"I personally never want to be responsible for a medication not going to market because of one site. Whatever I can do to minimize that noise — picking the right patients, not contributing to placebo effect — it's a very structured, simple thing. And if you're consistent from patient to patient throughout, it makes it easy."
Dr. Lara Shirikjian didn't start in medicine — she started in a lab, pipetting urine into tubes as a research assistant fresh out of undergrad. She worked her way up at the same company through medical school, returned during residency to help on clinic days, and never really left. By the time she completed her psychiatry training, research was already in her bones. She became a principal investigator 11 years ago and has been with the same organization — now CenExel, the largest dedicated CNS research network in the country — since 2002.
That longevity is the point. She has watched schizophrenia research evolve from a six-person operation running four-page inclusion criteria to a network of hundreds of sites processing 14-page protocols, managing billing departments, and building out dedicated regulatory infrastructure. Through all of it, she has maintained a clear view of what actually moves the needle — and what gets in the way.
This conversation covers the placebo effect problem in psychiatry, why sightless trials failed the moment COVID made them necessary, what breaks down when sponsors let CROs own all communication, and why the disconnect between community psychiatry and clinical research in Los Angeles is more adversarial than most sponsors realize.
The Placebo Effect Problem
Psychiatric trials live and die by placebo response. It's the variable that has sunk more studies than any protocol flaw, and Dr. Shirikjian has spent her career treating it like a controllable operational risk — not an inevitability.
Her approach starts before the patient ever sits down. The physical environment is designed to strip out the cues that signal a therapeutic relationship. Receptionists don't ask about weekends. Staff don't offer warmth beyond what's required. Every interaction is transactional, consistent, and explained to the patient upfront so the absence of warmth isn't mistaken for indifference. "This isn't therapy," she tells patients. "This is clinical research and you're part of it." That clarity, she argues, is actually reassuring — especially for patients with schizophrenia, who respond well to concrete, black-and-white relationships.
The consistency extends from patient to patient and visit to visit. Every PI has something close to a script — not a formal document, but a fixed sequence of what gets said and in what order. If a patient notices and comments that she didn't ask about allergies, she can say: allergies are always the first thing she asks, within the first thirty seconds. The repeatability is the point. Noise in psychiatric endpoints often comes from variability in how patients are handled, not just from the drug itself.
Why Psychiatry Trials Still Need Humans
Before COVID, there was momentum behind the idea of sightless and decentralized trials in CNS. Dr. Shirikjian was skeptical then. COVID made the case for her.
Administering psychiatric rating scales by phone or FaceTime didn't work. The nuance that makes those instruments valid — the clinical judgment applied in the room, the ability to observe behavior and affect, the interviewer's capacity to probe an ambiguous answer — doesn't survive a video call. In settings with milder endpoints or objective lab draws, remote participation makes sense. In psychiatry, you need the humans present.
The same logic applies to data quality. She describes a scenario that has actually happened in the field: a blood pressure reading of 127 gets scanned into an EDC system and inverted to 172. The patient doesn't meet inclusion criteria. The error isn't caught until the study closes. No clinical judgment was in the loop to flag the anomaly. This isn't an argument against technology — it's an argument for keeping human QC between the paper and the database. As she puts it: "Every employee within a clinical trial site is valuable, from the research assistant to data entry to quality control. That's why we can't outsource it to robots."
What Sponsors Get Wrong
Dr. Shirikjian's criticism of the sponsor-CRO structure is specific. It's not that CROs are bad at their jobs. It's that inserting a layer between the site and the sponsor creates a filtering problem: CRO staff decide which concerns are worth escalating, and they sometimes get it wrong. A platform problem that makes a cumbersome study even harder to run might not seem like a priority issue from a CRO's vantage point. From the site coordinator's perspective, it's the thing that makes them route the next eligible patient into a different study instead.
She's had her direct cell number in the hands of sponsors and medical monitors throughout her career and treats that access as a competitive advantage. When a problem surfaces at the site level, she can resolve it in a phone call instead of waiting through a ticketing chain. "I would rather someone call me directly, real time, any time of the day with the problem. I don't think a chain of command is necessary."
The administrative load on PIs has grown in step with protocol complexity. Inclusion and exclusion criteria have grown from four pages to fourteen. Lab kit expiration emails arrive ninety at a time. IRB submissions that used to be handled in-house are now farmed to third parties with enough hands in the process to generate six-week delays on a Schedule I delivery. Her view: simpler protocols run faster, and sites that have built strong internal infrastructure to manage the overhead will consistently out-execute sites that are still trying to outsource it.
Why Schizophrenia Research Matters Most
Dr. Shirikjian has been asked over her career whether she finds the emerging areas — psychedelics, biotech, personalized medicine — more exciting than schizophrenia. Her answer has not changed in ten years.
The approved standard of care for schizophrenia is not good. The drugs that exist don't work for everyone, they carry a metabolic burden — weight gain, elevated glucose, cardiovascular risk — that erodes quality of life for patients who are already managing a difficult chronic illness, and there is no drug that is a reliable slam dunk. The muscarinic compounds entering trials now are producing results she's finding genuinely exciting. Patients who weren't responding to anything are improving. Family members are noticing. "We're starting to crack the surface," she says, "because the outcomes for that patient population are so bleak."
Running high-quality schizophrenia trials in Los Angeles has not been operationally difficult for her site. The patient population is large. Her team has spent two decades building the infrastructure for wraparound support — transportation, meals, non-study medication refills when a patient can't reach their primary provider — that keeps patients retained through long study timelines without crossing the IRB line into therapeutic services. The IRB constraint is real: no social work, no group therapy, nothing that could improve outcomes independent of the drug. "If you follow the protocol, there's no gray area. It's very black and white. It's easy." But within those bounds, the difference between a site that keeps patients and a site that loses them is almost entirely about relationship infrastructure.
The Community Provider Disconnect
The biggest structural barrier to schizophrenia research in Los Angeles isn't recruitment — it's the broken relationship between research sites and the community psychiatrists who care for most patients with serious mental illness.
Community providers don't refer to clinical trials. The historical reason is straightforward: they've seen patients decompensate during inpatient studies, get dropped by the research site when the trial ends, and return to the community psychiatrist destabilized. For some sites, that is exactly what happens. The research episode ends, the site's financial relationship with the patient ends, and the clinical responsibility is tossed back.
Dr. Shirikjian's site operates differently. If a patient decompensates during an inpatient study, her site keeps them until they're restabilized — back to their baseline — before discharge. It doesn't always cost more. A follow-up visit and a quick refill for a patient who ran out of blood pressure medication two weeks after the study closed is a minor operational commitment. But it changes what the community relationship can look like.
The problem is that many community psychiatrists don't know that, and some don't want to know. She has gotten earfuls on calls that took fifteen attempts to connect — told she's a "bottom feeder" in the pockets of pharma. The hostility is real and it has material consequences for research. Sites that require medical records from the LA County Department of Mental Health have faced a process so convoluted — patients physically appearing at records offices, paying photocopy fees themselves, waiting six to eight weeks for records that need to arrive inside a two-to-four-week screening window — that some research sites decline studies with those documentation requirements entirely. Kaiser patients can pull their records on a phone in two minutes. The disparity shapes which studies get run and which patient populations get excluded.
Her ask of sponsors is simple: build the community relationship into the protocol. Not as an afterthought, but as a funded, structured part of study design. The sites that can show community providers a track record of responsible patient stewardship will unlock referral networks that sites without that track record cannot access.
