Need a statistician for analysis of what constitutes a "random sample" of billings

Case involves a so called random sample of billings in a health care billing dispute. Out of approximately 2,000 patients and 50,000 sampling events, a few hundred were picked. These were analyzed for regulatory compliance and then a "bill" made up to recoup money by means of projecting the sample over the entire patient population and billing history. Our contention is that the sample was not "random" but skewed so as to produce a larger billing error.