I also saw this study 
Both things are true. The study was funded by Axkid so the framing of it is clear that they are prioritising the results showing rear facing is protective. But the results aren't skewed, ie the information is accurate, it's just only one part of the picture. If you read the actual study, linked at the bottom of the article, rather than the bits presented on this page, it gives you a bit more of the full picture.
It is worth noting that this study is completely useless for assessing the relative protection of 5 point harness forward facing vs high back booster, partially because the numbers involved were so low that you can't realistically draw any conclusions, and partly (possibly for this reason) they have all been lumped together into one category.
There have been other studies of this nature which have been done previously and they all show fairly similar results, in fact this particular study shows less of an effect from rear facing. In the older studies, it was shown that roughly 3/4 of deaths in forward facing or unrestrained children could likely have been prevented by rear facing. In this study it is lower, more like 2/3. My guess as to why this is is that modern cars are better at protecting the occupants, and perhaps better usage of high back booster type seats for children in the in-between phase. In the past it was common to place children in booster seats, sometimes booster cushions, or just a seatbelt from around 3. Possibly also that modern forward facing seats tend to use isofix and are easier to use, as misuse is much more impactful in forward facing seats.
This is essentially a case study type where they look at fatality cases and analyse using recreations, crash test data, and expert analysis whether the child could have possibly survived if they were in a different kind of seat. These are only ever estimates, obviously we don't have a time machine to actually know, and because thankfully child passenger car deaths are low in number, the numbers will only ever give you a very rough idea.
@Walkacrossthesand These figures only make sense when you compare to what people normally do, and whether the group is larger or smaller for children who died in accidents. If car seat type had no effect at all, you'd expect to see that "normal use" of car seats roughly correlates with the distribution of type of car seat used when a child has died. Usually though what you see is a significant difference in that more children who died were travelling unrestrained or forward facing compared with rear facing, suggesting the rear facing seat is the most protective.
For some reason they didn't include the figures here, but in another fairly recent study from Sweden, the standard usage for different kinds of child car seats up to age 4/5 is as follows: 57% rear facing (although this will be higher in reality as this study mainly observed children aged 2+) and 2.5% unrestrained with presumably the other ~40% being in forward facing seats.
That does show a clear difference - over 57% normally RF, vs 28% of children who died, around 40% normally FF vs 34% of children who died, and around 2.5% usually unrestrained vs 33% unrestrained - the biggest difference is in the unrestrained group, which is expected, but there is also a large difference in the RF group.
However there is another statistical problem with comparing these kinds of stats - there is a correlation between generally risky or irresponsible driver behaviour such as driving without insurance/license, distracted driving, drunk/drug driving, and not using child restraints. So while it's absolutely true that driving with unrestrained children is putting them at massive risk if you were to crash, and using a rear facing seat significantly reduces risk compared with forward facing, there is some aspect especially in the unrestrained group of bias because if you're the kind of parent to use a car seat and particularly to select a car seat based on safety credentials, then you probably are also less likely to have a crash because you are more likely to care and think about safety when driving. That's not any direct effect from the type of restraint, it's just a correlation.
The case study approach which is used in this study is more comprehensive - it looks at each case in detail and makes a judgement for each. For this study they did not actually compare whether a child who was unrestrained and under 125cm would have been better protected in a forward facing seat. They ONLY compared the likely outcome if the child had been in a rear facing seat. So it simply doesn't really tell you anything about whether forward facing seats are protective or not. We know from other studies that forward facing seats are indeed protective mainly because being unrestrained is so high risk. But it's not what this study is looking at. This study only compared FF and unrestrained to RF, and unrestrained to FF in children over 125cm.
I need to do the school run, but I can come back later with other, older studies of this type if interested.