Estimating the cost of clinical trials accurately is often a challenging task, especially when dealing with diseases such as Tuberculosis (TB).
Tuberculosis is one of the most common and deadly infectious diseases globally, and new treatments are desperately needed to combat it. Humanitarian organization Médecins Sans Frontières (MSFs), or Doctors Without Borders, recently shared the detailed cost data for a clinical trial on drug-resistant tuberculosis in PLOS Global Public Health. They reported this trial cost a total of $39 million.[1]
MSF reported that 26% of this cost came from central activities, while operational expenses such as staffing accounted for 32% of the cost. The duration of the trial, the number of subjects, the type of intervention, and even the country where it’s conducted also greatly influence the overall expenditure. A large amount of the information needed to make an accurate trial budget is in certain sections of the protocol, such as the schedule of events.
Cost your tuberculosis clinical trial protocol
Predicting trial costs accurately is complicated. You can either take the past trials that you have cost data from, and try to extrapolate a model from the total final cost and apply this to a new trial that is in the planning stage, or you can attempt to itemise the costs and add them up. I would call these two approaches “top down” and “bottom up” respectively.
| Approach | Description | Advantages | Disadvantages |
|---|---|---|---|
| Top down | Take all your past trials and the final cost that was known, and train a model. This could be a very simple linear model such as a times number of subjects plus b times number of sites etc. | Tends to learn from the real final costs of past trials. So if “black swans” or unexpected costs arose in your past trial, such as high numbers of AEs, this would be baked into your model. | Doesn’t provide an itemised breakdown. Hard to understand. |
| Bottom up | Find all invoiceable items in the protocol and sum them into a total trial cost spreadsheet. | Understandable - you can see at a glance what will happen to the total cost if a particular visit or lab test is cut from the trial budget. | Tends to underestimate costs because many costs that arise in a trial are unforeseeable. Also, CRO costs are often not explicitly stated in a protocol, such as the number of CRA meetings, CRA training costs, so these can be hard to itemise and will need some intelligent guesswork. |
That’s where tools such as Fast Data Science’s Clinical Trial Risk Tool come into the picture. Our tool can estimate the risk of a trial failure, the total cost, and generate itemised trial budgets. The Clinical Trial Risk Tool can perform both a bottom-up and a top-down cost analysis.
This innovative tool uses machine learning algorithms to predict the cost of a clinical trial from its written protocol. It can analyse the schedule of events, number of subjects, interventions, and even the type of intervention to provide an estimation of the trial’s overall cost.
For example, trials conducted at sites in high-income countries also tend to cost more, and drug interventions are more expensive on average than medical devices or behavioural interventions.[2]
Find out about TB trial costs
Estimating the cost of a tuberculosis clinical trial becomes easier with the Clinical Trial Risk Tool - all you have to do is upload your TB study protocol to the tool and see the estimates it comes up with. If you disagree with the AI’s output, you are free to adjust the intermediate values.
Use the search feature below to find TB trials from the Clinicaltrials.gov registry.
