If the 1990s were about digitising hospital records, the early 2000s were about upgrading the software of our minds. In 2000, the Medico Friends Circle held its annual meeting at Yatri Niwas, a modest guest house across the road from Gandhi’s Ashram in Sevagram.
MFC meetings had a distinct flavour: simple arrangements, serious conversations, a fierce and unvarnished idealism. We sat cross-legged on the floor, arguing about social determinants of health over endless cups of tea, while ceiling fans pushed warm air from one corner of the room to another in a futile attempt at cooling. It was in that sweltering, intellectually charged room that I met Dr. Madhukar Pai — known to everyone simply as Madhu.
He was a young graduate of Madras Medical College, then pursuing his MD in Community Medicine at CMC Vellore. Madhu spoke with a clarity that was almost startling — not loud, not dramatic, just sharp, his thoughts organised like a perfect clinical note. We fell to talking about research, and I confessed my growing fascination with Evidence-Based Medicine, a term I had first come across in 1996 in the writings of David Sackett. Madhu’s eyes lit up. He was already running workshops on epidemiology and statistics of his own. In that small room, with the scent of chai and the low buzz of activists outside, we decided to work together. It was a casual decision, made between sips of tea, that quietly changed the direction of my professional life.
The workshop that woke the wards
In February 2000 we ran a three-day workshop on Statistics and Study Designs at Sevagram. Thirty residents signed up, from Medicine, Paediatrics, Obstetrics, and Surgery. Most arrived wearing the expression I knew well from thesis season — polite interest thinly stretched over dread. They expected the usual diet of p-values, t-tests, and academic tedium.
Madhu had other plans. He didn’t teach statistics as a punishment; he taught it as a torch. He showed the residents how bias bends the truth, why randomisation is simply fairness written into a study design, how to read a paper without being seduced by its own confident conclusions. He made confidence intervals feel like common sense rather than a mathematical hurdle. For eight hours a day, the residents stayed awake — fully, vibrantly awake. In Sevagram, that alone was a minor miracle. By the third evening the dread had softened into something like curiosity. Several residents asked whether Madhu could simply join MGIMS full-time. I remember thinking that we were no longer just treating patients; we were learning to interrogate our own habits.
Three years later the pull of that new world took me to its birthplace — McMaster University in Hamilton, Ontario, where Evidence-Based Medicine had been shaped into a global movement. That trip, and the missed connection at Heathrow, the workshop with Gordon Guyatt and Deborah Cook, the UK alumni tour that followed it, ending with two hours on snakebite in Oxford, is a story worth telling on its own, and it takes up the chapters that follow. I came home with a suitcase full of books and a new way of seeing my own work. The evidence revolution had begun in Sevagram, but for me it was only the prelude to a bigger one: a year at Berkeley.
Berkeley, With Madhu Already There
He had already taught me how to write a research paper, on an earlier visit to Sevagram in 2000, in a workshop that grew into another. He taught me to write a first draft without flinching, and introduced me to the statistical software Stata and the reference manager EndNote, tools that stayed with me for the rest of my career.
Our own run of papers together began in early 2003, with Priya Badam, working with Vaishali Solao, asking whether a stroke could be told apart from a bleed at the bedside. A few months later, my resident Manoj Singh took a question just as simple — could a nurse’s palm reliably catch a fever — and got it into a major international journal. The next year brought three more theses built the same way: Ravindra Indra asking whether a failing thyroid announced itself on examination before any blood test confirmed it, Samir Patil looking at risk factors for acute myocardial infarction in rural central India, and Rajnish Joshi, with Namita Jajoo and Amandeep Singh, asking whether percussion alone could detect an enlarged liver.
Then, in 2005, the manuscript I’d watched him revise in Berkeley — his own PhD work with my resident Kaustubh Gokhale, comparing a new blood test for TB against the old skin test in health workers across Sevagram — finally appeared in JAMA, the same year Geetha Devi asked whether an ailing brain announced itself on examination before any scan confirmed it. Two more theses followed in 2007: Trunal Lokhande asking whether the old signs of a fluid-filled chest could still be trusted at the bedside, and, built on the same stroke data as Priya Badam’s, a paper with Rajnish and Vaishali again, working out how to turn such bedside signs into a proper predictive score. Then, just as abruptly, Madhu drew a line: no more joint papers unless they were on tuberculosis. His reasoning was simple. He wanted a career built on TB, and anything outside it would make his CV diffuse — a risk he wasn’t willing to take. True to that, when two more residents, Udit Narang and Vishakha Jain, added their names to a paper of ours in 2011, it was on TB again — a follow-up study asking how well those same tests predicted who would actually go on to develop the disease.
Thirty-four papers, one long argument
If I sat down and counted every manuscript Madhu and I put our names to, jointly or with Rajnish and the rest of our circle — the QFT/TST reproducibility series, the healthcare-worker testing study, nosocomial TB in Indian hospitals, two reviews of TB diagnostics, a malaria diagnostic-accuracy meta-analysis, papers on pleural TB and meningitis, a study of cardiovascular risk in the rural elderly, and household health insurance with Rajnish — the count comes to thirty-four, spread across more than two decades and a dozen journals. What strikes me now, looking back over that trail, is how little of it was ever really about the papers. Each one was an argument, conducted in the open, about how medicine ought to know what it claims to know — the same argument Madhu and I had first had over tea at Yatri Niwas in 2000, simply carried on at a larger scale, with real patients and real stakes attached. That, I think, is the truest measure of what began in that sweltering MFC room: not that two people became friends, though we did, but that a friendship built out of evidence-based medicine kept generating more evidence to be based on.
Two States
Their marriage would have made an unlikely case study on its own. Madhu was from the South, Tamil to the bone; Nitika grew up in Allahabad, a UP girl who spoke the confident, fast Hindi of that city, and Madhu barely knew a word of it. It was, more or less, the plot Chetan Bhagat would sell a decade later as 2 States — except no novelist was needed to invent the ending.
Nitika was another pillar of those Berkeley days. She had her own work going — HIV in women, done jointly with our Ob-Gyn faculty back in Sevagram, published in its own right — but she still found room to host Bhavana for dinner more than once when Bhavana came to Berkeley for three weeks in December 2004, and when I finally left Berkeley for good, it was Nitika who packed my bags with the same care a family member would bring to the job. I have never quite been able to thank the two of them enough.
Whatever the odds on paper, the two of them made it work, professionally and at home, and raised a daughter, Annika, and a dog, Dora, in whatever city his research happened to plant them next. Years later, asked in an interview where he’d been happiest, Madhu didn’t name Montreal, or the Chair at McGill, or any of the awards — he named the Bay Area, the years the two of them spent training there together. Asked, in the same interview, whom he’d most like to apologise to, he named Nitika, Annika, and Dora, for all his globe-trotting — though he added, in the same dry register I’d first heard at Yatri Niwas, that they understood his global-health mission and hadn’t thrown him out of the house yet.
What the friendship carried
The support was never only scientific. Madhu insisted my name, and Rajnish’s, belonged on our papers — not as a courtesy but because he meant it. Once, after I sent him a compiled list of his own publications, he wrote back, half-joking, that there was at least one person in Sevagram who valued his work — the kind of small exchange that went both ways between us for twenty years. When my income tax paperwork needed sorting out one year, he did it, and joked about depositing the cheques himself. When a colleague’s mother died, or another collapsed with a heart attack boarding a train, Madhu was often the first to write asking what he could do, from three thousand miles away in Montreal. Whenever the platform was his to give — a book chapter, an UpToDate entry, a place in a WHO working group — he found a way to route it through Sevagram rather than around it. He did the same for others: quietly getting a young colleague of ours into a Berkeley summer programme, pushing a Fogarty fellowship through for another. Sevagram, to him, was always a talent pool worth investing in, not just a place he happened to know one doctor.
Behind the papers ran a second, less visible correspondence — thousands of emails, mostly about tuberculosis. Madhu had started the QuantiFERON studies, QFT1 through QFT5, with MGIMS as a site, and for the better part of a decade that project ran like a second department between us: assays, ethics renewals, wire transfers from UCSF and CIHR that Madhu chased for months so that a small institute in Wardha could keep its funding. “At last,” he wrote me once, after three months of paperwork, “I have managed to get UCSF to prepare the MOU.” He was, in effect, running grant administration for MGIMS from three continents away, simply because he believed the work was worth the friction.
A visit home
In August 2023, Madhu came to Sevagram to deliver an oration on tuberculosis and to accept an award for it. He turned down the institute guest house without a second thought and stayed with us instead, and spent the evening talking to Bhavana, Ashwini, and Amrita with the kind of warmth only an old friend manages. On stage the next day he was, as ever, precise rather than sentimental: TB, he said, is still the world’s leading infectious killer, and in India it can take sixty days, three visits, and several rounds of the wrong antibiotic before anyone thinks to test for it. That gap, between what a doctor knows and what he does, was the whole of his argument. It was the same gap Madhu had spent twenty years trying to close in me, one workshop, one manuscript, one exasperated email at a time.
Ambitious, persistent — stubborn, Nitika would correct — and hardworking: that was how he once summed himself up in three words, and watching him hold that room at fifty-six the way he’d once held a floor of residents at thirty, it was hard to disagree with any of them.
He had first shown up in my life as a young man arguing about confidence intervals on a floor in Sevagram. He left it, that week in 2023, as a Canada Research Chair at McGill, chair of its Department of Global and Public Health, editor-in-chief of a journal — a long way, and, sitting in my own living room listening to him talk to my daughter, not very far at all.