WorkflowMedical Affairs

Investigator-Initiated Study Review

Medical affairs KPIs for an IIS programme, and how to measure its ROI

An IIS programme's largest costs are not in its grants budget: the duplicate it funds, the study that never enrols, the weeks each concept waits and the hours its medical directors spend. Measure those by stage and by research priority and the ROI case has real numbers; here is the cost model, the twelve KPIs and how Investigator-Initiated Study Review is built to report them.

Updated · 12 min read

See one case, screen by screen →
Watch with transcript →

Medical affairs KPIs for an investigator-initiated studies (IIS) programme should count the evidence it buys, not the concepts it processes. The return on the programme is that evidence, per dollar and per reviewer hour: funded studies that enrol, report and answer a question the company's published research priorities ask. Medical affairs teams receive concepts from independent investigators who will sponsor and run the study themselves, while the company supplies drug, funding or both. A cross-functional committee decides which to support. The work is hard to value because its real costs sit outside the grants line, and no public benchmark exists to compare against. Botminds' Investigator-Initiated Study Review exists to make those costs visible: AI agents score each concept for priority fit, design, feasibility and overlap with existing studies, and build a cited packet for the people who decide.

So the ROI question is really four questions. How many funded studies duplicate work that was already under way? How many stall before they enrol? How long does each concept wait, and what does the wait cost? And how many medical director hours go into each decision? Count those, and "is the programme worth it" becomes "what does each answered question cost us".

What the public record says about the economics of investigator-initiated studies

25%

of 60 investigator-initiated oncology concepts approved by pharma sponsors; median 61 days to a decision

Gerber et al., 2025

18months

median from first submission to activation for the concepts that were approved

Gerber et al., 2025

26.5%

of investigator-sponsored trials stopped for poor recruitment, against 8.4% of industry-sponsored ones

Kasenda et al., 2014

0

fields in US Open Payments that mark a research payment as investigator-initiated

our query of 931,959 records, 2025

Gerber covers one 13-institution thoracic oncology consortium over 2014–2024; Kasenda covers 1,017 randomised trials approved by six ethics committees in 2000–2003.Source: Gerber et al., Clin Cancer Res 2025; Kasenda et al., JAMA 2014; CMS Open Payments research file

How to measure medical affairs value beyond activity counts

Measure what was bought, not what was handled: counts of concepts received, approved and funded measure throughput, not value. A programme can approve more concepts every year while funding duplicates, studies that never reach their sample size and studies that never report. The investigator-initiated sponsored research community has said so for years: the IISRA's metrics statement separates efficiency metrics such as "time from submission to review decision" from the value a programme should report, which it lists as publications, "data for potential future indications", supporting evidence for phase III design and expanded safety information (IISRA, 2011).

Nor can a head of medical affairs borrow a market figure for spend. US manufacturers reported $9.50 billion in research payments to Open Payments for 2025 (CMS), but most of that is company-sponsored trials, and our own query of the 931,959 research records found no field that identifies an investigator-initiated study. In the UK, R&D is 70% of disclosed transfers of value, £647.3 million in 2025 (ABPI, 2026), and under the EFPIA code it is disclosed only in aggregate. Every programme therefore has to build its own baseline. Approval rates do not travel either: in one oncology consortium the investigators' own peer review passed 88% of concepts, and the pharmaceutical sponsors then approved 25% of the same concepts (Gerber, 2025). A rate is a consequence of what arrives and who is judging it, not a measure of what was bought.

The upside is real, which is why the accounting matters. KEYNOTE-016, registered with Johns Hopkins as lead sponsor and Merck as collaborator (ClinicalTrials.gov NCT01876511), was one of five single-arm trials behind pembrolizumab's 2017 approval for MSI-high solid tumours, the first time the FDA approved a cancer treatment on a biomarker rather than the tumour's site of origin (FDA approval summary, 2019). Most IIS budgets fund nothing like that. The question for the programme owner is how much of the budget goes to studies that could, and how much leaks away before they start.

Four costs that never appear in the IIS budget

The hidden costs of an IIS programme are a funded duplicate, a study that stalls on enrolment, the time each concept waits, and the reviewer hours spent on concepts that should never have reached review. None appears as a budget line. All four can be counted.

The funded duplicate is the most expensive and the least visible. A concept that repeats a company phase 3b, another funded investigator study or a registered trial buys an answer that was coming anyway. It is also a compliance problem: the HHS Office of Inspector General lists research "that is unnecessarily duplicative" among the indicia of questionable research (OIG, 2003), and a Medical Affairs Professional Society presentation on running IIS programmes calls a "good tracking system necessary to identify overlap" (MAPS, 2018). How to run that check element by element is the subject of the overlap check.

"Indicia of questionable research include … research that is unnecessarily duplicative or is not needed by the manufacturer for any purpose other than the generation of business."

HHS-OIG Compliance Program Guidance for Pharmaceutical Manufacturers, 2003

The stalled study costs the payments made before it stopped, the drug shipped and the evidence it never produces. Investigator-sponsored trials were discontinued for poor recruitment at more than three times the rate of industry trials, and discontinued trials were more likely to stay unpublished, 55.1% against 33.6% (Kasenda, 2014). The usual cause is a forecast taken on trust; checking it against the site's own enrolment history is covered in the enrolment feasibility guide.

Waiting is a cost to the investigator, whose funding window and patient population do not wait, and to the company, whose question stays open. The pharma sponsors in Gerber's consortium took a median of 61 days to decide, and approved concepts took 18 months from first submission to activation. Concepts with shorter projected durations were more likely to be approved, which suggests sponsors already price time in, informally.

Reviewer time is the cost nobody books. The cross-functional committee the Medical Affairs Professional Society describes draws on medical affairs, clinical, regulatory, safety, legal and biostatistics, with clinical pharmacology and translational research joining as needed (MAPS, 2018). The scientific review itself is done by medical directors on top of a medical role. Every one of those hours is real, and each sits in another department's budget, uncounted.

A cost model you can fill in with your own numbers

Price each hidden cost as a count multiplied by a unit cost you already know, and leave the value of a day of waiting as your own judgement. The model below is deliberately blank where only your programme has the number. The illustrative fills come from a sample programme and from estimates, and are marked as such.

The hidden cost of an IIS programme, line by line

A four-line cost model for the hidden costs of an IIS programmeWaiting: concepts a year times days saved times your value of a day; illustrative 200 times 25 equals 5,000 concept-days a year. Funded duplicate: duplicates funded times grant plus drug plus review hours; illustrative sample concept asked for 448,000 dollars and overlapped the company's own phase 3b at 0.86. Stalled study: studies that stall times payments and drug released before the stall; illustrative sample site enrols 1.1 patients a month against 3.3 needed, 55 months instead of 18. Reviewer time: concepts reviewed times hours per concept times loaded hourly cost; time one real cycle. The cost per answered question is grants plus drug plus hidden cost, divided by funded studies that report results.1 · WAITINGconcepts a year × days saved× your value of a dayIllustrative: 200 × 25 =5,000 concept-days a year2 · FUNDED DUPLICATEduplicates funded ×(grant + drug + review hours)Sample concept: $448,000 asked,0.86 overlap with own phase 3b3 · STALLED STUDYstudies that stall × paymentsand drug released before itSample site: 1.1 a month, 3.3 needed,55 months, not 184 · REVIEWER TIMEconcepts reviewed × hours each× loaded hourly costYour number:time one real cycle firstCOST PER ANSWERED QUESTION(grants + drug + 1 + 2 + 3 + 4)÷ funded studies that report
Illustrative fills: 200 concepts a year is the midpoint of an estimated 150–250 for a mid-size company; 25 days is the gap between an estimated 52-day manual baseline and the solution's 27-day design target; the $448,000 concept and the 55-month projection come from fictional sample data. None is a customer result.

Two warnings about filling it in. First, do not import a cost of delay from drug development. A peer-reviewed Tufts Center study puts a day of delay at about $500,000 in lost prescription sales, averaged over 645 drugs launched since 2000, and the direct cost of a phase II or III trial at about $40,000 a day, from 409 trial budgets (Smith, DiMasi and Getz, Therapeutic Innovation & Regulatory Science, 2024). Both price a product's own development: the sales a marketed medicine forgoes, and the budget of a registration-stage trial. An IIS concept waiting for review is not a phase III programme waiting to file. Its delay rarely moves a launch date, and the study itself is sponsored and run by an outside investigator. Use your own figure, or leave the line in days and let the committee judge. Second, the stalled-study line depends on how you pay. Milestone payment, which a MAPS presentation describes as "typically 4 milestone payments", from contract execution through 50% and 100% enrolment to a final 10–15% on the manuscript or report, caps what a stalled study can cost (MAPS, 2018). A lump sum does not.

The bottom line of the model is the number worth reporting to a chief medical officer: total programme spend, hidden costs included, divided by the funded studies that actually report results. It falls when overlap is caught before funding, when infeasible studies are reshaped rather than approved as written, and when finished studies publish.

What KPIs should medical affairs track for an IIS programme?

Twelve, in four groups: three for speed, three for the quality of what is funded, three for compliance and three for the evidence that comes out. Each is a ratio or a duration that can be computed from timestamps and records the programme already keeps; none needs a survey.

The KPIs that show whether an IIS programme buys evidence

  1. Speed · Receipt to decision

    Median and 90th-percentile days from a complete submission to the signed decision. The median hides the concepts that wait three cycles.

  2. Speed · Days by stage

    Screening, scientific review, waiting for committee, letter. Shows whether the queue is the reviewers or the calendar.

  3. Speed · Approval to executed contract

    Days from decision to signature. At 29 academic sites the median to full execution was 91 days, sponsor-initiated and investigator-initiated contracts together.

  4. Quality · Overlaps caught before committee

    Concepts flagged as repeating a company, funded or registered study ÷ concepts reviewed. Track the misses too: duplicates found after funding.

  5. Quality · Decline and reshape reasons

    Share by reason: duplicate, outside priorities, not feasible, budget above fair market value. Tells you what to publish to investigators.

  6. Quality · Enrolment against forecast

    Actual patients a month ÷ forecast, per funded study, checked at the 50% enrolment milestone date.

  7. Compliance · Budgets above fair market value

    Lines above the benchmark range at submission, and the share approved with a written reason. Should trend to zero unexplained.

  8. Compliance · Commercial-involvement flags

    Concepts held for a sales or marketing link, and days to clear or hold. Zero flags in a year is a question, not a result.

  9. Compliance · Transparency completeness

    Research payments reported with the study name ÷ payments made, as 42 CFR 403.904 requires in the US.

  10. Evidence · Priority coverage

    Published research priorities with at least one funded or active study ÷ all priorities. The gaps are the agenda for the next cycle.

  11. Evidence · Results within a year

    Funded studies with results posted within one year of primary completion ÷ studies due, the deadline in both 42 CFR 11.44 and EU CTR Article 37.

  12. Evidence · Cost per answered question

    Total spend, hidden costs included ÷ funded studies that publish or post results. The single number for the business case.

Source: IISRA metrics statement, 2011; Kiriakis et al., Clin Transl Sci 2013; 42 CFR 403 subpart I; 42 CFR 11; Regulation (EU) 536/2014

The evidence measures are the ones most programmes do not keep, because the posting duty sits with the investigator-sponsor, not the company. The base rate is low. Only 40.9% of US trials due to report did so within the one-year deadline, and industry sponsors were 3.08 times as likely to comply as non-industry sponsors (DeVito et al., 2020). By our own count on ClinicalTrials.gov, completed phase 2–4 trials led by an academic sponsor with an industry collaborator posted results 54.9% of the time, against 68.5% for industry-led trials and 32.1% for academic trials with no industry collaborator (ClinicalTrials.gov API, 2015–2022 completions). And evidence takes time to arrive: across 691 trials in the IMPACT cohort, the median time to any publication was 4.07 years, and 26% reached a clinical guideline (Blümle et al., 2021). Report evidence KPIs on a rolling multi-year basis, or they will always look worse than the speed ones.

The quality measures need the most care, because they depend on how good your checks are. A programme that reports "no duplicates caught" may have none, or may not be looking. How to test an overlap finder's recall and a scorer's agreement with your committee is set out in measuring AI review accuracy.

Where the days go: receipt to decision, measured by stage

Measured by stage, the largest block of receipt-to-decision time is often waiting for the committee, not reviewing. That is why the stage split matters more than the total: a programme that only reports the median cannot tell whether to hire a reviewer or move the committee date.

Take a sample programme, a fictional mid-size company's fourth-quarter cycle, as it runs through Investigator-Initiated Study Review. The solution's design target is to cut the median from receipt to decision roughly in half, from an estimated 52 days by hand to 27. The split below shows where that saving is meant to come from.

Where the days go: Investigator-Initiated Study Review's target, by stage

By hand, sample programme (estimated baseline)52 days
Screening6
Scientific review18
Waiting for committee21
Letter7
  • Screening6
  • Scientific review18
  • Waiting for committee21
  • Letter7
With the agents (design target)27 days
Screening2
Scientific review9
Waiting for committee14
Letter2
  • Screening2
  • Scientific review9
  • Waiting for committee14
  • Letter2
Published: pharma sponsors, submission to decision61 days
Sponsor decision (median, IQR 31–183)61
  • Sponsor decision (median, IQR 31–183)61
The first two lanes are a design target split by stage on fictional sample data, not a customer result. The third is the published median for 60 thoracic oncology concepts submitted to pharmaceutical sponsors.Source: Gerber et al., Clin Cancer Res 2025

Read the lanes as a finance partner would. Screening falls because the agents read the concept form, budget sheet and cover email into a structured record and name what is missing, so the programme manager stops chasing. Scientific review falls because the medical director opens a packet already scored and searched, every finding linked to the passage it came from. Waiting for committee falls least, because the committee calendar does not move; it falls at all only because packets are ready before the docket closes. If your own stage split shows waiting as the largest block, the cheapest saving may be a rolling decision for low-risk concepts rather than any tool. The cycle itself is followed concept by concept in the one-cycle case study.

Evidence coverage: which research priorities your funded studies answer

Evidence coverage is the count, per published research priority, of company studies, funded investigator studies and concepts in the current cycle. It is the KPI that turns an IIS budget into a portfolio: it shows which priorities are crowded, which are covered and which have no study at all.

Evidence coverage in Investigator-Initiated Study Review: covered, crowded, open gap

A research priorities screen for a Q4 2026 review cycle. Each published priority, grouped by therapy area, lists the company studies, funded investigator studies, public registry studies and this cycle's concepts against it, and is labelled Covered, Crowded or Open gap. A side panel shows the committee date with 4 of 5 packets ready and funding for the second half of 2026 by therapy area against a 6.4 million dollar envelope.

Look at the crowded rows: each has a concept marked high overlap. Crowding is where duplicate spend comes from; open gaps are where the budget is not reaching.

Fictional sample data: 13 published priorities, 14 concepts this cycle, a $6.4M envelope for the half-year. One oncology priority is an open gap, with no concept and no funded study.

Coverage also tells the field team what to say. PhRMA's principles hold that companies "should not solicit research projects" from investigators, but "may publicly disclose the general areas or types of studies that it does or does not support" (PhRMA, 2014). An open gap is exactly such an area. Publishing it, and letting medical science liaisons share the published list in scientific exchange, is the compliant way to steer concepts toward unanswered questions. Measuring coverage each cycle shows whether that worked. Where the line between sharing an area and soliciting a study falls is covered in the compliance and fair market value guide.

What a programme dashboard should show, and what Investigator-Initiated Study Review reports

A programme dashboard for an IIS committee needs four panels: concepts received and decided over time, days by stage, the reasons concepts were declined or reshaped, and evidence coverage by priority. Investigator-Initiated Study Review builds these from the work itself: each concept's timestamps, the agents' findings, the reviewers' signed recommendations and the committee's decisions.

The raw material of the dashboard: every concept, its stage and its findings

The Proposals screen of Investigator-Initiated Study Review for a Q4 2026 cycle, with tabs Open 10, Assigned to me 3, On docket 4, Decided 4 and All 14. Ten rows, each showing the proposal name and reference, the investigator and site, a research priority code, a fit score, an overlap label (high, some or low), the budget and patient count, a stage bar and a day count. One concept, TYK-CLE, carries a Compliance hold label and a fit of 0.42 against a priority marked Outside.

Read the Overlap column: three of the ten open concepts are marked high overlap.

Fictional sample data: the proposals list for one sample review cycle, 10 of 14 concepts still open. Each row carries the stage the concept has reached, a day count, its priority fit and the overlap the agents found: the timestamps and findings the dashboard's panels are counted from.

The dashboard covers the speed and quality measures and evidence coverage up to the decision. The solution's journey ends when the letter is sent, at which point it updates the drug supply forecast and the transparency record. Enrolment against forecast, milestone payments and results posting happen after that, in the systems that manage contracts and studies, so the post-award KPIs in the list above still need their own source. A business case should say so rather than imply one screen answers every question.

Building the business case for IIS review software

A credible business case for IIS review software rests on your own baseline, the cost model above with your unit costs, and a measured comparison on a real cycle. Vendor percentages, including ours, are targets until your programme has measured them.

How long that side-by-side cycle takes to set up, and what the alternatives cost to run, is the subject of the go-live comparison. The design that makes each measure countable is on the Investigator-Initiated Study Review page.

Questions people ask

What is a good approval rate for an IIS programme?

There is no public benchmark. The best peer-reviewed figure, 25% of concepts approved by pharmaceutical sponsors in one thoracic oncology consortium, covers one therapy area and one group of investigators, and no large company publishes its own rate. Track decline reasons instead: a falling share of declines for duplication or for falling outside published priorities means investigators are being steered better.

How much does the industry spend on investigator-initiated studies?

Nobody can say from public data. US Open Payments reports $9.50 billion of research payments for 2025 but has no field that marks a study as investigator-initiated, and European codes disclose R&D only in aggregate. Your own programme's spend, measured against the evidence it produced, is the only number that means anything.

Should medical affairs use the Tufts cost of a day of delay for IIS?

No. The Tufts figures, about $500,000 a day in lost prescription sales and about $40,000 a day in direct costs for a phase II or III trial, price a product's own development, where a late day can move a launch. A concept waiting for IIS review rarely does. Keep waiting in days, or set an internal value per concept-day that your finance partner agrees with.

Should concepts sourced by medical science liaisons be a KPI?

Count them for context, but never set a target. PhRMA's principles say companies "should not solicit research projects, study descriptions, or protocols" from investigators, and a quota for MSL-sourced concepts rewards exactly that. The useful measure is the quality of what arrives, such as the share of concepts that fit a published priority and pass the compliance screen, not its volume.

Which single KPI should a head of medical affairs report?

Cost per answered question: total spend, hidden costs included, divided by funded studies that publish or post results. It moves only when the programme funds fewer duplicates, fewer infeasible studies and more studies that report, which are the three things an IIS budget should buy.

Investigator-Initiated Study Review.
See it on your own files.

An agentic solution from Botminds. Scores every investigator's study concept on priority fit, scientific quality, feasibility and overlap with existing studies, ranks the cycle on one scorecard, and builds a cited packet for the committee to decide and sign. Book a demo, or explore how it works.