Measuring the ratio of protein to creatinine in the urine has been shown to be a very useful way of estimating the amount of protein that a patient is losing in the urine.
Normal people are really good at keeping protein from being flushed away by the kidneys. Firstly, the kidneys are very resistant to filtering most of the proteins found in the blood stream. Those (mostly very small) ones which do get through are usually catabolised and reabsorbed by the tubular cells. As a result, healthy adults lose a maximum of 150 mg of protein in a day. This includes about 100 mg of Tamm-Horsfall protein (a protein produced by the loop of Henle), 5-15 mg albumin and 20-40 mg other proteins.
Almost all renal diseases increase the amount of protein lost in the urine. However, the test is quite a sensitive way of picking up subtle renal abnormalities. For instance, diabetics and hypertensives are both prone to renal insufficiency, and you'll know that this is starting by checking for elevated 24 hour proteinuria - this is abnormal long before, say, serum creatinine starts to rise.
The most obvious way to test for elevated protein excretion in the urine is to ask the patient to pee in the same container over a 24 hour period, and then simply measure the amount of protein therein. This works - but it is very cumbersome. The patient generally needs to be admitted to hospital, and accurate compliance with this regime is tricky.
A much cleverer way of estimating the 24 hour urinary protein excretion is the protein-creatinine ratio. The assumption behind this test is that people excrete 10 mmol of creatinine per day in their urine. This is a fairly valid assumption, since creatinine is freely filtered and its urinary concentration depends mostly on muscle mass. (If you want to be even more accurate, you could say that women excrete about 9 mmol, and men 12 mmol.)
Once you assume this, you can happily estimate the amount of protein that would have been found in a patient's urine over 24 hours from a single sample of his/her urine. For instance, say a patient would produce 10 mmol of creatinine per day. On a random sample of her urine, we find the following: protein: 60 mg, creatinine: 2 mg. A simple calculation (multiplying both figures by 5, in this case) allows us to deduce that this patient would have excreted 300 mg of protein into her urine over 24 hours - thus, she has proteinuria, and renal dysfunction.
It is obvious that this once-off measurement is enormously more practical than strictly collecting urine for 24 hours. Its accuracy has also been confirmed over wide ranges of renal functions. In fact, the 2005 UK Chronic Kidney Disease guidelines state that the urinary protein-creatinine ratio is a better test than 24 hour urinary protein measurement.