Bosses Harness Big Data to Predict Which Workers Might Get Sick | Hacker News Reader

Bosses Harness Big Data to Predict Which Workers Might Get Sick

wsj.com

76 pointsuptown54 comments

"Castlight recently launched a new product that scans insurance claims to find women who have stopped filling birth-control prescriptions, as well as women who have made fertility-related searches on Castlight’s health app."

Profoundly creepy.

I think that's an understatement. Unless that data is completely anonymized, isn't it a violation of HIPAA? Isn't this illegal?

HIPAA is probably a stretch - but certainly sounds like someone's asking for an employment lawsuit.

> Unless that data is completely anonymized, isn't it a violation of HIPAA?

It sounds to me from reading various articles that Castlight is used in various ways, including to provide information and recommendations to employees. This use is probably fine under HIPAA.

Information that gets sent back to employers that uses PHI from insurance claims, absent specific consent for the use intended, would seem on first glance to be a HIPAA violation on the part of the insurer.

And, actually using it in a discriminatory way (which paying to gather it would be prima facie, though not dispositive, evidence of) would violate a whole lot of employment law on the employers part.

Correct. PHI coming from insurance claims absent a signed HIPAA release for uses related beyond "treatment/medical education, payment, and operations" would be a HIPAA violation on the part of the insurance company.

Employers themselves are not HIPAA-covered entities. Insurers are however.

This is the part where I hope someone comes up with a charitable interpretation or has inside knowledge that refutes the claim. Wow.

So... Can employers use this to predict which employees might become depressed? How about which employees might get pregnant? (https://twitter.com/mims/status/699964784783990789)?

It’s a Brave New World of work. At Walmart, of course.

If the wellness program is offered as part of a group health plan, then the wellness-plan provider is a business associate of a covered entity under HIPAA, and PHI protection applies. So in these cases, the employer cannot have access to individual employees' health information and, more to the point, they usually don't want it -- they're trying to cut overall costs, not individual employees.

Having said that, I do think that more robust legislative firewalls are needed to prevent companies from trying to push the envelope in accessing individual employees' records (HIPAA controls covered entities' actions, but not employers').

Thank you for the elucidation. This is the kind of quality reply that makes HN great.

The problem is that you do have employers wanting to target individual employees and that just because something is illegal doesn't mean it won't happen. An employer could use this to identify what employees are about to become pregnant and then find reasons to fire them. And given that being fired is likely to change their chance of choosing to become pregnant, it becomes significantly harder to prove this was an illegal targeted attack instead of a justified firing of an employee who was violating policy (and given how policies are written, workplace policies are treated like laws these days where everyone is violating some policies and they are only selectively enforced). Add in that this is most likely to be deployed against lower income people if it is deployed, and that those it is deployed against will be in a very poor condition to fight back (hiring a lawyer right after losing a job isn't something most can afford), you have a nasty discrimination disaster ready to happen (or in reality, ready to be made worse than it currently is).

Policies that are 'selectively' enforced have limited value as protection from a lawsuit.

If it can be proven that there is no reason behind the selective enforcement and if it can be shown that it is selectively enforced. The thing is, the people most likely to be a victim of these policies are the least likely to afford the discovery to prove this, to say nothing of the other side effects (such as being black listed from an industry).

Caveat: In my experience, I've been subjected to a phone call from a Registered Nurse who reviewed my personal health records and called to ask me to switch to a cheaper medication because the company was going through some financial difficulty. I'm 100% not making this up. Your assertion that a company doesn't want personal health information of individual employees is one that I do not agree with.

If a company can eliminate the 5% of people driving XX% of the premium cost - through cancer, major surgeries, etc - then it stands to reason they would undertake that method of reducing overhead.

The employer itself may not have access, but the party they hired may have access, and I do agree that robust legislative firewalls and significant jail-time for violators should be a national priority.

Or, in other words, I very much disagree "Wellness Programs" are actually concerned with wellness (no assistance with dietary materials, gym memberships, incentives, etc) and should be portrayed as "Health Cost Containment Programs" which is significantly more honest.

Who was this RN working for exactly? Your PCP? Your insurance company? A broker? A third party?

I doubt they are employed at the company you work at, meaning your employer isn't look at your health record.

Employed directly? Maybe not. Employed by a company the employer asked to look into reducing health plan costs? Most likely. And just as creepy and scummy.

Or it could be their PCP's RN asking them to switch their ongoing Rx to a generic, because recent updates to their health plan means a higher co-pay or deductible.

This happens more than you realize, and doesn't mean their employer violated their privacy, which is why I ask.

Then they would have introduced themselves as such.

Then it's an easy question to answer, isn't it? Any more insights into their phone conversation, st3v3r?

Edit: I knew I recognized you, s73v3r - https://news.ycombinator.com/threads?id=s73v3r

I believe your scenario as described would be an accurate assessment of what I perceived to be happening. As in, a hired firm that could legally review records as medical professionals and then make that type of call. I'm a well-educated health consumer so I knew the product I was being requested to use was actually more risky long-term to my personal health, which I thoroughly explained at probably too loud a volume to be considered polite during the course of the call.

> If the wellness program is offered as part of a group health plan, then the wellness-plan provider is a business associate of a covered entity under HIPAA, and PHI protection applies. But whether that prevents disclosure depends on what specific consent is given as part of opting-in to the offered wellness plan.

As I understand, these plans often include consent for fairly extensive information sharing with employers as part of the opt-in paperwork (and that that is a big part of the reason that they are offered.)

Mims tweeted a correction shortly thereafter, saying that the employee gets the alert, not the employer [1]. However, employers do apparently get anonymized, aggregated data [2]. Still not a great situation but not quite as bad as originally reported.

[1]https://twitter.com/mims/status/700012777767698432

[2]https://twitter.com/mims/status/700018880022867968

I wonder how they anonymize and aggregate it. Given enough information and an established relationship (like a manager might have with you) you can make really solid educated guesses even though the data seems anonymous to a third party.

Yeah, it still seems ripe for abuse to those who are so inclined.

Should we swap out "Big Data" for "Personal Information" or "Spying" in the title of this article?

How is it spying if it's information that's already available to the company, either publicly or as a result of the person being an employee? It's not like they're sending goons to dig through trash cans.

Sorry, I agree, spying may be a stretch. But so too is Big Data. It doesn't take a complex algorithm to determine IsBuyingBirthControl vs. IsNotBuyingBirthControl. I would like to think that I have the option to keep this information private from my employer. The employer's best interest is to minimize costs not to improve my health, most of the time these two correspond but not always.

Whether or not it's called big data is non-sequitur. Employers have had access to claims data before the tools were available to process it en mass.

"How is it spying if it's information that's already available to the company, either publicly or as a result of the person being an employee?"

Inference might be a better word.

No amount of Economic benefit is worth the invasiveness of these practices. My boss doesn't need to know what medications I'm on, what kind of food I eat, or how many videogames I'm buying. If it's not occurring on company time it's none of his business, end of discussion.

This creepy ass practice needs to be stopped now. If my employer ever tries anything like this, I will quit on the spot. No redemption, no change, nothing will bring me back because once it's been done one time, it WILL be done again.

Sidebar: This is why I much prefer working at a small business where employees are treated like people, not resources to be extracted and then cast aside when they no longer produce enough.

This is why, after years of living in the US, it still staggers me that people think your employer controlling your access to healthcare is normal and not anything to be concerned about.

...whilst maintaining the conditions that actually make them sick.

This version of the article wasn't paywalled for me: http://www.nasdaq.com/article/bosses-harness-big-data-to-pre...

It is great to think that my employer is paying for this but just got rid of our gym membership reimbursement.

Either good ol' right hand/left hand lack of coordination/direction/information sharing or it wasn't "cost effective". Either way, you have to love corporate bullshit.

Ever heard about the manager who figured that about 40% of sick leaves take place on Mondays or Fridays?

Simple, we also need a : 'Employees Harness big data to predict which employers are most likely to do crummy predictions'

"We're not firing you because you're pregnant, we're firing you to cut costs!"

PS. We're are going to claim your work was sub-par to also save on unemployment!

They're trying to kill me," Yossarian told him calmly.

No one's trying to kill you," Clevinger cried.

Then why are they shooting at me?" Yossarian asked.

They're shooting at everyone," Clevinger answered. "They're trying to kill everyone."

And what difference does that make?

Catch-22

That would still give them unemployment benefits. We are firing them because they violated policies A (must be on time), D subsection 12 (no personal internet use), and Q subsection 7 clause 3 (15 minute break per every 4 hour shift). (That most people are allowed to get away with minor violations of the above as long as they are productive, that the fired employee was productive enough that we would've continued to ignore their minor violations, and that they are actually being targeted because we have information to indicate they are trying to become pregnant is going to be near impossible to prove. Especially since the fired employee is now unemployed and not likely able to afford the legal battle.)

Harness big data, such as typical flu season dates? Or maybe weather forecast?

What, do you need precision down to the last day for planning purposes?

So all I have to do is vote in a mid-term election and I'll be healthier?! </s>

I couldn't get past the paywall, but let me take a rough guess: those working in an open office after an illness hits the schools were most susceptible. First the parents, then a few waves of those who caught it from them.

Based on experience from the last few months, two major illnesses (one viral, one bacterial) flowed through the schools and into the parents, who came to work and subsequently took out 3/4 the workforce.

Nothing unexpected, sadly.

No, your post has nothing to do with the article, which starts with people getting back surgery (or not) depending on which doctors they see.

Walmart spotted that, and decided to see if they can predict other stuff. They now send messages to people "at risk of diabetes"

> Employee wellness firms and insurers are working with companies to mine data about the prescription drugs workers use, how they shop and even whether they vote, to predict their individual health needs and recommend treatments.

> Trying to stem rising health-care costs, some companies, including retailer Wal-Mart Stores Inc., are paying firms like Castlight Healthcare Inc. to collect and crunch employee data to identify, for example, which workers are at risk for diabetes, and target them with personalized messages nudging them toward a doctor or services such as weight-loss programs.

It has some other stuff too:

> Welltok Inc., whose clients include Colorado’s state employees, has found that people who vote in midterm elections tend to be healthier than those who skip them, says Chris Coloian, the firm’s chief solutions officer.

There are plenty of risks:

> “There are enormous potential risks” in these efforts, such as the exposure of personal health data to employers or others,” says Frank Pasquale, a law professor at the University of Maryland, who studies health privacy.

I'm reminded of the supermarket loyalty card scheme that was outing pregnant women to their family by sending them vouchers for baby products (nappies etc). Some of these women were young adults living with their parents. (I can only find a link to a US story about Target: http://www.nytimes.com/2012/02/19/magazine/shopping-habits.h... But I think it's happened in the UK with Boots too.)

There is also the risk of ruining the signal. If people who vote are more likely to do that because they are less likely to be depressed and non-depressed individuals are more likely to take care of themselves, then encouraging people to vote will only ruin the value of knowing if they are likely to vote, it won't cause anybody to need less health care.

We get random boxes of baby formula and wipes after having a registry at target.

side note: It's pretty hard to have a decent discussion when pay wall links hit the front page. Half of us can't even read them. There is probably a way around it but not worth my time.

I looked up workarounds today, and it's as simple as googling the article title, then clicking the link.

Or, on HN, clicking the "web" link, and then clicking one of the results.

Did not know the web link trick.

  I couldn't get past the paywall, but let me take a rough guess

Why? It's never likely to add much value, but quite likely to be pointless. (and unsurprisingly, you were completely off base)

>Nothing unexpected, sadly.

It shouldn't be too surprising that the version of the article you made up in your own mind said what you expected it to say.

Hint: those workers who regularly stay out all night drinking alcohol will miss many more days of work.

Reply on news.ycombinator.com