing the risk factors related to the most prevalent NCDs in East Azerbaijan province. The more details about the Azar cohort study has been explained in other published article .(14) The inclusion criteria were: (i) permanent residence in Shabestar district (minimum of 9 months); (ii) written informed consent; (iii) at least one Azeri parent; and (iv) age between 35 and 70 years at the time of enrolling in the study. Exclusion criteria were: (i) having a diagnosis of a disabling psychiatric disorder; and (ii) having a diagnosis of a disabling physical illness.
2-2-Dietary intake assessment
Nutritional status was assessed using Food Frequency Questionnaire (FFQ). The FFQ assessment asks about consumption of food and drink during the past year. The questionnaire includes 125 food items appropriate for the Iranian population, including bread, cereal, grains, meat and meat products, milk and dairy products, vegetables, fruits, types of oil and oilseeds, sugar, miscellaneous food products, spices and food supplements. Also some local foods were added into the questionnaire. The nutrition questionnaire also asks about cooking methods, food preservation, food storage, cooking styles and use of herbal medicines and drinks.
To help the respondent's memory and increasing accuracy and precision of participants responds, household scales, including glasses, teaspoons, tablespoons, and colored photographs of portion size were used during the interview. All participants provided an answer to their frequencies of food intake (daily, weekly, monthly and annually) according to standard portion size for each food item, then each participant’s reported intake was transformed to weight using standard Iranian household measures. Dietary information was converted to energy and nutrients using revised Nutritionist IV software [Nutritional Database Manager 4.0.1, Nutritionist IV, version 3.5.2](15). The data of 12126 participants from Azar cohort study were used for 2015 years. The general characteristics of population including age, sex, marital status, education, and SSB consumption are reported in Table 1. SSBs were defined as any sugar sweetened sodas, fruit drinks, sports/energy drinks, sweetened iced tea, or homemade SSBs, which contained at least 50 kcal per 8-oz serving, with 100% fruit juice being excluded.
2-3-Diabetes definition
The prevalence of type 2 diabetes were determined according to self-reporting of participants.
2-4-Effects of SSB Intake on Diabetes Mellitus
Effects of SSB consumption on diabetes mellitus were based on a meta-analysis of 8 prospective cohorts with a total of 310819 participants and 15043 cases of type 2 diabetes mellitus. In this meta-analysis, individuals in the highest category of SSB intake (1–2 servings/d) had a 26% greater risk of developing type 2 diabetes mellitus in comparison with those in the lowest category of SSB intake (none or <1 serving per month; risk ratio, 1.26; 95% CI,1.12–1.41). The association between SSB intake and risk of type 2diabetes mellitus in this meta-analysis was consistent across sex and ethnic groups, which included blacks, whites, and Asians. Although there was heterogeneity across the studies (I2=66%), all but one showed positive associations between SSB intake and risk of type 2 diabetes mellitus, with the strength of the association increasing with the study size and duration(4).
2-5-Cause-Specific Mortality by Age and Sex
According to the Derakhshan et al.’s study, the incidence rate of type 2 diabetes in Iranian men and women was 9.36 and 10.1 per 1,000 persons annually respectively. Note that these rate were calculated separately for different individuals according to age and sex(16).
The mortality rate of type 2 diabetes in the study population in 2015 was based on the data from the National and Subnational Burden of Diseases, Injuries, and Risk Factors in Iran (NASBOD)(1).
2-6-Selected Policy Scenario
According to our earlier report about priotrizing policy options related to reducing the burden of non-communicable diseases in Iran(17), we selected several policy scenarios for the simulation:
1- No policy scenario: assuming the current situation is maintained and no special policy is implemented, the rate of change in the SSB consumption and disease burden during the 20 years was simulated.
2- Applying 10% tax on SSB:According to Afshin et al.’s meta-analysis, a 10% increase in the price of SSB would reduce the consumption by 7%. Thus, the amount of sugar sweetened beverages consumed in a given year (for example, 2016) for each age and sex was estimated(18). Also, given the differences in consumption levels in response to tax across different socio-economic situations, individuals were divided into five groups based on different socio-economic where the changes in the SSB consumption following excising 10% tax were considered within 4-10%(19).
3- Replacing SSB with water:In this scenario, the SSB has been replaced with water, where zero consumption of these drinks has been considered(20, 21).
4- Changing the reformulation of SSB through reducing the sugar content of SSB by 30%(22).
5- Applying 10% tax on SSB along with changing the formulation of SSB via reducing the sugar content of sugary drinks by 30%.
3-Model output
3-1-Estimation of Deaths Prevented or Postponed (DPP)
This model has been used to estimate the number of deaths prevented or delayed because of type 2 diabetes after implementing a specific policy. According to the methodology of a similar study to obtain preventable or delayed death in the case of implementing a policy option in an age and sex group, the desired command code was written based on the following mathematical calculation and implemented in the software. As each policy option increases or decreases the amount of nutrients consumed, the effect of changes in the consumption on reducing or increasing the risk is calculated using the following formula:
![](data:image/png;base64,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)
Where, RR= Relative risk between SSB consumption level in baseline and type 2 diabetes extracted from meta-analyses; A=current amount of SSB consumed in the target state in terms of serving per day in the age and sex group; B= The amount of SSB consumption after implementing the desired policy option in terms of serving per day. The number obtained from the above formula is multiplied by the number of deaths due to type 2 diabetes while no policies are implemented. Also, the number death postpone or prevented under specific scenario were calculated by subtracting the total number of deaths due to the type 2 diabetes under specific scenario from the number deaths due to the type 2 diabetes under no policy scenario(19). This study was approved by Ethics Committee in Shahid Beheshti University of Medical Sciences (Ethics No. IR.SBMU. NNFTRI.1397.056).
4-Analyses
We used Monte Carlo simulation to quantify the uncertainty in the attributable deaths from SSB intake data (which includes both measurement and sampling error as well as modeling uncertainty), and uncertainty from the relative risks in our final estimates. Each policy scenario simulated 1000 times and mean and standard error of iteration were reported. For each mean exposure, population-representative standard deviations were predicted using coefficients from regressions performed on all available dietary survey data in our collection, where the standard deviation was the dependent variable while the mean was the independent variable(23, 24). All analyses were performed using R and Python software, version 2.15.0.