Does C++ have dynamically declared types based on integer width arithmetics?

The title may seem odd, but I'm not sure what else to call it. If you have suggestions please let me know and I'll edit.

Consider this

uint8_t lo = 0x11; // given
uint8_t hi = 0x22; // given

uint16_t word = (hi << 8) | lo; // constructed

But there's lots of given values, and they are either uint8_t OR uint16_t. That means the result (word) should be uint16_t OR uint32_t, depending on the makeup of the vars I give it.

Is there anything like this? (bad syntax I know)

TYPE lo = 0x11;
TYPE hi = 0X22;

decl_int_type(sizeof(lo) + sizeof(hi)) word = (hi << sizeof(hi)) | lo;

Edit: I am not trying to solve a problem, or putting this in production. This came up as a discussion with a coworker and I was just asking for educational purposes.

Depending on what you want to do with the result, std::bitset might help you accomplish what you need:

#include <bitset>
#include <iostream>
#include <iomanip>
#include <climits>

int main() { 
    char lo = 0x11;
    unsigned short hi = 0X33;

    using uint64_t = unsigned long long;

    constexpr auto bits = (sizeof(lo) + sizeof(hi)) * CHAR_BIT;

    std::bitset<bits> b = (uint64_t(hi) << sizeof(lo)*CHAR_BIT) | lo;

    std::cout << b << "\n";
}

But whether this is useful depends heavily (completely, really) on why you want to do this, and what you want to do with the result.