In 1848, Rudolf Virchow, one of public health's heroes, contended that "medicine is a social science, and politics is nothing other than medicine writ large." It would please me greatly to think that Virchow's point has been taken. Although I'm a physician, these past two years have been an object lesson about the difficulties of scaling up and of moving from caring for individual patients to building health systems in settings of privation and disarray.

A few years ago, building health systems was precisely what I thought I knew most about. But the January 2010 earthquake that ended so many Haitian lives and destroyed so much of its infrastructure was a grim reminder that we still lack the ability to translate goodwill and resources into robust responses. Reflect, for a minute, on the limits and the potential of the activity that used to be called "charity" or "foreign aid" but that I prefer to call "accompaniment."

"Accompaniment" is an elastic term. It has a basic, everyday meaning. To accompany someone is to go somewhere with him or her, to break bread together, to be present on a journey with a beginning and an end. There's an element of mystery, of openness, of trust, in accompaniment. The companion, the accompagnateur, says: "I'll go with you and support you on your journey wherever it leads; I'll share your fate for a while. And by 'a while,' I don't mean a little while." Accompaniment is about sticking with a task until it's deemed completed, not by the accompagnateur but by the person being accompanied.

I teach at Harvard but volunteer with Partners In Health (PIH), an organization dedicated to providing health care for the poor that I helped found over 25 years ago. PIH has sought to make accompaniment the cornerstone of our efforts, from rural Haiti to neighborhoods in Boston and prisons in Siberia. In each of these settings, there are people who need long-term help: patients with chronic diseases and families facing persistent poverty. But health officials serving those areas routinely lack resources to provide it. In other words, even the accompagnateurs need accompaniment.

I first heard the term "accompaniment" somewhere between Haiti and Harvard. In 1983, the year following my graduation from college, I found myself in Cange, a squatter settlement on a dusty hilltop in central Haiti. I've told this story many times, to my own students and to anyone who would listen, but it is worth repeating today. The people in the settlement had been displaced by a development project, a hydroelectric dam that flooded a fertile valley in order to improve irrigation downstream and send electricity to far-off Port-au-Prince. These may have been worthy goals. But the people of Cange blamed their misfortune on the project; at the time of its completion, the structure was one of the largest buttress dams in the world, located in one of the poorest countries in the world. This experience gave me a new view of official development -- a dim view, since the people displaced by the dam received neither water nor electricity from it. The experience taught me a lot about what sociologists call the "unanticipated consequences of social action."

In 1984, as a first-year medical student, I returned to Haiti to finish building a clinic that local health workers and I had broken ground on in Cange. Once the clinic was up and running, community health workers from all over the region began referring patients with acute illnesses to it. Eventually, it became a small hospital and then a fully functional medical center. Still, patients with chronic diseases -- especially tuberculosis, which we treated often -- needed a lot more than a clinic or medical center could provide; they needed long-term social support. Most families in the area also lived in chronic poverty. Without food, clean water, and shelter, all the medical attention in the world could not help them.

In Cange, we made financial and nutritional support central to our tuberculosis program and trained community health workers to deliver long-term care. You may have missed our paper in Seminars in Respiratory Infections 20 years ago, but I believe it introduced the term "donkey-rental fee" to the health policy literature. We called such complex wraparound services "accompaniment." Community health workers were more than distributors of medicine and keepers of records; they were the patients' accompagnateurs. We found that when good clinical care -- the right diagnoses and treatment plan -- was packaged with robust accompaniment, tuberculosis cure rates rose from around 50 percent to nearly 100 percent.

When the village saw its first AIDS cases in the late 1980s, we began accompanying those living with HIV. We even took the strategy back to the United States, to Roxbury, Dorchester, Mattapan, and other poor neighborhoods in and around Boston. These patients lived in the shadow of Harvard's great teaching hospital yet suffered from an incurable (though treatable) disease and chronic poverty. They needed accompaniment in their homes and neighborhoods if they were to stay on schedule with doctors' appointments and medications; they also needed help finding childcare and paying rent. Our accompaniment programs addressed these difficulties and improved their clinical prognoses.

Of course, all of these programs were small and touched only a few patients at a time. We looked around after years of hard work and saw that we still lived in a world where millions of people died each year of treatable infectious diseases. Medicine and wraparound services could only do so much. What was ultimately needed were policy decisions to change the way chronic illnesses -- including AIDS and tuberculosis -- were diagnosed and treated worldwide. In the late 1990s, the economist Jeffrey Sachs invited me to speak to a class at the Kennedy School on this topic. I returned the favor by inviting him to visit the squatter settlement in Haiti.

At the time, our AIDS program in Cange was "controversial." AIDS medications were deemed too costly to waste on poor people. Newsweek highlighted the issue by printing a cover with the image of a young man dying from AIDS with the title "Too Poor to Treat." Despite this consensus of indifference, Sachs decided to come see Cange for himself. Now Cange looked different in 2000 than it had in 1983: Trees and green had replaced the dust, tin-roofed homes had replaced lean-tos, and the city boasted schools and a hospital. It was far from perfect but was immeasurably better than it was only two decades before.

Sachs wasn't feeling well during that visit, and I, responsible doctor that I am, suggested that a brisk walk to a village about two hours away would do him good. I almost killed Jeffrey Sachs. Five hours in the blazing sun did not, in fact, improve his health. Nonetheless, he did get to see community health workers -- accompagnateurs -- tending to AIDS patients. Thanks to the fruits of modern medicine, these patients were doing well.

But there was no funding with which to scale our Cange programs up. This problem was partially addressed by the Global Fund to Fight AIDS, Tuberculosis, and Malaria, an initiative Sachs helped launch in 2002, and the George W. Bush administration's Emergency Plan for AIDS Relief (PEPFAR), which was launched the following year. These efforts have since saved millions of lives. To be sure, the Global Fund and PEPFAR remain underfunded, limited in scope, and insufficiently coordinated with efforts to meet other priorities, such as maternal and child health. But they have prevented millions of deaths and set a precedent for life-long accompaniment of people living in poverty and chronic ill health.



There is a theological literature on accompaniment, and if you have the temerity to plumb it, you will be reminded of the term's Latin origins: ad + cum + panis, which is one way of saying "breaking bread together." The term is popular in liberation theology, which, like PIH, has its roots in Latin America. PIH is a secular organization, but that does not mean that those of us who make up the organization's network of 13,000 people in 12 countries don't draw on the inspiring work of theologians such as Gustavo Gutiérrez, a Peruvian priest who has written compellingly of "the preferential option for the poor." In fact, his principle guides our work: Although everyone deserves good medical care, those living in poverty receive the lion's share of our attention.

So what does this mean in terms of health and development policy? Let me draw on experiences during the aftermath of Haiti's 2010 earthquake. Many governments and nongovernmental organizations (NGOs) offered to supply humanitarian relief after the earthquake, and one of the jobs of the United Nations' Office of the Special Envoy, headed by former U.S. President Bill Clinton, was to track incoming aid. Here are some startling numbers: Of the $2.4 billion committed or disbursed, 34 percent went to civil and military organizations of donor states, 30 percent to UN agencies and international NGOs, 29 percent to other NGOs and private contractors, 6 percent to unspecified recipients, and 1 percent to the Haitian government.

These figures are for direct relief, not reconstruction, and it's hard to push resources through a government that lay in ruins (28 of 29 Haitian federal buildings were damaged or destroyed, and around 20 percent of federal employees were killed or injured in the quake). To be sure, Haiti needed the logistical and medical support that the U.S. government and military provided: U.S.N.S. Comfort, which steamed into Haitian waters only eight days after the quake, treated a thousand Haitian patients. But the international humanitarian effort could have done more to accompany the local authorities in charge of direct relief and reconstruction.

Foreign contractors and international NGOs need to find a better way of accompanying development partners. Sometimes this will include more direct budgetary support for struggling public health and education authorities, more support for local firms, and more local procurement. Of course, competition for resources -- jobs, lucrative contracts, and the like -- can lead to strife. But certain projects, including providing medical care for the poorest and education for the young, should be supported by a broad consensus.

We've tried to use this approach to lessen food insecurity. As I noted, it is difficult, sometimes impossible, to treat patients who don't get enough food to eat. For as many as half of school-aged children in Haiti, and almost all those we meet in clinics and hospitals, hunger is a constant companion. One remedy for acute malnutrition is known as ready-to-use therapeutic food, or RUTF for short. Colleagues from Médecins Sans Frontières showed in Niger that one such RUTF, a miraculous and tasty peanut paste, could save the lives of most children with moderate and acute malnutrition. For years, we've used a similar recipe in central Haiti to make what we call "Nourimanba" (manba is the Haitian word for peanut butter). Instead of buying this paste from a foreign company, we made our own using mostly locally procured ingredients -- an obvious choice in a predominantly agricultural country like Haiti. At first, we made it in the pharmacy warehouse in Cange, but we soon put together a tiny production facility. As in Niger, we found that the paste worked just fine in treating moderate and severe malnutrition, and the more reliable market for peanuts helped local farmers in turn.

We're now working on a larger facility with improved food-processing capacity. Although this effort has demanded skills beyond our team's medical training, we have found many partners, including specialists from a U.S. pharmaceutical company and Haitian agronomists. The program will not only treat malnourished children, it will, we hope, create many jobs. It ends up being accompaniment not only for people living in food insecurity but also for local farmers and all those seeking to improve food-processing capacity in rural Haiti. It would have been easier perhaps to buy RUTF from international providers -- in that case, just a matter of filling out a purchase order. And if the ingredients were difficult to obtain or prepare -- as is the case with many vaccines, for example -- ordering from abroad might have been the only option. But Nourimanba is a good example of a product that can and should be prepared locally.

Another example concerns Haiti's hospitals after the earthquake. The General Hospital has the largest patient load in Port-au-Prince but not enough resources and medical personnel. Its staff was small and underpaid even before the disaster. In the weeks and months after January 12, 2010, many international teams set up shop on the General Hospital campus, helping bolster its surgical and acute-care capacity. Coordination was a challenge: Different relief groups sometimes competed for space or access to hospital facilities. But no one could deny their role in mitigating a great deal of suffering. When they packed their bags, however, the hospital was still overrun with patients requiring medical attention; its staff was still underpaid. Since then, our teams have tried to bring together a number of international partners to travel the longer and harder road of reconstruction, in this case, helping rebuild the General Hospital.

For its part, the American Red Cross agreed to send $3.8 million in "performance-based" salary support for the hospital's beleaguered employees. But Haitian institutions often lack even the infrastructure to evaluate staff. Only an accompaniment approach could help develop such platforms and put them under the control of the intended beneficiaries. The Red Cross decided to work with the General Hospital to create these systems, and the effort is already starting to bear fruit: hospital staff are better paid, and the hospital's new accountability platforms are taking root. This kind of accompaniment is among the best ways to help address the structural deficits preventing Haiti from rebuilding better. We are hoping to launch similar projects with other key development organizations, including UNICEF, too.


Accompaniment is both an objective that is set at the beginning of a task and a mode of follow-through. My lesson, in a nutshell, is that the great failures of policy and governance usually result from failures of implementation, and accompaniment is good insurance against such failures. There are, of course, many bad policies; they've scarred the world in diverse ways and have hurt the vulnerable most of all. But most policies that universities cook up are not bad policies. Most policies that UN agencies develop are not bad policies. When NGOs take the trouble to make policies, which is not all that often, they are also not bad policies. The problem is delivery.

Again, let's consider Haiti. A Haitian graduate of Harvard, Jerry Tardieu, told me about a project there called the "Haiti Caucus," a challenge to come up with innovative policy projects that might help Haiti recover from the quake. A large number of Harvard students have already come up with an even larger number of great ideas. "They get it," he said, and I believe it.

Although great ideas abound -- at Harvard and in Haiti -- plans for implementation are scarce. When Corail-Cesselesse, a windswept plain north of Port-au-Prince, was identified as a possible post-earthquake resettlement location, scores of architects and urban planners began developing plans by the dozens. But months dragged on, and still no one had broken ground. In fact, no one had even bothered to check whether the proposed site was suitable for resettlement. The tendency to "minister from safe enclosures" (to use the words of theologian Roberto Goizueta), rather than from the place itself, led planners to overlook the minor detail that Corail sat smack in the middle of a floodplain. Anything built there would have sunk into the mud during the rainy season.

The road from policy development to implementation is usually long and rocky, one that must be trod with companions. When travelers have diverged as dramatically as Haiti and the United States, communication between them becomes difficult. About ten years ago, one of my colleagues was examining a patient, a five-year-old girl named Maveline, and found a tumor in her abdomen. It turned out to be cancer on her right kidney, curable if removed before it spread elsewhere. That procedure, a nephrectomy, could be performed right there in rural Haiti -- at one of the hospitals we built, an investment that had been regarded by some policy experts as foolish. The procedure went smoothly enough, but an X-ray suggested that the cancer had already spread to Maveline's lungs, and she needed chemotherapy and radiation, which could not be performed in Haiti. Since I was then living in a faculty apartment in a Harvard dorm but often absent, my wife and I thought that the child and her mother might stay there for a month or two while receiving treatment in Boston.

The two months or so turned into a year, and we got pretty close to Maveline and her family. When in Boston working at the Brigham and Women's Hospital, I would come home at about nine o'clock at night, tired but looking forward to seeing Maveline. Her mother, who had never before left central Haiti, would make me dinner, and we would all sit and watch cartoons or whatever Maveline wanted.

One night, well after nine, I was eating a home-cooked meal and sitting with Maveline and her mother, and there was a rap on the door. In came two students: one from Harvard and one from MIT. The Harvard student, Emilio, was from Miami, and although he had never once traveled to Haiti, he'd mastered Haitian Creole with a touching fluency that said a lot about him (he has since become a Jesuit priest and has dedicated himself to the poor in Haiti and Brazil). I'd never seen the MIT student before, but he was Haitian, and the conversation -- completely in Creole -- went something like this:
MIT student: "You're Dr. Paul Farmer?"
Me (suppressed thought): "No, I killed him and took his apartment."
Me (actual words): "Yes, nice to meet you."
MIT student: "You wrote The Uses of Haiti?"
Emilio, somewhat proudly: "Yes, and he's written other books as well."
Maveline's mother (hands on hips, mildly offended): "Dr. Paul, you never told me you knew how to read and write!"

I tell this story to amuse, sure, but also to illustrate how inequality works in the modern world and thus what an accompaniment approach needs to address. Maveline is still alive and well, but I'm not confident her mother knows how to read and write.

Haiti needs and deserves proper cancer and surgical care facilities and new and better health systems. But for this to happen, we will have to move from aid to accompaniment. Maveline needed an open-ended commitment; her care did not last for a couple of months but for years. An accompagnateur would not say, "Sorry, Maveline, we can help with your surgery but not your chemo." Seeing Maveline through to a cure involved a great many people. And her disease uncovered gaps in local infrastructure: If she'd been seen earlier and her tumor had been found before it spread to her lungs, much needless suffering could have been avoided. If her mother had been able to attend school, she would have been able to read and write, and chances are that Maveline would not have been born to poverty, and so on. It is not a bad thing to think systemically; if you want to solve problems, do not fear so-called mission creep. If the biggest failures in the policy world concern implementation or delivery, the second-biggest cause of failure is that programs are too often stovepiped, even though the problems they seek to address -- health and illiteracy, for example -- are tightly interwoven.


Failures of imagination are the costliest failures. Malcolm Gladwell quotes an engineer speaking of his former employer: "Xerox had been infested by a bunch of spreadsheet experts who thought you could decide every project based on metrics. Unfortunately, creativity wasn't on a metric." Neither are goodness, decency, social justice, and patient accompaniment. But that does not mean that these are unnecessary in public policy and in service of the common good. Just because we cannot yet measure the value of accompaniment does not mean it cannot serve as an important guiding idea.

Another way of putting this is: Beware the iron cage. About 25 years ago, when I was a graduate student at Harvard, I bought a copy of sociologist Max Weber's collected works. It hurt my back and brain to even look at this giant tome, but his topic -- how the "iron cage" of rationality comes to suppress innovation -- remains relevant to this day. It occurs through "routinization," a process by which rationalized bureaucracies gradually assume control over traditional forms of authority. This is often a good thing: Rationalized procedures can improve efficiency and equity. (Atul Gawande made this insight the core of his "checklist manifesto.") When the World Health Organization launched its directly-observed therapy protocol for tuberculosis, many countries, such as Peru, made great strides against the ancient scourge.

But exceptional events -- black swans, in popular parlance  -- expose the limits of this form of efficiency. When patients began falling ill with drug-resistant strains of tuberculosis, WHO guidelines suggested they be treated with the same first-line drugs as non-resistant patients. Yet treating patients with the very drugs to which their disease had developed resistance not only failed to help them; it enabled the worse strains to spread unchecked among patients' families and co-workers. This is the double-edged sword of routinization: Rationalized treatment protocols first helped health providers increase the effectiveness and reach of treatment but later prevented them from taking necessary steps to curb the spread of drug-resistant strains. Increases in bureaucratic efficiency can come at the price of decreased human flexibility. In other words, as institutions are rationalized, and as platforms of accountability are strengthened, the potential for accompaniment can be threatened, since it is open-ended, elastic, and nimble.

When the iron cage of rationality leads to a poverty of imagination, cynicism and disengagement follow. It is easy to be dismissive of accompaniment in a world in which arcane expertise is advanced as the answer to every challenge. But expertise alone will not solve the difficult problems ahead. This was the long, hard lesson of the earthquake: We all waited to be saved by expertise, but we never were. True accompaniment does not privilege technical expertise above solidarity, compassion, and a willingness to tackle what may seem insuperable challenges. It requires cooperation, openness, and humility; this concept may, I hope, infuse new vitality into development work.

You are reading a free article.

Subscribe to Foreign Affairs to get unlimited access.

  • Paywall-free reading of new articles and a century of archives
  • Unlock access to iOS/Android apps to save editions for offline reading
  • Six issues a year in print, online, and audio editions
Subscribe Now
  • PAUL FARMER is Kolokotrones University Professor at Harvard University, chairman of Harvard Medical School's Department of Global Health and Social Medicine, and a Founding Director of Partners In Health, an international charity that provides health care to and undertakes research and advocacy on behalf of the sick and poor. This article is adapted from his May 2011 commencement speech at Harvard University's John F. Kennedy School of Government.
  • More By Paul Farmer