Another dot in the blogosphere?

Posts Tagged ‘graphs

Video source

The video above provides insights into what graphs about COVID-19 tell us and what they do not. It is a lesson on critical analysis.

One takeaway from the video might be this: The numbers do not lie, but people do. You can use numbers to tell the story you want. So it is important for all of us to learn to read in between the lines.

Consider another aspect of COVID-19 — the search for a therapy. The latest drug to emerge with potential for treating patients is remdesivir.

Remdesivir is an experimental anti-viral drug manufactured by the pharmaceutical company Gilead. The drug interferes with the replication of corona viruses by mimicking an RNA nucleotide (a building block for the virus).

Remdesivir was initially used on a compassionate case basis. So what is there to read in between the lines? The first line is an initial study and the second is another study.

Remdesivir did not provide statistically significant therapy in the first study. The sample size of patients was also too low, but it hinted that those who were treated earlier seemed to benefit.

The second study was larger and claimed that “… more than 1,000 patients showed those given remdesivir improved after an average of 11 days, compared with an average of 15 days for those not given the treatment”.

So we have found a cure, have we not? Remdesivir is not our trump card yet. We have several unknowns:

It might highly depend on when and at what stage of the infection patients receive the drug. We can say it is currently not clear who is benefitting from remdesivir. Is it helping patients who would have recovered anyway, recover quicker? Is remdesivir more beneficial for younger compared [with] older patients? At what stage of the infection does treatment yield the best outcomes?

A layperson needs to read these in between the lines.

  • The two studies cannot be compared because they do not have the same designs.
  • When the drug is administered might be important in speeding up recovery (earlier is better)
  • The confounding variables are not accounted for, i.e., people may recover for reasons other than the treatment.
  • The drug might help people recover faster but it does not prevent people from dying.

A brutal way of thinking about the last point is that the people who recover will do so faster with the help of remdesivir; the ones who will die will do so even if they receive the drug.

So a graph can show or it can hide. It does so at the will and whim of those who illustrate with numbers. Scientific studies are typically written not to hide, but they can be hard to understand in terms of language and interpretation of results. This is why we need basic science literacy — so we can read in between the lines of a graph or a paragraph.


Usage policy

%d bloggers like this: