In August 2017, roughly 700,000 Rohingya people crossed the border from Myanmar's Rakhine State into Bangladesh in a matter of weeks. By the time I arrived at the IRC compound in Cox's Bazar's Baharchara that December, the camps at Kutupalong had become the largest refugee settlement on earth — and the health system serving them was being built in real time, in the mud, with whatever we had.

I want to write honestly about what that experience was like — and more importantly, what it taught me about health systems, evidence, and the gap between what epidemiology says should work and what actually works in a field setting where the floor is literally unstable.

Arriving at the Edge of the System

Nothing prepares you for the scale of Cox's Bazar during the acute Rohingya response. The standard frameworks we use for humanitarian health — SPHERE standards, WHO minimum initial service packages, inter-agency coordination protocols — were all technically in place. And they were also, simultaneously, overwhelmed.

My role as Senior Health Manager at IRC meant overseeing the implementation of emergency health programmes across multiple sites. On paper, this involved monitoring project indicators, coordinating with donors, and ensuring that our work aligned with the Ministry of Health and Family Welfare's protocols. In practice, it meant making decisions in conditions of radical uncertainty, with imperfect data, every single day.

Building From the Ground Up: PHC and BEmONC

One of the most consequential things I was part of during those four years was establishing two Primary Health Care Centres and a standalone Basic Emergency Obstetric and Neonatal Care (BEmONC) facility. If you have only worked in academic or institutional settings, the phrase "establishing a health facility" might conjure images of procurement paperwork and staff onboarding. Let me describe what it actually involved.

It began with site selection — identifying land that was not on a flood plain, accessible to the highest-density areas of the camp, reachable by ambulance, and acceptable to both the refugee community and the local Bangladeshi government. It meant designing physical layouts that could accommodate patient flow while meeting minimum infection prevention standards in a context where running water was intermittent and electricity unreliable.

It meant recruiting and training health staff — a mix of Bangladeshi clinical officers, Rohingya community health workers, and international clinical staff — and ensuring that they were all operating from the same treatment protocols, even when those protocols had to be adapted from WHO guidelines written for health systems with laboratory capacity we simply did not have.

"A BEmONC facility is not a building. It is a set of competencies, supplies, and protocols working together. You can construct the building in a week. The rest takes longer — and the rest is what saves lives."

The BEmONC centre was particularly important. Maternal mortality among Rohingya women had been catastrophically high in Myanmar, and the camp setting created new risks — overcrowding, poor sanitation, limited antenatal care, and a population with deeply held cultural norms around male health workers attending births. We worked with female community mobilisers, adapted our referral pathways, and built trust incrementally. There is no dataset that fully captures what that trust cost to build, or what it was worth.

The Integrated Multisector Team: When Silos Kill

One of the things I am most proud of from Cox's Bazar was designing and establishing what we called the Integrated Multisector Team (IMT) — a structure that brought together health, nutrition, water and sanitation, and protection staff into a single coordinated response unit.

The logic was simple: a malnourished child with diarrhoea is not a nutrition problem or a WASH problem. She is a child whose recovery depends on five different interventions happening simultaneously, delivered by people who are actually talking to each other. Siloed programming — where health workers track health indicators, nutrition workers track anthropometrics, and WASH workers track latrine coverage, and none of them sit in the same room — is one of the most persistent structural failures in humanitarian response.

Getting the IMT to function required convincing programme managers from different sectors to share data and coordinate referrals in ways that cut against their institutional incentives. Donors fund sectors. Reporting frameworks are sector-specific. The whole architecture of humanitarian financing pushes towards silos. Breaking through that, even partially, took sustained negotiation and constant documentation of the benefits.

What the Field Taught Me About Evidence

I came to Cox's Bazar with an MPH. I had studied epidemiology, research methods, and evidence-based public health. And the field immediately showed me the limits of what I thought I knew.

Evidence-based medicine, as practised in high-income countries, rests on assumptions that do not hold in complex emergency settings. It assumes you have diagnostic capacity. It assumes your patient population matches the trial populations that generated the evidence. It assumes supply chains are reliable enough to actually deliver the intervention the evidence supports. None of these assumptions held consistently in Cox's Bazar.

This is not an argument against evidence. It is an argument for understanding what evidence is: a best available approximation, generated in specific contexts, that must be adapted — thoughtfully, systematically, and with humility — to the context you are actually working in. That is what praxis means to me. Not the rejection of theory, but its responsible enactment in the world as it actually is.

The Decision I Still Think About

There is one decision from Cox's Bazar that I think about more than any other. In 2019, during a cholera outbreak in the camps, we faced a resource allocation question: with limited oral rehydration supplies and insufficient staff, how do we prioritise? The textbook answer is straightforward — triage by severity, prioritise children under five and the elderly. In practice, the community dynamics, the layout of the camps, and the trust relationships we had built meant that the textbook answer would have left some of the most vulnerable people unreached.

We made a different decision. I believe it was the right one. I cannot prove it with a randomised trial. And that experience — of having to act with incomplete evidence, under time pressure, with real lives at stake — shaped my understanding of what public health research is ultimately for more deeply than any coursework ever did.

What I Carry Forward

I returned to academia — to McMaster — not to escape the complexity of field work, but to develop better tools for navigating it. The questions that drive my PhD research all have roots in Cox's Bazar: How do we measure what people eat in resource-limited settings? How do we understand the interaction of multiple diseases when our frameworks address them separately? How do we build evidence that is consequential — that actually changes what happens to the people it is about?

The Rohingya response was not a success story without qualification. Hundreds of thousands of people remain displaced, stateless, and without durable solutions. The health gains we made in those four years are real and meaningful — and they are also fragile, dependent on continued funding that is never guaranteed. Praxis is not a single act. It is an ongoing commitment to ensuring that knowledge serves the people who most need it to.

Imagine you are a health planner in Dhaka. You have tuberculosis rates that remain stubbornly high. You have a diabetes epidemic that is accelerating. You have coastal communities where the groundwater is saline and the rate of gestational hypertension is alarmingly elevated. You have mental health needs that the formal health system almost entirely ignores. And you have all of this happening simultaneously, in a country where the health budget per capita is among the lowest in Asia.

The standard public health response to this situation is to create vertical programmes: a TB programme, a diabetes programme, a hypertension programme, a mental health programme. Each with its own funding stream, its own indicators, its own staff training, its own reporting chain. Efficient, legible, and donor-friendly.

Syndemic theory says this is the wrong approach — not just operationally, but conceptually. It argues that treating these as separate problems misunderstands what they actually are.

The Concept: What Singer Actually Said

Medical anthropologist Merrill Singer introduced the concept of syndemics in the 1990s to describe a situation in which two or more epidemics interact synergistically — meaning they do not just co-occur but actively worsen each other — and where this interaction is driven and sustained by underlying social conditions. The word "syndemic" is a portmanteau of "synergistic" and "epidemic."

This third criterion is what gives syndemic theory its political edge. It insists that you cannot fully understand a syndemic without asking who bears the burden of it, and why. And in almost every documented case, the answer involves structural inequality.

Why Bangladesh Is a Textbook Syndemic Context

Bangladesh is in what demographers call a "protracted polarised" epidemiological transition. Unlike high-income countries, which largely completed the transition from infectious to non-communicable disease dominance over the twentieth century, Bangladesh is managing both simultaneously. TB kills tens of thousands annually. Diabetes prevalence in adults now exceeds ten percent. Cardiovascular disease is the leading cause of death. And infectious diseases have not gone away.

These diseases interact in ways that compound their harm. Consider tuberculosis and diabetes. TB damages lung tissue and triggers systemic inflammation. That inflammation impairs insulin signalling, contributing to glucose dysregulation. Conversely, the hyperglycaemia of diabetes impairs macrophage function — the body's primary defence against mycobacterium tuberculosis. A person with diabetes is two to three times more likely to develop active TB. And a person with active TB who also has diabetes is significantly more likely to fail treatment, relapse, or die.

Now layer in coastal Bangladesh. Salinity intrusion — driven by sea-level rise, cyclone damage, and upstream river diversion — has contaminated freshwater sources across Khulna, Satkhira, and Barguna districts. High dietary sodium from salinised water is directly associated with elevated blood pressure and gestational hypertension. The same communities dealing with hypertension are also dealing with arsenic contamination from tube wells — arsenic that acts as an endocrine disruptor impairing pancreatic beta-cell function and increasing diabetes risk.

"The syndemic is not a complication of the health crisis in Bangladesh. In many districts, it is the health crisis."

Mental Health: The Invisible Amplifier

Syndemic theory is perhaps most clarifying when applied to mental health. Depression and anxiety are not peripheral to TB or diabetes management. They are central to it. Depression impairs motivation to seek care, to take medications, to attend follow-up appointments. In TB, non-adherence to treatment — driven in significant part by the psychological burden of the disease and stigma — is the primary driver of multidrug-resistant TB (MDR-TB).

This is what I call the Adherence Loop: depression reduces adherence → non-adherence produces treatment failure → treatment failure generates MDR-TB → MDR-TB treatment is longer, harder, and more isolating → isolation worsens depression. It is a loop that Bangladesh's health system is not currently equipped to break.

What Syndemic Thinking Changes in Practice

If you accept syndemic theory, several things follow. First, screening must be bidirectional — TB clinics should screen for diabetes, diabetes clinics should screen for TB, antenatal care in coastal districts should screen for hypertension and arsenic exposure. Second, treatment protocols must be coordinated — the pharmacokinetic interaction between rifampicin and oral hypoglycaemics is well documented but almost never managed systematically in Bangladeshi clinical settings. Third, and most important: addressing the syndemic means addressing the structural conditions that drive it. Clean water access. Urban poverty reduction. Mental health integration into primary care. In a syndemic framework, these are primary interventions, not supplementary ones.

The Bangladesh Syndemic Series

This is the theoretical foundation of the research I am currently conducting: a sequence of three linked scoping reviews that examine each major syndemic cluster in Bangladesh through both a syndemic lens and a consequentialist filter. The goal is not just to describe the syndemics that exist, but to identify the interventions with the greatest potential to reduce burden, reduce inequity, and be realistically implemented within Bangladesh's health system capacity. That combination — syndemic diagnosis, consequentialist triage — is, I believe, what rigorous global health research in 2025 should look like.

There is a quiet but important debate running through academic epidemiology, and it rarely makes it into the methods courses or the journal articles. It is a debate about what epidemiology is fundamentally for. Is it a science of description and causal inference — concerned primarily with identifying relationships between exposures and outcomes? Or is it a moral enterprise — one whose ultimate justification lies in what it does to improve the health of populations?

Sandro Galea, in a landmark 2013 commentary in the American Journal of Epidemiology, made the case for the latter position and argued that epidemiology should explicitly adopt a consequentialist framework: one that evaluates the value of epidemiological knowledge by its consequences for population health.

The Problem with Pure-Science Epidemiology

The dominant tradition in academic epidemiology is concerned above all with internal validity: Did we adequately control for confounding? Was the exposure measurement reliable? These are important questions. But the paradigm can produce a kind of methodological perfectionism that is disconnected from consequence. You can run a methodologically flawless study on a question that, even if perfectly answered, will not meaningfully reduce burden of disease. You can publish extensively on exposures that affect tiny fractions of a population while large structural determinants of health go understudied because they are harder to randomise.

This framework does not tell epidemiologists what not to study. It provides a hierarchy of urgency — a way of thinking about where scarce research resources should be concentrated if the goal is to maximise population health impact.

Consequentialism as a Moral Philosophy

The philosophical roots of Galea's argument are worth understanding. Consequentialism — the tradition most associated with Jeremy Bentham and John Stuart Mill — holds that the moral value of an action is determined entirely by its outcomes. The right action is the one that produces the best consequences, typically understood as the greatest well-being for the greatest number of people.

Applied to epidemiology, this is demanding. It says that the moral worth of a research programme is not determined by its methodological elegance or citation impact — it is determined by whether it reduces suffering.

"If we believe that the point of population health science is to improve population health, then we must evaluate the science by that criterion — not by the elegance of the methods or the prestige of the journal."

Why This Matters for My Research

When I designed the Bangladesh Syndemic Series, I deliberately adopted Galea's consequentialist filter as an organising principle. The choice of which disease clusters to study was not arbitrary. Each was selected because the burden is large, the interaction is potentially modifiable, and addressing these syndemics would disproportionately benefit the most disadvantaged Bangladeshis. The final phase of the series — a policy brief aimed at Bangladesh's Ministry of Health and Family Welfare — is the moment at which the consequentialist commitment is fully enacted. The research is not finished when the papers are published. It is finished when the evidence has been translated into a form that policymakers can actually use.

The Limits of Consequentialism

I hold this framework seriously, but with appropriate scepticism. Consequentialism has well-known failure modes — it can, in principle, justify violating individual rights for aggregate benefit. Modern public health consequentialism guards against this by building in equity constraints. Research that produces large average effects but concentrates benefits among the already-privileged fails the consequentialist test, even if its total welfare impact appears positive. This is why syndemic theory and consequentialist epidemiology fit together so naturally. Syndemics cluster in the most disadvantaged communities. Consequentialist epidemiology demands that we focus precisely there.

Ask someone what they ate yesterday and they will probably give you an answer. Ask them what they ate on average over the past three months and the answer will be more approximate. Ask them to recall every ingredient, portion size, and preparation method — accounting for seasonal variation, social eating, and the difference between what they intended to eat and what they actually consumed — and you are asking for something that human memory simply cannot reliably provide.

This is the central problem of nutritional epidemiology, and it is more consequential than most people — including many health researchers — appreciate. The question of what people eat is not just a logistical challenge. It is a measurement validity problem that propagates through every analysis and every conclusion that nutritional science produces.

The Three Main Tools and Their Problems

1. The 24-Hour Dietary Recall

The 24-hour dietary recall (24HR) asks participants to report everything they consumed in the previous twenty-four hours, typically guided by a trained interviewer using a multiple-pass method. It is considered the gold standard for estimating usual intake at the population level — when repeated across multiple days. The problems: a single 24HR captures only one day. People under-report systematically. They under-report foods that carry social stigma and over-report foods they believe are healthy. And in low- and middle-income country settings, multi-pass protocols developed in Western populations do not map cleanly onto local food cultures and preparation methods.

2. The Food Frequency Questionnaire (FFQ)

The food frequency questionnaire asks participants how often they eat a pre-specified list of foods over a defined reference period. Because it captures habitual dietary patterns rather than single-day intake, it is the most commonly used tool in large epidemiological cohort studies. But the FFQ has its own deep problems. The food list must be culturally appropriate — a FFQ developed for a Canadian population of European origin will systematically miss the dietary exposures most relevant to a South Asian immigrant population. Lentil preparations, fermented rice dishes, mustard oil cooking, betel nut chewing — these are nutritionally meaningful and culturally central, and they appear on almost no standard Canadian FFQ.

Cognitive testing is a critical but underused step in FFQ development. Before a questionnaire is administered, it should be tested with representatives of the target population to determine whether they understand the questions as the developers intended, whether the food items are recognisable and culturally salient, and whether the response categories make sense in practice. My scoping review on FFQ validation methods found that cognitive testing is conducted inconsistently — and sometimes not at all — in newly developed questionnaires.

3. Biomarkers

Biomarkers — objective biological measures of nutrient status such as serum vitamin D or urinary sodium — are not subject to the recall and social desirability biases of self-reported dietary data. They are often used as validation criteria: if a FFQ score correlates with a relevant biomarker, that is evidence the FFQ is measuring what it claims to. But biomarkers have their own limitations. They reflect not just dietary intake but also absorption, metabolism, and physiological state. A woman in her second trimester of pregnancy has different serum ferritin kinetics than the same woman post-partum, independent of her dietary iron intake.

Measurement Error and What It Does to Your Results

When exposure is measured with error — and all dietary assessment tools have error — associations between diet and health outcomes are attenuated toward the null. This means that real effects appear smaller than they are. Studies reporting that a dietary pattern "has no significant effect" may be reporting a true null — or they may be reporting an attenuated effect that is real but too small to detect given the measurement error. Without knowing the magnitude and direction of the error, you cannot distinguish these two interpretations.

"Nutritional epidemiology does not have a replication crisis. It has a measurement crisis. Fix the measurement and many of the inconsistencies in the literature resolve themselves."

What Better Measurement Looks Like

Better measurement involves several things working together: culturally appropriate food lists developed with community input, multiple assessment methods used in combination, explicit measurement error correction models applied during analysis, biomarker validation where feasible, and population-specific tools rather than assumed transferability across cultural contexts. The questions that nutritional epidemiology is trying to answer — what should people eat to live longer, healthier lives; how does prenatal nutrition shape child development; how does dietary transition in immigrant populations alter chronic disease risk — are among the most consequential questions in public health. They deserve measurement that is adequate to the importance of the answers.

In the early months of the COVID-19 pandemic, hospitals around the world were making decisions under conditions of radical uncertainty. The pathogen was new, the clinical spectrum was poorly understood, and there was no evidence-based treatment protocol. In that context, a reasonable precautionary instinct led many clinicians to prescribe antibiotics empirically — as protection against the bacterial superinfections that had complicated influenza pandemics in the past, and to cover for the possibility that what looked like COVID-19 might also involve bacterial pneumonia.

The problem is that COVID-19 is caused by a virus. Antibiotics do not treat viral infections. And as my MSc systematic review documented, the scale of empiric antibiotic prescribing for hospitalised COVID-19 patients was, in global terms, extraordinary — and its long-term consequences for antimicrobial resistance may prove to be one of the most serious secondary harms of the pandemic.

What the Evidence Showed

My MSc thesis at McMaster involved a systematic review and meta-analysis of studies examining antibiotic use in hospitalised COVID-19 patients. The findings were striking: across the studies I reviewed, the majority of hospitalised COVID-19 patients received antibiotics — in many hospital settings, prescription rates exceeded seventy percent, despite published estimates suggesting that bacterial co-infection was present in only a small minority of cases. The most commonly prescribed classes included macrolides (particularly azithromycin), beta-lactams, and fluoroquinolones. In many studies, the rationale was empiric — the decision was made without microbiological confirmation of bacterial infection.

Why COVID-19 and AMR Are a Syndemic

The relationship between COVID-19 and antimicrobial resistance is not incidental. It is a syndemic relationship: two crises that interact and compound each other, driven by underlying structural conditions. COVID-19 created the pressure that drove antibiotic overuse — clinical uncertainty, overwhelmed health systems, inadequate diagnostic capacity. Those structural conditions are not unique to COVID-19; they are features of health systems globally that will remain in place for the next pandemic.

Meanwhile, the antibiotics consumed in COVID-19 wards in 2020 and 2021 contributed selective pressure for resistance — in hospital environments that are among the most important reservoirs for resistant organisms: wards where immunocompromised patients are concentrated, where horizontal gene transfer between bacteria occurs at high rates, and where the consequences of resistant infection are most severe.

"We treated a viral pandemic with antibiotics and called it cautious medicine. What we were actually doing was borrowing from the future — paying for short-term clinical comfort with long-term resistance risk that will be borne by patients who have not yet been born."

What Should Have Happened

The alternative to empiric antibiotic prescribing is not doing nothing. It is microbiological stewardship: using rapid diagnostics to identify whether bacterial co-infection is actually present, prescribing antibiotics only when indicated, and when prescribing, selecting the narrowest-spectrum agent appropriate for the likely organism. In many hospital settings during COVID-19, this capacity was overwhelmed. Antibiotic stewardship programmes — which exist in many high-income country hospitals but are far less consistently implemented in LMICs — were deprioritised during the acute phase of the pandemic. These are structural failures, not individual ones. The solution is investment in diagnostic infrastructure and antibiotic stewardship capacity before the next pandemic — not during it.

The Research Gap I Found

One of the most important findings from my systematic review was methodological: the quality of reporting in studies examining antibiotic use in COVID-19 patients was highly variable. Definitions of bacterial co-infection differed across studies. Antibiotic prescription rationale was inconsistently documented. Follow-up data on patient outcomes were often absent. This makes it very difficult to estimate the net harm of the prescribing patterns we documented, or to identify the clinical circumstances in which antibiotic prescribing might have been genuinely warranted. Better research design in this area is needed before the next pandemic — because there will be a next pandemic.

When I left IRC in June 2021 to begin my MSc at McMaster, more than a few colleagues asked me why. I was Senior Health Manager. I had successfully led major funded projects. I had a title, a field role, and the peculiar satisfaction that comes from solving real problems in real time in places where the problems are urgent. Why would I swap that for a student desk in Hamilton, Ontario?

The honest answer is that Cox's Bazar broke something open in my thinking. Not in a damaging way — in a productive one. Four years of managing health programmes in the world's largest refugee settlement had shown me the limits of my own analytical tools. I was good at implementing evidence. I was much less equipped to evaluate it critically, to generate it, or to identify the gaps in it that mattered most.

The Frustration of Good Intentions Without Good Evidence

Let me be specific. Around 2019, IRC was designing a new integrated mental health and psychosocial support component for its Rohingya health programming. There was a clear need — the population had experienced mass violence and ongoing uncertainty, and rates of depression, anxiety, and PTSD among Rohingya adults were extremely high by every available estimate.

The problem was that the evidence base for MHPSS interventions in humanitarian settings was thin, inconsistent, and almost entirely derived from populations that were not demographically or culturally similar to the Rohingya. For example, we were trying to design an intervention for a Muslim, Bangladeshi-speaking, predominantly rural population that had experienced a specific historical trauma — and the closest analogous evidence came from Syrian refugee programmes and studies conducted with non-Muslim populations.

This is not a failure of practitioners. It is a failure of research investment. The populations that most need mental health interventions are precisely the populations that are hardest to study. And so the evidence follows easier paths, and practitioners in the field make decisions based on incomplete and poorly transferable knowledge.

"I did not return to academia because I had fallen out of love with field work. I returned because I wanted to be the kind of researcher who makes field work better."

The Decision to Go Back

The decision crystallised over about six months in 2020 and 2021. I had been reading more — epidemiology journals, global health policy documents, systematic reviews. And I kept running into questions I could not answer. How do you measure diet in a population that prepares food communally and does not think of eating in terms of individual portions? How do you design a screening programme for two diseases simultaneously when your health system is built around single-disease vertical programmes? How do you know whether an intervention worked in a setting where randomisation is impossible and follow-up data is spotty? These questions led me to McMaster.

What Was Hard About the Transition

I want to be honest about what the transition was actually like, because the version of this story that gets told publicly — "practitioner goes back to school, flourishes intellectually, lives happily ever after" — leaves out a lot. Going from Senior Health Manager to first-year MSc student is a significant identity adjustment. I had been running multi-million-dollar programmes and managing large teams. Suddenly I was in a classroom being told I needed to understand the difference between internal and external validity. The knowledge was genuinely useful. The transition in status was genuinely disorienting.

There is also a particular loneliness that comes from being an older student with field experience in an academic environment where the dominant professional model is the straight-line trajectory: undergraduate → graduate → postdoc → faculty. People do not always know what to do with a CV that includes "established BEmONC facility in Rohingya refugee camp" between the MPH and the MSc.

What Academia Has Given Me

With that said: the return has been worth it in every way that matters. The MSc taught me systematic review — a methodology I now consider one of the most valuable analytical tools in global health. The PhD has given me the space to think carefully, over an extended period, about problems that cannot be solved in a six-month project cycle. It has also given me language and frameworks for things I already knew intuitively from the field. Syndemic theory, for instance, was already implicit in how I was thinking about the Rohingya health situation — the interconnection of nutrition, infectious disease, mental health, and structural poverty was not news to me. But having the theoretical language for it has made my thinking more precise.

What I Wish I Had Known

If I could say one thing to someone in a similar position — a practitioner considering a return to graduate school — it is this: the value of the transition is directly proportional to the clarity with which you can articulate the questions you are returning to answer. If you go back to academia because it feels like the right thing to do, you will drift. If you go back because four years in Cox's Bazar generated specific, concrete questions that you cannot answer without better analytical tools — that clarity will sustain you through the identity adjustment and the loneliness. I went back for the questions. I am still working on the answers. That feels like exactly the right place to be.

There is a well-documented epidemiological phenomenon called the "healthy immigrant effect": people who arrive in Canada from other countries tend, on average, to be healthier than the Canadian-born population at the time of arrival. They have lower rates of obesity, diabetes, cardiovascular disease, and many cancers. Within a decade of arrival, that health advantage erodes significantly — converging toward, and in some domains falling below, the health profile of the Canadian-born population.

Diet is one of the most important mechanisms driving this convergence. What people eat changes when they move countries. And not always in the direction that naive intuitions suggest.

The Dietary Transition

When I arrived in Canada from Bangladesh, I noticed my own dietary patterns shifting within months. Not because I intended to eat differently — but because the food environment I was embedded in was completely different. The corner store near my apartment in Hamilton did not stock mustard oil or hilsa fish or the varieties of dal I had grown up eating. Getting culturally familiar food required planning, travel, and in some cases paying premium prices at specialty stores.

Meanwhile, the foods that were cheaply, conveniently, and constantly available — fast food chains, packaged snacks, sweetened beverages, white bread — were exactly the ultra-processed products I had studied in my MPH as drivers of the non-communicable disease epidemic. The food environment was engineered to push me toward a dietary pattern that the evidence clearly identifies as harmful. And it was doing so to millions of immigrants and international students across the country, simultaneously.

South Asian Dietary Patterns and the Transition Risk

Traditional South Asian dietary patterns — when maintained — are associated with significant health benefits. High consumption of legumes provides fibre and slow-digesting carbohydrates. Fermented foods provide probiotic benefit. Turmeric and other spices have documented anti-inflammatory properties.

But South Asian dietary patterns also include features that become amplified in the immigration context in ways that increase cardiometabolic risk. White rice consumption — high in many South Asian diets — carries greater glycaemic implications when the accompanying dietary fibre from traditional preparations is replaced by processed alternatives. And the stress of immigration itself — social isolation, precarious employment, unfamiliar social norms — is associated with emotional eating and reduced diet quality independent of food environment changes.

The Measurement Problem

To understand how dietary patterns change after immigration, you need to be able to measure what people eat. And as I have written about elsewhere, dietary assessment is genuinely difficult. For South Asian immigrants in Canada, it is particularly difficult because the standard tools used in Canadian health research were not designed for this population. A food frequency questionnaire developed for a White European population will ask about bread consumption, pasta intake, and dairy servings. It will almost certainly not ask about roti frequency, dal variety, or the use of fenugreek seeds.

When you administer such a tool to a South Asian respondent, you systematically miss much of what they eat. The resulting dietary profile is a ghost — a partial image that reflects the Canadian food environment they have been partially absorbed into, but obscures the traditional dietary patterns they have maintained. This matters enormously for research: if your tool misclassifies South Asian immigrants as having "low legume intake" because it does not ask about the specific legumes they eat, any study built on that tool will produce biased estimates of the diet-health relationship in this population.

The Scale Development Project

One of my ongoing research projects addresses this from a different angle. Rather than asking what South Asian immigrants eat, it asks: how do their food habits change after arriving in Canada? And can we develop a validated, psychometrically sound scale to measure that change? This scale development project focuses initially on international students — a population that experiences dietary transition rapidly and intensively, and that is geographically concentrated, making research feasible. The scale asks about changes in cooking practices, ingredient substitution, meal timing, social eating patterns, and attitudes toward traditional versus host-country food.

Why This Is a Public Health Priority

The healthy immigrant effect is not inevitable. It is a product of specific, modifiable conditions — a food environment that makes poor nutritional choices cheap and convenient; social conditions that erode the community food culture through which traditional dietary patterns are transmitted; health systems that do not screen or counsel immigrant populations on the specific dietary risks of acculturation; and research tools that are not adequate to measure the exposure in the first place. My research contributes at the measurement and epidemiological level. But measurement is in service of something larger: the development of evidence-based interventions that support immigrant and newcomer populations in maintaining the health advantages they arrive with. That is a consequential goal. And it begins with being able to see, clearly, what people eat.