[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"post-reading-a-nutrition-study-without-getting-fooled":3,"post-reading-a-nutrition-study-without-getting-fooled-author":91,"post-reading-a-nutrition-study-without-getting-fooled-related":107},{"id":4,"title":5,"author":6,"body":7,"category":90,"description":13,"excerpt":91,"extension":92,"featuredImage":91,"meta":93,"navigation":94,"path":95,"publishedAt":96,"readingTime":97,"seo":98,"stem":101,"tags":102,"__hash__":106},"blog\u002Fblog\u002Freading-a-nutrition-study-without-getting-fooled.md","How to read a nutrition study without getting fooled","faculty-research-lead",{"type":8,"value":9,"toc":79},"minimark",[10,14,19,22,26,29,33,36,40,43,47,50,54,76],[11,12,13],"p",{},"Most nutrition headlines describe a study far more confidently than the study describes itself. The gap between the two is where careers in misinformation are built — and where careful practitioners earn their credibility.",[15,16,18],"h2",{"id":17},"start-with-the-design-not-the-conclusion","Start with the design, not the conclusion",[11,20,21],{},"Before reading the abstract's final sentence, find the design. A cross-sectional survey cannot establish that a food causes an outcome, no matter how large the sample. A randomised trial can, but only for the population, dose and duration actually studied.",[15,23,25],{"id":24},"ask-who-was-studied","Ask who was studied",[11,27,28],{},"A trial in 24 young male athletes tells you very little about a 68-year-old with chronic kidney disease. Generalisability is not a technicality; it decides whether the finding belongs in your consultation at all.",[15,30,32],{"id":31},"look-for-the-effect-size-not-just-the-p-value","Look for the effect size, not just the p-value",[11,34,35],{},"\"Statistically significant\" answers whether an effect is likely to be real. It does not answer whether the effect is large enough to matter. A 0.4 kg difference over twelve months is real and clinically irrelevant at the same time.",[15,37,39],{"id":38},"check-what-was-measured","Check what was measured",[11,41,42],{},"Self-reported dietary intake carries substantial measurement error. Surrogate endpoints — a biomarker moving in the right direction — are not the same as the outcome patients care about.",[15,44,46],{"id":45},"read-the-funding-and-conflicts","Read the funding and conflicts",[11,48,49],{},"Industry funding does not invalidate a study, but it belongs in your weighting, especially where the outcome measured is unusually favourable to a single product.",[15,51,53],{"id":52},"a-working-checklist","A working checklist",[55,56,57,61,64,67,70,73],"ol",{},[58,59,60],"li",{},"What was the design, and what can it support?",[58,62,63],{},"Who was studied, and for how long?",[58,65,66],{},"How large was the effect in absolute terms?",[58,68,69],{},"Was the outcome meaningful or a surrogate?",[58,71,72],{},"Who funded it, and were conflicts declared?",[58,74,75],{},"Does it agree with the wider body of evidence?",[11,77,78],{},"If a study fails several of these, it is not evidence you should be changing practice on — it is a hypothesis someone else needs to test properly.",{"title":80,"searchDepth":81,"depth":82,"links":83},"",2,3,[84,85,86,87,88,89],{"id":17,"depth":81,"text":18},{"id":24,"depth":81,"text":25},{"id":31,"depth":81,"text":32},{"id":38,"depth":81,"text":39},{"id":45,"depth":81,"text":46},{"id":52,"depth":81,"text":53},"Research",null,"md",{},true,"\u002Fblog\u002Freading-a-nutrition-study-without-getting-fooled","2026-06-18",9,{"metaTitle":99,"metaDescription":100,"title":5,"description":13},"How to read a nutrition study without getting fooled | Nexus Institute Blog","Headlines rarely survive contact with the methods section. A practical checklist for reading nutrition research critically.","blog\u002Freading-a-nutrition-study-without-getting-fooled",[103,104,105],"critical appraisal","evidence","research methods","LCmoNBnWz8TeerDqjqaQr1U4SIRTbl1eTk04WSUcmr8",[108],{"id":109,"title":110,"author":111,"body":112,"category":90,"description":116,"excerpt":91,"extension":92,"featuredImage":91,"meta":167,"navigation":94,"path":168,"publishedAt":169,"readingTime":170,"seo":171,"stem":174,"tags":175,"__hash__":179},"blog\u002Fblog\u002Fdesigning-a-food-frequency-questionnaire.md","Designing a food frequency questionnaire that actually works","faculty-biostatistics",{"type":8,"value":113,"toc":159},[114,117,121,124,128,131,135,138,142,145,149,152,156],[11,115,116],{},"A food frequency questionnaire is the workhorse of nutritional epidemiology, and the fastest way to produce a dataset that cannot answer your research question.",[15,118,120],{"id":119},"match-the-instrument-to-the-question","Match the instrument to the question",[11,122,123],{},"An FFQ built to rank participants by usual intake of a single nutrient looks different from one intended to capture whole dietary patterns. Decide which you need before drafting a single item.",[15,125,127],{"id":126},"build-the-food-list-from-local-data","Build the food list from local data",[11,129,130],{},"Borrowing an FFQ validated in another population is a common shortcut and a common source of error. The food list must reflect what your population actually eats, in the portions they actually eat.",[15,132,134],{"id":133},"keep-the-recall-period-realistic","Keep the recall period realistic",[11,136,137],{},"Asking about the past year gives usual intake but invites recall error. Asking about the past month reduces error and captures seasonality poorly. Both are defensible; the choice needs justifying in your methods.",[15,139,141],{"id":140},"portion-sizes-need-anchors","Portion sizes need anchors",[11,143,144],{},"Photographs, household measures or standard portions all work. Free-text estimation does not.",[15,146,148],{"id":147},"validate-before-you-rely-on-it","Validate before you rely on it",[11,150,151],{},"Compare against multiple 24-hour recalls or weighed records in a subsample, report correlation coefficients by nutrient, and be transparent about which nutrients your instrument measures poorly. Every FFQ measures something poorly.",[15,153,155],{"id":154},"pilot-with-the-real-population","Pilot with the real population",[11,157,158],{},"A pilot with twenty participants from your target group will find problems that months of internal review will not: confusing item wording, missing staple foods, and the point at which people stop concentrating.",{"title":80,"searchDepth":81,"depth":82,"links":160},[161,162,163,164,165,166],{"id":119,"depth":81,"text":120},{"id":126,"depth":81,"text":127},{"id":133,"depth":81,"text":134},{"id":140,"depth":81,"text":141},{"id":147,"depth":81,"text":148},{"id":154,"depth":81,"text":155},{},"\u002Fblog\u002Fdesigning-a-food-frequency-questionnaire","2026-05-20",10,{"metaTitle":172,"metaDescription":173,"title":110,"description":116},"Designing a food frequency questionnaire that actually works | Nexus Institute Blog","FFQs are cheap to administer and easy to design badly. What separates a usable instrument from an unusable one.","blog\u002Fdesigning-a-food-frequency-questionnaire",[176,177,178],"FFQ","dietary assessment","validation","FvoC8tdwkWP5ZGuZN9Efj1yl25BxTuK7_C00YHw5RNM"]