Falciparum malaria and HIV-1 infection represent two of the most important health problems facing sub-Saharan Africa. Despite their wide distribution, frequency and inevitable clinical co-existence, evidence to suggest an interaction has been lacking, other than by the direct consequences of HIV-contaminated blood transfusion in severe anaemic malaria. Studies in pregnant women have suggested higher parasite rates in HIV-1-infected women [1–3] and a possible clinically important negative impact on perinatal outcome . However, the small number of studies in children and non-pregnant adults have failed to show an association between falciparum malaria and HIV-1 infection [5–10]. We now report our findings from a well-characterized cohort of HIV-1-infected adults living in Entebbe, Uganda, and suggest that there is an important increase in the rates of falciparum malarial fever as a consequence of HIV-1-associated immunosuppression.
Materials and methods
Participants were recruited at two community clinics in Entebbe, Uganda, the Uganda Virus Research Institute and the AIDS Support Organisation. These clinics provide counselling, HIV testing, medical care and basic welfare. This region is hyper-endemic for malaria.
Between October 1995 and June 1998, HIV-1-infected adults (17 years or older) living within 15 km of the study clinics, were recruited into an efficacy trial of 23-valent pneumococcal polysaccharide vaccine . Medical care was free to study participants with transport subsidized for clinic attendance. Consent for home visiting was sought from all participants.
Parasite measurements were performed in asymptomatic and well HIV-uninfected control adults, with their consent, while attending the clinics for voluntary counselling and HIV testing. They were not followed-up except to receive HIV results and malaria therapy if parasitaemic.
At study entry, a baseline clinical assessment was conducted on each participant. A haemoglobin, white cell and CD4 T cell measurement was performed and serum and plasma were stored. Participants were reviewed every 6 months if well (routine visit). The clinical assessment and investigations performed at entry were repeated. In addition, participants were questioned about self-administered usage of antimalarial medication in the 2 weeks preceding the routine visit, and were asked about episodes of illness managed independently of the trial clinics. From April 1997 until the end of the study a malaria blood slide was performed at each routine visit. Individuals were considered well if they were asymptomatic, afebrile, and did not report antimalarial usage in the preceding 2 weeks.
If not well, the routine visit data collection was deferred until the clinical problem was resolved. Meanwhile, the clinic visit was categorized as a clinical illness visit, and the event protocol was used with appropriate data and clinical specimen collections.
If there was non-attendance at the 6 monthly routine clinic appointment, a study field-worker would visit a participant at home within the subsequent 4 weeks. The field worker would visit up to three times, one week apart, if a participant continued to be absent. If alive, they would be offered transport to the clinic, or if too sick to travel, they would be visited by a study physician. If dead, a verbal autopsy was performed. If they refused further participation in the cohort, had left the study area, or could not be traced by virtue of an inaccurate address, they were deemed to be defaulters. The established pneumococcal vaccine trial protocol then censored individuals from the study at the failed clinic visit or the date of departure from the study area .
Participants had open access to the study clinics when sick and were then investigated and managed according to set protocols (event visit) free of charge. If a relative informed the clinic that the participant was unwell but unable to travel, transport to the clinic would be arranged or a home visit by a study clinician would be carried out within 48 h. The intense, passive clinical surveillance of the cohort for febrile episodes was designed to maximize the identification of invasive pneumococcal disease, an event that can be difficult to diagnose. Standard clinical protocols were used to investigate each febrile illness episode according to the presenting syndrome. Protocols for symptomatic febrile patients (axillary temperature > 37.4°C) without focus of infection consisted of bacterial, fungal and mycobacterial culture of blood, malaria slides and chest X-ray. All other febrile events (e.g. cough and fever, fever and diarrhoea) also routinely included a malaria slide. Other investigations related to the clinical presentation (e.g. sputum Gram and Ziehl–Neelsen staining and culture for bacteria and mycobacteria in cases of respiratory infections).
Malarial fever was defined in the standard fashion as an acute (duration < 14 days) febrile illness with asexual stage malaria parasites in the blood, no other aetiology (e.g. tuberculosis, bacteraemia or fungaemia), and resolution with antimalarial therapy, routinely fansidar or quinine. In order to increase the specificity of a diagnosis of malarial fever, a stricter case definition incorporating a parasite density cut-off (the 75th centile for parasite density in well individuals) was used. Other events were diagnosed according to standard clinical and laboratory criteria.
HIV-1/2 testing was performed using a rapid latex agglutination technique (Capillus, Cambridge Biotech, Galway, Ireland), followed by confirmation using two enzyme immunoassays (Wellcozyme, Wellcome Diagnostics, Hertford, UK and Recombigen, Cambridge Biotech). CD4 T cell counts were performed using a FACS-count system (Becton-Dickinson, San Jose, USA). Total leukocyte counts were measured using the quantitative blood chemistry semi-automated system (Becton-Dickinson). Blood smears were stained using Leishman or Giemsa techniques, and examined according to conventional procedures. Parasite density was calculated as follows: actual leukocyte count × (asexual stage parasites per 200 leukocytes on the thick film/200). Quality control of malaria microscopy involved dual reading of approximately every 20th slide with less than 5% interobserver error noted. The study laboratory also participated in a UK-based national external quality assurance scheme for blood parasitology (91% success rate).
Rates and proportions
Parasite densities were calculated for all positive samples identified at routine or febrile event visits. Malarial fever rates were calculated for all events. Recurrent events within 28 days of the initial presentation were treated as relapses and not new events. To allow for HIV disease progression and CD4 T cell count decreases in each individual participant, person-years of observation (pyo) were calculated in 6 month blocks. These blocks were assigned to a CD4 T cell stratum on the basis of the cell count at the start of the 6 month period. By convention, the strata used were less than 200, 200–499 and 500 and above. Asymptomatic positive blood smears were assigned to the CD4 T cell stratum based on a measurement performed on the same blood sample.
Data were gathered on standardized proformas and dually entered into a database, foxpro (Microsoft Corporation, CA, USA). Stata statistical software (Stata Corporation, Release 6.0, College Station, TX, USA) was used to perform calculations. Linear trends of proportions were assessed using the χ2 test. Trends in densities were investigated using a linear regression of log10 transformed data. Rate data were compared using a negative binomial regression model. Likelihood ratio statistics were used to investigate models with and without CD4 T cell strata to estimate the interaction of trend in rates and CD4 T cell count.
The research was approved by the Ugandan Ministry of Health, AIDS research sub-committee and the ethical review board of the Liverpool School of Tropical Medicine. Study participants gave signed informed consent, after discussions in local languages. The study was carried out in accordance with the UK Medical Research Council's guidelines for good clinical practice in clinical trials.
The pneumococcal polysaccharide vaccine trial started in October 1995. By the end of June 1998, 1371 HIV-1-infected adults (70% women) had been enrolled: participants’ median age was 31 years (range 19–67); the median duration of follow-up was 1.2 years; 371 (27%) died (rate 253/1000 pyo); and 165 (12%) were lost to follow-up. In total, 766 febrile events were investigated at the study clinics and 153 episodes (20% of all febrile episodes) in 117 individuals fulfilled our standard case definition for Plasmodium falciparum malarial fever. All patients considered to have malarial fever received standard therapy (fansidar or quinine) within 48 h. Out of 92 bacteraemic episodes (Streptococcus pneumoniae 35, non-typhi Salmonella 37, other coliforms 16, other Gram positives 4) malarial parasites were identified in one case of non-typhi Salmonella bacteraemia (1.1% of bacteraemic episodes). In this case of dual infection, response to therapy was only achieved after the use of ciprofloxacin after the failure of the quinine/amoxycillin combination. Three episodes of non-falciparum malaria were also diagnosed and treated, but are excluded from the analysis.
There were no cases of cerebral or complicated malaria and no deaths clearly attributable to malaria. A cause of death was identified in 288 (78%) cases. The majority of deaths occurred in participants with CD4 T cell counts less than 200; 284, 77% of total, at a rate of 490 per 1000 pyo . Principal causes of death were: wasting syndrome 73 cases (20% all deaths); cryptococcal disease 61 cases (17% all deaths); and tuberculosis 23 cases (6% all deaths). One individual presented during their terminal illness with both falciparum parasitaemia and culture-confirmed cryptococcal meningitis. Therapy with quinine but not antifungal agents failed to halt a progressive encephalopathic/meningitic illness. The primary cause of death was attributed to cryptococcal disease. Of the 93 deaths with no identified cause, 68 (73%) were in those with CD4 T cell counts of less than 200.
As expected, a strong association between lower CD4 T cell count strata and rates of tuberculosis, non-typhi Salmonella bacteraemia, S. pneumoniae infection and cryptococcal disease was noted [13–15]. In addition, rates of malarial fever also increased with lower CD4 T cell counts (Table 1).
Approximately one in 10 well adults in Entebbe, a region hyper-endemic for falciparum malaria, are blood-slide positive at any one time, and this appears to be independent of the underlying HIV status or degree of immunosuppression (Table 2). When investigated for fever, patients may therefore be parasitaemic by chance rather than have malarial fever. The proportion of positive slides in sick (febrile/symptomatic) patients is consistently more than double that in well participants for all CD4 T cell strata (Table 2). However, increased sampling of the more immunosuppressed sicker patients may merely be identifying co-existent parasitaemia. Rates of any febrile illness increased significantly with decreasing CD4 T cell count; 140, 293 and 597 per 1000 person-years for CD4 T cell groups 500, 200–499 and less than 200, respectively (likelihood ratio test χ2 = 87.5, P < 0.001). This may have led to an overestimate of the association between the higher rates of malarial fever with lower CD4 T cell counts. To allow for ascertainment bias we used data on parasite densities in well individuals to generate a stricter case definition for malarial fever. There was a clear difference between malaria parasite densities in well (afebrile/asymptomatic) participants and patients with malarial fever for all CD4 T cell strata (Fig. 1). Furthermore, there was a trend towards increasing density with lower CD4 T cell count in the malarial fever group (Fig. 2a). This trend was not observed in the well/asymptomatic participants (Fig. 2b), although the regression coefficients were not significantly different. There was no change in parasite densities in asymptomatic/well patients according to HIV status (Fig. 3). Modifying the malarial fever case definition by incorporating a parasite density increased the specificity of the definition. Excluding malarial fever cases with parasite densities below the 75th centile for well individuals (2.8 × 109 parasites/l) strengthened the trend of increased rates at lower CD4 T cell counts (Table 1 b). Using more conservative density cut-offs did not affect this trend, e.g. 90th centile for asymptomatic individuals (8.4 × 109 parasites/l) rates of malarial fever were 16, 33, 76 per 1000 person-years, likelihood ratio test χ2 = 18.5, P < 0.001.
The total of 153 malarial fever events included 36 recurrences in 32 individuals. The overall recurrence rate was 394 per 1000 pyo. There was no association of recurrent events with CD4 T cell counts: 446, 348 and 411 per 1000 pyo for CD4 T cell groups 500, 200–499 and less than 200, respectively, using both the standard malarial fever definition (P = 0.85) and modified case definition (data not shown).
Pregnancy, a predisposing factor for malaria, was uncommon, with 14 deliveries in the study period despite many female participants. Reported self-medication for malaria was assessed at routine visits. Antimalarial usage by sick patients in the 2 weeks preceding the consultation was similar across the three CD4 T cell strata (21–25%, χ2 = 2.0, df = 2, P = 0.37). Questioning at routine follow-up also revealed that only 4% of severe clinical episodes were unreported to the clinic, with no relationship to CD4 T cell strata.
These data reveal an increasing risk of malarial fever with advancing HIV immunosuppression, suggesting an interaction between P. falciparum and HIV. The presence of an association would be expected based on our understanding of antimalarial immunity, and is entirely consistent with data from a separate cohort of HIV-infected adults living in a rural location in south-west Uganda. , but these are the first data to confirm the relevance of this knowledge and the importance of HIV-1 infection on falciparum malaria fever in sub-Saharan Africa .
The principal interaction was not with malarial infection per se, background rates of parasitaemia and densities were similar across CD4 T cell strata, but with the development of clinically manifest febrile disease. Our main concern, when ascribing causality to a case of fever was to avoid bias introduced by increased sampling of individuals at lower CD4 T cell counts and the co-existence of parasitaemia and fever. As expected the rates of any febrile illness increased dramatically as the CD4 T cell count decreased. The use of a specific parasite density to provide a greater positive predictive value in ascribing a febrile condition to malaria in endemic areas is accepted . We chose a level derived from our active asymptomatic surveillance data. The choice of the 75th centile was arbitrary, but is biologically plausible and, we believe, valid. Active surveillance of well individuals was conducted separately from the clinical investigation of sick patients and generated independent data. Moreover, using higher, more conservative densities or deriving a cut-off from more complex models such as that described by Smith et al.  may affect the absolute rates but will not affect the trend.
As we used parasite densities to increase the specificity of our malarial fever definition, the possibility that fever itself could increase parasite densities, making the standard assumptions about densities and causality of fever invalid, was also considered. Such an effect of fever seems unlikely. Parasite rates in culture-confirmed bacteraemic illness were only 1 in 92, significantly less than would be expected by chance, given the background rate of parasitaemia of 10%, suggesting a suppressive effect of bacteraemia on P. falciparum parasitaemia. In addition, the high malarial fever recurrence rates would not be explained by a non-specific effect of fever.
We have been unable to identify any inherent bias to question these results. Individuals were classified by signs and symptoms before preparing the malaria smear, and microscopy was performed blind of clinical data. Parasite densities were calculated using actual leukocyte counts, and not the standard white cell count (6 × 109/l) commonly used, to avoid an overestimation of trophozoite numbers in the presence of HIV-related leukopaenia. Despite many female participants, pregnancy was uncommon and was unlikely to have contributed to any susceptibility at lower CD4 T cell counts. The large proportion of female participants directly reflected the clinic population. The majority of these women were widowed, without stable income and were attracted to the clinic by basic welfare, foodstuffs and female comradeship. The clinic promoted abstinence and safe sex among its clients (distributing free condoms), and combined with the loss of a partner and often public knowledge of their HIV status, probably contributed to the low pregnancy rates. Free medication and subsidized transport were strong incentives to attend the study clinics when sick, and we believe that most febrile events were seen, with only a few reports of attendance at other health centres.
Mild self-limited/rapidly treated malarial fever may have been missed as a consequence of our passive surveillance for cases. It might thus be argued that the cases we saw represented the more symptomatic individuals, those responding less well to self-treatment, or perhaps health-seeking behaviour changed with decreasing CD4 T cell counts. Our data did not allow these issues to be fully explored. However, reported antimalarial usage by study participants was frequent, but did not significantly vary across the CD4 T cell strata, suggesting that behaviour with regard to self-treatment was similar. The malabsorption of antimalarial agents from late-stage enteropathy leading to poor response seems unlikely given the high bioavailability of chloroquine (the principal agent used in self-treatment) and the good response to our own prescribed oral therapy.
Our data are not inconsistent with other published reports [5–10] and a systematic review , which showed no convincing evidence of an association between asymptomatic malaria and HIV infection, except perhaps during pregnancy [1–3]. In Africa, parasitaemic adults will usually be asymptomatic at a given time and, in keeping with our findings, parasite rates and densities in cross-sectional sampling will show no difference. Most studies were cross-sectional [5,6,9,10] and sought differences in malaria rates between HIV-1-infected and uninfected individuals without considering HIV stage or, in some, the presence of symptoms. Interestingly, in the pregnancy studies, differences in parasite rates between HIV-infected and uninfected women showed a clear relationship between parity, with greater differences at higher parities. Multiparity is likely to be a crude indicator of the stage of HIV immunosuppression, by virtue of longer sexual activity.
Two prospective studies [7,8] also failed to show any association but had limitations. One study  compared uninfected and recently seroconverted HIV-1-infected individuals of all ages, but predominantly children, likely at this stage to be relatively immunocompetent. Anti-parasite immunity may have been relatively well preserved in the HIV group. No data were presented on parasite densities to improve diagnostic specificity. The other study , in children only, reported higher rates of malarial fever, with a non-significant trend towards higher mean parasite densities in children with AIDS. No data on background rates of parasitaemia were reported, rates of malarial fever were high irrespective of HIV status, the identification of other intercurrent disease events was limited and the observation time was short. Moreover, interference with the acquisition of anti-parasite immunity in children may not be directly comparable with the progressive loss of malaria-specific immune responses in adults, suggesting caution in applying findings from one group to the other.
The rates of recurrent disease were high irrespective of the CD4 T cell count. Similar observations have been made with other HIV co-infections, in particular invasive S. pneumoniae disease . Such findings suggest that once anti-parasite immunity is impaired there is a persistent defect. Despite this increased susceptibility, there was no apparent effect on the severity of malarial fever. No cases of severe and complicated malaria were seen in our study; all patients received prompt treatment, and it is a relatively uncommon sequela of malarial fever in adults in hyperendemic areas. Of the 93 deaths without an identified cause, most occurred in individuals with low CD4 T cell counts and usually in individuals who had been moved from the study area by relatives taking them back to their ancestral village before death. There is no reason to believe that they represent a highly selective group with regard to malaria susceptibility. Outside well-funded research projects, access to effective care may be limited. Severe malaria disease and death may also be associated with HIV infection, but conclusions cannot be drawn from our data. Three hospital-based cross-sectional studies [21–23] were too small or incomplete to draw conclusions about the rate of severe malaria, but two noted that the mortality rate was double in patients with HIV infection.
The interaction of HIV-1 with malaria is important. HIV-1 infection is widespread in the tropics, and the majority of infected individuals will be exposed to P. falciparum at some stage. Investigating the mechanism of the interaction may provide a new avenue for understanding the protective immune response to P. falciparum infection. Furthermore, our findings suggest an additional HIV-related public health problem in Africa, and have important implications for the clinical management of HIV/AIDS patients in malaria-endemic areas. An increased emphasis on mosquito avoidance with impregnated bed-net usage is the most relevant immediate action, while chemoprophylaxis strategies are considered. Many questions remain, such as the impact of HIV infection on gametocyte rates and its effect on the spread of malaria, and further studies to investigate this are clearly merited. The growing optimism that somehow the critical global malaria situation, especially in Africa, was largely unaffected by the HIV epidemic needs revision.
The authors would like to thank all staff in Uganda who participated in the trial and all the seropositive clients and their carers who helped in the project implementation. They also thank Dr Daan Mulder who assisted with the development of this work and who sadly died before its completion.
Presented in part at the Royal Society of Tropical Medicine and Hygiene, Manson House, Portland Place, London on 18 March 1999 .
1. Steketee RW, Wirima JJ, Slutsker L. et al
. Malaria parasite infection during pregnancy and at delivery in mother, placenta, and newborn: efficacy of chloroquine and mefloquine in rural Malawi. Am J Trop Med Hyg 1996, 55: 24 –32.
2. Steketee RW, Nahlen BD, Avisi J, Van Eijk A, Misore A. HIV and malaria overlap and do interact in sub-Saharan African pregnant women.Proceedings of the XIIth International Conference on AIDS
. Geneva, 1998 [Abstract no. 13298:145].
3. 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 1999, 4: 5 –12.
4. 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 1995, 9: 721 –726.
5. Nguyen-Dinh P, Greenberg AE, Mann JM. et al
. Absence of association between Plasmodium falciparum
malaria and human immunodeficiency virus infection in children in Kinshasa, Zaire. Bull WHO 1987, 65: 607 –613.
6. 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 1990, 84: 336 –338.
7. Colebunders R, Bahwe Y, Nekwei W. et al
. Incidence of malaria and efficacy of oral quinine in patients recently infected with human immunodeficiency virus in Kinshasa, Zaire. J Infect 1990, 21: 167 –173.
8. Greenberg AE, Nsa W, Ryder RW. et al
. Plasmodium falciparummalaria and perinatally acquired human immunodeficiency virus type 1 infection in Kinshasa, Zaire.A prospective, longitudinal cohort study of 587 children.
N Engl J Med 1991, 325: 105 –109.
9. Atzori C, Bruno A, Chichino G. et al
. HIV-1 and parasitic infections in rural Tanzania. Ann Trop Med Parasitol 1993, 87: 585 –593.
10. Nwanyanwu OC, Kumwenda N, Kazembe PN. et al
. Malaria and human immunodeficiency virus infection among male employees of a sugar estate in Malawi. Trans R Soc Trop Med Hyg 1997, 91: 567 –569.
11. French N, Nakiyingi J, Carpenter LM. et al
. 23-Valent pneumococcal polysaccharide vaccine in HIV-1-infected Ugandan adults: a double-blind, randomised and placebo controlled trial. Lancet 2000, 355: 2106 –2111.
12. French N, Nakiyingi J, Mujugira A, Mulder D, Janoff EN, Gilks CF. In HIV-1-infected Ugandan adults immunologic and clinical stage are comparable and reveal no evidence of rapid progression but poor survival with advanced disease. J Acquir Immune Defic Syndr 1999, 22: 509 –516.
13. French N, Gilks CF, Mujugira A, Janoff EN. Impact of febrile illness among patients with HIV-1 disease in Uganda.Proceedings of the 37th Interscience Conference on Antimicrobial Agents and Chemotherapy
. Toronto, September–October 1997, 37 [Abstract no. I–155].
14. Gilks CF, Brindle RJ, Otieno LS. et al
. Life-threatening bacteraemia in HIV-1 seropositive adults admitted to hospital in Nairobi, Kenya. Lancet 1990, 336: 545 –549.
15. Gilks CF, Ojoo SA, Ojoo JC. et al
. Invasive pneumococcal disease in a cohort of predominantly HIV-1 infected female sex-workers in Nairobi, Kenya. Lancet 1996, 347: 718 –723.
16. Whitworth J, Morgan D, Quigley 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 –1056.
17. French N, Gilks CF. Fresh from the field: some controversies in tropical medicine and hygiene.HIV and malaria, do they interact?
Trans R Soc Trop Med Hyg 2000, 94: 233 –237.
18. Greenwood BM, Bradley AK, Grenwood AM. et al
. Mortality and morbidity from malaria among children in a rural area of The Gambia, West Africa. Trans R Soc Trop Med Hyg 1987, 81: 478 –486.
19. Smith T, Scellenberg JA, Hayes R. Attributable fraction estimates and case definitions for malaria in endemic areas. Stat Med 1994, 13: 2345 –2358.
20. Chandramohan D, Greenwood BM. Is there an interaction between human immunodeficiency virus and Plasmodium falciparum
? Int J Epidemiol 1998, 27: 296 –301.
21. Simooya OO, Mwendapole RM, Siziya S, Fleming AF. Relation between falciparum malaria and HIV seropositivity in Ndola, Zambia. BMJ 1988, 297: 30 –31.
22. Leaver RJ, Haile Z, Watters DA. HIV and cerebral malaria. Trans R Soc Trop Med Hyg 1990, 84: 201. 201.
23. Niyongabo T, Deloron P, Aubry P. et al
. Prognostic indicators in adult cerebral malaria: a study in Burundi, an area of high prevalence of HIV infection. Acta Trop 1994, 56: 299 –305.