AnFI,
Let's deal with the engine failure first: it has been established (empirically) that the failure rate of helicopter (turbine) engines is about 1:100,000/flight hour. When two engines are fitted, and certificated, the rules require that each engine is protected (to the extent possible) from failure of the other.
If we exclude failure from (common) external factors such as salt water ingestion, ash ingestions, fuel starvation, FOD ingestion etc., we can assume the failure of each engine is an independent event. The probability of an engine failure is 1:100,000 (or 1 x 10^-5) therefore the failure of one engine followed by the failure of the other is 1 x 10^-5 and (times) 1 x 10^-5 or 1 x 10^-10.
Hence the probability of failure of 'a' followed by 'b' is 1 x 10^-10.
The result of an engine failure for a single is a forced landing; the result of an engine failure for a CAT A twin operating in Performance Class 1 is continued flight on the other engine.
However, the important issue is the consequence of the failure; if we assume superior skill in the pilot performing an auto-rotation and landing, what remains is the characteristic of the (forced) landing site. If the landing
can be performed without harm to the occupants of the helicopter it is said to be a safe-forced-landing. Where the landing cannot be performed without harm to the occupants of the helicopter, it is said to be hostile.
Hostility is ICAO is defined as:
Hostile environment. An environment in which:
a) a safe forced landing cannot be accomplished because the surface and surrounding environment are inadequate; or
b) the helicopter occupants cannot be adequately protected from the elements; or
c) search and rescue response/capability is not provided consistent with anticipated exposure; or
d) there is an unacceptable risk of endangering persons or property on the ground.
Clearly: b) above can be mitigated with the use of life-vests (with splash hoods), life-rafts and survival suits; c) is dependent upon b)
and the efficacy of search and rescue.
Where hostility of the environment is transient - i.e. at sea - but occurs frequently or can happen very quickly, some States take the precaution of defining it as hostile.
Where hostility of the environment is transient but occurs infrequently and with warning, the operator is left with the Risk Assessment and provision of appropriate 'adverse weather' policy/procedures.
As you have said, there are other reliability issues that have to be addressed; these are dealt with in certification and build standards, and component lifing; or, with redundancy. Some States also take the additional precaution of permitting flight over a hostile environment only in Performance Class 1 (certification in Category A being a more quantitative process).
All of this has been considered in detail for a number of decades; States and operators (and oil companies) have used the above tools to Risk Assess their operations and make appropriate provisions in their regulations/procedures.
It all comes down to the old adage, horses for courses.
This is not a single v twin but a risk assessment discussion. Mathematics /statistics /probability just provides us with the tools with which to make the best (and informed) decisions.
Jim