Clinical trials · Real-world evidence · Epidemiology
Frameshift designs and runs studies for pharmaceutical, biotech and research organisations: clinical trials, registry-based real-world evidence, and the epidemiological and statistical work underneath both. Studies in the Danish national registers are our deepest area.
Capabilities
Three areas of work. The data source changes between them; the standard for designing a study does not.
The statistician's half of a trial.
We write the analysis plan, run the primary and secondary analyses, and draft the methods and results for the paper.
Trial work →We work inside the approved Danish data environment ourselves.
We run six recurring study types there, target trial emulation among them, on registers that cover the whole population.
Why Danish registry data →The methods work underneath both.
We design cohort and case-control studies, run survival, competing-risks and causal-inference analyses, and work in statistical genetics and prediction modelling.
Selected work →The six questions clients bring us most often.
Establish the size and shape of the patient population.
We quantify incidence, prevalence, survival, disease trajectories and healthcare use across the whole national population carrying the diagnosis, so the figures describe routine care.
What actually happens after diagnosis.
We reconstruct treatment initiation, sequencing, switching, adherence and persistence from prescription and hospital records, and show the pathways patients follow over years of routine care.
Bring trial-level design discipline to real-world data.
We define the trial we would ideally run, then emulate its design in longitudinal registry data to estimate comparative treatment effects in routine clinical practice.
Explore Target Trial Emulation →Give a single-arm study something to compare against.
We build comparator and natural-history cohorts from registry data, applying your trial's own eligibility criteria and follow-up windows so the control arm is drawn from patients who would have qualified for it.
After launch, in the patients who were never enrolled.
We design and analyse post-authorisation safety and effectiveness studies (PASS), following the treated population for as long as the registers run, usually years past the trial's database lock. Where the study is a regulatory commitment, the protocol is written to GVP Module VIII and registered in the EU PAS Register.
Find out whether the question can be answered at all.
We establish which registers hold the variables you need, how complete they are, what sample size is achievable and which design is realistic, before anyone commits to a full study.
The data environment
Denmark has recorded every hospital admission since 1977 and every prescription filled at a pharmacy since 1995, linked to diagnoses, cancer data, causes of death and socioeconomic information through a personal identifier issued at birth or on immigration and used for life. The other Nordic countries hold comparable registers, so a design developed here can often be extended, where the question warrants it and the approvals allow.
Study complete or near-complete national populations, followed across years or decades, with emigration and death almost the only routes out of the cohort.
The Civil Registration System links parents to children, and the same identifier links health records to education and income data. That is what makes sibling-matched and within-family designs possible.
The National Patient Register, the National Prescription Register, the Danish Cancer Register, the Register of Causes of Death, and Statistics Denmark's education and income data, all linked on the same identifier.
Frameshift holds its own authorisation to work with Danish national register data, so we execute registry studies directly inside the approved Danish data environment. Individual studies still require project-level approval, and we tell you at scoping what that adds to the timeline.
Clinical trials
We run the statistics on clinical trials: design and sample size, statistical analysis plans, primary and secondary analyses, reporting and publication.
Estimands, endpoints and analysis populations are settled before the data are examined. In a trial the protocol forces that on you; in a registry study, usually nothing does. We hold registry studies to the same rule anyway.
All three of the client engagements we have delivered are randomised trials, published in NEJM Evidence, the International Wound Journal and Wound Repair and Regeneration. The analyses we led are there to read.
Trial and registry work often answer neighbouring questions for the same programme. We take on either, and say which one the question actually needs.
Track record
Registry and population-based studies from our team, alongside randomised trials, genetic epidemiology and clinical prediction work, and client engagements where we led the statistics. The published record is heaviest in haematology and cancer.
Nationwide decomposition separating demographic change from treatment progress in years of life lost to cancer across Denmark.
View Article →A nationwide Danish data resource linking lymphoid cancer diagnoses, treatment and outcomes across national health registers for research use.
View Article →Nationwide Danish cohort and genetic risk score study finding shared aetiological pathways between migraine and pre-eclampsia, with implications for cardiovascular risk stratification.
View Article →Randomised trial across 255 patients testing fish skin grafts against standard care for deep diabetic foot ulcers. Fish skin achieved wound closure in 44% versus 26% at 16 weeks.
View Article →Machine-learning models that sharpen risk stratification and predict treatment failure in diffuse large B-cell lymphoma from routine clinical data.
View Article →Nationwide Danish cohort linking routinely collected health data to examine breastfeeding duration and subsequent childhood cancer risk.
View Article →Registry-linked prescription and hospital data used to quantify how concurrent medication burden predicts survival, admissions and infection risk.
View Article →Why Frameshift
A study is only as good as its design. In a trial that means the endpoints, the estimands and the analysis populations are fixed before anyone sees the data; in a registry study it means eligibility, time zero, exposure and comparators are fixed on the same terms. No amount of sample size fixes a cohort defined wrong, a time zero that leaks information from the future, or an endpoint chosen after someone has seen the curves.
Statistical epidemiologist with a PhD from the University of Copenhagen. He trained in statistical genetics at deCODE genetics and held research posts at Statens Serum Institut and the Danish Cancer Institute before co-founding Frameshift. He has led the statistical work on randomised clinical trials, nationwide Danish registry studies and genetic epidemiology, including research published in NEJM Evidence, the British Journal of Cancer and The Lancet Obstetrics, Gynaecology & Women's Health. At Frameshift he leads study design, analysis strategy, statistical programming and reporting.
Statistical geneticist with a PhD in immunology and infectious diseases from the University of Copenhagen. He spent nearly a decade at deCODE genetics working on large-scale human genetics, with further experience in cancer epidemiology at the Danish Cancer Institute, spanning gene discovery, population cohorts and registry-based research. At Frameshift he leads the practice and works on study design, genetic and observational analyses, and turning complex biomedical questions into tractable statistical ones.
He works on machine learning and computational analysis for clinical and health research, built on a background in software engineering and data science. His published work spans machine-learning models for clinical risk prediction and the assembly of research-ready health-data infrastructure from real-world records. At Frameshift he leads computational and machine-learning work, bringing model development and validation to problems where standard statistical methods fall short.
Bias, timing and comparability decide whether a result means anything, in a trial or in a register. We settle the design questions first, and say plainly what the data can and cannot support.
Whoever designs a study writes its analysis and drafts its methods section. Questions about why an estimate moved go to the person who produced it.
A study runs from the feasibility read through cohort and endpoint definitions to the analysis and the write-up. Tables, figures and methods text are delivered in the state they need to be in for a submission.
The analysis plan is signed off before outcome data are examined. Primary results are independently checked by a second team member, and any deviation from the plan is logged and reported alongside the findings.
Frameshift Insights
Methodological essays on epidemiology, biostatistics and study design.
A new Danish nationwide register study followed 1.5 million children and found no association between prenatal acetaminophen exposure and autism, even in a sibling-matched analysis. A look at what the study shows, what the registers made possible, and why the evidence has only converged further.
The Guardian reported that CRC is now the #1 cancer killer in Americans under 50. We pooled 44 years of NORDCAN data across five countries. The trend is real, but the ranking is not: CRC is #3 in the Nordics for both sexes, and the rise is hitting young men hardest.
A 2013 Danish register study found that skin cancer patients had half the all-cause mortality of the general population. Swap the diagnosis for a lottery prize and the same design manufactures the same survival advantage, because the exposure is defined by having lived long enough to receive it.
Get in touch
We are most useful when you already have the question and need the design, the statistics or the reporting behind it, whether the data come from a randomised trial or from the national registers. Tell us what you are trying to establish and we will usually reply the same or next working day.
