The trusted source for
healthcare information and
Patient adherence data is key to improve trial efficiencies
Nonadherence poses risks to study results, patients
Researchers and clinicians like to believe that patients will do as they instruct, taking their medications, showing up for appointments, and keeping notes on symptoms, etc.
But the reality is that adherence can be far lower than imagined, even for a highly-educated population, experts say.
Thirty years ago, a study on glaucoma patients and adherence found that only 25 percent of university-educated patients with the eye disease would take their eye drop therapy as prescribed, says John Urquhart, MD, DrHC, FRCPE, FAAAS, FISPE, FBMES, FRSE, chief scientist with AARDEX Ltd. of Union City, CA, and Zug, Switzerland. Urquhart also is an adjunct professor of biopharmaceutical sciences at the Center for Drug Development Science, University of California - San Francisco, CA, and an emeritus extra-ordinary professor of pharmacoepidemiology at Maastricht University in Maastricht, the Netherlands.
This adherence study was possible because of the use of a bulky electronic device in the eye drop bottle, and its results stunned ophthalmologists, Urquhart says.
"Nobody saw what real patients do with their medicine before that," he says.
In principle, research participants are supposed to follow the instructions in the protocol, but to what extent do they do this and what are the implications of deviations, asks Carl C. Peck, MD, an adjunct professor at the Center for Drug Development Science in the School of Pharmacy, department of biopharmaceutical sciences, University of California - San Francisco.
"Since this is a human activity, there is reason to expect that perfect adherence to the protocol will not happen," Peck says. "What's surprising is that deviations from clinical trials [involving adherence] are presumably known and measured, but in fact they are not."
Adherence is the missing link in therapeutics, Urquhart says.
Medications are developed with high levels of quality control in specifying content and dosage form, but once they're given to patients, it's like sending it down a black hole, he says.
The key in both clinical work and research trials is to monitor patients' adherence and use these data effectively.
Traditional methods for measuring adherence include pill counting and having the patient keep a diary, Peck notes.
For example, investigators would have patients bring in their pill bottles, and then they'd count how many pills were left, compared with how many pills should have been left if the patient had been taking his or pills as prescribed.
The problem with this method is that it was too easy for patients to dump out their pills before coming to the clinic, Urquhart says.
"If you don't believe they'll do that then give them 50 percent more pills than they need and see how many still come back with an empty bottle," Urquhart says. "So there's a false sense of security in the minds of most investigators because they've been lulled into that by a method that has been thoroughly discredited by both electronic monitoring and chemical marker data."
Likewise, when investigators ask participants to keep a diary, there's a bias for reporting good adherence, Peck says.
"Patients have a desire to please their doctors, and clinical trial participants have a desire to please investigators, so there's always bias in their reporting how much drug they took," Peck says. "The deviations are huge."
In the past decade, researchers have relied more heavily on other methods of measuring adherence, including the electronic pill bottle device called medication-event monitoring systems (MEMS® cap) and testing participants' blood for drug levels.
The MEMS® cap, which Urquhart's former company designed, has been used in many HIV medication trials, and it's designed to record the times and dates of when patients open the pill bottle to take their medication. Studies have shown that the MEMS® cap's data corresponds with biological markers, indicating the participants have adhered to their medication regimen as well as predicted by MEMS® cap data.
MEMS cap may even be a better measurement of adherence than some types of directly-observed therapy, Peck suggests.
"During the Vietnam period, I was a physician and did some research," Peck says. "All the men who went to Vietnam had to take once-a-week chloroquine pills."
The drill sergeant would put the pill in their mouths and tell them to swallow them, Peck recalls.
When the soldiers left, they'd find two-thirds of the pills on the ground. The soldiers didn't want to prevent malaria because that diagnosis could get them back to the states, Peck says.
Once investigators establish that adherence is low, what are the implications, Peck says.
"The results in clinical trials are believed and used for decisions," Peck explains. "So to the extent that deviations from the nominal description or regimen are occurring, there's opportunity for a lot of misinformation out there."
This is where researchers can turn the lemons into lemonade: "If you could actually know what a patient takes in a clinical trial, then you have powerful new information that permits greater learning and knowledge retrievable from a dataset," Peck says. "This is true so long as you take the deviations into account."
The trouble is that investigators, physicians, and the FDA interpret clinical trial data without taking into account the participants' actual exposure patterns and drug-taking behavior, Peck notes.
Incomplete and inaccurate information obtained during a clinical trial can prove dangerous to patients taking the medication later, he says.
For example, in the 1980's, Peck saw a young woman patient who had potentially fatal cardiac arrhythmia. Her other doctors thought it was related to an antifungal medication she was taking, he recalls.
Peck was skeptical and didn't want to re-challenge her to the antifungal medication as the other doctors had advised. The case eventually led to an FDA investigation in which it was discovered that her allergy medication, a drug called Seldane, had caused the uncharacteristic arrhythmia, Peck says.
"They found that Seldane had low potential, but does slightly prolong the QT interval and can lead to fatal arrhythmia," Peck explains.
Investigators found that Seldane complications occurred when patients took both Seldane and either erythromycin or the anti-fungal agent ketoconazole. So the FDA issued warnings and label changes. Then in January 1997, the FDA announced that Seldane would be withdrawn from the market.
This case made the research community aware of QT prolongation and its connection to fatal arrhythmia, and led to the discovery of other drugs with the same problem, ultimately causing nine drugs to be withdrawn from the market, Peck says.
"Now, the FDA requires all drugs to undergo QT prolongation study in phase I with healthy volunteers," Peck says.
In these phase I studies, variations in participant adherence can impact the data about QT prolongation, he says.
"If you don't know whether patients are taking their medications or whether they're taking it at irregular times, it could lead to false positives or negatives," Peck says.
One theory for ignoring the patient adherence problem is that adherence is a real world problem, and so the trial data will more accurately anticipate how patients use the medication in the real world, Peck notes.
"But if the data are based on erratic compliance, how do you inform the patient who is compliant, and a small handful of patients actually follow instructions, that they're likely to have a greater safety problem," Peck says. "The safety information would be downward biased."
Also, drug efficacy could be downward biased and even lead to a drug being turned down when it would have been proven effective if it had been properly used, Peck says.
"So, if you have compliance data that are objective and reliable, and you use it, then you learn more," Peck says.
Investigators who use adherence data to assist with their studies could handle the information this way, Peck suggests:
On the negative side, some drug developers worry that if investigators start to explain the actual patterns of the medications people take, it may slow down the review process, Urquhart says.
"Against that is the risk of failing to confirm the effectiveness of the drug," he adds.
"My advice to people doing trials is to be much more careful about patient selection than they have been traditionally, selecting people who will comply satisfactorily," Urquhart says. "The best predictor of a patient's future adherence is his/her past adherence. The more variance you have in a clinical trial, the less statistical power you have, and the number one source of variance in drug response is variable adherence."