Gallbladder cancer is ten times commoner in Delhi than in Chennai. Chronic pancreatitis here presents in the fourth decade and is mostly idiopathic. These are not the diseases international trials were designed around, and no amount of reading will tell us how they behave. A collaborative of Indian centres, building the prospective cohorts that can.
Two examples, from opposite ends of the hepatobiliary and pancreatic workload. Both are conditions where the Indian pattern differs sharply from the one the literature assumes.
A tenfold gradient across one country, and among the highest rates recorded anywhere. Whether and when to operate on gallstones in a woman from the Gangetic belt is therefore not the same question it is in Rotterdam.1
A patient in their late thirties, no drinking history, already diabetic, in pain. That description was published in 2008, from a survey that saw each patient once and followed none of them.2
This is not for want of effort. Indian units publish steadily, societies meet, and EPICAP-India is now screening 110,000 people across ten states to establish how common pancreatitis actually is.3 What is missing is something less glamorous and harder to sustain: standing prospective cohorts with a shared data dictionary, a defined follow-up schedule, and the same fields recorded the same way at every site, year after year.
Without that, every Indian question is answered by a single-centre retrospective series whose findings are inseparable from that centre's referral pattern, and which the next unit cannot pool with its own.
The epidemiology of these conditions are not well-known in India.EPICAP-India protocol, BMJ Open Gastroenterology 20243
What they needed was agreement on definitions and a shared place to put the data. Note the range: the same collaborative infrastructure supports a registry that never ends, a snapshot audit done in eight weeks, and a randomised trial.
Every cholecystectomy and ERCP in participating Swedish units, entered prospectively, indefinitely. No hypothesis at the outset.
Consecutive cholecystectomies across the UK and Ireland over two months, collected by surgical trainees in their own units. Case ascertainment 95.2%.
A restrictive strategy, requiring five specific pain criteria before cholecystectomy, against usual surgeon judgement.
Surgical drainage within six weeks against an endoscopy-first strategy, with pain measured on the Izbicki score and integrated over eighteen months.
Percutaneous drainage first, escalating to minimally invasive necrosectomy only if needed, against primary open necrosectomy.
The Netherlands has a population smaller than Karnataka. Sweden has fewer people than Delhi. Neither has an advantage over India in numbers. They agreed on definitions first.
A busy unit might see twenty-five patients with chronic pancreatitis, or forty gallbladder cancers, in a year. Ten such units make two hundred and fifty. That is the difference between a descriptive series and a confidence interval narrow enough to act on.
CholeS exists because one hospital's practice looks normal from inside it. A result from a single centre is inseparable from its referral pattern, its endoscopist and its theatre list. The same result across eight centres is a finding about the disease, which is why reviewers treat multicentre data differently.
Collaborations fail on definitions, not goodwill. If one centre records pain as present or absent and another uses the Izbicki score, the data will not pool, and no amount of enthusiasm afterwards will fix it. We do that work before recruitment opens: every field drawn from published common data elements, every permitted value fixed in advance, and the dictionary versioned so it can be revised without invalidating what came before.
Contribution is counted by the registry. The publication policy is agreed before the first analysis, and every contributing centre can see where it stands at any time.
The registry was built to hold more than one study, because rebuilding the infrastructure for each question is how collaboratives die. A new study is a new data dictionary on the same foundation, with the same centres, the same accounts and the same governance.
Prospective registry with baseline, intervention and follow-up. Aetiology, pain, exocrine and endocrine failure, endotherapy and surgery, and outcomes over time.
What happens to the incidentally detected stone in an Indian population, in a country carrying a tenth of the world's gallbladder cancer. A question Western cohorts cannot answer for us.
The infrastructure is the group's. Any member centre may propose a study; the steering committee decides what the collaborative takes on and in what order.
What every centre sees beyond its own data is the national picture as counts: recruitment by centre, and distributions across the collaborative, with any category of fewer than five patients suppressed so that an aggregate cannot narrow to a person. No centre can read another centre's records, and there is no query available to it that returns one.
Go to registry.iddsg.org and enter your institutional email. A one-time link arrives; there is no password. Your account is created pending and can see nothing at all until a principal investigator approves it.
Write to k.gautham@gmail.com with your name, centre and the email you signed up with. We assign your centre, grant access to the study, and set up your site code. Study identifiers then read CP-JIP-0001 onwards.
We supply the protocol, the full data dictionary, the consent template and the technical and security dossier. You can enter data from the day approval comes through.
