In general, when talking about bicycle helmets we should remember that in countries with nearly 0% helmet adoption rate, after enforced helmets laws were enacted, causing nearly 100% helmet adoption rate, we usually see only very slight reduction in cyclist fatalities. Usually that reduction is similar to the reduction of cycling in general (every mandatory helmet law causes a reduction in cycling, if enforced). So based on this it appears that bicycle helmets do not seem to be particularly effective.
The studies that show otherwise (claiming that helmets are X% efficient with a large value of X) usually have a very wide definition of "head injury". For example, if your worst fear is that you could have a bleeding wound in your head, a helmet is approximately 100% efficient against that. But most people want something more from a helmet -- not just protection against minor injuries, but protection against major injuries and death.
The cause for this wide definition of "head injury" is that cycling is an extraordinarily safe activity, and so getting enough cases of "head injury" that are actually severe injuries would require a whole population and not just a subset. A scientist cannot reasonably collect enough data from other sources than nationwide statistics, so they expand the definition of "head injury" used in studies (and thus come to a false conclusion).
We also know that since a typical helmet does not have MIPS, that helmets are not actively harmful. One of the theories behind MIPS is that a helmet could cause rotation of the skull, causing a brain injury, and therefore riding with a helmet would be more harmful than riding without a helmet. This theory is false, because otherwise the countries with suddenly enacted enforced helmet laws would see a rapid increase in severe head injuries.
Yet the MIPS theory might have some merit: for example if a helmet with MIPS causes 50% reduction in severe injuries and deaths, it could be possible that missing MIPS would increase injuries by 50%, so a non-MIPS helmet could cause approximately 0% reduction in severe injuries and deaths (which we have observed in countries with sudden helmet wearing spike due to enforced helmet laws). So I can't conclusively prove that MIPS would be unnecessary. Getting enough data would require a country with 0% helmet adoption rate and lawmakers willing to enact legislation that would make riding illegal without a MIPS helmet, and actively enforcing that legislation (so every cop seeing a helmeted cyclist would need a way to quickly tell if it's a MIPS helmet).
If shopping for a helmet, I would consider this list a bare minimum:
- Better protection than the minimum helmet requirements. In Europe, the minimum is EN 1078, which is crap (a falling rod with one end fixed to ground and another end freely falling from 1.8 meter height has its head hit the pavement with as much energy as a dropped apple from 2.7 meter height; but EN 1078 drops severed heads only from 1.5 meter height). In US, the CPSC has 2.0 meter drop height which is far better, but Snell standards have a drop height of 2.2 meters which is the best of the current standards. I'd say buy a Snell certified helmet at least as some of those are available.
- MIPS. It has a valid idea behind it, that the observed helmet non-effectiveness is caused by increased rate of rotational injury. We don't know if the theory is true, but at least it could be true.
- In-mold construction. They have a polycarbonate shell around which the shock absorbing layer is molded. By placing the shell inside the mold, it becomes more rigidly attached to the shock absorbing layer, thus potentially increasing the robustness of the helmet in scenarios above that of the testing (which are not strict enough as a 2.7 meter drop height should be used at least).
Also remember that even if you satisfy this shopping list, helmet effectiveness is still extraordinarily poor. Is it 0% effective against severe injury / death? Or 10% effective? Or 20% effective?
The fact is, we don't know. The only things we know are:
- Cycling is an extraordinarily safe activity. The health benefits of riding a mile far exceeds the health risks of riding a mile, helmet or no helmet.
- A helmet cannot and will not be nowhere as effective as for example a car seatbelt