Case study: Formulary Pull-Through When the Coverage Already Exists

On 13 August 2026, according to its page metadata, Indegene published a case study on a nephrology formulary programme. A specialty pharmaceutical company had broad payer coverage for its therapy. Many physicians underestimated it. The vendor describes the problem as "the gap between actual formulary access and physician perception", and the fix as a segmented email and direct-mail campaign to about 5,000 healthcare professionals.

That makes it a clean formulary pull-through case study, and an unusual one. Formulary insights products, such as MMIT's, are built to find the restrictions that block a drug. This programme assumed the restrictions were mostly absent and went after a belief instead. The page reports a 3.9% increase in prescribing and a 4:1 return on investment. This post separates what was measured from what was claimed, and sets out the US rules that bind formulary messages to prescribers.

The situation: broad coverage, and prescribers who did not believe it

The customer is unnamed. The page describes a specialty pharmaceutical company with a therapy used in nephrology. It gives no product, no country and no dates for the programme itself. Its vocabulary of health plans, payer coverage and formulary placement is the US market's, and this post reads it that way.

According to Indegene, three things were going wrong at once:

• Belief, not coverage. Many physicians thought patient access was more limited than it was, so they hesitated to prescribe.
• Coverage was local. It varied across health plans and geographies, so a national "widely covered" message would have been wrong for some physicians.
• No new field force. The client wanted a data-driven fix that did not mean hiring more representatives.

Hold that diagnosis up against MMIT's oncology case, which sits in our library as a midsize oncology company using IDN formulary insights. The symptom was identical: favourable formulary placement with most major payers and low adoption. The cause was different. MMIT reports that the brand was off formulary at three integrated delivery networks (IDNs). At four more it was listed but held back by competing order sets, separate patient pathways in the electronic record and extra lab testing. Telling those prescribers that coverage was good would have been true at the payer level and misleading at the point of care.

So the first question about the nephrology programme in our library is not how to copy it. It is which of the two problems you have.

What was deployed

BioPharm Communications, which Indegene describes as one of its companies, built the programme. The page lists five components.

The page never mentions artificial intelligence, models or algorithms. This is segmentation, templated personalisation and campaign analytics. The expensive word is localised. Every combination of cohort, geography and plan mix is, in regulatory terms, a separate promotional piece. That is where AI tools are now sold into this work, and where the review burden lands. Our pieces on agentic MLR review and on the recall an MLR agent actually delivers cover that side.

What it achieved, and what the numbers measure

The page reports four figures. All four are the vendor's, and none comes with a method.

The last two figures need care. If 2,794 engagements came from 2,794 different physicians, that is 56% of 5,000, which matches the 50%+ figure. But the same vendor counts engagements as events elsewhere. In a sibling Indegene launch case, also in our library as a launch programme for low-access territories, it reports 51,583 engagements across 38,202 targeted HCPs. That is more engagements than physicians. If the nephrology figures use the same convention, the claim that more than half engaged is only true if almost every engaged physician engaged exactly once. The page does not say which convention applies.

The 3.9% is the number that matters, and it is the one with least support. A rise in prescribing over a campaign period can come from market growth, a payer adding the drug, a competitor's supply problem or seasonality. Only a comparison group separates those from the campaign. Nothing on the page says one existed.

Attribution looks like a cluster-randomised trial published in JAMA Internal Medicine in 2022. Practices in one academic health system were randomised to show prescribers patient-specific costs and cheaper alternatives at the point of prescribing. Out-of-pocket costs fell by an adjusted 11.2% against control practices. The authors also stated the limit: only 4.2% of 867,757 prescriptions had an alternative to recommend. The intervention differs, but the design is the point. Randomisation, a named metric and a stated denominator make a number portable.

In fairness to Indegene, 3.9% is a modest claim. It is more plausible than the 38.4% prescription lift the same vendor reports for its launch programme. MMIT's 17% utilisation gain over six months is a vendor number with no stated control too. None of these is evidence of failure. They are simply not evidence of size.

What made it work, and what limits it

The logic is sound where its premise holds. A physician who expects a patient to be refused may not write the prescription at all. Correcting that belief with plan-specific facts costs little and needs no field visit. The programme was also aimed at a narrow barrier rather than at general promotion, which the page stresses.

The limits follow from the same logic.

• Wrong diagnosis, wrong message. Where access really is restricted, as at MMIT's IDNs, a message saying the drug is covered sets up a rejected claim or an abandoned prescription, and risks teaching the physician to distrust the next mailing.
• Coverage data go stale. Formularies, tiers and utilisation management change during a plan year. A printed piece outlives the data it was built on. Every coverage statement needs a source and an "as of" date.
• Prescribers now have another route to the same facts. Since 1 January 2021, Medicare Part D plans must offer prescribers at least one real-time benefit tool giving patient-specific formulary and benefit information. From 1 January 2027 those tools must use the NCPDP Real-Time Prescription Benefit standard, version 13. For Part D patients, the case for mailing coverage claims to prescribers weakens as point-of-prescribing tools spread.
• Variants multiply review. Localisation is where generative tools can save production time and add review time. Each variant can drift from approved wording, which is what OPDP cited in a July 2026 letter about HCP emails.

The rules it had to satisfy: what binds formulary messages to HCPs

The page does not discuss compliance. These are the US rules that apply to a programme of this shape, position current at 12 October 2026.

The payor guidance is the wrong shelter. Teams sometimes treat formulary content as access information rather than promotion. FDA's 2018 guidance says plainly that it does not apply to "health care providers who are making individual patient prescribing decisions". The June 2026 revised draft, which updates the guidance for the Consolidated Appropriations Act, 2023, repeats the same exclusion. A formulary message to a prescriber is promotional labeling. It is submitted on Form FDA 2253 like any other piece.

On-label consistency still applies. The programme's assets put formulary access alongside clinical evidence. Any clinical claim that goes beyond the prescribing information is judged by the three-factor test in FDA's consistent-with-labeling guidance. That guidance is explicit: if a communication "fails to satisfy any one of these factors", it is not consistent with the labeling.

OPDP reads HCP emails closely. OPDP's stated job is to make sure prescription drug promotion is truthful, balanced and accurately communicated. Its untitled letter to Sanofi on 15 July 2026 concerned HCP emails for Beyfortus. OPDP found that the shorthand "RSV disease" gave a misleading impression of an indication limited to lower respiratory tract disease, even though the emails also stated the full indication. Short, localised variants are where shorthand creeps in. Our post on net impression and AI-generated content explains why claim-by-claim checking misses this.

Affordability content carries anti-kickback risk. The page says the assets included affordability resources. OIG's 2014 bulletin says copay coupons "constitute remuneration offered to consumers to induce the purchase" of a drug. Where a federal programme pays, the anti-kickback statute is implicated. A message that implies a coupon works for a Medicare Part D patient is a compliance problem as well as a promotional one.

Email has its own law. The FTC's CAN-SPAM compliance guide states: "The law makes no exception for business-to-business email."

What does not bind. States have limited room to stop segmentation on prescribing history. In Sorrell v. IMS Health, decided 23 June 2011, the Supreme Court struck down Vermont's law barring the use of prescriber-identifying information for marketing without the prescriber's consent. No FDA rule addresses AI used to generate or personalise promotional content. The ordinary promotional rules apply to whatever the tool produces.

What this means in practice

For a brand team that wants this result, the steps run in this order, longest lead time first.

1. Diagnose the barrier before you buy a channel. Compare prescribing against coverage at plan and health-system level. Where coverage is broad and uptake is low, check for IDN formulary status, order sets and testing rules first. A short survey of a sample of target physicians tells you whether the gap is belief or restriction. If it is restriction, you need account work, not a mailing.
2. Build a coverage file you can defend. One source per plan, an "as of" date on every statement, and a refresh cadence agreed with market access. Decide in advance what happens to live pieces when a plan changes its tier or utilisation rules.
3. Design the measurement before the first send. Randomly hold out a share of physicians or territories. Fix the metric, such as new patient starts, and the window. Agree the return-on-investment formula with finance, including agency, print and postage costs. Report unique physicians reached separately from engagement events.
4. Plan the review load, not just the content. Count the variants your segmentation will create. Each goes through MLR and onto Form FDA 2253 at first use. Settle the affordability wording for government-insured patients once, centrally, and make every vendor honour email opt-outs.
5. Read the result against the holdout. If the difference is small, that is still an answer. It tells you the barrier was elsewhere, which is worth knowing before the next launch.

The nephrology programme may well have worked. The page does not contain the evidence that it did, and the cheapest moment to collect that evidence is before the first email goes out.