The map hidden in the sales ledger: what India’s medicine sales could show about the patients nobody is treating

Drug companies know, town by town and month by month, where their blood-pressure pills sell. Read the other way, the same ledger shows where they don’t, and in India that is often where people with high blood pressure are going untreated. Governments, NGOs and CSR funders could use that map. Someone has to decide who holds it.

By Ishan Kashyap, Founder & Editor · 6 November 2026 · 17 min read

About this essay. The people in the opening and closing scenes are composites. Every number is real and sourced at the end. The district map is our own calculation from published survey data, and where we infer rather than report, we say so.

A quarterly review in Mumbai

On the last Thursday of the quarter, a brand manager we will call Neha opens the review deck for her company’s biggest blood-pressure brand. Slide fourteen is the one her director always asks about. It ranks every district in her region by sales per thousand adults, against what the district “should” sell given its population and income. Marketers call the gap white space.

At the bottom of the ranking sit three districts in the east of her territory. Sales there are a fraction of the regional average and have barely moved in two years. The recommendation on the slide is the sensible commercial one: low potential, deprioritise. Two medical representatives will move to the state capital, where doctors write more prescriptions and chemists stock more brands.

Neha has just found something more important than a weak market, and her deck has no way to say so. A district where blood-pressure medicine does not sell is either a district where few people have high blood pressure, or one where many people have it and nobody is treating them. Her ledger cannot tell the two apart on its own. Put it next to a health survey, and it can.

A pharma company’s white space is a public-health blind spot. They are the same rows in the same spreadsheet.

The disease that leaves no receipt

High blood pressure is the clearest test of this idea, because it is common, cheap to treat and almost silent. Most people who have it feel nothing until a stroke or heart attack. The largest study of the problem in India, published in JAMA Network Open in 2023, used the fifth National Family Health Survey to measure the blood pressure of about 1.7 million adults in 2019–21. It found that 28.1% had hypertension. Of those, 36.9% had been diagnosed. Of the diagnosed, 44.7% were taking medicine, which means only 17.7% of everyone with hypertension was being treated. Only 8.5% had it under control.

The World Health Organization’s first global report on hypertension, in 2023, put the number of Indian adults aged 30 to 79 with the condition at 188.3 million, of whom 37% had been diagnosed and only about 15% had it under control. It estimated that India would need to treat another 67 million people effectively to bring half of all patients under control.

The government is working hard on the problem. Its “75/25” initiative, launched in May 2023, aimed to put 7.5 crore people with hypertension or diabetes on standard care by the end of 2025. By March 2025, the health ministry told Parliament, 4.2 crore were being treated for hypertension and 2.5 crore for diabetes, close to 90% of the target.

The trouble is the people outside those numbers. A patient on treatment leaves a trail: a clinic visit, a prescription, a strip of tablets bought every month. An untreated patient leaves nothing at all. You cannot count them from the records of a system they never entered. You can only find them by noticing the transactions that should have happened and didn’t.

What the ledger already knows

Noticing missing transactions is what a sales team does for a living. India’s retail pharmaceutical market was worth ₹2,40,672 crore in 2025, according to Pharmarack, which audits what chemists buy and sell. Heart and blood-pressure medicines were its largest growth engine: the cardiac segment was worth about ₹33,115 crore in the year to December 2025, and antihypertensives made up roughly half of that. By our arithmetic, Indians spend something like ₹16,000 crore a year in pharmacies on blood-pressure pills alone.

Every one of those sales is recorded at least twice: once by the stockist who sells to the chemist, and again by the auditors who sample chemists across the country. Large companies combine the two into monthly views of every molecule, brand and town they sell in. They use that view for three things that matter here.

• Demand sensing: what is selling, where, this month, compared with last year.
• White-space analysis: where a category sells far less than population, income and doctor numbers predict.
• Lapse tracking: where repeat purchases fall away, which in a chronic medicine usually means patients have stopped taking it.

Each answers a commercial question. Turned around, each answers a public-health one. Where is the category under-selling? Where are patients undiagnosed or untreated? Where do repeat purchases fall away? Where are patients giving up on treatment? Pharma companies do not set out to collect public-health intelligence. They collect it anyway, as a by-product of selling, and almost nobody outside marketing ever reads it that way.

Need against demand: the quadrant

The idea at the centre of this essay is simple enough to draw. Plot every district twice: once for how much a disease is there (need), and once for how much treatment for it is being bought or dispensed (demand). Four kinds of district appear, and each calls for a different response from a different owner.

The quadrant is not new as a shape. Every brand team has a version, usually labelled “invest”, “defend”, “harvest” and “ignore”. What is new is the question it is asked. A brand team asks where the next rupee of sales will come from. A health funder would ask where the next rupee of help will do the most good. The bottom-right corner is unattractive to the first and the whole point for the second.

Drawing the map with public data

You do not need any company’s sales data to see that the map is worth drawing. The authors of the JAMA Network Open study published their district estimates, and they include a stand-in for demand: the share of people with hypertension who say they take medicine for it. We plotted all 707 districts in the survey against two medians: 27.7% of adults with hypertension, and 19.5% of them on treatment.

Three things stand out.

Need and care barely track each other. If treatment followed disease, the dots would slope upwards: more patients, more prescriptions. They don’t. Across the 707 districts, the correlation between the two is about 0.1, which is close to none. Where people are sick and where they are treated are almost separate questions.

The underserved corner is crowded. 161 districts sit in the bottom-right box, above the median for hypertension and below it for treatment. In 98 districts, fewer than one in ten people with high blood pressure take anything for it. Amritsar has one of the highest rates of hypertension in the country, at 47% of adults, and only 15% of them are treated. In Mirzapur in Uttar Pradesh it is 40% and 8%. In Anjaw, in Arunachal Pradesh, 44% and 5%.

The worst gaps hide inside good state averages. Funders tend to think in states. The map says they should think in districts. The same state can contain some of the best-served and worst-served places in India.

In Gujarat, 46% of people with hypertension in Ahmedabad take medicine for it, against 6% in The Dangs. In Odisha the range runs from 37% in Khordha, home to Bhubaneswar, to 9% in Koraput. In Maharashtra, from 40% in Mumbai Suburban to 15% in Nandurbar. Many of the lowest-ranked districts, including The Dangs, Nandurbar, Koraput and Alirajpur in Madhya Pradesh, have large tribal populations. The study’s data show the same pattern nationally: only 24% of people with hypertension from Scheduled Tribes had been diagnosed, against 40% in the general category.

The study’s authors concluded that local data are critical for planning and monitoring hypertension care. We agree. The difficulty is that this map is already five years old. The fieldwork ran from 2019 to 2021. The sixth survey’s fact sheets came out in May 2026, but analyses of this kind take years to follow, and a district that started a screening drive last spring will not see its effect in a national survey until the end of the decade. A sales ledger refreshes every month.

Where the ledger would sharpen the map

Put pharmacy sales next to the survey, and the map gains three things it does not have today.

Time. A survey every five years cannot tell a district officer whether last quarter’s screening camps turned into patients who keep buying their tablets. Monthly sales can, at least roughly. A rise in new prescriptions for amlodipine or telmisartan after a screening drive is a sign the drive found people. A rise that fades after three months is a sign they were not kept on treatment.

Detail. A survey of about 2,400 adults per district can just about rank districts. Sales are recorded by town and by chemist, which in principle allows a view below the district, down to the blocks and market towns where help would actually be delivered.

Breadth. The same method works for any chronic disease with a medicine and a survey estimate: diabetes, asthma, thyroid disease, and conditions like depression whose treatment gap is even larger and harder to see.

Where the ledger would mislead

The ledger is a proxy, and proxies fail in predictable ways. Anyone who builds this map has to plan for at least four.

Free medicine leaves no sale. Ayushman Arogya Mandirs, India’s upgraded primary health centres, are meant to give more than 170 essential medicines free, including those for blood pressure. Tamil Nadu goes further: its Makkalai Thedi Maruthuvam programme, launched in 2021, takes blood-pressure and diabetes medicines to patients’ homes, and the state says it had reached two crore people by December 2024. In a district like that, low pharmacy sales may mean the state is doing its job, not that patients are neglected. The same is true, in part, of more than 20,000 Jan Aushadhi generic-medicine stores, whose sales the government records but private audits may not. The map needs public dispensing data alongside private sales, or it will send help to the places that need it least.

Sales are booked where the shop is, not where the patient lives. A villager who buys tablets in the district town shows up in the town. Patients cross district and state lines for doctors and cheaper stockists. That smears the map at its edges, and favours aggregating to the level where it can be trusted.

A sale is not a patient. One patient may buy a month’s supply, another a week’s. Doctors switch brands, combine molecules and hand out samples. Converting sales into “patients on treatment” takes assumptions about dose and duration, and those assumptions should be published.

Signals drift. The best-known failure of this kind of data is Google Flu Trends, which estimated flu from search queries. Between August 2011 and September 2013, it overestimated flu in the United States in 100 of 108 weeks, because people’s searching habits, and Google’s own search engine, changed underneath the model. Sales change too: a price cut, a new combination pill or a distributor strike can move them without any change in disease. The answer is the one the flu researchers gave. Commercial data should complement surveys, not replace them, and the model should be recalibrated against every new survey round.

It has been done elsewhere

Reading commercial pharmacy data for public health is not a thought experiment. In the United States, the Centers for Disease Control and Prevention publish county-by-county maps of opioid prescribing built on IQVIA’s retail prescription data, which come from about 54,600 pharmacies. Those maps helped show where the opioid epidemic was being fed, years before overdose statistics caught up. In England, the NHS publishes anonymised data every month on what every general practice prescribes. An Oxford team turned it into OpenPrescribing, a free tool that lets anyone compare practices and spot unusual prescribing, and it is now funded by NHS England.

Neither example is a perfect fit. The American data are bought by the government from a commercial supplier, and English prescriptions come from a single public payer. India has neither arrangement. We found no public evidence that commercial pharmacy audits are shared with Indian health authorities for this purpose. That is the gap, and it is an organisational gap, not a technical one.

Follow the money

The map matters most to the people who decide where help goes, and Indian health money still moves largely by habit and headquarters. Companies spent ₹34,908.75 crore on corporate social responsibility in 2023–24, of which ₹7,150.81 crore, about a fifth, went to health. Projects in Maharashtra received ₹6,065.95 crore of the total, about 17%. Uttar Pradesh, home to about one in six Indians, received ₹1,545 crore, or 4.4%. Bihar received ₹260.53 crore. In 2022–23, the 112 aspirational districts that the government has picked out as the least developed received ₹1,402.89 crore, under 5% of the total.

The law encourages this. The Companies Act asks firms to “give preference to the local area and areas around it where it operates”, though the corporate affairs ministry has said the preference is directory, not mandatory. Plants and offices are where they are, and CSR follows them.

Lay that spending next to the hypertension map, and the mismatch is stark. Uttar Pradesh contains 37 of the 161 underserved districts, nearly a quarter. Maharashtra contains one. Madhya Pradesh, Chhattisgarh and Arunachal Pradesh together contain another 55. A CSR head who wants to fund hypertension care where it is missing would find the answer in this map, not in the company’s plant list. With sales data refreshed every month, she could also see whether her money had moved anything.

Who should hold the map

The most important question is not technical. Sales data are commercially sensitive, health data are personal, and a map of where poor people go untreated is exactly what an aggressive marketer would like to have. If the map is worth drawing, its governance has to be designed before the first file changes hands. We would build it on five principles.

What it means for you

Why it could go wrong

The map could be used to sell, not to help. A list of districts where many people are untreated is also a list of untapped markets. That is not entirely bad, since more pharmacies stocking cheap generics would help, but a commercial push of expensive branded combinations into poor districts would raise costs for the people least able to pay. The firewall in principle three is the safeguard, and it needs teeth.

The map could replace the people who should check it. A dashboard is cheaper than a household survey and much cheaper than a health worker. If funders start trusting the sales map instead of the survey and the ASHA, the Google Flu Trends lesson will be relearned at patients’ expense.

The map could punish the honest. Districts that report their dispensing carefully will look better served than districts that don’t. Unless public data are complete, the map will reward poor record-keeping with attention it may not deserve, and the reverse.

The map could stop at diagnosis. Finding untreated patients is the easy half. Keeping them on treatment for decades is the hard half, and it needs clinics, supplies and people. A map that moves money to the right districts but not into follow-up will simply produce a new crop of patients who start treatment and stop.

What should happen next

RavenArc tests every decision against six questions. For a company board or a CSR committee deciding whether to contribute to a map like this, they make a short agenda.

The same review, a year later

Run the quarter again. Neha’s deck still has slide fourteen, and the same three districts are still near the bottom. This time there is a second slide behind it, drawn from the shared map. Two of the three districts sit in the underserved corner: high blood pressure is common there, and treatment is rare. The third turns out to be well served after all, by a state scheme that hands out free tablets her ledger never saw.

Nothing about Neha’s job has changed. She still has to sell. But her company’s aggregated figures now feed a monthly map that a district health officer, a CSR committee in Mumbai and an NGO in the state capital all read. One of them has already moved a screening programme into the two underserved districts. In six months, the map will show whether the people it found are still buying their tablets.

India already keeps a monthly record of where its medicines sell. The map of where they don’t is in the same ledger. Someone only has to read it the other way.

RavenArc tests every decision against six questions. See the RavenArc Decision Method.