Reproducibility of a designed Semi-Quatitative food frequency questionaire in general populations in the north Viet Nam

Southeast-Asian J. of Sciences Vol. 6, No. 2 (2018) pp. 191-200 REPRODUCIBILITY OF A DESIGNED SEMI-QUATITATIVE FOOD FREQUENCY QUESTIONAIRE IN GENERAL POPULATIONS IN THE NORTH VIETNAM Le Tran Ngoan1,5∗, Le Xuan Hung1, Pham Van Phu1 Nguyen Quang Dung1, Tran Hieu Hoc1, Pham Thi Oanh2, Nguyen Thanh Binh3, Dinh Thi Minh1, Nguyen Thu Hoai1, Pham Thi Ha Phuong1, Trinh Thi Duc Hanh1, Lam Van Chien1, Phan Van Can4 1 Hanoi Medical University, Vietnam; 2 National Cancer Hospital, Vietnam; 3 Bac Ninh province office of food safety and hygiene, Vietnam; 4Viet Nam general statistical office, Vietnam; 5School of Medicine, International Univ. of Health and Welfare, Japan e-mail: letngoan@hmu.edu.vn Abstract Objectives: To examine a reproducibility of a semi-quantitative food frequency questionnaire (SQFFQ) in general populations in the North Viet Nam in 2017. Methods: A Cross-sectional survey using methods of 24-hour recall records of food intake (24-FR, the reference group) in 300 households of three commune general populations that have 1,334 persons. SQFFQ was tested for 600 participants which found 310 partic- ipants eligible in linking with persons of 300 households. The R2, which was used to compare the energy and estimated nutrients obtained by both methods, was calculated before and after adjustment for total en- ergy intake, age and sex, performed in STATA 10. Results: For macro 5∗Corresponding author. Key words: SQFFQ, validation, general populations, Viet Nam. 191 192 Reproducibility... nutrients intake estimated by two methods of 24-FR and SQFFQ, the intake amounts consisted of energy (kcalo, 1934.5 versus 1901.4); protid (gram, 87.8 versus 81.4); lipid (gram, 43.5 versus 41.7); and glucid (gram, 295.5 versus 294.5). there is a small correlation for lipid (adjusted R2= 0.20); a moderate correlation for protid (adjusted R2= 0.38) and glu- cid (adjusted R2= 0.36), a strong correlation for energy (adjusted R2= 0.53). Conclusions: The designed SQFFQ was being validated having good characteristics of feasibility, practical and reliability in the general populations located in the North Viet Nam. Background Characteristics of data collection tool of epidemiological studies have ability to reproduce but also have a considerate sensitivity, thus a specific study is essentially required [6]. To address the association between diet and cancer, a semi-quantitative food frequency questionnaire, (SQFFQ) has been designed and used to collect data [8]. The reproducibility of SQFFQ has been validated in the United States [3] and in Japan [8]. The designed SQFFQ has also been repeatedly used for a periodic every 4-year from 1976 to date to collect data of dietary habits of study participants [4]. Applying the methods of developing SQFFQ of the Japanese [7, 8] and the U.S. at the Harvard T.H. Chan School of Public Health [3, 12], the SQFFQ was developed in general populations located in the Red River Delta in the Northern Viet Nam. In brief, the development of database of nutritional status of 300 households by 24-hour recall records of food intake was completed. Then based on a contribution of major nutrients up to cumulative 90%, food items were selected. Average amount intake and standard deviation was estimated for each selected food item. Number of food items were selected respectively for beer and liquor (3); fat and oil (2); a polished rice (1); cereal and products (6); nuts and seeds (4); vegetables (15), fruits (14), meat (13); seafood (13), condiments and sauces (5); milk and eggs (6); dried fishes, sausage, spring-roll (3). In total, there were 85 food items included in the SQFFQ. Food frequency intake was designed to collect dietary habits during the last year, included 7 levels of frequency, such as seldom or never intake, 6-11 times/year, 1-3 times/month, 1-2 times/week, 3-4 times/week, 5-6 times/week, and 1-3 times/day. Objective: To examine a reproducibility of the SQFFQ in general popu- lations in the North of Viet Nam in 2017. Le Tran Ngoan et al. 193 I. Methods 1. Design A Cross-sectional survey using 24-hour recall records of food intake (24- FR, the reference group) and SQFFQ (being tested) were applied [3, 8, 12]. Study sites were at 3 communes represented 3 provinces located in the Red River Delta in the North Viet Nam, named Hung Yen, Ha Nam, Bac Ninh (in proximity to the Hanoi city). Study time was in the 4th quarter of 2017. Sample size was calculated using STATA 10. Referred to the result of the latest Viet Nam National Nutritional Survey in 2009-2010, the estimated energy intake (mean±S.D) was 1925 kcalo±587 [1]. Assume: a=0,05 (two sides); a=0,90; m1=2005 kcalo, m2=1845 kcalo; sd1=617, kcalo; sd2=557 kcalo; n1/n2=1; 24-FR completed 1 time, completed SQFFQ was 1 time; the desired R2 = 0.50. The STATA statement command of sampsi 2005 1845, sd1(617) sd2(557) method(change) pre(1) post(1) r1(.5), the required samples were 284 study participants. 2. Conducted 24-hour recall records of food intake A household survey of 24-hour recall records of food intake was conducted in 3 consecutive weekdays for 300 households in Hung Yen, Ha Nam and Bac Ninh province by using the designed questionnaire from the National Nutrition Insti- tute and its guidelines in combination with in person interview at participants house by the trained interviewers. All households members were registered for their breakfast (main=1/addition=2); lunch (main=3/addition=4); dinner (main=5/addition=6); places of intake could be at home or outside. A ration was weighted by age and sex for each family member, then sum up for total rations daily and for three days. Every food item in each meal was registered its weight(gram), then converted into a clean-raw weight(gram). All clean-raw food items were converted into 86 nutrients for every household, for one single day and for three days in average. Total amount intake of each nutrient of each household was divided to number of rations to obtain an average intake of nutrients per person per day (one ration). The participants of 300 households consumed 257 food items during three days. After excluded one household having incomplete survey, total el- igible 299 households with 1,334 persons had successfully completed 24-hour recall records of food intake. During the three days, 3,989 person-day were registered and the number of rations was 3,417. We excluded 278.0 rations that had breakfast or lunch or dinner outside of their home. The remaining is 3,138 rations that had breakfast, and lunch and dinner at their home. 3. Conducted a nutritional survey using the designed SQFFQ After completion of the household survey of 24-hour recall records of food 194 Reproducibility... intake for 3 consecutive weekdays, we conducted a nutritional survey using the designed SQFFQ by in person interview at participants households by the trained interviewers. The study participants comprised 600 persons recruited from 3 communes represented 3 provinces Hung Yen, Ha Nam, and Bac Ninh (located around the Hanoi city). The SQFFQ inquired dietary habits dur- ing one previous year (for one year) for 85 foods/recipes and frequency in 7 categories: seldom or never intake, 6-11 times/year, 1-3 times/month, 1-2 times/week, 3-4 times/week, 5-6 times/week, and 1-3 times/day. Nutrients intake per person per day were estimated by the following formulation, which based on the Vietnamese food composition table, version 2007. n = f∗n∗a(w/100) ∗k/365 Where f was frequency intake during last year; na was amount of nutrient per 100 gram; w was average weight intake for one time; k was weighted size of intake (small=0.8, medium=1.0, large=1.2). 4. Linked participants ID of two studies We linked persons who completed the SQFFQ with persons who partici- pated in the 24-hour recall records of food intake. Inclusion criteria was persons had fully participated three meals, that was breakfast (main=1/addition=2); lunch (main=3/addition=4); and dinner (main=5/addition=6) at their home. As the result, 310 participants were included in the final model of data analysis. 5. Data analysis The food intake registered by the 24-hour recall records of food intake and the SQFFQ were converted into energy and 85 nutrients (based on the Viet- namese food composition table version 2007) The R2, which compared the en- ergy and nutrient values obtained by both methods, was calculated before and after adjustment for total energy intake, age and sex, performed in STATA 10. The absolute value of R2 indicating the strength of relationship was grouped as following labels: 0− < 0.1: none or very weak relationship, 0.1− |R2| < 0.25: weak relationship; 0.25 − |R2| 0.5: strong relationship [5]. Due to lack of data in the Vietnamese food composition table, version 2007, we did not estimate the correlation R2 when the estimated amount of nutrients was to be one unit or lower. II. Results The study comprised 310 participants (men 153, women 157), who completed both 24-hour recall records (24-FR) of food intake and SQFFQ. Men and women combined for age, the youngest was 12 and the oldest was 81; a mean age was 47.6. For body high (cm), the lowest was 145 and the highest was 182, Le Tran Ngoan et al. 195 an average was 160.5. For body weight (kg), the lowest was 35, highest was 85, an average was 55.4 which illustrated in the Table 1. Table 1. Characteristics of study participants Note: there were un-known data of age of four participants and high and weight of three participants; S.D.: Standard deviation; Min: Minimum; Mdn: median; Max: Maximum. For macro nutrients intake estimated in two methods of 24-FR and SQFFQ, the amounts consisted of energy (kcalo, 1934.5 versus 1901.4); protid (gram, 87.8 versus 81.4); lipid (gram, 43.5 versus 41.7); and glucid (gram, 295.5 ver- sus 294.5). Similar observation was seen for major nutrients of other groups of vitamin, mineral, fatty acids, and amino acids (Table 2). Due to lack of data in the Vietnamese food composition table, version 2007 and the esti- mated amount intake was too small, the nutrients were not analyzed for the correlation R2, as following: galactoza, maltoza, lactoza, B2, B9, D, glycetin, margaric, arachidic, behenic, lignoceric, myristolei, palmitolei, linolenic, arachi- donic, eicosapen, dososahex, sumtrans, and sacaroza (Table 2). For the macro nutrient, there was a small correlation (relationship) for lipid (adjusted R2= 0.20); a moderate correlation for protid (adjusted R2 = 0.38) and glucid (adjusted R2 = 0.36), a strong correlation for energy (adjusted R2 = 0.53) (Table 2 and Figure 1). For the sugar group, the very weak correlation was seen (adjusted R2 = 0.05), fructoza (adjusted R2 = 0.04), and glucoza (adjusted R2 = 0.03). Similar observation for mineral such as calci (adjusted R2 = 0.06), mangan (adjusted R2 = 0.04), natri (adjusted R2 = 0.03); For vitamin groups, in detailed of vitamin C (adjusted R2 = 0.06), B12 (adjusted R2 = 0.03), A (adjusted R2 = 0.07), D (adjusted R2 = 0.04), K (adjusted R2 = 0.04), bcaroten (adjusted R2 = 0.04), acaroten (adjusted R2 = 0.01); for folat (adjusted R2 = 0.07) (Table 2). A small (weak) correlation was seen for vitamin group of B1 (adjusted R2 = 0.20), B2 (adjusted R2 = 0.18), PP (adjusted R2 = 0.23), and vitamin H (adjusted R2 = 0.22). A moderate correlation was seen for mineral group of phosphor (adjusted R2 = 0.38), kali (adjusted R2 = 0.32), zinc (adjusted R2 = 0.33), selen (adjusted R2 = 0.34); for vitamin group of B5 (adjusted R2 = 0.37); for amino group of methionine (adjusted R2 = 0.31), tryptophan (adjusted 196 Reproducibility... Table 2. Estimated nutrients intake of 24-hour re- call records and SQFFQ, and their correlations Le Tran Ngoan et al. 197 198 Reproducibility... R2 = 0.31), phenylalan (adjusted R2 = 0.36), threonine (adjusted R2 = 0.34), valin (adjusted R2 = 0.35), leucine (adjusted R2 = 0.34), isoleucine (adjusted R2 = 0.30), arginine (adjusted R2 = 0.35), histidine (adjusted R2 = 0.32), cystin (adjusted R2 = 0.35), tyrosin (adjusted R2 = 0.32), alanine (adjusted R2 = 0.30), glutamic (adjusted R2 = 0.32), (Table 2). Figure 1. Scatterplots and correlation between 24-hour recall records and SQFFQ for energy (kcal, adjusted R2=0.53), protid (protein, adjusted R2=0.38), glucid (glucid, adjusted R2=0.36), and Vitamin B5 (b5, adjusted R2=0.37) III. Discussions The reproducibility of the designed semi-quantitative food frequency question- naire (SQFFQ) in general populations in the North of Viet Nam was validated and the findings show that a strong positive correlation for energy and a mod- erate positive correlation for major macro nutrient of protid and glucid and number of minerals, amino acids. Among vitamins, only a moderate positive correlation was seen for the vitamin B5. The listed macro nutrients and listed micro nutrients having a strong or major positive correlation have highlighted the designed SQFFQ being feasibility, practical and reliability in the general participated populations. These findings in here have consisted with the studies performed in Japanese population for the designed SQFFQ [7, 8]; and stud- ies performed in the United State at the Harvard T.H. Chan School of Public Health and Harvard School of Medicine. The designed SQFFQ was used to Le Tran Ngoan et al. 199 collect data for the Nurses Health Study (from 1976 to date), the Health Pro- fessionals Follow-up Study (from 1986 to date) and the Nurses Health Study II (from 1989 to date) [3, 12]. The designed SQFFQ has ability to reproduce in term of energy, protid, glucid, zinc, selen, and some major amino acids, vitamins and mineral that are related as protective factors against cancer in the human [13, 14]. The present findings also suggest promoting observation study on the association between diet and cancer to explore the underlying of the association between them in low- and middle-income countries in general and in Viet Nam. This possibility can progress in understanding the mechanism of the development of cancer (diet related to about 35% of all cancer [2]), and other non-communicable diseases (NCD) of diabetes and cardiovascular diseases relating to diet. Unhealthy diet has been recognized being major risk factors of NCD, which occurred in low- and middle-income countries as much as about 80% worldwide [9-11]. The present work has calculated the adjusted R2 for 67 of 86 nutrients according to Vietnamese food composition table, version 2007 thus it will promote the advanced study in the near future in the country and in the region for the association between diet and NCD. The present study certainly has a limitation, such as about 19 nutrients were unable to analysis due to missing of the data in the Vietnamese food compo- sition table, version 2007 and group of sugar and fatty acids showing a null or small correlation between 24-hour recall records of food intake and the de- signed SQFFQ. This minor correlation might be generated from the database collected from one survey only due to the limited constraint time of work. The next survey is initiating and new findings are expected to be published soon. Conclusion The designed semi-quantitative food frequency questionnaire (SQFFQ) was be- ing validated has positive characteristics of feasibility, practical and reliability in general populations located in the Red River Delta in the North of Viet Nam. The SQFFQ can be used to collect data for the analytical epidemiologi- cal studies of cancer and other non-communicable diseases. Acknowledgements. We would like to thank the participants and staffs of the Da Trach, Dong Du and Lien Bao commune health stations for the assistance in performance nutritional surveys for this study. Funding. This research was partly funded by the World Bank Foreign Talents STI Grants, Agreement No.: 18/FIRST/1a/HMU. Conflicts of Interest. None to declare 200 Reproducibility... References [1] “Ministry of Health, National Institute of Nutrition, UNICEF, National Nutritional Survey 2009-2010”, in: Le Thi Hop, Le Danh Tuyen, (Eds.), National Institute of Nutrition, Hanoi, Medical Publishing House, 2010, pp. 96. [2] R. Doll, R. Peto, The causes of cancer: Quantitative estimates of avoidable risks of cancer in the united states today, JNCI 66 (1981), 1193-1265. [3] D. Feskanich, E.B. Rimm, E.L. Giovannucci, G.A. Colditz, M.J. Stampfer, L.B. Litin, W.C. Willett, Reproducibility and validity of food intake measurements from a semi- quantitative food frequency questionnaire, J. Am. Diet Assoc. 93 (1993), 790-796. [4] G. Liu, G. Zong, K. Wu, Y. Hu, Y. Li, W.C. Willett, D.M. Eisenberg, F.B. Hu, Q. Sun, Meat Cooking Methods and Risk of Type 2 Diabetes: Results From Three Prospective Cohort Studies, Diabetes Care 41 (2018), 1049-1060. bibitem[5] D.S. Moore, W.I. Notz, M.A. Flinger, “The basic practice of statistics”, 6th ed., Freeman and Company, New York, NY: W. H., 2013. [5] I.D.S. Silva, “Cancer epidemiology: Principle and methods”, IARC - WHO, Lyon, France, 1999. [6] S. Tokudome, M. Ikeda, Y. Tokudome, N. Imaeda, I. Kitagawa, N. Fujiwara, Develop- ment of data-based semi-quantitative food frequency questionnaire for dietary studies in middle-aged Japanese, Jpn. J. Clin. Oncol. 28 (1998), 679-687. [7] Y. Tokudome, C. Goto, N. Imaeda, T. Hasegawa, R. Kato, K. Hirose, K. Tajima, S. Tokudome, Relative validity of a short food frequency questionnaire for assessing nutri- ent intake versus three-day weighed diet records in middle-aged Japanese, J. Epidemiol 15 (2005), 135-145. [8] WHO, “Global NCD Network: a new network to combat non-communicable diseases”, World Health Organization, Geneva, Switzerland, 2009. [9] WHO, Non-communicable Diseases Progress Monitor 2015. Publication date: Septem- ber 2015. ISBN: 978 92 4 150945 9. progress-monitor-2015/en/, World Health Organization, Geneva, Switzerland, 2015. [10] WHO, World Health Organization Non-communicable Diseases (NCD) Country Pro- files of Viet Nam, 2014., World Health Organization, Geneva, Switzerland, 2018. [11] W.C. Willett, L. Sampson, M.J. Stampfer, B. Rosner, C. Bain, J. Witschi, C.H. Hen- nekens, F.E. Speizer, Reproducibility and validity of a semiquantitative food frequency questionnaire, Am. J. Epidemiol 122 (1985), 51-65. [12] World Cancer Research Fund, Food, Nutrition, Physical Activity, and the Prevention of Cancer: a Global Perspective, World Cancer Research Fund / American Institute for Cancer Research, Washington DC AICR, 2007. [13] World Cancer Research Fund, Continuous Update Project, Colorectal Cancer 2011 Re- port, World Cancer Research Fund / American Institute for Cancer Research, Wash- ington DC AICR, 2011.nt of