"You'll thank me someday ..."
August 12, 2021 9:54 AM   Subscribe

Some patients have found themselves suddenly treated like criminals because of their dogs' medications, their past sexual abuse, and seeing doctors a long way away from their house. Why?
"A sweeping drug addiction risk algorithm has become central to how the US handles the opioid crisis. It may only be making the crisis worse."
posted by Countess Elena (65 comments total) 28 users marked this as a favorite
 
Given how well we've done with every other aspect of the response to this so-called crisis, I'm somehow not surprised. (So-called because there wasn't much of a crisis before the crackdown, just a bunch of moralizing shits clutching their pearls. We've got quite the crisis now, though! Good job on that, DEA!)
posted by wierdo at 10:10 AM on August 12, 2021 [9 favorites]


Interesting - that explains why the new Vet I'm taking my dog to asked for my birthday. (They even mentioned the DEA in explaining it).
posted by drewbage1847 at 10:12 AM on August 12, 2021


This phrase from the article:

the problem of algorithmic unfairness—the tendency of AI to obscure and weaponize the biases of its underlying data

is eminently quotable.
posted by gimonca at 10:15 AM on August 12, 2021 [33 favorites]


25 years ago, I was devastated that there weren't more pharmacy flags in place as my loved one doctor-shopped to his death at age 29. Now we've gone too far in the other direction. One of these centuries, perhaps humans will get it right.

Another big risk factor is mental illness, which affects at least 64 percent of all people with opioid use disorder.

DING DING DING YA THINK?? There's a reason why they call it self-medicating. I'm too exhausted to rant about this, but the lede is buried just a little bit in TFA ...
posted by Melismata at 10:19 AM on August 12, 2021 [27 favorites]


On the plus side, the AI has ensured that the utilization of paperclips is optimal!
posted by SPrintF at 10:27 AM on August 12, 2021 [4 favorites]


Prescriptions for animals are put under their owner's name.

Is this still true? I worked in a pharmacy in high school and regularly filled prescriptions for Fido Smith or whatever the pet's name was. I believe the pharmacist would fill controlled substances but I don't remember as this was in early 2000s and things were much less strict.
posted by geoff. at 10:36 AM on August 12, 2021


Let me guess - this company makes a fortune while creating a hell for individuals, and not addressing any root causes of drug abuse or overprescription of pain medication. Yes, we all need yet another gatekeeping layer to keep us from medical care, especially when there are profits to be made. Sigh.
posted by 41swans at 10:42 AM on August 12, 2021 [33 favorites]


Is this still true? I worked in a pharmacy in high school and regularly filled prescriptions for Fido Smith or whatever the pet's name was. I believe the pharmacist would fill controlled substances but I don't remember as this was in early 2000s and things were much less strict.

I just emergency filled a prescription for gabapentin for my cat Arthur Dent at a human pharmacy while driving cross-country. The prescription I got was absolutely listed under "Arthur LASTNAME," not my name. (They also refused to give me my pharmacy slip back, which.... fine, okay, it's not like insurance pays for his meds. I can probably get a new scrip from my old vet if I need to in order to keep his meds intact, given that he's on them for a pretty nasty case of pica.)
posted by sciatrix at 10:45 AM on August 12, 2021 [3 favorites]


When I had my knee surgery in October, my surgeon prescribed me Percoset for the post-surgery pain, along with a second prescription for a laxative in case the Percoset gave me constipation. But the hospital pharmacy only gave me one bottle; I was a little too out of it to question that, though, and just went on home, taking one of the pills that was in the bottle I'd been given before trying to go to sleep.

It did absolutely nothing to treat the pain that finally started kicking in an hour later when the anesthesia wore all the way off. I spent a miserable sleepless night as all the pain from the surgery set in, watching the clock like a hawk to check whether enough time had elapsed for me to try taking another pill in the hopes that maybe it would finally do something this time. But no dice. The following morning, when my surgeon's office finally opened, I called him immediately to holler that the pain meds are doing nothing, for the love of God help me.

He asked me to check the bottle I'd been given - and to both of our surprise, we saw that I had instead been given Tramadol. He apologized profusely, told me that I could supplement that with regular old over-the-counter ibuprofen, like I had already ("take three pills instead of just one each time if you're doing that," he said) and said he would call the pharmacy right away to find out what the hell happened.

I immediately downed three Advil, and it was already starting to kick in when my surgeon called me back. Apparently, he said, there had been "a flag on my record" which caused them to overrule my surgeon's prescription and give me something that wasn't an opioid; they had simply forgotten to tell him that they'd overruled him, so he had no idea it had happened. He apologized again, but I was so exhausted and relieved the pain was finally going away that I just said "okay, it's all good," and we agreed I'd go the Tramadol-and-ibuprofen route - I'd probably only need it a few days anyway, and sure enough, within three days I didn't need anything at all.

At the time I thought that the mysterious "flag on my record" came from my having casually mentioned to a nurse that as a teenager I'd once thrown up after taking Codeine. But now....I wonder.
posted by EmpressCallipygos at 10:46 AM on August 12, 2021 [44 favorites]


Slight conflict here. My mother was a drug addict and I wish this had been part of her record, cause then she would have done heroin more which she was nicer on. On the other hand, I once had a nurse make me chew a Tylenol because I couldn't swallow it whole (wisdom tooth surgery that went wrong) and I cried the whole time because I had just had wisdom tooth surgery. I had IV PAIN MEDS prescribed but she said that she "didn't want me getting addicted". I called the charge nurse and they pulled her off my room, fixed me up, and I've never had a problem with meds. I need to go process this article in the fetal position.
posted by lextex at 11:00 AM on August 12, 2021 [20 favorites]


which caused them to overrule my surgeon's prescription and give me something that wasn't an opioid

Better not tell them how Tramadol works!
posted by atoxyl at 11:11 AM on August 12, 2021 [19 favorites]


Better not tell them how Tramadol works!

Funny (“funny”) that people still think prescribing weak opioids with messy pharmacology that causes more interactions and side effects is the way to prescribe less opioids, though! See shit like dextropropoxyphene (off the market because turns out it’s as good as or better than other opioids at killing people, while being less good at killing pain).
posted by atoxyl at 11:15 AM on August 12, 2021 [20 favorites]




I am rethinking what might have gone wrong with several of my friends who had endometriosis or other sources of chronic pain and wound up buying relief on the street. We all had pets we kept till the end and I don't know any women who were not abused.
posted by Mr. Yuck at 11:20 AM on August 12, 2021 [13 favorites]


I get that this article is about the injustice of the way flags/scores work in the first place, but what's up with the reaction to the flags? Her gynocologist dumped her? Why? That doesn't even make any sense. Presumably the gynecologist had seen actual clinical evidence of her endometriosis and knew it was true. But ok, even if she hadn't...even if she WERE an opioid addict, don't opioid addicts need gynecological care sometimes?

Also, my experience is that I could not walk out of MajorResearchHospital in Toronto without an opioid prescription. I've been offered OxyContin prescriptions after multiple surgeries. I said to the doctor handing me one once, "I'm not going to fill that*. That stuff will ruin your life." and he didn't know what I was talking about. This is maybe 8 years ago. HE DID NOT KNOW WHAT I WAS TALKING ABOUT. That was the scariest thing really. Though withholding a prescription if he had somehow found out my childhood dog was epileptic would not have been a better thing, either.

*Note that my pain was going to be due to surgery and would pass in a few weeks. That's a very different thing from what a lot of people who use opioids for pain experience, I know. I did not fill it because I could tolerate some pain for a few weeks and I had the luxury of staying home and taking it easy. I know that these things do not apply to many people.
posted by If only I had a penguin... at 11:24 AM on August 12, 2021 [11 favorites]


I was in the hospital for over a week once in their equivalent of a high-dependency unit after having had most of a major abdominal organ removed. That first night was the worst pain I've ever had, to the point that I gained a visceral rather than a theoretical understanding of why people would choose euthanasia rather than go on suffering. It's so traumatic I can barely remember it.

There was one nurse who grumbled that I should only be getting some muscle relaxant rather than the Percocet.
posted by praemunire at 11:29 AM on August 12, 2021 [21 favorites]


Yup, I've got a friend with endometriosis that finally culminated with her needing a hysterectomy last year, who also has had many animals with complex medical needs -- and from her descriptions of interactions with doctors, it's very clear that they've frequently suspected her of drug seeking. I wonder if she's got a high NarxCare score. We need better regulations around these obscure, unaccountable "scores" that govern people's lives without their knowledge.
posted by biogeo at 11:29 AM on August 12, 2021 [23 favorites]


Her gynocologist dumped her? Why?

Fear of the DEA, most likely. Less likely to lose their prescribing privileges and get tossed in jail if they can say "we dumped her as soon as we found out she might be abusing pain pills."
posted by wierdo at 11:30 AM on August 12, 2021 [24 favorites]


I am of the firm belief that we need a law that says that for an algorithm to have legal force, its details must be publicly accessible. Don't care if it's your "trade secret" - you want to impact the lives of others, you put your cards on the table.
posted by NoxAeternum at 11:40 AM on August 12, 2021 [96 favorites]


Fear of the DEA, most likely. Less likely to lose their prescribing privileges and get tossed in jail if they can say "we dumped her as soon as we found out she might be abusing pain pills."

Doctors have a lot more leeway than you think. My friend is an gynecologal oncologist and routinely prescribed opiates. A lot of his work unfortunately revolves around end of life care or very invasive surgery. As such addiction isn't an issue.

That said he says that except for short-term treatment he wouldn't prescribe opiates for endometriosis and often other health issues show if endometriosis is present. To be clear he stopped short of calling this story's presentation false and said that he doesn't diagnose from an article, but that there's likely more to this story. Long-term maintenance of endometriosis isn't typically opioids.

He did recoil at the idea that one of his patients would not be filled medication he prescribed, but didn't seem to think NarxCare was an issue, going so far as to question why the doctor didn't follow up with the false positive. It would be akin to a pharmacist denying a prescription and the doctor following up with it -- which is already done and has been done since pharmacists existed. Given my brief time at a pharmacy I routinely saw people turned away and never saw someone that would fit your average middle to upper class white male turned away. So maybe we're trading one bias for another? Determining who is an addict, or what an addict is, and who needs it is difficult.
posted by geoff. at 11:48 AM on August 12, 2021 [1 favorite]


This is maybe 8 years ago.

Yeah that’s about when the (most recent) abrupt shift in the opposite direction happened.
posted by atoxyl at 11:59 AM on August 12, 2021 [1 favorite]


That said he says that except for short-term treatment he wouldn't prescribe opiates for endometriosis and often other health issues show if endometriosis is present. To be clear he stopped short of calling this story's presentation false and said that he doesn't diagnose from an article, but that there's likely more to this story. Long-term maintenance of endometriosis isn't typically opioids.

Perhaps I got the wrong impression from the article, but it sounded to me like the opioids in question were not meant to be taken daily, but on particularly bad days when NSAIDs weren't cutting it.
posted by wierdo at 12:06 PM on August 12, 2021 [7 favorites]


I really want to get on my soapbox about this but I'm afraid that may somehow negatively impact my NarxCare score.
posted by Servo5678 at 12:06 PM on August 12, 2021 [23 favorites]


Equifax to buy Appriss Insights for $1.83bn

I have a feeling they're here to stay.
I am of the firm belief that we need a law that says that for an algorithm to have legal force, its details must be publicly accessible. Don't care if it's your "trade secret" - you want to impact the lives of others, you put your cards on the table.
And you're liable if it turns out your algorithm is racist.
posted by fullerine at 12:14 PM on August 12, 2021 [28 favorites]


Sounds like a new chapter in "Weapons of Math Destruction" where Cathy O'Neil talks about biases built into math models and algorithms due to a variety of reasons and how they affect our world.
posted by kschang at 12:14 PM on August 12, 2021 [11 favorites]


A close family member suffers from a bad back, multiple surgeries. They have been prescribed Vicodin for about 4 years plus. No longer will they give a 3 month supply. Now a new prescription every month after a required doctor visit. At least during the pandemic, telehealth visit. Every year they make this person take a urine test. I asked the doctor what they were testing for. I mean it should be obvious that it will show positive for opiods. That is the point the doctor said. They want to make sure that you HAVE IT in your system and you are not selling them on the street. One year this person had a false positive for marijuana. Trust me when I say it was definitely a false positive. Anyway, that was causing all sorts of conniptions at the doctor's office. I told my relative to offer to retake the test today. THe doctor so did not believe them that they said it would be better and hopefully more accurate if they came back 30 days from now. She was hinting to clean out your system. The relative took the test 2 days later and it proved to be a false positive. THe doctor was more relived than we were. Doctor was worried about prescribing to a drug user. THis is a doctor who is in Pain Management. Must write a lot of prescriptions for opiods. Either that or epidurals.

The system is fooked. Over reaction to a problem that affects the legit people and barely touches the not so legit pill seekers.
posted by AugustWest at 12:38 PM on August 12, 2021 [11 favorites]


Oh, algorithms- is there anything you can't make totally fucking worse?
posted by TheWhiteSkull at 1:28 PM on August 12, 2021 [13 favorites]




I think part of the massive scale of the damage wreaked on us by the Sacklers is this sickening new belief by doctors & nurses that pain relief doesn't exist, has never existed, and is not even important or worth treating. That's what I can't deal with.
posted by bleep at 1:49 PM on August 12, 2021 [19 favorites]


I am of the firm belief that we need a law that says that for an algorithm to have legal force, its details must be publicly accessible. Don't care if it's your "trade secret" - you want to impact the lives of others, you put your cards on the table.

This couldn't hurt, but it's really much more important to surface the data (not the training set, but the data per individual) that went into the score and the data points that were the strongest predictors.

This is probably a hard sell for companies like Appriss, because the data going into the algorithm stands a good chance at being garbage at worst and incomplete at best, and if doctors saw the strongest predicting data points they would see mostly random bits of data that hardly relate to drug abuse.

Predictive scores like this should be treated like a bar chart or a pie chart. Helpful in boiling down information, but not a legitimate substitute for the complete data set.
posted by TurnKey at 1:49 PM on August 12, 2021 [4 favorites]


In remote sensing (photos, LiDAR, radar, etc.) we often do classifications of ground cover using various algorithms. What usually happens is that objects you can see with the eye are left out or boundaries are drawn in what appears to be the wrong place. The upshot is, what ever "algorithm" the eyes and brain is using to classify things is complex and subjective and often disagrees with the simpler, objective mathematical model. For people who don't understand the utility or limitations of models, this is an anxiety-inducing problem. Scientists who do not specialize in remote sensing will often want to reject the model because it disagrees with their eye.

This is usually OK, because we don't really care whether one individual rock or tree is inside or outside of a category, because these models are about population-level statistics: how likely is any given rock, from the set of all rocks, to reside in a category? That's the number we want.

When you apply that logic to people, though, you run into a problem. People, unlike rocks, have individual rights and expectations. You can assess the probability that an individual person resides in a category, but if you are wrong, you've violated their human rights. This isn't a feature of bad models, it's a feature of all models.

tl;dr: mathematical models and human rights are fundamentally incompatible when models are used to target individuals rather than populations. It's an instance of the ecological fallacy.
posted by klanawa at 1:52 PM on August 12, 2021 [72 favorites]


Mefi's own Maias.

And if you like this, you should very definitely read her new book, Undoing Drugs!

Side note -- I continue to be saddened and dismayed by the way mefites throw around all kinds of stigmatizing language and beliefs about people who use drugs and even the drugs themselves.
posted by gingerbeer at 2:02 PM on August 12, 2021 [33 favorites]


I think part of the massive scale of the damage wreaked on us by the Sacklers is this sickening new belief by doctors & nurses that pain relief doesn't exist, has never existed, and is not even important or worth treating.

Yes, although to be honest I think this is a self-serving belief borne out of a) the American Puritan tradition and b) the criminal penalties that doctors face.

Culturally, Americans have always been horribly prone to positive thinking and "mind over matter," especially as it applies to other people. Add this to the popular belief, even among medical people, that black people feel less pain, and you have a recipe for an enormous bullshit cake.

And then there's the fear that a doctor could be criminally liable, not to mention civilly liable, for their prescriptions. I love doctors. I grew up around doctors. They have huge loans and kids. They do not want to go to jail. The calculus is simple.

This article struck me hard because I have almost all of the indicators that would flag me under the NarxCare system, and I have never had a substance abuse problem. Nor would it be right for doctors to treat me or anyone else like a felon if I did. (Really, we shouldn't treat felons like felons, but leave aside.) Plus, I had been thinking the other day about how little pain relief I had when I had dry socket, and yet I had read that people had gone to the dentist for dry socket and ended up dead from opiate addiction. It never even occurred to me that this was a problem that computers could solve, because, as demonstrated above, they can't.
posted by Countess Elena at 2:15 PM on August 12, 2021 [8 favorites]


As a medical professional, I really want to know if anyone with any medical expertise whatsoever is involved in designing and implementing these algorithms. In my wildest fantasies, anyone with a hand in their design who isn’t a doctor/NP/PA gets prosecuted for practicing medicine without a license. Sure the algorithms come with disclaimers that they’re not to be used as prescribing guidelines in theory, but in practice they absolutely are because of the way the DEA surveils providers, so clearly only medical professionals should be involved in their creation. I know those disclaimers probably shield anyone at Apriss from this particular flavor of prosecution, but I will continue to dream.

See also—judges having any hand at all in determining treatment for people who use drugs, including the right to mandate that people not use suboxone or methadone.
posted by I am a Sock, I am an Island at 2:22 PM on August 12, 2021 [24 favorites]


I was surprised there was no mention of surveillance capitalism in the article. I saw one link to the book The Age of Surveillance Capitalism in the comments. There is an absolute need for new law to address this era we have entered. This is Buttle Tuttle with a vengence, for profit.
posted by Pembquist at 2:29 PM on August 12, 2021 [1 favorite]


All this is just another reminder that the War On (Some) Drugs is a civil war that that has been being fought against the population by the government since before the vast majority of us were born (allowing for some outlier pre-prohibition metafites.) It is a war that can never be won, so every setback is met with the typical anger and shame driven doubling-down on prohibition tactics yet again despite them having never worked. The only ones who should be making decisions about medical treatment are the doctor and the patient: not the insurance companies, the police (local, state, national, or extra-national), computer algorithms, politicians, or curtain twitchers.
posted by Blackanvil at 2:36 PM on August 12, 2021 [15 favorites]


Dumping patients because you think theyre misusing opiates is malpractice, no? I mean, wouldn't a more sensible and ethical approach be to change the treatment plan on the basis that you have new data (correct or incorrect is another point) that contraindicates using Percocet or whatever? Pure unadulterated hatred of drug users is all I can see motivating this move.
posted by flamk at 3:00 PM on August 12, 2021 [2 favorites]


Dumping patients because you think theyre misusing opiates is malpractice, no?

How would that be malpractice?
posted by Anonymous at 3:57 PM on August 12, 2021


The war on drugs is a war on people with disabilities.
posted by bile and syntax at 4:15 PM on August 12, 2021 [33 favorites]


I can't emphasize enough that if you are a white cis man, your experience with doctors is vastly different from the experience of everyone else.
posted by hydropsyche at 4:20 PM on August 12, 2021 [38 favorites]


How would that be malpractice?
Because even if someone is addicted to opiates that doesn't make them an un-person and they're not sent into exile. They still are entitled to the same level of care & concern as everyone else. Addiction is something that happens TO YOU as a result of the conditions your big sack of chemistry finds itself in, not some kind of sinful indulgence that deserves the death penalty.
posted by bleep at 4:32 PM on August 12, 2021 [25 favorites]


*And I mean that's IF addiction is the actual problem and not just "needs medicine we decided to reclassify as taboo because we got spooked like a bunch of nervous horses."
posted by bleep at 4:34 PM on August 12, 2021 [10 favorites]


Yeah, last time I checked, opiate misuse is seen by medicine as a disease (opiate use disorder). With what other medically recognized disease would it be considered appropriate and ethical to dump a patient via a mailed letter? Isn't consultation over the treatment plan, obtaining referrals, etc a part of ethical/legal medical practice? Even if the gynecologist felt he couldn't ethically or safely treat this patient shouldn't there be more than just a random letter terminating treatment?
posted by flamk at 5:03 PM on August 12, 2021 [8 favorites]


Many doctors I've seen have just stopped prescribing opiates all together.

As someone with chronic pain which has not responded to a laundry list of other treatments, surgeries, etc, I've mostly run out of options. It's almost impossible to get opiods these days (and yet somehow people are still under the impression that doctors hand them out like candy... lol). Without them it's been incredibly hard to work, and I doubt I will be able to continue much longer. I've pretty much given up on ever having pain relief again.
posted by thefoxgod at 5:21 PM on August 12, 2021 [14 favorites]


“ This is Buttle Tuttle with a vengence, for profit.” Oh, yes!

When you’re done reading Shoshana Zuboff and Cathy O’Neil and, oh, Virginia Eubanks (Automating Inequality) and Ruha Benjamin (Race After Technology) and are pining for a movie, try Brazil. Made in 1985, but prescient about technology’s foibles. A fly falls into a keyboard and changes one letter - Tuttle/Buttle - and that error sets off a cascade of injustices. Though it’s imagining a future War on Terror rather than on drugs it riffs on how such wars tend to amplify technological overreach and dehumanization.
posted by zenzenobia at 6:51 PM on August 12, 2021 [2 favorites]


People should not be denied medical treatment because they are addicts or suspected of being so, obviously. But if a doctor feels the scope of practice is entering addiction medicine and they feel they can't provide that standard of care then I don't think it's a good idea to start threatening them with malpractice suits to get them to do it. Mandating better addiction medicine education in schools so they CAN provide the necessary standard of care, yes. Don't use bullshit black box proprietary algorithms built on trash data to flag "addict potential", absolutely.
posted by Anonymous at 7:14 PM on August 12, 2021


THe doctor so did not believe them that they said it would be better and hopefully more accurate if they came back 30 days from now. She was hinting to clean out your system.

I wish there was a cheat sheet for coded messages like this. I would have totally thought that the test would be more accurate in thirty days. I remember once being told that a particular balcony was a good place to go if I wanted fresh air. I mean, it was a perfectly nice balcony, but apparently that meant “management turns a blind eye to people smoking there”.
posted by Joe in Australia at 7:28 PM on August 12, 2021 [5 favorites]


As a medical professional, I really want to know if anyone with any medical expertise whatsoever is involved in designing and implementing these algorithms.

Whether or not there is (there likely isn't), the more important question is whether the methods used in "the algorithm" are ever updated based on actual evidence (and they most certainly are not!). When it flags a non-drug user as a high likelihood of abuse candidate, is that false positive added to the dataset so it can learn to avoid those? Incredibly unlikely!

We live in a bizarre dystopia where everything is measured, but incorrectly, and the measurements are used to make inferences, but illogically, without applying anything resembling scientific accountability. Data science: The form of science with none of the content. And its conclusions govern our lives and deaths.
posted by dis_integration at 8:07 PM on August 12, 2021 [18 favorites]


Because even if someone is addicted to opiates that doesn't make them an un-person and they're not sent into exile. They still are entitled to the same level of care & concern as everyone else.

If a doctor decides they don't believe the patient is telling the truth about fundamental ailments, they aren't going to treat them for that. It's not like "Oh, you have a broken leg that we can see on an x-ray but you don't deserve treatment."

It's "you've been coming in claiming mysterious pain we can't see and obviously you're doing it for the drugs." The issue is the doctor being confident in this erroneous belief; if the belief were correct stopping treatment would be the ethical choice.

When it flags a non-drug user as a high likelihood of abuse candidate, is that false positive added to the dataset so it can learn to avoid those? Incredibly unlikely!

It almost certainly is being done. People who set up algorithms like this routinely include retraining and reweighting on new data.

The issue is predicting a rare event like opioid addiction is really hard, and it can't ever learn to avoid false positives. Per TFA an independent researcher (Kilby) trying to work with these had a 90% false positive rate even while missing many genuine addicts, numbers which are completely unsurprising to anyone who deals with this stuff.

Scientifically Kilby had a decent algorithm--someone they flagged was 20 times more likely to be an addict than if you picked at random from the general population. If it was something like "locations to take mineral samples from" it would have been useful, you'd save your client lots of money. With humans of course it means you either don't act on the results, or you deny 10 people critical pain medication for every addict you make seek out black market fentanyl or something. (Kilby understands this, but then they weren't selling anything.)
posted by mark k at 11:07 PM on August 12, 2021 [2 favorites]


If a doctor decides they don't believe the patient is telling the truth about fundamental ailments, they aren't going to treat them for that. It's not like "Oh, you have a broken leg that we can see on an x-ray but you don't deserve treatment." It's "you've been coming in claiming mysterious pain we can't see and obviously you're doing it for the drugs." The issue is the doctor being confident in this erroneous belief; if the belief were correct stopping treatment would be the ethical choice.

This doesn't quite match the situation, which is a woman with an existing diagnosis which she is being treated for then getting dropped. So this would be more like, "we saw that broken leg on your x-ray a month ago and we've been treating you for it, but now we've decided that you faked the x-ray somehow so buy-bye."
posted by EmpressCallipygos at 3:49 AM on August 13, 2021 [15 favorites]


I'm a physician. I'm required by state law to look at the state's prescription drug monitoring program (PDMP) before prescribing any controlled substance. (A surprising number of antiseizure medications are controlled. Don't be misled that controlled substances are only narcotics and sedatives.)

My state's PDMP uses Narx scores. I don't care what the score is. I really hope other physicians don't, either.
posted by adoarns at 6:54 AM on August 13, 2021 [3 favorites]


Whether or not there is (there likely isn't), the more important question is whether the methods used in "the algorithm" are ever updated based on actual evidence (and they most certainly are not!). When it flags a non-drug user as a high likelihood of abuse candidate, is that false positive added to the dataset so it can learn to avoid those? Incredibly unlikely!

We live in a bizarre dystopia where everything is measured, but incorrectly, and the measurements are used to make inferences, but illogically, without applying anything resembling scientific accountability. Data science: The form of science with none of the content. And its conclusions govern our lives and deaths.
I think that the book Weapon of Math Destruction by Cathy O'Neil, mentioned above, does a good job of summarizing what takes a model from being merely bad to really destructive. She has three criteria, described in more detail here and here, that this model clearly meets: opacity, scale, and damage. In the book, O'Neil calls out a number of WMDs that are rarely or never re-trained, but still used to make big destructive decisions. Even if they are continuously re-trained, many models create a feedback loop where they can become self-reinforcing - for example, if someone with a high NarxCare score doesn't get access to opiates, then they may need to buy drugs elsewhere to manage their pain, which may be harder to manage and may lead to abuse that wouldn't have happened otherwise*. And then it appears to anyone evaluating the model's predictive power that it was in fact an effective predictor of abuse. And so on and so forth.

*Please forgive my oversimplification of the patterns of abuse and addiction for the purposes of this explanation
posted by mosst at 6:57 AM on August 13, 2021 [2 favorites]


Algorithms to solve it... make it more efficient, machinelike. Remove the human element.

Because the problem with drug addiction and people who have it is that they aren't feeling alienated enough.
posted by symbioid at 8:25 AM on August 13, 2021 [2 favorites]


It's not like "Oh, you have a broken leg that we can see on an x-ray but you don't deserve treatment."

Even in this very thread there are accounts of this very thing just like in every internet thread on this topic.
posted by bleep at 8:56 AM on August 13, 2021 [5 favorites]


Apriss: Types of Personal Information Processed

The Personal Data Content we provide to our Customers is aggregated from public records, publicly available information such as news and business information, and other third-party sources.

Our third-party sources may include data providers, biographical sources, broadcast content providers, social media providers and public source information. In some cases, our source materials contain personal information and, where this is the case, that personal information will also appear in our Personal Data Content.

Such Personal Data Content may include, but is not limited to:

name, age, date of birth, gender, country of residence;

government identification numbers such as social security or national insurance numbers;

employment and education details, which may include details of public, religious, political or trade union roles;

personal and professional affiliations;

information from newspapers blogs, social media and other websites;

financial information relating to income or wealth, such as net worth, bankruptcy or insolvency filings;

inclusion on a sanctions list or a public list of disqualified directors or other positions of responsibility;

criminal background or history;

medical and prescription history;

reported allegations of theft, money laundering, bribery and corruption and similar criminal activities.

posted by snuffleupagus at 9:16 AM on August 13, 2021 [1 favorite]


Also I really am scared of the idea that you are considered guilty of "faking pain" until you can "prove yourself innocent" before you're worthy of a doctors attention. It's really sick & twisted. Just treat the pain and do your best to find out where the pain is coming from. Let people feel safe enough to tell the truth that they're addicted and treat them with kindness & respect and then they won't need to stage an elaborate fiction at all. The fact that doctors don't have time for this or interest in it is the actual problem.
posted by bleep at 9:59 AM on August 13, 2021 [7 favorites]


There was a Sawbones episode on the opioid crisis (link w/ transcript) in which Sydnee talks about her medical training and the push to take pain seriously in the '70s and '80s. "Pain is the fifth vital sign" was apparently a mantra in that period.

The episode was about the crisis and so she is saying this in the context of saying this led to overprescribing medication, especially when combined with oxycontin marketing machine. Looking at the transcript it's interesting that the numbers she quoted then on addiction rates are way higher than the ones in TFA. I'm assuming TFA is more correct, but certainly her numbers would be influencing doctors' behaviors.

Even in this very thread there are accounts of this very thing just like in every internet thread on this topic.

I wasn't saying that doesn't happen; I'm saying there are circumstances where dropping a patient would be ethical (partly in response to the apparent belief that doing so would be "malpractice").

It doesn't excuse doctors for doing so incorrectly. And it makes doctors not believing women/POC worse.
posted by mark k at 10:06 AM on August 13, 2021


A surprising number of antiseizure medications are controlled. Don't be misled that controlled substances are only narcotics and sedatives

Most of those share a chemical class/mechanism of action with “sedatives” though, no? Benzos, barbiturates, other GABA drugs?
posted by atoxyl at 10:53 AM on August 13, 2021


Come to think of it, does this algorithm have any way of detecting false positives and correcting for them?

Recently I was watching some YT videos, and one of the law channels was highlighting how police, supposedly trained in "drug recognition" and certified and all that, basically accused people who have NO history of drug use, NEGATIVE results in a drug test administered at the time of arrest, of being a druggie, and the charges were not dropped for MONTHS, resulting in loss of job and much more.

When challenged by ACLU, the police department claimed their drug recognition experts are more accurate than drug tests.

ACLU lawyers managed to dig up the training, which proved that while the cops can recognize the signs properly in people already determined to be intoxicated (inmates), they were NEVER trained to spot false positives. I.e. ANYONE who exhibited these signs MUST be guilty.

Wonder if something like that is happening here?
posted by kschang at 10:58 AM on August 13, 2021 [2 favorites]


Not to pretend that the schedules make much sense. The difference between I and II is at least clear, if not fair - you got the recreational heavy-hitters without medical approval, and with - but I dare somebody to try to justify III vs. IV, compound-by-compound, as anything but a historical contingency.
posted by atoxyl at 11:03 AM on August 13, 2021


My daughter provides psych care to homeless folks and is able to prescribe Buprenorphine which is an opiate but it's used to treat addiction, like Methadone, but without as much hassle. When I described this article to her she said "oh, that addiction score thingy? Yeah, I see it, but I don't pay attention to it." But she's more into harm reduction than ... monitoring possible off-label drug use to protect herself, I guess? Her patients aren't in a position to doctor-shop anyway. And nobody has given her grief other than pharmacies that just aren't set up to serve people who don't have credit cards and fixed addresses. It can be really hard for folks to fill a scrip, especially if someone at a counter isn't interested in helping them work through a snafu. They're so used to not being helped, they kind of expect it.

But it's creepy to know that this company collects all of this data on just about everyone. We could arrest the Sacklers instead, but how would shareholders make money from that?
posted by zenzenobia at 11:16 AM on August 13, 2021 [4 favorites]


I'm saying there are circumstances where dropping a patient would be ethical (partly in response to the apparent belief that doing so would be "malpractice").
Next time you find yourself in a position of not being unable to obtain medical care that you have a documented need for because of "ethics" I'm sure that will be a cold comfort.
posted by bleep at 2:22 PM on August 13, 2021 [5 favorites]


Presumably the gynecologist had seen actual clinical evidence of her endometriosis and knew it was true.

Unfortunately endometriosis is difficult to confirm definitively without laparoscopic surgery. The primary symptom is pain.
posted by bq at 4:21 PM on August 13, 2021


The episode was about the crisis and so she is saying this in the context of saying this led to overprescribing medication, especially when combined with oxycontin marketing machine. Looking at the transcript it's interesting that the numbers she quoted then on addiction rates are way higher than the ones in TFA. I'm assuming TFA is more correct, but certainly her numbers would be influencing doctors' behaviors.

I think it depends on where they're pulling the studies from. This article titled "Opioid Abuse in Chronic Pain — Misconceptions and Mitigation Strategies" seems like a pretty good overview and in the very last section on minimizing addiction risk there's this:
Although published estimates of iatrogenic addiction vary substantially from less than 1% to more than 26% of cases, part of this variability is due to confusion in definition. Rates of carefully diagnosed addiction have averaged less than 8% in published studies, whereas rates of misuse, abuse, and addiction-related aberrant behaviors have ranged from 15 to 26%.
It sounds like the author of the article was pulling from the study or studies that presented the very lowest addiction risk, whereas McElroy was reading the ones that presented the highest.


Unfortunately endometriosis is difficult to confirm definitively without laparoscopic surgery. The primary symptom is pain.

I never had pain from my endometriosis so the only reason it was discovered is because my surgeon noticed it when she was in there removing my tubes. Diagnostic methods for "lady problems" are pretty shit.
posted by Anonymous at 4:59 PM on August 13, 2021


It's not like "Oh, you have a broken leg that we can see on an x-ray but you don't deserve treatment." It's "you've been coming in claiming mysterious pain we can't see and obviously you're doing it for the drugs."

lol

to be fair I do not have a broken leg, god forbid I should be taken for a liar. no, all I have is six herniated discs, now down to four, two of them having been replaced with prosthetics and the rest just hanging out having a good time on their nerve roots; one botched* rotator cuff repair; two more torn Parts awaiting MRI confirmation. all but the last two have pretty pictures to prove it, and the last two will have pretty pictures very soon. none of these things are mysterious, none of them have causes unknown. I don't even have any distastefully feminine ailments or official mental illnesses or addiction history or tricky diseases to complicate the picture. nothing but a very long list of awful injuries.

no, I cannot and could not get pain medication from any neurosurgeon or orthopedic surgeon or PM&R doc, save for the seven and three days after each surgery, respectively. if you were wondering whether the pain of spine surgery, which involves punching out bits of your neck ligaments and moving your trachea & esophagus several inches out of the way for a few hours, among many much worse things, fades to manageability after three days, the answer is No.

I did get the full slate of painful and useless injections offered by "pain management" these days, because I always do what doctors tell me to do. most of them dangerous when done in your neck and none of them helpful. but hey, you take what you can get.

if you were wondering if the little story you tell yourself above is true in spirit or in generality, let alone in specifics, the answer is No.

I signed a consent form acknowledging that surgery was not guaranteed to fix the pain I was having surgery to fix; acknowledging that, indeed, pain has been known to get worse permanently as a result of these surgeries. Not only has no doctor ever denied the reality of my physical pain or failed to believe in its causes, they believe in it so vehemently they make me sign papers attesting that I believe in it too.

they still don't prescribe any painkillers.

that is because there is no doctor living who values his patient's wellbeing above his own medical license. And I don't even blame them for it, it's perfectly rational. Refusing to take a minor risk on a patient in pain enables them to go on treating thousands of less miserable patients for decades and decades. They ease more suffering in the long run by making some people's problems not their problem, is what they figure, and it's a fair position, I guess. no, I don't blame them, just as long as they don't lie about it.

sorry to correct you at such length but if you've learned anything, it has been well worth it.

*per a different surgeon. you wouldn't want to take my own word for it.
posted by queenofbithynia at 8:36 PM on August 13, 2021 [18 favorites]


Near the end of the article...
Moreover, even among people with known addiction, there is little evidence that avoiding appropriate medical opioid use will, by itself, protect them. “I think undertreated pain in someone with a history of addiction is every bit, if not more, of a risk factor for relapse,” says Wakeman. She calls for better monitoring and support, not obligatory opioid denial.
posted by spamandkimchi at 10:06 PM on August 13, 2021 [11 favorites]


« Older "Shakespeare played [the ghost]...it's the...   |   An organizing motto for their grief Newer »


This thread has been archived and is closed to new comments