Two years ago, on a gray January afternoon, I visited the Ridge Avenue homeless shelter in Philadelphia. I was looking for poor people who had been paid to test experimental drugs. The streets outside the shelter were lined with ruined buildings and razor wire, and a pit bull barked behind a chain-link fence. A young guy was slumped on the curb, glassy-eyed and shaky. My guide, a local mental health activist named Connie Schuster, asked the guy if he was okay, but he didn't answer. "My guess is heroin," she said.

We arrived at the shelter, where a security guard was patting down residents for weapons. It didn't take long for the shelter employees to confirm that some of the people living there were taking part in research studies. They said that the studies are advertised in local newspapers, and that recruiters visit the shelter. "They'll give you a sheet this big filled with pills," a resident in the shelter's day room told me the next day, holding up a large notebook. He had volunteered for two studies. He pointed out a stack of business cards on a desk next to us; they had been left by a local study recruiter. As we spoke, I noticed that an ad for a study of a new ADHD drug was running on a television across the room.

If you're looking for poor people who have been paid to test experimental drugs, Philadelphia is a good place to start. The city is home to five medical schools, and pharmaceutical and drug-testing companies line a corridor that stretches northeast into New Jersey. It also has one of the most visible homeless populations in the country. In Philly, homeless people seem to be everywhere: sleeping in Love Park, slumped on benches in Suburban Station, or gathered along the Benjamin Franklin Parkway, waiting for the free meals that a local church gives out on Saturdays.

On another occasion, I met former subjects at Chosen 300, a storefront church that serves meals to homeless people. The service had already started by the time I arrived, and raucous gospel music filled the bleak room. The congregation consisted of several dozen black men sitting on folding chairs. Many stared at the floor.

After the service I spoke to a thin young man in a dirty T-shirt who told me he had done an outpatient study for an anxiety drug. "Some kind of new benzo," he said, as he devoured a bowl of Cheerios. (Benzodiazepines like Valium, Xanax, and Ativan are often prescribed for anxiety.)

Outside, an older man named Steve told me he was trying to get into a depression study. "They ask you a lot of questions and see if you're approved for it," he said. "If you're approved for it they'll pick you up wherever you're at."

Later I walked round the corner to a shelter, where I talked to an elderly white man. "I'd say the majority of guys here take advantage of that," he told me, "because they get a lot of money and they're broke as hell."

Addiction treatment studies are one popular option. Last November, I visited the Sunday Breakfast Association shelter, where I met a man named George. He had a wispy goatee and a Letterman-like gap between his front teeth. George talked with such familiar, ironic congeniality that I was taken aback when he told me he had spent time in prison and once tried to commit suicide. "This city is fucking tough, and it is getting worse," he said. I mentioned a recruitment flyer I'd seen outside the shelter asking for subjects with "cocaine dependency." George nodded. He told me that a lot of people start taking drugs just so they can qualify for those studies. "You take that shit two days before to get it into your blood." He mentioned that he had recently screened for a trial at a research site running addiction studies. "There were people in the waiting room high as a kite," he said. "They were incoherent."

But the studies I heard about most often were for psychiatric drugs: antipsychotics, antidepressants, anxiety drugs, and stimulants. George used to take Risperdal, an antipsychotic. "That drug will turn you into a zombie," he said. He mimed falling over sideways in his chair. "I couldn't sit up without falling asleep." He gestured toward the other shelter residents: "Ninety-five percent of the population here has some kind of mental problem."

Most people think of pharmaceutical research as a highly technical activity that takes place in world-class medical centers. The reality is somewhat different. This is apparent in a grainy video that I watched a few years ago. It had apparently been recorded on a cell phone, and the camerawork started off wobbly. A tanned man wearing sunglasses and a necklace appeared and was introduced as Dr. Johnny Edrozo, a psychiatric researcher. His shirt was unbuttoned partway down his chest. "The latest stimulant coming out of the market is Vyvanse, which is a Dexedrine preparation," Edrozo told the interviewer, pausing occasionally to chew gum. For reasons that were not explained, the interview took place in a parked car.

This was my introduction to South Coast Clinical Trials, a chain of private research sites in Southern California that specializes in testing psychiatric drugs. Pharmaceutical companies now typically outsource clinical studies to contract research organizations like South Coast, which run trials faster and at lower cost than universities do. Their job is simply to follow the instructions of their sponsors. This formula is working: The contract research industry has grown steadily since the early 1990s and may now generate over $100 billion in annual income, according to the Tufts Center for the Study of Drug Development. At the top of the heap are corporations like Quintiles, which has 28,000 employees and operates in about 100 countries. At the other end are private physicians and small companies like South Coast, which are often based in strip malls or suburban office parks.

Dan Sfera, the owner of South Coast, has produced scores of web videos like this one, the ostensible purpose of which is to demystify drug research. (The unstated purpose, of course, is to generate business for their psychiatric research facilities.) I visited Sfera and his colleague Don Walters not long after watching the video, and they introduced me to a research subject named Steve, who vouched for the good intentions of South Coast clinic staffers. "I love this place," he said. "It's awesome. They don't treat you like you have a mental illness." A middle-aged man with a short, gray-flecked beard, Steve was starting an outpatient study of Depakote, a seizure drug that is sometimes prescribed for bipolar disorder. He had arrived at the clinic wearing red gym shorts and bedroom slippers. Over the summer, Steve told me, he'd been hospitalized for four weeks and had received eight rounds of electroconvulsive therapy. As he spoke, his hands trembled so violently that he spilled his coffee on the floor. He seemed preoccupied with his roommate, who he said hadn't showered for weeks. "The man's got toenails this long," he said, holding his fingers inches apart.

Steve told me he was staying at a room-and-board, an unlicensed facility where mentally ill people are given a room and meals. At Sfera's suggestion, I visited one that was home to some South Coast research subjects. It was located in South Central Los Angeles, a bleak neighborhood of chain-link fences and graffiti. The furniture in the house was worn, but a vase of flowers had been placed on the coffee table. Herbert Norman, the house manager, told me that he had 21 residents living there, and that many had been in clinical trials. In fact, Norman was enrolled in a South Coast trial himself. "My diagnosis is bipolar II," he said, surprising me a little. "Yeah, bipolar with a little bit of schizophrenia." A broken smoke alarm chirped in the background.

Soon, a very large man in a black Lakers T-shirt lumbered out of a bedroom, giving me a halfhearted fist bump before easing himself onto the couch. He was introduced as Harold. His diagnosis was paranoid schizophrenia, he told me, and he was enrolled in an outpatient study of Abilify, an antipsychotic drug. As we talked, an older black woman kept wandering in and out of the room, her lips smacking and her face twitching. I wanted to find out more about the study Harold was in and whether he understood the risks, but he spoke in a nearly inaudible mumble. "I'm always nervous about taking the pills," he said. "You kind of feel like a guinea pig all of a sudden." He said, though, that he had not suffered any side effects. When I asked Harold how much he was being paid, he hesitated and looked around the room, as if he did not want anyone else to hear. Then he asked for a piece of paper and wrote down the number 65.

To find people like Harold, some contract research organizations have employees visit room-and-boards and homeless shelters. In Philadelphia I met a man named Ed Burns, who explained to me how these recruiters work. Burns and his wife had been on the street for over two years when we spoke; he said they had trouble getting space in shelters, even though his wife is pregnant and Burns has bipolar disorder and depression. "I was on Depakote and I almost killed someone out of anger," he said. "It made me a wrecking machine." Burns was living in a shelter when he got a message saying that someone from the Veterans Affairs hospital was waiting outside for him. But when he went outside, he said, he was met by a representative of a research company known as CRI Worldwide.

"I was tired, I was hungry, and half an hour earlier the police had treated us like crap," Burns said. "And this woman is saying, 'Imagine, in 40 days you'll have $4,000!' The recruiter made testing drugs sound like a vacation in a five-star hotel, Burns said. "It's like a resort selling time shares. They talk about all the benefits first, and it sounds great, but then you start to ask: What do I have to do?"

Not long ago, such offers would have been considered unethical. Paying any volunteer was seen as problematic, even more so if the subjects were poor, uninsured, and compromised by illness. Payment, it was argued, might tempt vulnerable subjects to risk their health. As trials have moved into the private sector, this ethical calculus has changed. First came a hike in the sums that volunteers could be paid: Many clinical trial sites now offer over $6,000 for an inpatient drug study. Eligibility requirements have changed, too. For years, trial sites paid only healthy volunteers, mainly to test new drugs for safety. These days people with asthma, diabetes, kidney disease, liver disease, and other conditions can be paid take part in trials.

More startling is that recruiters are able to approach patients with serious mental illnesses, such as bipolar disorder and schizophrenia. "We are looking for individuals 18 and over who are diagnosed with schizophrenia or schizo affective disorder," read a recent Craigslist ad in St. Louis. "Earn up to $2,800.00." Around the same time, I saw an ad for a Los Angeles site that was offering three times as much for a related trial. South Coast tests experimental schizophrenia drugs, too, but only outpatient trials of drugs that have passed initial safety tests. Volunteers are typically paid $40 to $50 per visit. "The payments are low enough to not be coercive, but they're enough to supposedly compensate them for the time they've spent here, and give them an incentive to come back," Sfera said. Still, Walters, who has since left South Coast, added that money is what motivates most subjects. "I'd say at least 85 to 90 percent of clients, that's why they do studies."

The main ethical issues here, of course, are the competence and judgment of the prospective subjects. "When you say 'money,' everything else goes out the window," said Hanif Jackson, a former program supervisor at the Ridge Avenue shelter in Philadelphia, which recently closed down. I heard the same thing from Harvey Bass, a chaplain who has worked at the Sunday Breakfast Rescue Mission shelter for 15 years. He said drug study recruiters often park outside the shelter and approach residents on the sidewalk. Although Bass didn't think it was his place to warn residents away from the studies, it was clear that he was not exactly a fan. "These guys have no job, no home, and a habit," he said. "You have people at their lowest state, and they'll say yes to anything."

Sherman Howerton was sitting on the steps of the Station House shelter on North Broad Street when I met him in Philadelphia last year. A polite, well-spoken man in early middle age, he was wearing black pants, a white shirt with a large black stripe, and black-and-white tennis shoes. He looked like a referee. "I suffer from schizophrenia," he told me, explaining that he used to have memory lapses and hear voices. Howerton did his first antipsychotic study in 2010, and has taken part in three more since, including an inpatient study of Abilify. "The medicine helps a heck of a lot," he said. Yet he was carrying quite a lot of excess weight, and I noticed that his mouth twitched every few seconds.

Abilify was the best-selling drug in America in 2013, with sales of $6.5 billion. It is also the most visible representative of an extraordinarily profitable class of drugs. Antipsychotics have been around since the 1950s, but for the first 40 years of their existence they were reserved for patients with serious mental illnesses, such as schizophrenia. (Medical journals advertised them with slogans such as "Reduces need for shock therapy and lobotomy.") There was good reason for this caution: Antipsychotics can cause potentially dangerous or disabling side effects, such as muscle stiffness, tremors, extreme restlessness, and tics. The most notorious is tardive dyskinesia, a writhing, twitching motion of the mouth and tongue that can be permanent.

About 20 years ago, pharmaceutical companies began rolling out new and improved, or "atypical," antipsychotics—drugs like Risperdal, Zyprexa, Seroquel, and later Abilify. The atypicals were expensive, but the companies' marketing materials described them as safer and more pleasant to take. By the mid-2000s, physicians were prescribing the new antipsychotics for a dizzyingly broad range of conditions, including insomnia, depression, anxiety, bipolar disorder, agitation, autism, ADHD, and post-traumatic stress disorder. Prescriptions of antipsychotics for conditions other than serious mental illness doubled, and the drugs found frequent use in nursing homes, juvenile care facilities, and prisons.

The reputation of the atypicals was so stellar that many psychiatrists were shocked by the publication in 2005 of a large, federally funded trial known as the CATIE study. Researchers compared the best-selling atypicals to a 1960s-era antipsychotic and discovered that, despite the marketing claims, the newer drugs were no more effective than the old ones and still carried some risk of the same side effects. That same year, the FDA added a warning notice to antipsychotics labelling, after the drugs were linked to increased mortality in elderly patients with dementia. Soon afterward, British researchers who had tracked patients on atypicals for the course of a year showed that the patients had no better quality of life than those on older antipsychotics, despite the far greater cost of the atypicals.

Then there were the metabolic problems. Almost from the start, doctors reported seeing patients on antipsychotics experience sudden weight gains. Some developed diabetes. At the end of the 2000s, it was alleged that the manufacturers of atypicals had used misleading marketing strategies and downplayed the side effects. Bristol Myers Squibb settled federal charges over the illegal marketing of Abilify in 2007. Two years later, Eli Lilly pled guilty to criminal and civil charges that it had illegally marketed its blockbuster atypical, Zyprexa, and was forced to pay a $1.4 billion penalty. Soon afterward came large settlements or penalties against AstraZeneca (for Seroquel), Pfizer (Geodon), and most recently Johnson & Johnson, which paid $2.2 billion this past November for illegally marketing Risperdal.

Still, the drugs remain on the market, and drug companies are continually trying to develop new versions. In most fields of medicine, healthy volunteers are used for the initial tests of drugs, where the focus is on side effects. After all, it's easier to link side effects to a drug if the patient is healthy. It can be hard otherwise to know whether a symptom is the result of the illness itself, or even of another medication the subject is taking. There are also ethical reasons for using healthy volunteers. The initial safety studies, known as Phase I trials, are almost pure risk and no benefit. A sick patient who signed up for such a study might need to stop taking an approved drug—a drug that might well be helping—in order to test the toxicity of an experimental treatment.

Yet this is precisely what happens with antipsychotics, which are tested first on mentally ill patients who will often need to stop taking medication in order to join a trial. And these are trials that offer little chance of therapeutic benefit, because most drugs fail at some point during testing. One study looked at central nervous system drugs that emerged from a company's own labs and entered human tests; only 8 percent were eventually approved.

If you talk to the researchers who do the testing, they'll say that this makes sense because patients with schizophrenia tolerate antipsychotics better, and at much higher doses. "You see different side-effect profiles," said Charles Bailey, medical director at Accurate Clinical Trials, a trial site in Kissimmee, Florida. "The more ill they are, the more they suck up the drug." This view is widely shared, but reliable data on the issue is thin—mainly, small studies of only a few hours' duration, and limited to older, generic drugs. A more cynical view is that healthy volunteers will not take such risky, unpleasant drugs, even if they are paid a lot of money. Even a single small dose of an antipsychotic will cause some subjects to experience a sudden drop in blood pressure and faint. Many people say the drugs make them feel miserable. Bob Helms, the founder of the now-defunct magazine for clinical trial volunteers, Guinea Pig Zero, who was a healthy volunteer on over 80 trials, describes antipsychotic studies as "getting half-retarded in exchange for payment."

There are 10 different atypical antipsychotic drugs now available in the United States, each of which had to be tested in humans. Several are available in long-acting or injectable formulations, which require additional human testing. When the patent on a drug expires, rival companies often release a generic version, which requires yet more testing. Add to this all of the experimental antipsychotics that were tested on volunteers but never made it to market, and you begin to understand the economic forces behind what I saw in Philadelphia. Drug companies need mentally ill research subjects, and homeless shelters are full of them.

At a shelter called My Brother's House, I met a resident I will call Anthony, a friendly man in his late 40s with a cheerful smile and a machine-gun laugh. He told me he had taken part in more than 20 studies of antipsychotics. "They call me a professional," he said. He spoke so quickly that I could barely keep up. "I only do schizophrenic research studies, even though I'm schizophrenic and bipolar…I tried to do the one for severe patients but they wouldn't let me in. You have to hear voices every day of the week, and I only hear once or twice a month."

"They treat you well," Anthony said when I asked him how he felt about the studies. "You watch TV. They have DVD, CD, PlayStation, Xbox. They order out meals three, four times a week. Chinese food, cheese-steak hoagies, Buffalo wings, pizza. They gave me a birthday party a few years ago. I had to cut the cake. They sang me 'Happy Birthday,' and they were on-key."

Anthony appeared to suffer from many side effects that you might expect to see in someone who had taken so many antipsychotics: diabetes, significant weight gain, and limb stiffness, for which he takes a drug called Cogentin. He mostly didn't attribute these problems to the research studies, however, with the exception of his tendency to sometimes spit up. "But," he said, "that didn't happen until I did about 12 or 13 studies."

Anthony's feelings aren't unusual; other homeless people I met told me they appreciated the comforts of the inpatient research unit. Often the side effects are worth it, at least in the short term. But not always.

On a sunny morning a couple of years ago, I visited a research unit at Lourdes Medical Center of Burlington County in Willingboro, New Jersey. The unit was operated by CRI Worldwide, the same company that Burns told me he had spoken with. (The company is now known as CRI Lifetree.) Its focus was on inpatient Phase I trials, which often involve gradually increasing the dose of a drug until subjects begin to feel toxic side effects. Some Phase I studies also require painful or unpleasant invasive procedures. For these reasons, the payment to subjects in Phase I trials is usually much higher than it would be for an outpatient study.

My first thought about the CRI unit: Its appearance did not exactly suggest clinical excellence. Most of the furniture looked as if it had been rescued from a Salvation Army store. Homemade notices with titles such as "Smoke breaks" and "Money requests" hung on the walls. No studies seemed to be going on, but a few people were wandering around or watching television, presumably waiting to be screened or assessed.

My tour guide was Steven Glass, a psychiatric researcher in his early 60s who had been with CRI for 12 years. He wore a goatee and a baby blue blazer. Glass reluctantly agreed to show me around the locked unit after I called from the lobby, uninvited and unannounced, explaining that I wanted to see what a Phase I psychiatric unit looked like. I was pretty sure that he would not have let me in if I acknowledged my other reason for visiting, which was to have a firsthand look at the place where Walter Jorden died.

Jorden had been admitted to the CRI trial unit in April 2007. He was a 47-year-old veteran, the father of three children, and a widower whose wife had died six years earlier. His diagnosis was paranoid schizophrenia. Jorden heard voices that told him to commit suicide, and once tried to hang himself. He had been hospitalized for depression, substance abuse, and a potential heart problem. At the time of the trial he had a monthly income of $845 in disability benefits and was living in a run-down recovery house outside Philadelphia called New Desires.

CRI offered Jorden a paycheck to test an antipsychotic drug called ASP2314, an experimental compound then in development by Astellas Pharma. Exactly how much Jorden was being paid is unclear, but if the rates at CRI were comparable to those at other sites, it was probably somewhere around $2,000. According to documents from a malpractice lawsuit subsequently filed by Jorden's family, he was expected to spend about a month in the unit. During that time he would be given the drug and monitored for side effects as the dosage was gradually increased. In particular, the study called for frequent measurement of Jorden's heart rate, blood pressure, and heart rhythm, using an EKG machine. Jorden also had to wear a Holter monitor, which researchers used to continuously record his cardiac status.

There was a good reason for this attention to the heart. Antipsychotics can cause heart arrhythmias, including a very dangerous condition called torsade de pointes, which can lead to sudden death. In 1996, an experimental atypical antipsychotic called sertindole failed to make it to market in the United States after it was associated with 12 sudden, unexplained deaths in clinical trials. Mellaril, a once frequently prescribed antipsychotic, now has a black box warning for risk of arrhythmias and sudden death. For this reason, many Phase I trials of antipsychotics now include careful monitoring for potential heart problems.

On day eight of the study, shortly after 10 a.m., Jorden told an attendant he was having chest discomfort. According to the court documents, he was sweaty and short of breath. Doctors were called, including Glass, the researcher in charge of the study. They decided Jorden was having a panic attack and prescribed Ativan, an anxiety drug. An hour or so later Jorden left his room and went to the nurse's station, again in distress, where another CRI physician instructed him to breathe into a bag. At noon Jorden was given a second dose of Ativan; he immediately started to tremble and shake, and then fell backwards, unconscious. The staff tried to resuscitate him, but at 12:35 p.m. Jorden was pronounced dead. According to an autopsy performed later, he died of "myocardial infarction," or a heart attack.

It appears unlikely that Jorden's heart attack was caused by the experimental drug. But according to Jeffrey Fierstein, a cardiologist retained as an expert witness by Jorden's family, the physicians involved deviated from expected standards of care by not more seriously considering the possibility that Jorden was having a heart attack. In Fierstein's opinion, they not only ignored classic signs of a heart attack, but also neglected to perform an EKG and missed the opportunity to give Jorden the clot-busting drugs that might have saved his life. As Fierstein points out, "My understanding is that it [the emergency department] was around the corner; it was right there."

I followed up by phone and email after my visit, but none of the people involved were willing to talk. Representatives at PRA, the clinical research organization that acquired CRI Lifetree last November, did not respond to inquiries. Joseph Goldberg, the attorney who represented Steven Glass, would say only that all charges against his client were withdrawn after the court approved a settlement between the other parties in the case. I even tried calling the number of a CRI recruiter that I found on a business card I picked up at one of the shelters in Philadelphia. I got through, but the conversation was brief, and the recruiter didn't take my subsequent calls.

Many people assume that medical research studies are tightly regulated to ensure the safety of the subjects, but that's not the case. In the 1970s, after a series of notorious research abuses, legislators pushed for a central federal agency with the power to protect human research subjects. The medical research establishment fought this idea, however, and when the National Research Act was passed in 1974 a very different alternative followed: a patchwork system of small ethics committees known as Institutional Review Boards. The boards were originally located in hospitals and medical schools, but clinical research has since moved into the private sector. Many are now for-profit companies that review studies in exchange for a fee.

Still, most trials have to be studied by a review board, and that includes studies that use homeless people. The boards could take a stand against the practice, and perhaps some do. Finding out is next to impossible, because both the FDA and the for-profit boards regard many of the records associated with clinical trials as commercial secrets. Even the name of the board that reviewed a trial is confidential. This secrecy means that it is hard to determine whether reviewers even know where trial sites are recruiting their subjects. There's also a conflict of interest to consider: Perhaps review boards don't ask too many questions, because a board may start losing customers if it gets a reputation for being too strict.

Federal protection is little better. The FDA's oversight, for instance, can be porous: One report found that between 2000 and 2005 the agency had only 200 inspectors to police an estimated 350,000 testing sites. And while there are federal guidelines that cover clinical trials, it's unclear whether recruitment of mentally ill subjects from homeless shelters and recovery houses violates these rules. Those guidelines require selection of subjects to be "equitable," and special protection is required for subjects who are "economically disadvantaged" or "mentally disabled," since these people are "vulnerable to coercion or undue influence." But the guidelines make no mention of mental illness per se. Nor do they specify what constitutes being "economically disadvantaged." The FDA declined to discuss the issue with me, but said in a statement that neither federal law nor its own regulations prohibit people living in homeless shelters from taking part in clinical trials.

None of the bioethicists or review board managers I spoke to were willing to publicly defend paying mentally ill homeless subjects to take part in clinical trials, although most did not seem especially surprised to hear that the practice was occurring. But some prominent bioethicists do not see homelessness as a barrier to research. When The Wall Street Journal reported in 1996 that the pharmaceutical company Eli Lilly was recruiting homeless alcoholics for research studies, Lilly responded by hiring an expert bioethics panel led by Tom Beauchamp of Georgetown University. The panel argued that not only was testing drugs for safety on homeless people reasonable, provided proper procedures were followed, but also that "it would be unfair to exclude homeless persons categorically as a group." Beauchamp and another panelist, Robert Levine of Yale, went on to become paid ethics consultants for Lilly.

The guidelines governing the selection of research subjects also tend to have a deeper conceptual flaw. In the notorious research scandals of the 1960s and '70s, the common element was exploitation. With the Tuskegee syphilis study it was exploitation of poor black men in Alabama; with the Willowbrook hepatitis study it was exploitation of disabled, institutionalized children; with the Holmesburg Prison experiments it was exploitation of prisoners. In each case, researchers with power took advantage of vulnerable populations, getting them to "volunteer" for studies that most people would refuse.

Yet you will not see the word "exploitation" in the federal guidelines governing research. Nor will you see it in the Declaration of Helsinki, a foundational ethics document first signed in the Finnish capital in 1964, or most other codes of research ethics. What you will see instead are instructions to avoid "coercion" and "undue influence."

Concepts like "coercion" and "undue influence" are poorly suited for economic transactions, however. Offering desperate people money to take risks to their health may be wrong, but nobody is being coerced. No one is threatening to harm people if they refuse to become test subjects. One parallel would be sweatshop labor. The ethical problem is not that people are coerced into working in sweatshops—people are desperate to work there, under horrific conditions, for pennies. The ethical problem is whether it is acceptable to take advantage of their desperation.

I saw the same desperation on one visit to the Sunday Breakfast Association shelter in Philadelphia. I was looking for former research subjects in the day room. It was a large, open space, filled with people slumped over in chairs, eyes glazed, or sitting with their heads down on a table. Many of them seemed sleepy or drugged. There was no television in the room, no conversation, just silence. The stench of urine and sweat was overpowering.

As I moved from table to table, people started calling out to me. It wasn't that they wanted to talk to a reporter. They thought I was a recruiter for clinical trials. "I want to be in a study," a young woman in a hoodie yelled, bundled up in a corner. I tried to explain that I wasn't a recruiter, just somebody looking for information, but it made no difference. Word spread quickly across the room. People were still calling to me as I went out the door.

This story is the first in a two-part investigative special on problems in the clinical trials industry. The second, which asks why disgraced doctors are allowed to test drugs on human volunteers, is available here.

This first part was written by Carl Elliott, edited by Jim Giles, fact-checked by Kyla Jones, and copy-edited by Lawrence Levi. Photography by Jeffrey Stockbridge. Illustrations by Matt Rota. Aleks Krotoski narrated the audio version.

Audiobook | Kindle | iBook | Nook