Read on Twitter
Patients said they wanted more info about the possible side effects of their breast cancer therapies, but there wasn’t good evidence so we started doing meta-analyses. The first is now out, on bisphosphonates. We found 24 possible side effects...
https://doi.org/10.1371/journal.pone.0246441
https://doi.org/10.1371/journal.pone.0246441
This is based on looking at the data from 103 different adverse outcomes, reported in 56 trials, including 29,248 breast cancer patients.
We looked at whether effects were different in metastatic v non-metastatic patients. They were generally very similar.
We looked at whether effects were different in metastatic v non-metastatic patients. They were generally very similar.
We also looked to see whether the effects were different for menopausal and non-menopausal women. They were essentially the same.
We did a network meta-analysis for each of the 103 side effects: https://bit.ly/2Z3RFQT
[check out that site - @cjackstats amazing work!]
We did a network meta-analysis for each of the 103 side effects: https://bit.ly/2Z3RFQT
[check out that site - @cjackstats amazing work!]
This helped us determine whether the drug type or the dose affected each side effect.
All 103. Again, check out that link! @cjackstats is amazing!
All 103. Again, check out that link! @cjackstats is amazing!
We also analysed the data from a single, large, trial both as a check of our meta-analysis and also to look at drug interactions: did the other therapies a woman was on affect her bisphosphonate-related side effects? The answer: on the whole, no.
(Thanks AZURE team!)
(Thanks AZURE team!)
The benefits from different breast cancer therapies are also independent of each other, which makes it possible to make models like https://breast.predict.nhs.uk . If side effects are also independent, it means we can add them to the Predict model and also give likelihoods for patients.
We had a quick look at clustering of side-effects too – the ones that clustered were not generally ones to do with bisphosphonates, but you can see the clusters most likely related to chemotherapy and hormone therapy.
More helpful info for patients!
More helpful info for patients!
Finally, we got hold of baseline absolute risk levels for these side effects in a comparable population of women so that we could give patients the number of women out of 100 who would be expected to suffer each effect if they were and were not taking bisphosphonates.
Amazing work & graphs from @cjackstats; data extraction by @zsofiaszlamka; a lot of good advice from @d_spiegel. Reporting of adverse effects is variable - it makes this work hard. All our data is in 10.17605/OSF.IO/GZ47H for those who want to update it in the future.
