Search

Objective: To develop a human immunodeficiency virus (HIV) staging system for sub-Saharan Africa on the basis of an evaluation of the World Health Organization (WHO) system and predictors of mortality., Design: Prospective cohort study with 4 years of follow-up., Setting: Kigali, Rwanda., Patients: 412 HIV-infected women recruited from prenatal and pediatric clinics., Measurements: Clinical signs and symptoms of HIV disease, laboratory assays (including complete blood count and erythrocyte sedimentation rate), and cumulative mortality., Results: The WHO staging system includes a clinical and a laboratory axis. The clinical axis was revised by inclusion of oral candidiasis, chronic oral or genital ulcers, and pulmonary tuberculosis as "severe" disease (clinical stage IV); in addition, body mass index was substituted for weight loss in the definition for the wasting syndrome. The 36-month cumulative mortality was 7% for women in modified clinical stage I ("asymptomatic"), 15% for those in stage II, 19% for those in stage III, and 36% for those in stage IV (P < 0.001). The laboratory axis was revised by replacing lymphocyte count with hematocrit and erythrocyte sedimentation rate. The 36-month mortality was 10% for women in modified stage A ("normal" laboratory results) and 33% for those in stage B (erythrocyte sedimentation rate > 65 mm/h or hematocrit < 0.38) (P < 0.001). A single staging system combining clinical and laboratory criteria is proposed, with a 36-month mortality of 7% for women in combined stage I, 10% for those in stage II, 29% for those in stage III, and 62% for those in stage IV (P < 0.001)., Conclusions: On the basis of this analysis, a staging system relevant for sub-Saharan Africa is proposed that reflects the range of HIV-related outcomes, has strong prognostic significance, includes inexpensive and available laboratory tests, and can be used by both clinicians and researchers.

Our objective was to describe associations among male circumcision, behavioral and demographic variables, ulcerative and nonulcerative sexually transmitted disease (STD), and human immunodeficiency virus (HIV) infection via a cross-sectional study in Kigali, the capital of Rwanda. Our subjects were 837 married men who volunteered for HIV testing and counselling. Uncircumcised men had a relatively low-risk profile in that they reported fewer lifetime sexual partners and prostitute contacts than circumcised men and were more likely to live in rural areas with lower HIV prevalence rates. Uncircumcised men were also less likely to report a history of sexually transmitted disease (64% versus 73%, p = 0.01), although they were more likely to report genital ulceration (GUD) (24% versus 17%, p < 0.03) and to have inguinal adenopathy noted on physical exam (42% versus 29%, p = 0.009). Despite the low-risk profile, uncircumcised men had a higher prevalence of HIV infection than circumcised men (29% versus 21% HIV positive, p = 0.02), which was most marked in men reporting five or more lifetime sex partners (36% versus 23% HIV positive, p = 0.005) or contact with prostitutes (35% versus 23% HIV positive, p = 0.009). Circumcision remained a predictor of HIV infection in multivariate analyses (multivariate odds ratio 1.69, 95% confidence interval 1.16-2.47). Lack of circumcision is associated with a higher risk of HIV infection in Rwandan men. Further research is needed to determine whether this higher risk is due in part to poor hygiene or to complex mechanisms operating through the acquisition of other sexually transmitted diseases. Circumcision may be an appropriate risk reduction approach for men with known exposures to the virus when there are constraints to alternatives, such as condom use.

To approximate the contributions of in utero, intrapartum, and postnatal transmission of human immunodeficiency virus type-1 (HIV-1) and to evaluate polymerase chain reaction (PCR) as a diagnostic tool for pediatric HIV infection, blood was collected at birth (cord blood), and at 3, 6-12, and 13-24 months in 218 children born to HIV-1-seropositive mothers in Kigali, Rwanda. Proviral DNA was detected by a double PCR using two sets of three primers (gag, pol, and env). Pediatric HIV-1 infection was defined according to serological and clinical criteria. The probability of having a positive PCR at a given time was calculated by a nonparametric method. Among children with unequivocal evidence of infection (n = 47), it was 30.5% on cord blood and 80.6% at 3 months. Thus, in children born to HIV-1-infected mothers, the estimated rate of transmission in the late postnatal period is 4.9%, and the rate of transmission in the intrapartum plus postnatal periods is 17.6%. Among 117 HIV-1-uninfected children born to HIV-1-infected mothers, six (5%) had a false-positive PCR on cord blood. These results should be taken into account in designing intervention trials aimed at reducing mother-to-child transmission of HIV-1.

Objective: To estimate the seroincidence of HIV-1 infection among women of reproductive age in Kigali, Rwanda., Design: Fixed prospective cohort followed for 36 months between November 1988 and June 1992, as part of an ongoing study of mother-to-child transmission of HIV-1., Setting: Centre Hospitalier, Kigali, Rwanda., Subjects: A total of 216 HIV-seronegative women were enrolled at delivery between November 1988 and June 1989., Methods: A blood sample was obtained at delivery to test for HIV antibodies (by enzyme-linked immunosorbent assay and Western blot). Serum was tested every 3 months during follow-up. Incidence density rates of HIV seroconversion were estimated., Results: The follow-up rate after 3 years was 89%, assessed by the maximum person-years method. The seroincidence density rate was 3.5 per 100 women-years (95% confidence interval, 1.9-5.0). It decreased linearly from 7.6 during the first 6-months postpartum to 2.5 per 100 women-years during the last 6 months of the third year of follow-up. Maternal age did not affect HIV incidence rates. We examined the role of the cohort, counselling, and the first 6-month postpartum effects on this estimate., Conclusion: This fixed cohort provided an overall estimation of the HIV infection incidence rate and its dynamics. These figures could be used for programming future HIV preventive vaccine efficacy trials in Rwanda.

Five questions were raised in 1986 regarding routine immunization of children infected by the HIV: do vaccines protect these children, both in terms of immunogenicity and clinical efficacy? is immunization, particularly with live attenuated vaccines associated with an increased risk of adverse events? could the stimulation by vaccine antigens precipitate the course of paediatric HIV infection and therefore be dangerous? what are the clinical and epidemiological features of vaccine preventable diseases among HIV-infected children? what is the risk of nosocomial transmission of HIV associated with immunization practices? Based on the best available information, the WHO formulated recommendations in 1987 and updated them in 1989. These recommendations are in general agreement with those proposed during the same period in the USA and in France (table 1). This paper provides an update on the scientific knowledge in this field, focusing on routine childhood immunization in the context of HIV infection, especially in developing countries. The cases of bacillus Calmette-Guérin (BCG), measles vaccine, diphtheria-tetanus-pertussis and poliomyelitis vaccines are reviewed. For each of these antigens, the experience of the authors in Kigali, the capital city of Rwanda, is used as an example. A brief overview of the issue of adult immunization in the context of HIV infection concludes this review. Paediatric HIV infection should not be considered as a limiting factor in the implementation and the progression of the EPI worldwide. Experience accumulated over the last seven years, particularly in Africa, indicates that the WHO recommendations should not be modified.

To investigate nutritional status and heterosexual human immunodeficiency virus (HIV) transmission, we performed a nested case-control study of sexually active, adult women in Kigali, Rwanda. Forty-five women who seroconverted during the 24-month study period were compared to 74 women who remained seronegative throughout the study. Seroconvertors and nonseroconvertors did not differ in preseroconversion serum levels of vitamin A, carotenoids, vitamin E, selenium, albumin, ferritin, or cholesterol. Weight loss, however, was a significant predictor of eventual HIV seroconversion. Subsequent seroconvertors lost an average of 1.5 kg during the first 6 months of the study compared with a 1.0-kg gain (p = 0.001) for nonconvertors. Nine of 27 (33%) seroconvertors, compared with one of 44 (2%) controls, lost at least 5 kg in the 6-month period beginning 1 year before their seroconversion (odds ratio, 21.5, 95% confidence interval 4.1-112). The association between weight loss and seroconversion was independent of other potential risk factors such as socioeconomic status, pregnancy, and genital ulcer disease. In addition to these findings for measured weight loss during follow-up, reported weight loss before enrollment was also a risk factor for subsequent seroconversion. Additional studies of heterosexual HIV transmission are needed to determine whether or not weight loss is causally related to susceptibility for HIV infection.

Objective: We evaluated the impact of human immunodeficiency virus (HIV) testing and counseling on self-reported condom and spermicide use and on corresponding HIV seroconversion and gonorrhea rates in urban Rwandan women., Design: Prospective cohort study with 2 years of follow-up, comparison of outcome variables before and after an intervention, and condom use measured in a control group that did not receive the intervention., Setting: Outpatient research clinic in Kigali, the capital of Rwanda., Participants: One thousand four hundred fifty-eight childbearing women, 32% of whom were infected with HIV, were enrolled in a prospective study in 1988, and followed at 3- to 6-month intervals for 2 years. Follow-up was available for 95% of subjects at year 1 and 92% at year 2., Interventions: An acquired immunodeficiency syndrome (AIDS) educational videotape, HIV testing and counseling, and free condoms and spermicide were provided to all participants and interested sexual partners., Main Outcome Measures: Self-report of compliance with condom-spermicide use and observed incidence of HIV and gonorrhea., Results: Only 7% of women reported ever trying condoms before the intervention, but 22% reported condom use with good compliance 1 year later. Women who were HIV-positive were more likely to adopt condom use than HIV-negative women (36% vs 16%; P < .05). Independent predictors of condom use, both in HIV-positive and in HIV-negative women, included HIV testing and counseling of the male partner, having a nonmonogamous relationship, and believing condoms were not dangerous. Human immunodeficiency virus seroconversion rates decreased significantly (from 4.1 to 1.8 per 100 person-years; P < .04) in women whose partners were tested and counseled. The prevalence of gonorrhea decreased substantially (13% to 6%; P < .05) among HIV-positive women, with the greatest reduction among condom users (16% to 4%; P < .05)., Conclusion: A confidential HIV testing and counseling program was associated with increased use of condoms and reduced rates of gonorrhea and HIV in urban Rwandan women. The lack of risk reduction in HIV-negative women whose partner's serostatus was unknown was of concern. Interventions that promote HIV testing and counseling for both members of a couple should be considered in other high-prevalence areas.

Objective: To determine whether HIV testing and counselling increased condom use and decreased heterosexual transmission of HIV in discordant couples., Design: Prospective study., Setting: Kigali, the capital of Rwanda., Subjects: Cohabiting couples with discordant HIV serology results., Main Outcome Measures: Condom use in the couple and HIV seroconversion in the negative partners., Results: 60 HIV discordant couples were identified, of whom 53 were followed for an average of 2.2 years. The proportion of discordant couples using condoms increased from 4% to 57% after one year of follow up. During follow up two of the 23 HIV negative men and six of the 30 HIV negative women seroconverted (seroconversion rates of 4 and 9 per 100 person years). The rate among women was less than half that estimated for similar women in discordant couples whose partners had not been serotested. Condom use was less common among those who seroconverted (100% v 5%, p = 0.01 in men; 67% v 25%, p = 0.14 in women)., Conclusions: Roughly one in seven cohabiting couples in Kigali have discordant HIV serological results. Confidential HIV serotesting with counselling caused a large increase in condom use and was associated with a lower rate of new HIV infections. HIV testing is a promising intervention for preventing the spread of HIV in African cities.

Objective: To better characterize the natural history of disease due to human immunodeficiency virus (HIV) infection in African women., Design: Prospective cohort study over a 2-year follow-up period., Participants: A total of 460 HIV-seropositive women and a comparison cohort of HIV-seronegative women recruited from prenatal and pediatric clinics in Kigali, Rwanda in 1988., Measurements: Clinical signs and symptoms of HIV disease, AIDS, and mortality., Main Results: Follow-up data at 2 years were available for 93% of women who were still alive. At enrollment, many seropositive women reported symptoms listed in the World Health Organization (WHO) clinical case definition of AIDS, but these were nonspecific and often improved over time. The 2-year mortality among HIV-infected women by Kaplan-Meier survival analysis was 7% (95% CI, 5% to 10%) overall, and 21% (CI, 8% to 34%) for the 40 women who fulfilled the WHO case definition of AIDS at entry. In comparison, the 2-year mortality in women not infected with HIV was only 0.3% (CI, 0% to 7%). Independent baseline predictors of mortality in seropositive women by Cox proportional hazards modeling were, in order of descending risk factor prevalence: a body mass index of 21 kg/m2 or less (relative hazard, 2.3; CI, 1.1 to 4.8), low income (relative hazard, 2.3; CI, 1.1 to 4.5), an erythrocyte sedimentation rate exceeding 60 mm/h (relative hazard, 4.9; CI, 2.2 to 10.9), chronic diarrhea (relative hazard, 2.6; CI, 1.1 to 5.7), a history of herpes zoster (relative hazard 5.3; CI, 2.5 to 11.4), and oral candida (relative hazard, 7.3; CI, 1.6 to 33.3). Human immunodeficiency virus disease was the cause of death in 38 of the 39 HIV-positive women who died, but only 25 met the WHO definition of AIDS before death., Conclusions: Human immunodeficiency virus disease now accounts for 90% of all deaths among child-bearing urban Rwandan women. Many symptomatic seropositive patients may show some clinical improvement and should not be denied routine medical care. Easily diagnosed signs and symptoms and inexpensive laboratory tests can be used in Africa to identify those patients with a particularly good or bad prognosis.

Sixteen human immunodeficiency virus type 1 (HIV-1)-seropositive children aged 5 to 12 years (nine girls and seven boys), born to HIV-1-infected mothers, were diagnosed between 1984 and 1987 in Kigali, Rwanda. They were compared with a group of age- and sex-matched HIV-1-seronegative children consecutively selected from the outpatient department. Two subjects were asymptomatic. Chronic cough was the most frequent symptom (seven of 16 patients). The most common signs were short stature (12 of 16 patients), low weight for age (seven of 16 patients), chronic parotitis (eight of 16 patients), persistent generalized lymphadenopathy (seven of 16 patients), and pulmonary tuberculosis (four of 16 patients). Lymphoid interstitial pneumonitis was diagnosed on radiologic grounds in five of 16 patients. Evidence of perivasculitis in the fundus was noted in three of 16 patients. Two children died during the study period (mean duration of follow-up, 40 months; range, 27 to 62 months); none of the other children had life-threatening infection or loss of developmental milestones. Immunologic assessment in the 16 children revealed high levels of IgG, decreased CD4+/CD8+ ratio, and skin test anergy. Endocrinologic investigations revealed normal thyroid function and normal basal human growth hormone levels but low basal insulinlike growth factor I levels (0.21 +/- 0.07 vs 0.44 +/- 0.20 U/mL for controls). In Kigali, perinatally HIV-1-infected children surviving beyond 5 years of age often present with moderate signs and symptoms, principally pulmonary involvement, chronic parotitis, and persistent generalized lymphadenopathy. Short stature is the major clinical manifestation in these patients and may be due, in part, to low growth hormone secretion rather than to malnutrition.

Objective: --To determine behavioral and demographic risk factors for human immunodeficiency virus (HIV) infection in central Africa., Design: --Cross-sectional survey., Setting: --Kigali, Rwanda., Participants: --A representative sample of 1458 childbearing women aged 19 to 37 years who were recruited from outpatient prenatal and pediatric clinics at the only community hospital in the city., Main Outcome Measure: --Antibodies to HIV assessed by enzyme immunoassay and confirmed by Western blot or indirect immunofluorescence., Results: --The HIV seroprevalence was 32% overall. Infection rates were higher in women who were single, in those in steady relationships that began after 1981, and in the 33% of women reporting more than one lifetime sexual partner. Women in legal marriages or monogamous partnerships had lower rates of infection, but even low-risk women had prevalences on the order of 20%. History of venereal disease in the past 5 years, although the strongest risk factor in a multiple logistic analysis (odds ratio, 2.7; 95% confidence interval, 2.0 to 3.7), was reported by only 30% of those infected. Having a male sexual partner who drank alcohol or who had higher income were significant risk factors for HIV infection in the multivariate analysis, but use of oral contraceptives and having an uncircumcised partner were not., Conclusions: --The epidemic of the acquired immunodeficiency syndrome in Rwanda has spread beyond high-risk groups to the general population of women without known risk factors. For most of these women, a steady male partner is the source of their HIV risk and therefore a vital target for intervention efforts.

We examined factors associated with perceived risk of AIDS, behavior change, and HIV infection in a representative sample of 1458 child-bearing urban women in Rwanda, central Africa. Although 68% of women reported only one lifetime partner, and the majority (87%) lived with a husband or steady partner, the prevalence of HIV antibodies was still high (32%). Before receiving their HIV test results, the women completed a questionnaire about AIDS knowledge, attitudes, and practices. Knowledge about HIV transmission was high, with 96-98% of women correctly identifying the three primary routes of infection. However, only 16% of women reported taking any action to avoid AIDS in the previous year, and most (11%) had done so merely by asking their male partners to change their behavior. Only 7% of women had ever tried condoms, and many (68%) thought they could be dangerous to use. Women who perceived themselves at risk of AIDS (57%) were more likely to report changing behavior; they were also more likely to be infected with HIV. Other factors associated with behavior change included having known someone with AIDS, having discussed AIDS with a male partner, and believing that condoms are not dangerous. Future interventions should enhance perception of risk, encourage male sexual partners to reduce risky behavior, and increase familiarity with condoms.

In 1986-1987 a consecutive sample of 3702 women presenting to prenatal and pediatric clinics at the only hospital in Kigali, Rwanda, was screened for human immunodeficiency virus (HIV) and malaria infection. The prevalence of HIV antibodies was 29%, and that of malaria parasites was 9%. HIV antibodies were more prevalent in women from the urban center than in those from the outskirts (31% vs. 20%, P less than .001), and malaria parasites showed the opposite prevalence pattern (8% vs. 15%, P less than .001); after stratifying by location, there was no association between HIV and the presence or degree of malaria parasitemia. HIV prevalence was 45% in women who had received a blood transfusion between 1980-1985 (before screening of donated blood began), and 28% among the great majority (94%) who had never been transfused. HIV prevalence was 44% in single mothers. 34% in women in common law unions, and 20% in those in legal marriages. These high rates of infection in the general population of Kigali highlight the need to develop effective programs for preventing further spread of sexually transmitted HIV.

We present the baseline results of a prospective cohort study on the perinatal transmission of HIV-1 in Kigali, Rwanda. HIV-1-antibody testing was offered to all women of urban origin delivering a live newborn at the maternity ward of the Centre Hospitalier de Kigali from November 1988 to June 1989; 218 newborns of 215 HIV-positive mothers were matched to 218 newborns of 216 HIV-negative mothers. The matching criteria were maternal age and parity. No differences in socioeconomic characteristics were observed between HIV-positive and HIV-negative women. HIV-positive mothers more frequently reported a history of at least one death of a previously born child (P less than 0.01) and a history of abortion (P less than 0.001). Most of the HIV-positive women were asymptomatic, but 72.4% of them had a CD4; CD8 ratio less than 1 versus 10.1% in the HIV-negative group (P less than 0.001). The frequency of signs and symptoms was not statistically different in the two groups, except for a history of herpes zoster or chronic cough, which was more frequent among HIV-positive women. The rates of prematurity, low birth weight, congenital malformations and neonatal mortality were comparable in the two groups. However, infants of HIV-positive mothers had a mean birth weight 130 g lower than the infants of HIV-negative mothers (P less than 0.01). The impact of maternal HIV-1 infection on the infant seems limited during the neonatal period.

Children with multiresistant Salmonella typhimurium (MRST) systemic infections, in total 246, were diagnosed during the study period. Of these, 220 had MRST without metastatic focal infections and 26 had metastatic focal infections (including 12 patients with meningitis). The median age of the children was 10 months. Diarrhoeal disease, measles and severe malnutrition were the most frequent causes of admission. Fever was found in 99% and diarrhoea in 72% of the patients, with respiratory symptoms in 72%. In 199 (81%) of the patients, the MRST infection was considered to be hospital-acquired. Of the 246 children, 159 were treated with cefotaxime. In this group, 16 of 152 patients died (10.5%). However, of the 87 children who did not receive cefotaxime, 64 died (74%). Relapses occurred in 4% of the patients with bacteraemia treated with cefotaxime. Our study confirms the high efficiency of cefotaxime in treating severe systemic infections with MRST.

In Rwanda, both HIV infection and bacteraemia represent major health problems among paediatric populations. We carried out of prospective study of determine if bacteraemia is a marker of HIV infection among ambulatory and hospitalized Rwandese children. All children presenting at the Department of Paediatrics of the Center Hospitalier de Kigali who had their blood cultured during a two-month period were eligible for the study. One hundred and thirty-five children were included in the study. A pathogen was isolated from 36 children (26.7%): S. typhimurium (10 cases), S. enteritidis (6), S. typhi (4), Str. pneumoniae (9). H. influenzae (6) and S. aureus (1). No association was found between bacteraemia and HIV seropositivity when all the children were considered. However, among patients less than 2 years old, bacteraemic subjects were more frequently (p less than 0.05) HIV seropositive (44%) than those with negative blood cultures (19%). Our study shows that in young children in Central Africa, the presence of bacteraemia may be an important marker of HIV seropositivity.

During a 21-month period, we observed an outbreak of severe systemic infections due to multiresistant Salmonella typhimurium among 66 children in the in-patient Department of Paediatrics of Kigali, Rwanda. These infections were more likely to occur in subjects who had stayed for a long time in the hospital for severe illness and/or malnutrition. The children usually presented first with mild to moderate diarrhoea and fever. Later, sever pulmonary involvement was often noted (rales: 58%; respiratory distress: 42%). Moreover, there were four cases of abscess, three arthritis and one meningitis. Of the 66 children, 48 were treated with cefotaxime. The fatality-rate among this group was 10.4%. The fatality-rate among the 18 other untreated patients was 77.9%, suggesting a high efficiency of cefotaxime against these strains of multiresistant Salm. typhimurium.

Nineteen out of 139 children with severe systemic disease due to multiresistant Salmonella typhimurium observed during a 34-month period in an in-patient department in Rwanda had focal metastatic infections. More than 80% of the invasive Salmonella infections were acquired in the hospital. Focal metastatic infections occurred after longer hospital stays than bacteremia (29.1 +/- 17.4 days as against 13.5 +/- 9.0 days, p less than 0.01) and were diagnosed more time after the first sign of infection (3.28 +/- 1.41 days as against 1.86 +/- 1.10 days, p less than 0.01). Bacteremia was documented in 13 of the 17 children with focal infection from whom blood cultures were obtained. Seven of 12 had positive stool cultures. The sites of metastatic focal infection were meninges (7 cases), soft tissue (5 cases), joint or bone (4 cases), pleura (2 cases), eye (1 case). The clinical course of meningitis was fulminant and 6/7 patients died before receiving adequate antimicrobial therapy. One child with meningitis and 9 patients with focal infections at other sites were treated with cefotaxime and were cured or improved.

From February to June 1986, 150 heterosexual couples with at least one HIV-seropositive member were recruited in the 'Centre Hospitalier de Kigali', Kigali, Rwanda. Of the 138 HIV-seropositive couples in whom both members were tested, 124 (90%) were sexual unions between two antibody-positive partners, illustrating the high efficiency of the heterosexual transmission of HIV. A comparison of these 124 couples with 150 HIV-seronegative couples showed that, in the husbands, seropositivity is significantly associated with sexual contacts with prostitutes and history of sexually transmitted disease (STD) within the past 2 years. Seropositive wives were less likely to be in their first marriage and reported more episodes of STD than seronegative ones. Seropositive couples were similar to seronegatives in their history of blood transfusion, male circumcision and overall use of contraception but not in their use of oral contraceptive pills, which was more frequent; they were more often in polygamous unions. Discriminant analysis showed that STD, sexual contacts with prostitutes and the number of previous unions are the most powerful independent variables associated with the seropositivity of the couples. Most of the risk factors for the couples were the risk factors for the husband, suggesting that in most cases the husband acquired the HIV infection and passed it to his wife.

To examine the frequency of community acquired bacteraemia in children in Kigali, Rwanda, blood cultures were obtained from 900 consecutive febrile children (T degrees greater than or equal to 39 degrees C) seen at an outpatient clinic over the course of a year. A pathogen was isolated from 112 children (12.4%): Salmonella typhi from 47, S enteritidis from 23, S typhimurium from 13, Streptococcus pneumoniae from 14, Staphylococcus aureus from 9, and Haemophilus influenzae from 3. Salmonella species represented 74% of the isolates. The children with S typhi bacteraemia were older (mean age 75 months) than those with bacteraemia due to other organisms. Controls consisted of febrile, nonbacteraemic children without (group I) or with (group II) Plasmodium falciparum parasitaemia. Bacteraemic children were older and presented more frequently with diarrhoea, vomiting, and dehydration, but less frequently with convulsions than controls. The rate of hospital admission was higher among bacteraemic children (61%) than among group I (39%) or group II (46%) controls. The case-fatality rate was similar in the three groups (9.3% versus 2.9% and 7.3%). Community-acquired bacteraemia in Rwandese children is common and is mainly caused by Salmonella species.