We’re in the really early days of artificial intelligence (AI) being used extensively in health care, and while I can see some great potential, I worry that it’s all built on flawed assumptions.
Research by AI often depends with billing codes, known as ICD codes (the most current of which are known as ICD10). There’s supposed to be one for every imaginable condition, but many rare conditions, including TIO, have no specific billing code. The TIO situation is mentioned in “Epidemiology of Tumor Induced Osteomalacia in Denmark” in Calc Tiss Int (April 2021). To learn more about these codes in general, check out the EveryLife Foundation’s ICD Code Roadmap with lots of resources for both patients and advocacy groups who are trying to get a code for their rare disorder.
XLH and autosomal hypophosphatemias are lucky to have a code, although it encompasses all the familial hypophosphatemias, E83.31, rather than distinguishing among the various genetic forms of transmission. Now that you know it, I encourage you to make sure your clinician uses the correct code in your records if at all possible. It can affect your insurance coverage, in addition to more theoretical issues.
The problem for research and eventually for widespread use of artificial intelligence is that clinicians use the wrong code all the time. And then it’s almost impossible for a patient to get the codes corrected in their files, since clinicians seldom have an incentive to change them as long as they’re getting paid, and using the wrong codes doesn’t always prevent payment. My current primary uses the correct code, but the previous one used the one for nutritional rickets, a purely pediatric disorder! (Yeah, I’m a little embarrassed that here I am, a patient advocate, and I have such an egregious error in my records, but it’s not worth my time and energy to fix old records, so I concentrate on current/future data.)
Coding can also be incomplete, rather than wrong, as when, say, a rheumatologist treating an XLHer for the osteoarthritis that is the consequence of XLH, uses a billing code for arthritis without also listing the code for the underlying diagnosis of XLH, which is what causes the arthritis. At least anecdotally, it appears that a significant number of XLH adults don’t see any health care provider specifically for their XLH, but instead seek treatment for individual symptoms, like the arthritis, so the XLH diagnostic code may not appear anywhere in their records. As a result, if someone were researching anonymized data, looking for the incidence of XLH patients in a region, the total number of XLH patients would be underreported.
Wrong billing codes can cause problems in In the opposite direction too, where conditions are wrongly diagnosed and thus over-reported. Patients with rare disorders tend to accumulate a significant number of misdiagnoses over the years. My own records are littered with suggestions that I have rheumatoid arthritis, DISH (Diffuse Ideopathic Skeletal Hypertrophy), fibromyalgia, and ankylosing spondylitis. Those are all common misdiagnoses made by clinicians who aren’t aware that XLH causes (directly or indirectly) osteoarthritis, skeletal hypertrophy (enlarged bones), whole-body pain (like fibromyalgia), and the types of spinal changes that are characteristic of ankylosing spondylitis. I wouldn’t be surprised if there were also (false) indications in my records and those of many XLHers that we’re drug-seeking (describing pain that is diffuse and moves from spot to spot), crazy (imagining non-existent pain or other odd symptoms), or mentally unbalanced (emotional in the face of not being heard or the frustration of not having any medical options for very real problems).
And there’s no way for us to get any of the irrelevant information out of our records, short of, perhaps, a court order, which is beyond the reach of most patients!
On the plus side, I figure if AI ever gets a look at my records, the data will crash the system from all the anomalies. It amuses me to picture a medical supercomputer melting down with an “out of cheese error” like the magical computer, Hex, does in Terry Pratchett’s Discworld series, when faced with impossible data. Maybe then researchers and clinicians will take the radical step of, yanno, actually listening to patients.
Still, it would be better if the records were correct rather than amusing, so if it’s at all possible, I encourage you to check your records and see if there’s a correct XLH diagnostic code in them. If it’s wrong, you may not be able to get it changed/erased, but you might be able to get the correct one added, especially if you can tell the clinician (or their assistant) what the correct code is.
***
Please note that the author is a well-read patient, not a doctor, and is not offering medical or legal advice.
If you’d prefer to read this blog as a newsletter, sent straight to your inbox, please sign up here.