Global Health Business: The Production and Performativity of Statistics in Sierra Leone and Germany

Abstract

The global push for health statistics and electronic digital health information systems is about more than tracking health incidence and prevalence. It is also experienced on the ground as means to develop and maintain particular norms of health business, knowledge, and decision- and profit-making that are not innocent. Statistics make possible audit and accountability logics that undergird the management of health at a distance and that are increasingly necessary to the business of health. Health statistics are inextricable from their social milieus, yet as business artifacts they operate as if they are freely formed, objectively originated, and accurate. This article explicates health statistics as cultural forms and shows how they have been produced and performed in two very different countries: Sierra Leone and Germany. In both familiar and surprising ways, this article shows how statistics and their pursuit organize and discipline human behavior, constitute subject positions, and reify existing relations of power.

Key words:

• data performativity
• global administration
• global bureaucracy
• global health
• statistics

Global administrative apparatus like health statistics “work to do what”¹ exactly? Tracking disease incidence and prevalence is certainly some of the work health statistics do. But what might we make of the other work of statistics, the behind-the-scenes decision-enabling administrative work, bringing order, commensurability, and the avowed efficiencies of neoliberalism to the messiness of human health? What of the new ways health statistics work, increasingly, to enable health endeavors to become business enterprises? In this article, I explore the roles health statistics play in the global economic transformations we have seen in the past decades as market fundamentalisms have restructured the provision of health services worldwide. The logistics of commodity capitalism require global administrative apparatus, among them raw quantitative data collection and the production of statistics, to operate both as the things and the structures of “best practice” health praxis (AbouZhar and Boerma 2005). Within medicine and public health, statistics increasingly accord the sense of business-like discipline and accountability in otherwise unruly health domains, even when this is not so. These transformations operate within and beyond geopolitical dominions, replicating, freeing up, constricting, and rerouting human behavior in ways familiar and surprising. The data in this article from Sierra Leone may be somewhat expected; the story from Germany is likely not.

Health statistics have proven invaluable in various health promotion campaigns; the epidemiology generated in Germany linking smoking to cancer is a poignant (albeit troubled) example (Proctor 2000). Worldwide, as a result of epidemiologically informed improvements in public health, statistics have generally operated as “maximally apolitical, authoritative, and safe” while playing a central role in shaping health priorities (Briggs and Mantini-Briggs 2003:256). That statistics are powerful is not new (Porter 1995; Hacking 1982). Statistics—from a German word, Statistik, a systematic accounting and enumeration for statecraft (Rose 1999:201)—have long been used by heads of state to govern. Using health statistics as a technology of global governance is a new aspect of an old practice.

We live in a politically and economically transformative time, and scientific objectivity changes and shifts with historical transformations (Schiebinger 1993, Daston and Galison 2007). The political and economic formations since the 1970s have been accompanied by new forms and intensifications of administrative oversight, audit, and accountability. A profusion of private sector development throughout the world has resulted from the rise of neoliberal ideology (Graeber 2010), the massive deregulation of financial institutions and instruments (Pfeiffer and Chapman 2010), and the increase of juridical enhancements and acceptance of corporations as legal persons (Scherer, Palazzo, and Baumann 2006). New technologies of governance have been required as divestitures from government provision and the shift to private provision has had a fractal effect on health in many places, weakening, to varying degrees, government claims as the master choreographer of health services for its own citizens. New forms of health provision have arisen, many nongovernmental. In any locale, there is now likely a mix of insider and outsider governors, public (government), private (for-profit organizations, corporations, and foundations), and quasi-governmental (“hosted” by government, but operating with private funds).

Newcomers to health provisioning like the Bill and Melinda Gates Foundation embrace and promote the change, holding that new configurations loosen the stranglehold governments had on public health and free up innovation. Others are adamant that privatization has led to “unruly” health care provisioning (Buse and Walt 1997), increased health disparities (Castro and Singer 2004), and class stratifications (Navarro 2007). Regardless, the fragmentation of health provision worldwide has necessitated global administrative apparatus aimed at organizing new kinds of donors and disparate public health agendas. There has been significant global growth in business management consultancy firms who manage health provisioning, like KPMG and McKinsey & Company. Statistics meet administrative needs in this emergent world (dis)order, operating as mediums of common categorization and exchange across diverse systems and modalities.

This shift to health business has necessitated new means to collect more numbers. (For the purposes of this analysis, I have comingled data, statistics, health information, and indicators. They are not synonymous, but the work they do as administrative apparatus is of the same kind.) An internal review at the World Health Organization found the organization recommended 3500 health indicators—numerical characterizations of health conditions—as necessary to lever financial and political support of its initiatives (Murray 2007:862). This profusion of numbers operate as administrative apparatus at work in a global bureaucratization of health service provision, with the aim of providing coherence, commensurability, and increasing efficiency in health praxis.

In this article, told in two parts, I problematize statistical production and performativity, terms explored next, to make points about governance, neoliberal privatization as an organizing modality, and the relations of power that typically go without saying in “global health.” Both sections feature the indicator Infant Mortality Rate (IMR; deaths per 1000 0–1 year), widely considered one of the most powerful gauges of a population's health. The Sierra Leonean section is about a health information system (HIS) initiative that includes IMR, and the German section centers on in-hospital production of IMR data. I show how processes of statistical production and utilization are not the same but are affected by similar modes of organizing thought. In both cases, IMR statistics are shown to be “false numbers” (Lampland 2010); they operate as administrative apparatus that shape health futures by reducing contextualizing “noise” and enabling business management rationalizations and decision-making.

I use cases from very different cultural contexts to make points not only about the business of global health but also to interrogate “that obscure object of global health” (Fassin 2012). Throughout the article I treat global health as a worldwide social field within which local health likelihoods are constituted through relations of power. Based on long-term engagement in both countries, I find the approbation to designate global health to low- and middle-income countries means we often miss how applications of emerging capitalist rationalities and business management techniques work contemporaneously. Being “global” about global health, by conducting same-themed research in rich and poor countries, provides insights that may otherwise remain obscured about the relations of power that shape human well-being. In this article, I show that pressures and prioritizations experienced locally are not simply random and that neoliberal techniques can catalyze new systems of advantage and intensify old systemic inequalities in both resource-replete and resource-limited countries.

ANTICIPATING BRIGHT GLOBAL HEALTH FUTURES

Numbers make it possible to evaluate the business investment risk of population health risks as if the risks are one and the same. We now see new kinds of linkages between statistical teleologies and health decision- and profit-making. “Investment,” “return,” “audit,” and “efficiency” (e.g., Boerma and Stansfeld 2007) are now commonplace descriptors of how and why money moves around the globe, makes good (on investment), and makes more (health). Actuarial science—the mathematics of risk—does double duty in health scenarios. Increasingly we witness actuarial sensibilities infusing health systems, merging “old” concerns about population health risk with “new” concerns about investment risk. Health probabilities—the chances that disease occurs—are merged with investment probability—the chances that profit will be made. Audit—the examining and verifying of accounts and programming—works alongside these sensibilities about risk, employing templates of neoliberal business management based on numbers. Tautological and increasingly mutually constitutive, in rich and poor countries alike, health risk and investment risk use some of the same numbers to measure and evaluate before (baseline) and after (outcomes) to determine if particular health interventions, ventures, and speculations were “worth it.”

There is a temporality to how health statistics work. They are used to look back in time, often to track the national status of disease or indicator prevalence in relation to international agendas, and then to look forward in time, to project future programmatic direction, and prophesize success and salvation. Ian Harper illustrated this phenomenon in descriptions of the disease he “performed” (2005:128) in Nepal, in which local data were sent away, then rendered and returned as global dictates. Naming a “new political object within an emerging global order” (137), Harper coined the term “statistico-tuberculosis.” This describes his clinic's tuberculosis statistics after they were mixed with regional figures in Katmandu, then sent off to Geneva to produce global figures based on “sputum conversion rates,” and then returned to Nepal to set standards for anticipated programmatic success. Tracking the number of tubercular bacilli in phlegm smears on a glass slide, he shows, reduced the eligibility of a large cohort of seriously ill people in Nepal to only those whose sputum records provided “real” evidence of TB. Statistical rendering may have “unintentionally, yet paradoxically, den[ied] significant numbers of people who have TB access to treatment” (144). International health statistics regularly operate as if they are freely formed, objectively originated, and accurate, yet Harper shows that an administrative abstraction like “statistico-tuberculosis” had profound real-time in-place effects on the degree of human suffering unchecked by treatment.

Part 1: On Health Data in Postwar Sierra Leone

At a maternal health clinic in Freetown, nurse-midwives giving a tour stopped in front of a small closet, unlocking and opening its door wide to show its contents. Inside, thousands of paper forms with health data about women and their babies were stacked high and deep, filling shelves top to bottom. Showing the data documents seemed important, on par with showing the antibiotics and other medicines kept locked in a metal cabinet. Two days later at a hospital 250 miles away, I was shown, again unbidden, health data forms within minutes of arrival. These forms, similar to those in the Freetown clinic, are typical for maternity hospitals throughout the world: they collect names, dates, weights, and other quantifiable measures, risk indicators, and complications. I had worked with similar forms at the same rural hospital in Sierra Leone two decades earlier and in hospitals in Germany and the United States where, despite differences in relative wealth, maternal and infant health hospital forms include mostly the same categories. What was different during this visit to Sierra Leone, however, was the presence data had acquired in 20 years. At the Freetown clinic, sitting at a small school desk too small for her lanky frame, one nurse-midwife entered data from that morning's birth, intently and painstakingly, for most of her shift. When the other nurse-midwives brought out oversized bound patient record ledgers, placing them on a desk before us and pointing out labor and delivery data patterns from the prior year, something new was being communicated. The free-flowing personalized narratives about this or that woman and her infant were gone. In their place were narratives of monthly and yearly numerical amalgamations of births and a few deaths. More than mere paperwork, health data and statistics had become a type of capital in stratagems of health praxis—privileged signs and symbols of modern health knowledge production and surveillance. They were something else as well. Those ledgers contained formal “proof” that women and infants in the clinic were receiving maternal care that met international standards, that donor money was being responsibly spent, and that what few pharmaceuticals the clinic had were being responsibly dispensed. Informally, those ledgers served as hedges against any future accusations of corruption and mismanagement.

In Sierra Leone, the postindependence government-to-government international health assistance of the 1960s has metamorphosed to a fragmented, often wheel-reinventing, mostly NGO provision of global health services. The nature of the bilateral funding has changed as well; once there were, for example, US Agency for International Development field officers directly managing development projects in the field. Now project work is mostly outsourced to an unknown number of NGOs: estimates for Sierra Leone are around 3000, about a quarter health focused.

In 2008, during a visit to an NGO compound on Wilkinson Street in Freetown, I spied a faded dog-eared memo pinned on a wall. It announced a health information systems (HIS) initiative, funded by the World Health Organization (WHO), the World Bank, and more than 15 other organizations, promising “a computerized system that systematically captures, collates and analyses data from peripheral health units in the country” for greater efficiencies in (1) health service delivery, (2) evidence-based decision-making, (3) resource utilization, (4) drug management, and (5) overall transparency. An editorial in The Lancet had announced commencement of the same initiative in 2005, concluding, “Without [global health statistics], we will continue to stumble around in the dark, making bad decisions on the basis of bad information” (The Lancet 2005:1983). The memo on the wall described a “revitalization.” This was a peculiar characterization, inferring a reanimation of an existing system. Although paper records had long been kept and stored at hospitals in Sierra Leone, no health information system, in the contemporary sense, had previously existed.

Over the past 30 years in Sierra Leone some census, vital statistics, and household surveys have been conducted, mostly ad hoc. Some twenty-odd years earlier, at a hospital in the Eastern Province run by Irish Catholic nuns, I had collected maternal and infant data, serving as Mende women's first point of contact when they came to the prenatal and infant care clinic. I had done the same at the rural clinics scattered within a 20-mile radius of the hospital, in buildings generally not much more than a tin-roofed shelter with a table. For 18 months in the mid-1980s, I filled out hundreds of 4 × 6 file cards, in English, with a woman's name, her birth date, information about parity and living children, along with her abdominal circumference measurement in inches on that clinic day. I weighed both sleeping and screaming babies in hanging scales, their watchful mothers nearby, and duly created another store of 4 × 6 card infant health data records. A nurse-midwife and I brought those cards with us, trekked them out again in our motorcycle packs, and deposited them in a hospital office closet. As far as I can ascertain, nothing was ever done with the data collected.

From the outset of the WHO initiative, local data collection resources—handwritten clinic intake and hospital patient records—were deemed “not good enough” for the purposes of building a health information system in Sierra Leone. The WHO initiative use “before” images of paper-laden shelves and closets similar to those I had helped fill as disparaging points of reference. These troves of paper data—a valuable local resource—are represented in some consulting documents as a significant factor in “the mess” (e.g., Kossi et al. 2009b:13).

Farcasting

Freetown's hotels are second homes to international cadres of NGO and international development workers escaping the oppressive midday heat and humidity. Such venues have long hosted ex-patriots. Health workers, education consultants, diamond smugglers, priests, nuns (during the war, gun-runners) mingle, unlikely comrades exchanging friendly tips on transportation, lodging, and food. Hotel lobbies and restaurants are filled with outsiders, hunched over computers; wires carrying AC current and Internet server lifeblood connect humans to electrical outlets. In the health and education sectors, the number of outsiders doing the work previously conducted by the Sierra Leonean government workers has increased significantly over 20 years.

The magnitude and intensity of health sector work in Sierra Leone has also changed. “Global health” is global business now; low estimates of global net worth start at about US $35 billion (Cohen 2006), although many people I have asked think $80–100 billion is more likely. Business needs numbers. Foundations, bilateral organizations, and NGOs compete against one another to “do the most good,” “most efficiently,” “in the shortest amount of time.” Organizations that can afford it, like the Gates Foundation, employ Masters of Business Administration graduates who use and spread a gospel of business model cost–benefit analyses to plan, monitor, and evaluate health interventions on strict investment-and-return timelines. Numbers are the bottom-line standard bearer for measuring both return on financial investments and improvements in health outcomes.

Business metrics have fostered the trading and sometimes the secreting of a new commodity: health statistics and information. Contact information for health NGOs in Sierra Leone (2006 compilation) once appeared for free on the Ministry of Health and Sanitation website. Now it is for sale on www.docstoc.com for US $19.95/month membership. When a public health professional whom I will call Sarah, from a large UK foundation, introduced herself to me in a Freetown hotel lobby after she heard me speaking Krio to a waitress, she offered me health statistics for trade for information. She had “good numbers,” she said, “recent.” Upon my return to Canada, she sent me the report she had drafted in the hotel lobby. Statistics had enabled her to write an authoritative report and make future recommendations without leaving the hotel or meeting a human being who met her population health criteria.

Statistics enable a multitude of possible processes. Among the most common is the write-up of proposals at a distance without going the distance. In this case, statistics had enabled this well-intentioned health professional to conduct global health business behind the gated and guarded walls of air-conditioned hotel compounds in a poor country. This is, after all, what statistics do by design. Statistics’ predictive value to public health professionals is that they can do the work they are supposed to do at a distance. Designed to forecast and farcast (see Murray 2007:862), statistics project in time and place.

Distance is a key analytic for thinking through fuller social meanings of statistics. “At a distance,” Latourian in origin, is a phrase that shows up in several of the health systems initiative program documents and publications. Scientific instruments are designed to work at a distance. Knowledge collected by the powerful about the less powerful using scientific instruments—collected in cycles of going away and coming back—creates centers of accumulated knowledge that create asymmetries of knowledge and thus power, e.g., Geneva, Atlanta, North American, and European universities (Latour 1987). As scientific instrument and product, statistics’ appeal is the way they enable the sense that the distant is familiar (when it is not), that the distant is manageable (when it is chaotic), and that they provide a means to gauge from afar whether further engagement would be profitable. The distance statistics overcomes advantageously aligns some members of the global community with extant power structures while disadvantaging others. Knowing people is not necessary.

Data Performativity

The opening gambit of Lampland and Star's Standards and Their Stories (2009:3) is about an American in Holland trying to make an appointment. She cannot make the appointment because she does not have a phone number and the computer software needs one. The American gives a number—1-2-3-4-5-6-7—and schedules her appointment. Lampland and Star explore how ordinary people deal with standards as well as to make a point about the importance of studying the unconscious use of standardizations and numbers. The give-one-get-one strategy is an appropriate analog to the multiple forms of data performativity manifest in Sierra Leone.

Performativity as a characterization of collective behavior in this sense moves from the theoretical space occupied by discursive and linguistic constructs of performativity (Butler 1990) to include performativity as it has been theorized in economics (Callon 2007; Çalışkan and Callon 2009). Certain economic models work to make themselves true when people repeatedly use particular scripts that add weight and robustness to the models’ use. The business of global health in Sierra Leone is fortified when people enact statistical health scripts through acts and processes of data collection, analysis, storage, and disbursement. Whether statistics are accurate or enough to improve health is less important than whether statistics are performed and work to enable economic systems. In this way, hollowness in the numbers—the numbers are inaccurate or are too disparate from people's bodies, human complexities, or communities to be meaningful representations—may not matter as much to the participants as the fact that statistics enable other things of value, like gainful employment and profit-making.

Data performativity is enacted in different ways. Sarah in the hotel lobby performed data in a way that is typical of many of the global public health professionals I meet in Sierra Leone: Her data do her work for her, anchoring and legitimizing her report. Economies of contemporary global health make this performative type of global health work the norm: statistics made Sarah's assignment accomplishable. Sarah was in a rush, on a short-term contract for a week. Oversight of her work is done at the UK office, not in-country, which means she is likely writing for outsiders who know little of life or health in Sierra Leone. She did not speak local languages, and while she appeared to know almost everything study-able about Sierra Leone, she had not met a single pregnant Sierra Leonean woman, community health worker, or prenatal health provider. She spent four days in the hotel lobby ferreting maternal health information from other public health workers like her, and then she went home to the United Kingdom to write her report.

A second set of performative data practices can be found among an elite and educated class of Sierra Leoneans. Public health professionals new to Sierra Leone usually require a Sierra Leonean contact, someone who knows at the very least where to buy a cell phone and exchange money. With the significant influx of outside money coming into the country after the war (over half of the federal budget comes from outside sources), there is a new cohort of Sierra Leoneans who piece together a livelihood moving from project-to-project as behind-the-scenes international development project administrators. They are often found in the same hotel lobbies where I met Sarah. This counterpart is usually different from university-based Sierra Leonean PhDs, who are in very short supply. Most are well-educated middle-aged men although there are some women

I met one of these men, Mr. K, when he drove a group of us around in a spacious air-conditioned SUV. During a quick stopover, we crossed from his home's foyer into a large living room where a cricket game between England and Bangladesh was satellite-streamed to a huge flat-screened monitor mounted on the wall, the color commentary blaring crystal clear through impossibly small high-end speakers. His two teenaged children, just home from their expensive English-language private high school, were snacking in the well-appointed kitchen. This cohort lives well, and they view “finding statistics”—for health, education, and democracy projects—as part of their job. I was told the data “do not always translate” to verifiable on-the-ground population health circumstances, but au foh du (“oh, well”). There are Sierra Leoneans who were incensed by the hypocrisies of aid industries that put so much value in statistics at the expense of meaningful and long-term involvement, but others view industry realities with ironic dispassion. In postconflict Sierra Leone, adhering to aid industry data requirements in exchange for a regular wage, however temporary, is among the better employment options. Performing data is just part of the job.

Sierra Leoneans savvy to the data production requirements in international development and humanitarian projects are not the only ones performing data. In August 2008, the World Health Organization (WHO) and its subsidiary Health Metrics Network (HMN) announced: “Sierra Leone's Health Information System Revitalized … The district health information system includes a computerized system that systematically captures, collates and analyses data from peripheral health units” (WHO/HMN 2008c). In October 2008 at an HMN board meeting in Switzerland, Sierra Leone was characterized as having a “National Health Information System (NHIS) [which] is a computerized system, including an integrated data warehouse for the management and presentation of health data” (WHO/HMN 2008b). The WHO followed with a Progress Report November 2008, commending HMN's “successful efforts to revitalize Sierra Leone's health information system” (WHO 2008), as if it was a done deal.

One grandiose claim led to another. In March 2009, HMN posted a press release that “Sierra Leone's district health information system goes nationwide [and] extended throughout the nation of five million people” (HMN 2009). By May 2009 a Global Health Council conference panel presentation characterized Sierra Leone's health information system revitalization as one of three “success stories” that “harmon[ized] various components into a functional HIS [health information system]” and “integrated all existing data elements and forms into one large system” (WHO/HMN 2009).

Behind the self-congratulatory reports and press releases put out by the triumvirate of WHO-HMN-The Sierra Leonean Ministry of Health and Sanitation, a different narrative emerges. “Going digital” means using computers and telecommunication technologies, and possessing, at the very least, network connectivity. In postwar Sierra Leone, most medical facilities still use paper data collection systems and few clinics and hospitals had computers. “[O]nly a few [health] officers actually had any experience with computers at all”; “[M]ost of the districts are not connected to the Internet”; and there is “no network connectivity between most of the districts and the national server” (Kossi et al. 2009a). “The national power supply system is down” and human capacity in IT skills is a “key issue that has to be tackled” (Tohouri, Asangani, and Braa 2008:4).” “[P]oor power supply, lack of Internet, poor roads to remote areas” and “a hazardous climate for technical equipment” are also challenges (Tohouri et al. 2010:6). Sierra Leone has no national power grid and most of the country has no power supply except for private and intermittent diesel-powered generators and a few solar panels (regularly targets for theft). Taking the health information system digital in Sierra Leone is complicated by more than power source and theft issues. I met a recruiter for a Ghanaian IT company who wanted to employ Sierra Leonean IT experts. Fifty people applied and he estimated that a full third could not turn the computer on during the interview. Most of the others did not know how to use basic software. He explained it this way: “The ‘normal’ cohort of youth who played around with computers” hardly exists in Sierra Leone. There is a hardware supply issue, but there is also the issue that “when everyone else in West Africa was going digital, there was a war on.”

Impetus for the health information system initiative in Sierra Leone dates to 2003 when the Bill and Melinda Gates Foundation announced a US $250 million program to fund 14 global health challenges. Challenge 13 is to “measure disease and health status accurately and economically in developing countries” (Walgate 2003). With little fanfare, infrastructure to support Challenge 13 was put in place in Geneva in the form of HMN, and in Seattle at the Institute for Health Metrics and Evaluation at the University of Washington. HMN is a Gates-by-proxy organization, directed by a former Gates Foundation associate director, and received US $50 million from Gates as start-up money. HMN's initial ambitions include “that by 2011, the HMN Framework will be the universally accepted standard for guiding the collection, reporting and use of health information by all developing countries and global agencies” (WHO/HMN 2008a).

HMN's motto—“Better Information. Better Decisions. Better Health.”—has taken on the characteristic of a performative script, one repeated again and again. The motto has been taken up by the many partners involved in the Sierra Leonean initiative, including the Sierra Leonean Ministry of Health and Sanitation (WHO/MOHS 2009). Scholarship on the impact of health information systems on national health outcomes fails to support the script, however (Littlejohns, Wyatt, and Garrican 2003; Tomasi, Faccini, and Maia 2004; Wears and Berg 2005; Kimaro and Nhampossa 2005; Jha et al. 2009). Separate stand-alone health systems within nation-states have resulted from decades of vertical programming. They are typically independent of national health information systems rather than integrated into them (WHO Maximizing Group 2009). “Better Information. Better Decisions. Better Health” masks jagged, fragmented and segregating effects of health initiatives (see Whiteford and Manderson 2000), and in resource-poor settings, health information data collection and data entry consumes disproportionately high degrees of scarce human and technological resources (Al-Samarrai 2010; Nichter 2008:170).

If a health information system in Sierra Leone is mostly myth, and if health statistics there are mostly hollow representations of human health, then what are they good for? In Part 1, I have shown that health statistics are good for writing reports and proposals from a distance. Even as statistics sometimes undermine community health efforts and initiatives, they enable governments, NGOs, and foundations to engage in the performance of planning and evaluation at a distance, and to forecast, farcast, audit, and account for health monies and programming. For outsiders, statistics are especially “good to think with,” and this has implications for the power dynamics of decision-making. For business-minded users, this other work that statistics does is worthwhile for gauging where to find financial return and justify investments. As we turn to Part 2, the German case, we find some similarities. There too we find a recent change in the kinds of numbers used for health decision-making. In the hospitals where I conducted research the shift has been from health indicators to profitability indexes. Neoliberal schematics of “what counts” has shifted evaluations of outcomes increasingly away from the health of humans and more toward the profitability of health care for shareholders. Numbers as global technologies of neoliberal governance prove impressively fungible.

Part 2: On the Vagaries of Health Statistics in Germany

The field of statistics has significant historical links to Germany and German thinkers (Porter 1995:54). Before beginning my fieldwork, I was surprised when a director of the International Statistics Program at the US National Center for Health Statistics in Hyattsville, Maryland, cautioned me in an e-mail that “Germany is one of those places where national statistics (at least in the area of health) are hard to come by.” Indeed, when I began my research in a western German hospital, simple public health data inquiry and retrieval was surprisingly confounding.

Not so in former East Germany, I was told emphatically by Dr. V, a statistician who had worked before reunification as the head of the statistics bureau in East German Berlin. Unapologetically nostalgic for the statistical production processes of the former East, Dr. V told me that West Germany had been self-conscious about health data collection, and for good reason. With more than a little schadenfreude, she said that as recently as the mid-1990s, the international scientific community had taken issue with the data samples from Holocaust victims still in use in West German teaching universities (Jesani 1994). Health statistics collection and analysis in West Germany had been burdened, she said, by fascist Nazi legacies; East Germany had not. Official East German discourses successfully promoted a communist state vision that “wiped the slate clean” of Nazi nationalism. When I conducted research at an eastern German hospital, I came upon the collections of which Dr. V was so proud. Neatly bound books of East German health statistics going back to the 1950s—archived, easily accessed, and comprehensibly formatted—were housed in the university library.

After reunification, Dr. V had been relocated from her beloved Berlin to a small western German city in what she considered Germany's hinterland to work in a statistics bureau that did not meet former East German standards. Her professional diminutions were, in her opinion, not the only regressions that had occurred since German reunification. Health statistics collected in reunified Germany were dependent on physician self-reporting by state, and she estimated underreporting of about 50 percent. After reunification, all the former East German states had obstetrical health data, but a decade after reunification, only Bavaria and Hessen among the former West German states were publishing state compilations, which troubled her work in the federal bureau. Her professional life in East Germany, she intimated, had been one of overall collective order and discipline, and she was now responsible for the “disorder” of an entire federal system. Over the ten years of my research in Germany, 1998–2008, obstetrical data have been collected and compiled on both state and federal levels. That change highlights an important point about health statistics: They possess histories and variances, in Germany and elsewhere. That day in 1999 Dr. V brought our interview to an abrupt end when, as if she could no longer tolerate talking about the current sorry state of affairs, she stood up, strode toward the door, and said rather pointedly that statistical rigor comes and goes everywhere, “even Germany.”

Numbers Made and Not Made

Three ultrasound department nurses in the German hospital where I conducted research had been talking about Rita all morning, in the way that people do when they cannot quite believe a person would do something and they keep repeating the same narrative. Rita was pregnant and her fetus had been diagnosed with Trisomy 18-Edwards Syndrome, a condition that usually affects all bodily systems, from neurological to osteological. The vast majority of fetuses with Trisomy 18 die before birth or soon after, but there are children with Trisomy 18 who live into late adolescence (Petek et al. 2003). Rita, a 33-year-old nurse, had prompted a fury of speculation as well as judgment when she and her husband, a doctor, decided months earlier to carry her fetus to term and to deliver with no intervention. Choosing not to abort after receiving test results like hers was not considered normal. The nurses were not alone in thinking this way. Among the obstetricians and nurses with whom I worked, there was an understanding that once detected, fetal anomalies would be “terminated.” The vast majority of women comply. Klinkhammer (1997) reported an overall (not anomaly specific) postdiagnostic abortion rate of 90% in Germany, and Mushaben, Giles, and Lennox (1997) reported a 98% abortion rate of fetuses diagnosed with Down syndrome.

Rita was close to full-term (40 weeks), when I met her. Dr. B, the obstetrician who examined her, spent more than 15 minutes scanning Rita's belly with the ultrasound transducer, talking softly and intermittently, as when he pointed out the beating fetal heart to Rita, and measured and calculated fetal body length. He was kind throughout, but there was a flare of tension when he asked Rita if she had ever seen a child with Trisomy 18. Without waiting for an answer, he described his experience of births of babies with the condition.

When Dr. B finished talking, Rita swung her legs over the side, sat fully upright, took a deep breath, and reiterated her decision to go to term. She seemed weary of having to explain one more time, but she was firm in her resolve to go ahead with the birth. Rita and Dr. B talked energetically back and forth for a while, the antagonistic energy dissipating as they talked. Dr. B later told me later that he felt bad about agitating her. He was only trying to prepare her for what she would see when the baby was born, not convince her that she should have an abortion. (In Germany, a diagnosis of any anticipated disability makes abortion legal up to any point before delivery.)

Two weeks later, Rita arrived at the hospital three centimeters dilated and having regular contractions. At check-in she was still intent on delivering without intervention. The contractions were steady at first, but stalled after about an hour when the baby died, due to the stress of the labor, the obstetricians said. Rita consented to an injection of pitocin to accelerate the contractions, but after several hours of protracted labor, Rita was put under general anesthesia and her four-pound girl was delivered by vacuum extraction. When I asked about the birth the next day, one of the young doctors who had been present at the birth began with, “That baby was not meant to be born.”

I had been granted official research access to the hospital's official Record of Birth, which provide the raw data from which hospital statistics are compiled. I was cross-checking data (e.g., age, weeks pregnant, number of prenatal visits) when I came to Rita's record. I noted that the section of the form where the stillborn birth (Totgeburt) is usually recorded was absent. Not just blank, but nonexistent. That part of the form template had been deleted. No date of death was recorded (Todesdatum), nor was a time of death (Urhzeit) noted. Also irregular in a second section for recording a “deceased child” (Kind verstorben), it read simply nein (no) as if Rita's baby had simply gone home with her mother after the birth and begun her life like other children born at the hospital. When I compared this category to the records of other women who had stillbirths or infant deaths that section read ja (yes). The form had other information, making it possible to infer a stillbirth: APGAR scores of 0 were listed at the one-, five-, and ten-minute intervals, and it was noted that the child did not breathe at delivery and that the pediatrician did not conduct the usual examination. But none of these other data categories count like data in the deceased child category, and no aggregate of other data counts as a death. The prevailing opinion that Rita's baby should not have been born at all appeared manifest in the data not entered in the birth record.

Including Rita's child, there were five deaths among my patient research cohort. Among those five, two more Record of Birth irregularities evidenced. One was the case of a baby that died before it reached the hospital. As a statistical event, dying before reaching hospital places a death in a kind of infant mortality no-man's land; the hospital does not claim responsibility. The final irregularity in the records represented a birth that began about a month into my stay at the hospital when a 40-year-old mother of four, Kamile, came to the hospital complaining of contractions. Not an unusual occurrence in an obstetrical hospital, but Kamile was only five and a half months pregnant. She gave birth two hours later to a one-pound baby boy. The baby did not breathe, was limp on delivery, and presumed dead. The obstetrician, Dr. Y, and the attending midwife turned their attentions to the mother, leaving the baby wrapped loosely in blankets on the warming table as was the norm. After a minute, Dr. Y turned and saw the baby move. He immediately intubated him, opening his airway for oxygen, and the baby was resuscitated. The baby, weighing slightly more than a pound and suffering from multiple infections, lived for ten more days. His birth record too did not reflect a hospital death, although there is no obligation in Germany to report deaths on that form unless the baby dies within seven days. Spaces for the date and time of death had been left blank. The entire deceased child category (Kind verstorben) had been erased; there was white space where a yes or no was indicated. Like with the other two deaths Kamile's child's death had been disappeared.

Publics and Privatizations

The year I conducted research, the hospital was struggling to regain lost status as a premier university hospital. In interviews patients told me unsolicited stories circulating about the hospital. Several women had heard from friends that one of the staff obstetricians had nicked and cut a baby's bottom with a scalpel when he made the incision for a C-section. This is not an unknown occurrence in obstetrics, but interviewees expressed dramatic expressions of panic and outrage over the incident. The most damaging story in circulation was that the neonatal intensive-care unit (NICU) had outbreaks of sepsis resulting in 28 infant infections, and the hospital could not identify the source. There was a police investigation and a formal prosecution. Two premature infants died and a third suffered a severe permanent disability. My research came at the end of the 30-month sepsis outbreak, which was eventually linked to a bacteria present in a bottle of cleaning solution that had been diluted too much, and had been used every day for months to clean common area surfaces, air systems, and infusion pumps. Several of the neonatologists involved published the case in a renowned English-language medical journal, and justified publicizing the case further as a measure of their commitment to medical transparency. I later learned that the lead author was coming up for tenure and needed the publication for his dossier.

What does death data mean to hospital business? In this case, meanings registered on several levels. This unusual and widely publicized incident fostered a heightened sense of diligence about infant mortality by hospital staff during a time when, coincidentally, the overall profitability of the university hospital was also under scrutiny. Even without Rita and Kamile's children's data, the hospital had infant mortality rates about twice the regional rate. Patient intakes dropped almost 20 percent the following year. Interviewees reported that women with other options could choose to deliver their babies elsewhere. Local obstetricians likewise referred their patients to other hospitals. At the hospital, reactions to normative decision-making were reportedly more intense that year. Rita's choice to go to term with a baby the obstetricians thought was highly likely to die, for example, was a greater problem that year. At the very least, pressure for death data to “go missing” was enhanced. When IMR numbers are low, as they are in Germany, each individual infant death bears significant impact on the overall numbers.

Patient loads declined further, policy reforms brought funding cuts, and by the early 2000s the (now) biggest private hospital corporation in Germany, Rhön Klinikum, targeted the hospital for a buyout and bought it. The ensuing private takeover of a public university research hospital was economically and symbolically significant, perceived by many German social democrats as a serious offence to the century-old system of universal health care.

Selling off public hospitals is an issue in Germany. Eighty percent of former East Germans and more than 50 percent former West Germans polled in one study say that the government should operate hospitals (Legge and Rainey 2003), not private enterprises. The privatization we are seeing now is actually the second wave of privatization. The first occurred in the former East German states in the 1990s after reunification when public state hospitals were privatized. Twenty-seven percent of all hospitals in former East Germany and 11 percent in former West Germany are now fully private (Schwierz 2010). Some public hospitals have turned too to private investors for specific upgrades and facilities initiatives (Maarse 2006:1002).

With privatization, the numbers changed. Public health indicators like IMR had been hospital physician-directors’ primary means for measuring success. In the two hospitals where I conducted research, budget and management numbers were of far less interest to the hospital administrators than the maternal and infant public health indicators; profit had not yet become a primary concern for administrators.

When I began the research project, hospital yearly reports published raw maternal and infant health data, typically about 25 to 30 pages’ worth. The Jahresbericht (annual review) gave physicians, administrators, and researchers like me access to raw data. I knew the data were imperfect, but, nevertheless, the data were a boon to me because I was able to think through my research questions and compare my own quantitative research results with “official” data. By 2004, year six of my research project, maternal and infant health data in the Jahresbericht had been reduced to four pages of summary tables, mostly patient counts. In 2006, the eighth research year and the first official year of the hospital takeover, the Jahresbericht became the Qualitätsbericht (quality record), using Rhön Klinikum's reporting template standard for its nearly 50 hospitals. It reflected that, as was the case, two university hospitals had merged into one in the takeover, but there was no data originating from the hospital where I had worked. Three pages of scant tables were devoted to the other hospital's obstetrical data, mostly patient counts. Gone were the columns of more than 90 obstetrical data points, by year and compared to state data as a whole. In the new Qualitätsbericht, the published health data now consisted of new “quality indicators,” which included patient survey responses to 20 questions such as “Are you happy with your current course of treatment?” and “How was our signage?”

Rhön Klinikum's reports provided much less raw data and operated more like a marketing tool than a public health data document. By 2008, Qualitäts-, the German word for quality, was so oft recurring and self-referential in the yearly report—used 284 times—as to be rendered meaningless, a hollow linguistic alibi-making that evoked a (surprising) nostalgia in me for the Jahresbericht. The report is replete with self-validating assurances that quality indicators are improving quality health care and increasing data transparency. With Qualitäts- we see a script being performed as if real, again invoking Çalışkan and Callon's performative actionization.

The (over)use of Qualitäts- was likely instigated by a new German law in 2007 directing hospitals to enumerate “quality assurance” data in 24 areas. The value of the indicators is reportedly their ability to link “quality” to “performance.” Hospitals can be “linked” to substandard care based on, for example, counts of complications after surgery. Complications “hurt a hospital's financial performance” (Mohlmann, Then, and Wichels 2009:59), and hospital survival is dependent on profitability. In brief, hospitals will be closed when patients are too sick to make the hospital profitable; the narrative is that the hospitals must have provided negligent care.

The western Germany hospital where I worked may be one of the first adversely affected by the new profitability standards. An early 2011 article in the weekly peer-reviewed German Medical Association journal reports that three physicians from that hospital have gone public to complain about the “medical irregularities” they attribute directly to the privatization of the hospital and the “commodification of the patients.” Citing deteriorating hygienic conditions and too few staff, the physicians went on television to report patient complaints. In response, they have been censured by Rhön Klinikum and threatened with a defamation law suit. A German study correlating poor hospital performance with the risk of insolvency (Mohlmann et al. 2009) may foretell that should Rhön Klinikum's profitability be further affected, there may be hospital for sale.

CONCLUSION

As business apparatus, health statistics can operate as global operands—manipulatable objects—in the service of the worldwide market fundamentalisms that have bureaucratized health provision in our advanced capitalist epoch. A health-as-commodity mindset has shaped new hegemonies of normative health praxis and an unprecedented proliferation of numbers. In global health today, statistics are produced at ecclesiastical proportions not so much because their use routinely improves the most entrenched and challenging health problems (see Adams 2010), but because they work so well at organizing people to perform in particular economic and political ways, making calculable the chaotic, and constituting subject positions familiar to the world system.

There are several arguments in this article: Health statistics the world over are usually at best incomplete; the pursuit of acontextual aggregated quantitative measures can have unintended and negative social affects, such as the deepening of classed and raced divides; and that rendering human suffering statistical robs it of intimacy and depth necessary for remedy. The inescapable point of this article, however, is that global political and economic transformations of the past decades have required governance that exerts power “not so much over a fixed territory as over a [global] multitude” moved along channels by administrative apparatus. At this particular historical moment, statistics serve as “a central knowledge-based activity indispensable for [global] salvation” (both quotes, Foucault in Rabinow and Rose 1994:259, 260). In places as different as Sierra Leone and Germany, statistical imperatives have intensified and moved us beyond traditional geopolitical modalities of data production in ways that work toward restructuring global health labor, information, and capital, if not always improving global health. Global applications of health statistics have multiple effects, but few are more compelling and fundamental than this: If it can't be enumerated, it won't work.

ACKNOWLEDGMENTS

For essential insights on earlier versions of this article I am indebted to Carole Browner, Dara Culhane, Elizabeth Dunn, Michael Hathaway, Margaret Lock, Rob Lorway, Stacy Pigg, Liz Roberts, and three anonymous reviewers. The National Science Foundation, the Wenner-Gren Foundation for Anthropological Research, and Simon Fraser University supported research for this article.

Additional information

Notes on contributors

Susan L. Erikson

Dr. SUSAN L. ERIKSON is a medical anthropologist and, since 2007, has been an assistant professor of global health in the Faculty of Health Sciences at Simon Fraser University, near Vancouver, British Columbia, Canada. Prior to becoming an anthropologist, she had a career in international development. She is the founding director of the Global Health Affairs program at the Korbel School of International Studies at the University of Denver.

Notes

1. See Graeber 2010.