One of the problems with determining how well a drug works in a study is figuring out whether the participants are actually taking it. Ideally, when scientists are comparing a treatment group to a control group in a trial, they want to be getting a good picture of the effects of the drug; however, their results only accurately show those effects if everyone is doing what they're told... and people don't. They forget the drug, they use it in the wrong way or at the wrong time, and those things make it very hard to figure out if it's the medication that doesn't work in a study or if it's the way that people are (not) using it.
In theory, there should be relatively straightforward ways to figure out if people are actually taking their pills - keeping track of the number they use, monitoring the number of times they open the pill bottle, having them write down each dose in their diary - but these ways don't always work. In one study where scientists counted the number of pills used and compared those results to blood tests of drug levels, they found that almost half of the people they thought had taken the drug correctly had gotten rid of the pills without swallowing them. Whether that's because the pills got lost or because the study participants wanted to please the researchers by making it look like they'd taken their medication doesn't really matter. What's important is that it showed that simple, straightforward measures of drug adherence don't necessarily give accurate results. They only tell you what people say they did - which may have little to do with their actual drug adherence.
That said, is the fact that people are bad about taking their drugs during a trial necessarily a bad thing? There is good a reason why most studies base their efficacy analyses on an intent to treat model - where a group is analyzed as though they have used the drug the way they were told, whether or not they actually did. That reason is that intent to treat models are thought to best reflect the way that drugs will actually work in the real world. If half the people in your study's treatment group forget to take their pills every day, or stop using the lubricant because it smells funny, it's likely that the same thing will be true when the drug is approved for more widespread use. To a certain extent, while it is useful for researchers to know if a drug isn't working in a trial because it's inactive or because people aren't taking it, for consumers and insurers the results are in many ways the same.
