A: I think one thing we learned with hindsight is that an NIMH submission or an NIMH grant is not a protocol to run a study by. You write up a grant and put in inclusion criteria/exclusion criteria, all the stuff you need to do a clinical trial, but you're writing primarily for the reviewers to know that you've taken care of all the issues that you need to take care of to run a clinical trial. It's not enough detail to actually run the trial.

When you do industry trials that are supported by pharmaceutical companies, you consult with them around the design but they come up with the actual protocol. There they know how to write proper protocols. Some exposure to industry trials has been helpful in term of they have that methodology down pretty well. So for people starting out, it would be helpful probably to have some exposure to industry trials to see protocols, start-up meeting, training of raters, that whole process, the quality assurance stuff they put in with outside monitors, which you don't do from a single-site study that you get funded through an outside source.

A: The biggest issue has to do with the data and your access to it. For a junior researcher, while being involved in an industry trial is probably good experience in terms of how to conduct trials, your ability to get any data out of it is non-existent. Even researchers who help the companies design the trial and act as consultants have a hard time getting data out of the trials. That's been a criticism of industry trials because negative ones don't get published and even if they were published, who would write up the data? The company, not the researcher, owns the data.

An alternative is to do an investigator-initiated trial where you ask a drug company to do a particular study in which you're interested. In that context, you then own the data, and it can be pilot data to start up a study. Pharmaceutical companies might agree to that kind of arrangement when you're dealing with an area they're not selling the drug or when they want some early data before they launch a larger study themselves.

A: If a junior investigator is interested in drug effectiveness trials, then those are often better funded with some collaboration between universities and industry. It's not a bad idea for someone to get some pilot data from a drug company or some other means. That's one way to do it. Some people want to design the full study first, and they often don't have the initial pilot data needed to get that started and to get funded.

A: In any kind of clinical trials research, recruitment is the biggest problem. It's the biggest headache I have on a day-to-day basis. When you resort to advertising, which we didn't do in the past but do now, the types of patients you recruit from advertising may be different than other patients. Designing the study and even getting funding has been easier than recruiting patients. We've actually been one of the better recruiting sites, and that's helped us in getting more studies funded.

There are a lot of factors in the difficulty to recruit patients. Sometimes, it's the disconnect between researchers and clinicians. The first study I ever did was in an inpatient unit where I was the director. I was responsible for the admissions, and I was actually doing a lot of the admissions myself. Just asking them to be part of a research study was not that difficult. I was going to see the patients anyway. As I get more distanced from the clinical service and don't see a lot of patients other than the ones in research, then I have to ask clinicians to refer patients to me.

Recruitment of underrepresented populations is also difficult. The patients we recruit from public health clinics are much more diverse than patients we see in our routine clinical trials. You just don't get that kind of diversity within samples where you're in a university and you either advertise or use your clinical services to recruit people into research. If you go out to the community where the people are being seen and recruit from that setting, it makes a big difference. Most of our samples are 80% or more Caucasian, and we need more data from diverse samples to answer some of our questions and to be able to generalize to the community at large.