Home › About
About TruePositive
A small, senior consultancy built around a single idea: that the value of evidence lies in whether it survives being checked.
The name
Why TruePositive
In diagnostics and in statistics, a true positive is a signal that is really there — as opposed to a false positive, which looks identical until someone checks. Health evidence produces a great many of both, and the difference between them is rarely visible in the result itself. It is visible in the design, in what was pre-specified, and in what happened when the finding was stress-tested.
That is the whole premise of the company. We would rather tell a client early that the data cannot support the claim they wanted than help them build something that falls over in assessment — which is slower, more expensive, and considerably more embarrassing.
How we work
Small on purpose
TruePositive is deliberately not a pyramid. The person who scopes an engagement is the person who does the work. There is no proposal team handing over to a delivery team, and no layer of account management between a question and the person who can answer it.
The trade-off is honest: we cannot staff twelve workstreams at once, and we say so when a project is bigger than the right shape for us. For work where depth matters more than headcount — a difficult estimand, a contested comparator, an analysis that has to be defensible line by line — it is the better arrangement, and it is the work we take.
Where a project needs more hands, we work alongside a client's internal team, or as a named specialist under a CRO or consultancy's contract, rather than sub-contracting to people the client never meets.
Principles
Four things we hold to
Pre-specify, then analyse
The plan is written and dated before the outcome data is seen. It is the cheapest available protection against a result that cannot be defended later.
Attack your own finding
Sensitivity analyses, negative controls and quantified bias analysis are part of the work, not an appendix. If a conclusion is fragile, we would rather find out first.
Reproducible by default
Analysis code is organised so a third party can re-run it from raw data to final table. It is what makes a result checkable a year later.
Say what the data cannot do
Limitations stated plainly, in the report and in the room. Assessors find them anyway; finding them first is what credibility is made of.
Who you will be working with
Klaus Kaae Andersen
Founder and principal consultant
Twenty-five years at the point where statistical method meets a health decision — from both sides of it. A PhD in statistics from the Technical University of Denmark, where he taught for a decade and was named Teacher of the Year; nine years leading the Statistics and Pharmacoepidemiology unit at the Danish Cancer Society Research Center, a group of fourteen researchers; then Medical Evidence Scientific Lead at AstraZeneca Nordics, working with medical affairs, health economics and market access on evidence generation. In 2022 he founded his own biostatistical consultancy and grew it to acquisition in 2024.
The research record is 164 peer-reviewed papers — including in Nature, JAMA, The Lancet Oncology and the Journal of the National Cancer Institute — and an h-index of 58. Most of it runs through Danish and Nordic national registers: cancer epidemiology and survivorship, pharmacoepidemiology and confounding by indication, stroke, and the methodological problems that come with all of them — competing risks, matched cohort survival, register linkage, and the difference between a correlation a register can show you and an effect it cannot.
Teaching has run alongside the research throughout: course lead at DTU across statistics, design of experiments, time series and quality control, research practice for medical doctors at the University of Copenhagen, director of the Nordic summer school in cancer epidemiology, and advisor to nine PhD students. That continues as TruePositive's course catalogue — real-world evidence for HTA, observational study design, analysing health data in secure research environments such as the NIH All of Us Researcher Workbench, and the responsible use of AI assistants in regulated settings.
Selected credentials
| Education | PhD, Statistics, Technical University of Denmark (2001); MSc, Mathematical Modelling, DTU (1997) |
|---|---|
| Research | 164 peer-reviewed publications, 18 as first author and 22 as last; h-index 58; over 100 international conference presentations |
| Previously | Medical Evidence Scientific Lead, AstraZeneca Nordics (2019–2021); Chief Statistician and Head of Unit, Statistics and Pharmacoepidemiology, Danish Cancer Society Research Center (2010–2019) |
| Academic | Associate professor, IMM, DTU (2004–2010); external associate professor, Faculty of Pharmaceutical Sciences, University of Copenhagen (2004–2019); Head of the DTU Statistical Consultant Center (2008–2010) |
| Service | Associate Editor, Journal of Cancer Survivorship; board member, Danish Society for Pharmacoepidemiology; board member, Danish Society of Theoretical Statistics (2008–2012) |
| Supervision | Advisor to nine PhD students and sixteen master's students |
| Recognition | Teacher of the Year, Technical University of Denmark, bachelor and master programmes (2004) |
- Based in Copenhagen, Denmark — working across the Nordics and Europe
- Languages English and Danish
Company details
The formal bits
| Based in | Copenhagen, Denmark |
|---|---|
| contact@truepositive.dk | |
| Working languages | English and Danish |
| Invoicing | Danish-registered; VAT handled according to your location and status |
| Contracting | Mutual NDA before confidential discussion; comfortable working under client SOW, framework or consultancy agreements |
Start with a conversation
No pitch deck. Describe the decision that has to hold up, and you will get a straight view on whether it is answerable and what it would take.