J&J's Tylenol-PM Clinical Doomed to Fail? Print E-mail
Written by Donald Riker, PhD   
Monday, 10 May 2010 00:00

The design of clinical protocols has to be right to succeed.  The protocol to support Tylenol-PM, as described in FDA's letter to the CHPA [link], suggests that the clinical protocol was doomed from the start.

Any new attempt must invest in the right vehicle.

Our previous comments on the lack of support for Tylenol-PM (J&J) efficacy was based on experience, not having read FDA's letter to CHPA.  Having now read the letter our comments were well taken and there is now reason to further believe it was a doomed endeavor.  As I see it, as one who has managed over 150 clinical trials of OTC products, several design features were ill advised.  To be fair we have not reviewed the protocol itself.

First, this design relies on the subjects themselves collecting the critical primary data concerning the time it takes them to go to sleep.  Think about it.  How do you accurately know how long it takes you to fall asleep?  Think about last night.  How long did it take?  It gets worse...

In this protocol the time it takes to fall asleep is what's identified as the "primary variable", the number to which the whole study is committed.  Obviously it's hard to punch a stopwatch noting when you fall asleep since you are already, in fact, asleep.  Since there is no one watching you then all you can do is try to guess when you fell asleep the night before.  It's cheaper to do it this way.  It turns out sleep researchers have compared insomniac's estimates of sleep latency (time to sleep) with real data taken by observers.   Invariably the insomniac overestimates sleep latency (time to sleep).  Only 20% of these estimates are accurate.  Some even underestimate it [Vanable et al, SLEEP, Vol. 23, No. 1, 2000] .

But wait, this study not only tested insomniacs but it evaluated an anti-histamine with a long "hang time" at high doses (50mg) known to sedate people starting 3 hours later.  The effect of a sedating anti-histamine on memory and recall the next morning badly biases any study design that relies on the subject's recall.  Since not all subjects took the anti-histamine the study suffers from a bad case of "recall bias".

Typically it takes people about 15min to fall asleep.  This study focused on subjects with pain from osteoarthritis and who had trouble falling asleep (takes 64min).  That aspect of the trial appears well done.  However, each active reduced sleep latency by about 14%, from 64 to about 56 minutes, hardly a big improvement and far from the normal period.  The measure of success of any drug therapy must be its clinical significance, or in this case what the investigator found to be what users considered a significant degree of benefit before initiating the study.

Lastly, the comments in our previous story [J&J's Other Tylenol Problem: Tylenol-PM] remain.  The characteristics of the two drugs in this combination, acetaminophen and diphenhydramine, are different enough in their timing-to-effect as to further bias any study that must focus, by definition, on the first hour that these subjects seek sleep.  The analgesic acetaminophen is more likely to reduce pain in this first hour than is diphenhydramine likely to induce meaningful sedation.

There is a reasonable chance, if truth be known, that this combination of an analgesic and an old anti-histamine is not entirely rational.

[Dr. Riker is a former Associate Director, Personal Healthcare at Procter & Gamble, and Chattem's last Vice President of R&D & Chief Scientific Officer.  He is a member of the American Society of Pharmacology and Experimental Therapeutics, American Society of Clinical Pharmacology & Therapeutics, and the Society of Neuroscience]

OTC Product News logo

On Point Advisors, LLC, or its affiliate OTC Product News, provide unbiased opinion on companies, products and events in the OTC industry and receive no financial incentive to promote or influence its opinion of any company or product, and disclose and disclaim any potential conflict of interest herein  (Read full disclaimer)