Release Bad News on a Friday

Politicians have long known to release bad news on a Friday and it appears that pharmaceutical firms may do likewise.

Safety alerts are announcements made by health regulators warning patients and doctors about new drug-related side effects. However, not all safety alerts are equally effective. We provide evidence that the day of the week on which the safety alerts are announced explains differences in safety alert impact. Specifically, we show that safety alerts announced on Fridays are less broadly diffused: they are shared 34% less on social media, mentioned in 23% to 66% fewer news articles, and are 12% to 51% less likely to receive any news coverage at all. As a consequence of this, we propose Friday alerts are less effective in reducing drug-related side effects. We find that moving a Friday alert to any other weekday would reduce all drug-related side effects by 9% to 12%, serious drug-related complications by 6% to 15%, and drug-related deaths by 22% to 36%. This problem is particularly important because Friday was the most frequent weekday for safety alert announcements from 1999 to 2016. We show that this greater prevalence of Friday alerts might not be random: firms that lobbied the U.S. Food and Drug Administration in the past are 49% to 56% more likely to have safety alerts announced on Fridays.

From The Friday Effect in Management Science by Diestre, Barber and Santalo.

Hat tip: Kevin Lewis.


This post is bad news,


it's Friday.


Part of the reason for Fridays being the most common day is probably that people are more likely to put things off from one week day to the next than from Friday to Monday.

My first thought also.

The day of the bad news release not being random does not imply intentionally timing the release for least exposure.

Friday is the natural end of a work week, and many projects or tasks use that as a deadline. It also has a psychological finality that allows one to enjoy the weekend without lingering guilt of unfinished work.

Similarly, many places set end of year deadlines as a psychological goal, milestone deadline; accounting, tax, reporting, or performance eval closure; and finality before long holidays.

Yes, but it being correlated with lobbying might well suggest it's deliberate.

Moving announcements to another day would reduce drug related deaths by 22%!?!?!? Doesn't this fail a basic sanity check?

Yes it does fail a basic sanity check. I suspect that there is some math based upon some wild ass guessing for core assumptions behind it all.

They probably aren't great at conveying the idea, I would think that the idea is "drug-related deaths DUE TO KNOWN SIDE-EFFECTS". And yes, that does not seem so far-fetched. If people are 10 to 50% less likely to know about a certain (fatal in this case) side-effect, then they are more likely to die. Sure, most drugs do not have fatal side-effects. But for those, that we suddenly learn certain bad thing, making it widely known is most important. Basic sanity is quite correct there.

That still doesn't pass a basic sanity check. What they are saying that people are up to 50% less likely to know and up to 22% more likely to die, because the information was released upon Friday instead of being pushed off till the next Monday. That's just dumb.

People/services compile databases of critical drug information. It doesn't fail to make it in because they kick off early on Friday and never check last week. The idea is dumb.

Heh. I'm not going to defend the numbers because I don't know. But disabuse yourself of the notion that compiled databases of critical drug information are adequate. Anyone in any field with a certain level of complexity cannot keep up with the flow of information. If an important bit of information doesn't attract attention it is lost in the flow.

I read a book about 9/11 and the author said that multiple government agencies across the western world published or released bad news the next day knowing it wouldn't be seen.

"But disabuse yourself of the notion that compiled databases of critical drug information are adequate."

How are they adequate for Monday through Thursday but not magically for Friday?

Let me tell you a story. A few years ago one of the manufacturers in my industry had a flaw in one of their major components. It manifested itself by not working. I saw 5 of them, and with 4 of them I got a new component and installed it. By the time I saw the fifth, some information came out where I could hit the thing with a 2x4 and it would work. I hit it with a 2x4 and it worked, so I didn't change it out. Something would seat correctly if it was hit, but otherwise it didn't work.

I found out later that the information had been available about 6-8 months before I got it, and my suppliers didn't see it until they started asking questions seeing a bunch of returns.

That is the real world. Having a database is fine. A doctor will prescribe how many times over a day? Here doctors have about 15 minutes for each consultation. If something is headline news they hear about it, if it isn't it takes time for the information to trickle down throughout the system.

Situations like this are descriptive; they show how the system actually works, not the way it is supposed to work. Quibble about the numbers, but the real communication channels that affect decisions are not very effective. These are very hard problems to solve.

Your arguing a different thing. Your arguing that a database missed some data for a long time. Sure databases will miss a certain amount of data. That's normal. But why are they capturing all the Monday -Thursday data to such a greater extent than the Friday data.

It's not as if the people that collate the data knock off early on Friday and never record anything. Even if they did, they would see the data on Monday. There's a time lag to everything and there always going to be entering data from the previous days, weeks, months, etc.

I could possible believe some slight difference but these numbers are implausibly large.

Based on these numbers the overall amount of attention paid to Friday announcements is the same as other days, there are just more of them.
(1-0.34)*1.53 = 1

So it’s not really a “Friday effect” just a clustering effect. Plausibly spreading the announcements would just lead to other announcements being diluted

Ok, that could explain it. Perhaps there tends to be more data and it's not as detailed. Maybe people responding to a deadline to get the data out the door and not being as complete as the average output on other days of the week by people who aren't against a deadline?

Why? Do folks not read, listen to, or watch the news on Fridays? Or is it that folks don't want bad news on Fridays because it might ruin the weekend? Or is it that so much bad news is released on Fridays that folks can't absorb it all? Or is it that bad news is often framed as though it's good news so folks don't know it's bad news they need to know? I understand that Trump will have a "fireside chat" on live tv to explain his excellent call to Ukraine. Will the "fireside chat" be today (Friday)? If so, will Trump's explanation of his call be watched only by his loyal followers and construed as the excellent call it and all of Trump's calls are? Will Trump explain that those listening in on his excellent call committed a crime for eavesdropping and will his excellent attorney general prosecute them?

"Safety alerts are announcements made by health regulators warning patients and doctors about new drug-related side effects."

Well, Alex, who ARE these health regulators, in which case?

Are they ALL pharmaceutical industry employees? (Are ANY employed by any health insurance companies?)

Are ANY of them "government employees"? (If so, can the entire sorry class of "government health regulators" be duly FIRED?)

These are not ordinary "product recalls": the class would seem to comprise any number of announcements amounting to MEDICAL EMERGENCIES. ("Oh, but let's wait until Friday . . .")

Culpable corrupt and corrupting Media Establishment executives seem to play an ample role in the engineering of tardy American "health regulation", too, who'd've thunk?

Recalls are generally voluntary, and initiated by the company, not the regulators. Especially in food.

It's actually horrifying to learn how hard it is for the regulators to force a recall on a recalcitrant company. Known defects and contamination incdetns are allowed to perpetuate for weeks/months/years until finally a tipping point is reached and the company decides finally to do the recall.

Read the notices carefully and you will see the language" Company A announces a recall..." The regulator is merely publishing the company's announcement.

I guess I'm amazed that our food hazard warning system works as well as it does. (Although I continue to wonder whether or when we can anticipate alerts to "microplastic contamination".)

We seem to get timely warnings of outbreaks of E. coli, listeria, and salmonella contamination. Who are our trusted food safety regulators? If their processes and systems work as well as they do, how could the pharma product alert system be so shoddy by comparison? (With ALL the brilliant geniuses working in our trusted pharmaceutical industries?)

American systems engineering seems to be yielding expectations that a single default mechanism will catch whatever goes astray within whatever system. I begin to lose the impression that the multiple stakeholders are exercising their due interests in seeing systems and their products or outcomes well-managed: whatever happened to overlapping constituencies?

My views may already be quaint: but if I were a pharma exec, I'd want my prescription-prone public as aware as possible. If I were a "health care regulator", I would want to be as timely in my alerts to the public as any producer or distributor of tainted food. Even if I were a corrupt and corrupting Media executive, I might want to perform at least one public service in my lifetime and help see that the public is alerted in timely fashion to whatever pharma product warning is being issued.

"We seem to get timely warnings of outbreaks of E. coli..."

Actually, no. The truth is horrifying. By the time you hear about it, industry insiders, regulators, etc often have known for a (relatively) long time that there was a problem.

Study as an example the peanut recalls. People went to jail over one of those.

I suspect that the last thing on earth that a food/drug maker wants - literally the last thing - is to announce a recall. It destroys brands, kills companies, and sends people to jail. At least on the food side. I think maybe the drug guys are like Wall Street bankers.

And while field inspectors may actually take their jobs seriously, their superiors are political appointees with an eye towards the revolving door.

Sure, the press loves to print recalls, part of the drum beat of fear and hyperbolic drama they like to push out the nightly firehose. But they aren't usually out crusading for better sunshine laws or doing proactive investigations.

I presume that large healthcare organizations have people who watch such announcements, and that they don't miss them for a Friday. It's a strength of organizational, institutional, medicine.

The patient management database is updated.

I further presume that small practices or hospitals handle such thing in a more ad hoc basis.

I'd cap my presumption with the confirmed bias that HMOs rule, PPOs .. not so much.

As medicine increasingly becomes data management, you need an organization of sufficient size to handle it.

I don't think you know what the difference is between a PPO and an HMO.

I think I do, and I think I know the implied difference in data management.

The same organization can offer both PPO and HMO plans. I can't speak with absolute certainty, but there's no reason the data management practices of the organization would differ materially, whether a patient is enrolled in either a PPO or HMO plan.

An HMO is a single-contact organization. It manages not just appointments across its operations, but also the complete patient record. When you call the toll free number, the duty nurse has access to that full patient record, etc. And it's more than data, it's an expert system. Anyone who sees you is guided by the embedded decision architecture.

Contrast that to a PPO, where you pick preferred physicians. Who is preeminent? Who has all the data? Who is implementing a decision architecture?

That is not what your typical HMO is. It would apply to organizations like Kaiser, where they're both your insurer and care provider, but that's not all or even most HMO's. What you're referring to is called an ACO, ICO, or sometimes a Medical Home, not an HMO.

Oh, well. Kaiser was certainly my model for understanding the strengths of integrated care.

If "most don't" provide this level of data management that seems, to this old programmer, a real problem.

I should note that while I did spend some years in biomedical engineering, that was some time ago. Current practitioners are going to have more current data. But again being clear that this is my presumption, I'd presume that data interchange practices (especially at small practices and hospitals) have not yet evolved to incorporate expert systems in they way they have at HMOs.

This morning, I was talking with [Redacted], my daughter, who is 8 years old. She told me her best friend, [Redacted] told her Representative Gabbard is not electable because she far to the left of America' current Zeitgeist. She asked me if it is true.

I wonder what schools are teaching children these days. Representative Gabbard's views are essentially in line with the common American's ideas and aspirations. Representative Gabbard supports a people' capitalism, where Wall Street can't control our society. She will bring American jobs back.

She opposes dictatorships. She supports a strong military. She will be the first President since 1992 with any meaningful military experience. She has vowed to "bring back a government that is truly of, by, and for the people!"

She supports "a new century free from the fear of nuclear war, in a world where there is real peace, where our people have time to pursue happiness rather than being forced to work constantly just to survive, where parents have time to spend with their children, and where we build strong communities that care for each other and the planet".

Your 8 year old daughter and her friends sure do follow politics closely! And they know what Zeitgeist means. You're doing a good job raising her!

I think it is the Zeitgeist. Everyone is worried about America' future.

By the way Alex,

"Almost 800,000 customers are left without power from Maine to Virginia"

Just because of a little cold? Speaking as a Californian, and an instant expert on other people's problems .. can't they just heat the wires?

Truly the East Coast is a failed state.

I can not imagine it happening in Brazil.

I can not imagine all the crime I experience happening in America.

Actually crime is pretty rare in Brazil, and the crime rate ia falling likena stone.

What's the best day to release Fake News?

Don't leave for tomorrow the fake news you can release today.

You are such a SnowFlake.

Actually, that's not really correct. No, you are a partisan whiner. You haven't demanded that Fox News be removed from the air which is what a SnowFlake would do. You just whine that it exists.

I'd say that they guy who can't take the casual joke is the "snowflake," wouldn't you?

And SQL is easy.

Not bad anon. Linking to Hannity would have been too obvious but linking to the morning “news-lite” program was well played. I’ll give ya a thumbs up on that one.

Well spotted Prior. It appears Mr Lewis has lost a little status.

I*f no work is done week ends then the arrival of side effect reports will accumulate on Monday thru Wednesday. With one day to prepare,the notices go out Thursday and Friday.

The agglomeration effects of culturally defined weeks cannot be avoided and this is the Null hypothesis for all businesses.

So the question is: Do pharmacuticals reinforce this more or less than expected from the Null hypothesis? I doubt there is a noticeable effect. But the authors are really pointing to the culturally defined work week, not pharacuticals in particular.

So in other words. The post says that pharma companies should act more like emergency fire departments, deliberately violating the cultural work week.

Comments for this post are closed