Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Fortified fish sauce : a novel means of improving thiamin status in rural Cambodia Whitfield, Kyly C 2016

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2016_may_whitfield_kyly.pdf [ 15.98MB ]
Metadata
JSON: 24-1.0300419.json
JSON-LD: 24-1.0300419-ld.json
RDF/XML (Pretty): 24-1.0300419-rdf.xml
RDF/JSON: 24-1.0300419-rdf.json
Turtle: 24-1.0300419-turtle.txt
N-Triples: 24-1.0300419-rdf-ntriples.txt
Original Record: 24-1.0300419-source.json
Full Text
24-1.0300419-fulltext.txt
Citation
24-1.0300419.ris

Full Text

 i FORTIFIED FISH SAUCE: A NOVEL MEANS OF IMPROVING THIAMIN STATUS IN RURAL CAMBODIA  by  Kyly C Whitfield  B.Sc. (Hons), The University of Guelph, 2010 M.Sc., The University of Guelph, 2012  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES (Human Nutrition)  THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver) April 2016  © Kyly C Whitfield, 2016    ii Abstract  Background: Infantile beriberi, a consequence of maternal thiamin deficiency, is not uncommon in Cambodia. The Cambodian diet consists largely of thiamin-poor, polished white rice and contains few thiamin-rich foods. Objectives: 1) Determine the thiamin status, assessed by erythrocyte thiamin diphosphate concentration (eTDP), among Cambodian women of childbearing age; 2) develop a stable and acceptable thiamin-fortified fish sauce; and 3) test the efficacy of thiamin-fortified fish sauce to increase eTDP among two groups in rural Cambodia: i) women and their youngest child aged 12-59 mo, and ii) pregnant and lactating women and their breastfed infants.  Methods: eTDP was determined in samples of women 20-45 y from Prey Veng (n=121) and Phnom Penh (n=117), Cambodia, and for comparison, Canada (n=47). Thiamin stability in fish sauce was assessed under various conditions, and acceptability was determined through sensory evaluation. Women (18-45 y; n=354, 276 non-pregnant and non-lactating, and 78 pregnant and lactating) and their families in Prey Veng, Cambodia were randomized to: control, low (LC, 2 g/L) or high (HC, 8 g/L) concentration thiamin-fortified fish sauce. Results: Mean ± SD eTDP was significantly lower among women in Prey Veng (149 ± 36 nM) than Phnom Penh (156 ± 32 nM), which, in turn, was lower than in Vancouver, (179 ± 37 nM; P<0.05). Thiamin was stable in fish sauce, and accepted by Cambodian women. Among non-pregnant women, endline eTDP (mean; 95% CI) was higher among those in LC (232; 220, 244 nM) and HC (231; 219, 244 nM) groups versus control (175; 163, 188 nM; P<0.001). Similar results were found in children (P<0.05). Endline eTDP was higher among lactating mothers in LC (276; 246, 306 nM) and HC (238; 207, 268 nM) groups versus control (194; 163, 224 nM; P<0.05). Infants of mothers in HC group had higher eTDP (257; 215, 298 nM; P<0.05) versus LC (205; 175, 235 nM) and control (181; 153, 210 nM) groups.  Conclusions: Thiamin-fortified fish sauce is an efficacious means of improving dietary thiamin intake and biochemical thiamin status in rural Cambodia, and as such highlights downstream potential to reduce mortality from a totally preventable disease, infantile beriberi.   iii Preface This research is the result of a collaborative effort. Though I took the lead on study design, protocol preparation, data collection, statistical analysis, and dissemination, this research would not have been possible without the support of many faculty and staff from the University of British Columbia, Helen Keller International, and the Cambodian Ministry of Health and Ministry of Planning.  This research grew out of a research partnership between the University of British Columbia and Helen Keller International – Cambodia, and I gratefully acknowledge the contributions of my colleagues. I led a successful grant application to Grand Challenges Canada (Stars in Global Health) and developed study protocols in collaboration with my doctoral supervisory committee members from the University of British Columbia, Drs. Tim Green, David Kitts, Eunice Li-Chan, Larry Lynd, and Judy McLean, Helen Keller International partners Mr. Aminuzamman Talukder, Mr. Hou Kroeun, Ms. Ly Sokhoing, Dr. Ame Stromer, and Mr. Keith Porter, and Cambodian government partners Dr. Prak Sophonneary from the Ministry of Health and Mr. Mam Borath from the National Sub-Committee for Food Fortification, Ministry of Planning. I oversaw data collection, quality control and data management in-country (conducted by the Cambodia-based research team) with Mr. Sim Chhoeun, Mr. Phon Phearom, and Mr. Rem Ngik (Helen Keller International). Laboratory assessment of eTDP was conducted by Mr. Benny Chan (UBC), and breast milk thiamin was analyzed by Drs. Daniela Hampel and Lindsay Allen (Agricultural Research Service, United States Department of Agriculture, University of California, Davis). I performed all data analysis and writing in close collaboration with my supervisory committee.   All studies documented in this dissertation obtained ethics approval in both Canada and Cambodia: the University of British Columbia Clinical Research Ethics Board (H12-02847, H13-01319, H14-02173, H14-00103), the University of British Columbia – Children’s and Women’s Health Centre of British Columbia Research Ethics Board (CQ14-0204/H14-01654), and the Cambodian National Ethics Committee for Health  iv Research (0004NECHR and 0245NECHR). The randomized controlled efficacy trial was registered with clinicaltrials.gov (NCT02221063).   A version of Chapter 2 has been published: Whitfield KC, Karakochuk CD, Liu Y, McCann A, Talukder A, Kroeun H, Ward M, Lynd LD, Kitts DD, Li-Chan ECY, McLean J, Green TJ. Poor thiamin and riboflavin status is common among women of childbearing age in rural and urban Cambodia. J Nutr 2015;145:628-33 and Whitfield KC & Green TJ. Erratum for Whitfield et al. Poor thiamin and riboflavin status is common among women of childbearing age in rural and urban Cambodia. J Nutr 2015;145:628-33. J Nutr 2016;146:147-8. “KCW, TJG, JM and AT designed research; KCW drafted the research protocol and TJG, JM, AT, LDL, DDK, and EYCL-C reviewed and edited the final protocol; KCW conducted research; HK and AT provided essential logistic support for study execution; YL, DDK, and ECYL-C modified and executed method for measuring TDP; AM, MW, and HM advised measurement of riboflavin, and developed the method and measured EGRac; KCW and CDK analyzed data; KCW wrote the manuscript; KCW and TJG had primary responsibility for final content. All authors read and approved the final manuscript.”      v Table of Contents Abstract .......................................................................................................................................... ii Preface ........................................................................................................................................... iii Table of Contents ...........................................................................................................................v List of Tables ................................................................................................................................ xi List of Figures .............................................................................................................................. xii List of Abbreviations ..................................................................................................................... i Acknowledgements ...................................................................................................................... iii Chapter 1: Introduction, literature review, and research objectives and hypotheses.............1 1.1 Introduction ............................................................................................................ 1 1.2 Literature review .................................................................................................... 2 1.2.1 Cambodia ........................................................................................................ 2 1.2.2 Thiamin ........................................................................................................... 4 1.2.2.1 Thiamin absorption, transport, and metabolism ...................................... 7 1.2.3 Dietary thiamin requirements ......................................................................... 9 1.2.3.1 Women of childbearing age ..................................................................... 9 1.2.3.2 Infants .................................................................................................... 11 1.2.4 Dietary thiamin sources and anti-thiamin factors ......................................... 12 1.2.5 Rice as a dietary staple .................................................................................. 13 1.2.6 Assessment of thiamin status ........................................................................ 15 1.2.6.1 Blood ...................................................................................................... 15 1.2.6.2 Breast milk ............................................................................................. 17 1.2.7 Thiamin deficiency and beriberi ................................................................... 18  vi 1.2.7.1 Thiamin deficiency ................................................................................ 18 1.2.7.2 Historical context ................................................................................... 20 1.2.8 Infantile beriberi in Southeast Asia: a disease of white rice consumption ... 21 1.2.9 Interventions to combat infantile beriberi ..................................................... 24 1.3 Summary of rationale ........................................................................................... 27 1.4 Research objectives and hypotheses .................................................................... 28 1.4.1 Objectives ..................................................................................................... 28 1.4.2 Hypotheses .................................................................................................... 29 Chapter 2: Poor thiamin status is common among women of childbearing age in rural and urban Cambodia .........................................................................................................32 2.1 Summary .............................................................................................................. 32 2.2 Introduction .......................................................................................................... 32 2.3 Methods................................................................................................................ 34 2.3.1 Participants .................................................................................................... 34 2.3.2 Sampling design ............................................................................................ 34 2.3.3 Data and blood collection ............................................................................. 35 2.3.4 Laboratory analysis ....................................................................................... 36 2.3.5 Data analyses ................................................................................................ 37 2.4 Results .................................................................................................................. 37 2.5 Discussion ............................................................................................................ 39 2.6 Implications and conclusions ............................................................................... 41 Chapter 3: Thiamin-iron fortified fish sauce: stability and sensory perceptions in rural Cambodia ............................................................................................................................42  vii 3.1 Summary .............................................................................................................. 42 3.2 Introduction .......................................................................................................... 43 3.3 Methods................................................................................................................ 45 3.3.1 Fish sauce formulation and production ......................................................... 45 3.3.2 Thiamin stability ........................................................................................... 48 3.3.2.1 Laboratory exposure experiments .......................................................... 48 3.3.2.2 Household shelf-stability ....................................................................... 49 3.3.3 Thiamin hydrochloride analysis .................................................................... 50 3.3.4 Sensory evaluation ........................................................................................ 50 3.3.4.1 Participants ............................................................................................. 50 3.3.4.2 Study procedure ..................................................................................... 51 3.3.4.3 Triangle test ........................................................................................... 51 3.3.4.4 Paired preference test ............................................................................. 52 3.3.4.5 Nine point hedonic scale ........................................................................ 52 3.4 Statistical analysis ................................................................................................ 53 3.5 Results .................................................................................................................. 53 3.5.1 Fish sauce fortification at factory ................................................................. 53 3.5.2 Exposure to light, oxygen, and heat .............................................................. 54 3.5.3 Household storage ......................................................................................... 54 3.5.4 Sensory analysis ............................................................................................ 56 3.6 Discussion ............................................................................................................ 58 3.7 Conclusions .......................................................................................................... 63  viii Chapter 4: Household consumption of thiamin-fortified fish sauce increases erythrocyte thiamin concentrations in rural Cambodian women and their children under 5 years: a randomized controlled trial ............................................................................64 4.1 Summary .............................................................................................................. 64 4.2 Introduction .......................................................................................................... 65 4.3 Methods................................................................................................................ 66 4.3.1 Study design .................................................................................................. 66 4.3.2 Fish sauce ...................................................................................................... 67 4.3.3 Randomization .............................................................................................. 68 4.3.4 Weighed fish sauce record ............................................................................ 68 4.3.5 Data and blood collection ............................................................................. 69 4.3.6 Biochemical thiamin analysis ....................................................................... 70 4.3.7 Data analysis ................................................................................................. 70 4.4 Results .................................................................................................................. 71 4.5 Discussion ............................................................................................................ 79 Chapter 5: Perinatal consumption of thiamin-fortified fish sauce in rural Cambodia: a randomized controlled efficacy trial ...................................................................84 5.1 Summary .............................................................................................................. 84 5.2 Introduction .......................................................................................................... 85 5.3 Methods................................................................................................................ 87 5.3.1 Study design .................................................................................................. 87 5.3.2 Intervention: fortified fish sauce ................................................................... 88 5.3.3 Randomization .............................................................................................. 90  ix 5.3.4 Data and biological sample collection .......................................................... 91 5.3.5 Erythrocyte thiamin diphosphate analysis .................................................... 92 5.3.6 Breast milk thiamin analysis ......................................................................... 93 5.3.7 Statistical analysis ......................................................................................... 93 5.4 Results .................................................................................................................. 94 5.5 Discussion .......................................................................................................... 100 5.6 Conclusions ........................................................................................................ 105 Chapter 6: Conclusions, discussion, and future research ......................................................106 6.1 Introduction ........................................................................................................ 106 6.2 Discussion of key findings ................................................................................. 106 6.2.1 Efficacy of thiamin-fortified fish sauce ...................................................... 106 6.2.2 Effectiveness of thiamin-fortified fish sauce .............................................. 110 6.2.3 Fish sauce as a fortification vehicle ............................................................ 111 6.2.4 Salt: an alternative thiamin fortification vehicle? ....................................... 112 6.3 Limitations ......................................................................................................... 113 6.3.1 Need to optimize thiamin fortification dose ............................................... 113 6.3.2 Lack of infantile beriberi prevalence data .................................................. 114 6.3.3 Defining thiamin deficiency ....................................................................... 115 6.3.4 Reaching the poorest groups ....................................................................... 115 6.4 Future research ................................................................................................... 116 6.4.1 Determine the prevalence of infantile beriberi ........................................... 116 6.4.2 Effectiveness study of thiamin-fortified fish sauce .................................... 117 6.4.3 Thiamin-fortified salt .................................................................................. 118  x 6.4.4 Multiple micronutrient fortification ............................................................ 118 6.5 Concluding summary ......................................................................................... 119 Bibliography ...............................................................................................................................120 Appendices: English and Khmer versions of data collection tools .............................................135      Appendix A: Cross-sectional questionnaire ………………………………………..135      Appendix B: Sensory Evaluation of thiamin-fortified fish sauce…………………..158 Appendix C: Semi-structured focus group discussion guide for fish sauce bottle design           ………………………………………………………………………………..……..165      Appendix D: Weighed fish sauce record questionnaire ………………………..… 175 Appendix E: Baseline and endline questionnaires for randomized controlled efficacy trial (non-pregnant cohort) …………………………..…………………………….181 Appendix F: Baseline and endline questionnaires for randomized controlled efficacy trial (pregnant cohort) ……………………………………….…………………….223   xi List of Tables Table 1-1: Thiamin content of select foods ..................................................................... 12 Table 1-2: Clinical presentation of thiamin deficiency .................................................... 19 Table 2-1: Characteristics and eTDP of women aged 20-45 y in Cambodia and Canada 38 Table 3-1: Demographic characteristics and general attitudes towards fish sauce of rural Cambodian women ........................................................................................................... 57 Table 3-2: Correctly identified ‘different’ sample in Triangle Test by rural Cambodian women ............................................................................................................................... 57 Table 3-3: Sensory evaluation of thiamin-fortified fish sauce by rural Cambodian women using the Nine-Point Hedonic Scale ................................................................................. 59 Table 4-2: Maternal and child eTDP at endline (6 mo) ................................................... 76 Table 4-3: Daily fish sauce consumption (mL/d) and thiamin intake (mg/d) from fish sauce of women, their husbands, and their children (12-59 mo) collected from the three day weighed fish sauce records, by meal preparation method.......................................... 78 Table 5-1: Baseline demographic characteristics and eTDP of women (18-45 y) in the pregnant and lactating cohort ............................................................................................ 96 Table 5-2: Antenatal care and delivery outcomes of rural Cambodian women (18-45 y) in the pregnant and lactating cohort, and characteristics of their newborn infants ........... 97 Table 5-3: Endline (6 mo) eTDP of mothers and their breastfed infants in the pregnant and lactating cohort ........................................................................................................... 98 Table 5-4: Thiamin concentrations of mature breast milk, and estimated daily total thiamin intake of infants fed this milk, from rural Cambodian mothers (18-45 y) in the pregnant and lactating cohort .......................................................................................... 100  xii List of Figures Figure 1-1: Chemical structures of thiamin vitamers thiamin, TMP, TDP, and TTP. ...... 6 Figure 1-2: Conversion between thiamin and the biologically active derivative, thiamin diphosphate. ........................................................................................................................ 7 Figure 1-3: Thiamin diphosphate (TDP) as a co-factor in cellular energy-producing pathways glycolysis, the Kreb’s Cycle, and the Pentose Phosphate Pathway. ................... 8 Figure 1-4: Principle of erythrocyte transketolase activity coefficient (ETKac) assay ... 15 Figure 3-1: Schematic of hypothetical intake distributions using the EAR cut-point approach for thiamin fortification. .................................................................................... 46 Figure 3-2: Thiamin hydrochloride concentrations (g/L) of fish sauce sample duplicates after laboratory exposure to light, oxygen, light and oxygen and heat. ............................ 55 Figure 3-3: Thiamin hydrochloride concentration of fish sauce (g/L) collected fortnightly from households in Prey Veng province, Cambodia over 6 months ................................ 56 Figure 4-1: Participant flow and follow-up of Cambodian women (18-45 y) and their youngest child (12-59 mo) in Prey Veng, Cambodia ....................................................... 72 Figure 4-2: Mean (95% CI) erythrocyte thiamin diphosphate concentrations (eTDP, nM) of women and their children aged 12-59 mo at baseline and endline (t=6 mo) stratified by baseline eTDP tertile ......................................................................................................... 77 Figure 5-1: Participant flow and follow-up for pregnant and lactating Cambodian women (18-45 y) and their newborn breastfed infants. ................................................................. 89  i List of Abbreviations AI  adequate intake ANOVA analysis of variance BMI  body mass index CI  confidence interval CV  coefficient of variation d  day(s) EAR  estimated average requirement EDTA  ethylenediaminetetraacetate eTDP  erythrocyte thiamin diphosphate GLM  general linear model h  hour(s)  HC  high concentration thiamin-fortified fish sauce HPLC-FLD high performance liquid chromatography with fluorescence detector IFA  iron folic acid supplements ITT  intent-to-treat analysis LC  low concentration thiamin-fortified fish sauce LSD  least significant difference min  minute(s) mo  month(s) NaFeEDTA ferric sodium ethylenediaminetetraacetate (iron fortificant) NGO  non-governmental organization NIPH  National Institute of Public Health (Phnom Penh, Cambodia)  ii NSCFF National Sub-Committee for Food Fortification (Phnom Penh, Cambodia) NTD  neural tube defect PPS  probability proportional to size sampling RDA  recommended dietary allowance SD  standard deviation TDP  thiamin diphosphate THCl  thiamin hydrochloride TMP  thiamin monophosphate TTP  thiamin triphosphate UBC  University of British Columbia (Vancouver, Canada) UL  tolerable upper intake level UN  United Nations USDA/ARS  Agricultural Research Service, United States Department of Agriculture wk  weeks(s) y  year(s)     iii Acknowledgements My first, biggest, and most enthusiastic thanks goes to Dr. Tim Green. Tim, I couldn’t have dreamed of a better mentor to guide me through my PhD. On our first day working together we rode through Phnom Penh on a moto; in many ways a Cambodian moto ride is the perfect metaphor for my doctoral degree – well-planned and purposeful, eye-opening, thrilling, and full of (mostly) enlightening bumps! Thank you for being my guide on this adventure, mentoring, teaching, and simply supporting me. You’re brilliant, Tim! – a phenomenal researcher and teacher, and all-round stellar person. I am so lucky to have been able to learn from you over the last four years. Thank you.   I was surrounded by extraordinary academics throughout my degree, but was especially lucky to be formally guided by Drs. David Kitts, Eunice Li-Chan, Larry Lynd, and Judy McLean. I cannot thank my supervisory committee enough for their constant support and insight. David, you always had time for a chat; thank you for supporting me academically and professionally. Eunice, I always appreciated your warm and sunny attitude, and amazing attention to detail! Larry, your stats lessons were invaluable, and I’ll not soon forget our many enlightening research chats over beers in Canada and Cambodia. And Judy, your passion for global health is contagious; thank you for introducing me to Cambodia. In addition to my exceptional supervisory committee, I would like to thank my other academic mentors, Drs. Dan Ramdath, Alison Duncan, Rickey Yada (my first academic mentor!), Susan Barr, Alastair Summerlee, Mary Ward, Helene McNulty, Murray Isman, and most especially, Candice Rideout.   I would like to thank my fellow graduate students: I learned so much from you, academically and otherwise, and am lucky to call this outstanding group of curious, bright, and uproarious characters my friends. Debating, chatting, and laughing over too much coffee (or wine) in the Green Lab student office, in Cambodia, and around the world, my PhD wouldn’t have been the same without you. Thank you Crystal Karakochuk, Kristina Michaux, Aviva Rappaport, Vashti Verbowski, Tina Li, Zach Daly, Phil Chebaya, Abeer Aljaadi, Rebecca Mercer, Amynah Janmohamed, Theresa Schroder, Teo Quay, Kaela Barker, and Jen Foley.   iv Thank you to my colleagues in at Helen Keller International in Cambodia, especially Hou Kroeun, Ly Sokhoing, and Zaman Talukder - I cannot thank you enough for your warm welcome and seemingly endless patience. Thank you also to Sim Chhoeun and Phon Phearom for your dedication in the field.     I am surrounded by a phenomenal group of family and friends. My sisters, Laura and Morgan, helped me through this PhD with humour, love, and the kind of support only sisters can give. Whether I was calling from Vancouver for a gossip break from writing, or after an especially nutty field day in rural Cambodia, you two made me laugh and kept me grounded. Mom and Dad, this world has never seen more supportive, ‘bursting with pride’ parents! I cannot thank you enough for raising me to love travel, or for your never-ending confidence in my abilities. Alongside my parents, an incredible group of friends kept me happy and sane throughout my PhD. Nathan Lachowsky, I’m so lucky to have such a brilliant, hilarious, and loving friend in my life. I’ll never forget all of the nights of alternating giggles and research debates over wine in our living room, and look forward to many, many more! Gavin Armstrong, what fun it was to have a confidant and fellow researcher in Cambodia; I am so happy we could live, learn, and adventure together! Sarah Chown, your delicious cakes and supportive notes are the tip of the iceberg; thank you for being a supportive and fabulous friend. Thank you to all of my friends that supported me in innumerable ways through this journey: Kaitlin Town, Martin Straathof, Carly Isman, Craig Murray, Naseam Ahmadi, Danika Meunier, Sarah Hunter, Nicola MacNeil, Erika Neilson, Jordan Willcox, Lo Smith, Kristen Bennett, and Chris Charles.   I would like to thank all of the Cambodian women and their families that participated in these studies, without whom this research would not have been possible. អរគុណ  This research was supported by funds provided from Grand Challenges Canada, the Canadian Institutes of Health Research, and the International Development Research Centre. I gratefully acknowledge these funding agencies: thank you for seeing potential in this research program and supporting potentially life-saving research in Cambodia. 1 Chapter 1: Introduction, literature review, and research objectives and hypotheses 1.1 Introduction Infantile beriberi, a disease caused by thiamin (vitamin B1) deficiency, remains a public health concern in Cambodia and other parts of Southeast Asia. Infantile beriberi usually presents during the exclusive breastfeeding period and without treatment commonly results in death within hours of clinical presentation. The Cambodian diet is low in thiamin, consisting mostly of thiamin-poor, polished white rice. Low maternal thiamin intake directly impacts breast milk thiamin content, putting exclusively breastfed infants at risk of thiamin deficiency and infantile beriberi. Therefore, maternal thiamin intake must be improved to prevent infantile beriberi. Food fortification is an inexpensive, sustainable, and passive means of improving the diet when micronutrients are lacking. Fish sauce could be an ideal vehicle for thiamin fortification as it is a popular condiment consumed by most Cambodians, and is already being used for iron fortification.   What follows is a review of the literature, followed by four chapters outlining the results of my doctoral research. Chapter 2 reports results of a cross-sectional study highlighting the low biochemical thiamin status among rural Cambodian women of childbearing age. In Chapter 3, I detail the formulation and evaluation of thiamin-fortified fish sauce, reporting on thiamin stability and consumer acceptability. Chapters 4 and 5 report the results of two concurrent double-blind randomized controlled efficacy trials of thiamin-fortified fish sauce. The first included non-pregnant, non-lactating women of childbearing age (18-45 y) and their youngest children 12-59 mo, and the second included a group of  2 pregnant and lactating women (18-45 y) and their newborn infants. The results of these chapters, as well as next steps and future directions for further research, are discussed in Chapter 6.   1.2 Literature review 1.2.1 Cambodia Cambodia is a rice-growing country in Southeast Asia, bordering Thailand, Vietnam, and Laos (1). Home to 13.4 million people, 80% of Cambodians live in rural areas; the capital city of Phnom Penh only has a population of 1.3 million people (1). Cambodia has a tumultuous history. Cambodia was a French colony until it gained independence in 1953. In 1975, Pol Pot led the Khmer Rouge regime (known locally as Pak Kommunis Kampuchea) in overthrowing the government. Over the next four years (1975-1979) the ensuing genocide resulted in an estimated two million deaths (approximately one quarter of the Cambodian population) both from execution and indirectly due to starvation and disease (2). Cambodia became a constitutional monarchy after the first free elections were held under the supervision of the United Nations Transitional Authority in 1993 (1). Despite political stability since 1993, Cambodia remains one of the least economically developed countries in Asia, with a per capita gross domestic product of ~US$800, and an estimated 28% of the population living under the poverty line (1).   Small-scale subsistence rice farming is the main economic activity in Cambodia (1), so food security is tightly linked with rice production. In the wet season (May to October), heavy rains provide ideal rice-producing conditions, while extremely dry conditions and  3 lack of irrigation limit rice production in the dry season (<10% of total Cambodian production; November to February) (3). Beyond household consumption, the marketing of rice for cash as well as other commodities make rice the major driver of household economic status, especially around harvest at the end of the wet season (3). Food insecurity, defined as the inability to acquire nutritionally adequate, safe, and acceptable foods in a socially acceptable way (4) is common in Cambodia. A recent survey of 900 households in Prey Veng province, Cambodia found that only 18% of households were food secure (5). The same study highlighted low dietary diversity: the mean Household Dietary Diversity Scale score (out of 12) was 4.7 (5). Both food insecurity and poor dietary diversity increase the risk of malnutrition and, in turn, poor health outcomes (6). In Cambodia, malnutrition typically manifests as ‘hidden hunger’, a term coined to refer malnutrition caused by micronutrient deficiencies while energy intake is adequate (7). Recent analysis of 2011 Food and Agriculture Organization food balance sheets estimate that milled (white, polished) rice makes up 1520 of the 2411 kcal/day per capita, or 63% of daily energy intake (8), so while people are not hungry in the traditional sense (inadequate energy intake), the low micronutrient content of rice means that micronutrient deficiency diseases remain prevalent.   One commonly employed malnutrition-related health indicator is childhood mortality; indeed Sustainable Development Goal 3.2 aims to end all preventable mortality of newborns and children under 5 years by 2030 (9). In 2013, post-neonatal mortality (death between 29-364 days) was 15.6 deaths per 1,000 live births in Cambodia (10). While high, this is a vast improvement as infant mortality (defined as death before the first  4 birthday) was 95 deaths per 1,000 live births in Cambodia in 2000 (1). However, this is higher than both the 2013 regional estimate of 7.9 deaths per 1,000 live births in Southeast Asia, and the current global estimate of 13.2 deaths per 1,000 live births (10). Therefore, there is room for improvement in infant mortality prevention in Cambodia. A portion of these infant deaths in Cambodia can be attributed to infantile beriberi (11,12), a ‘hidden hunger’ disease in infants caused by thiamin (vitamin B1) deficiency.  1.2.2 Thiamin Thiamin is an essential micronutrient required for normal human metabolism (13). Thiamin is found in four forms: thiamin, thiamin monophosphate (TMP), thiamin diphosphate (TDP), and thiamin triphosphate (TTP) (see Figure 1-1) (14). TDP is the biologically active derivative of thiamin (see Figure 1-2 for conversion). TDP plays a major role in glucose metabolism and cellular energy generation (13), and is also responsible for maintaining normal mitochondrial structure and function (15). TDP is a co-factor for 24 metabolic enzymes; of key  5 importance are pyruvate dehydrogenase and α-ketoglutarate dehydrogenase, which aid in the formation of acetyl-CoA and succinyl-CoA, respectively, for use in the Krebs cycle (14,16). TDP is a co-factor with transketolase in the pentose phosphate pathway, where it forms intermediates that can be channeled into the glycolysis pathway for energy production (13,17,18). Figure 1-3 summarizes the role of TDP in these pathways. As a co-factor in these cellular energy-forming pathways, thiamin is required for any conversion of glucose, or simple carbohydrates, to useable energy in humans (15). TDP also plays a major role in the central nervous system, modulating neuronal and neuromuscular transmissions (19).  6  Figure 1-1: Chemical structures of thiamin vitamers thiamin, TMP, TDP, and TTP (public domain).    Thiamin Thiamin monophosphate Thiamin diphosphate Thiamin triphosphate  7 Figure 1-2: Conversion between thiamin and the biologically active derivative, thiamin diphosphate.    1.2.2.1 Thiamin absorption, transport, and metabolism  Thiamin is a water-soluble vitamin and, as such, is not stored long-term in any tissue (13). Free thiamin is absorbed in the small intestine via two concentration-dependent mechanisms: below 1 μM, thiamin is transported through an active, carrier-mediated, sodium-dependent mechanism; at higher concentrations thiamin is absorbed via passive diffusion (20). While the majority of thiamin is absorbed in the jejunum (13) thiamin is absorbed throughout the gastrointestinal tract (20); colonic microbiota also synthesize considerable thiamin and TDP, both of which are absorbed via a sodium-independent, pH-sensitive, carrier mediated process in the colon (15). Once absorbed, phosphorylated thiamin undergoes transport to the liver via portal blood (13), mediated by SLC19A2 or SLC19A3 (14). Excess non-protein-bound plasma thiamin is rapidly dephosphorylated, filtered by glomerulus, and cleared by the kidneys (21). Little is known about the transport and incorporation of thiamin into breast milk.   8  Figure 1-3: Thiamin diphosphate (TDP) as a co-factor in cellular energy-producing pathways glycolysis, the Kreb’s Cycle, and the Pentose Phosphate Pathway. 9 TDP is the most abundant derivative in human tissue, making up nearly 90% of total body thiamin (20). Healthy adults have an estimated total body thiamin content of 25-30 mg (13), found mostly in the brain and skeletal muscle (22). At high doses only a small portion of thiamin is absorbed (23), and of that little is retained (21). For instance, a single oral dose of thiamin higher than approximately 2.5 to 5 mg has been shown to go largely unabsorbed in healthy adults (20). An experimental thiamin supplementation study among six healthy, thiamin-replete Canadian males in the 1960s found 21% absorption of a 2.5 mg thiamin dose, but decreased absorption at higher doses: 9% and 4% absorption of 5 mg and 20 mg doses, respectively (23). Of the thiamin that is absorbed, little is retained. An Australian study found that, among 6 test volunteers (22-43 y; n=3 men and 3 women) who consumed a 11 mg dose of thiamin hydrochloride, there was rapid active excretion of excess plasma thiamin in the urine (21). However, rapid metabolism and turnover of this vitamin, with a biological half-life between 9 and 18 days (24), necessitates regular dietary intake (13).   1.2.3 Dietary thiamin requirements 1.2.3.1 Women of childbearing age The Institute of Medicine’s recommended dietary allowance (RDA) for thiamin was established based on few studies with small sample sizes using outdated analytical techniques (studies conducted between 1940s-80s) (18). The heaviest weighting of evidence for the thiamin DRIs was placed on a thiamin depletion-repletion study conducted among only seven men (age unknown) in a metabolic unit in the late 1970s, using urinary thiamin excretion as the biomarker (18,25). As such, the DRIs should be  10 considered under this context. The EAR and RDA for thiamin for women aged 19-50 y is 0.9 mg/d and 1.1 mg/d, respectively (18), however the Canadian Community Health Survey data indicate that usual thiamin intake among women of childbearing age is much higher at 1.48 mg/d (aged 19-50 y) (26). During pregnancy and lactation the DRIs were calculated to allow for a 10% increase for higher energy utilization in pregnancy and increased energy costs for milk production in lactation, increasing the RDA to 1.4 mg/day (18). During the third trimester of pregnancy, thiamin is preferentially sequestered by the fetus (27): umbilical cord blood of thiamin-replete mothers has up to three times higher thiamin concentrations than maternal blood at birth (28). This sequestration highlights the importance of adequate maternal thiamin intake, especially during the latter stages of pregnancy, to allow for a build up of infant stores in utero. During lactation, increased maternal thiamin requirements ensure adequate growth and development of the infant (29,30). Lactating women transfer an estimated 0.16 mg thiamin daily to breast milk, and additional thiamin is required by the mother herself to meet increased energy needs for milk production (18).   There is no tolerable upper intake level (UL) for thiamin because there have been no reports of adverse effects of excess thiamin intake (17,18), probably because absorption declines rapidly at dietary intakes at and above 5 mg (20,23) and excess thiamin is excreted in urine (21). Indeed, some daily over-the-counter supplements can contain 50 mg thiamin (18) or more, greatly exceeding the RDA.    11 Southeast Asian-specific dietary thiamin recommendations are identical to those noted above, except for needs during lactation (18,31). The Southeast Asian RDA for lactating women is 1.5 mg/day: in addition to the 1.1 mg/day for women of childbearing age, increased thiamin needs of 0.2 mg  (transferred to milk) and 0.2 mg (increased needs for for milk production) were estimated (31). However, the Institute of Medicine recommends 1.4 mg/day as it estimates only 0.1 mg/day requirements for milk production (18).   1.2.3.2 Infants Since there have been no documented reports of full-term, exclusively breastfed American or Canadian infants developing beriberi, the Institute of Medicine set an adequate intake (AI) for thiamin for infants aged 0-6 months using observed thiamin intakes by exclusively breastfed infants consuming milk from healthy, purportedly thiamin-replete mothers (18). AIs are developed when there is not sufficient evidence available to establish an estimated average requirement (EAR) and RDA, and are expected to meet or exceed the needs of individuals in that age group (18). Using an observed mean thiamin concentration of 210 μg/L in milk produced by well-nourished American mothers (n=24) in the 1980s (32–34), and estimating infants of this age consumed 780 mL milk/day, the Institute of Medicine set the thiamin AI for infants aged 0-6 mo at 200 μg thiamin/day (18). This AI is contentious due to the small sample size of women used to set this AI, and likely suboptimal (outdated) analytical techniques (35); see 1.2.6.2 Breast milk, below.   12 1.2.4 Dietary thiamin sources and anti-thiamin factors Thiamin is found in a wide range of foods, but is particularly high in whole grain products, legumes, soy, beans, pork, and organ meats (13). In North America, wheat flour is enriched with thiamin to offset processing losses (36,37), and other foods, including ready-to-eat breakfast cereals, are fortified with thiamin voluntarily. The thiamin content of these high-thiamin foods (and thiamin-poor alternatives) can be found in Table 1-1.   Table 1-1: Thiamin content of select foods Food Thiamin content (mg/100g)1 Kellogg’s Corn Flakes (fortified breakfast cereal) 1.34 Pork loin (cooked) 0.599 Bread       Whole wheat bread 0.522      White bread (thiamin enriched) 0.510 Beef liver (broiled) 0.194 Tofu (fried) 0.170 Red lentils (boiled) 0.169 Kidney beans (cooked) 0.160 Beef kidney (simmered) 0.160 Rice       Brown rice (cooked) 0.102      White rice (cooked) 0.020 1 Values derived from National Nutrient Database for Standard Reference (Release 28, software v.2.3.4.2), United States Department of Agriculture Agricultural Research Service.  Thiamin is sensitive to neutral and alkaline pH, oxygen, light, and heat (38), as well as exposure to thiaminases (enzymes that degrade thiamin) (39). In addition, thiamin absorption and metabolism can be disrupted by the presence of anti-thiamin factors. There are three types of anti-thiamin factors: thiaminase, polyphenols (namely tannins), and catechols (39). Thiaminase, an enzyme that degrades thiamin, was the first ‘antivitamin’ to be discovered, in the 1930s (40). Found in several fish species as well as clams and crabs, thiaminase cleaves thiamin at its methylene linkage (39,40). However,  13 thiaminase is heat-liable, so thiaminase-containing foods are only a risk when consumed raw (13). Unfortunately, prahoc, a fermented fish paste that is often viewed as the second most important food in Cambodia after rice, is eaten both cooked and raw (41). Gallic acid-containing tannins, such as those found in tea and betel nuts, destroy thiamin under neutral and basic conditions (39). The anti-thiamin effects of these tannins can be prevented by delaying contact with tannins (i.e. consuming tea hours after a thiamin-containing meal, rather than during) (42), or by cooking with ascorbic acid to lower pH (39). In Cambodia, betel nut chewing has lost popularity due to permanent staining of teeth, however some women chew betel during pregnancy to alleviate morning sickness (43). Finally, catechols, found in green coffee beans, sunflower seeds, and bracken fern also have anti-thiamin activity (40), but these are not common in the Cambodian diet.   1.2.5 Rice as a dietary staple As shown in Table 1-1, white rice is a poor source of thiamin because thiamin is found only in the outer husk and bran, the vast majority of which is removed during the milling process (44). Further polishing removes any remaining traces of thiamin from the rice as only the thiamin-poor endosperm remains (45). This is problematic because in most rice-consuming cultures globally, well-milled, polished white rice is preferred (44,46) for several reasons: organoleptic qualities, white rice as a status symbol (45), and because removal of the lipid-rich outer bran increases shelf-life (44). In Cambodia, brown rice is a food traditionally consumed by prisoners and livestock (i.e. fish, pigs); as such, white rice is preferred.    14 Parboiling, a process of soaking and steaming rice before milling, is one means of improving the thiamin content of white rice (44). While originally employed to aid in de-husking (bran is softer to grind through hand milling), this process forces thiamin and other B vitamins from the bran into the endosperm during the soaking step (45). Unfortunately parboiling rice is not commonplace in Asia because traditionally this process resulted in a musty aftertaste caused by mold growth over long-term solar drying (45). Even today, with parboiled rice being dried in commercial driers (hence no musty taste), the parboiled rice remains out of favour because this processing causes a darkening of the endosperm, decreasing consumer acceptability (44).    As noted above, thiamin is required for carbohydrate metabolism, therefore high consumption of carbohydrates (such as polished white rice) increases thiamin requirements (14). A short-term trial in 12 healthy Austrians (n=6 women, 6 men) reported significantly lower plasma and urine thiamin concentrations when diets moved from 55% carbohydrate (control diet) to 65% (4 day exposure) and 75% carbohydrates (4 day exposure), while keeping other factors constant (i.e. thiamin intake, diet, exercise), indicating higher thiamin usage at increased carbohydrate intakes (47). Higher requirements have been reported in the field as well: a 1988 outbreak of beriberi among 140 people in The Gambia highlighted important links between manual labour, high consumption of carbohydrates, and thiamin deficiency (48). Beriberi prevalence was three times higher among men compared to women, and one third of those men affected were aged 20-29 y. The outbreak occurred during the ‘hungry season’ (the time between last season’s food running out and the start of the current year’s harvest), so researchers  15 attributed the disproportionate number of young men affected to a combination of a heavy workload with high consumption of imported, polished white rice (48). Cambodian women face similar circumstances: physical labour is required to maintain the homestead, and approximately 60% of dietary energy is thought to come from polished white rice (8).    1.2.6 Assessment of thiamin status 1.2.6.1 Blood Erythrocyte transketolase activity coefficient (ETKac) is widely regarded as the best functional indicator of thiamin deficiency (17,18). TDP is a cofactor with transketolase in the pentose phosphate pathway (which takes place in erythrocytes), and therefore measuring the activity of transketolase before and after the addition of excess TDP reveals the original level of TDP saturation within the erythrocytes, indicating thiamin adequacy. If the cells had sufficient TDP, the ETKac would not change after the addition of excess TDP, and an ETKac ratio of 1 would be calculated. If the erythrocytes were originally deficient in thiamin, then the addition of excess TDP would increase ETKac, and the ETKac ratio would rise above 1 (17,18) (see Figure 1-4). ETKac >1.25 is commonly used as a cut-off for thiamin deficiency (18).   Figure 1-4: Principle of erythrocyte transketolase activity coefficient (ETKac) assay  16  Although ETKac is regarded as the best indicator of thiamin status because it provides a functional measure of thiamin, this test has several downfalls including poor inter-assay precision, and the rapid inactivation of transketolase during sample processing and storage (49). In addition, there have been reports of this assay underreporting activity coefficients among chronically deficient populations (1,19).  To address these shortcomings, an alternate method is now more commonly used to measure TDP directly: high performance liquid chromatography with a fluorescent detector (HPLC-FLD) (49,51). The vast majority (95%) of thiamin in blood is present as TDP (52), and it depletes at approximately the same rate as other tissues (18,53), making it an ideal candidate for measurement. TDP is stable in blood for at least 24 hr at room temperature, and for up to 7 months at -70°C (49). HPLC also yields high inter-assay precision, and can be easily standardized with commercially available TDP. TDP measurement with HPLC-FLD also correlates well with ETKac (49). However, direct assessment of blood thiamin makes interpreting results more difficult as compared to a functional measure because the TDP available for use remains unknown.   Indeed, interpretive criteria for sufficient thiamin status using erythrocyte TDP (eTDP) is still unclear as cut-offs vary widely (17). The Institute of Medicine defines thiamin deficiency as eTDP < 70 nM (18); however this guideline is based on values from one group of Dutch blood donors (n=98) in the 1980s (18,54). Interestingly, these cut-off values are described by the Institute of Medicine as ‘erythrocyte thiamin’ (18), but in the  17 original report are referred to as cut-offs for ‘whole blood or red cells’ (54). These terms should not be used interchangeably; if these cut-offs are based on whole blood analysis, the erythrocyte cut-offs should be ~40% higher (taking into account the hematocrit). Wilkinson and colleagues put forth a much higher cut-off for suboptimal thiamin in erythrocytes of <140 nM; this was calculated as a reference limit for the bottom 2.5% of healthy New Zealander blood donors (n unknown) (55). An even higher cut-off of <148 mM has been suggested as it is at the lower range of ‘normal’ among British adults (17,56). In practice, reference ranges of eTDP vary widely. A study among British adolescents aged 13-14 y (n=35 girls, 19 boys) reported eTDP (mean [range]) of 227 nM (101-950) and 206 nM (120-445) among girls and boys, respectively (56). Among a sample of healthy British laboratory staff (n=29 male, 16 female) aged 20-60 y, mean eTDP was 174 nM (52). In Cambodia, Coats et al reported whole blood TDP adjusted for hematocrit (eTDP equivalent) of 150 nM (95% CI: 134, 166; n=27) among a group of healthy lactating mothers (11). Unfortunately, to the best of my knowledge, there are no known reference ranges for healthy, thiamin-replete, non-pregnant, non-lactating Cambodian women of childbearing age.   1.2.6.2 Breast milk Three thiamin vitamers are found in human milk: thiamin, TMP (which is present in highest concentrations), and TDP (57). For the last 60 years, thiamin concentrations in milk have been measured using the thiochrome method, wherein a derivatizing agent is used to alter the molecular structure of thiamin (2 rings, see Figure 1-1) to thiochrome (3 rings), which fluoresces (35). More recently, researchers have built upon this method by  18 embedding the thiochrome derivatization within online pre- or post-column HPLC-FLD, allowing for precise and highly repeatable measurements (35). The same advantages of TDP measurement in blood with HPLC-FLD apply to breast milk.   As noted in 1.2.3.2 Dietary thiamin requirements (infants), the AI for thiamin for infants aged 0-6 mo is 200 μg/d. It is important to note, however, that the AI is contentious as more recent reports of breast milk thiamin concentrations are lower than these estimates, likely due to improvements in analytical techniques for quantification of thiamin in biological samples during the past 35 years (35). For instance, a recent report of median (range) total thiamin concentrations in mature milk (≥2 weeks) from women worldwide highlighted wide variation in breast milk thiamin concentrations by region, with no group approaching 210 μg/L: Cameroon, 116 μg/L (86-221; n=5); China, 31 μg/L (15-127; n=5); India, 11 μg/L (4-75; n=24); Malawi, 21 μg/L (2-152; n=18); United States, 37 μg/L (5-66; n=28) (58). In addition, infant breast milk consumption changes with age (59). More research is required to better understand the usual thiamin content, measured using valid and reliable analytical techniques, of breast milk produced by healthy, thiamin-replete, and overall well-nourished mothers of infants aged 0-6 mo.  1.2.7 Thiamin deficiency and beriberi 1.2.7.1 Thiamin deficiency Thiamin deficiency manifests with several different clinical presentations, in adults as wet, dry, and Shoshin beriberi, and Wernicke encephalopathy (Wernicke-Korsakoff  19 syndrome), and in infants as infantile beriberi (13). Clinical presentations of these different forms of thiamin deficiency can be found in Table 1-2.   Table 1-2: Clinical presentation of thiamin deficiency  Clinical Presentation References Dry beriberi x peripheral neuropathy with symmetric, bilateral impairment of motor functions in distal limbs, most notably legs x paresthesia of feet and toes, calf muscle tenderness  x acute encephalopathy (13,60,61) Wet beriberi x high cardiac output x tachycardia x cardiomegaly, right ventricular hypertrophy x vasodilation in the extremities x sweating, warm skin x renal water retention, leading to edema x heart failure (13,60,62,63) Shoshin beriberi x same as wet beriberi, but with rapid onset (13,60,61) Wernicke-Korsakoff syndrome x associated with chronic, excessive alcohol intake x ataxic gait x loss of ocular control (ophthalmoplegia, nystagmus) x confusion, memory impairment x peripheral neuropathy (13,64,65) Infantile beriberi x presents among breastfed infants aged 2-4 mo x dysphonia (distinctive hoarse cry) x tachypnea and dyspnea, tachycardia x cardiomegaly, right ventricular hypertrophy x hepatomegaly x heart failure x anorexia, vomiting, oliguria x generalized edema x convulsions (11,13,19,66,67)  As a vitamin deficiency disease beriberi is reversible, however if left untreated it has serious consequences (13,65,68): without treatment, Wernicke-Korsakoff syndrome  20 causes long-term, irreversible memory damage (65), and infants can die of infantile beriberi within hours of clinical presentation without rapid thiamin administration (68).   Interestingly, humans are much more sensitive to thiamin deficiency than animals: thiamin concentrations of human brains are much lower than other animals, potentially highlighting the sensitivity of thiamin deficiency in humans compared to other species, most notably rodents (22). In addition, unlike rodents, who have a large hepatic thiamin storage capacity, humans have limited storage of thiamin (22) with stores depleting on a thiamin-free diet within two weeks (31).     1.2.7.2 Historical context Beriberi has a long history in the medical literature, with references found in Chinese medical texts as early as 2600 BC (45,68). However, it was not until the 1880s that beriberi underwent systematic investigation. Kanehiro Takaki, a surgeon in the Japanese navy, became interested in kakké (Japanese term for beriberi) after estimating that three quarters of the patients in the Tokyo Naval Hospital were suffering from this disease when he began his tenure in 1870 (45). Takaki studied medical records from all the Japanese naval ships, and identified that there was no pattern in kakké prevalence or mortality based on ship population density, routes, or weather. As such, he turned to diet, conducting controlled dietary experiments on naval vessels. Takaki postulated nitrogen/protein deficiency was the cause of kakké after noting the higher quality of British and German naval meals, so replaced some of the white rice with meat, condensed milk, bread, and vegetables on Japanese naval ships. Although his nitrogen hypothesis  21 was incorrect, the replacement of thiamin-poor white rice with thiamin-rich foods eradicated kakké mortality in the Japanese navy by 1887 (45). During the same time period, Christiaan Eijkman conducted a series of experiments with chickens in the Dutch East Indies (69). Eijkman showed that consumption of white rice caused beriberi among chickens, while brown rice consumption prevented the disease (69). It is worth noting, however, that beriberi was not confined to the Dutch East Indies (modern-day Indonesia) or Japan. As noted above, beriberi was long known as kakké in Japan, which is actually translated from the Chinese term for ‘leg disease’ (45); in Java, Dutch physician Jacobus Bonitus first described the disease as beriberi, which translates to ‘sheep’ in local language, in reference to patients’ tottering, knee-shaking gait (45,60). As early as 1913, beriberi was treated with an extract derived from rice polishings in the Philippines (45).   Beyond beriberi, thiamin itself is also of historical note due to several ‘firsts’: thiamin was the first vitamin to be isolated and synthesized. Casimir Funk isolated the ‘anti-beriberi’ factor from rice bran in 1911 and dubbed it ‘vitamine’ (60), and in 1936 R. R. Williams identified the correct chemical structure of thiamin and successfully synthesized it in the lab (69).     1.2.8 Infantile beriberi in Southeast Asia: a disease of white rice consumption After the discovery of thiamin and its role in the prevention and treatment of beriberi, this disease became relatively uncommon until an outbreak of infantile beriberi in refugee camps on the Thai-Burmese border in the 1980s (68,70). Between 1987 and 1990, 40% of the extremely high 18% infant mortality among Karen infants residing in these refugee  22 camps could be attributed to infantile beriberi, despite access to high quality antenatal care (68). This was surprising to healthcare practitioners as reports of beriberi among adults were rare. Researchers went on to show that women did not have a thiamin-rich diet, and were exposed to several thiamin antagonists: diets consisted of polished white rice and fermented fish paste, and betel nut chewing was extremely common (79% of women; 92% of whom chewed betel nuts within one hour of a meal) (68).   Mothers with low dietary thiamin intake and/or poor biochemical thiamin status produce breast milk low in thiamin, putting their infants at risk of developing thiamin deficiency and infantile beriberi (57). Infantile beriberi presents in breastfed infants aged two to four months consuming thiamin-poor breast milk from thiamin deficient mothers (13,57). With recommendations that breast milk be the sole source of nutrition for infants under 6 months (71), maternal dietary thiamin intake must be improved to prevent infantile beriberi and related mortality (45).  Infantile beriberi was addressed in the Karen refugee camps through perinatal maternal thiamin supplementation, as well as intramuscular thiamin administration to infants presenting with clinical symptoms (68,72). Similarly, supplementation of thiamin deficient lactating women has been shown to improve breast milk thiamin concentrations during research projects in rural Cambodia (73), the Gambia (74), and India (75).   While infantile beriberi was addressed in these refugee camps, it remains an often-overlooked cause of infant mortality throughout Southeast Asia, where well-polished,  23 non-parboiled rice remains the dietary staple (11,29,76–79). In 2008 there were 3,038 cases of infantile beriberi in the Kantha Bopha Hospitals in Phnom Penh and Siem Reap, Cambodia alone (personal communication, Kantha Bopha Hospital). A recent study among ethnic groups in northern Laos reported a high infant mortality rate of 50 deaths of infants aged 0 – 6 mo among 468 live births in the 22 villages included in the survey (77). Of these 50 infant deaths, 36 occurred when infants were aged 1-3 mo, and 17 were suspected to be caused by infantile beriberi (77). Similarly, thiamin deficiency appears to be prevalent throughout Laos, including in the capital city Vientiane where 13% biochemical thiamin deficiency (defined as basal ETKA < 0.59, without clinical signs of beriberi) was reported among a cohort 778 of sick infants <12 mo (78). The authors of that study referred to infantile beriberi as ‘a forgotten disease in Asia’ (78). Indeed, a major gap in the current literature is the lack of infantile beriberi prevalence data. Much of the current documentation of infantile beriberi comes from case reports, or focuses on specialized groups including refugee populations (11,66,68,70). Although anecdotal reports of infantile beriberi are common among rural Cambodian health care providers, the lack of representative prevalence data hinders our understanding of the magnitude of this public health issue.  As shown in Table 1-2, many of the signs of infantile beriberi that would be more obvious to mothers, such as vomiting, oliguria, anorexia, dyspnea, tachypnea and tachycardia (11,13,19,66,67), are general symptoms that may not seem of immediate risk to infants, and as such may not prompt a quick trip to the health centre. Since death can  24 occur within hours of clinical presentation (68), it is possible that infantile beriberi-related mortality goes improperly diagnosed.     1.2.9 Interventions to combat infantile beriberi Thiamin-rich foods are not typically consumed in Cambodia due to cost (i.e. expensive animal-source foods including pork, organ meats), or availability and cultural norms (i.e. legumes, soy). In addition, an estimated 60% of daily calories come from non-parboiled (44), polished white rice (8), which is a poor source of thiamin (13). A recent report used Food Balance Sheet data to highlight probable micronutrient deficiencies across the Western Pacific Region (80). Of the 17 countries surveyed, Cambodia had the lowest estimated amounts of thiamin per capita in the available food supply. Dietary thiamin intake was estimated at 0.6 mg, well below the RDA of 1.4 mg/day for pregnant and lactating women (18,80). As such, changing dietary behaviours to increase thiamin intake is a long-term, multi-sectorial challenge. With evidence of infantile beriberi cases in Southeast Asia, a timely intervention to increase thiamin intake of Cambodian women is warranted to combat this disease.   Thiamin supplementation is an obvious solution, as supplementation among thiamin deficient lactating women has been shown to improve breast milk thiamin concentrations (68,72–75). Supplementation is advantageous as it provides the correct dose only to those who require thiamin (deficient individuals; pregnant and lactating women). In addition, perinatal micronutrient supplements are already recommended: Cambodian Ministry of Health guidelines indicate iron-folic acid supplements (IFA) as standard of care  25 throughout pregnancy (90 tablets distributed over two antenatal care appointments) and post-partum (42 tablets provided at first postpartum contact) (81). Adding thiamin to IFA could be one means of improving perinatal thiamin intake. However supplementation is a targeted, costly, and resource-intensive intervention that relies heavily on individual compliance. A recent assessment of IFA in two Cambodian provinces showed that only 47% of women were adherent (82). While thiamin could be incorporated into IFA, uptake would likely be similarly low due to known adverse side effects, namely gastrointestinal discomfort, from iron.  Food fortification, or adding vitamins and minerals to commonly consumed foods, is another option for improving thiamin status. Food fortification has been successfully employed in Canada since the mid-twentieth century; in conjunction with a diverse diet, salt iodization, the addition of vitamins A and D to milk, and B vitamin enrichment of white flour have nearly eradicated micronutrient deficiency diseases (36). In Cambodia, fortification is gaining popularity: salt iodization is mandatory (83), and iron (fish and soy sauce) and vitamin A (vegetable oil) fortification programs are being piloted (84–86). Fortification has several advantages: it is a sustainable, cost-effective, and passive intervention that requires no behaviour change (87–90). As a population-wide intervention, there is potential to reach large sections of the population with relatively low cost and effort (91). All consumers would receive thiamin, which is beneficial in Southeast Asia where thiamin consumption is known to be low (13,29,45).    26 However, fortification does have several downfalls, including a lack of consumer choice, risk of over-consumption of the nutrient, and potential for interaction with other nutrients (87). While overall considered an inexpensive intervention, there are related costs including start-up infrastructure and equipment maintenance, as well as a need for quality assurance programs both at the factory and through the government (for mandatory fortification) (87). Finally, fortification can not reach everyone; due to unequal household food distribution, the highest consumers of fortified products in Cambodia would likely be adult men, rather than those most in need, pregnant and lactating women, and the cost of a fortified product may be inhibitive to those most in need (low socioeconomic status, most food insecure).   While these downfalls should be addressed, in the case of thiamin in Cambodia, a large proportion of the population would likely benefit from increased thiamin intake since marginal thiamin deficiency is known to cause apathy, fatigue, loss of appetite, and dizziness (45), and may be common among other at-risk groups such as the elderly (22) and those with high carbohydrate intakes (14,47). There is a low risk of adverse effects from high thiamin consumption (18), so fortification has the potential to be a passive and sustainable intervention to improve dietary thiamin intake in rural Cambodia.   Rice is an obvious staple to fortify with thiamin as recent food balance sheet analysis estimate it makes up approximately 60% of dietary energy (8). Indeed, the Khmer translation of ‘eat’ is ‘eat rice’. However, thiamin fortification of rice would be challenging and costly. Previous attempts to fortify rice with thiamin in Southeast Asia  27 were unsuccessful as rice is typically washed three times before consumption, removing all of the powdered or spray-coated B vitamins (45). In addition, there are rice millers in nearly every Cambodian village, therefore fortification would need to take place locally, increasing costs and impacting quality control. Fish sauce, however, is a popular traditional Cambodian condiment consumed by 90% of the population (85), and since this condiment is already a fortification vehicle for iron (84–86), fortification infrastructure is already present in Cambodian factories (92). In addition, 44 fish sauce factories reach 78% of the Cambodian population (92). While a segment of the population makes their own fish sauce at home, fish sauce fortification has potential to improve the thiamin intake of a wide group of Cambodians.   1.3 Summary of rationale Thiamin deficiency and infantile beriberi remain an often overlooked public health issue in Cambodia and the wider region (11,77–79,93,94). Improving maternal thiamin intake can prevent infantile beriberi among breastfed infants (13,57). An estimated 90% of Cambodians consume fish sauce (85), and this condiment is already a fortification vehicle for iron (84–86) so fish sauce fortification infrastructure is already present (92). As such, thiamin fortification of fish sauce could be an ideal means of improving the dietary thiamin intake of pregnant and lactating women and, in turn, their breastfed infants, for the prevention of infantile beriberi. The objectives of my doctoral research were to: 1) confirm that rural Cambodian women of childbearing age did indeed have low biochemical thiamin status; 2) formulate shelf-stable thiamin-fortified fish sauce that would be acceptable to rural Cambodia women; and 3) determine the efficacy of this  28 thiamin-fortified fish sauce to increase the biochemical thiamin status of rural Cambodian women of childbearing age and their children, compared to a control fish sauce containing no thiamin. These objectives, and my hypotheses are outlined in detail, below.  1.4 Research objectives and hypotheses Infantile beriberi remains an often-overlooked cause of infant mortality in Cambodia (29,45,78); as such, the aim of this research program was to increase the dietary thiamin intake of Cambodian women with the overall goal of preventing infantile beriberi. Using an iterative approach, I examined the biochemical thiamin status of women of childbearing age, and then explored thiamin fortification of fish sauce as a means of increasing dietary thiamin intake to improve biochemical thiamin status in women and their children.   1.4.1 Objectives 1. To determine the biochemical thiamin status, as determined by eTDP, of representative samples of women of childbearing age (20-45 y) from rural and urban Cambodia, as compared to a convenience sample of purportedly thiamin-replete Canadian women.  2. If women did have poor biochemical thiamin status, to develop a fortified fish sauce in which:  a) thiamin degradation was minimal, and  b) organoleptic properties were acceptable to rural Cambodian consumers.  29  3. To determine whether ad libitum consumption of the developed thiamin-fortified fish sauce at two concentrations (2 and 8 g/L) over six months could improve eTDP (and lactating women’s breast milk) compared to a control sauce containing no thiamin, among two groups of women (18-45 y) in rural Cambodia: a) non-pregnant, non-lactating women and their youngest children aged 12-59 mo (non-pregnant cohort), and  b) pregnant and lactating mothers and their newborn breastfed infants (pregnant and lactating cohort).  1.4.2 Hypotheses 1.  Null hypothesis (H0): eTDP concentrations do not differ between women of childbearing age living in rural Cambodia, urban Cambodia, and Canada.    Research hypothesis (HA): Purportedly thiamin-replete Canadian women will have higher eTDP concentrations than Cambodian women, and Cambodian women living in an urban centre will have higher eTDP than rural women.   2. a) Null hypothesis (H0): Thiamin will be stable in a fish sauce matrix.   Research hypothesis (HA): Thiamin will not be stable in a fish sauce matrix.    30 2. b) Null hypothesis (H0):  The addition of thiamin to fish sauce will not cause any organoleptic changes that affect consumer acceptance of thiamin-fortified fish sauce.  Research hypothesis (HA): The addition of thiamin to fish sauce will cause organoleptic changes that affect consumer acceptance of thiamin-fortified fish sauce.  3. a)  Null hypothesis (H0): Women and their children consuming thiamin-fortified fish sauce over six months will have the same eTDP as those consuming a control fish sauce containing no thiamin; there will be no dose response in eTDP corresponding to different fortified fish sauces (2 g/L and 8 g/L).   Research hypothesis (HA): Six month consumption of fish sauce fortified with thiamin will increase maternal and child eTDP in a dose-dependent manner (8 g/L > 2 g/L > control).   3. b)  Null hypothesis (H0): Women consuming thiamin-fortified fish sauce over six months, and their breastfed infants will have the same eTDP (and breast milk thiamin concentrations) as those consuming a control fish sauce containing no thiamin; there will be no dose response in eTDP corresponding to different fortified fish sauces (2 g/L and 8 g/L).    31 Research hypothesis (HA): Six month consumption of fish sauce fortified with thiamin will increase maternal and child eTDP and breast milk thiamin concentrations in a dose-dependent manner (8 g/L > 2 g/L > control).   32 Chapter 2: Poor thiamin status is common among women of childbearing age in rural and urban Cambodia 2.1 Summary  Background: Thiamin deficiency in infancy is the underlying cause of beriberi, which can be fatal without rapid treatment. Reports of infantile beriberi are not uncommon in Cambodia, however population representative data are unavailable.  Objective: To determine the biochemical thiamin status among women of childbearing age in rural and urban Cambodia.  Methods: We measured erythrocyte thiamin diphosphate (eTDP), an indicator of thiamin status, in a representative sample of Cambodian women (20-45 y) in urban Phnom Penh (n=117) and rural Prey Veng (n=121), Cambodia, and for comparison purposes, in a convenience sample of women in urban Vancouver, Canada (n=47).  Results: Mean ± SD eTDP was significantly lower among women in Prey Veng (149 ± 36 nM) than Phnom Penh (156 ± 32 nM), which, in turn, was lower than in Vancouver, (179 ± 37 nM; P<0.05).  Conclusions: Cambodian women have significantly lower eTDP values than purportedly thiamin-replete women from Vancouver, Canada. Strategies may be needed to improve thiamin status among women in Cambodia.   2.2 Introduction Beriberi, a disease caused by severe thiamin deficiency, is rare in economically developed regions but is thought to be more common in countries where dietary sources of thiamin are low, particularly those in Southeast Asia (11,68,78,95). Infantile beriberi is  33 characterized by a persistent hoarse cry, vomiting, anorexia, generalized edema, oliguria (96), convulsions, and heart failure (11), and is often fatal without rapid thiamin administration (67,78). Beriberi is most serious and life-threatening in infants due to the rapid growth and development that occurs during the first months of life (13). Beriberi typically presents in exclusively breastfed infants whose mothers have suboptimal thiamin status and consequently have low breast milk thiamin concentrations (57). Interestingly, it is not uncommon for infants to present with beriberi while their mothers remain asymptomatic (11,13,67). In Cambodia, low dietary thiamin intake is likely a result of multiple factors: the high consumption of polished white rice, which has been removed from the B-vitamin-containing husk (8,46), a lack of parboiling rice (97), and low dietary diversity (1). In a recent study, Coats et al. (11) found evidence of thiamin deficiency among mother-infant pairs in rural Cambodia based on levels of whole blood thiamin diphosphate concentrations (TDP), an indicator of thiamin status.   Although there has been some documented evidence of thiamin deficiency in Cambodia, there is a lack of population representative data from both urban and rural areas. The prevalence of thiamin deficiency among Canadian women of childbearing age is thought to be extremely low due to the mandatory fortification of grain products (18,36,87) and access to a diverse diet including thiamin-rich animal source foods. There is little data highlighting the erythrocyte thiamin diphosphate concentrations (eTDP) of healthy, B vitamin-replete women of childbearing age to which we could compare the Cambodian values; therefore, we recruited a small convenience sample of healthy women in Vancouver, Canada.    34  We aimed to measure eTDP, an indicator of thiamin status, among women of childbearing age in rural and urban Cambodia, and for comparison purposes, urban Vancouver, Canada.   2.3 Methods 2.3.1 Participants In this observational, cross-sectional study, representative samples of women of childbearing age (20-45 y) were recruited in urban Phnom Penh (n=160) and rural Prey Veng (n=160), Cambodia. For comparison purposes, a convenience sample of purportedly B vitamin-replete women of childbearing age (20-45 y) was also recruited in urban Vancouver, British Columbia, Canada (n=51). Women who were pregnant, lactating, or taking B vitamin-containing supplements were excluded. Ethics approval was obtained from the University of British Columbia Clinical Research Ethics Board (H12-02847 and H13-01319) in Canada and the National Ethics Committee for Health Research (0004NECHR) in Cambodia. Written informed consent was obtained from all women.  2.3.2 Sampling design A list of all villages in Prey Veng and Phnom Penh provinces was obtained from the Cambodian Ministry of Planning. Using a computer-generated random number list, simple random sampling was employed to select 16 villages from each region. In each of the 32 villages the Village Chief used the Village Registry to compose a list of all eligible  35 women, from which 10 women per village were selected using a random draw and invited to participate. If a selected woman was unable or unwilling to participate, an additional woman was randomly selected until 10 women per village were successfully recruited.   In Vancouver, British Columbia, Canada, a convenience sample of women of childbearing age (20-45 y) was recruited from The University of British Columbia by posters, emails, and word of mouth. Eligibility for study participation was confirmed via email or over the telephone. The first 51 eligible women who contacted the study team were recruited to participate.    2.3.3 Data and blood collection In Cambodia, data and blood collection took place from January 29 to February 7, 2013. Trained enumerators administered a short demographic questionnaire in the woman’s home, and height and weight were measured (see questionnaire in Appendix A). The next morning, a non-fasting venous blood sample was collected from each woman at her local health centre into an evacuated tube containing EDTA (Becton Dickinson Vaccutainer, Mississauga, ON). Response rates for blood collection were 91% in Phnom Penh (n=146 of 160) and 98% (n=156 of 160) in Prey Veng. Blood was transported on ice to the National Institute of Public Health (NIPH) in Phnom Penh daily for processing and storage. Blood samples were spun in a refrigerated (4°C) centrifuge, plasma and buffy coat were removed, erythrocytes were washed three times with phosphate buffered saline  36 (Amresco, Solon, OH), aliquoted, and stored at -80°C. All samples were shipped on dry ice to the University of British Columbia, Canada.   In Vancouver, data and blood collection took place from July 4 to July 15, 2013. Women completed a short demographic questionnaire under the supervision of a research assistant. All other procedures were identical to those in Cambodia. All blood samples were immediately processed and stored at -80°C, and batch analyzed with the Cambodian samples.  2.3.4 Laboratory analysis TDP was measured in erythrocytes because the majority (~90-95%) of thiamin in blood is present as TDP (49,52), and it declines in erythrocytes at approximately the same rate as other tissues (18,53). eTDP was measured at UBC using reverse-phase high performance liquid chromatography with a fluorescence detector (HPLC-FLD) according to Lu & Frank (51), with modifications. Briefly, 500 µL of 10% wt/vol trichloroacetic acid in deionized water was added to a mixture of 250 µL previously frozen packed erythrocytes and 250 µL deionized water. Samples were vigorously vortex mixed, placed on ice for 15 min, and then centifuged (13000 g, 10 min). An aliquot of supernatant (500 µL) was washed twice with 750 µL of water saturated methyl-tert-butyl ether. From the aqueous layer, a 150 µL aliquot was then transferred to a 96 well plate. HPLC analysis was performed using an Agilent 1260 Infinity system with a Poroshell 120 EC-C18 column (3.0 x 50 mm with 2.7 µm; Agilent Technologies, Mississauga, ON) at 25°C with the detector set at an excitation wavelength 375 nm and emission wavelength 435 nm.  37 Mobile phase A consisted of 25 nM sodium phosphate (pH 7.0) and methanol (90:10 vol/vol), while mobile phase B consisted of 25 nM sodium phosphate (pH 7.0) and methanol (30:70 vol/vol). Prior to injection, online sample derivatization (methanol, sample, and 1.2 mmol/L potassium ferricyanide in 15% wt/vol sodium hydroxide) was performed by automated injector programming. Quantitation of eTDP was based on peak area and external standardization using TDP calibration solutions (~20-800 nM thiamin pyrophosphate ≥ 95%, Sigma-Aldrich, Oakville, ON).   2.3.5 Data analyses Data for eTDP are expressed as means ± SD. Differences in women’s characteristics and eTDP were determined using one-way ANOVA and Tukey’s test with three groups (Phnom Penh, Prey Veng, Vancouver) or an independent samples t-test (Cambodia and Canada). Results were considered significant at P<0.05. All analyses were performed using SPSS for Macintosh version 22.0 (IBM Corp., Armonk, NY).  2.4 Results All Cambodian Village Chiefs we approached agreed to participate, and of the women recruited to participate, only one woman refused (361 women were invited to participate). Blood samples available for analysis were as follows: Phnom Penh, n=117; Prey Veng, n=121; Vancouver, n=47. A comparison of demographic characteristics and eTDP is outlined in Table 2-1. In this study, women were significantly older in Cambodia compared to women in Vancouver (P<0.001), likely due to the different sampling methods used in Canada and Cambodia. Approximately two thirds of Vancouver women  38 were of European descent, and one third were Chinese; all Cambodian participants were Khmer. The majority of participants had a body mass index (BMI) within the normal range. eTDP was significantly different among women living in different regions: Prey Veng, 149 ± 36 nM, Phnom Penh 156 ± 32 nM, and Vancouver, 179 ± 37 nM (P<0.05).  Table 2-1: Characteristics and eTDP of women aged 20-45 y in Cambodia and Canada1,2  Phnom Penh,  urban Cambodia Prey Veng,  rural Cambodia Vancouver,  urban Canada  n=117 n=121 n=47 Age, y 33 ± 7a 36 ± 6b 26 ± 5c Body Mass Index, kg/m2 23.1 ± 4.0a 21.7 ± 2.9b 22.0 ± 2.0ab      Underweight (≤ 18.5) 13 (11) 15 (12) 1 (2)      Normal (18.51 – 24.99) 69 (59) 92 (76) 42 (89)      Overweight (25 – 29.99) 28 (24) 12 (10) 4 (9)      Obese (≥ 30) 7 (6) 2 (2) 0 (0) Ethnicity         Khmer 117 (100) 121 (100)  0 (0)      European 0 (0) 0 (0) 32 (68)      Chinese 0 (0) 0 (0) 15 (32) Live Births3, number 2.4 ± 1.3a 2.9 ± 1.5a 1.7 ± 0.58a Education         None 5 (4) 11 (9) 0 (0)      Primary (years 1-6) 66 (56) 88 (73) 0 (0)      Secondary (years 7-12) 43 (37) 22 (18) 0 (0)      Higher education 3 (3) 0 (0) 47 (100) Number in Household 5.6 ± 2.5a 4.8 ± 1.6b 2.6 ± 1.2c Annual Household Income4,5, $US 2,943 ± 2,327a 1,482 ± 1,469a 65,251 ± 49,878b      Bottom 20% 59 (50) 93 (78) 3 (7)      Middle 60% 42 (36) 22 (18) 35 (83)       Top 20% 16 (14) 5 (4) 4 (10) eTDP, nM 156 ± 32a 149 ± 36b 179 ± 37c 1 Values are mean ± SD or number (%) 2 For continuous variables, means in a row without a common letter indicate a significant difference (P<0.05; Tukey post-hoc test) 3 Of women who have ever given birth: Phnom Penh, n=100; Prey Veng, n=107; Vancouver, n=3 4 Prey Veng, n=120; Vancouver, n=42 5 Participants classified using wealth quintiles (2011) for Canada (98) and Cambodia (99)   39 2.5 Discussion Thiamin deficiency has been described in Southeast Asia for many years, historically in Myanmar (68,70), Laos (67,78), Thailand (95), and more recently in Cambodia (11,73), but many of the studies were case reports, or focused on refugee populations, or individuals with clinical disease. In Prey Veng, Coats et al. (11) compared the whole blood TDP concentrations of mother-infant pairs of Cambodian infants with clinical symptoms of beriberi (n=27 pairs), without beriberi (n=27 pairs), and an American reference group (n=20 pairs). In their study, low whole blood TDP was prevalent among all Cambodian infants and mothers, regardless of the beriberi status of the infant. However, that study was limited to one district within Prey Veng, and recruited a convenience sample of mother-infant pairs seeking medical treatment at one health centre (11). Similarly, no women in the current study presented with symptoms of beriberi, so we echo Coats and colleagues’ call for more research to better understand the relationship between blood thiamin concentrations and the presentation of clinical disease.  Our study has a number of strengths. Most notably it used a population-representative sample of women in urban (Phnom Penh) and rural (Prey Veng) Cambodia with a very high response rate. However, we acknowledge a number of limitations. First, our sample of Vancouver women was a convenience sample of younger women recruited from a university campus, and therefore does not reflect the thiamin status of women of childbearing age in Canada. Second, we used simple random sampling instead of probability proportional to size sampling (PPS) to select the villages. PPS takes into consideration the size of the village and thus the potential importance of larger sampling  40 units. Fortunately villages in the sampling regions were roughly the same size, and there is little reason to believe thiamin intake differs dramatically between villages within the same region.   We did not collect information on women’s dietary intake in this study, so while we presume the lower eTDP among Cambodian women was caused by a lower dietary intake of these vitamins, this cannot be confirmed. However, a recent analysis of Food Balance Sheets was used to predict likely micronutrient deficiencies throughout the Western Pacific Region, and reported that thiamin is likely lacking in the Cambodian diet (80). Of the 17 countries investigated, Cambodia had both the lowest estimated daily per capita amount of thiamin (0.58 mg/d) and the lowest estimated daily per capita thiamin density per 1,000 kcal in the available food supply (0.29 mg) (80).   In this study we chose not to include pregnant and lactating women. Dietary thiamin requirements are higher in pregnant and lactating women (18,31), so while we did not include this group of women we presume that this group would also have lower eTDP than the Canadian women surveyed. The investigation of thiamin status among pregnant and lactating women in Cambodia has important public health implications for the prevention of infantile beriberi and warrants further investigation.  We acknowledge that at present there is a lack of established standard cut-offs for thiamin deficiency. The Institute of Medicine thiamin cut-offs are controversial as they are based on values from a small group of Dutch blood donors (n=98) in the 1980s  41 (18,100). Interestingly, these cut-off values are described in the Institute of Medicine guidelines as ‘erythrocyte thiamin’ (18), but in the original citation are referred to as cut-offs for ‘whole blood or red cells’ (100). These terms should not be used interchangeably; if these cut-offs are based on whole blood analysis, the erythrocyte cut-offs should be ~40% higher (taking into account the hematocrit). With this, we did not employ cut-offs in this study. It was also for this reason that we recruited Vancouver women to act as a reference group. The Institute of Medicine values as well as other published eTDP reference ranges came from relatively small sample sizes (for example, British laboratory staff, n=29 men, 16 women) (52), a convenience sample of male and female blood donors of unknown sample size (55), or adolescent boys (n=19) and girls (n=35) (56), and all were published in the mid-1990s before the current HPLC method was employed (51), making comparisons difficult. However, while the Vancouver group who formed the reference group were all women of childbearing age, there are also limitations: the Vancouver women were recruited using convenience sampling, and no dietary data was collected so may not all have actually had adequate thiamin intake. Future research should focus on the development of clinically meaningful cut-offs for thiamin deficiency.     2.6 Implications and conclusions We found lower eTDP among women of childbearing age in rural and urban Cambodia compared to women in Vancouver, Canada. Given the absence of infantile beriberi in Canada, strategies such as supplementation, fortification, and/or food-based interventions may be warranted to improve the thiamin status of Cambodian women.   42 Chapter 3: Thiamin-iron fortified fish sauce: stability and sensory perceptions in rural Cambodia 3.1 Summary Background: Infantile beriberi remains a cause of infant mortality in Cambodia as low maternal thiamin (vitamin B1) consumption results in production of thiamin-poor breast milk. We formulated two thiamin-fortified fish sauces, low concentration (LC; 2 g thiamin hydrochloride (THCl)/L fish sauce) and high concentration (HC; 8 g/L) as a means of increasing maternal thiamin intake with potential to prevent infantile beriberi. Objective: To test the stability of thiamin in fish sauce after exposure to light, oxygen, and heat in the laboratory, and under typical storage and use conditions in rural Cambodian households. In addition, we aimed to test consumer acceptability of newly formulated fish sauces through sensory evaluation, as compared to a control sauce containing no thiamin.  Methods: Thiamin-fortified fish sauce was exposed to light, oxygen, and light + oxygen for up to 21 days, and exposed to heat at 100°C for up to 35 minutes. Sensory evaluation was conducted among 90 women aged 18-45 y in Prey Veng, Cambodia using the Triangle Test, Paired Preference Test, and a Nine Point Hedonic Scale.   Results: THCl was shelf stable in fish sauce. Mean ± SD of THCl in household samples was 2.3 ± 0.4 and 9.7 ± 1.0 g/L for LC and HC fish sauces, respectively. Light, oxygen, light + oxygen, and heat exposure did not cause more than 10% degradation in most samples. Women enjoyed the taste of the newly formulated fish sauces, and only approximately one third of women could correctly identify the different sample in the Triangle Test, indicating that women were unable to differentiate fortified and control  43 fish sauces. Conclusions: Thiamin is stable in a fish sauce matrix, and thiamin-fortified fish sauce is acceptable to rural Cambodian consumers.   3.2 Introduction Thiamin is an essential vitamin involved in glucose metabolism and neuronal transmission (13). Thiamin deficiency is rare in western countries where thiamin-rich foods are consumed from a variety of sources, and some foods may be fortified (13,36,64,101). Unfortunately thiamin deficiency is still common in Southeast Asia, in part due to a heavy reliance on non-parboiled (44,45), thiamin-poor, polished white rice (11,68,78,95,97). We recently reported significantly lower blood thiamin concentrations among a representative sample of women of childbearing age in rural Prey Veng province, Cambodia, compared to a similar group of purportedly vitamin-replete Canadian women (93,94). Thiamin deficiency presents as infantile beriberi among breastfed infants aged two to four months consuming thiamin-poor breast milk from thiamin deficient mothers (13,57). Since breast milk is recommended as the sole source of nutrition for infants under 6 months (71), maternal dietary thiamin intake must be improved to increase breast milk thiamin, and prevent infantile beriberi and related mortality (45).   Major dietary sources of thiamin include whole grains including brown rice and rice bran, legumes, and pork (13), foods that are either not commonly consumed or are costly in Cambodia. Fortification is an ideal means of increasing population-wide dietary thiamin  44 intake because it is a sustainable, low cost, and passive intervention that requires no behaviour change (87–89). Fish sauce is consumed by an estimated 90% of Cambodians (85), and has been used as a fortification vehicle for iron in Vietnam (102,103) and Cambodia (84–86). Fortification of fish sauce with iron and thiamin is a potential means of addressing both micronutrient deficiencies using pre-existing factory infrastructure.   Thiamin degrades when exposed to neutral and alkaline pH, oxygen, heat (38,104,105), and although there is still some debate, light (13,31,38,105,106). Although fish sauce is acidic (107), typical Cambodian storage and use conditions impose risk of thiamin degradation through exposure to oxygen, light, and heat (cooking), potentially reducing efficacy of this product. In addition to stability, consumer acceptability of thiamin-fortified fish sauce is essential for uptake of this intervention. Two thiamin degradation products, 2-methyl-3-furanthiol (MFT) and bis(2-methyl-3-furyl) disulfide (MFT-MFT) are produced when thiamin degrades at high temperatures, producing a meaty aroma and flavour (108,109). Although typically associated with cooked meats, such as a recent report of thiamin degradation to MTF and MTF-MTF in hams cooked at 69°C for 120 min (109), they have been shown to appear at even lower temperatures, for example, in aqueous solution after storage for 7 days at 35°C (108), an ambient temperature which is not atypical in Cambodia. Therefore the objective of this study was twofold: first, to determine the stability of thiamin within a fish sauce matrix after exposure to common degraders (light, oxygen, heat) in a laboratory setting (mimicking household storage), and during real life, household storage conditions in rural Cambodia; and secondly, to  45 determine consumer acceptability through sensory evaluation of thiamin-fortified fish sauce.   3.3 Methods In collaboration with Leang Leng Enterprises (Phnom Penh, Cambodia), we formulated three fish sauces for this study, two thiamin-fortified and one control (no detectable thiamin); all three fish sauces contained iron as per Cambodian government guidelines (86). These fish sauces were later employed in a randomized controlled efficacy trial in Prey Veng, Cambodia to determine the effect of consumption on biochemical thiamin status.   3.3.1 Fish sauce formulation and production The World Health Organization recommends using the EAR cut-point method to determine optimal fortificant doses (87). This approach requires the target population’s usual intakes of both the fortification vehicle and the micronutrient of interest (87). Using both usual intakes, the optimal dose is calculated so that only approximately 2.5% of the target population consumes less than the EAR, while only 2.5% consume quantities of the micronutrient above the UL (87). Note that thiamin does not have a UL as no adverse events have been reported from excess thiamin intake (18). Following this approach for thiamin-fortified fish sauce, the thiamin content of fish sauce should ideally increase thiamin consumption among the target population, pregnant and lactating women, to shift usual intake of thiamin upwards. If, for example, we assumed 100% thiamin deficiency among pregnant and lactating women, and assumed no thiamin sources in the diet, daily  46 consumption of 10 mL of fish sauce fortified at 0.14 g/L would provide 1.4 mg/d. See Figure 3-1 for a hypothetical example of a usual intake distribution highlighting the thiamin EAR and RDA for pregnant and lactating women (18). Unfortunately, there are no published dietary intake data reporting thiamin or fish sauce intakes of Cambodian women. However, recent food balance sheet data indicates that only 0.58 mg/d thiamin is available per capita in Cambodia (80), an amount well below the EAR for pregnant and lactating women of 1.1 mg/d (18). Regarding fish sauce intake, a study evaluating iron-fortified fish sauce in Kampot province, Cambodia, required school children consume 10 mL fish sauce daily for 5 months as part of a school meal program (84). The authors of this study did not describe how this value was obtained, but did acknowledge that usual fish sauce consumption may be much lower than 10 mL/d, even among adults (84).   Figure 3-1: Schematic of hypothetical intake distributions using the EAR cut-point approach for thiamin fortification. Black dashed line: potential current thiamin intake distribution from food balance sheet estimates (80); grey line: resulting upward shift in usual thiamin intake distribution after introduction of thiamin-fortified fish sauce so that only 2.5% of target population consumes less than the EAR (18).  With limited data and no distribution ranges for usual intake, we were unable to employ the EAR cut-point method. Instead, because this fortified fish sauce was intended a  47 proof-of-concept study, relatively high concentrations of 2 and 8 g/L thiamin in fish sauce were formulated in order to identify an effect, if present. Further, as highlighted in 1.2.3.1, the EAR for thiamin was based on a very weak evidence base. Since the fish sauce would be employed for a short-term (6 mo) intervention with regular fortnightly check-ins, and since there have not been reports of adverse events from excess thiamin intake, even with over-the-counter supplements containing 50 mg and more (18), we were not as concerned about harm from excess intake over 6 mo.    Fish sauce was fortified with thiamin as thiamin hydrochloride (THCl; ≥ 98% purity; Huazhong Pharmaceutical Co., Ltd, China) and iron as ferric sodium ethylenediaminetetraacetate (NaFeEDTA; Ferrazone®, Akzo Nobel Functional Chemicals B.V., The Netherlands). Three fish sauces with differing thiamin concentrations were formulated and bottled (one day, one 2300 L batch per formulation): low concentration thiamin (LC, 2 g/L THCl), high concentration thiamin (HC, 8 g/L THCl), and control fish sauce (no detectable thiamin).   Fish sauce was pasteurized at 100°C for 90-120 min then cooled to room temperature. Fortificants were added and stirred with a mechanical mixer for 30 min before bottling in identical 750 mL transparent plastic bottles. A 10% overage of THCl was added to LC and HC fish sauces to counter potential manufacturing losses. Fish sauce samples (5 mL) were collected every 15 min from the bottling line (~2 h; n=10, 10, and 7 for LC, HC, and control, respectively) to verify fortification. All fish sauce was fortified with 2.8 g/L NaFeEDTA according to Cambodian government guidelines (86), and received  48 Cambodian Ministry Health Certificates for Hygienic Food Production. Within 48 h of production, fish sauce was transported from the Leang Leng Factory in Phnom Penh to the Svay Antor District of Agriculture Office (~100 km east) for storage in a dark, dry room at ambient temperature (~25-28°C) until distribution to households (for household stability experiment). Several bottles of fish sauce were transported to Vancouver, Canada for controlled exposure experiments; there, fish sauce was stored in a dark, dry cupboard at ambient temperature (~20°C).  3.3.2 Thiamin stability 3.3.2.1 Laboratory exposure experiments Thiamin stability in LC and HC fish sauces was assessed after exposure to light, oxygen, light + oxygen, and heat (100°C) under laboratory conditions mimicking household storage. Duplicate samples of LC and HC fish sauces (5 mL in capped 10 mL transparent plastic centrifuge tubes; Eppendorf, Mississauga, ON) were UV-exposed on a southeast-facing windowsill (clear glass) in Vancouver, Canada (approximately 15 h sunlight daily in July 2015) at room temperature (~20°C). Identical uncapped (oxygen exposed) fish sauce sample duplicates were stored in a dark cupboard at room temperature (~20°C). A third experiment exposed duplicate samples to both oxygen and light: uncapped samples were placed on the same southeast-facing windowsill. For all three experiments, samples were exposed for 0 h, 8 h, 24 h, and 2, 4, 7, 10, 14, and 21 d, and then stored at -80°C until batch THCl analysis. Exposure times were based on self-reported time to finish one-750 mL fish sauce bottle in households with 3 to 6 people in Prey Veng, Cambodia (mean ± SD days to finish one bottle = 14 ± 7 d, n=107 families). For oxygen and light +  49 oxygen exposure experiments, samples were weighed before exposure; samples were returned to their original volume with deionized water to control for volume loss due to evaporation.  Fish sauce was also exposed to heat to simulate cooking at 100°C (i.e. soup). Aliquots of 5 mL fish sauce in glass tubes were placed in a 100°C heating block (Analog Dry Block Heater, VWR International, Mississauga, ON) for 1, 3, 5, 7, 10, 15, 20, 25, 30, and 35 min. Samples were removed from heat after the allotted time, cooled to room temperature, brought back to their original volume with deionized water, and then stored at -80°C. Heat exposure duration was based on self-reported cooking time for Khmer soup for mean ± SD of 22 ± 14 min (n=83 women); note, however, that many women add fish sauce as final ingredient immediately before consumption.  3.3.2.2 Household shelf-stability Fish sauce samples (0.5 mL) were collected from randomly selected households every two weeks between October 20, 2014 and April 6, 2015 (n=160) to determine thiamin stability under actual household storage and use conditions (storage temperatures ~26-28°C). Fish sauce samples were collected from the fish sauce bottle currently in use during regularly scheduled fortnightly home visits. Sealed samples were stored in a dark container and transported to the National Institute of Public Health in Phnom Penh within three days of collection for long term storage at -80°C. Samples were batch shipped on dry ice to Vancouver, Canada for THCl analysis.   50 3.3.3 Thiamin hydrochloride analysis Fish sauce samples were mixed vigorously and diluted (LC, 160,000x; HC, 400,000x) before a 150 µL aliquot was transferred to a 96 well plate. HPLC analysis was performed using an Agilent 1260 Infinity system with a Poroshell 120 EC-C18 column (3.0 x 50 mm with 2.7 µm; Agilent Technologies, Mississauga, ON) at 25°C with the detector set at an excitation wavelength 375 nm and emission wavelength 435 nm. Mobile phase A consisted of 25 mM sodium phosphate (pH 7.0) and methanol (90:10 vol/vol), while mobile phase B consisted of 25 mM sodium phosphate (pH 7.0) and methanol (30:70 vol/vol). Prior to injection, online sample derivatization (methanol, sample, and 1.2 mM potassium ferricyanide in 15% wt/vol sodium hydroxide) was performed by automated injector programming. THCl (as equivalents in fish sauce) was quantified based on peak area and external standardization using THCl calibration solutions (0-20 g/L; Sigma-Aldrich, Oakville, ON).  3.3.4 Sensory evaluation 3.3.4.1 Participants A convenience sample of 90 women of childbearing age (18-45 y) was recruited from the Prey Veng provincial market in Prey Veng Town, Prey Veng, Cambodia, over two days, 7 days after fish sauce was bottled. Women were selected as consumer panelists because they are the primary end-users (cooks) in rural Cambodia, and therefore deemed most likely to perceive sensory differences between fish sauces. Ethics approval was obtained from the National Ethics Committee for Health Research (0245NECHR) in Cambodia,  51 and the University of British Columbia Clinical Research Ethics Board (H14-02173) in Canada. Written informed consent was obtained from all women.  3.3.4.2 Study procedure Women completed a discriminative test, the Triangle Test, and two affective tests, the Paired Preference Test and the Nine Point Hedonic Scale (110,111). These simple sensory evaluation tests were chosen because they can be completed by minimally trained consumer panelists (112). Before sensory evaluation, women were asked demographic questions including their age, and questions regarding general attitudes towards fish sauce and typical fish sauce consumption behaviours (see questionnaire in Appendix B).    Women determined the time between tests based on their feeling of preparedness for the next tasting. Two trained interviewers conducted sensory evaluation in Khmer language and recorded all responses. Women were blinded to fish sauce samples, and all sauce was served in identical 30 mL plastic cups containing 5 mL of fish sauce at ambient temperature (~25-28°C). A bottle of water and unsalted crackers were provided to women to cleanse their mouths between samples. Women were informed that we were seeking an honest and free expression of their opinions towards the fish sauce samples, and that there were no correct answers to any of the tests employed.  3.3.4.3 Triangle test The goal of the Triangle Test is to determine whether statistically significant sensory differences exist between two fish sauces (112). Women were given three samples, two  52 the same and one different, and asked to identify the different sample (110). The probability of the woman selecting the different sample by chance alone is one third (110). Women were randomly assigned to receive one of six possible triangle test sample combinations (sets). Three coded samples were provided to each woman at one time, and she was asked to identify the different sample.   3.3.4.4 Paired preference test In the Paired Preference Test, women were asked to taste two products and identify the preferred sample (111). Only control and HC fish sauces were assessed because if any difference in sensory preference truly existed due to the added THCl, these differences would likely be most apparent between control and HC rather than LC. The order of presentation of control and HC fish sauce was random.   3.3.4.5 Nine point hedonic scale Neither the Triangle Test nor the Paired Preference Test allows women to describe whether they ‘liked’ or disliked’ the products sampled (111), so a Nine Point Hedonic Scale was also employed. This scale is a simple, accurate, and reliable means of assessing likes and dislikes of a wide range of foodstuffs and beverages (112) across a range of populations of untrained consumer panelists (113). In this test women were asked to taste the three fish sauce samples, one at a time, and assess their degree of liking or disliking each sample using the following nine-point scale: (1) dislike extremely; (2) dislike very much; (3) dislike moderately; (4) dislike slightly; (5) neither like nor dislike; (6) like slightly; (7) like moderately; (8) like very much; or (9) like extremely (113). In Khmer  53 language, liking or disliking food translates to how delicious that food is perceived to be, therefore the Khmer translation of this scale ranged from “extremely not delicious” to “extremely delicious”. Women tasted samples one at a time in a randomized order and were asked to verbally score the samples.    3.4 Statistical analysis Descriptive statistics were computed for THCl analysis and expressed as means ± SD (g THCl /L fish sauce). For sensory analysis, means ± SD or n (%) were computed for demographic characteristics, general attitudes and consumption patterns of fish sauce, and all sensory tests. For the Triangle Test, results were divided into groups based on the different sample in the set. The critical number of correct responses to determine whether the different sample was chosen by chance alone (one tailed p=1/3; n=30) was 15 women (P=0.043) (114). That is, if 15 or more women correctly identified the different sample, there is said to be a discernable sensory difference. For the Paired Preference test analysis, a binomial test with a test proportion of 0.5 was used to determine whether there was a statistically significant difference in the proportion of women who preferred control or HC fish sauce. Results were considered significant at P<0.05. All analyses were performed using SPSS for Macintosh version 22.0 (IBM Corp., Armonk, NY).   3.5 Results 3.5.1 Fish sauce fortification at factory The average THCl concentration of LC and HC fish sauces (n=10 each, analyzed in triplicate) was 2.0 ± 0.15 and 8.7 ± 0.3 g/L, respectively. The control fish sauce (n=7) did  54 not contain any detectable thiamin. There was random variation in the thiamin concentrations throughout the bottling process: the THCl content of samples compared to the added fortificant ranged between 81 – 101% in LC fish sauce (of 2.2 g/L) and 93 – 104% in HC fish sauce (of 8.8 g/L).  3.5.2 Exposure to light, oxygen, and heat Thiamin concentrations of fish sauce samples after exposure to light, oxygen, both light and oxygen, and heat, can be found in Figure 3-2. Duplicate samples were measured at each time point; ± 10% THCl from baseline mean (calculated from triplicate samples) is shown with dotted lines. There was minimal THCl loss greater than 10% across all exposures. Little THCl degradation was seen after cooking at 100°C for up to 35 min, especially with HC fish sauce.  3.5.3 Household storage THCl concentrations of fish sauce stored and utilized in rural Cambodian households over 6 months can be found in  Figure 3-3 (expressed as mean ± SD for each time point). THCl varied more amongst HC samples than LC samples. Mean ± SD over six months was 2.3 ± 0.4 g/L and 9.7 ± 1.0 g/L for LC and HC fish sauces, respectively (n=53 each). Interestingly, these means are higher than the mean TCHl concentrations from samples off the bottling line (LC: 2.0 ± 0.15 g/L and HC: 8.7 ± 0.3 g/L), indicating variability across samples. However, only two LC fish sauce samples (collected from households at weeks 20 and 24) measured less than 2 g/L, and no HC samples were less than 8 g/L.   55  Figure 3-2: Thiamin hydrochloride concentrations (g/L) of fish sauce sample duplicates after laboratory exposure to light (A), oxygen (B), light and oxygen (C) and heat (D; 100°C). LC,|; HC ,z; dotted lines show baseline mean ± 10%; m, minutes; h, hours; d, days. 56  Figure 3-3: Thiamin hydrochloride concentration of fish sauce (g/L) collected fortnightly from households in Prey Veng province, Cambodia over 6 months. Mean ± SD at each timepoint: LC, grey line; HC, black line; n=53 each, n=4-5 samples per collection timepoint.   3.5.4 Sensory analysis Demographic characteristics of women who participated in sensory evaluation of the formulated fish sauces can be found in Table 3-1. As expected based on the convenience sampling method deployed, the majority (94%) of women had a positive attitude towards fish sauce, while no women had negative feelings. All 90 women reported consumption of fish sauce on a regular basis, and 81% purchased fish sauce regularly.      57 Table 3-1: Demographic characteristics and general attitudes towards fish sauce of rural Cambodian women (18-45 y; n=90) Demographic Characteristic  Mean ± SD or n (%) Age, years  34 ± 8 Pregnant  4 (4) General Attitude Towards Fish Sauce        Negative  0 (0)      Neutral  5 (6)      Positive  85 (94) Fish sauce consumed regularly  90 (100) Fish sauce purchased regularly  73 (81)  For HC, LC, and control, respectively, 11, 8, and 11 women correctly identified the different sample in the Triangle Test (Table 3-2). Since fewer than 15 women correctly identified the different sample we reject the assumption that statistically significant sensory differences existed: women could not discern the fish sauces. While more women preferred control (n=54; 60%) to HC fish sauce (n=36; 40%) in the Paired Preference Test, this was not significantly different (P=0.07).   Table 3-2: Correctly identified ‘different’ sample in Triangle Test by rural Cambodian women (18-45 y; n=90) Different Sample in  Triangle Test Set n Different Sample Correctly Identified1 P2 Control 30 11 (37) 0.415 Low concentration  30 8 (27) 0.833 High concentration  30 11 (37) 0.415 1 Values are number (%) 2 Probability of obtaining X or more correct responses in Triangle Test (one tailed P=1/3; n=30) (114)  All fish sauces fared similarly in the Nine Point Hedonic Scale test, with the same mean and standard deviation, median, and range for all three sauces (Table 3-3). In general, women tended to like all fish sauces, with nearly 70% obtaining scores ≥ 6 for each fish  58 sauce. Indeed, 31 women (34%) scored all three fish sauces ≥ 6. Fewer than 20% of women disliked any of the fish sauces.   3.6 Discussion THCl was stable in fish sauce, both during routine storage and use in rural Cambodian households, and under laboratory exposure conditions designed to mimic household conditions. Based on scores provided during the Nine Point Hedonic Scale test, women liked all three fish sauces evaluated in this study. There was not a significant difference in the proportion of women who preferred control or HC fish sauces, and women were unable to discern the three different fish sauces.   59 Table 3-3: Sensory evaluation of thiamin-fortified fish sauce by rural Cambodian women (18-45 y; n=90) using the Nine-Point Hedonic Scale  Fish Sauce Formulation Mean ± SD Range Median Mode Categories of Nine-Point Hedonic Scale1,2 Dislike Neutral Like Control 6.5 ± 2.1 1 - 9 7 9 14 (~16) 17 (19) 59 (66) Low concentration 6.5 ± 2.1 1 - 9 7 7 14 (~16) 14 (~16) 62 (69) High concentration 6.5 ± 2.1 1 - 9 7 9 12 (13) 16 (18) 62 (69) 1 Values are n (%). 2 Categories of Nine-Point Hedonic Scale: Dislike, scores 1-4; Neutral, score=5; Like, scores 6-9. 60 The within-batch variation of THCl concentrations from the bottling line (LC, 81 – 101%; HC, 93 – 104%) is not uncommon or unexpected. However, THCl concentrations of LC and HC fish sauce samples collected from households had much wider sample variability, ranging from 1.3 – 3.5 g/L (59 – 159% of target) and 8.2 – 12.0 g/L (93 – 136% of target), respectively. This sample variability could be explained by several factors, including that samples were collected from 106 unique households with different storage conditions (storage on table, in cupboard, sealed, or not properly capped) from bottles at different stages of use (newly opened bottles to final dregs). It could also be that the samples from the bottling line did not capture the true variability in the fish sauce either because an insufficient number of samples were collected, or due to factors such as incomplete fortificant mixing, or potential dilution with fish sauce already in the bottling line. Compared to Ferrazone®, which is similar in consistency to granulated sugar, THCl is much lighter and powdery, more similar to the consistency of powdered sugar. With values as high as 3.5 and 12.0 g/L in LC and HC sauces respectively, it may be that while 30 minutes of mechanical stirring in a 2300 L drum is adequate for Ferrazone®, the THCl may not have fully mixed. However, no factory workers, field staff, or study participants saw clumps of THCl in the mixing drums or in individual bottles. Alternatively, some of this variation may have occurred in the household: THCl may have settled at the bottom of the bottle, so samples taken from nearly empty bottles may have had higher THCl concentrations. Or, if bottles were improperly sealed, fish sauce evaporation may have artificially increased THCl concentrations. Further research should investigate the optimal mixing time of THCl in fish sauce to minimize sample variability, and should examine the effects of evaporation and THCl precipitation.    61 As shown in Figure 3-3 very few samples showed losses of more than 10% of baseline measurements. Although thiamin is known to be sensitive to oxygen and heat (38), exposures to these conditions in the laboratory had little effect on thiamin stability within a fish sauce matrix, potentially due to the pH and amino acid content of fish sauce. While nearly 100% of thiamin has been shown to degrade after boiling for 20 minutes at pH 9, little degradation occurs under acidic conditions (38). Therefore the acidic pH of fish sauce (107) was likely protective against thiamin degradation. In addition, the presence of α-amino acids has shown to enhance thiamin stability slightly, even at neutral and alkaline pHs (most notably between pH 4.5 – 7) (104). McIntire & Frost reported 27% thiamin degradation after exposure to 100°C for 4 h at pH 6, but in the presence of glycine this loss decreased to 21% (104). Fish sauce is a rich source of amino acids, containing ~16 g/L fish sauce; a recent analysis revealed 426 mg glycine per 100 mL fish sauce (115). There is debate in the literature surrounding light-induced thiamin degradation. While general consensus is that thiamin, like other B vitamins, is sensitive to light (13,31,38,105), Chen and colleagues reported 26% loss of THCl in a parenteral nutrition solution after exposure to direct sunlight for 8 h, but no degradation after exposure to indirect sunlight and fluorescent light (106). In the context of this study, thiamin was likely protected by the acidic pH and amino acid content of fish sauce, as well as indirect sunlight exposure.    The Nine Point Hedonic Scale can be employed with little training (112), however there is evidence that Asian participants use the scale differently than American panelists, the population group more commonly engaged in sensory analysis (116). Yeh and colleagues compared the use of the Nine Point Hedonic Scale among American, Chinese, Korean, and Thai participants and found that while Americans tend to score more frequently on the extreme ends of the scale  62 (scores 1-3 and 7-9), Asian participants often use the central scores (scores 4-6) (116). However, while only 9 (10%), 11 (12%) and 10 (11%) of women in this study indicated a score in the lowest tertile of the Nine Point Hedonic Scale (scores 1-3) for HC, LC, and control fish sauces, respectively, a large proportion of women, 50 (56%), 55 (61%), and 51 (57%), scored HC, LC, and control sauces in the highest tertile (scores 7-9), respectively. Therefore, the Cambodian women in this study appeared to make full use of the scale. This difference may be attributed to the popularity of fish sauce among Cambodians (85) especially among this convenience sample of women, of whom 94% reported having a positive attitude towards fish sauce.  MTF and MTF-MTF, common degradation products of thiamin-rich foods have strong meaty flavours (108,109) that may affect consumer acceptability of fortified fish sauce. Since we report minimal thiamin degradation, it is unlikely that MTF and MTF-MTF were produced in high enough quantities to have affected the smell and taste of the fish sauce (117). Indeed, since fish sauce also has a powerful, somewhat meaty flavour (118) it may be that off-flavours caused by MTF and MTF-MTF, if present, were not detected by participants. In addition, the strong salty taste of fish sauce may mask thiamin fortification and thiamin degradation products. Since fish sauce must contain no less than 200 g/L salt (107), it may be that the addition of thiamin as a salt (THCl) at relatively low concentrations (2 and 8 g/L) does not greatly alter the taste of fish sauce, especially considering the strongest taste of fish sauce is saltiness (118).  Fish sauce is one of the most commonly consumed condiments in Cambodia and throughout Southeast Asia (84), making it an ideal vehicle for micronutrient fortification. Both thiamin-iron- and iron-fortified fish sauces were equally well accepted and not discernable by rural women of  63 childbearing age, the household cooks of Cambodia. As such, there is potential for other micronutrients to be added along with thiamin and iron as a means of combating multiple micronutrient deficiencies in Cambodia and the wider region with a single intervention using pre-existing factory infrastructure.   3.7 Conclusions This study documents the first stability and sensory evaluation of thiamin-iron-fortified fish sauce. THCl was relatively stable in a fish sauce matrix both during 6 months of typical storage and use in rural Cambodian households, and after exposure to light, oxygen, and heat under laboratory conditions. The fortified fish sauces were well accepted by rural women aged 18 – 45 y in Prey Veng province, Cambodia, with >65% of women liking the sauces (scores ≥ 6 on a Nine-Point Hedonic Scale). There was no significant difference in the preference for control or HC fish sauces, and only approximately one third of women correctly identified the different sample in the Triangle Test, indicating that women were unable to differentiate the three fish sauces. Since the two concentrations of thiamin-iron-fortified fish sauces were both well accepted and not discernable from control fish sauce fortified with only iron, we expect these formulations will be well accepted by future rural Cambodian consumers. If brought to scale, thiamin-iron-fortified fish sauce has the potential to be a sustainable means of combating infantile beriberi in rural Cambodia, and potentially throughout Southeast Asia.     64 Chapter 4: Household consumption of thiamin-fortified fish sauce increases erythrocyte thiamin concentrations in rural Cambodian women and their children under 5 years: a randomized controlled trial  4.1 Summary Background: Poor thiamin status is prevalent in rural Cambodia due to heavy consumption of thiamin-poor white rice and a lack of dietary diversity, which could lead to potentially fatal infantile beriberi among breastfed infants. In regions where dietary thiamin intake is adequate, women of childbearing age are purportedly thiamin-replete and beriberi is uncommon.  Objective: To determine whether ad libitum consumption of thiamin-fortified fish sauce over six months could increase erythrocyte thiamin diphosphate concentrations (eTDP) among women of childbearing age and their youngest children aged 12-59 mo, compared to control sauce containing no thiamin. Methods: In this double-blind, randomized controlled efficacy trial 276 non-pregnant, non-lactating women aged 18-45 y and their families in Prey Veng, Cambodia, were individually randomized to receive one of three fish sauce formulations: low concentration (LC, 2 g/L), high concentration (HC, 8 g/L), or a control sauce. eTDP was measured using HPLC-FLD. Results: Fish sauce consumption did not differ among treatment groups (P=0.19). In intent-to-treat analysis, women’s baseline-adjusted endline eTDP (estimated marginal mean; 95% CI) was significantly higher among women in LC (232; 220, 244 nM) and HC (231; 219, 244 nM) groups compared to control (175; 163, 188 nM; P<0.001). Similarly, baseline-adjusted eTDP was significantly higher among children in LC (243; 233, 254 nM) and HC (229; 219, 240 nM) groups compared to control (212; 202, 223 nM; P<0.05).    65 Conclusions: Women and their children who consumed thiamin-fortified fish sauce at concentrations of 2 and 8 g/L for six months had significantly higher eTDP compared to those consuming a control fish sauce.   4.2 Introduction Infantile beriberi, a potentially fatal disease caused by a lack of thiamin (vitamin B1) among breastfed infants aged 2-4 months (13,57), was common amongst Karen refugees living in camps on the Thai-Burmese border in the 1980s (68,70). Thiamin deficiency is most serious and life-threatening in infants (67) because the developing brain is highly sensitive to the effects of deficiency (13). Low maternal dietary thiamin intake results in thiamin-poor breast milk (57); unfortunately mothers of infants with beriberi are often asymptomatic (11,13,67). Increasing the thiamin intake and status of lactating women could prevent infantile beriberi and related infant mortality. Maternal thiamin supplementation addressed infantile beriberi in these refugee camps (68), but thiamin deficiency remains a public health concern in this region (11,29,45,77,78,93,95) where B-vitamin poor, polished white rice is the dietary staple (119,120). Elsewhere, infantile beriberi has been nearly eradicated through diverse diets, and/or thiamin fortification or enrichment (36,37,64,87,88). In Cambodia, where non-parboiled (44), polished white rice is estimated to account for up to 60% of dietary energy (8) and dietary diversity is low (1,5,120), reports of beriberi are not uncommon (11). We recently reported that erythrocyte thiamin diphosphate concentrations (eTDP) were significantly lower among non-pregnant, non-lactating rural Cambodian women of childbearing age than purportedly B-vitamin replete Canadian women (93).    66 Strategies may be needed to improve thiamin intake in Southeast Asia (89). In rural Cambodia where there is limited access to thiamin-rich foods such as whole grains and animal-source foods (1,46), thiamin fortification could be a sustainable, low cost (102), and passive approach to improve thiamin status (87). Fortification of fish sauce, a popular traditional condiment already used as a fortification vehicle for iron (84,85,102,103,121), is a population-wide intervention that has the potential to improve the thiamin intake not only of lactating women for the prevention of infantile beriberi, but of all consumers. Since beriberi is uncommon in populations with adequate thiamin intake, the aim of this study was to determine whether ad libitum consumption of a newly formulated thiamin-fortified fish sauce for six months would yield higher eTDP among rural Cambodian women of childbearing age and their children under 5 years compared to those who consumed a control fish sauce containing no thiamin.   4.3 Methods 4.3.1 Study design This was a community-based, double-blind, randomized controlled efficacy trial, where we recruited women and their families from villages in Prey Veng province, Cambodia. Villages participating in other non-governmental nutrition-related programs were excluded. Using village registries, all female-heads-of-household aged 18-45 y with at least one child between 12-59 mo were invited to participate in the study. Women were excluded if they were taking a thiamin-containing supplement, planned to leave their village (i.e. for work) during the next six months, or if they did not agree to exclusively use the study fish sauce. Women provided informed written consent on behalf of themselves and their child. Ethical approval was obtained from the  67 National Ethics Committee for Health Research in Cambodia (0245NECHR) and the University of British Columbia Clinical Research Ethics Board in Canada (H14-00103).   4.3.2 Fish sauce  In partnership with local fish sauce producer Leang Leng Enterprises (Phnom Penh, Cambodia), we produced three fish sauce formulations for this study: low concentration thiamin (LC; 2 g/L thiamin hydrochloride (THCl; ≥ 98% purity), Huazhong Pharmaceutical Co., Ltd, China, donated by Natural Factors, Coquitlam, British Columbia, Canada), high concentration thiamin (HC; 8 g/L THCl), and control, which had no detectable thiamin. Thiamin fortification dose in fish sauce is discussed in 3.3.1. All fish sauces were fortified with iron as ferric sodium ethylenediaminetetraacetate (2.8 g/L Ferrazone®, Akzo Nobel Functional Chemicals B.V., The Netherlands) as per Cambodian Ministry of Planning guidelines (86). Fish sauce was pasteurized for 90-120 min at 100°C in 2300 L drums, cooled to room temperature, and then fortificants were added and mixed for 30 min before bottling. All fish sauce used in the study received Cambodian Ministry of Health Certification for Hygienic Food Production.   Fish sauce packaging was identical among the three groups except for a unique serial number stamp on each bottle, and different coloured caps for different study arms intended to prevent confusion amongst field staff and participants. Green, orange, and purple caps were selected based on feedback from 12 focus group discussions (n=60 women aged 18-45 y from 12 villages within 3 communes in Prey Veng, Cambodia; see semi-structured guide in Appendix C), which highlighted these colours as pleasant yet culturally and politically neutral in Cambodia. The blinding code was known to Leang Leng Enterprises, and was stored in a sealed envelope held  68 by the Principal Investigator in case of emergency. Sensory evaluation using the triangle test (110) indicated that rural Cambodian women (n=90) were unable to discern the three formulations of fish sauce.  4.3.3 Randomization Before study recruitment, a master ID list was created with group assignments using online randomization software (GraphPad Software Inc., La Jolla, CA) to allow for non-stratified, individual, equal randomization to arms. Cambodian field staff then assigned ID numbers sequentially during baseline data collection. Women received their first fish sauce after baseline blood collection, then every two weeks thereafter during household visits from field staff. Women were instructed to consume the study fish sauce ad libitum as they normally would; since Cambodian families eat family-style from a common pot, the entire household consumed the study fish sauce during the 6 month intervention. In November 2014 (t=1 mo), all study participants and their family members (husbands, sisters, mothers, mothers-in-law etc.) were invited to a nutrition education workshop in their village where they learned about thiamin deficiency and beriberi, as well as infant and young child feeding practices according to Cambodian Ministry of Health guidelines (81).  4.3.4 Weighed fish sauce record To obtain accurate fish sauce intake data we conducted a weighed fish sauce record with a randomly selected subset of 28 households from the three treatment groups. A trained enumerator weighed all fish sauce consumed by the mother, husband, and youngest child aged 12-59 mo between dawn and dusk (~12 hours) on three non-consecutive days (two week days,  69 one weekend day; different enumerators for each day) within a two week window. The family was instructed to continue with their day normally; if the family cooked with fish sauce the enumerator weighed the meal and individual meal servings using a battery-powered electronic balance (Salter Aquatronic, Oak Brook, IL, USA). Fish sauce consumption was calculated by multiplying the weight of fish sauce-containing food consumed by the amount of fish sauce added to the meal (see Appendix D). Data were collected in December 2014 (t=2 mo) to document consumption during early dry season to balance seasonal dietary intake (mid-point between lean and rich seasons), and to avoid the potential increased fish sauce consumption at the beginning of the study due to the novelty of receiving free fish sauce.   4.3.5 Data and blood collection Demographic, dietary, and subjective information regarding fish sauce was collected at baseline (October 6-17, 2014) and endline (t=6 mo; April 22-29, 2015) using an interviewer-administered questionnaire in women’s homes (see questionnaire in Appendix E). The next morning a non-fasting venous blood sample was collected into evacuated EDTA-coated tubes (Vacutainer, Becton Dickinson; Mississauga, ON) from each woman and her youngest child aged 12-59 mo at a central village location. Blood was stored on ice, and was transported within 5 h to the National Institute for Public Health (NIPH) laboratories in Phnom Penh for processing. Blood samples were centrifuged at 3000 rpm for 15 min at 4°C, and plasma and buffy coat were removed.  Erythrocytes were washed three times with phosphate buffered saline (Amresco; Solon, OH) and stored at -80°C. Baseline and endline blood samples were batch shipped on dry ice to the University of British Columbia (UBC) in Vancouver, Canada in June 2015.   70 4.3.6 Biochemical thiamin analysis eTDP was measured at UBC using reverse-phase high performance liquid chromatography with a fluorescence detector (HPLC-FLD) according to Lu & Frank (51), with modifications. Briefly, 500 µL of 10% wt/vol trichloroacetic acid in deionized water was added to a mixture of 250 µL previously frozen packed erythrocytes and 250 µL deionized water. Samples were vigorously vortex mixed, placed on ice for 15 min, and then centifuged (13000 g, 10 min). An aliquot of supernatant (500 µL) was washed twice with 750 µL of water saturated methyl-tert-butyl ether. From the aqueous layer, a 150 µL aliquot was then transferred to a 96 well plate. HPLC analysis was performed using an Agilent 1260 Infinity system with a Poroshell 120 EC-C18 column (3.0 x 50 mm with 2.7 µm; Agilent Technologies, Mississauga, ON) at 25°C with the detector set at an excitation wavelength 375 nm and emission wavelength 435 nm. Mobile phase A consisted of 25 nM sodium phosphate (pH 7.0) and methanol (90:10 vol/vol), while mobile phase B consisted of 25 nM sodium phosphate (pH 7.0) and methanol (30:70 vol/vol). Prior to injection, online sample derivatization (methanol, sample, and 1.2 nM potassium ferricyanide in 15% wt/vol sodium hydroxide) was performed by automated injector programming. Quantitation of eTDP was based on peak area and external standardization using TDP calibration solutions (range ~20-800 nM; Sigma-Aldrich; recoveries of low (20 nM) and high (800 nM) standards in deionized water were 102.5% and 93.2%, respectively). Quality controls were conducted on each run using pooled erythrocytes from ten healthy Canadian adults (inter-run CVs <9%, n=17).   4.3.7 Data analysis We estimated a sample size of n=92 mothers per group would be sufficient to observe a 10% difference between fortified and control groups at endline assuming a minimum baseline of 38  71 nM (93), standard deviation=18 nM (122), 80% power, alpha=0.05, with a 2-sided test, assuming a 15% attrition rate. Participant characteristics and attitudes were summarized as mean ± SD for continuous variables, and as n (%) for categorical variables. In intent-to-treat analysis (ITT), baseline eTDP values were carried forward when endline data were missing. In as-treated analysis, only participants with complete data were included in analysis. All analyses refer to ITT unless otherwise stated.  General linear models (GLM) were used to assess differences between fortified groups (LC and HC) compared to the control group (reference). GLM were used to assess endline eTDP (adjusted for baseline), and fish sauce consumption. Post-hoc analysis with least significant difference (LSD) adjustment for multiple comparisons was used to assess differences between the three treatment groups: LC, HC, and control. GLM (run separately for LC and HC groups) were used to assess differences in both endline eTDP and eTDP change between baseline and endline, by baseline eTDP tertile (lowest tertile as reference group; post-hoc analysis with LSD for comparison between three tertiles). Estimated marginal means (95% CI) are reported for GLM. All analyses were performed with SPSS for Macintosh version 23.0 (IBM Corp., Armonk, NY), with a significance level of P<0.05.   4.4 Results Participant recruitment, flow and follow-up is shown in Figure 4-1. Of the 427 women screened for participation, 276 mother-child dyads were enrolled in the study and randomly assigned to control, LC, and HC (n=92, n=93, and n=91, respectively). At the end of the study 197 mothers remained resulting in an overall attrition rate of 29%, with no differences by treatment group  72 (control, n=27; LC, n=26; HC, n=26). Moving away for work, usually temporarily, was the main reason women were lost to follow-up (18%, n=49: control, n=13; LC, n=19; HC, n=17).    Figure 4-1: Participant flow and follow-up of Cambodian women (18-45 y) and their youngest child (12-59 mo) in Prey Veng, Cambodia  Baseline characteristics of mothers, children, and their households can be found in   Table 4-1. The mean age of women was 30 ± 6 y, and most of the women had a normal BMI (18.51-24.99 kg/m2). Households typically had 5 ± 2 people, with a relatively high mean annual  73 household income of US$1,906 ± 1,903 with 49% (n=136) of households within the top 20% of incomes in Prey Veng province (99).  Table 4-1: Baseline characteristics of Cambodian women (18-45 y), their youngest children (12-59 mo), and their households1  Control n=92 Low Concentration n=93 High Concentration n=91 Mothers    Age, y 30 ± 6 30 ± 6 30 ± 7 Body mass index, kg/m2      Underweight (≤ 18.5)      Normal (18.51-24.99)      Overweight (25-29.99)      Obese (≥ 30) 21.8 ± 3.5 12 (13) 63 (69) 15 (16) 2 (2) 21.8 ± 3.3 13 (14) 67 (72) 11 (12) 2 (2) 22.3 ± 3.1 8 (9) 67 (74) 14 (15) 2 (2) Education       None      Primary (years 1-6)      Lower Secondary (years 7-9)      Upper Secondary (years 10-12)      Higher Education  8 (9) 55 (60) 19 (21) 10 (10)  -   7 (8) 47 (50) 29 (31) 9 (10) 1 (1)  12 (13) 35 (38) 36 (40) 7 (8) 1 (1) eTDP, nM 150 ± 39 172 ± 53 179 ± 59 Children    Sex, female2 Age, mo 29 (26) 35 ± 14 33 (40) 32 ± 14 32 (43) 31 ± 13 eTDP, nM3 172 ± 43 194 ± 55 211 ± 65 Households    Household population 5 ± 2 5 ± 2 5 ± 2 Annual household income, US$4      Bottom 20%      Middle 60%      Top 20% 1,678 ± 1,614 12 (13) 39 (42) 41 (45) 2,119 ± 2,013 9 (10) 31 (33) 53 (57) 1,917 ± 2,047 14 (15) 35 (39) 42 (46) 1 mean ± SD or n (%). eTDP, erythrocyte thiamin diphosphate concentration. 2 Data collected at endline; control, n=81; LC, n=82; HC, n=75. 3 Control, n=87; LC, n=88; HC, n=87 4 Households were classified using 2011 income quintiles from Prey Veng province, Cambodia (99)   Maternal and child endline eTDP can be found in Table 4-2. In intent-to-treat analysis, after 6 months women in LC and HC groups had significantly higher baseline-adjusted eTDP (mean,  74 95% CI) of 232 nM (220, 244) and 231 nM (219, 244), respectively, than control, 175 nM (163, 188; P<0.001). There was no significant difference in baseline-adjusted eTDP between women receiving the two fortified fish sauces, LC and HC (P=0.97). Similarly, at endline children’s eTDP was significantly higher in the LC (243; 233, 254 nM) and HC (229; 219, 240 nM) groups compared to control (212; 202, 223 nM; P<0.05), and there was no difference between HC and LC groups (P=0.06). As-treated analysis did not markedly change the findings (see Table 4-2).   Endline eTDP for mothers and children stratified by baseline eTDP tertile is shown in Figure 4-2. LC women in the highest tertile at baseline had significantly higher endline eTDP (P<0.001) compared to the lower tertiles; those in the lower two tertiles at baseline did not have different mean endline eTDP (P=0.07; Panel A). Among HC women, those in the highest tertile at baseline had significantly higher endline eTDP than those who started in the lowest tertile (P<0.001; Panel B). Among women in the LC group, the rate of change between baseline and endline eTDP did not differ by baseline tertile (P=0.50). However, among women in the HC group, the mean (95%) change in eTDP between baseline and endline was significantly higher among women in the lowest tertile, 88 nM (61, 115) compared to women in the highest baseline tertile, 34 nM (7, 61; P=0.006). Mean endline eTDP among both LC and HC children differed significantly among all three tertiles (P<0.001; Panels C and D). Neither HC nor LC group children had different rates of change in eTDP between baseline and endline by baseline tertile (LC, P=0.52; HC, P=0.19).     Results of the weighed fish sauce record are given in Table 4-3. Overall, consumption did not differ by treatment arm (mothers, P=0.80; husbands, P=0.24; children, P=0.54). Fish sauce  75 consumption was highly variable, with ranges in consumption between 0 – 81 mL per day among mothers. The overall number of bottles of fish sauce consumed over the six month study was not significantly different between treatment groups: control and LC, 13 (12, 14) bottles, and HC, 12 (11, 13) bottles (P=0.19). Most participants (n=142, 72%) self-reported increased household fish sauce consumption during the study period, with the majority indicating a slight (n=51, 36%) or a large (n=84, 59%) increase. Only 8 women (6%) attributed this increase in fish sauce consumption to the fact that fish sauce was free: the majority of women (n=114, 86%) credited the “delicious taste” of the study fish sauce. Ninety-six percent of women indicated that they would purchase thiamin-fortified fish sauce if it became available at market, mostly due to its organoleptic properties (n=140, 74%), but also because they believed it could improve the health of their family (n=45, 24%).    76 Table 4-2: Maternal and child eTDP at endline (6 mo)1   β SE P value  n Estimated Marginal Means (95% CI)2 Mothers  Intent-to-Treat3      Control      Low concentration      High concentration   (Ref) 57 56  - 9 9  - <0.001 <0.001   92 93 91  175 (163, 188)a 232 (220, 244)b 231 (219, 244)b As Treated4      Control      Low concentration      High concentration   (Ref) 78 75  - 11 11  - <0.001 <0.001   65 67 65  183 (167, 198)a 260 (246, 275)b 258 (243, 273)b Children Intent-to-Treat3             Control      Low concentration      High concentration  (Ref) 31 17 - 8 8 - 0.031 <0.001  87 88 87 212 (202, 223)a 243 (233, 254)b 229 (219, 240)b* As Treated4      Control      Low concentration      High concentration    (Ref) 46 42  - 9 10  -  <0.001 <0.001   70 66 55  215 (203, 227)a 261 (249, 273)b 257 (243, 271)b 1 A general linear model adjusted for baseline eTDP was used to assess differences between fortified groups (LC and HC) compared to control group (reference). eTDP, erythrocyte thiamin diphosphate (nM) 2 Post-hoc analysis (with LSD adjustment for multiple comparisons) assessed differences between LC, HC, and control groups; values that do not share a common superscript differ significantly, P<0.001; *P<0.05.  3 In intent-to-treat analysis, baseline values were carried forward for missing endline values. 4 In as-treated analysis, a total of n=79 women and n=71 children were excluded from analyses due to attrition, blood draw refusal, and inability to draw blood (on behalf of phlebotomist).      77  Figure 4-2: Mean (95% CI) erythrocyte thiamin diphosphate concentrations (eTDP, nM) of women and their children aged 12-59 mo at baseline and endline (t=6 mo) stratified by baseline eTDP tertile (lowest tertile: __  __  __; middle tertile: _ _ _ ; highest tertile: ___ ). Panels: A, women low concentration group; B, women high concentration group; C, children low concentration group; D, children high concentration group.A BC DErythrocyte thiamin diphosphate (eTDP; nmol/L)Erythrocyte thiamin diphosphate (eTDP; nmol/L)Baseline Endline (t=6 mo) Baseline Endline (t=6 mo)Baseline  Endline (t=6 mo) Baseline Endline (t=6 mo) 78 Table 4-3: Daily fish sauce consumption (mL/d) and thiamin intake (mg/d) from fish sauce of women, their husbands, and their children (12-59 mo) collected from the three day weighed fish sauce records, by meal preparation method  Woman Husband Child  n Mean ± SD Range n Mean ± SD Range n Mean ± SD Range Control               Daily Fish Sauce Consumption 27 14 ± 17 1 - 60 18 12 ± 11  0 - 34 26 4 ± 6 0 - 24           Stir fry 13 3 ± 2 0.7 - 7 8 4 ± 3 0.7 - 9 9 1 ± 1 0.1 - 3           Condiment 12 24 ± 16 2 - 55 8 16 ± 8 7 – 27 10 8 ± 6 3 – 23           Soup 14 4 ± 4 0.4 - 11 8 4 ± 4 0.4 – 11 12 1 ± 2 0.1 – 7           Other1 3 3 ± 4 0.4 - 7 2 4 ± 6 0.2 - 8 2 2 ± 2 0.3 - 3      Daily Thiamin Intake - -  - - - - - - - Low Concentration                Daily Fish Sauce Consumption 26 18 ± 21 0 - 81 10 25 ± 28 5 - 93 27 4 ± 5 0 - 26           Stir fry 14 5 ± 5 0.4 – 22 6 5 ± 1 2 - 6 13 2 ± 2 0.3 – 7           Condiment 5 32 ± 24 11 – 69 2 14 ± 8 8 – 20 4 4 ± 2 2 - 6           Soup 13 12 ± 13 1 - 33 6 23 ± 27 3 - 73 10 4 ± 5 0.1 – 15           Other1 10 8 ± 10 0.3 - 31 4 14 ± 14 0.9 - 32 6 3 ± 4 0.2 - 11      Daily Thiamin Intake 26 36 ± 42 0 - 162 10 49 ± 55 10 - 185 27 8 ± 11 0 - 52 High Concentration                Daily Fish Sauce Consumption 25 16 ± 19 0 - 78 14 18 ± 22 0 - 68 26 6 ± 8 0 - 33           Stir fry 13 8 ± 10 1 – 41 7 8 ± 6 3 - 17 12 5 ± 4 0.3 - 33           Condiment 2 12 ± 0.7 11 – 12 1 2  -  1 2  -            Soup 12 14 ± 13 1 – 47 4 33 ± 16 21 – 54 12 5 ± 4  0.5 – 14           Other1 6 16 ± 11 4 - 31 3 18 ± 14 4 - 32 6 5 ± 8 0.3 - 22      Daily Thiamin Intake 25 127 ± 153 0 - 626 14 147 ± 178 0 - 540 26 47 ± 67 0 - 262 1 Includes dishes such as Teuk Trey Ph’aem or Teuk Kroeung  79 4.5 Discussion An estimated 90% of the Cambodian population consumes fish sauce (85), making it an ideal vehicle for micronutrient fortification. Indeed, fish sauce fortified with iron has been tested and is currently being sold in Cambodia (84–86) and throughout the region (102,103). We found that thiamin-fortified fish sauce is an efficacious means of increasing dietary thiamin intake and biochemical thiamin status among women of childbearing age and their young children in rural Cambodia. Women consuming a thiamin-fortified fish sauce had higher baseline-adjusted eTDP at endline compared to women consuming the control fish sauce (P<0.001). Since infantile beriberi, a potentially fatal disease in breastfed infants (13), remains an issue in Cambodia (11), increasing the thiamin intake and status of women of reproductive age could prevent these infant deaths (57). While this target group is a priority, thiamin fortification has the potential to benefit all consumers by preventing marginal thiamin deficiency, which causes apathy, fatigue, loss of appetite, and dizziness (45), and may be common in other at-risk groups such as the elderly (22).    Neither maternal nor child baseline-adjusted endline eTDP differed significantly between LC and HC, indicating a plateau: among this sample, consumption of fish sauce containing more than 2 g/L thiamin did not confer higher eTDP. Indeed, dietary consumption data from the weighed fish sauce records highlight that women in the high concentration group could have received up to 626 mg thiamin in one day from fish sauce alone (Table 4-3). Healthy adults have an estimated total body thiamin content of 25-30 mg (13). At high doses only a small portion of thiamin is absorbed (23), and of that, little  80 is retained, with high concentrations of thiamin appearing in urine shortly after administration (21). For instance, a single oral dose of thiamin higher than approximately 2.5 to 5 mg has shown to go largely unabsorbed in healthy adults (20). However, rapid metabolism and turnover of this vitamin, with a biological half-life of 9-18 days (24), necessitates routine dietary intake (13). However, given the average fish sauce intake of 16 mL/d among women, a fortification level of 0.09 g/L would have provided 1.4 mg/d thiamin; at concentrations of 2 and 8 g/L, 32 and 128 mg/d, respectively, were consumed. While there is no UL for thiamin and daily over-the-counter supplements can contain 50 mg thiamin (18) or more, the lowest effective concentration of thiamin is likely well under 2 g/L and should be determined to limit excess intake and keep costs low for consumers (87).   Among women in the HC group, the change in eTDP between baseline and endline was influenced by women’s initial thiamin status. HC group women in the lowest baseline eTDP tertile had a significantly larger improvement in eTDP (mean change between baseline and endline of 88 nM) compared to women in the highest eTDP tertile at baseline (mean change of 34 nM). Although this trend was not found among women in the LC group, these results suggest that with the same intervention those women starting off with lower thiamin status have higher potential to benefit and will likely show greater improvement than those who are adequate to begin with.  Most Cambodian producers sell three types of fish sauce: high quality, with high protein content, dark colour, and rich fish flavours; medium quality, which is a slightly watered  81 down version of high quality; and low quality, which is more dilute, lighter in colour, and contains artificial colours and flavours. We fortified medium quality fish sauce (market value ~US$0.85 per 750 mL bottle). Before the study, approximately half of women made fish sauce at home (n=148, 54%), and based on self-reported costs, those who did purchase at market likely selected low quality fish sauce (mean ± SD, 2,056 ± 568 riel, or ~US$0.50 ± 0.15 per 750 mL bottle). Therefore, additional research is required to determine the effectiveness of thiamin-fortified fish sauce, investigating both a wider quality range of fish sauces, and requiring participants to purchase sauce.   Adherence to study fish sauce was high, and fortnightly check-ins showed limited fish sauce sharing outside the household: only 7 women reported sharing fish sauce, and each only once over the six month study. Fish sauce was distributed for free for ad libitum consumption, so intake was likely higher than usual, potentially replacing other condiments such as soy sauce or salt.   This study had several strengths: we recruited women from randomly selected villages not involved in other nutrition interventions, had frequent fortnightly follow-ups with participants, measured detailed fish sauce consumption in a subset of households with the weighed fish sauce record, and had a control group, which was very important as we saw an increase in eTDP across all three treatment groups over the 6 month study. The latter may be attributed to seasonal economic status and/or food availability: baseline data collection took place in October, but wet-season rice, the major economic activity and a main driver of food security for rural Cambodians, is harvested in December/January (3).  82 With this, it would be of interest to follow-up over one year to identify the true seasonal differences in thiamin intake and biochemical status.     Limitations of this study include restriction to one rural province in Cambodia, a higher than expected attrition rate, and a lack of dietary intake data beyond fish sauce consumption. Future research should aim to determine usual fish sauce intake throughout the year outside of a controlled study environment to obtain better consumption data from which to base the thiamin fortification level, and rather than distribute fish sauce for free, should make it available at market for women to purchase. In addition, research to determine the usual intake of thiamin would be beneficial in optimizing the fortification dose; while women and children consuming fish sauce fortified at 2 g/L had significantly higher eTDP than those in the control group, the optimal fortification dose may be lower than 2 g/L. More research is required to optimize the dose to avoid unnecessary excess intake and lower production costs.      Another major limitation of this research is a lack of interpretive criteria for sufficient thiamin status using eTDP. Appropriate deficiency cut-offs are unclear as they vary widely: >70 nM (54), >140 nM (55), and >148 nM (56). Therefore, cut-offs were not employed in this study, making it more difficult to quantify the fortification level required for adequate maternal status to prevent infantile beriberi. Curiously, women’s baseline eTDP was higher than previously reported: 167 ± 52 nM (n=276), compared to 149 ± 36 (n=121 women aged 20-45 in Prey Veng; see Table 2-1). This difference may be attributable to seasonal variation (baseline data was collected in April in this study,  83 while the previous data was collected in January), or differences in socioeconomic status of the women enrolled (women in this study were wealthier and better educated, see Table 2-1 and Table 4-1), but this cannot be confirmed.    Fish sauce is an efficacious means of improving thiamin intake, and therefore has potential to be a simple and sustainable vehicle for other micronutrients as well. Given that women may make their own fish sauce, presumably especially those who are very poor and can use fish caught freely from rice paddies in the rainy season, salt fortification (83), a major ingredient in fish sauce that is already fortified with iodine (83), should also be explored as a thiamin fortification vehicle in tandem with fish sauce to reach those at highest risk of inadequate thiamin intake.      84 Chapter 5: Perinatal consumption of thiamin-fortified fish sauce in rural Cambodia: a randomized controlled efficacy trial 5.1 Summary Background: Infantile beriberi, a fatal disease caused by thiamin deficiency, remains a public health concern in Cambodia and regions where B-vitamin poor, polished white rice is a staple food. Low maternal thiamin intake limits breast milk thiamin content, putting breastfed infants at risk of beriberi. We fortified fish sauce, a popular Cambodian condiment, to increase maternal dietary thiamin intake. Objective: To determine whether maternal consumption of thiamin-fortified fish sauce could improve the erythrocyte thiamin diphosphate concentration (eTDP) of mothers and their breastfed infants, and breast milk thiamin concentrations, compared to mothers consuming a control fish sauce.  Methods: In this community-based, double-blind randomized controlled efficacy trial, 90 pregnant women (18-45 y; 23 ± 7 wk gestation at enrolment) were recruited in Prey Veng province, Cambodia. Women were randomized to one of three treatment groups (n=30 per group) for ad libitum consumption of fish sauce for 6 months: control, low (LC, 2 g/L) or high concentration (HC, 8 g/L) thiamin-fortified fish sauce. Maternal blood was collected at baseline. Maternal blood, breast milk, and infant blood were collected at endline (t=6 mo). eTDP and breast milk thiamin were measured using HPLC-FLD.  Results: Baseline-adjusted endline eTDP (estimated marginal mean, 95% CI) were significantly higher among mothers in the LC (276; 246, 306 nM) and HC (238; 207, 268 nM) groups compared to control (194; 163, 224 nM; P<0.05); LC and HC did not differ (P=0.08). Total breast milk thiamin concentrations were significantly higher among  85 women in both LC (211; 187, 236 μg/L) and HC (180; 152, 209 μg/L) groups compared to control (136; 110, 162 μg/L; P<0.05). Infants of HC mothers had significantly higher eTDP (257; 215, 298 nM; P<0.05) compared to LC (205; 175, 235 nM) and control (181; 153, 210 nM).  Conclusions: Perinatal consumption of thiamin-fortified fish sauce for 6 months improved eTDP and breast milk thiamin concentrations among pregnant and lactating women in rural Cambodia, and in turn improved eTDP of their breastfed infants. Regular consumption of thiamin-fortified fish sauce has the potential to improve maternal thiamin status, and in turn prevent infantile beriberi in this population.   5.2 Introduction Infantile beriberi is caused by thiamin (vitamin B1) deficiency (13) and typically presents among breastfed infants (57) around 3 months of age (68) with a persistent hoarse cry, vomiting, anorexia, generalized edema, oliguria (19), convulsions, and signs of heart failure (liver enlargement, rapid breathing, and rapid heart rate) (11). Without rapid thiamin administration infants can die within hours (68). Mothers with poor dietary thiamin intake produce breast milk low in thiamin, putting their infants at risk of developing thiamin deficiency and infantile beriberi (57). During infancy, a period of rapid growth and development, thiamin needs are high relative to body size placing infants at risk of beriberi (13,67,78) while mothers remain asymptomatic (13). Umbilical cord blood of thiamin-replete mothers has up to three times higher thiamin concentrations than maternal blood at birth (28) due to preferential thiamin sequestration to the fetus  86 during the third trimester (27). Thus, low maternal thiamin intake throughout pregnancy can impose further increased risk of developing infantile beriberi (57).  Infantile beriberi-related mortality was common among breastfed infants of Karen refugees living in camps on the Thai-Burmese border in the 1980s (29,68,70). While addressed there through supplementation (68), poor thiamin status likely remains a public health concern in Cambodia due to a lack of dietary diversity (1). Thiamin is found in high concentrations in foods that are either not typically consumed (legumes and whole grains, for example, brown rice), or which are costly (pork). Finally, food balance sheet analysis estimate that approximately 60% of daily calories (8) come from non-parboiled (44), polished white rice, a very poor source of thiamin (13). We recently reported significantly lower erythrocyte thiamin diphosphate concentrations (eTDP), an indicator of thiamin status, among a representative sample of non-pregnant, non-lactating Cambodian women of childbearing age living in rural Prey Veng province compared to healthy, purportedly vitamin-replete Canadian peers (93,94). With this, pregnant and lactating women would likely also have lower eTDP than the Canadian women sampled since thiamin requirements increase during these life stages (18,31).   Supplementation of thiamin deficient lactating women has been shown to improve breast milk thiamin concentrations in the Karen refugee camps (68,72), rural Cambodia (73), the Gambia (74), and India (75); however supplementation is a targeted, resource-intensive intervention that relies heavily on individual compliance. With a combination of diverse diets and thiamin fortification/enrichment of white wheat flour (36,37,64),  87 infantile beriberi is virtually non-existent in the West. An estimated 90% of Cambodians consume fish sauce (85), and this condiment is already a fortification vehicle for iron (84–86). As such, thiamin fortification of fish sauce could be a sustainable, low-cost, and passive intervention (87–89) to improve the dietary thiamin intake of pregnant and lactating women and, in turn, their breastfed infants.   The aim of this study was to determine whether consumption of thiamin-fortified fish sauce over six months during pregnancy and early lactation yielded higher eTDP and breast milk thiamin concentrations among rural Cambodian women compared to those consuming a control fish sauce containing no thiamin. A second aim was to determine whether eTDP differed between infants of mothers consuming thiamin-fortified fish sauce compared to infants of mothers consuming control fish sauce (no detectable thiamin).     5.3 Methods 5.3.1 Study design This was a community-based, double-blind, randomized controlled efficacy trial conducted in two groups of women concurrently: 90 pregnant, and 270 non-pregnant women. We report here on the 90 pregnant women recruited through convenience sampling from Prey Veng province, Cambodia. Eligibility criteria were as follows: 18-45 y and 3-8 mo pregnant with a singleton fetus (self-report); the female head of their household; planned to exclusively breastfeed their infant for 6 mo; no prior history of preeclampsia, pre-term delivery, or birth defects; not involved in other non-governmental  88 nutrition programs; not consuming thiamin-containing dietary supplements (not standard of care in Cambodia (81)); agreement to exclusively consume the study fish sauce in their household; and no plans to leave their village (e.g. for seasonal work) for the duration of the study (6 mo). See Figure 5-1 for participant recruitment, flow, and follow-up. Women provided written informed consent to participate. Ethical approval was obtained from the Cambodian National Ethics Committee for Health Research (0245NECHR) and the University of British Columbia – Children’s and Women’s Health Centre of British Columbia Research Ethics Board in Canada (CQ14-0204/H14-01654). Clinicaltrials.gov Identifier: NCT02221063.  5.3.2 Intervention: fortified fish sauce Study fish sauce was produced by Leang Leng Enterprises (Phnom Penh, Cambodia) and received Cambodian Ministry of Health Certification for Hygienic Food Production. Fish sauce was pasteurized for 90-120 min at 100°C, cooled to room temperature, then fortified as follows: low concentration thiamin (LC, 2 g/L thiamin hydrochloride (THCL; ≥ 98% purity), Huazhong Pharmaceutical Co., Ltd, China); high concentration thiamin (HC, 8 g/L THCl); and control, which contained no detectable thiamin. Thiamin fortification dose in fish sauce is discussed in 3.3.1. All fish sauces were also fortified with iron as ferric sodium ethylenediaminetetraacetate (2.8 g/L Ferrazone®, Akzo Nobel Functional Chemicals B.V., The Netherlands) as per Cambodian Ministry of Planning guidelines (86).   89   Figure 5-1: Participant flow and follow-up for pregnant and lactating Cambodian women (18-45 y) and their newborn breastfed infants. 90 Fish sauce packaging was identical among the three groups except for a unique serial number stamp on each bottle, and different coloured caps for different study arms intended to prevent confusion amongst field staff and participants. Green, orange, and purple were selected for bottle caps after consultation with local women highlighted these as pleasant yet culturally and politically neutral colours (n=60 women aged 18-45 y from focus group discussions in 12 villages within 3 communes in Prey Veng, Cambodia; see semi-structured guide in Appendix C). The blinding code was known to Leang Leng Enterprises, and was stored in a sealed envelope held by the Principal Investigator in case of emergency. Sensory evaluation using the triangle test (110) indicated that Cambodian women (n=90) were unable to discern the three study fish sauces.   5.3.3 Randomization Using a pre-determined master list of individual randomized ID numbers, field staff assigned IDs sequentially at enrollment. Women received fish sauce after their baseline study visit, and fortnightly thereafter during household visits by field staff. Since Cambodian families eat from a common pot, participants and their entire household were instructed to consume the study fish sauce ad libitum, as they normally would, throughout the six-month study. At 1 mo (November 2014), study participants attended a nutrition education workshop in their village to learn the signs, treatment, and prevention of infantile thiamin deficiency and beriberi, as well as the Cambodian Ministry of Health infant and young child feeding curriculum (81). Participant’s family members (husbands, sisters, mothers, mothers-in-law etc.) were invited to attend.     91 5.3.4 Data and biological sample collection At baseline (October 6-17, 2014) and endline (t=6 mo, April 22-29, 2015) demographic information was collected using an interviewer-administered questionnaire in women’s homes (see questionnaire in Appendix F). Participants gave birth during the study; field staff visited each participant within 72 hours of birth to measure the infant’s length and weight using a length board and calibrated weight scale, respectively (123).  The morning following data collection, non-fasting venous blood samples were collected into evacuated EDTA-coated tubes (Vacutainer, Becton Dickinson, Mississauga, ON) from women and their infants (endline only) at a central village location. Blood samples were then transported on ice to the National Institute for Public Health (NIPH) laboratories in Phnom Penh within 5 h of collection. Blood samples were centrifuged at 3000 rpm for 15 min at 4°C, plasma and buffy coat were removed, and erythrocytes washed three times with phosphate buffered saline (Amresco, Solon, OH) before storage at -80°C.   Breast milk samples were collected using a battery-powered single breast pump (Swing Breast pump, Medela, Mississauga, ON). One full breast expression was collected from the breast that women self-identified as being more ‘full’ (the breast not most recently fed from). Two women chose to manually hand-express milk rather than use the breast pump. Breast milk was transported along with the blood samples to NIPH daily on ice. Total milk volume was recorded, then milk was mixed well, aliquoted into amber cryovials, and stored at -80°C. All blood and breast milk samples were batch shipped on dry ice to the University of British Columbia (UBC) in Vancouver, Canada in June 2015.  92  5.3.5 Erythrocyte thiamin diphosphate analysis eTDP was measured at UBC using reverse-phase high performance liquid chromatography with a fluorescence detector (HPLC-FLD) according to Lu & Frank (51), with modifications. Briefly, 500 µL of 10% wt/vol trichloroacetic acid in deionized water was added to a mixture of 250 µL previously frozen packed erythrocytes and 250 µL deionized water. Samples were vigorously vortex mixed, placed on ice for 15 min, and then centifuged (13000 g, 10 min). An aliquot of supernatant (500 µL) was washed twice with 750 µL of water saturated methyl-tert-butyl ether. From the aqueous layer, a 150 µL aliquot was then transferred to a 96 well plate. HPLC analysis was performed using an Agilent 1260 Infinity system with a Poroshell 120 EC-C18 column (3.0 x 50 mm with 2.7 µm; Agilent Technologies, Mississauga, ON) at 25°C with the detector set at an excitation wavelength 375 nm and emission wavelength 435 nm. Mobile phase A consisted of 25 mM sodium phosphate (pH 7.0) and methanol (90:10 vol/vol), while mobile phase B consisted of 25 mM sodium phosphate (pH 7.0) and methanol (30:70 vol/vol). Prior to injection, online sample derivatization (methanol, sample, and 1.2 mM potassium ferricyanide in 15% wt/vol sodium hydroxide) was performed by automated injector programming. Quantitation of eTDP was based on peak area and external standardization using TDP calibration solutions (range ~20-800 nM; Sigma-Aldrich, Oakville, ON; recoveries of low (20 nM) and high (800 nM) standards in deionized water were 102.5% and 93.2%, respectively). Quality controls were conducted on each run using pooled erythrocytes from ten healthy Canadian adults (inter-run CVs <9%, n=17).    93 5.3.6 Breast milk thiamin analysis Free thiamin, thiamin monophosphate (TMP), and thiamin diphosphate (TDP) in breast milk were analyzed at the USDA/ARS Western Human Nutrition Research Centre, University of California, Davis using HPLC-FLD and pre-column derivatization of the analytes to their thiochrome esters. 12 μL of perchloric acid (HClO4, 70%, Sigma Aldrich, St. Louis, MO) was added to 100 μL breast milk for protein precipitation. After vortexing for 1 min the samples were centrifuged at 4°C for 10 min at 14000 rpm, then 75 μL of the supernatant was transferred into a fresh 1.5 mL centrifuge tube. The thiamin vitamers were derivatized by adding 27 μL of an aqueous solution of 12 mM potassium ferricyanide (K3[Fe(CN)6]) in 3.35 N sodium hydroxide (NaOH). After briefly mixing, the reaction was quenched with 24 μL 1M phosphoric acid (H3PO4) to neutral pH. After filtration, 30 μL of the sample was analyzed via an Agilent 1200 series HPLC-FLD using a Phenomenex Kinetex C18 column (100 Å, 150 x 4.6 mm, 5 μm) protected by a Phenomenex SecurityGuard C18 pre-column (4 x 30 mm; Torrance, CA). A solvent gradient of 0.15 M dipotassium hydrogen phosphate (K2HPO4; mobile phase A) and methanol (mobile phase B) was employed as follows: 0 min: 85% A, 3 min: 80% A, 6 min: 50% A, 6.5-8min: 85% A at a flow rate of 1.4 mL/min. Excitation and emission wavelengths for detection were set to 367 and 435 nm, respectively. A pooled breast milk sample with previously established thiamin concentrations was used as an internal control. This control was measured four times; CVs for thiamin, TMP, and TDP were 2%, 2%,  and 5%, respectively.    5.3.7 Statistical analysis We estimated that a sample size of n=30 mothers per group would be sufficient to detect a 30% difference in maternal eTDP between fortified and control groups at endline assuming a  94 minimum baseline of 38 nM (93), SD of 18 nM (122), 80% power, and alpha=0.05 using a 2-sided test. Demographic characteristics were summarized as mean ± SD or n (%). Maternal eTDP analyses were conducted using both intent-to-treat (ITT) and as-treated analyses. In ITT, baseline eTDP values were carried forward when endline data were missing (all analyses refer to ITT unless otherwise stated); in as-treated, only participants with complete data were included in analysis.   General linear models (GLM) were used to assess differences between fortified groups (LC and HC) compared to the control group (reference) for the following: maternal endline eTDP (adjusted for baseline eTDP), infant eTDP, breast milk thiamin concentrations, infant age and lactation duration, and fish sauce consumption. Post-hoc analysis with least significant difference (LSD) adjustment for multiple comparisons was used to assess differences between the three treatment groups (LC, HC, and control). Estimated marginal means (95% CI) are reported for GLM. All analyses were performed with SPSS for Macintosh version 23.0 (IBM Corp., Armonk, NY), with a significance level of P<0.05.   5.4 Results Participant flow and follow-up can be found in Figure 5-1. Ninety-three pregnant women in Prey Veng, Cambodia were screened; 90 met eligibility criteria, agreed to participate, and were randomized to the three study arms, n=30 per group. Five participants were excluded from analyses: 3 due to stillbirth or miscarriage, and 2 who had not yet delivered by endline data collection. Retention in the three groups was n=27, 28, and 23 for control, LC, and HC, respectively.   95  Baseline demographic characteristics of participants can be found in Table 5-1. Mean ± SD age of women at enrollment was 26 ± 5 y, the majority of women were in their second trimester (mean gestation, 23 ± 7 wk), and approximately half of women were pregnant with their first child. Most women attended 5 antenatal care visits throughout pregnancy, and delivered at their local health centre (Table 5-2). Mean ± SD birth length and weight was 49 ± 2 cm and 3.1 ± 0.6 kg, respectively, and only 5 infants had low birth weight (<2.5 kg). Household fish sauce consumption was similar across all three treatment groups: households in LC, HC, and control groups consumed 15 (13, 16), 13 (11, 14), and 16 (14, 17) 750 mL bottles, respectively, over the 6 mo study (P=0.07). 96 Table 5-1: Baseline demographic characteristics and eTDP of women (18-45 y) in the pregnant and lactating cohort1  Control n=28 Low Concentration n=29 High Concentration n=28 Age, y 27 ± 5 26 ± 5 25 ± 5 # weeks pregnant, wk 23 ± 8 22 ± 7 25 ± 8 Parity      Primipara      Multipara  13 (46) 15 (54)  16 (55) 13 (45)  14 (50) 14 (50) Number in household 5 ± 1 5 ± 2 5 ± 1 Education      None      Primary      Lower Secondary      Upper Secondary      Higher Education  2 (7) 11 (39) 12 (43) 2 (7) 1 (4)  3 (10) 15 (52) 8 (28) 3 (10) -  1 (~4) 12 (43) 14 (50) 1 (~4) - Annual household income, US$2      Bottom 20%      Middle 60%      Top 20% 1,625 ± 1,156 2 (7) 11 (39) 15 (54) 1,440 ± 1,233 3 (10) 15 (52) 11 (38) 1,388 ± 1,301 4 (14) 11 (39) 13 (47) Erythrocyte thiamin diphosphate, nM 152 ± 44 175 ± 62 179 ± 63 1 mean ± SD or n (%). eTDP, erythrocyte thiamin diphosphate concentration (eTDP; nM) 2 households were classified using 2011 income quintiles from Prey Veng province, Cambodia (99)  97 Table 5-2: Antenatal care and delivery outcomes of rural Cambodian women (18-45 y) in the pregnant and lactating cohort, and characteristics of their newborn infants1  Control Low Concentration High Concentration  n=27 n=28 n=23 Antenatal    Antenatal care visits 5 ± 2 5 ± 2 5 ± 3 Iron folic acid tablets consumed 85 ± 19 78 ± 25 85 ± 23 Deworming treatment 23 (85) 20 (71) 20 (87) Delivery/Postpartum    Delivery location      Local Health Centre      District Hospital      Provincial Town Hospital      Private Clinic or Hospital      Other: garment factory   13 (48) 1 (4) 3 (11) 10 (37) -   17 (61) 1 (~4) 1 (~4) 8 (29) 1 (~4)  9 (39)  6 (26) 2 (9) 6 (26) -  Infant characteristics     Female Birth weight, kg2 Low birth weight (<2.5kg)2 Birth length, cm2 14 (52) 3.1 ± 0.5 3 (11) 49 ± 2 13 (46) 3.0 ± 0.4 1 (4) 49 ± 2 9 (39) 3.1 ± 0.9 1 (4) 49 ± 3 Age at endline, wk 16 ± 8 17 ± 7 14 ± 8 1 mean ± SD or n (%) 2 LC, n=27  Maternal endline eTDP can be found in Table 5-3. Baseline-adjusted endline eTDP (estimated marginal mean; 95% CI) was significantly higher among women in LC (276; 246, 306 nM) and HC (238; 207, 268 nM) groups compared to women in the control group (194; 163, 224 nM; P<0.05); women in LC and HC groups did not differ (P=0.08). As-treated analysis did not change the findings markedly (see Table 5-3). Infant eTDP at endline is also found in Table 5-3. Infants of mothers in the HC group had significantly higher eTDP (257; 215, 298 nM; P<0.05) compared to mothers in the LC (205; 175, 235 nM) and control groups (181; 153, 210 nM).  98 Table 5-3: Endline (6 mo) eTDP of mothers and their breastfed infants in the pregnant and lactating cohort1   β SE P value  n Estimated Marginal Means (95% CI) Mothers  Intent-to-Treat2      Control      Low concentration      High concentration   (Ref) 82 44  - 22 22  - <0.001 0.048   28 29 28  194 (163, 224)a 276 (246, 306)b 238 (207, 268)b As Treated3      Control      Low concentration      High concentration   (Ref) 87 61  - 22 23  - <0.001 0.011   27 28 23  195 (163, 227)a 282 (252, 312)b 256 (222, 290)b Infants      Control      Low concentration      High concentration  (Ref) 24 75 - 1 3 - 0.263 0.002  25 24 17 181 (153, 210)a 205 (175, 235)a 257 (215, 298)b 1 General linear models were used to assess differences in eTDP between fortified groups (LC and HC) and control group (reference). Maternal model was adjusted for baseline eTDP. Post-hoc analysis (with LSD adjustment for multiple comparisons) assessed differences between LC, HC, and control groups; values that do not share a common superscript differ significantly, P<0.05. eTDP, erythrocyte thiamin diphosphate concentration (eTDP, nM)  2 In intent-to-treat analysis, baseline values were carried forward for missing endline values. 3 In as-treated analysis, n=7 women were excluded from analyses due to attrition and blood draw refusal. 99 Concentrations of the three thiamin vitamers found in human milk, thiamin, TMP, and TDP, as well as total thiamin (calculated as the amount of thiamin from TMP, TDP, and thiamin), can be found in Table 5-4. TMP was the most abundant vitamer in these breast milk samples. Breast milk TMP was significantly higher among women in the LC group (166; 146, 187 μg/L) compared to the control group (133; 111, 154 μg/L; P=0.03). Total breast milk thiamin was significantly higher among women in both LC (211; 187, 236 μg/L) and HC groups (180; 152, 209 μg/L) compared to the control group (136; 110, 162 μg/L; P<0.05); LC and HC did not differ (P=0.10). Thiamin concentration followed the same pattern: women in LC and HC groups produced breast milk with significantly higher thiamin than those in the control group (P<0.05). TDP, which made up <5% of total thiamin in breast milk, did not differ between groups (P=0.28).     100 Table 5-4: Thiamin concentrations of mature breast milk, and estimated daily total thiamin intake of infants fed this milk, from rural Cambodian mothers (18-45 y) in the pregnant and lactating cohort1  Control n=23 Low Concentration n=26 High Concentration n=19 Thiamin         μg/L 18 (3, 33)a 63 (49, 77)b 54 (38, 71)b      nM 66 (10, 123)a 237 (184, 290)b 204 (142, 266)b Thiamin monophosphate         μg/L 133 (111, 154)a 166 (146, 187)b 140 (116, 163)ab      nM 385 (232, 448)a 483 (424, 542)b 406 (337, 475)ab Thiamin diphosphate         μg/L 4 (2, 6)a 5 (3, 7)a 6 (4, 9)a      nM 9 (4, 14)a 12 (7, 17)a 15 (9, 20)a Total thiamin2         μg/L 136 (110, 162)a 211 (187, 236)b 180 (152, 209)b      nM 512 (414, 609)a 797 (705, 888)b 680 (572, 787)b      Estimated Infant Intake (μg/d)3 104 (83, 124)a 165 (145, 185)b 138 (115, 161)b 1 General linear models were used to assess differences between fortified groups (LC and HC) and the control group (reference). Post-hoc analysis (with LSD adjustment for multiple comparisons) assessed differences between LC, HC, and control groups; values in rows that do not share a common superscript differ significantly, P<0.05. Results expressed as estimated marginal means (95% CI).  2 Total thiamin calculated as amount of thiamin from thiamin monophosphate, thiamin diphosphate, and free thiamin 3 Estimated infant intake (μg thiamin/d) calculated based on daily breast milk consumption of exclusively breastfed infants in developing countries: 0-2 mo, 714 mL/d; 2-5 mo, 784 mL/d; 6-8 mo, 776 mL/d (59)    5.5 Discussion Perinatal consumption of thiamin-fortified fish sauce for six months yields higher eTDP among pregnant and lactating women and their breastfed infants in rural Cambodia compared to mothers consuming a control fish sauce without thiamin. Breast milk total thiamin concentrations were significantly higher among women consuming fortified fish sauce compared to those in the control group.    101 It has been established that maternal thiamin supplementation of deficient mothers increases breast milk thiamin concentrations (57). Prentice and colleagues reported a significant increase in breast milk thiamin concentration from 160 to 220 μg/L (P<0.001) among Gambian women consuming a supplemental food containing 1.36 mg thiamin daily for one year (74). Similarly, breast milk thiamin of 5 nutritionally inadequate Indian women increased from 109 to 268 μg/L after daily micronutrient supplementation of increasing doses (0.2 – 20 mg/d) over 8 months (75).   While no infants in this study showed clinical symptoms of infantile beriberi, a lack of clear biomarker cut-offs hinders assessment of biochemical thiamin adequacy. The Institute of Medicine used observed thiamin concentrations of 210 μg/L in milk produced by well-nourished mothers (32–34) to set the adequate intake (AI) for infants aged 0-6 mo at 200 μg thiamin/day (18). Using estimated daily breast milk intakes for exclusively breastfed infants in developing countries (59) we predicted daily thiamin intake of infants in our study (see Table 5-4). Only 9 infants (LC, n=7, HC, n=2) would have consumed ≥200 μg/d. Curiously, a recent report of thiamin concentrations in mature milk (≥2 weeks) collected from women globally (including the United States) and no infants consuming this milk would have consumed the AI (18,58). However, a limitation of this source is that breast milk thiamin rather than total thiamin (thiamin + TMP + TDP) or TMP (the most abundant vitamer in breast milk), is reported. However, the findings echo a recent study in Malawi reporting that only 50% of anti-retroviral-treated mothers living with HIV who had been consuming thiamin-containing dietary supplements for 6 months produced breast milk (expressed as breast milk total thiamin) that would meet the infant thiamin AI (124).   102  The report of breast milk thiamin concentrations globally also highlighted wide variation in breast milk thiamin concentrations by region: India, 11 μg/L (n=24), Malawi, 21 μg/L (n=18), China, 31 μg/L (n=5), United States, 37 μg/L (n=28), and Cameroon, 116 μg/L (n=5) (58). These breast milk thiamin concentrations (apart from Cameroon) align relatively well with those of women in the control group (18 μg/L, n=23). Our breast milk thiamin of 63 and 54 μg/L for LC (n=26) and HC (n=19) groups, respectively, are higher than values in the United States, which may be due to supplement consumption, but unfortunately this and other dietary intake data were not reported (58).  We recognize that AIs are developed when there is not sufficient evidence available to establish an estimated average requirement and recommended dietary allowance, and are expected to meet or exceed the needs of individuals in that age group (18). As such, thiamin intakes lower than the AI of 200 μg/d could be sufficient and may not impose risk of infantile beriberi. Actually, the AI was rounded up from 160 μg/d (18), but even when using a cut-off of 160 μg/d, only 17 infants (LC, n=13; HC, n=4) would have had sufficient thiamin intake. The current infant thiamin AI was determined from data from only 24 American women in 1980 (33,34). Improvements in analytical techniques for quantification of thiamin in biological samples during the past 35 years (35) may explain some of the discrepancy between the AI cut-off and recently published reference values; as such, thiamin AIs for infants aged 0 – 6 mo should be re-visited.   The amount of thiamin in breast milk is known to increase with lactation duration (58), therefore the age of the child likely influences breast milk thiamin concentration. Infant age did not differ  103 by treatment group (P=0.38) and total thiamin in breast milk was not correlated with infant age in our study (P=0.07), but this may be attributed to the small sample size.   While eTDP was significantly higher among women consuming fortified fish sauce compared to control, there was no significant difference between women in the LC and HC groups (P=0.08). Total breast milk thiamin concentration also did not differ between LC and HC groups (P=0.10). This suggests that consumption of fish sauce fortified at, or even below, 2 g/L is sufficient for maternal thiamin adequacy. However, infant eTDP was significantly higher among HC compared to LC (P=0.03). Since LC and HC breast milk total thiamin concentrations did not differ, and infant age or lactation status was similar between groups (P=0.36), this finding may highlight the importance of maternal thiamin adequacy in late pregnancy (13,27,57) as HC infants may have built up thiamin stores in utero.   This study had several strengths, including both antenatal and postpartum consumption of fish sauce, frequent fortnightly follow-ups with participants, biological sample collection from mothers (both blood and breast milk) and their breastfed infants, and a control group, which was shrewd as we saw an increase in maternal eTDP between baseline and endline. Limitations include a lack of dietary intake data and representation from only one Cambodian province. The women in this study were relatively well-off, with 44% in the top wealth quintile for Prey Veng province (99). Women participating in other non-governmental nutrition programs were ineligible to participate in this study to avoid potential contamination, and our convenience sample was likely recruited from main roads where paid labour is more easily accessible compared to more remote households. Future research should target poor households, where  104 infantile beriberi is presumably a larger issue. While we collected a full breast expression from the breast mothers self-reported as not most recently emptied, we did not have information on the last feed, and this one expression is not necessarily representative of usual milk. Finally, there is a lack of interpretive criteria for normal or healthy eTDP for women, as cut-offs in the literature vary widely: >70 nM (54), >140 nM (55), and >148 nM (56). A recent study in Prey Veng, Cambodia reported no significant difference in blood thiamin concentrations between mothers of infants with and without beriberi (11).  Unexpectedly, baseline eTDP found in this study, 169 ± 58 nM (HC, LC, and control women, n=85) were similar to values reported among a convenience sample of Vancouver women, 179 ± 37 nM (n=47, 20-45 y; see Table 2-1). These unexpected values, alongside the range in cut-offs in the literature highlight the need for more research to develop clinically meaningful cut-offs for thiamin deficiency.   Adequate maternal thiamin intake throughout pregnancy and lactation allows for thiamin sequestration in utero (27) and production of thiamin-replete breast milk (18,57), preventing infantile thiamin deficiency and beriberi among breastfed infants. Here we showed that 6 month perinatal consumption of thiamin-fortified fish sauce, a condiment consumed by the majority of Cambodians (85), is an efficacious means of increasing maternal eTDP and breast milk thiamin, and in turn, infant eTDP. Therefore, thiamin-fortified fish sauce has the potential to prevent infantile beriberi and its related infant mortality in Cambodia and the region among women consuming commercially produced (rather than homemade) fish sauce. This intervention is facilitated by existing fortification infrastructure within factories as fish sauce has already been successfully fortified with iron in Cambodia and Vietnam (84–86,102,103). As such, fish sauce could be a simple and sustainable vehicle for other micronutrient fortification as well. Further,  105 fortification is a population-wide intervention, so there is potential for all consumers of thiamin-fortified fish sauce to improve thiamin intake,, potentially preventing beriberi outbreaks (95) and improving thiamin status of women (93,94) before conception.   5.6 Conclusions Perinatal consumption of thiamin-fortified fish sauce for six months was an efficacious means of improving eTDP and breast milk thiamin concentrations among pregnant and lactating women in rural Cambodia, and in turn, eTDP in their breastfed infants. More research is needed before this intervention can be scaled-up, but this research highlights the potential for thiamin-fortified fish sauce to improve thiamin intake and status, which in turn may provide a sustainable means of preventing infantile beriberi in this region.  106 Chapter 6: Conclusions, discussion, and future research  6.1 Introduction  Infantile beriberi remains a cause of infant mortality in Cambodia and throughout Southeast Asia despite its relatively easy treatment and prevention with thiamin. After identifying that Cambodian women of childbearing age had lower biochemical thiamin status than purportedly thiamin-replete Canadian women in Vancouver, the overall objective of my doctoral research was to identify and test the efficacy of an inexpensive, sustainable, and culturally-appropriate intervention to improve thiamin intake and status of these women with the long term goal of eradicating infantile beriberi in Cambodia. This research has contributed to the body of knowledge on thiamin deficiency, and, through close collaboration with nutrition stakeholders in-country, including the Cambodian Ministry of Health and Ministry of Planning (National Sub Committee for Food Fortification), has brought the issue of thiamin deficiency and infantile beriberi to the attention of public health nutrition stakeholders, health practitioners, and policy makers in Cambodia.    6.2 Discussion of key findings 6.2.1 Efficacy of thiamin-fortified fish sauce Using an iterative approach, after developing a stable and well-accepted thiamin-fortified fish sauce, I aimed to determine whether six-month ad libitum consumption of thiamin-fortified fish sauce could improve eTDP compared to a control sauce (containing no thiamin) among rural Cambodian women and their children. In Chapters 4 and 5 I reported that thiamin-fortified fish sauce was an efficacious means of increasing eTDP among all the populations we studied: baseline-adjusted endline eTDP was significantly higher among individuals in the HC and LC  107 groups compared to the control group for non-pregnant, non-lactating women (P<0.001), their children aged 12-59 months (P<0.05), and pregnant and lactating women (P<0.05); LC and HC groups did not differ. Similarly, total breast milk thiamin concentrations were significantly higher among women in both LC and HC groups compared to control (P<0.05). eTDP of infants of HC group mothers had significantly higher eTDP than those infants of mothers in the LC and control groups (P<0.05), which did not differ. These results demonstrate the potential for thiamin-fortified fish sauce to increase dietary thiamin intake and biochemical thiamin status of rural Cambodians, which in turn could prevent infantile beriberi.   I hypothesized a dose response to the thiamin-fortified fish sauce, postulating HC > LC > control for eTDP in all populations studied. However, I found that there was no statistical difference between eTDP of individuals in the HC and LC groups; the exception being infants. These results suggest a plateau, as consumption of fish sauce containing more than 2 g/L thiamin did not confer higher eTDP. Since a single oral dose of thiamin higher than ~ 5 mg has shown to go largely unabsorbed in healthy adults (20), this finding is not unexpected as this would require fish sauce intakes of only 2.5 mL and 0.625 mL for individuals in LC and HC groups, respectively.   One means of preventing infantile beriberi is to increase the intake of all individuals in the target population, pregnant and lactating women, so that the entire thiamin intake distribution will shift upwards, decreasing the proportion of the population that is at risk of inadequate intake (18,87). As noted in Chapter 3, the ultimate goal of many fortification programs is for 97.5% of the target population to consume more than EAR while also limiting excess intake (87); this overall shift is  108 of particular importance to that those at highest risk of inadequate intake before fortification. However, the lowest consumers are likely those of lowest socioeconomic status that may not consume as much commercially-produced fish sauce because they choose to make some fish sauce at home; therefore, the target dose must be adjusted to meet the needs of those lowest consumers. As noted in Chapters 4 and 5, more research investigating the optimal fortificant dose is required. In this study, even if individuals were thiamin deficient to begin with the majority of the population likely received more thiamin than required. However, this could be advantageous.   Due to unequal household food distribution, men are most likely to be the highest fish sauce consumers, therefore consuming the highest amount of thiamin. While not the target population for the prevention of infantile beriberi, this high thiamin intake likely poses no harm as there is no UL for thiamin consumption (18). In addition, thiamin needs increase with high carbohydrate intake (14,47). Since polished white rice is the staple in Cambodia, making up an estimated 60% of daily energy intake (8), increased thiamin consumption is likely helpful to ensure optimal energy metabolism. Marginal thiamin deficiency, which causes apathy, fatigue, loss of appetite, and dizziness (45), could also be prevented with improved population-wide thiamin intake.   Despite this, excess thiamin administration should be avoided. A balance must be struck between providing adequate thiamin to the target group, women of childbearing age who are likely the lowest consumers in the population, while protecting the highest consumers, typically men, against excess intake. Fortunately, there is no known toxicity or adverse events from excess thiamin intake, however it is unclear whether long term, chronic excess intake poses risk for  109 adverse events for an entire population. Economically, it is wasteful to include excess thiamin as it increases costs for producers and consumers. With the long-term goal of sustainability, higher costs for a fortified product compared to the traditional, non-fortified alternative, may decrease sales, especially among the population’s poorest and likely most at risk for thiamin inadequacy.   The case of thiamin fortification is not dissimilar to folic acid fortification of white wheat flour in Canada and the United States for the prevention of neural tube defects (NTD) (125). Coats and colleagues showed that not all infants of women with low biochemical thiamin status go on to develop infantile beriberi (11), just as not all women with low folate status will give birth to a NTD-affected baby (126). Fortification of a Canadian staple, white wheat flour, increased the folic acid intake of all Canadian women of childbearing age; this in turn shifted the red cell folate distribution upwards (126) and reduced the incidence of NTD (125,127). The aim of this research was to determine if fish sauce fortification could have the same effect on thiamin status for the prevention of infantile beriberi. One major difference, however, is that there is concern that population-wide folic acid fortification, while ideal for women of childbearing age for the prevention of NTD, poses a risk to elderly as it can mask vitamin B12 deficiency and result in irreversible neurological damage (128). There are no such known risks or adverse nutrient interactions associated with high thiamin intakes (18). Therefore, population-wide thiamin fortification can likely shift the eTDP distribution upwards, decreasing the proportion of women at risk of inadequate intake (18,87), and preventing infantile beriberi (57).   110 6.2.2 Effectiveness of thiamin-fortified fish sauce This was a proof-of-concept study designed to determine the effect of thiamin-fortified fish sauce consumption on eTDP among rural Cambodian consumers under controlled conditions. Both the randomized design and use of a control group in this study provide convincing evidence that the increase in eTDP seen among individuals in the LC and HC groups can be attributed to consumption of thiamin-fortified fish sauce rather than other factors such as seasonal differences in thiamin intake. As such, this study’s design was ideal to assess the objective of this study; I can conclude with confidence that thiamin-fortified fish sauce is an efficacious means of increasing eTDP among consumers. However, this study design has limitations and there is a need to determine the effectiveness of this intervention. While there were attempts made to mimic ‘real world’ conditions, including ad libitum consumption rather than a set consumption rate of fish sauce, more research is required to determine whether thiamin-fortified fish sauce is an ideal means of increasing dietary thiamin intake outside a controlled study environment.   In this study, participants were encouraged to consume fish sauce as they normally would. Although erythrocytes undergo a complete turnover in 120 days, the study duration of 6 months was selected to allow for a ‘run in’ period of high consumption so that if there were increased fish sauce consumption due to the novelty of free sauce, this would normalize after a few months, therefore allowing biomarkers to reflect more usual intake.   As noted in Chapter 4, 72% of women in the non-pregnant cohort felt that typical household fish sauce consumption changed during the study period; of those who indicated an increase in consumption, the majority indicated a large (59%) or a slight (36%) increase. While only 6% of  111 women attributed this change to the fact that fish sauce was free, more research is required to predict consumption trends when fish sauce is purchased at market. Such research could help to identify usual fish sauce intake, and aid in determination of the optimal thiamin fortification level of fish sauce in rural Cambodia.   There was no economic assessment of this product in my research. THCl was donated and fish sauce was made on a small scale in one fish sauce factory already outfitted with fortification equipment, therefore it is difficult to estimate the true cost of thiamin-fortified fish sauce. However, we can look to iron-fortified fish sauce for approximate costs. Iron fortification of fish sauce in Cambodia confers an increase of only 2 cents (US$) per litre of fish sauce, which was deemed affordable by >80% of Cambodians in a recent survey (n=2,797) (85). Importantly, NaFeEDTA is an expensive fortificant, whereas THCl fortification is much cheaper (87). Since iron fortification equipment and infrastructure are already present in the majority of Cambodian fish sauce factories (85,92), start-up costs to implement thiamin fortification would be minimal. However, future research should investigate the least expensive means of procurement of thiamin in Cambodia at scale.   6.2.3 Fish sauce as a fortification vehicle In Chapter 3 I reported that THCl was stable in a fish sauce matrix, and that thiamin-fortified fish sauce was well accepted by rural women of childbearing age. One major benefit of fortification is that it requires no behaviour change (87–90); with careful selection of a culturally-appropriate vehicle (a foodstuff commonly consumed by the majority of the population), routine consumption allows for passive consumption of the micronutrient of interest. Fish sauce has the  112 potential to be an ideal vehicle in Cambodia as it is consumed by 90% of the Cambodian population (85), and is already fortified with iron (84,86). However, there are some limitations of fortified fish sauce in Cambodia that must be considered.   Not all women consume commercially-produced fish sauce; poorer women may make their own fish sauce. Unfortunately there are no data on home production of fish sauce in Cambodia, or whether this practice is seasonal. In addition, as noted in Chapter 4, there is a wide range in quality of commercially-produced fish sauce in Cambodia, with three major categories of fish sauce available at market: high quality, with high protein content, dark colour, and rich fish flavours; medium quality, which is a slightly watered down version of high quality; and low quality, which is more dilute, lighter in colour, and contains artificial colours and flavours. I chose the medium quality fish sauce as a fortification vehicle and, as shown in Chapter 3, thiamin was stable in this matrix. However, the quality of the fish sauce may have been protective, with a higher protein (and therefore α-amino acid) content stabilizing thiamin (104) and the darker colour protecting thiamin against direct sunlight exposure (13,31,38,105). Low quality fish sauce would likely not confer these protective effects, and therefore all qualities of fish sauce should be investigated in the future to determine the stability of thiamin across all potential fish sauce vehicles in Cambodia.  6.2.4 Salt: an alternative thiamin fortification vehicle? Fish sauce is consumed by, but not necessarily purchased by, 90%