What are the odds of data mysteriously changing in your database?

On rare occasions, we receive a missive like the one below, where you’re absolutely 100% certain that data is mysteriously changing and that it’s not even remotely possible a human in your office could be responsible (inadvertently or otherwise):

“The wrong diagnosis is being sent for the patients condition. What is being sent to Aetna is for migraine headaches. What he is being treated for is low back pain. The old diagnosis was sent for his current treatment. Somehow the patient’s DX is different. The DX in the printed DX is for back, not migraines. Is this a computer malfunction? Why could this happen?”

Let’s discuss how ECLIPSE data is actually represented internally & saved on your computer. We’ll use dates & last names as an example:

A date in ECLIPSE is represented internally by 4 bytes or 32 bits – each of which can have a binary value of 0 or 1. That’s over 4 billion (4,294,967,295) possible values.

A last name is represented internally by 26 bytes, each of which can have a decimal value between 0 and 255. For instance, the value “0” means no data. A value of “32” is a blank space. And a lower case “a” has a value of 97.

• The odds of any single character in the name changing are 1 in 255.
• The odds of 2 characters changing to a specific value (e.g. “Xi”) are 255 * 255, or 1 in 65,025.
• The odds of just 5 characters simultaneously & randomly changing (e.g. from the name “Smith”) to some other specific value (e.g. they’re all gone, all spaces, or some other recognizable name, like “Doe”) are 255 * 255 * 255 *255 * 255, or 1 in 1,078,203,909,375. (And that’s actually a simplification.)
  Virtually all of the possible values would result in what appeared to be corrupt data (as opposed to the wrong date… or name) appearing on your screen. A different way of looking at it would be to think about the odds of getting hit by lightning, which are approximately 1 in 700,000 during a calendar year & twice in your lifetime… about 1 in 9 million.

The odds of you being hit by lightning twice in your lifetime are far better than the data in those bytes randomly changing to a valid value (e.g. another date, no name at all, or a different name).

It’s indeed more likely that someone has inadvertently changed the information. We use ECLIPSE in our office too and are unaware of similar issues anywhere… However, it’s perfectly natural to be certain you didn’t alter data. We have all done this at one point or another! Regardless, keep an eye on it. And let us know. Every year, someone does get hit by lightning…