Study participants: This hospital-based case-control study was conducted on newly diagnosed patients (maximum one month elapsed since the detection) in Kabul, Afghanistan in 2019. Using the formula suggested for case-control studies[37], and hypothesizing that almost 50% of Afghan population were consuming inflammatory diet[38], we calculated required sample size. Given that those with inflammatory diet would have 2.5 times greater risk of gastric cancer[34, 36], we considered the odds ratio of 2.5. We assumed the type I error of 5%, study power of 80% (β = 0.20) and the ratio of controls to cases as 2. Therefore, the required sample was calculated to be 81 cases and 162 controls. However, we recruited 90 patients with gastric cancer and 180 controls to account for probable drop-outs. Cases were chosen by using convenience-sampling method from the patients with pathologically confirmed gastric cancer during the prior month that had been referred to Jamhuriat Hospital, Kabul, Afghanistan. Cases were aged between 20 and 75 years. The control participants were aged between 20 and 75 years and were relatives of patients. Cases and controls were matched in terms of age (± 5) and sex. The individuals with a history of any type of pathologically confirmed cancers (except gastric cancer) and those with a history of chemotherapy or radiotherapy (due to cancer) were not included in the study. All cases and controls provided written informed consent. To conduct the study, permission was first acquired from authorities to get into the hospital. Then the details of the study were explained to the hospital chairman. Cases and controls, that met the inclusion criteria were asked to voluntarily participate in the study. The study was ethically approved by the TUMS Ethics Committee (code: 1398.460).
Assessment of dietary intakes
Common dietary intakes of participants during a year before the diagnosis of gastric cancer in cases and during a year before the interview in controls were examined by a pre-tested Willett-format food frequency questionnaire (FFQ). We used a specific FFQ that was designed for the current study. This FFQ was consisted of food items, with standard portion sizes, usually consumed by Afghan people. A trained interviewer administered the FFQ through face to face interviews at the presence of individuals who were involved in the preparation and cooking of foods. All reported consumption frequencies were converted to grams per day by using household measures. Daily intakes of energy and nutrients were calculated for each person by using the US Department of Agriculture food consumption database.
Construction of DII: Dietary data derived from FFQ was used to calculate DII scores for all subjects. We calculated DII score, using the approach developed by Shivappa et al[21], based on 29 food parameters (instead of 45 items that was used by Shivappa et al.). We did not consider some items, suggested by Shivappa et al[21], because some items were not available in our data set. The items we used in DII construction include: energy, carbohydrate, fat, protein, fiber, cholesterol, mono-unsaturated fatty acids (MUFAs), poly unsaturated fatty acids (PUFAs), saturated fat (SFAs), vitamin B12, vitamin B6, folic acid, niacin, riboflavin, thiamin, vitamin A, vitamin D, vitamin E, β-carotene, vitamin C, zinc, selenium, magnesium, iron, caffeine, pepper, onion, garlic and green/black tea. To do this, first energy-adjusted amounts of all 29 nutrients were calculated using residual method[39], to avoid misclassification. Then, we calculated the z score for 29 food parameters by subtracting the “standard global mean” from the amount consumed by each subject. The obtained value was then divided by the “global standard deviation”, reported by Shivappa et al[21]. To reduce skewness, we then converted this value to a centered percentile score. The DII score or each food item was then computed by multiplying this value to the respective food parameter effect score reported by Shivappa et al[21]. The overall DII score for each participant was calculated by summing up all foods’ DII scores. A higher DII score indicates a more inflammatory diet and a lower DII score indicates a less inflammatory diet.
Assessment of other variables
To collect information about covariates, a pretested questionnaire was used, in which data on age, sex, marital status, place of residence, education, job, family history of cancers, smoking status, ethnicity, supplements use, drug usage, H. pylori infection, medicine use, socio-economic status, drinking tea, history of diabetes, consumption of kebab food, fried foods, outdoor food usage, boiled food usage, oils used, salt intake, fatty foods, intra-meal water drinking, intra-meal drinking of other beverages were collected. Short form of International Physical Activity Questionnaire (IPAQ) was used for measuring physical activity of participants through face-to-face interview. All results of the IPAQ were then expressed as Metabolic Equivalents per week (METs/week). Weight was measured to the nearest 0.1 kg using a digital scale with minimal clothes and without shoes (Seca, Hamburg, Germany). Height was measured to the nearest 0.1 cm in a standing position, without wearing shoes, using a tape measure and shoulders touching the wall and looking straight forward. Body mass index (BMI) was calculated as weight (kg) divided by height squared (m2). All measurements were completed by a trained dietitian.
Statistical analysis
Cases and controls were compared in terms of several variables related to general characteristics. To do this, we applied Student’s t test and chi-square, where appropriate. Then, tertile cut-off points of DII score were derived based on control participants. This was done to reduce error due to the possibility of changed dietary intakes in patients with gastric cancer. All study participants were categorized based on these cut-off points. Comparison of general characteristics across tertiles of DII was done using one-way ANOVA for continuous variables and chi-square test, for categorical variables. Age- and gender-adjusted intakes of foods and nutrients across tertiles of DII were computed using General Linear Model and compared using ANCOVA. Binary logistic regression was applied to examine the association between DII and gastric cancer in different models. First, we controlled for age (continuous) and sex (male/femal). Further adjustments were made for physical activity (categorical), and family history of cancer (yes/no), SES (low, middle and high), consumption of barbecued foods (categorical), smoking (categorical), drugs use (yes/no) and H. pylori infection (yes/no).
We also adjusted for BMI in the last model to identify obesity-independent association. In all these analyses, the first tertile of DII was considered as the reference category and odds ratios and 95% CIs in other tertiles were calculated. The overall trend of ORs across increasing tertiles of DII was examined by considering tertiles of DII as an ordinal variable. All the statistical analyses were carried out using SPSS (SPSS Inc., version 24). P values were considered significant at < 0·05.