This week, we had a few charming examples of risk management in clinical trials with several of our customers. I started thinking about what we could do to get things to run real fast and avoid some of the inevitable potholes and black swans that crop up in clinical trials.
Engaged in basic science and stuck in data traffic
There is something very disturbing about an industry that develops products using advanced basic science.
It is disturbing because the industry uses 40-year old processes and information technology.
This industry accepts a reality of delays of a year or more due to manual data processing.
This industry is called life sciences.
That’s what disturbs on a personal and strategic level. We can and should do better. The disconnect between basic science and modern software should disturb anyone involved with clinical research because the cost to society is enormous. We are enamoured with Instagram, Uber and WeWork but we choose to pretend that life science research exists in a parallel untouchable universe protected by ICH GCP, FDA, MDR and a slew of other TLAs.
Alright. I am Israeli and trained as a physicist. Let’s look for some practical, real-world solutions. Let’s try them out and iterate.
5 ways to make your clinical research run real fast
1. Data model
Before designing your eCRF, design your data model. If you do not know what data modelling means, then 4 weeks before the study starts is a bad time to start learning. Hire a specialist in data modelling, preferably someone who does not work in life sciences. Pay them $500/hour. It’s worth every penny. The big idea is to design an abstract data model for your study for speed of access and usability by patients, site coordinators, study monitors and statisticians before designing the eCRF.
2. Discipline equals speed
Start early. Go slow and speak softly and then run fast. There is a story about the difference between a Japanese wood sculpture artist and an Israeli artist. The Japanese artist goes into his studio and looks at a big piece of wood. He walks around the wood and observes. He goes home. The next day and for the next month, he observes the wood in his studio, without touching his tools. After a month of observation, he comes in, picks up a . hammer and chisel and chop, chop chop, produces a memorable work of art. The Israeli goes into his studio and looks at a big piece of wood. He starts carving away and improvising all kinds of ideas from his head. He goes home. The next day and for the next month, he chops away at wood and replaces raw material several times. After a year, he has a work of art.
The big idea is that discipline equals speed. It prepares you for the unexpected. See point 6 below.
A good book that presents this approach in a very practical way is Discipline equals Freedom by Jocko Willink.
3.Date and time
Date/time issues can be visualised as a triangle.
Side 1 of the triangle is the site coordinator who collects data into the EDC.
Side 2 of the triangle is the CRA who monitors CRC work and data quality and performs SDV.
Side 3 of this triangle is the subject who needs to come and visit the doctor on certain days that study coordinator scheduled for her when she started the trial.
Pay attention to your date and time fields. This is a much neglected part of data design in clinical trials.
The challenge is that you need to get your clinical data on different timelines. Most people ignore the fact that clinical trials have several parallel timelines.
One timeline is the study schedule. Another timeline is adverse events. Another timeline is patient compliance. You get it. If you collect high quality date times in your data model, you can facilitate generating the different time-series.
One of the most popular pieces on this blog is an essay Jenya wrote on dates and times in clinical data management. You can read it here.
4.Do not DIY your EDC
You can DIY a chair from Ikea but not your clinical trial. The notion of a researcher or clinical manager, untrained in data modeling, data analysis and user interface design using a DIY tool to develop the most important part of your study should make you stop and think. To put this in different perspective, if you are spending $5,000/month to monitor 3 sites, you should not be paying $450/month for a DIY EDC. It’s called penny-wise and pound foolish.
5.Prioritize deviations.
While it is true that protocol deviations need to be recorded, not every protocol deviation is created equal. I was stunned recently to hear from a quality manager at one of the big CROs that they do not prioritise their deviation management. Biometrics, dosing, patient compliance and clinical outcomes should be at the top of list when they relate to the primary clinical endpoint or safety endpoint. This is related to the previous points of not DIY, data modelling and observing before cutting wood.
6.Do some up-front risk assessment but don’t kid yourself.
Before you start the study, any threat analysis you do is worthless. A risk analysis without data is worthless. You may have some hypotheses based on previous work you or someone else did but do not kid yourself. First collect data, then analyse threats. I’ve written about how to do a risk assessment in clinical trials here, here, here and here. Read my essay on invisible gorillas.
