Unit testing logic that looks similar to its implementation

What is the general convention for unit testing logic that looks similar to its implementation?

For example, say I have a function like this:

function getNumberOfElements() {
  return document.childNodes[0].childNodes.length
}

If I tested this routine in a fixture, I would be repeating the logic in the routine:

it('should return the number of nodes', function() {
  var numberOfElements = document.childNodes[0].childNodes.length
  expect(getNumberOfElements()).toBe(numberOfElements)
})

So is there a better way? I can see this approach as being prone to all sorts of problems, but how else would you test a function like this?