Transactions of the Royal Society of Tropical Medicine and Hygiene (2005) 99, 561—567

Treatable factors associated with severe anaemiain adults admitted to medical wards in Blantyre,Malawi, an area of high HIV seroprevalence

David K. Lewisa,∗, Christopher J.M. Whittya,b, Amanda L. Walshc,Henry Epinoa, Nynke R. van den Broekc,d, Elizabeth A. Letskye,Clyton Munthali a, Joshua M. Mukiibi f, Martin J. Boereea

a Department of Medicine, University of Malawi College of Medicine, Private Bag 360, Chichiri,Blantyre 3, Malawib London School of Hygiene and Tropical Medicine, University of London, Keppel Street,London WC1E 7HT, UKc Malawi—Liverpool—Wellcome Trust Clinical Research Programme, P.O. Box 30096, Blantyre 3, Malawid Liverpool School of Tropical Medicine, Pembroke Place, Liverpool L3 5QA, UKe Department of Haematology, Queen Charlotte’s Hospital, London, UKf Department of Haematology, University of Malawi College of Medicine, Blantyre, Malawi

Received 22 July 2004; received in revised form 6 December 2004; accepted 19 January 2005Available online 12 May 2005

KEYWORDSAnaemia;HIV;AIDS;Tuberculosis;Bacterial infections;Sub-Saharan Africa

Summary Severe anaemia is a common presentation in non-pregnant adults ad-mitted to hospital in southern Africa. Standard syndromic treatment based on datafrom the pre-HIV era is for iron deficiency, worms and malaria. We prospectivelyinvestigated 105 adults admitted consecutively to medical wards with haemoglobin<7 g/dl. Those with acute blood loss were excluded. Patients were investigatedfor possible parasitic, bacterial, mycobacterial and nutritional causes of anaemia,including bone marrow aspiration, to identify potentially treatable causes. Seventy-nine per cent of patients were HIV-positive. One-third of patients had tuberculosis,which was diagnosed only by bone marrow culture in 8% of HIV-positive patients. In21% of individuals bacteria were cultured, with non-typhi salmonella predominatingand Streptococcus pneumoniae rare. Iron deficiency, hookworm infection andmalaria were not common in HIV-positive anaemic adults, although heavy hook-worm infections were found in 6 (27%) of the 22 HIV-negative anaemic adults. Inconclusion, conventional treatment for severe anaemia in adults is not appropriatein an area of high HIV prevalence. Occult mycobacterial disease and bacteraemiaare common, but iron deficiency is not common in HIV-positive patients. In addition

* Corresponding author. Present address: Vauxhall Primary Health Care, Limekiln Lane, Liverpool L5 8XR, UK.Tel.: +44 151 298 2246; fax: +44 151 207 1272.

E-mail addresses: david.lewis@livgp.nhs.uk, mariadavid@callaghanlewis.fsnet.co.uk (D.K. Lewis).

0035-9203/$ — see front matter © 2005 Royal Society of Tropical Medicine and Hygiene. Published by Elsevier Ltd. All rights reserved.

doi:10.1016/j.trstmh.2005.01.002

562 D.K. Lewis et al.

to iron supplements, management of severe anaemia should include investigation fortuberculosis, and consideration of antibiotics active against enterobacteria.© 2005 Royal Society of Tropical Medicine and Hygiene. Published by Elsevier Ltd.All rights reserved.

1. Introduction

In Blantyre, Malawi a primary diagnosis of anaemiacontributes to 10% of deaths on adult medicalwards (unpublished audit). In a study of 344 febrileadults (58% medical admissions) on our wards, two-thirds were anaemic (haemoglobin <11 g/dl), and18% severely anaemic (haemoglobin <7 g/dl) (Lewiset al., 2002). For women attending an antenatalclinic in Blantyre, the figures were 57% (anaemia)and 4% (severe anaemia) (van den Broek et al.,2000). These figures are probably broadly represen-tative of southern African hospital practice. Despiteits importance, there are very few recent studies onthe causes of anaemia in adults who are not preg-nant.

In routine practice few investigations areavailable and blind treatment for anaemia oftenconsists of iron and anthelmintics with or withoutantimalarials. In classical tropical medical practicethis was reasonable; for example, in the pre-HIVera about half the adults with severe anaemia inhospital in Harare were iron-deficient, often withhookworm in their stool (Gelfand, 1968). Little

ing especially on treatable causes. Inflammation al-ters biochemical measurement of iron status, andexamination of a stained bone marrow aspirate isthe gold standard for assessing iron status (vanden Broek and Letsky, 2000). The study includedbone marrow examination, and set out to identifyparasitic, bacterial, mycobacterial and nutritionalcauses of anaemia as a way of helping cliniciansfaced with this problem to determine best treat-ment.

2. Materials and methods

Patients were adults consecutively admitted to themedical wards of Queen Elizabeth Central Hospi-tal, the main provider of secondary care to Blan-tyre, the largest city in Malawi. The Departmentof Medicine admits around 800 patients per month,of whom 70% are HIV-positive (Lewis et al., 2003).Patients with haemoglobin <7 g/dl were eligible forthe study. Those known to have had acute blood lossor those too ill to consent to bone marrow aspira-tion were excluded.

is known about the causes of severe anaemia innon-pregnant adults in sub-Saharan Africa sinceHIV became widespread. As over half of adultadmissions to medical wards have HIV in thisregion, it is likely that HIV has affected the patternof causes of anaemia.

Links between HIV, infections and anaemia arewell recognized, but the extent to which they con-tribute to the burden of anaemia in non-pregnantadults in clinical practice is not known. HIV and tu-berculosis (TB) are common in the adult populationin southern Africa, and most people with AIDS orsevere TB are anaemic (Morris et al., 1989; Sullivanet al., 1998). Decreasing haemoglobin correlatesstrongly with increasing mortality in African andWestern cohorts of HIV-positive individuals (Elliottet al., 1995; Moore, 1999; Sullivan et al., 1998),and recovery from anaemia is linked to improvedoutcome (Mocroft et al., 1999).

As severe anaemia (defined as haemoglobin<7 g/dl [Stoltzfus, 1997]) is a common cause ofadult hospital admission and there are currentlyfew data to guide treatment, we prospectively in-vestigated adults admitted to medical wards withsevere anaemia for underlying factors, concentrat-

After obtaining informed consent (including pre-test counselling for HIV test) a standardized historywas taken, and physical examination performed.Blood was drawn for: full blood count; bacterialculture; peripheral blood film, including reticulo-cyte count and examination for malaria parasites;erythrocyte sedimentation rate; HIV testing; andserum vitamin B12. CD4+ counts were measured onHIV-positive samples. All patients had a chest X-rayand were asked to give urine, stool and sputum sam-ples. After a full explanation of the procedure, abone marrow sample was taken from the sternumunder local anaesthetic. Bone marrow aspirate wasprocessed for bacterial and mycobacterial culture,and histological, fungal, acid-fast bacilli and ironstaining.

Chest X-rays were viewed by a chest physicianblind to other findings, with blind audit by an in-fectious diseases physician. Definite TB was definedas any of the following: acid-fast bacilli in sputumor lymph node aspirate; lymphocytic pleural effu-sion with fever; and cavities or miliary pattern onchest X-ray. Probable TB was defined as a highlysuggestive chest X-ray plus chronic (>1 month) un-explained cough and fever or weight loss.

Causes of severe anaemia in Malawi 563

Full blood count was performed with a Coul-ter counter. Thick films for malaria parasites werestained with Field’s stain and at least 20 fieldswere examined. HIV testing was performed with arapid test (HIVSPOT, Genelabs Diagnostics, Singa-pore) and confirmed by ELISA (Organon Vironostika,Geneva, Switzerland or Capillus HIV/HIV2, Cam-bridge Diagnostics, Galway, Ireland); patients witha positive result in both tests were regarded as HIV-positive. CD4+ counts were made with a BectonDickinson FACS Count.

For bacterial cultures, 5ml blood (or 1—2mlmar-row) in a single 50ml (20ml for marrow) bloodculture bottle (Brain Heart Infusion Broth contain-ing sodium polyanethol sulphonate, E&O Labora-tories, Bonnybridge, UK) was incubated overnightat 37 ◦C before venting. Cultures were examinedmacroscopically every day, followed by Gram stain-ing if turbid or haemolysed. Isolates were identifiedaccording to standard techniques, including seroag-glutination and biochemical tests. Sputum smearswere stained with Ziehl—Neelsen (ZN).

A 1ml aliquot of marrow was inoculated into asupplemented Middlebrook 7H9 broth (Difco). Bot-tles were incubated for 6 weeks, and examined byZ

swd

f5owo

swcawm1iimmca

pvt(c

These were age, sex, urban/rural residence, yearsof education and literacy. Education and literacywere used as surrogate markers of socio-economicstatus, which could affect prevalence of HIV andother infections.

Ethical approval was given by the Research Com-mittee of the College of Medicine, University ofMalawi.

3. Results

During the study period, 128 consecutive patientsfulfilled the entry criteria, and 105 (82%) of thoseconsented and entered the study. The demograph-ics of those who declined investigation were similarto those who consented.

The median age of patients was 35 years (range18—88, interquartile range 28—43); 47 (45%) weremen, and 49 (47%) lived in an urban residence.Eighty-three (79%) patients were HIV-positive, ofwhom 77 had CD4+ measured and 62 (81% of thosemeasured) had a CD4+ count <200. Fifty-six (53%)patients met the expanded WHO case definition forAt(H

sl1Ppich(omFf

t(srtdcfptsa

N-staining once a week.Marrow slides, which had been prepared as the

amples were collected, were air-dried and stainedith Wright’s stain. These were examined for evi-ence of dimorphic fungi.A formol ether concentration method was used

or examination of stool; a random selection of over0% of stools were read independently by a sec-nd reader blind to the first reading. Terminal urineas filtered and examined microscopically for ovaf Schistosoma haematobium.Bone marrow aspirates were considered repre-

entative if they contained fragments, and theyere all examined by a haematologist blinded tolinical and laboratory details, and confirmed bysecond independent haematologist. Marrow filmsere stained for iron using Prussian Blue (Perls’ethod) and classified as: 0, no iron present;, trace of iron; 2, normal iron; 3, plentifulron; 4, excessive iron. Films were also exam-ned for megaloblastic change (change classified asegaloblastic, masked megaloblastic, mild maskedegaloblastic and partly megaloblastic), and forhanges of chronic inflammatory disease (classifieds present, absent, or partial iron block).Data were analysed with Stata 6.0 (Stata Cor-

oration, College Station, TX, USA). Categoricalariables were compared with �2 or Fisher’s exactest. Odds ratios (OR) were calculated and adjustedAOR) determined in a logistic regression model in-luding predefined potential confounding factors.

IDS surveillance (WHO, 1994). No patients wereaking antiretroviral drugs. While in hospital, 2625%) of patients died (28% HIV-positive and 14%IV-negative patients).One-third of patients had TB, often diagnosed in

everal ways (Table 1). Those with TB were moreikely to be HIV-positive (38% of HIV-positive and4% of HIV-negative patients, OR 3.6, 95% CI 1—13,= 0.05; AOR 3.2). In eight patients, where it wasossible to repeat haemoglobin 1 month after start-ng anti-TB treatment, haemoglobin rose in all ex-ept one, by a mean of 2.1 g/dl. Sixteen patientsad positive mycobacterial marrow culture; seven8% of HIV-positive patients), all with clinical AIDSr CD4+ count <200, had TB diagnosed only by bonearrow aspirate. In three, TB was not suspected.our of these patients had CD4+ counts <60. Weound no fungi in marrow aspirates.Twenty-two patients had positive blood cul-

ures, and all but two of these were HIV-positiveP = 0.124) (Table 2). Twelve of these had theame organism isolated from their marrow aspi-ate, most commonly non-typhi salmonella. One pa-ient’s marrow culture grew non-typhi salmonellaespite a negative blood culture; another had Es-herichia coli in the blood but Cryptococcus neo-ormans in marrow and cerebrospinal fluid. Thoseatients with documented fever (axillary tempera-ure ≥37.4 ◦C on admission or enrolment) were nottatistically more likely than others to have bacter-emia (OR 1.34, 95% CI 0.5—3.5). Most patients with

564 D.K. Lewis et al.

Table 1 Methods of diagnosing TB in HIV-positive and HIV-negative anaemic patients

Diagnostic method HIV− (% of 21) HIV+ (% of 77) Total (% of 98)

Sputum positive for acid-fast bacilli 2 (10) 8 (10) 10 (10)Marrow culture 1 (5) 15 (19) 16 (16)Lymph node aspirate positive for acid-fast bacilli 0 4 (5) 4 (4)Lymphocytic pleural effusion 1 (5) 3 (4) 4 (4)Pericardial effusion 0 2 (3) 2 (2)

Chest X-raya

Miliary 1 (5) 8 (10) 9 (9)Classical pattern 4 (19) 18 (23) 22 (22)Classical pattern ‘probable’ 0 1 (1) 1 (1)Non-classical pattern 0 8 (10) 8 (8)Non-classical pattern ‘probable’ 0 7 (9) 7 (7)

Total definite TB 3 (14) 29 (38) 32 (33)Total probable TB 0 4 (5) 4 (4)Total definite plus probable TB 3 (14) 33 (43) 36 (37)

Note that many patients were diagnosed by more than one method. Three patients already taking anti-TB treatment, and fourwho died before undergoing bone marrow and chest X-ray, were excluded from the analysis.a ‘Classical pattern’ defined as pleural effusion, hilar lymphadenoapthy, cavities, upper lobe fibrosis or retraction atelectasis, orbilateral upper lobe infiltrates; ‘non-classical pattern’ defined as diffuse or middle plus lower zone infiltrations.

salmonella bacteraemia (7 of 11) had a palpablespleen, and splenomegaly was strongly associatedwith salmonella infection (OR 3.9, 95% CI 1.06—14,P = 0.04; AOR 3.4).

Hookworm infection was not common in HIV-positive anaemic patients. It was significantly more

common in HIV-negative patients, especially heavyinfection likely to cause severe anaemia, usinga scoring system with a cut-off correspondingto roughly 300 eggs/g (OR for heavy infection ifHIV-negative 38.4, 95% CI 4—353, P = 0.001; AOR95, 95% CI 3—3359, P = 0.012). Malaria parasites

Table 2 Bloodstream and stool infections in HIV-positive and HIV-negative anaemic patients

Organism HIV− (% of 22) HIV+ (% of 83) Total (% of 105) P for difference betweenHIV+ and HIV−

Non-typhi salmonella 0 11 (13) 11 (10) 0.06Escherichia coli 1 (5) 6 (7) 7 (7)Streptococcus pneumoniae 0 1 (1) 1 (1)Klebsiella pneumoniae 0 1 (1) 1 (1)�-haemolytic streptococcus 0 1 (1) 1 (1)Streptococcus agalactiae 1 (5) 0 1 (1)Malaria parasitesa 3 (14) 12 (14) 15 (14)Hookworm infectionb 6 (38) 9 (14) 15 (19) 0.03Heavy hookworm infectionc 6 (38) 1 (2) 7 (9) <0.001

TotalsBacteraemia 2 (9) 20 (24) 22 (21) 0.12Tuberculosis 3 (14) 33 (43) 36 (37) 0.02Parasitic cause of anaemiad 7 (32) 7 (8) 14 (13) 0.009Any infectious cause of anaemia 11 (50) 52 (63) 63 (60) 0.28

6 (67

showtients1) anpositi

Any treatable cause of anaemiae 16 (64) 5a Classified as 7 ‘1+’, 6 ‘2+’, and 2 ‘4+’. In addition, one filmb Other stool findings included Strongyloides stercoralis (4 pasoma mansoni (2), Entamoeba coli (2), Entamoeba histolytica (patients who provided stool samples: 16 HIV-negative, 65 HIV-c Heavy hookworm infection approximates to >200 eggs/g.d

Includes malaria only if more than scanty parasitaemia, and hooke Includes TB, bacteraemia, iron deficiency and B12 deficiency.

) 70 (67)

ed gametocytes only., all HIV-positive), Ascaris lumbricoides (3 patients), Schisto-d Giardia lamblia (1). Percentages for hookworm are of thoseve and 81 in total.

worm only if heavy infection.

Causes of severe anaemia in Malawi 565

Table 3 Bone marrow histology in HIV-positive and HIV-negative anaemic patients

HIV− n (%) HIV+ n (%) Valid samples P for difference betweenHIV+ and HIV−

Iron deficiency 10 (59) 10 (16) 81 <0.001Iron excess 2 (12) 12 (19) 81 0.5Chronic inflammatory diseasea 4 (25) 29 (51) 73 0.06Megaloblastic change: yes 3 (17) 8 (12) 83 0.59Masked 6 (33) 12 (19) 83Mild masked 0 2 (3) 83Partial 1 (6) 4 (6) 83

a Chronic inflammatory disease excluded four patients with partial iron block.

were found in similar proportions in HIV-positiveand -negative cases. Eight patients had significantmalaria parasitaemia (more than scanty parasitesseen), but half of these had other significant mor-bidity (two bacteraemia, one pneumonia and onepericardial TB).

Four patients were vitamin B12-deficient (twoHIV-positive, two -negative). One had a high meancell volume (MCV), macrocytic blood film, pancy-topaenia and the clinical features of subacute com-bined degeneration of the cord. In the others nospecific clinical diagnosis was made, nor clinicalcause for anaemia found; one had microcytosis withconfirmed iron deficiency but masked megaloblas-tic change, while the other two had normal MCVsbut megaloblastic bone marrow.

Table 3 shows bone marrow findings. In HIV-positive patients iron excess was more commonthan iron deficiency. Half the HIV-positive patientshad chronic inflammatory disease. Iron deficiencywas more common in HIV-negative than HIV-positivepatients (OR 7.7, 95% CI 2.4—25, P = 0.001; AOR 9.9,95% CI 2.2—45, P = 0.003).

4

Chtsittpomaewci

no symptoms to suggest this infection, and in somethe diagnosis was only made on bone marrow cul-ture. Community-based studies might find a differ-ent pattern of co-morbidity, and the results of thisstudy should not be taken to represent the causesof anaemia at a community level. We aimed togive the hospital clinician some idea of likely treat-able causes when faced with this common, life-threatening, but potentially treatable problem, andlimited diagnostic facilities.

TB was the commonest treatable condition wefound; one-third of patients had a firm diagnosisof TB, and in some of them haemoglobin rose dra-matically with treatment. Other studies have showndisseminated TB to be the prime cause of death inHIV-positive patients (Lucas et al., 1993). Anaemiamay be a stronger predictor than HIV of pulmonaryTB and mycobacteraemia (Lewis et al., 2002; Lucaset al., 1993). Most TB was diagnosed using routinemethods, but in a few cases (8% of HIV-positive pa-tients) bonemarrow culture was needed. This raisesthe possibility that occasionally a trial of treat-ment for TBmay be justified in HIV-positive severelyanaemic patients when other causes of anaemia areruled out. In Brazil, 7% of patients with AIDS, fever

. Discussion

urrent syndromic treatment for anaemia in mostospitals in Africa reflects historical data, and inhe small number of HIV-negative patients in thistudy significant hookworm infection was foundn over one-third. In this study of hospital inpa-ients, however, the great majority had HIV infec-ion with low CD4+ counts. This partly reflects theopulation admitted to medical wards. The causef anaemia in HIV-positive individuals is usuallyultifactorial, with both nutritional deficienciesnd infections contributing (Costello, 1988; Morrist al., 1989; Sullivan et al., 1998). Nevertheless,e found a potentially treatable infection, whichould cause anaemia, in over half of those stud-ed; in a significant number of these there were

and anaemia had TB in their bone marrow, com-pared with 19% of our HIV-positive anaemic patients(Barreto et al., 1993).

On our wards blood is routinely cultured fromunwell, febrile patients, and 16% are bacteraemic(Gordon et al., 2001). Bloodstream infection wasmore common than this (21%) in patients in ourstudy. These infections were predominantly of non-typhi salmonella and E. coli, but by contrast toHIV-positive febrile patients, few grew pneumo-cocci. In children, non-typhi salmonella is also asso-ciated with severe anaemia, and we have repeatedthe finding that splenomegaly is common in adultsand children with salmonella bloodstream infec-tion (Gordon et al., 2002; Graham et al., 2000).The clinical importance of this is that splenomegalyand anaemia are often assumed to suggest malaria.

566 D.K. Lewis et al.

Although malaria is associated with anaemia inAfrican children and pregnant women, it did notseem to be an important cause in our patients(Guyatt and Snow, 2001; Rogerson et al., 2000;Stoltzfus et al., 2000). In January 2000 (the sameseason as this study), 15% of surgical inpatientshad parasitaemia, which is similar to the preva-lence in this study (unpublished data) and probablyrepresents the background prevalence in otherwisehealthy adults in this catchment area.

Four patients in our study had vitamin B12 de-ficiency, of whom one had macrocytosis and grossneurological abnormalities. In Zimbabwe B12 defi-ciency caused 86% of megaloblastic anaemia, andmost (70%) had neurological dysfunction (Savageet al., 1994). It is concerning that three of our pa-tients, despite megaloblastic change in their mar-row, had B12 deficiency with normal or low MCVsand no neurological dysfunction; we do not knowhow their deficiency could have been detectedwithout measuring B12 levels. We were not able tomeasure folate levels, but in the Zimbabwe studyisolated folate deficiency was rare.

We found that iron deficiency and hookworm in-fection were common only in HIV-negative patients.

tive. Management should include investigation forTB, and consideration of antibiotics active againstenterobacteria.

Conflicts of interest statementThe authors have no conflicts of interest concerningthe work reported in this paper.

Acknowledgements

We would like to thank the patients who agreed totake part in this study, and Freda Nsamala and thestaff in the Wellcome Trust Research laboratories.Funding for this project came from the National Tu-berculosis Programme of Malawi.

References

Barreto, J.A., Palaci, M., Ferrazoli, L., Martins, M.C., Suleiman,J., Lorenco, R., Ferreira Jr., O.C., Riley, L.W., Johnson Jr.,W.D., Galvao, P.A., 1993. Isolation of Mycobacterium aviumcomplex from bone marrow aspirates of AIDS patients inBrazil. J. Infect. Dis. 168, 777—779.

Boelaert, J.R., Weinberg, G.A., Weinberg, E.D., 1996. Altered

C

E

G

G

G

G

G

K

L

In other studies, anaemia is mainly associated withheavy hookworm infection and is often mild withhaemoglobin levels above 9 g/dl (Latham et al.,1983; Olsen et al., 1998). In the anaemia of chronicdisease, serum iron levels may fall while iron accu-mulates in bone marrow, and iron stores usually risewith HIV progression (Boelaert et al., 1996; Kentet al., 1994). In settings with a high prevalence ofHIV and infectious disease, routinely treating non-pregnant adults who have severe anaemia with ironsupplements may have little impact unless the un-derlying infection is treated.

5. Conclusions

This study set out to find factors associated withanaemia in adults admitted to hospital where HIVis common. In this study most adults admittedto medical wards with severe anaemia had HIV,but two-thirds had a treatable condition proba-bly contributing to their anaemia, most commonlyTB or bloodstream infection. Intervention stud-ies would be needed to prove definitively thattreating these reduced morbidity, but it is likelythat treating these infections would have a posi-tive impact. Iron deficiency and hookworm infec-tion were rare, except in HIV-negative patients;current syndromic treatment with only antimalari-als, anthelmintics and iron is unlikely to be effec-

iron metabolism in HIV infection: mechanisms, possible con-sequences, and proposals for management. Infect. AgentsDis. 5, 36—46.

ostello, C., 1988. Haematological abnormalities in human im-munodeficiency virus (HIV). J. Clin. Pathol. 41, 711—715.

lliott, A.M., Halwiindi, B., Hayes, R.J., Luo, N., Mwinga, A.G.,Tembo, G., Machiels, L., Steenbergen, G., Pobee, J.O.M.,Nunn, P., McAdam, K.P.W.J., 1995. The impact of human im-munodeficiency virus on mortality of patients treated for tu-berculosis in a cohort study in Zambia. Trans. R. Soc. Trop.Med. Hyg. 89, 78—82.

elfand, M., 1968. A clinical study of severe anaemia admittedto medical wards, Harare hospital. Rhodesia. J. Trop. Med.Hyg. 71, 118—121.

ordon, M.A., Walsh, A.L., Chaponda, M., Soko, D., Mbwinji, M.,Molyneux, M.E., Gordon, S.B., 2001. Bacteraemia and mor-tality among adult medical admissions in Malawi - predomi-nance of non-typhi salmonellae and Streptococcus pneumo-niae. J. Infect. 42, 44—49.

ordon, M.A., Banda, H.T., Gondwe, M., Gordon, S.B., Boeree,M.J., Walsh, A.L., Corkill, J.E., Hart, C.A., Gilks, C.F.,Molyneux, M.E., 2002. Non-typhoidal salmonella bacteraemiaamong HIV-infected Malawian adults: high mortality and fre-quent recrudescence. AIDS 16, 1633—1641.

raham, S.M., Walsh, A.L., Molyneux, E.M., Phiri, A.J.,Molyneux, M.E., 2000. Clinical presentation of non-typhoidalSalmonella bacteraemia in Malawian children. Trans. R. Soc.Trop. Med. Hyg. 94, 310—314.

uyatt, H.L., Snow, R.W., 2001. The epidemiology and burdenof Plasmodium falciparum-related anemia among pregnantwomen in sub-Saharan Africa. Am. J. Trop. Med. Hyg. 64 (1—2Suppl.), 36—44.

ent, S., Weinberg, E.D., Stuart-Macadam, P., 1994. The etiol-ogy of the anemia of chronic disease and infection. J. Clin.Epidemiol. 47, 23—33.

atham, M.C., Stephenson, L.S., Hall, A., Wolgemuth, J.C.,Elliot, T.C., Crompton, D.W.T., 1983. Parasitic infections,

Causes of severe anaemia in Malawi 567

anaemia and nutritional status: a study of their interrelation-ships and the effect of prophylaxis and treatment on workersin Kwale District Kenya. Trans. R. Soc. Trop. Med. Hyg. 77,41—48.

Lewis, D.K., Peters, R.P.H., Schijffelen, M.J., Joaki, G.R.F.,Walsh, A.L., Kublin, J.G., Kumwenda, J., Kampondeni, S.,Molyneux, M.E., Zijlstra, E.E., 2002. Clinical indicators ofmycobacteraemia in adults admitted to hospital in Blantyre.Malawi. Int. J. Tuberc. Lung Dis. 6, 1067—1074.

Lewis, D.K., Callaghan, M., Phiri, K., Chipwete, J., Kublin, J.G.,Borgstein, E., Zijlstra, E.E., 2003. Prevalence and indicatorsof HIV and AIDS among adults admitted to medical and surgi-cal wards in Blantyre. Malawi. Trans. R. Soc. Trop. Med. Hyg.97, 91—96.

Lucas, S.B., Hounnou, A., Peacock, C., Beaumel, A., Djomand,G., N’Gbichi, J.-M., Yeboue, K., Honde, M., Diomande, M.,Giordano, C., Doorly, R., Brattegaard, K., Kestens, L., Smith-wick, R., Kadio, A., Ezani, N., Yapi, A., De Cock, K.M., 1993.Themortality and pathology of HIV infection in a West Africancity. AIDS 7, 1569—1579.

Mocroft, A., Kirk, O., Barton, S.E., Dietrich, M., Proenca, R.,Colebunders, R., Pradier, C., dArminio Monforte, A., Lederg-erber, B., Lundgren, J.D., 1999. Anaemia is an independentpredictive marker for clinical prognosis in HIV-infected pa-tients from across Europe. AIDS 13, 943—950.

Moore, R.D., 1999. Human immunodeficiency virus infection,anemia, and survival. Clin. Infect. Dis. 29, 44—49.

Morris, C.D.W., Bird, A.R., Nell, H., 1989. The haematologicaland biochemical changes in severe pulmonary tuberculosis.Quart. J. Med. 272, 1151—1159.

adults in western Kenya. Trans. R. Soc. Trop. Med. Hyg. 92,643—649.

Rogerson, S.J., van den Broek, N.R., Chaluluka, E., Qongwane,C., Mhango, C.G., Molyneux, M.E., 2000. Malaria and anemiain antenatal women in Blantyre, Malawi: a twelve-month sur-vey. Am. J. Trop. Med. Hyg. 62, 335—340.

Savage, D., Gangaidzo, I., Lindenbaum, J., Kiire, C., Mukiibi,J.M., Moyo, A., Gwanzura, C., Mudenge, B., Bennie, A.,Sitima, J., et al., 1994. Vitamin B12 deficiency is the pri-mary cause of megaloblastic anaemia in Zimbabwe. Br. J.Haematol. 86, 844—850.

Stoltzfus, R., 1997. Rethinking anaemia surveillance. Lancet349, 1764—1766.

Stoltzfus, R.J., Chwaya, H.M., Montresor, A., Albonico, M., Savi-oli, L., Tielsch, J.M., 2000. Malaria, hookworms and recentfever are related to anemia and iron status indicators in 0-to 5-y old Zanzibari children and these relationships changewith age. J. Nutr. 130, 1724—1733.

Sullivan, P.S., Hanson, D.L., Chu, S.Y., Jones, J.L., Ward, J.W.,Adult/Adolescent Spectrum of Disease Group, 1998. Epidemi-ology of anemia in human immunodeficiency virus (HIV)-infected persons: results from the multistate adult and ado-lescent spectrum of HIV disease surveillance project. Blood91, 301—308.

van den Broek, N.R., Letsky, E.A., 2000. Etiology of anemia inpregnancy in south Malawi. Am. J. Clin. Nutr. 72 (Suppl.),247S—256S.

van den Broek, N.R., Rogerson, S.J., Mhango, C.G., Kambala,B., White, S.A., Molyneux, M.E., 2000. Anaemia in pregnancyin southern Malawi: prevalence and risk factors. BJOG 107,

W

Olsen, A., Magnussen, P., Ouma, J.H., Andreassen, J., Friis, H.,

1998. The contribution of hookworm and other parasitic in-fections to haemoglobin and iron status among children and

445—451.HO, 1994. WHO case definitions for AIDS surveillance in adultsand adolescents. Wkly Epidemiol. Rec. 69, 273—275.