Malaria in Injection Drug Abusers in Vietnam

MAJOR ARTICLE Malaria in Injection Drug Abusers in Vietnam T. T. H. Chau, 2 N. T. H. Mai, 2 N. H. Phu, 2 C. Luxemburger, 1,3 L. V. Chuong, 2 P. P. Loc, 2 T. T. M. Trang, 2 H. Vinh, 2 B. M. Cuong, 2 D. J. Waller, 1,3 D. X. Sinh, 2 N. P. J. Day, 1,3 T. T. Hien, 2 and N. J. White 1,3 1 Wellcome Trust Clinical Research Unit, 2 Centre for Tropical Diseases, Ho Chi Minh City, Vietnam; and 3 Centre for Tropical Medicine, Nuffield Department of Clinical Medicine, John Radcliffe Hospital, Oxford, United Kingdom A prospective case-control study was conducted in a referral hospital in Ho Chi Minh City, Vietnam, to compare the clinical and laboratory features and outcome of severe falciparum malaria in injection drug abusers (IDAs) with those of patients who had acquired malaria by mosquito bite. From 1991 to 1996, 70 IDAs were admitted to the hospital, of whom at least 32 had acquired malaria by needle sharing. Although IDAs were more likely than control patients with severe malaria to be malnourished and to have coincident hepatitis B, hepatitis C, and human immunodeficiency virus infections, the overall rates of mortality, complications, and recovery were similar in the 2 groups. The route of malaria acquisition did not affect the outcome of severe malaria. The management of severe malaria in IDAs is similar to that for other patients. In 1928, the Professor of Clinical Medicine at the Egyptian University of Cairo reported an outbreak of malignant malaria in young men who were addicted to heroin [1]. All of these men were injecting the drug. Because none of the injection drug abusers (IDAs) originated from the area of Egypt where malaria was endemic at that time, the professor concluded that malaria had been transmitted by the communal use of unsterilized syringes containing residual infected blood. Since 1928, there have been several case reports [2 9] and small series describing both falciparum and vivax malaria in IDAs, as well as more-extensive investigations of transfusion malaria [10] and detailed descriptions of blood-passaged and drug-attenuated infections in malaria therapy [11]. Genetic, behavioral, epidemiological, and parasitological factors determine whether an individual infection with Plasmodium falciparum will progress to vital organ dysfunction and death. Severe malaria is an un- Received 8 August 2001; revised 7 January 2002; electronically published 23 April 2002. Financial support: The Wellcome Trust of Great Britain. Reprints or correspondence: Dr. Tran Thi Hong Chau, Centre for Tropical Diseases, 190 Ben Ham Tu Quan 5, Ho Chi Minh City, Vietnam (bschautung@hcm.vnn.vn); or Prof. Nick White, same address (jeremyjf@hcm.vnn.vn). Clinical Infectious Diseases 2002; 34:1317 22 2002 by the Infectious Diseases Society of America. All rights reserved. 1058-4838/2002/3410-0004$03.00 usual outcome, but once it has developed, it is associated with a mortality rate of 20% [12]. The outcome is determined by the degree of vital organ dysfunction, the delay in administration of antimalarial therapy, and, probably, the choice of antimalarial drug. Serious infections in IDAs may have a high mortality rate. There have been no controlled, systematic studies to investigate whether the presentation and outcome of severe malaria in IDAs differ from the presentation and outcome in patients infected through the bite of an anopheline mosquito. We report a prospective case-control study of malaria among IDAs admitted to the Center for Tropical Diseases, Cho Quan Hospital, Ho Chi Minh City, Vietnam. METHODS Background. At the Center for Tropical Diseases, Cho Quan Hospital, Ho Chi Minh City, Vietnam, patients from the city and surrounding provinces are treated. This hospital is the referral center for malaria in the southern region of Vietnam. There has been no transmission of malaria within urban central Ho Chi Minh City for 130 years. Most patients admitted to this hospital acquired their infections in regions 50 100 km away where malaria is endemic. Patients. This prospective case-control study was conducted in a clinical research ward at the Center for Malaria in Injection Drug Abusers CID 2002:34 (15 May) 1317

Tropical Diseases in Ho Chi Minh City, Vietnam, from 1991 through 1996. All patients admitted to the ward with severe falciparum malaria who acknowledged that they were IDAs or who were reported by their relatives to be IDAs (the IDA group) were included in the study if they or their relatives gave informed consent. This study was approved by the Ethical and Scientific Committee of the Center for Tropical Diseases. Severe malaria was defined according to a modified version of the World Health Organization criteria [13], as described elsewhere [14]. The criteria were the presence of asexual forms of P. falciparum on a blood smear and 1 of the following characteristics: Glasgow Coma score of!11 (cerebral malaria); hyperparasitemia (110% parasitized RBCs); anemia (hematocrit,!20%) or jaundice (serum bilirubin level, 12.5 mg/dl [150 mm]), with a parasite count of 1100,000 parasites/ml; renal impairment (urine output of!400 ml per 24 h and a serum creatinine level of 13 mg/dl [250 mm]); hypoglycemia (blood glucose,!40 mg/dl [2.2 mm]); and/or shock (systolic blood pressure,!80 mm Hg with cool extremities). Each IDA was matched with the next patient to be admitted to the ward who had severe malaria, who denied being a drug abuser, and who had no clinical signs suggestive of addiction (control group). Clinical procedures and treatment. On admission, a detailed questionnaire was completed regarding the patient s history of malaria, past hospitalizations, travel history, previous antimalarial treatment, drug use, alcohol intake, and smoking habits. This information was obtained from the patient or relative; if it was obtained from the latter, the information was supplemented by questioning the patient after recovery of consciousness. Information on use of antimalarial drugs before admission was also obtained from the transfer letters accompanying patients referred from other hospitals. The same physician (T.T.H.C.) assessed the socioeconomic and nutritional statuses for all IDAs and control patients. A full clinical examination was performed, and venous blood samples were obtained for quantitative parasite count, full blood count, routine biochemistry, determination of plasma glucose and lactate levels, and blood culture. Blood samples were also obtained (with permission) for serological testing for HIV infection and infection with hepatitis A, B, C, and/or E viruses. A lumbar puncture was performed on all unconscious patients. Patients were treated with intravenously administered quinine (loading dose, 20 mg/kg of dihydrochloride salt, followed by 10 mg/kg q8h) or intramuscularly administered artemether (4 mg/kg initially followed by 2 mg/kg q8h; Kunming Pharmaceutical Company). This was part of a large, double-blind, randomized trial comparing quinine and artemether for the treatment of severe malaria, which has been reported elsewhere [13]. Complications were treated according to standard recommendations. Clinical monitoring was performed every 6 h. Time to resolution of fever, time to recovery of consciousness, and occurrence of complications not present at admission were carefully recorded, as was the likely cause of death. Statistical analysis. Statistical analysis was performed with SPSS software, version 6.0 (SPSS). Proportions were compared by calculating x 2 with Yates correction or by Fisher s exact test. Student s t-test and analysis of variance were used to compare normally distributed continuous data. Data that not conform to a normal distribution were compared by use of the Mann- Whitney U test and Kruskall-Wallis analysis of variance. Because the main demographic characteristics of the 2 groups differed, analyses were stratified for these factors when comparing admission clinical features and laboratory findings. Treatment was also taken in account in the analysis of outcomes. Logistic regression was used to identify factors associated independently with clinical findings and outcomes. RESULTS From January 1991 through November 1996, 70 IDAs with severe falciparum malaria were admitted to the study. Most of those for whom the information was available (30 [93%] of 45 IDAs) had most recently injected drugs during the month before admission (median time elapsed since last injection, 5 days; range, 1 120 days). The main drug of abuse was heroin. The IDAs acknowledged a median of 14 drug injections per week (range, 3 28). Thirty-two IDAs (46%) were living in an area of Ho Chi Minh City where malaria is not endemic and denied any travel outside the city. They were considered to have definitely acquired malaria infection via injection with contaminated needles. The remaining 37 IDAs (there was no information on 1 patient) lived or had travelled recently to areas where malaria is endemic and, therefore, could have been infected either by the intravenous route or by mosquito bite. Demographic data, medical history, and serological status for HIV and hepatitis A, B, C, and E virus infections are summarized in table 1. IDAs were more often male, were more likely to drink alcohol and smoke tobacco heavily, were more likely to be residents of Ho Chi Minh City, and had been hospitalized previously with malaria more frequently than had the control group. More control patients than IDAs had received antimalarial drugs before admission into this study. The distribution of antimalarial drug use in both groups was similar: of 56 patients for whom there was available data, 38 (68%) had taken quinine, 14 (25%) had taken artemisinin derivatives, and 4 (4%) had taken chloroquine or sulfadoxine-pyrimethamine. Most (90%) of the IDAs for whom serologic testing for hepatitis was performed were hepatitis C virus carriers; they were also more often seropositive for hepatitis B and E virus than were control patients. Previous hepatitis A infection was equally common in both groups. Nearly one-half (45%) of the IDAs were seropositive for HIV, whereas all control patients 1318 CID 2002:34 (15 May) Chau et al.

Table 1. Admission characteristics of injection drug abusers (IDAs) and control patients with severe falciparum malaria. Characteristic IDAs Control patients P Age, median years (range) 30 (17 60) 30 (15 79).87 Male sex 68/70 (97) 59/70 (84).020 Residence in Ho Chi Min City a 62/69 (90) 22/70 (31)!.001 Homelessness 7/69 (10) 0/70 (0).019 Socioeconomic status Very poor 16/70 (23) 19/70 (27) Poor 33/70 (47) 47/70 (67) Average or rich 21/70 (30) 4/70 (6).001 Travel during previous month 20/69 (29) 34/70 (49).028 Regular tobacco smoking 59/63 (94) 46/69 (67)!.001 Regular alcohol use 53/63 (84) 39/69 (57).001 Medical history Previous hospitalizations 36/63 (57) 23/68 (34).012 Previous episodes of malaria 36/62 (58) 26/68 (38).037 Previous severe or cerebral malaria 8/62 (13) 4/69 (6).27 Receipt of antimalarial treatment before present hospital admission 14/62 (23) 44/66 (67)!.001 Positive serologic test result, by virus HIV 22/49 (45) 0/61 (0)!.001 Hepatitis A 36/47 (77) 46/61 (75) 1.000 Hepatitis B 17/61 (28) 7/60 (12).045 Hepatitis C 45/47 (96) 26/61 (43)!.001 Hepatitis E 17/47 (36) 4/61 (7)!.001 NOTE. Data are no. of patients with characteristic/no. with available data (%), unless otherwise indicated. a There is no malaria transmission within the city. were HIV negative. Apart from their residence, IDAs who were infected via injections did not differ from other patients in the IDA group with regard to demographic data, medical history, and results of serologic testing for hepatitis viruses and HIV (data not shown). Admission Findings Clinical features. Patients from both groups were admitted to the hospital after a similar duration of illness (median for IDAs, 5 days [range, 1 30 days]; median for control patients, 5 days [range, 1 14 days]). The presenting symptoms and findings at physical examination were also similar in both groups (table 1). Palpable spleen was observed more frequently among IDAs (36 [54%] of 70) than it was among control patients (13 [19%] of 70; P!.001). This clinical finding was also associated with residence in Ho Chi Minh City, a history of hospitalization and/or malaria episodes, and HIV or hepatitis C virus seropositivity. In a multivariate analysis, a past history of malaria (OR, 1.5; 95% CI, 1.2 2.7) and drug addiction (OR, 1.2; 95% CI, 1.1 1.7) remained independently associated with the presence of an enlarged spleen. Malnutrition was also more common among IDAs (43 [61%] of 70 patients) than it was among control patients (5 [7%] of 70; P!.001) and was associated with residence in Ho Chi Minh City, history of hospitalization and/or malaria episodes, alcohol consumption, and hepatitis B, C, or E virus seropositivity. Only drug addiction and hepatitis C virus seropositivity remained independently associated with malnutrition if all these factors were included in a multivariate model. In the IDA group, the development of cerebral malaria was more common among those patients who definitely acquired falciparum malaria by injection drug use than it was among other IDAs ( P p.014 ). In all other respects, the patients who acquired malaria by injection drug use rather than via a mosquito bite were similar with respect to admission characteristics, laboratory findings, and outcome (data not shown). Laboratory findings. As expected, the use of antimalarial drugs before admission in the study reduced parasite counts: the geometric mean parasite count was 13,695 parasites/ml Malaria in Injection Drug Abusers CID 2002:34 (15 May) 1319

Table 2. Laboratory findings at admission to the hospital for injection drug abusers (IDAs) and control patients with severe falciparum malaria. Laboratory finding, group IDAs Control patients P Parasite count, geometric mean parasites/ml (range) All patients 78,050 (40 2,443,924) 22,622 (20 974,656).015 Pretreated patients 148,145 (40 708,886) 20,598 (20 943,381).322 Non-pretreated patients 172,099 (100 2,443,924) 136,276 (40 974,656).184 Hematocrit, mean % SD 29.8 9.1 30.4 7.7.63 WBC count, median cells 10 9 /L (range) 9.5 (2.8 44.0) 9.8 (3.0 38.5).97 Platelet count, mean cells 10 9 /L (range) 88 (20 260) 50 (18 272)!.001 Serum albumin level, mean g/dl SD 3.5 0.8 3.4 0.9.97 Blood urea nitrogen level, median mg/dl (range) 26 (12 190) 38.5 (1 190).21 Serum creatinine level, median mg/dl (range) 1.7 (0.7 10) 2.4 (0.9 8.7).007 Alanine aminotransferase level, median m/l (range) 110 (20 489) 156 (32 1150).023 Malnourished patients 100 (20 377) 156 (50 510).42 Nonmalnourished patients 147 (24 289) 156 (32 1150).39 Aspartate aminotransferase, median m/l (range) 150 (40 470) 185 (30 1040).023 Malnourished patients 150 (40 470) 149 (97 520).67 Nonmalnourished patients 155 (60 420) 190 (30 1040).22 Total bilirubin level, median mg/dl (range) 2.4 (0.4 19.0) 4.6 (0.6 21.0).001 Malnourished patients 2.2 (0.4 13.6) 2.8 (1.3 12.4).36 Nonmalnourished patients 3.4 (1.0 19.0) 4.6 (0.6 21.0).062 Blood glucose level, median mg/dl (range) 97.5 (10 263) 92.5 (23 400).96 (range, 20 943,381 parasites/ml) in patients already treated and 100,208 parasites/ml (range, 40 2,443,924 parasites/ml) in those who had not received antimalarial drugs ( P!.001). After adjusting for this factor, the parasite counts at admission were similar in the 2 groups (table 2). IDAs and control patients had different liver function test results, but these were related to nutritional status. When we controlled for nutritional status, bilirubin, aspartate aminotransferase, and alanine aminotransferase levels were similar in both groups. Platelet counts were lower in the control group than they were in the IDA group, and this difference remained significant when the patients were compared by sex, the presence of palpable spleen, and nutritional status. At admission to the study, signs of severe malaria (as defined previously) were present in similar proportions in the 2 groups (table 3). Outcome Sixty patients (86%) were treated with quinine and 10 (14%) were treated with artemether in each group. The overall mortality rate was 20% (28 of 140 patients) and was similar in the 2 groups. The IDAs died later after admission than did control patients (median time from admission to death for IDAs, 50 h [range, 4 502 h]; median for control patients, 16 h [range, 4 254 h]). All but 1 of the factors (hyperparasitemia) identified in an univariate analysis ( P!.10 ) as a potential risk factor for mortality remained associated independently with a fatal outcome in a multivariate analysis. These risk factors were as follows: an elevated serum creatinine level at admission (OR, 1.07; 95% CI, 1.02 1.26), hypoglycemia at admission (OR, 1.10; 95% CI, 1.01 2.33), malnutrition (OR, 1.25; 95% CI, 1.08 2.00), and period of admission in the study (1991 1993 and 1994 1996; OR, 1.33; 95% CI, 1.09 2.62). More IDAs than control patients developed coma after admission to the hospital, but the time to regain consciousness was similar in both groups. The incidence of other malaria-related complications was the same in each group (table 4). IDAs stayed a shorter time in the hospital than did control patients, mainly because IDAs tended to discharge themselves (12 [22%] of 55 survivors). Relationship of Outcome to HIV Status Nearly one-half (45%) of the IDAs were seropositive for HIV infection, whereas none of the control patients were ( P!.001). However, HIV-positive patients did not differ from HIV-negative patients with regard to patient history, symptoms at admission, clinical findings on examination, abnormal laboratory values, and outcome. DISCUSSION Blood transfusion and needle sharing among drug addicts readily transmits malaria. In areas where malaria is endemic, the 1320 CID 2002:34 (15 May) Chau et al.

Table 3. Signs of severe malaria at admission to the hospital for injection drug abusers (IDAs) and control patients with severe falciparum malaria. No. of patients with sign/ no. with available data (%) Sign (definition) IDA group Control group Cerebral malaria (Glasgow Coma score,!11) 23/70 (33) 32/70 (46).17 Hyperparasitemia (110% parasitized RBCs) 10/70 (14) 12/70 (17).81 Anemia (hematocrit,!20%) and parasitemia (1100,000 parasites/ml) 5/70 (7) 1/70 (1).21 Jaundice (serum bilirubin level, 12.5 mg/dl) and parasitemia (1100,000 parasites/ml) 21/64 (33) 21/68 (31).96 Renal impairment (urine output,!400 ml/24 h; serum creatinine level, 13 mg/dl) 11/70 (16) 21/70 (30).07 Hypoglycemia (blood glucose level,!40 mg/dl) 6/66 (9) 2/68 (3).16 Shock (systolic blood pressure,!80 mm Hg, and cool extremities) 6/70 (9) 1/70 (1).12 Positive blood culture result 1/59 (2) 5/67 (8).14 majority of the population may have malaria parasites in their blood, and in a significant proportion of these patients, parasites are detectable by microscopic examination (i.e., densities of 50 ml). As for transmissible viral infections, small parenteral inocula only are sufficient to transmit large numbers of organisms [14]. For example, an inoculation of 0.05 ml of blood from a donor with parasitemia (i.e., 5000 parasites/ ml) via an infected needle would yield approximately the same parasite numbers as the merogony of 24 hepatic schizonts (i.e., 24 successful sporozoites, assuming 20,000 merozoites per hepatic schizont and 50% invasion efficiency). This represents a Table 4. Survival among and development of malaria-related complications in injection drug abusers (IDAs) and control patients with severe falciparum malaria. Outcome or complication large inoculum and would exceed the number of parasites usually generated by a single mosquito bite in the majority of cases (the sporozoite inoculum has been estimated to be 8 15 sporozoites [15]). Obviously, if the donor has a higher parasite count, or if a larger volume of blood is transfused, then the inoculum will be even greater. The consequences of a larger inoculum to a nonimmune host would be a shorter prepatent period and less time for specific host defences to be mobilized. These reasons, in addition to the generally poor physical condition of IDAs in developing countries and their common comorbidity, have all given rise to an impression that IDAs tend IDAs Control patients P Death 15 (21) 13 (19).68 Time to death, h 50 (4 502) 16 (5 254).022 Cerebral malaria a 29 (41) 17 (24).048 Time to recovery of consciousness, h 44 (16 264) 72 (6 270).55 Acute renal failure 17 (24) 26 (37).14 Hypoglycemia 8 (11) 8 (11) 1.000 Convulsions 12 (17) 5 (7).12 Bleeding 17 (24) 20 (29).70 Infection 24 (34) 20 (29).44 Dialysis 6 (9) 11 (16).30 Time to clearance of fever, h 72 (24 264) 87 (4 486).37 Duration of hospitalization, days 8 (4 500 12 (6 44)!.001 P NOTE. Data are no. (%) of patients or median value (range). a Glasgow Coma score,!11. Malaria in Injection Drug Abusers CID 2002:34 (15 May) 1321

to have a poor outcome when they develop severe malaria. Both hepatitis B virus infection and HIV infection influence the severity of falciparum malaria, although hepatitis B virus appears to increase the risk of developing severe disease rather than altering the course of disease once it develops [16, 17]. For HIV, there is evidence for some loss of malaria immunity with advanced disease and AIDS-related immunosuppression, although the interaction is not marked. In this large series, despite the high prevalence of infection with hepatitis B virus, hepatitis C virus, and HIV, the clinical features and outcomes of IDAs did not significantly differ from those of patients who acquired malaria via mosquito bites and who did not have these viral coinfections. In this series, some of the IDAs may indeed have acquired malaria via a mosquito bite, because approximately one-half of these patients had visited areas where malaria is endemic in the recent past. However, comparison of these patients with those who had not left the urban area (where there is no transmission), and, therefore, who must have acquired malaria via needle sharing, did not suggest any significant differences, except with regard to cerebral malaria, which was more common in those IDAs who definitely acquired malaria via needle transmission. The difference between IDAs who died and other patients who died of severe malaria with regard to the time to death is unexplained, although the numbers are small (15 vs. 13, respectively) and may be a chance finding. It is of interest that, despite their urban residence and readier access to hospitals, the duration of illness was similar for IDAs and other patients. This suggests that IDAs were generally reluctant to present to the hospital, and, therefore, these patients in the community may have died of fulminant malaria and have been underrepresented. Two factors may attenuate severe malaria in IDAs. The first is malnutrition, which is considered to protect from severe malaria [14], although this theory is controversial. The mechanism of apparent protection is unclear. Iron deficiency has been suggested as a contributory factor. IDAs are continually inoculating infectious and noninfectious particulate material into the circulation. Thus, host defence and, in particular, splenic function may be augmented. Thus, the patients defences against malaria may be primed and the large inoculum and disadvantage of a short incubation period may be balanced by preexisting activated host defences. Whatever the mechanism, the management of malaria in IDAs is similar to that for patients with naturally acquired disease, although venous access may be difficult and the risks of spreading HIV and hepatitis virus to nurses and physicians are higher. The route of malaria acquisition did not affect the outcome of patients with severe malaria. The management of severe malaria in IDAs is similar to that for other patients. Drug withdrawal, behavior problems in the wards, and the high rate of self-discharge from the hospital may further complicate clinical management. Acknowledgments We are grateful to the Director and staff of the Center for Tropical Diseases, Cho Quan Hospital, Ho Chi Minh City, Vietnam. We are grateful to Dr. Jeremy Farrar for his critical reading of this manuscript. References 1. Biggam AG. Malignant malaria associated with the administration of heroin intravenously. Trans R Soc Trop Med Hyg 1929; 23:147 53. 2. Dover AS. Malaria in a heroin user. JAMA 1971; 215:1987. 3. Rosenblatt JE, Marsh VH. Induced malaria in narcotic addicts. Lancet 1971; 2:189 90. 4. Lyman DO, Boese RJ, Shearer LA. Malaria among heroin users. Health Serv Rep 1972; 87:545 9. 5. Shaw PK, Brodsky RE, Schultz MG. Malaria surveillance in the United States: 1974. J Infect Dis 1976; 133:95 101. 6. Bastin R, Charmot G, Brumpt I, Sansonetti P, Solal-Celigny P. Transmission par seringue de paludisme a plasmodium falciparum chez un heroinomane. Nouv Presse Med 1979; 8:699 700. 7. Brown JD, Khoa NQ. Fatal falciparum malaria among narcotic injectors. Am J Trop Med Hyg 1975; 24:729 33. 8. Baker JE, Crawford GP. Malaria: a new facet of heroin addiction in Australia. Med J Aust 1978; 2:427 8. 9. Gonzales Garcia JJ, Arnalich F, Pena JM, et al. An outbreak of Plasmodium vivax malaria among heroin users in Spain. Trans R Soc Trop Med Hyg 1986; 80:549 52. 10. Dodd RY. Transmission of parasites and bacteria by blood components. Vox Sang 2000; 78(Suppl 2):239 42. 11. Kitchen SF. Symptomatalogy: general considerations and falciparum malaria. In: Boyd MF, ed. Malariology. Philadelphia: WB Saunders, 1949:996 1017. 12. World Health Organization, Division of Control of Tropical Diseases. Severe and complicated malaria. Trans R Soc Trop Med Hyg 1990; 84(Suppl 2):1 65. 13. Hien TT, Day NPJ, Phu NH, et al. A controlled trial of artemether or quinine in Vietnamese adults with severe falciparum malaria. N Engl J Med 1996; 335:76 83. 14. White NJ. Malaria. In: Cook GC, ed. Manson s tropical diseases. London: WB Saunders, 1996:1087 164. 15. Ponnudurai T, Lensen AHW, van-gemart GJA, Bolmer MG, Meuwissen JH. Feeding behaviour and sporozoite ejection by infected Anopheles stephensi. Trans R Soc Trop Med Hyg 1991; 85:175 80. 16. Whitworth J, Morgan D, Quigly M, et al. Effect of HIV-1 and increasing immunosuppression on malaria parasitaemia and clinical episodes in adults in rural Uganda: a cohort study. Lancet 2000; 356:1051 6. 17. Thursz MR, Kwiatkowski D, Torok ME, et al. Association of hepatitis B surface antigen carriage with severe malaria in Gambian children. Nat Med 1995; 1:374 5. 1322 CID 2002:34 (15 May) Chau et al.