Acta Tropica 83 (2002) 151–158

Morbidity and nutritional impairment in relation to CD4count in a Zambian population with high HIV prevalence

Paul Kelly a,b,c,*, Isaac Zulu a,b, Beatrice Amadi d, Mwansa Munkanta a,Jacqueline Banda a, Laura C. Rodrigues c, David Mabey c, Roger Feldman b,

Michael J.G. Farthing b

a Department of Medicine, Uni�ersity of Zambia School of Medicine, Lusaka, Zambiab Department of Adult and Paediatric Gastroenterology, St Bartholomew’s and Royal London School of Medicine, Turner Street,

London, UKc London School of Hygiene and Tropical Medicine, London, UK

d Department of Paediatrics, Uni�ersity of Zambia School of Medicine, Lusaka, Zambia

Received 30 March 2001; received in revised form 15 February 2002; accepted 14 March 2002

Abstract

The HIV epidemic has greatly increased morbidity in many African cities and severe undernutrition is a prominentfeature of the clinical presentation. However, there is little information about the relationship of morbidity ornutritional status to immune damage at a population level. We report a cross-sectional study of morbidity andnutritional status in relation to CD4 count in an impoverished urban community in Lusaka, Zambia, at enrolmentinto a longitudinal study. Over a 2 month period in 1999, 261 (52%) of 506 adults resident in one area wereinterviewed and examined. Of 186 adults who consented to testing, 33 (51%) of 65 who were HIV seropositivereported symptoms of disease compared to 39 (32%) of 121 who were HIV seronegative (OR 2.2, 95%CI 1.1–4.2;P=0.02). Peripheral blood CD4 counts in HIV seronegative individuals were broadly similar to norms in developedcountries, but 8 (7%) had CD4 counts below 500 cells/�l. Morbidity in HIV seropositive adults was dominated bytuberculosis (n=11), other respiratory infections (5) or persistent diarrhoea (4), and affected individuals had a widerange of CD4 counts. Nutritional impairment was evident in HIV seropositive adults with clinical evidence ofopportunistic infection (OI), not those with asymptomatic HIV infection. Unexpectedly, we also noted that systolicblood pressure was reduced progressively in HIV infection and in those with OI. In conclusion, HIV-relatedmorbidity was dominated by a small number of treatable infectious diseases occurring over a wide range of CD4count. Nutritional impairment was associated with OI. © 2002 Elsevier Science B.V. All rights reserved.

Keywords: HIV; Africa; Nutrition; Morbidity; CD4 count; Blood pressure; Public health

www.parasitology-online.com

1. Introduction

The HIV epidemic in sub-Saharan Africa hashad its greatest impact in cities, but few studies

* Corresponding author. Tel.: +44-20-7882-7191; fax: +44-20-7882-7192.

E-mail address: m.p.kelly@qmul.ac.uk (P. Kelly).

0001-706X/02/$ - see front matter © 2002 Elsevier Science B.V. All rights reserved.

PII: S0 001 -706X(02 )00095 -5

P. Kelly et al. / Acta Tropica 83 (2002) 151–158152

have analysed the burden of morbidity in thegeneral urban population. The large longitudinalstudies of morbidity and mortality in Africanpopulations in the 1960s and 1970s dealt withrural populations and tended to focus on mater-nal and child health (Feachem and Jamison,1991). The information which is available onHIV-related morbidity, from 1983 to date, islargely hospital based (Reeve, 1989; Grant et al.,1997; King et al., 1994) or applies to rural popula-tions (Morgan et al., 1997) and these studies havejustifiably tended to focus on morbid events inlate stage HIV infection.

There are few data on CD4 counts in Africanpopulations, and it is still uncertain whether mor-bidity in HIV infection in Africa occurs earlierthan in Europe or North America. Nutritionalstatus almost certainly has prognostic significancein AIDS in any population, but again few dataare available on the nutritional status of Africanpopulations severely affected by HIV. It remainsto be established, at least in populations withmarginally inadequate food supply, whether nutri-tional impairment begins early in HIV infectionor whether it develops only with the onset ofsecondary infections.

We set out to examine the relationship betweenmorbidity, nutritional impairment and CD4 countin patients with HIV infection in a Zambianurban community which probably exemplifiesmany urban African populations. We also reportunexpected findings relating to blood pressure(BP) in this population.

2. Methods

2.1. Study setting

Misisi is an unplanned residential area in thesouthern part of Lusaka which was illegal until1995. As a consequence, civic amenities are few,housing is of poor quality, and overcrowding isintense. In 1995 we found evidence of high mor-tality rates in adults (Kelly et al., 1998) whichseem to be attributable to the severity of the HIVepidemic: in Lusaka the HIV seroprevalence isbetween 21 and 35% (Fylkesnes et al., 1997) but

this may be an under-estimate (Fylkesnes et al.,1998; Gray et al., 1998). Authorisation was ob-tained to study part of one sector of Misisi fromthe Lusaka Urban District Health ManagementBoard, and approval was obtained from the Re-search Ethics Committees of the University ofZambia (UNZA) and the London School of Hy-giene and Tropical Medicine. The study was con-ducted from March–May 1999.

2.2. Recruitment of adult participants

In order to be able to compare the eligiblepopulation and the population willing to partici-pate in the study, one half of one sector (approx-imately one quarter) of the compound was chosenas apparently representative of the housing condi-tions in the township as a whole. Recruitment ofa cohort within a cluster of closely groupedhouses was felt to be important to maximiseretention in the study over several years of followup. At the outset, a house to house survey wasconducted to include all households in the studyarea, during which the purpose of the study wasexplained, and one respondent per household wasinterviewed to obtain a demographic profile.

Every adult (18 years of age or more) was theninvited to the nearest urban health centre, a 10min walk from the study area, for interview andexamination. At this interview, the purpose of thestudy was explained in more detail and a twopage information sheet read out in one of fourlocal languages. This information included ex-plicit description of the procedures required forthe study and the follow up. The recruitment andconsent process included information about HIVtesting, and participants who consented to inclu-sion in the study were offered the option of HIVtesting (together with CD4 count) with full pre-and post-test counseling, but they were also freenot to have the test. This conforms with thepolicy of the UNZA Research Ethics Committee.

2.3. Inter�iew and examination

To minimise bias, recruitment was conductedthrough the week, in the early evenings and week-ends, and appointments made at suitable times for

P. Kelly et al. / Acta Tropica 83 (2002) 151–158 153

those with work to go to. At the recruitmentinterview, further demographic information wasobtained and participants were asked if they hadany current health problems or any disability.This was followed by routine physical examina-tion including BP measurement, nutritional as-sessment and urine analysis using dipstick testing.Pulmonary tuberculosis (TB) was defined as thepresence of characteristic symptoms and signs andabnormal chest radiograph, with a good responseto antituberculous therapy. Antituberculous ther-apy was commenced only if the patient did notrespond to a course of broad-spectrum antibiotics(amoxycillin/erythromycin).

Nutritional assessment comprised three anthro-pometric measures, height, weight and mid upperarm circumference (MUAC) as described previ-ously (Kelly et al., 1996a), and these parameterswere used in the analysis as height (m), body massindex (BMI; kg/m2) and MUAC (cm).

2.4. HIV antibody tests and CD4 counts

HIV antibody testing was performed using arapid test (Capillus, Trinity Biotech, Ireland) andan ELISA (Recombigen HIV 1+2 EIA), andresults declared positive only if both were posi-tive. Discrepant tests were resolved by Westernblot (New LAVBLOT1, Sanofi Diagnostics). Pe-ripheral blood CD4 counts were carried out usinga FACSCount (Becton Dickinson) flow cytomet-ric assay according to the manufacturer’sinstructions.

2.5. Statistical analysis

To analyse the influence of HIV infection onBP and nutritional status, three groups were cre-ated: HIV seronegative, HIV seropositive withoutclinical evidence of opportunistic infection (OI)and HIV seropositive with clinical evidence of OI.The latter group included patients with TB, per-sistent diarrhoea, Kaposi’s sarcoma, or dissemi-nated dermatosis: one an impetiginisedscabetiform rash and one widespread planarwarts. Hypothesis testing was carried out usingANOVA as the distributions of measuredparameters were approximately normal.

3. Results

In the initial house to house survey, 203 house-holds were interviewed, and 506 adults wererecorded as being resident in this area and there-fore eligible for inclusion. 266 (53%) of theseadults volunteered for inclusion but five laterdecided to withdraw, leaving 261 adults (88 menand 173 women), 52% of all those available forinclusion in that defined area.

3.1. Composition of the cohort and selection bias

In order to determine whether the study cohortwas representative of the eligible population thecohort data were compared with data from thehouse-to-house survey with respect to demo-graphic composition and socio-economic mea-sures (education and occupation), and HIVseroprevalence was compared to previous esti-mates for Lusaka (Fylkesnes et al., 1997). Demo-graphic data from the house-to-house surveyrevealed the expected population structure of anadult African population (Fig. 1). Fig. 1 alsoshows the population structure of those adultswho volunteered to join the study, and demon-strates under-representation of men, especiallyyounger men, in the cohort. Educational achieve-ment and source of income of the individuals inthe study cohort did not differ significantly fromthe eligible population.

3.2. HIV infection and CD4 counts

One hundred and eighty six of the 261 adultsinterviewed consented to HIV testing and of these65 (35% of those tested) were seropositive. Sev-enty five declined testing. CD4 counts were simi-lar to norms in industrialised countries, but eight(7%) of 115 HIV seronegative adults had countsbelow 500/mm3. There was no significant differ-ence in CD4 count between men and women.Predictably, CD4 counts were lower in HIVseropositive than in HIV seronegative adults (Fig.2). HIV seroprevalence was comparable to previ-ous estimates of 22–35% for Lusaka (Fylkesnes etal., 1997).

P. Kelly et al. / Acta Tropica 83 (2002) 151–158154

3.3. Morbidity in the population

Ninety three of those interviewed (34%) re-ported that they had symptoms of illness at thetime of interview. 33 (51%) of 65 HIV seropositiveadults reported symptoms compared to 39 (32%)of 121 HIV seronegative adults (OR 2.2, 95%CI1.1–4.2; P=0.02) or 21 (28%) of 75 who were nottested (OR 2.7, CI 1.2–5.7; P=0.01). The mostfrequently diagnosed clinical conditions in HIVinfected individuals were TB (n=11), upper res-piratory infection (n=5), persistent diarrhoea(n=4), pelvic inflammatory disease (PID; n=4),abdominal pain unexplained despite endoscopy(n=3), unexplained chest pain (n=2), dissemi-nated skin infections (n=2), Kaposi’s sarcoma(n=1), unexplained fever (n=1), pan-uveitis(n=1) and minor complaints (n=3). CD4 countsof less than 200 cells/�l were found in 6/11 indi-viduals with TB, 2/3 with persistent diarrhoea(one result missing), 2/4 with PID, one with dis-seminated dermatosis and one with unexplainedfever.

The profile of complaints presented by HIVseronegative adults was different, dominated byabdominal pain (n=8) musculo-skeletal chestpain (n=6), dysmenorrhoea (n=4), PID (n=3),osteoarthritis (n=2), or minor complaints (n=18). Three of those who complained of abdominal

pain had duodenal ulcers seen at endoscopy. Onlyone HIV seronegative patient had TB, manifest asa pleural effusion.

Fourteen cases of TB were discovered includingtwo who did not have HIV tests, five of whichhad previously been diagnosed. Sputum smearexamination was performed in five of the ninenew cases and two were positive; cultures werenot performed. Three had extra-pulmonary dis-ease: one pericardial, one pleural and one peri-toneal. All nine newly diagnosed cases hadabnormal chest radiographs, and all had failed torespond to a combination of amoxycillin anderythromycin.

BP measurements were made on 212 adults (70men, 142 women) on recruitment and overallthese indicate low population values. Mean (SD)systolic BP was 118.3 (17.1) mmHg in men and109.2 (15.5) mmHg in women (P=0.0001). Mean(SD) diastolic BP was 76.2 (15.1) in men and 69.5(13.1) in women (P=0.001). Mean systolic BPwas 114 (16) mmHg in HIV seronegative adults,109 (13) in HIV positive adults and 100 (15) inHIV seropositive adults with OIs (P=0.005).Post-hoc statistical testing revealed that the differ-ence between HIV infected adults without OIsand the HIV seronegative adults was also statisti-cally significant (P�0.05). Diastolic BP did not

Fig. 1. Population structure of study cohort compared to that of eligible population. Populations are represented as the number ofindividuals in each 5 year age group, divided by sex. All individuals recruited are included.

P. Kelly et al. / Acta Tropica 83 (2002) 151–158 155

Fig. 2. Distribution of peripheral blood CD4 counts in adults who were HIV seronegative (shaded bars) or HIV seropositive (openbars). Only adults who consented to testing are included.

differ significantly with HIV status. After stratifi-cation by age, the difference in BP between HIVseropositive and seronegative individuals was stilltrue in adults over 40 years (P=0.04) and ap-proached significance in those under 40 years ofage (P=0.09). Using the WHO definition ofdefinite hypertension (160/95 mmHg), eight of therecruited adults were hypertensive.

Urinalysis, followed by urine microscopy whereappropriate, revealed four uncomplicated urinarytract infections and three cases of trichomoniasisbut no cases of urinary schistosomiasis. There wasno relation to HIV.

3.4. Nutritional status

Height and BMI were low by comparison withnorms in industrialised countries (Lentner, 1984),suggesting lifelong undernutrition in the popula-tion as a whole. Mean height was 1.58 m inwomen and 1.69 m in men (P�0.0001). MeanBMI was 22.0 kg/m2 in women and 19.9 in men(P�0.0001). MUAC also differed: mean MUACwas 27.3 cm in women and 26.3 in men (P=0.04). BMI and MUAC were reduced in HIVseropositive adults with OIs (Table 1).

4. Discussion

The population studied in this crowded, impov-erished residential area of Lusaka had a highburden of untreated infectious disease, much ofwhich appeared to be HIV-related. These illnesseswere found at all stages of CD4 count, but themajority of those with TB or persistent diarrhoeahad CD4 counts less than 200 cells/�l. Nutritionalimpairment was seen in advanced HIV disease.

The data presented here were derived fromcross-sectional analysis of one impoverished ur-ban population 20 years into its HIV epidemic,but probably reflect the current profile of morbid-ity in much of this region. Eligible residents andrecruited adults were similar in terms of educa-tional achievement and occupation, and the HIVseroprevalence was close to previously reportedpopulation-based estimates. Selection bias is in-evitable in a study of this nature, and young menwere the most under-represented. The under-rep-resentation of younger men is a consequence oftheir relative inaccessibility during daylight hours,a common pattern in population surveys. On thewhole this group does not have heavy familyresponsibilities and these men move around thecity freely looking for employment and social

P. Kelly et al. / Acta Tropica 83 (2002) 151–158156

opportunities. The proportion of women recruitedwas high, probably reflecting their concern toobtain access to health care for themselves andtheir children. These data therefore represent areasonably representative cross-section of healthproblems in an impoverished urban African popu-lation, and the data regarding nutritional status,BP and peripheral blood T cell subsets in HIVseronegative adults approach ‘normal’ values forsuch a population.

Overall the profile of morbidity in this urbanpopulation has many similarities to the pattern ofHIV-related disease seen in hospital (Grant et al.,1997), but there are differences which probablyreflect the relatively small number of adults withlate stage disease. In this study we did not dis-cover any cases of progressive neurological dis-ease, meningitis or oro-pharyngeal candidiasiswith dysphagia. Our findings differ from those inrural Uganda where TB was uncommon (Morganet al., 1997).

The heavy burden of TB currently seen in citiesin sub-Saharan Africa contrasts sharply with datafrom large longitudinal studies conducted in sixrural locations in sub-Saharan Africa before theemergence of the HIV epidemic, in which TB wasnot dominant (Feachem and Jamison, 1991). Ithas been known for over a decade that HIVinfection in Africa strongly predisposes to TB(Elliott et al., 1990; Lucas et al., 1994). Themajority of the cases identified here had not beentreated. A minority of these cases of TB wereconfirmed by sputum smear microscopy and thediagnosis was clinical. Appropriate response toanti-tuberculous chemotherapy was one criterionused for establishing the diagnosis in sputumsmear negative cases, but this was only given after

failure of treatment with a combination ofamoxycillin and erythromycin.

The prevalence of persistent diarrhoea (2.3%)was in line with previous estimates in Lusaka(Kelly et al., 1996b), and underscores the burdenof intestinal infectious disease in the community.We have previously found that the proportion ofpersistent diarrhoea cases attributable to OI ishigh (Kelly et al., 1996b), which is why we haveincluded it as a defining criterion for inclusion inthe OI group. Endoscopy failed to identify acause for the abdominal pain in those HIVseropositive adults who complained of it (Fer-nando et al., 2001).

Our data provide some of the first unselectedpopulation norms for peripheral bloodlymphocyte subsets and indicate that CD4 countsin apparently healthy HIV seronegative adultsliving in poor parts of Lusaka may be slightlylower than would be expected in healthy Eu-ropean adults and 7% had counts below 500cells/�l. Africans living in sub-Saharan Africahave lower neutrophil and higher lymphocytecounts than Europeans (Fleming, 1996). It is un-likely that this difference is related to nutritionalstatus as there was no relationship between CD4count and BMI. Hospital based data from Tanza-nia indicate that low CD4 counts in HIV seroneg-ative adults occur predominantly in men (Urassaet al., 1999) but we found the same proportion ofmen and women in this group. Further work isunder way to establish the degree to which theselow–normal CD4 counts fluctuate over time.

Since the earliest days of the HIV epidemic inAfrica, severe wasting has been prominent as amanifestation of the infection (Serwadda et al.,1985). In our study, nutritional impairment, as

Table 1Nutritional status and HIV infection

HIV seropositive with OI (n=17)HIV seronegative PHIV seropositive (n=48)(n=121)

21.6 (5.1) 18.7 (2.6)Mean (SD) 21.0 (2.7)BMIMedian (IQR) 20.3 (18.7–23.2) 20.8 (19.5–22.5) 19.1 (16.5–21.5) 0.04

24.8 (3.4)Mean (SD)MUAC 27.3 (2.5)27.1 (3.9)26 (25–29)Median (IQR) 27 (25–29) 0.0625 (23–27)

P. Kelly et al. / Acta Tropica 83 (2002) 151–158 157

detected by the simple measures used, was associ-ated with the onset of OI rather than with early(asymptomatic) HIV infection. This is consistentwith evidence from industrialised countries, inwhich weight loss and consequent nutritional im-pairment were associated with infectious compli-cations (Macallan et al., 1995; Grunfeld, 1995).We have already reported that the severity ofnutritional impairment is a major risk factor formortality in patients with persistent diarrhoea(Kelly et al., 1999).The findings we report herewould suggest that, as in developed countries,effective treatment of OI is likely to be importantin preventing or reversing nutritional failure, evenwhen food availability is limited as it is in thepopulation under study (data not shown).

We found unexpectedly that systolic BP waslower in HIV seropositive than HIV seronegativeadults. Although some of these differences may beconfounded by age (as HIV infection was com-moner and BP was lower in those under 40 yearsof age), the differences remained after stratifica-tion by age. A study in North America indicatesthat BP does not exhibit its normal age-relatedincrease in HIV infected adults (Mattana et al.,1999). This may represent a degree of autonomicfailure, but adrenal cortical failure also occurs inHIV infected adults in late stage disease, and hasbeen described in patients with TB (Mugusi et al.,1990). Clearly more work needs to be done toclarify this issue. The prevalence of hypertension,defined using the WHO definition, in this popula-tion (4%) was considerably lower than in urbanCameroon (12–16%, Mbanya et al., 1998) or inurban Nigeria (11%, Kaufman et al., 1996). Thedifference between the sexes has been observedbefore (Kaufman et al., 1997).

These data are among the first normal valuesfor CD4 counts in an African population withhigh HIV seroprevalence. Two major syndromesof HIV-related morbidity were associated withCD4 counts under 200 cells/�l and nutritionalimpairment was associated with the onset of OI.It seems likely that prevention or early treatmentof these major HIV-related syndromes may pre-vent the severe nutritional problems commonlyseen in wake of the epidemic in Africa.

Acknowledgements

We would like to dedicate this paper to thememory of our colleague Emmanuel Kunda whowas a key member of the study team. We aregrateful to Rose Banda, Vera Yambayamba, Coil-lard Kaunga and Samson Mbewe. MaxKatubulushi was responsible for data manage-ment. Dr Grace Chipalo kindly evaluated some ofthe ophthalmological problems seen. Grant sup-port: The Wellcome Trust, London, UK.

References

Elliott, A.M., Luo, N.P., Tembo, G., Halwiindi, B., Steenber-gen, G., Machiels, L., Pobee, J., Nunn, P., Hayes, R.J.,McAdam, K.P.W.J., 1990. Impact of HIV on tuberculosisin Zambia: a cross-sectional study. BMJ 301, 412–415.

Feachem, R.G., Jamison, D.T., 1991. Disease and Mortality insub-Saharan Africa. Oxford University Press for the WorldBank, New York.

Fernando, N., Holton, J., Zulu, I., Vaira, D., Mwaba, P.,Kelly, P., 2001. Helicobacter pylori infection in an urbanZambian population. J. Clin. Microbiol. 39, 1323–1327.

Fleming, A., 1996. Haematological problems in the tropics:reference ranges. In: Cook, G.C. (Ed.), Manson’s TropicalDiseases, 20th ed. Saunders, London, pp. 101–107.

Fylkesnes, K., Musonda, R.M., Kasumba, K., Ndhlovu, Z.,Mluanda, F., Kaetano, L., Chipaila, C.C., 1997. The HIVepidemic in Zambia: socio-demographic prevalence pat-terns and indications of trends among child-bearingwomen. AIDS 11, 339–345.

Fylkesnes, K., Ndhlovu, Z., Kasumba, K., Musonda, R.M.,Sichone, M., 1998. Studying dynamics of the HIV epi-demic: population-based data compared with sentinelsurveillance in Zambia. AIDS 12, 1227–1234.

Grant, A.D., Djomand, G., De Cock, K.M., 1997. Naturalhistory and spectrum of disease in adults with HIV/AIDSin Africa. AIDS 11 (Suppl. B), S43–S54.

Gray, R.H., Wawer, M.J., Serwadda, D., Sewankambo, N.,Li, C., Wabwire-Mangen, F., Paxton, L., Kiwanuka, N.,Konde-Lule, J., Quinn, T.C., Gaydos, C.A., McNairn, D.,1998. Population-based study of fertility in women withHIV-1 infection in Uganda. Lancet 351, 98–103.

Grunfeld, C., 1995. What causes wasting in AIDS? N. Engl. J.Med. 333, 123–124.

Kaufman, J.S., Owoaje, E.E., James, S.A., Rotimi, C.N.,Cooper, R.S., 1996. Determinants of hypertension in WestAfrica: contribution of anthropometric and dietary factorsto urban-rural and socioeconomic gradients. Am. J. Epi-demiol. 143, 1203–1218.

Kaufman, J.S., Asuzu, M.C., Mufunda, J., Forrester, T.,Wilks, R., Luke, A., Long, A.E., Cooper, R.S., 1997.

P. Kelly et al. / Acta Tropica 83 (2002) 151–158158

Relationship between blood pressure and body mass indexin lean populations. Hypertension 30, 1511–1516.

Kelly, P., Summerbell, C., Ngwenya, B., Mandanda, B., Hosp,M., Fuchs, D., Wachter, H., Luo, N.P., Pobee, J.O.M.,Farthing, M.J.G., 1996a. Systemic immune activation as apotential determinant of wasting in Zambians with HIV-re-lated diarrhoea. Q. J. Med. 89, 831–837.

Kelly, P., Baboo, K.S., Woolf, M., Ngwenya, B., Luo, N.,Farthing, M.J.G., 1996b. Prevalence and aetiology of per-sistent diarrhoea in adults in urban Zambia. Acta Trop.61, 183–190.

Kelly, P., Feldman, R., Ndubani, P., Baboo, K.S., Timaeus, I.,Farthing, M.J.G., Wallman, S., 1998. High adult mortalityin Lusaka, Zambia. Lancet 351, 883.

Kelly, P., Musonda, R., Kafwembe, E., Kaetano, L., Keane,E., Farthing, M.J.G., 1999. Micronutrient supplementationfor the AIDS diarrhoea-wasting syndrome in Zambia: arandomised controlled trial. AIDS 13, 495–500.

King, R., Fox, E., Twagirukristu, J.B., Karita, E., Allen, S.,1994. Excess morbidity related to HIV infection. Trans.Roy. Soc. Trop. Med. Hyg. 88, 295.

Lentner, C., 1984. Geigy Scientific Tables, vol. 3, eighth ed.Ciba-Geigy, Basle, pp. 297–328.

Lucas, S.B., DeCock, K.M., Hounnou, A., Paecock, C.,Diomanda, M., Honde, M., Beaumel, A., Kestens, L.,Kadio, A., 1994. Contribution of tuberculosis to slimdisease in Africa. BMJ 308, 1531–1533.

Macallan, D., Noble, C., Baldwin, C., Jebb, S.A., Prentice,A.M., Coward, A., Sawyer, M.B., McManus, T.J., Griffin,

G.E., 1995. Energy expenditure and wasting in HIV infec-tion. N. Engl. J. Med. 333, 83–88.

Mattana, J., Siegal, F.P., Sankaran, R.T., Singhal, P.C., 1999.Absence of age-related increase in systolic blood pressurein ambulatory patients with HIV infection. Am. J. Med.Sci. 317, 232–237.

Mbanya, J.C., Minkoulou, E.M., Salah, J.N., Balkau, B.,1998. The prevalence of hypertension in rural and urbanCameroon. Int. J. Epidemiol. 27, 181–185.

Morgan, D., Maude, G.H., Malamba, S.S., Okongo, M.J.,Wagner, H.-U., Mulder, D., Whitworth, J.A., 1997. HIV-1disease progression and AIDS-defining disorders in ruralUganda. Lancet 350, 245–250.

Mugusi, F., Swai, A.B., Turner, S.J., Alberti, K.G., McLarty,D.G., 1990. Hypoadrenalism in patients with pulmonarytuberculosis in Tanzania: an undiagnosed complication?Trans. Roy. Soc. Trop. Med. Hyg. 84, 849–851.

Reeve, P.A., 1989. HIV infection in patients admitted to ageneral hospital in Malawi. BMJ 298, 1567–1568.

Serwadda, D., Mugerwa, R.D., Sewankambo, N.K., Lwegaba,A., Carswell, J., Kirya, G.B., 1985. Slim disease: a newdisease in Uganda and its association with HTLVIII infec-tion. Lancet ii, 849–852.

Urassa, W., Mbena, E., Mhalu, F., Biberfeld, G., 1999. Refer-ence values of lymphocyte subsets and natural killer cellsamong healthy HIV seronegative adults in Dar es Salaam.Abstract 13PT52-1, XI International Conference on AIDSand STDs in Africa.