I came across this 2002 article by Richard Altschuler that answers the following questions I’ve had for a while:

Does the expiration date on a bottle of a medication mean anything? If a bottle of Tylenol, for example, says something like “Do not use after June 1998,” and it is August 2002, should you take the Tylenol? Should you discard it? Can you get hurt if you take it? Will it simply have lost its potency and do you no good?

In other words, are drug manufacturers being honest with us when they put an expiration date on their medications, or is the practice of dating just another drug industry scam, to get us to buy new medications when the old ones that purportedly have “expired” are still perfectly good?

These are the pressing questions I investigated after my
mother-in-law recently said to me, “It doesn’t mean anything,” when I
pointed out that the Tylenol she was about to take had
“expired” 4 years and a few months ago. I was a bit mocking in my
pronouncement — feeling superior that I had noticed the chemical
corpse in her cabinet — but she was equally adamant in her reply, and
is generally very sage about medical issues.

So I gave her a glass of water with the purportedly “dead” drug, of
which she took 2 capsules for a pain in the upper back. About a half
hour later she reported the pain seemed to have eased up a bit. I said
“You could be having a placebo effect,” not wanting to simply concede
she was right about the drug, and also not actually knowing what I was
talking about. I was just happy to hear that her pain had eased, even
before we had our evening cocktails and hot tub dip (we were in
“Leisure World,” near Laguna Beach, California, where the hot tub is
bigger than most Manhattan apartments, and “Heaven,” as generally
portrayed, would be raucous by comparison).

Upon my return to NYC and high-speed connection, I immediately
scoured the medical databases and general literature for the answer to
my question about drug expiration labeling. And voila, no sooner than I
could say “Screwed again by the pharmaceutical industry,” I had my
answer. Here are the simple facts:

First, the expiration date, required by law in the United States,
beginning in 1979, specifies only the date the manufacturer guarantees
the full potency and safety of the drug — it does not mean how long
the drug is actually “good” or safe to use. Second, medical authorities
uniformly say it is safe to take drugs past their expiration date — no
matter how “expired” the drugs purportedly are. Except for possibly the
rarest of exceptions, you won’t get hurt and you certainly won’t get
killed. A contested example of a rare exception is a case of renal
tubular damage purportedly caused by expired tetracycline (reported by
G. W. Frimpter and colleagues in JAMA, 1963;184:111). This
outcome (disputed by other scientists) was supposedly caused by a
chemical transformation of the active ingredient. Third, studies show
that expired drugs may lose some of their potency over time, from as
little as 5% or less to 50% or more (though usually much less than the
latter). Even 10 years after the “expiration date,” most drugs have a
good deal of their original potency. So wisdom dictates that if your
life does depend on an expired drug, and you must have 100% or so of
its original strength, you should probably toss it and get a refill, in
accordance with the cliché, “better safe than sorry.” If your life does
not depend on an expired drug — such as that for headache, hay fever,
or menstrual cramps — take it and see what happens.

One of the largest studies ever conducted that supports the above
points about “expired drug” labeling was done by the US military 15
years ago, according to a feature story in the Wall Street Journal
(March 29, 2000), reported by Laurie P. Cohen. The military was sitting
on a $1 billion stockpile of drugs and facing the daunting process of
destroying and replacing its supply every 2 to 3 years, so it began a
testing program to see if it could extend the life of its inventory.
The testing, conducted by the US Food and Drug Administration (FDA),
ultimately covered more than 100 drugs, prescription and
over-the-counter. The results showed that about 90% of them were safe
and effective as far as 15 years past their original expiration date.

In light of these results, a former director of the testing program,
Francis Flaherty, said he concluded that expiration dates put on by
manufacturers typically have no bearing on whether a drug is usable for
longer. Mr. Flaherty noted that a drug maker is required to prove only
that a drug is still good on whatever expiration date the company
chooses to set. The expiration date doesn’t mean, or even suggest, that
the drug will stop being effective after that, nor that it will become
harmful. “Manufacturers put expiration dates on for marketing, rather
than scientific, reasons,” said Mr. Flaherty, a pharmacist at the FDA
until his retirement in 1999. “It’s not profitable for them to have
products on a shelf for 10 years. They want turnover.”

The FDA cautioned there isn’t enough evidence from the program,
which is weighted toward drugs used during combat, to conclude most
drugs in consumers’ medicine cabinets are potent beyond the expiration
date. Joel Davis, however, a former FDA expiration-date compliance
chief, said that with a handful of exceptions — notably nitroglycerin,
insulin, and some liquid antibiotics — most drugs are probably as
durable as those the agency has tested for the military. “Most drugs
degrade very slowly,” he said. “In all likelihood, you can take a
product you have at home and keep it for many years, especially if it’s
in the refrigerator.” Consider aspirin. Bayer AG puts 2-year or 3-year
dates on aspirin and says that it should be discarded after that.
However, Chris Allen, a vice president at the Bayer unit that makes
aspirin, said the dating is “pretty conservative”; when Bayer has
tested 4-year-old aspirin, it remained 100% effective, he said. So why
doesn’t Bayer set a 4-year expiration date? Because the company often
changes packaging, and it undertakes “continuous improvement programs,”
Mr. Allen said. Each change triggers a need for more expiration-date
testing, and testing each time for a 4-year life would be impractical.
Bayer has never tested aspirin beyond 4 years, Mr. Allen said. But Jens
Carstensen has. Dr. Carstensen, professor emeritus at the University of
Wisconsin’s pharmacy school, who wrote what is considered the main text
on drug stability, said, “I did a study of different aspirins, and
after 5 years, Bayer was still excellent. Aspirin, if made correctly, is very stable.

Okay, I concede. My mother-in-law was right, once again. And I was
wrong, once again, and with a wiseacre attitude to boot. Sorry mom. Now
I think I’ll take a swig of the 10-year dead package of Alka Seltzer
in my medicine chest — to ease the nausea I’m feeling from calculating
how many billions of dollars the pharmaceutical industry bilks out of
unknowing consumers every year who discard perfectly good drugs and buy
new ones because they trust the industry’s “expiration date labeling.”
___________________________________________________

The above article appears on the MedScape website after this introduction by Thomas A. M. Kramer, MD, in Aug 2003:

This month’s Psychopharmacology Today column will be our second guest
column. It is a piece that has been available on the Web for about a
year but was brought to my attention recently. It answers a question
that I have asked and been asked multiple times. Before I found this,
no one had ever given me a straight answer about what the expiration
dates on medications mean and how seriously they should be taken. This
is an important issue, and I think that psychopharmacologists, if not
all practitioners and patients, will find this column immensely
helpful. It is well researched, well written, and I wish that I had
written it myself.