What's important is how the conversion is defined between the two worlds (unmanaged vs managed) in the native or managed code side.

If conversion is defined using a COM / Automation Type Library (in DLL or TLB for example), then it will be declared as the corresponding type automatically, especially for very well known types such as the unsigned 16-bit integer type.

So a VBA 16-bit type will be declared as the equivalent (if it's a standard well-known type) .NET 16-byte. For example, VBA integer will be declared as .NET's Int16 (like the short keyword in C#).

If conversion is defined manually, for example using .NET interface with COM annotations, then it's the developer's task to ensure the declaration is consistent for both sides (or problems/bugs/crashes may happen).

I suggest two links in the official doc about this: Default Marshaling Behavior and Blittable and Non-Blittable Types to determine what can be considered as standard or well-known types.