The etiologic agent associated with the epidemic is Leishmania donovani, and the sand fly vector is Phlebotomus orientalis (whose habitat is Acacia-Balanites woodland) [10]. Vector-control strategies have not been feasible because of logistic and war-related constraints. It is not known whether humans are the main reservoirs of L. donovani in Western Upper Nile and whether treating patients has been an effective control measure.

Medecins Sans Frontieres established visceral leishmaniasis treatment centers in Western Upper Nile in the village of Ler in May 1989 and in the village of Duar (80 km north of Ler and reportedly near the initial focus of the epidemic) in July 1990 (Figure 1). Approximately 19,000 patients were treated from 1989 to 1995, and about three quarters of these were treated in Duar. To identify patient characteristics associated with various treatment outcomes, we analyzed data for a cohort of 3076 patients who were treated in Duar with the pentavalent antimonial compound sodium stibogluconate (Pentostam, Wellcome Foundation Ltd., London, United Kingdom) [11] during the first year the center was operational (August 1990 to July 1991).

View larger version (14K): [in this window] [in a new window]   Figure 1. Map of Sudan. The sites (Duar and Ler) of Medicins sans Frontieres-Holland's two treatment centers for visceral leishmaniasis in Western Upper Nile province, southern Sudan, are shown. Khartoum is in northern Sudan.

The patient census peaked at about 800 patients at the end of the cohort period (July 1991) and thereafter increased to more than 1500 patients (August to November 1991). The treatment center was staffed by an average of two expatriates and about 25 Sudanese health auxiliaries trained by the expatriates. To reach Duar, patients typically walked or were carried for days. After "admission," they stayed in crowded conditions, without access to latrines or adequate shelter, food, or water. The light aircraft that was the center's only means of transporting supplies could not accommodate bulky shipments of food or other provisions. In "clinics" held outdoors under shade trees, patients received daily parenteral therapy for about a month. Despite their severe debility and the exceptionally difficult circumstances under which they were treated, most fared remarkably well.

Methods

Top Methods Results Discussion Author & Article Info References

Diagnosis of Visceral Leishmaniasis and Study Cohort

The diagnosis of visceral leishmaniasis was considered for persons who reportedly had been febrile for at least 2 weeks and had either palpable splenomegaly or wasting and lymphadenopathy. The diagnosis was supported serologically with a direct agglutination test [12, 13] by analysis of finger-prick blood specimens collected on filter paper. The definition of a positive serologic result varied depending on the time elapsed since the transmission season and the batch of antigen used. A Giemsa-stained smear of a specimen obtained by splenic aspiration (lymph node aspiration was used if splenic aspiration was considered too dangerous [for example, in young children]) was examined for parasites 1) if expeditious diagnosis was needed, 2) if the serologic result was negative or borderline but leishmaniasis was the likely diagnosis, or 3) if supplies for serologic testing were unavailable.

Of the approximately 5000 persons evaluated for visceral leishmaniasis in Duar during the study period, about 3335 (66.7%) were admitted with this diagnosis and received at least some antimonial therapy. The study cohort consisted of 3076 patients (92.2% of 3335) previously untreated for visceral leishmaniasis whose medical records were found. Data were extracted from these records, clinic registration books, and monthly summary reports.

Treatment of Visceral Leishmaniasis

Patients received once-daily intramuscular injections of sodium stibogluconate (20 mg of pentavalent antimony/kg of body weight) for 30 consecutive days. By protocol, the minimum and maximum daily doses were 200 mg and 850 mg, respectively. Therefore, the 1383 adults ( 18 years of age) in the cohort (median weight, 47.0 kg) received a median daily dose of 18.1 mg of antimony/kg. Vitamin A was given on admission to nonpregnant patients; other vitamins, folic acid, and ferrous sulfate were provided daily; and chloroquine prophylaxis for malaria was given weekly. Supplemental nutrition could not be provided routinely.

A test-of-cure specimen was obtained, typically on the 25th day of treatment, through aspiration of the spleen or a lymph node. If the specimen was positive for parasites, treatment was continued until two consecutive weekly specimens were negative. If parasites were still detectable after 60 days of therapy, the treatment regimen was changed from a once-daily intramuscular dose of 20 mg of antimony/kg (maximum daily dose, 850 mg) to twice-daily intravenous doses of 10 mg of antimony/kg (no maximum dose in mg). Treatment was continued until two consecutive specimens were negative.

Discharged patients who reported recurrent symptoms (including fever for more than equals 2 weeks) and had a positive splenic or lymph node aspirate were retreated and classified as having relapsed; 15 such patients were readmitted on clinical grounds, most of them during periods when expatriate staff had been evacuated. Patients readmitted by the end of 1994 (3.5 years after the end of the study period) were included in the analyses. After treatment, patients could not be routinely monitored for relapse.

Study Variables

The following patient characteristics were assessed on admission (unless otherwise specified) and included in the analyses: age (usually estimated by the staff), sex, self-reported duration of illness, hemoglobin level, body mass index (measured in patients aged 18 years who did not have obvious signs of pregnancy; measured as weight in kg/height in m² [14]), spleen size (Hackett classes 1 to 5 [15]), grade of parasite density in a splenic aspirate (negative = 0, positive = 1 to 6 [16, 17]), and symptoms during the treatment course. Patients were asked each day about vomiting, diarrhea ( 3 loose or watery stools in 24 hours), and bleeding (epistaxis, bleeding gums, and hematemesis). Data on diarrhea and bleeding were available only for patients admitted from March to July 1991.

We assessed four mutually exclusive treatment outcomes:

1. Default: treatment stopped against medical advice without receipt of at least 25 doses of sodium stibogluconate, including, if indicated, the additional doses specified after a test-of-cure specimen was positive for parasites (number of patients who defaulted/number of patients admitted).

2. Death: death for any reason during the treatment course (number of patients who died/number of patients admitted).

3. Relapse: discharge after receipt of at least 25 drug doses (including, if indicated, the additional doses specified after a test-of-cure specimen was positive for parasites) and readmission by the end of 1994 because of relapsed visceral leishmaniasis (number of patients who relapsed/[number of patients admitted – numbers of patients who defaulted or died during the first treatment course]).

4. Successful treatment: discharge after receipt of at least 25 drug doses (including, if indicated, the additional doses specified after a test-of-cure specimen was positive for parasites) and no readmission by the end of 1994 for retreatment (number of patients successfully treated/number of patients admitted).

Statistical Analysis

We explored the associations between patient characteristics and treatment outcomes by doing univariable analyses and calculating risks. To assess statistical significance, we used approximate 95% CIs and, when appropriate, the chi-square test for trend. We used the Wilcoxon two-sample test to compare the ranked distributions of ordinal variables. Associations between characteristics and outcomes were further assessed by multivariable analyses. We used MULTLR software for unconditional multiple logistic regression [18] with maximum likelihood fitting [19] and used adjusted odds ratios with 95% CIs to approximate risk ratios for death (see below for variables included in the models). The word "significantly" indicates that the CI excludes 1.0 or that the P value is less than 0.05.

We studied nutritional status (body mass index) only for adults because children were not measured in a standardized way or with sufficient care to justify analysis of their nutritional data. Of all 1383 adults in the cohort, 1269 (91.8%) had their height and weight recorded (that is, their body mass index could be calculated), as well as age, sex, duration of illness, hemoglobin level, and spleen size. Of these, 1207 (95.1%) did not default or relapse and thus constitute the so-called adult cohort. The values (medians or proportions) of the patient characteristics for the adult cohort were not significantly different from the values for those patients of the 176 adults not included in the adult cohort for whom data about the given characteristic were available. The exception was that the adult cohort included patients who were somewhat more malnourished (median body mass index of 15.2 kg/m² compared with 15.7 kg/m²; P = 0.04). We used this cohort and a subset (the so-called complete-data cohort [n = 385 adults]) in multivariable analysis to address the risk for death. For the adults in the complete-data cohort, information was available not only about the adult-cohort variables but also about parasite density on admission (determined only under certain conditions) and about vomiting. This cohort included 52.4% of the 735 adults for whom information about vomiting was reliably recorded and 27.8% of all 1383 adults in the study cohort. Compared with those patients in the group of 998 adults who were not included in the complete-data cohort for whom data about the given characteristic were available, the complete-data cohort included more women (54.3% [CI, 49.3% to 59.3%] compared with 37.6% [CI, 34.6% to 40.6%]); it also included patients who were more anemic (median hemoglobin level, 75 g/L compared with 85 g/L), were more malnourished (median body mass index of 14.9 kg/m² compared with 15.5 kg/m²), and had larger spleens (14.5% compared with 8.0% had a spleen size of 4 or 5) (P < 0.001 for these three comparisons).

Results

Top Methods Results Discussion Author & Article Info References

The study cohort of 3076 patients ranged in age from younger than 1 year to 79 years (median, 15 years; Table 1). Young children (< 5 years of age) and older adults ( 45 years of age) were underrepresented (9.5% and 6.4% of the cohort, respectively, compared with the 1985 estimates for Sudan of 18.2% and 13.2%, respectively [20]). Slightly more than half (53.5%) of the patients were male; male overrepresentation was most prominent in the 25-to 34-year-old age group (63.5% males). Most patients (91.0%) had been sick less than 5 months. Overall, the patients were notably anemic (median hemoglobin level, 77 g/L; 16.9% had a hemoglobin level less than 60 g/L) and malnourished (median body mass index for adults, 15.2 kg/m²; 9.9% had a body mass index less than 13 kg/m², and 3.6% had a body mass index less than 12 kg/m²).

Most patients (2562 [83.3%]) were successfully treated (Table 1); these patients received a median of 30 drug doses (range, 25 to 72 doses; 115 [4.5%] received more than 40 doses). Only 99 patients (3.2%), 14 of whom were later readmitted for treatment, defaulted during their first admission; 336 patients (10.9%) died during their first admission; and 79 patients are known to have relapsed (3.0% of the 2641 patients who were discharged alive [that is, who had a final treatment outcome of successful treatment or relapse] and 2.6% of the entire cohort of 3076 patients). One patient classified as having relapsed and eight patients classified as having defaulted are known to have died, for a total of 345 known deaths (11.2%). Of the 99 patients who defaulted, 25 (25.3%) left the treatment center after receiving fewer than 5 drug doses; 27 (27.3%), after 5 to 14 doses; and 19 (19.2%), after 15 to 24 doses. Twenty-eight (28.3%) did not stay for extended therapy after a test-of-cure specimen was found to be positive for parasites and thus were conservatively classified as having defaulted (26 patients defaulted after receiving more than equals 30 doses).

The 79 patients who are known to have relapsed received a median of 30 drug doses during their first admission (range, 28 to 97 doses; 6 [7.6%] received more than 40 doses) and were readmitted 1 to 28 months after they had been discharged (median, 4 months; more than 24 months for one patient). In univariable analysis Table 1, the risk for relapse was higher (but not significantly) for the youngest and oldest patients and for patients who had been ill for at least 5 months before the first admission. The risk was similar for male and female patients and did not vary much in relation to hemoglobin level, spleen size, or the presence of symptoms (such as vomiting). Most adults with a body mass index of at least 13 kg/m² had a similar risk for relapse; about one third of the more malnourished patients (body mass index less than 13 kg/m²) had died during the initial admission.

Of the 336 patients (10.9%) who died during their first admission, 95 (28.3%) died after receiving fewer than 5 drug doses; 146 (43.5%), after 5 to 14 doses; 63 (18.8%), after 15 to 24 doses; and 32 (9.5%), after 25 or more doses. Patients who died after receiving fewer than 5 doses were more likely to have been young children than were patients who died after receiving 5 or more doses: 26 of 95 (27.4% [CI, 18.4% to 36.4%]) compared with 32 of 241 (13.3% [CI, 9.0% to 17.6%]). During the 5-year period for which data were available (1990 to 1994), the risk for death did not vary consistently by season or vary appreciably with changes in the patient census (the latter measure was used as a surrogate for workload).

In univariable analysis, young children; older adults; and patients who had been sick for a long time ( 5 months), were severely anemic, were markedly malnourished, had a large spleen, or had a high parasite density were overrepresented among those who died (Table 1). In contrast, similar proportions of male and female patients died. For patients who died and patients who were successfully treated, the median ages were 20.0 years and 15.0 years, respectively (P = 0.008); the median hemoglobin levels were 69 g/L and 79 g/L, respectively (P < 0.001); and the median body mass indexes were 14.3 kg/m² and 15.4 kg/m², respectively (P < 0.001). For patients 5 years of age or older, the risk for death increased as age increased (test for trend, P < 0.001), as the hemoglobin level decreased (P < 0.001), and as body mass index decreased (P < 0.001).

Young children were at increased risk for death, particularly the 91 who were severely anemic (hemoglobin level less than 60 g/L; risk, 0.41 [CI, 0.31 to 0.51]); nearly two thirds of the young children who died were severely anemic. The risk for death was also high for older adults, particularly the 20 with severe anemia (risk, 0.45 [CI, 0.23 to 0.67]) and the 11 with extreme cachexia (body mass index less than 12 kg/m²; risk, 0.64 [CI, 0.35 to 0.92]).

Patients who died were also more likely to have vomited at least once during their treatment course, but bleeding and diarrhea were not associated with death. About half (44.4% [CI, 38.3% to 50.5%]) of those who died had vomited compared with only 26.6% (CI, 24.3% to 28.9%) of those who were successfully treated. No clear seasonality was noted for vomiting (1989 to 1994).

In multivariable analysis of data for the adult cohort (1207 adults; Table 2), the approximate risk ratios for death were significantly elevated at 2.2 for patients who had been sick for at least 5 months, 3.6 for those 45 years of age or older (see bottom of Table 2 for data on patients of all ages), 4.6 for those with a hemoglobin level less than 60 g/L, and 12.2 for those with a body mass index less than 12 kg/m². In multivariable analysis of data for the complete-data cohort (385 adults; 75 [19.5%] of whom died), the approximate risk ratios for death were significantly elevated at 2.0 (CI, 1.1 to 3.6) for patients with a parasite density grade of 5 or 6 (compared with patients whose grade was 0 to 4), 2.4 (CI, 1.2 to 4.9) for those 45 years of age or older (compared with patients 18 to 44 years of age), 2.6 (CI, 1.4 to 4.9) for those with a hemoglobin level less than 60 g/L (compared with patients whose level was more than equals 60 g/L), 2.8 (CI, 1.6 to 5.0) for those who had vomited (compared with patients who had not), and 3.0 (CI, 1.7 to 5.2) for those with a body mass index less than 14.0 kg/m² (compared with patients whose index was more than equals 14 kg/m²). The approximate risk ratio was decreased for female patients compared with male patients (0.4 [CI, 0.2 to 0.7]). Spleen size and duration of illness were not significant in this model and therefore were not retained in it; interaction terms were also not significant. Overall, 141 (36.6%) patients in the complete-data cohort had vomited at least once during the treatment course (compared with 256 [34.8%] of the 735 adults for whom information about vomiting was recorded). Sex was the only patient characteristic significantly associated with vomiting (approximate risk ratio for female patients compared with male patients, 1.9 [CI, 1.2 to 2.9]).

Discussion

Top Methods Results Discussion Author & Article Info References

Our cohort included persons of all ages, consistent with the recent introduction of visceral leishmaniasis into Western Upper Nile. Young children (< 5 years of age), older adults ( 45 years of age), and female patients were somewhat underrepresented; however, after the cohort period, the proportion of young children increased to 17.9% in 1992, to 14.9% in 1993, and to 30.5% in 1994. The female-to-male ratios were 0.93 in 1992, 0.77 in 1993, and 1.14 in 1994 [9].

Although it may be intuitive that patients with certain characteristics (such as severe anemia) were least likely to survive, we were surprised that most patients fared well. Overall, 2562 (83.3%) were successfully treated, despite severe debility, therapy with a potentially toxic compound and no laboratory monitoring, inadequate nutritional support, no blood transfusions, and minimal treatment of concurrent illnesses. Even severely debilitated patients usually survived. In fact, 11 (50.0%) of the 22 men and 17 (68.0%) of the 25 women with a body mass index less than 12 kg/m² were successfully treated; a body mass index less than 12 kg/m² has been considered incompatible with life, at least for males [21].

It is problematic to compare the death rate we observed (about 11%, regardless of whether only first or both first and second admission deaths are included; the death rate is 14.0% if we conservatively assume that patients who defaulted and whose outcome is unknown later died) with the wide range of previously reported rates (0% to more than 50%) for patients with visceral leishmaniasis treated in other settings [11, 22-25]. Most published data were for healthier patients treated under more favorable circumstances. The closest comparison group is probably the group of patients displaced, because of the civil war, from Western Upper Nile and treated in a provisional hospital in Khartoum (Figure 1). In a study done from January 1989 to February 1990 (before the July 1990 opening of the center in Duar) of 693 such patients [23], the overall case-fatality rate was 11.3% (12% for 623 patients receiving a once-daily intravenous dose of 10 mg of antimony/kg for 30 days and 6% for 70 patients receiving twice-daily intravenous doses of 10 mg of antimony/kg for 15 days). Unlike patients treated in Duar, patients in Khartoum received 3000 to 3500 kcal/d; patients in Khartoum with symptomatic anemia and a hemoglobin level less than 40 g/L also received blood transfusions. The death rate in Duar has decreased from about 11% (our 1990-1991 cohort) to about 8.5% (162 of 1916 patients) in 1992, 7.2% (64 of 892 patients) in 1993, and 6.6% (96 of 1448 patients) in 1994 (these rates reflect the periods during these years for which data are available). These decreases probably occurred in large part because the patients' status on admission and the quality of medical care have improved.

Because we could not thoroughly evaluate our patients for concurrent illnesses, we could not distinguish dying "with" from dying "of" leishmaniasis. Fewer patients may have died if the patients had had fewer comorbid illnesses and less severe cachexia and if such conditions could have been aggressively treated. On the other hand, had our patients been less debilitated, it is probable that fewer would have developed such severe visceral leishmaniasis, which is usually fatal if not appropriately treated. Although some of our patients' deaths may have been partially attributable to drug toxicity, many patients died early in the treatment course; antimonial toxicity, however, is usually cumulative [11]. Fortunately, coinfection with the human immunodeficiency virus (HIV) apparently has not contributed to the severity of the outbreak; serosurveys conducted as late as the fall of 1994 did not identify any HIV-positive patients (6, 23; Seaman J. Unpublished data).

We identified the following risk factors for death during treatment: young and older age, long duration of illness, severe anemia, markedly low body mass index, high parasite density, and vomiting. Some factors were interrelated. For example, young children and older adults were at even higher risk if they were severely anemic (hemoglobin level less than 60 g/L), as were patients of all ages. We could not determine the extent to which the observed anemia was directly or indirectly caused by leishmaniasis rather than by other prevalent diseases (such as malaria). The monthly proportions of patients with severe anemia ranged from 7.3% to 27.9% during the study period and ranged from 5% to 9% in 1993-1994.

Malnutrition has previously been shown to be both a risk factor for and a consequence of visceral leishmaniasis [26-30]. Our patients' cachexia, like their anemia, was probably multifactorial. However, we believe that the most markedly cachectic persons in the region were those with leishmaniasis. Only 101 (7.7%) of the adults in our cohort had a body mass index of 18 kg/m² or more, whereas 866 (65.6%) had a body mass index less than 16 kg/m². In 1993 in Ler, the proportion of patients 16 years of age or older who had a body mass index less than 16 kg/m (²) decreased to less than 40%; in Duar, however, the proportion was still about 50% to 60%. A body mass index less than 16 kg/m² is generally considered to indicate severe malnutrition [14]. The appropriateness of this standard for persons of the Nilotic tribes of Sudan, who are typically tall and slender, may be debatable; however, a body mass index less than 16 kg/m² certainly indicates the presence of some wasting, and a body mass index less than 14 kg/m² indicates marked wasting. Although our patients may have fared better had supplementary nutrition been provided, most patients survived despite receiving only antileishmanial therapy.

Vomiting during the treatment course was another risk factor for death. The analysis did not consider the severity, frequency, or cause (for example, pancreatitis from drug toxicity [31]) of the vomiting. Although patients who vomited were more likely also to have had diarrhea sometime during their treatment course, having diarrhea (with or without vomiting) did not increase the risk for death. However, perhaps only certain subgroups of patients with diarrhea had a worse prognosis. For example, analysis of data for 2 months of the study period suggested that patients who had diarrhea associated with at least 10% loss of body mass were at increased risk for death (data not shown). The results of a study done in Duar in 1992 suggested that much of the nosocomial diarrhea may have been bacillary (Herwaldt BL. Unpublished data).

The relapse rate of 3.0% (of patients discharged alive) that we observed for Duar in 1990-1991 is similar to the rate of 4.6% seen in Khartoum in 1989-1990 [23]. Because relatively few patients had relapsed, our ability to address risk factors for relapse was limited. We could not monitor our patients after discharge and therefore sought to minimize the risk for relapse by continuing to treat if a test-of-cure specimen was positive for parasites, even though persistence of some parasites does not necessarily predict failure to cure [11].

We may have somewhat underestimated the relapse (and death) rate; however, given the lack of alternatives, patients who relapsed probably returned to Duar unless they were restrained by such factors as severe debility, death, seasonal flooding, or war-related insecurity. In fact, a 1994 follow-up study of 191 persons from a relatively accessible district who were evaluated for visceral leishmaniasis in 1991 [8] showed that none of the 45 patients treated in the interim had subsequently relapsed without seeking treatment (Seaman J. Unpublished data). Thus, our passive post-treatment monitoring system probably does not substantially underestimate the relapse rate. In addition, the system can detect changes in the rate. For example, the relapse rate among the patients from the three most accessible districts temporarily increased from 4.8% (CI, 3.7% to 5.9%; this is the rate for the cohort period for these three districts, which account for 58.6% of the patients who were discharged alive) to 5.9% (CI, 4.3% to 7.5%) for patients admitted in October 1991 and to 9.9% (CI, 7.1% to 12.7%) for patients admitted in November 1991 (post-cohort period). In these 2 months, the expatriates were evacuated for several weeks, the patient census was high, and the treatment course was shortened to 24 days or less.

Although not all patients in the post-cohort period have responded to antimonial therapy [32], continued high rates of successful treatment suggest that L. donovani remains highly susceptible to therapy. In contrast, in the epidemic in India, decreasing responsiveness to antimonial therapy has been more problematic and may in part be attributable to the inappropriate use of sodium stibogluconate by local practitioners [33]. Despite the chaotic circumstances in Western Upper Nile, treatment has been provided in a relatively controlled way and according to standardized protocols. This approach may have hindered the development of drug resistance.

The characteristics of our patients and the limitations of our data and analyses should be considered. Only persons who accessed medical care and met specified criteria were included in our clinic-based cohort. Persons who did not or could not seek or reach medical care were not enumerated. We did not have the luxury of obtaining detailed medical histories or of doing thorough physical examinations and sophisticated laboratory testing. Regrettably, we did not have sufficiently complete data on the numbers of patients who developed post-kala-azar dermal leishmaniasis to include this outcome in our analyses. Some of the poor prognostic factors we identified were probably only partially attributable to leishmaniasis.

Despite such limitations, we could identify useful indicators for determining which patients are at particularly high risk for adverse outcomes. We now monitor such patients more carefully and have made low-technologic changes in our treatment protocols. Although giving blood transfusions is still infeasible, severely anemic patients and pregnant women receive vitamin K therapy because of the possibility of concomitant vitamin K-responsive coagulopathy. To prevent the development of severe anemia and other sequelae of Plasmodium falciparum infection, which may be chloroquine-resistant, patients 1 year of age or older are now treated on admission with pyrimethamine-sulfadoxine instead of chloroquine.

Although routine provision of nutritional support remains problematic, data on our patients' poor nutritional status have helped persuade donors to provide food aid when possible. Malnourished children younger than 5 years of age are treated for 5 days with a broad-spectrum antibiotic because of their increased risk for bacterial infection [34]. Patients who vomit are treated aggressively with oral rehydration therapy and antiemetic drugs. We temporarily withhold antimonial treatment if vomiting is intractable and sometimes substitute an alternative antileishmanial agent [13, 32]. Although treating certain comorbid illnesses remains problematic, therapy for tuberculosis is now available.

Treating patients who have leishmaniasis remains challenging, in part because the ongoing civil war restricts relief efforts. Although the patient census decreased somewhat in 1992 and 1993 after peaking during the late summer and fall of 1991, it increased again beginning in July 1994. Some may question whether the effort entailed in providing parenteral antileishmanial therapy in this remote and unstable area of the world is warranted, especially because those who survive in the short-term return home to face death from other diseases as well as from war-related causes. However, the long treks the patients make to Duar, to be treated for what they call the "killing disease," presumably manifest a desire to live, despite their dire predicament. Just as the full extent of the morbidity and mortality attributable to the epidemic is incalculable, so are the long-term community-wide benefits of treating patients fortunate enough to access medical care. We trust that the treatment provided to the patients in this cohort and to the more than 16 000 other patients with visceral leishmaniasis treated by Medecins Sans Frontieres in southern Sudan have helped not only to save individual lives and to lead to improved medical care but also to give hope to families and to preserve communities.

Mr. Mercer: Centre for Development Studies, University of Wales Swansea, Swansea SA2 8PP, Wales.

Dr. Sondorp: HealthNet International, Singel 540, 1017 AZ Amsterdam, the Netherlands.

Dr. Herwaldt: Centers for Disease Control and Prevention, Division of Parasitic Diseases, Mailstop F-22, 4770 Buford Highway NE, Atlanta, GA 30341-3724.