Malaria and HIV, with tuberculosis, are considered the most important public health problems facing Africa. While Plasmodium falciparum is recognised as an important cause of fever in HIV-infected individuals [1,2], there is still substantial uncertainty about the extent to which the two pathogens interact even after 20 years of the AIDS epidemic. Initial descriptions suggested no more than clinical coexistence [3–6]. More recently, studies in regions of stable malaria transmission have noted higher parasite rates in HIV-1-infected multigravidae [7–9], a possible clinically important negative impact on perinatal outcome  and a trend for increasing rates of malarial fever with advancing HIV-associated immunosuppression [11,12]. Two small case series of hospital inpatients [13,14]. have suggested an increase in instances of cerebral malaria in HIV-infected adults, but case detection and HIV surveillance were incomplete in both studies. Increased severity of disease has not been a feature of cohort studies from areas of stable transmission, even when rates of malarial fever have increased; however, relatively few cases were reported [11,12].
The identification of any interaction is complicated by the nature of the immune response to malaria. In endemic areas, repeated childhood exposure and disease episodes generate partial immunity, which ameliorates infection but does not fully prevent parasitaemia or clinical disease. In a sick patient with fever and who is also parasitaemic, the malaria infection can either be causal or coincidental. It is not always easy to be certain, leading to potential ascertainment bias from imprecise case definitions. In contrast, in non-immune individuals with little exposure to P. falciparum, parasitaemia is diagnostic as infection almost invariably leads to symptomatic disease. Therefore, a prospective study of malaria in an area of unstable transmission and high HIV prevalence would better permit testing of the hypothesis that HIV coinfection is associated with severe malaria.
Since 1995, we have been studying the impact of HIV on a rural district in northern KwaZulu-Natal, South Africa . In Hlabisa, malaria has been sporadic since the intensive control efforts implemented after the 1932 Natal epidemic . However, a marked increase in rates of falciparum malaria was recognised during early 1999. Cases presented over 5 months until transmission by the principal vectors (Anopheles funestus and A. arabiensis) was disrupted by cooler weather. In early December 1999, a further major increase in falciparum malaria was noted. A process of systematic surveillance of patients with fever was instituted in early 2000 to identify cases of malaria, record outcome and link this to HIV infection status. At the time of the study, HIV seroprevalence amongst antenatal clinic attenders in KwaZulu-Natal was 32.5% .
The Hlabisa district hospital and four community-based clinics serving the areas with the highest rates of malaria in the previous year were chosen as study sites. All adult patients (defined as 14 years or older) presenting between January and May 2000 with an acute febrile illness and appropriate history (acute onset fever, headache, diarrhoea, confusion, fits, jaundice and/or impaired level of conciousness) were screened for falciparum malaria with a rapid malaria antigen test (Kat-Quick; Katmedical, Pretoria, South Africa). Individuals with a positive test were enrolled in the study following informed consent given by the patient or their guardian. A standardized form recording basic demographic and clinical data was completed. If participants wanted to know their HIV status, they were counselled and tested separately. Information was sought on duration of illness, pregnancy status, previous malaria and treatment prior to presentation. Further specimens were taken for processing at a later stage to ascertain HIV status (a blood spot collected on filter paper) and to confirm the presence of malaria (a thick blood smear). HIV testing was performed anonymously. Results could only be linked to clinical data by study number, not by name.
Patients were treated according to disease severity using national guidelines . Those who were relatively well, ambulatory and not pregnant or vomiting received outpatient therapy with sulfadoxine–pyrimethamine (SP). Those who were less well, pregnant or did not respond to oral therapy could be started on quinine and were admitted to hospital. Additional clinical data were recorded on disease severity, further treatment and outcome. Patients with severe/complicated disease or vomiting, and all pregnant women, were admitted to a high-dependency ward for intravenous quinine. Less-ill patients received oral quinine or SP. The study was approved by the ethics committees of the University of Natal, South Africa and the Liverpool School of Tropical Medicine, UK.
HIV testing and microscopy of blood smears was undertaken at the Liverpool School of Tropical Medicine blind from clinical data. Blood spots were tested for HIV using a commercial kit according to the manufacturer's instructions (Wellcozyme GAC-ELISA, Abbott, Abbott Park, Illinois, USA) following elution of blood from filter paper (0.34 mm chromotography, Whatman, Maidstone, UK) using a standardized method. Samples producing indeterminate results were excluded from analysis. Blood slides were read in duplicate by experienced microscopists, who were subjected to a quality assurance process. Final reporting of discordant results was based on the decision of a third independent microscopist. Parasite density was measured semiquantitatively. The number of parasites were counted against 200 white blood cells.
The primary case definition of malaria was a positive rapid antigen test confirmed by microscopy. Asexual-stage trophozoites, malarial pigment or gametocytes were taken to indicate recent infection. Severe disease was defined according to World Health Organization criteria by the presence of one or more of the following features: impaired renal function (urea > 15 mmol/l or creatinine > 190 μmol/l), recurrent fits, unrousable coma (Glasgow Coma Scale < 9), anaemia (haemaglobin < 5 g/dl), acute confusion or drowsiness, bleeding disorder, pulmonary oedema, acidosis (serum bicarbonate < 15 mmol/l) or aspiration pneumonia .
Clinical and laboratory data were entered in duplicate into a Microsoft Access database. Statistical analysis was performed using STATA version 5 (College station, Texas, USA). The primary outcome measures were the presence of severe/complicated disease and death from malaria. Variables showing an association with the primary outcome measures on univariate analysis were incorporated into a logistic regression model to estimate their adjusted effect. To achieve the most appropriate model, a stepwise procedure was used. Variables were sequentially added to the model, starting with those with the strongest univariate association and removed if non-significant at the 5% level.
A total of 1378 acutely febrile adult patients with positive malaria antigen tests and appropriate clinical history were identified. Of these, 195 individuals (14%) were excluded because of incomplete clinical data and 74 individuals (5%) were excluded because of incomplete sets of blood samples, leaving 1109 study subjects (Fig. 1). Despite staff training, the quality of blood slides prepared by clinic staff at enrolment was poor. The majority were too thick and required prolonged removal of haemoglobin and staining. This appeared to reduce considerably the sensitivity of subsequent microscopy : of the 1109 malaria antigen-positive adults, only 613 (55%) had slide evidence of malaria. However, HIV prevalence was similar in both groups and 29.9% of all enrolled adults were HIV positive (Table 1). Where speciation was possible, P. falciparum was the Plasmodium sp. identified in all patients.
A greater number of elderly adults (age > 60 years) were found in the HIV-uninfected group with slide-confirmed malaria than in the HIV-positive group, otherwise the age distribution was similar. Associations were found between HIV and presentation with either vomiting or acute confusion/drowsiness. Although HIV-infected individuals were more likely to be admitted to hospital, in regression analysis this effect was explained by the presence of vomiting and acute confusion/drowsiness. Neither semiquantitative asexual parasite density nor gametocyte density varied with HIV status (Table 2). However, irrespective of HIV status, parasite density was higher in those with severe disease (Wilcoxon rank sum P < 0.01; data not shown).
Overall, 51% (310/613) of adults with slide-confirmed malaria were admitted to hospital. Pre-hospital treatment with either SP or quinine was unrelated to HIV status. However, HIV-infected adults admitted were more likely to have impaired renal function, coma, acidosis or jaundice (Table 3). Overall, 47% of HIV-infected adult patients had severe disease compared with 30% of HIV-uninfected adults (P = 0.003). Facilities for microbiological investigation were limited; empirical use of parenteral antibiotics was similar in HIV-infected and HIV-uninfected patients with malaria. There was a higher rate of death in HIV-infected than in HIV-uninfected patients with malaria (20% versus 3.8%; P < 0.001). However time to death and overall duration of inpatient stay was unrelated to HIV status.
Severe disease was identified in 110 (35%) of patients with slide-confirmed malaria admitted to hospital. Being HIV positive was associated with having severe as opposed to non-severe disease [adjusted odds ratio (OR), 2.3; 95% confidence interval (CI), 1.4–3.9]. Similarly, being older, of male sex, having had treatment prior to admission with quinine or receiving parenteral antibiotics during the hospital admission were also independently associated with severe disease; receiving SP treatment before admission was associated with non-severe disease (Table 4). Of the independent risk factors for severe disease identified in patients with malaria who were admitted to hospital, only HIV infection was associated independently with an increased risk of death (adjusted OR 7.5; 95% CI, 2.2–25.1). Acidosis and coma were also strong independent risk factors for death (Table 5). Quinine was the therapy given to 90% (43/48) of the HIV-infected and 90% (56/62) of the HIV-uninfected patients with severe disease.
A further analysis was undertaken to incorporate all 1109 adults with positive malaria antigen tests. HIV and age parameters did not differ (Table 1). The previously defined associations with severe disease and death did not change. The respective OR values for the association of severe disease with HIV infection were 1.8 (95% CI, 1.2–2.8) unadjusted and 2.0 (95% CI, 1.3–3.2) adjusted; associations with death were 5.3 (95% CI, 2.5–11.4) unadjusted and 6.6 (95% CI, 2.2–20.1) adjusted.
The interaction between HIV and malaria has long been contentious, in part because of the poor specificity of diagnostic criteria for malaria in HIV-infected individuals living in regions of stable transmission. We postulated that any impact, particularly on severity of malaria, would be more readily identifiable in a non-immune population where parasitaemia more predictably indicates disease. We were able to investigate this in a recent malaria epidemic in KwaZulu-Natal, sampling over 1000 of the estimated 20 000 adult cases during the study period . The results suggest underlying HIV infection is associated with a twofold higher risk of severe malaria in adults, and a six- to eightfold increase in the risk of death.
There are several reasons why malaria could be more severe in adults with HIV infection. There could be relative failure to control parasite multiplication. Against this, parasitaemia density was unrelated to HIV status, although this appears not to be the case in adults with preexisting antimalarial immunity [11,12]. Worse outcome could relate to an aberrant host response triggered by malaria in those infected with HIV or to secondary complications of malaria such as bacteraemia, itself a well-recognised HIV-related problem in Africa . Other underlying HIV-related comorbidity (unrelated to acute malaria) may also have contributed to the excess mortality. Although HIV-infected adults have higher inpatient mortality than those who are not uninfected, the increased risks of death would be expected to be of the order of twofold  rather than the sevenfold increased risk reported in the patients with malaria. Furthermore time to death was unrelated to HIV status, an observation consistent with a similar underlying process leading to death. The study was not designed to investigate mechanisms and further work to characterize the pathophysiology of P. falciparum infection in non-immune HIV-infected adults is required.
Difficulties were experienced with data collection and fulfilment of the a priori malaria case definition (antigen detection confirmed by microscopy) because of the speed required to establish the study in response to the evolving epidemic. Out of the 1378 participants initially enrolled, 269 (19.5%) were excluded as a result of incomplete records or investigations. Missing records and samples were spread across all sites (data not shown), suggesting a problem of random omissions. We do not believe these exclusions can be linked to participants’ HIV status. Few people knew their HIV status. Of those who did, none disclosed it and only 11 missing samples were blood spots for HIV. It is unlikely, therefore, that adults with underlying HIV infection excluded themselves from the study. Patients were not systematically assessed for symptoms and signs suggestive of underlying HIV infection, nor for whether they fulfilled any surveillance definition for clinical AIDS. Such screening is not routinely conducted in clinics or in the hospital and it is unlikely that staff excluded patients because they suspected HIV infection.
Confirmation of malaria by microscopy was lacking in 45% of those positve by the rapid test. Antigen tests based on the P. falciparum HRP2 protein have high specificity , but the Kat-Quick test relies on correct use and interpretation of the band pattern after addition of sample and buffer. User error may, therefore, have produced false-positive results. Discussions with clinic staff during and after the study suggested that the rapid tests were being used appropriately and interpreted correctly, although real-time quality assurance could not logistically be implemented during the study. It is more likely that the large number of microscopically unconfirmed cases resulted from the poor quality of the blood slides. Prior to the study, clinic staff had no experience of slide preparation and training was brief. Consequently, inappropriately thick smears were produced, which required prolonged haemoglobin removal and staining. As a result, considerable smear material was lost from the slides, reducing the sensitivity of microscopy to detect parasites . There is little likelihood of this introducing bias, and HIV prevalence rates in the groups with confirmed and unconfirmed malaria were equal. Hospital staff had more supervision and prepared better slides than clinic staff, explaining the disparity in rates of microscopically confirmed malaria in hospitalized patients. Nevertheless a post hoc analysis was conducted, incorporating all adults with a positive rapid test irrespective of smear status. The results of this analysis were similar to those of the primary analysis: HIV infection remained strongly associated with death, reassuring us of the robustness of our original findings.
Self-medication with antimalarial drugs might have introduced bias; however, these agents are unavailable to the public as over-the-counter medicines in the study area. Treatment in both referral health centres and at the district hospital was prescribed in accordance with government treatment guidelines , promoting SP for non-severe disease. We noted that pre-hospital treatment with SP was associated with non-severe disease whereas quinine treatment was related to having severe disease. This suggests that clinic staff were able reliably to identify those patients more sick with malaria and correctly start patients on quinine prior to admission.
Severe disease and death associations are based on hospitalized individuals rather than the whole adult cohort. The study focused only on acute disease and for simplicity we did not record preexisting chronic morbidity or features of clinical AIDS. The decision to treat in the clinic or to refer to hospital was taken on presenting clinical features assessed by the nurse or doctor on duty, or according to national guidelines (pregnancy or vomiting). Those with HIV infection were more likely to be admitted. Vomiting and acute confusion/drowsiness were associated with HIV infection, and the presence of these symptoms (suggesting the need for parenteral therapy or severe disease rather than underlying chronic HIV/AIDS infection) were the likely reasons for admission. Outpatients could not be systematically followed-up at home and it was assumed that they presented again with clinical deterioration. The usual health-seeking behaviour in the local community is to return to the clinics or hospital should a condition worsen or fail to respond to therapy [25,26]. We believe that unrecognised morbidity and mortality from malaria in the outpatient group was uncommon. Similarly, inpatients when discharged were requested to attend an outpatient follow-up appointment, but non-attenders could not be followed up at home.
The demonstration of a strong association between HIV infection and poor outcome in adults with P. falciparum malaria living in areas of unstable transmission has important implications for clinical case management, malaria control and the direction of future research. During the malaria season, or specifically with the recognition of a malaria epidemic, malaria chemoprophylaxis and the adoption of comprehensive measures to limit mosquito bites (bed-nets, repellents, clothing and domestic insecticide use) may be warranted for the population at large and particularly important for HIV-infected adults. Alternative or more aggressive treatment of adult malaria, such as drug combinations and the use of artemisinin derivatives may be beneficial, and merits further investigation. In areas of high HIV prevalence and unstable malaria, all appropriate mosquito control measures with effective insecticides should be considered. In the Hlabisa study area, selective DDT (dichlorodiphenyltrichloroethane) spraying to control the pyrethroid-resistant mosquito population was credited with successfully averting a further outbreak of malaria during the 2000–2001 season .
HIV and malaria overlap extensively in sub-Saharan Africa, and even small interaction effects may have potentially massive public health and individual clinical consequences. We have demonstrated an unexpectedly large association between HIV and the outcome of malaria infection in a region of unstable transmission. The previously held belief that HIV and malaria do not significantly interact needs to be modified. Urgent investment in research and public health action is needed.
We are indebted to Dr S. Drysdale, N. Drysdale, Z. Buthelezi, G. Mdamba and D. Mthembu for their contributions to staff training, supervision and collection of samples and data in Hlabisa; to Dr W. Bailey, K. Moi and M. Moore in the laboratory in Liverpool, and the staff and patients of Hlabisa health district.
Sponsorship: This study was financed by the UK Department For International Development (DFID) HIV/AIDS/STI and Malaria Knowledge Programmes at the Liverpool School of Tropical Medicine.
Note: DFID does not accept responsibility for any information supplied or views expressed in this article.
1. Leroy V, Msellati P, Lepage P,
Batungwanayo J, Hitimana DG, Taelman H, et al. Four years of natural history of HIV-1 infection in African women: a prospective cohort study in Kigali (Rwanda),
1988–1993. J Acquir Immune Defic Syndr Hum Retroviro
l 1995, 9
2. Anglaret X, Chêne G, Attia A, Toure S, Lafont S, Combe P, et al. Early chemoprophylaxis with trimethoprim-sulphamethoxazole for HIV-1-infected adults in Abidjan, Côte d'Ivoire: a randomised trial. Lancet,
3. Nguyen-Dinh P, Greenberg AE, Mann JM, Kabote N, Francis H, Colebunders RL, et al. Absence of association between Plasmodium falciparum malaria and human immunodeficiency virus infection in children in Kinshasa, Zaire . Bull World Health Org
4. Colebunders R, Bahwe Y, Nekwei W, Ryder R, Perriens J, Nsimba K, et al. Incidence of malaria and efficacy of oral quinine in patients recently infected with human immunodeficiency virus in Kinshasa, Zaire. J Infect
5. Muller O, Moser R. The clinical and parasitological presentation of Plasmodium falciparum malaria in Uganda is unaffected by HIV-1 infection. Trans R Soc Trop Med Hyg
6. Greenberg AE, Nsa W, Ryder RW, Medi M, Nzeza M, Kitadi N, et al. Plasmodium falciparum malaria and perinatally acquired human immunodeficiency virus type 1 infection in Kinshasa, Zaire. A prospective, longitudinal cohort study of 587 children. N Engl J Med
7. Steketee RW, Wirima JJ, Bloland PB, Chilima B, Mermin JH, Chitsulo L, et al. Impairment of a pregnant woman's acquired ability to limit Plasmodium falciparum by infection with human immunodeficiency virus type-1. Am J Trop Med Hyg
8. Steketee RW, Wirima JJ, Slutsker L, Breman JG, Heymann DL. Comparability of treatment groups and risk factors for parasitemia at the first antenatal clinic visit in a study of malaria treatment and prevention in pregnancy in rural Malawi. Am J Trop Med Hyg
9. Verhoeff FH, Brabin BJ, Hart CA, Chimsuku L, Kazembe P, Broadhead RL. Increased prevalence of malaria in HIV-infected pregnant women and its implications for malaria control. Trop Med Int Health
10. Bloland PB, Wirima JJ, Steketee RW, Chilima B, Hightower A, Breman JG. Maternal HIV infection and infant mortality in Malawi: evidence for increased mortality due to placental malaria infection. AIDS
11. Whitworth J, Morgan D, Quigley M, Smith A, Mayanja B, Eotu H, et al. Effect of HIV-1 and increasing immunosuppression on malaria parasitaemia and clinical episodes in adults in rural Uganda: a cohort study. Lancet
12. French N, Nakiyingi J, Lugadda E, Watera C, Whitworth JAG, Gilks CF. Increasing rates of malarial fever with deteriorating immune status in HIV-1–infected Ugandan adults. AIDS
13. Simooya OO, Mwendapole RM, Siziya S, Fleming AF. Relation between falciparum malaria and HIV seropositivity in Ndola, Zambia. Br Med J
14. Leaver RJ, Haile Z, Watters DA. HIV and cerebral malaria. Trans R Soc Trop Med Hyg
15. Gilks CF, Haran D, Wilkinson D. Coping with the impact of the HIV epidemic: the Hlabisa–Liverpool HIV link. S Afr Med J
16. Moorthy V, Wilkinson D. Severity of malaria and level of Plasmodium falciparum transmission. Lancet
17. South African Department of Health. South African Health Review 2000.
Pretoria: South African Department of Health; 2001.
19. World Health Organization. Severe falciparum malaria. World Health Organization, Communicable Diseases Cluster. Trans R Soc Trop Med Hyg
20. Shute GT. The microscopical diagnosis of malaria.
In Principles and Practices of Malariology.
Edited by Wernsdorfer J, McGregor I. Edinburgh: Churchill-Livingstone, 1988:781–814.
22. Gilks CF, Brindle RJ, Otieno LS, Simani PS, Newnham RS, Bhatt SM, et al
. Life-threatening bacteraemia in HIV-1 seropositive adults admitted to hospital in Nairobi, Kenya.
Lancet 1990, 336
23. Arthur G, Nduba VN, Kariyuki S, Kimari J, Bhatt S, Gilks CF. Trends in bloodstream infections among human immunodeficiency virus-infected adults admitted to a hospital in Nairobi, Kenya, during the last decade.
Clin Infect Dis 2001, 33
24. Beadle C, Long GW, Weiss WR, McElroy PD, Maret SM, Oloo AJ, et al. Diagnosis of malaria by detection of Plasmodium falciparum HRP-2 antigen with a rapid dipstick antigen-capture assay. Lancet
25. Wilkinson D, Gcabashe L, Lurie M. Traditional healers as tuberculosis treatment supervisors: precedent and potential. Int J Tuberc Lung Dis
26. Pronyk RM, Makhubele MB, Hargreaves JR, Tollman SM, Hausler HP. Assessing health seeking behaviour among tuberculosis patients in rural South Africa. Int J Tuberc Lung Dis
Keywords:© 2004 Lippincott Williams & Wilkins, Inc.
falciparum; epidemic unstable malaria; HIV disease; outcome; death; Africa