Rudimentary medicine: HIV testing kit at KIWAKKUKI clinic Photo:Chris Hildreth

Durack remembers when the first fellows arrived: "Two young doctors got off the plane completely dazed. We took them to the supermarket. They were stunned. We told them, 'Go get whatever you need.' They brought back a packet of coffee and a packet of rice." To date, fourteen Tanzanians have received training at Duke and more than 200 Duke students, residents, and fellows have worked in Tanzania.

It was an arrangement with an unusual emphasis on reciprocity, the service end of a "research-with-service" ethos central to Duke efforts and yet at odds with prevailing practice. "The model that we invoke is not like a lot of the research in international settings," Thielman, the Duke epidemiologist, told me as we walked through KCMC's wards. "Too often, U.S. investigators parachute in, grab the blood, if you will, and helicopter out. And they leave very little capacity in the developing country."

As Thielman spoke, the steady clamor of a construction crew echoed through the hospital halls. Outside, workers were putting the finishing touches on a much larger, Duke-funded HIV clinic, aimed at attracting entire families for treatment at the same time. "We want to have a presence here. Not just with machines and technology," he said, gesturing toward the clinic, "but with the training that will sustain this sort of work."

In the nearly two decades since the collaboration's inception in Tanzania, the division of infectious diseases at Duke has grown in size and stature: Its faculty has more than doubled; it's moved into the ranks of the top ten programs in the country; and its once-meager financial support has ballooned with major federal funding. Last August, in a move that signaled Duke's arrival at the forefront of the field, the National Institutes of Health awarded the medical school a $4-million grant for the study of AIDS co-infections in Tanzania, among the largest ever given for such work.

John Bartlett, professor of medicine and, until recently, director of Duke's infectious-diseases clinic, is principal investigator on the co-infections study in Moshi. He's been traveling to the region, alternating with Thielman every three months or so, for the past ten years. When I met him in his office at Duke last April, he was still jet-lagged from a recent trip, and I asked him why going there was so important--why the research can't be done here.

"It's important," he said, "because the geographic manifestations of disease vary greatly. Take, for example, HIV. A principal manifestation of HIV in a resource-limited environment is tuberculosis. Since you rarely see tuberculosis in the United States, you have to go there to study it." And so, he said, "We might test a specific drug in patients. Or we might follow patients over time to understand the clinical manifestations of their disease. But the research doesn't serve to help Tanzanians unless it's done in Tanzania."

Measuring up: Carl James '03, a Hart Fellow, checks the height of an orphan at Fukemi Primary School Photo:Chris Hildreth

That was the clinical reason. Then Bartlett gave what he calls "the big picture" reason: "Less than 1 percent of the world's population that needs treatment has access to medications; eight to nine thousand people die a day of AIDS; and we have in our hands the tools to prevent it. That," he said, "is not acceptable." Bartlett's voice grew even more impassioned as he described what I would encounter at KCMC: the 40 percent of patients on the wards who are HIV-positive; the mortuary at the back of the hospital where a brightly decorated truck is loaded, at dusk, with the day's corpses, and where family members dressed in white begin the funeral procession to the graveyard down the street.

It was scenes of suffering like these that moved Thielman and Bartlett in 2001 to expand the DUMC-KCMC collaboration. They set out to identify places in the region conducive to the sort of clinical trials that would achieve their two priorities: producing meaningful research and directly aiding the desperately ill.

They went first to the Kibongoto National Tuberculosis Hospital, where more than 75 percent of patients are infected with both tuberculosis and HIV. (Tuberculosis is estimated to be responsible for half of all AIDS-related deaths worldwide.) Relying in large part on John Crump, an assistant professor of medicine and Duke's only epidemiologist residing in Moshi, Thielman and Bartlett proceeded to set up a trial at Kibongoto evaluating a new anti-HIV drug regimen. The study, sponsored by GlaxoSmithKline, provided medications for more than seventy patients.

As the Kibongoto trial got under way, the Duke team was approached by a woman named Dafrosa Itemba, a Tanzanian who heads an organization called KIWAKKUKI (the Swahili acronym stands for "Women Fighting Vigorously Against AIDS in Kilimanjaro"), for help in developing an HIV testing facility for her clinic. Thielman and Bartlett obliged, soliciting funding for the construction of the laboratory space and the purchase of equipment, and months later the service was up and running. But they didn't stop there. KIWAKKUKI, they realized, could be a valuable partner. It was, after all, the only organization delivering care to the sick in a way that hospitals like KCMC and Kibongoto weren't. The all-volunteer staff was taking it to them in their homes--wherever that might be.

Stephen lived in Majengo district, a five-minute drive from the center of town, in a complex of dirt streets and banana trees and crude huts. Barefoot children in rags skittered about. A group of men crowded around a billiards table in the shade, drinking cups of the potent local brew. They looked up as the KIWAKKUKI van rumbled by. Inside, Helen Chu braced for bumps in the road.

A third-year Duke medical student, Chu had come to Moshi on a Fulbright scholarship the previous fall. For the past six months she had been conducting research under Thielman and Crump on two of KIWAKKUKI's core services--home-based care and voluntary counseling and testing--while helping to administer both.

Chu cautioned me not to present her as a medical expert, and by degree requirements, of course, she is not. But to the HIV-infected patients she treated in the slums or the rural villages outside the city, Chu was the closest to a doctor most had ever seen. That she wore a stethoscope and carried a penlight was itself a source of comfort, and comfort, she said, was the real point of the trip. Without antiretroviral drugs, care for the HIV-infected, she told me, "is largely a matter of easing the pain."

When we arrived, Stephen, who looked to be about thirty, was lying on a mat in the dark of his single-room shack. His sister stirred a pot of ugaali, a staple dish of ground corn, over a fire outside. In preparation for our visit, four plastic chairs had been set up in a semi-circle as though for a morning tea and chat. But instead of tea we'd brought antibiotics, and rather than chat, Chu mostly listened.

She sidled up to him, putting her head close to his chest and a hand on his back, and asked him to take deep breaths. "He's wheezing," Chu said. "That could be pneumonia." She turned back to Stephen. She asked him what was hurting and for how long, and he answered in whispers. Before we left, Chu gave him Septra for his diarrhea. "Take one a day, three days a week," she told him. Lightness Kaale, a nurse practitioner at KIWAKKUKI, translated. And then we prayed. "It's amazing how much simply praying with them helps," said Chu. "You leave, and they're smiling."

At KIWAKKUKI, Chu contributed what Thielman calls "groundbreaking research" to an issue at the heart of global health, no small feat for a med student conducting her first study in the field. The issue was testing, and as with any number of prevention or treatment efforts, talk of strategy had become mired in talk of "cost-efficiency" and "sustainability," all revolving around a single dollar: the price of an HIV test.

For three months, Chu studied the effect of the cost of an HIV test on client volume at KIWAKKUKI. First, she offered the test for a dollar, then briefly for free, and then for the dollar fee again. "What we found, at first, was that more clients presented when testing was offered for free than when offered for a dollar, indicating that cost may be a significant barrier," she recalled. "After we bumped it back up to a dollar, though, we had fewer clients presenting, but we sustained a much higher number than before the free period. So I think we showed--and I think policy changes should reflect this--that if you can't make testing free indefinitely, if you provide it for just a window of time, you ultimately get a significant increase in participation."