Even here in this article, I am looking at the graphs they present as evidence of the thesis. Maybe I'm crazy but looking at the line for male early-onset incidence I don't see anything new or worrying in that graph. If we ignore the puzzling "estimated" section, it wanders up and down. It's not currently the highest it's been in the recent past, and increases in e.g. the 90s were followed by decreases later. The rates in women meanwhile have been roughly the same since 1995 after a small increase in the early 90s. This is presumably why they pick 1990 as their base year, as otherwise they would not be able to report a big percentage change.
So the story here is really being carried by the estimated/modeled future incidence, but it's not a good idea to pay attention to public health modeling. This is epidemiology, a field that we learned during COVID is not robust or honest. These people constantly claim to be able to predict the future incidence of disease, are constantly being proven wrong by miles, always in the direction of over-predicting incidence, and their response is to just ignore that fact, attack their critics and carry on as if everything is fine. Basic methodological errors are everywhere and nobody inside the field cares.
So we look at this prediction and immediately notice a couple of things. Firstly, it's a prediction of the past. Where is the actual cancer data for 2020, 2021, 2022 and 2023? The source they use only runs a survey every two years, and the last was in 2021 but appears to not report cancer in that round for some reason. There certainly are figures for cancer available for many countries but they don't deign to verify their predictions against that data. Replacing data that exists with modeled guesses is completely typical for epidemiology, I've seen that dozens of times.
The second problem is that their estimate isn't a simple extrapolation of whatever was happening previously. Instead it has an immediate sharp upswing at the moment the model begins and then a very steep climb quite unlike anything seen in the past. This is surely an artifact of the rather complicated statistical methodology they picked. When we look at Figure 5 in their source paper we see that the CIs of this prediction are at any rate enormous (and presented confusingly of course).
Is there anything to worry about here? Probably not. In other contexts the steady rates of cancer over the past 30 years are used to debunk claims about mobile phones causing tumors. If there was genuinely a steady and large rise in cancers over a long period of time they would be able to show that clearly using actual collected data, instead of relying on their usual bag of tricks, and the conversation around mobile phones would look very different.
Even here in this article, I am looking at the graphs they present as evidence of the thesis. Maybe I'm crazy but looking at the line for male early-onset incidence I don't see anything new or worrying in that graph. If we ignore the puzzling "estimated" section, it wanders up and down. It's not currently the highest it's been in the recent past, and increases in e.g. the 90s were followed by decreases later. The rates in women meanwhile have been roughly the same since 1995 after a small increase in the early 90s. This is presumably why they pick 1990 as their base year, as otherwise they would not be able to report a big percentage change.
So the story here is really being carried by the estimated/modeled future incidence, but it's not a good idea to pay attention to public health modeling. This is epidemiology, a field that we learned during COVID is not robust or honest. These people constantly claim to be able to predict the future incidence of disease, are constantly being proven wrong by miles, always in the direction of over-predicting incidence, and their response is to just ignore that fact, attack their critics and carry on as if everything is fine. Basic methodological errors are everywhere and nobody inside the field cares.
So we look at this prediction and immediately notice a couple of things. Firstly, it's a prediction of the past. Where is the actual cancer data for 2020, 2021, 2022 and 2023? The source they use only runs a survey every two years, and the last was in 2021 but appears to not report cancer in that round for some reason. There certainly are figures for cancer available for many countries but they don't deign to verify their predictions against that data. Replacing data that exists with modeled guesses is completely typical for epidemiology, I've seen that dozens of times.
The second problem is that their estimate isn't a simple extrapolation of whatever was happening previously. Instead it has an immediate sharp upswing at the moment the model begins and then a very steep climb quite unlike anything seen in the past. This is surely an artifact of the rather complicated statistical methodology they picked. When we look at Figure 5 in their source paper we see that the CIs of this prediction are at any rate enormous (and presented confusingly of course).
Is there anything to worry about here? Probably not. In other contexts the steady rates of cancer over the past 30 years are used to debunk claims about mobile phones causing tumors. If there was genuinely a steady and large rise in cancers over a long period of time they would be able to show that clearly using actual collected data, instead of relying on their usual bag of tricks, and the conversation around mobile phones would look very different.