The British Journal of Radiology, 80 (2007), 302–306

Comparison of plain chest radiography and high-resolution CT in

human immunodeficiency virus infected patients with commu-
nity-acquired pneumonia: a sub-Saharan Africa study
1 1
K NYAMANDE, MBChB, FCP (SA), MD, U G LALLOO, MD, FCCP, FRCP (London) and 2F VAWDA, FC Rad (SA)

Departments of 1Medicine and 2Radiology, Nelson R Mandela School of Medicine, University of KwaZulu-Natal, Durban,
South Africa

ABSTRACT. The objective of the study was to determine the proportion of patients with
missed lesions on plain chest radiographs compared with high-resolution computed
tomography (HRCT) in 49 human immunodeficiency virus (HIV) infected patients with
community-acquired pneumonia (CAP). Patients underwent plain chest radiography
and HRCT scans of the chest at admission. Microbiological investigations for CAP were
performed. An experienced radiologist, without knowledge of clinical or pathological
data, reported the chest radiographs and HRCT scans. The study group included 26
females and 23 males, aged 18–53 years (mean age 36 years). Organisms were isolated
from 26 patients (53%). In 40 patients (82%), the HRCT scans demonstrated lesions not
visualized on the plain chest radiographs. There was 100% correlation between plain
radiographic and HRCT scan findings in nine cases (18%). Lesions that were not
visualized on the plain radiographs but elucidated on HRCT included: pleural effusion
(n514), ground-glass opacification (n520), pericardial effusion (n58), cavitation (n54),
cysts (n54), bullae (n54), abscess (n51) and pneumothorax (n51). In 20 of 23 cases,
hilar lymphadenopathy, identified on HRCT, was not recognized on plain chest
radiographs. In patients in whom an organism was isolated, a correct HRCT diagnosis of Received 29 March 2006
pulmonary tuberculosis, bacterial pneumonia and Pneumocystis carinii pneumonia Revised 21 July 2006
(PCP) was made in 80%, 84% and 100% of cases, respectively. The proportion of Accepted 15 August 2006
patients with missed lesions on plain chest radiographs in HIV infected patients with
DOI: 10.1259/bjr/15037569
CAP was high. This has important implications for management and prognosis. HRCT
scans correlate well with the microbiological diagnosis when reported by an ’ 2007 The British Institute of
experienced radiologist. Radiology

Sub-Saharan Africa has the largest number of human
immunodeficiency virus (HIV) infected subjects world-
wide. The HIV pandemic in the region is still increasing
and has not yet reached a plateau. The number of HIV
infected patients presenting with pulmonary infections
has also increased dramatically.
The imaging technique of choice in patients with
clinical symptoms and signs of community-acquired
pneumonia (CAP) has traditionally been the chest
radiograph [1–7]. This is largely because chest radio-
graphy is easy to perform, widely accessible, cheap and
associated with low radiation. High-resolution CT
(HRCT) scanning is reserved for the analysis of complex
cases, particularly when the chest radiograph is equivo-
cal with regard to associated central obstruction, cavita-
tion, lymphadenopathy, or empyema [8, 9]. The role of
HRCT, however, is rapidly evolving. In febrile neutro-
penic patients, CT scanning is more sensitive than plain
Address correspondence to: Prof. Umesh Gangaram Lalloo, Internal
Medicine, University of KwaZulu Natal, 719 Umbilo Road,
Congella, Durban, KwaZulu Natal, 4013 South Africa. E-mail:
lalloo@:ukzn.ac.za
This study was funded by a grant from the Medical Research
Council of South Africa.
film in early detection of lung infections [10]. HRCT may
have a role in patients whose chest radiographs are non-
revealing or non-diagnostic [11]. HRCT is helpful in the
differential diagnosis of infectious from non-infectious
acute parenchymal disease in the immunocompetent
patient, but is of limited value in making a specific
diagnosis [12].
There are no studies from sub-Saharan Africa on the
utility and value of HRCT scanning of the chest in HIV
infected patients with CAP. We compared HRCT with
plain chest radiography, and the usefulness of HRCT in
the microbiological diagnosis of PCP, pulmonary tuber-
culosis and bacterial CAP. We hypothesized that, in HIV
infected patients presenting with CAP, clinically impor-
tant lesions may not be evident on plain chest radio-
graphy.
Patient recruitment and methods
The study was performed at King Edward VIII
Hospital, Durban, South Africa. This is a tertiary
teaching hospital of the Nelson R Mandela School of
Medicine. The hospital serves a predominantly black
African population from the townships of Umlazi,

302 The British Journal of Radiology, May 2007

Comparison of plain CXR and HRCT in HIV infected patients with CAP
KwaMashu, Inanda, Clermont and Chesterville. The
ethics committee of Nelson R Mandela School of
Medicine, Durban, South Africa, granted permission to
perform the study. From June 2000 to July 2001, 54
patients were recruited into the study. All the patients
were black Africans. They were randomly selected from
inpatients presenting with symptoms and signs of CAP
with a chest radiograph showing consolidation or an
infiltrate compatible with the diagnosis. None of the
patients was on antiretroviral therapy. Microbiological
tests performed following induction and expectoration of
sputum included Gram stain and culture, and Ziel–
Neelsen stain for acid-fast bacilli. Blood cultures and
urine tests for Legionella pneumophila and Streptococcus
pneumoniae were performed. The PCP immunofluores-
cence test was performed on induced sputum to detect
Pneumocystis jirovecii. CD4 counts were assessed by flow
cytometry.
A HRCT scan of the chest was performed after
admission to the medical ward, consisting of 1.5 mm
collimation sections at 10 mm intervals reconstructed
with a high spatial frequency algorithm. All scans were
performed without intravenous contrast medium at
suspended end-inspiration with the patient in a supine
position. Scans were reviewed at a setting appropriate
for both lung parenchyma and mediastinum. The chest
radiographs and HRCT scans were evaluated by a
radiologist who had no prior knowledge of the aetiology
of the pneumonia, duration of symptoms, severity of
symptoms, degree of dyspnoea, presence or absence of
fever or leukocytosis. The HRCT scans were reported
without the concurrent availability of the chest radio-
graphs.
Evaluated HRCT findings included consolidation,
ground-glass opacification, nodular opacification, pleural
effusions, pericardial effusions, abscess formation, cavita-
tion, bullae, cysts, mediastinal lymphadenopathy (nodes
greater than 1 cm in short axis diameter) and the presence
or absence of a pneumothorax. The radiologist was asked
to record the most likely aetiological diagnosis as bacterial
pneumonia, Mycobacterium tuberculosis, Pneumocystis jirove-
cii pneumonia or any other diagnosis. The chest radio-
graphic and HRCT scan findings were then compared.
Statistical analysis
Two by two tables were used to calculate the
sensitivity, specificity and positive and negative pre-
dictive values of HRCT diagnosis vs the microbiological
diagnosis.
Results
Of the 54 patients recruited into the study 5 were
excluded because they were HIV seronegative. Thus data
from 49 patients were analysed. There were 26 females
and 23 males. The mean age of the patients was 36 years
(range 18–53 years). The CD4 count was deter-
mined in 37 patients. The mean CD4 count was
184 cells ml21 (range 0–1223 cells ml21).
The findings on chest radiography correlated with the
HRCT scan findings in only nine patients (18%). In 40
The British Journal of Radiology, May 2007
patients (82%), abnormalities were missed on the chest
radiograph, with a total of 76 missed lesions. The most
commonly missed abnormalities included ground-glass
opacification (50% of patients) and mediastinal lympha-
denopathy (50% of patients), followed by pleural
effusions (35% of patients) and pericardial effusions
(20% of patients) (Table 1). Analysis of the missed lesions
as a proportion of the total number (n576) revealed that
26% were mediastinal lymphadenopathy, 26% ground-
glass opacification, 18% pleural effusions, 10% pericar-
dial effusions, 5% cavitation, 5% bullae, 5% cysts and 1%
abscess formation and pneumothorax.
Microbiology
There was a total of 31 isolates from 26 patients (53%).
No organisms were isolated in the remainder (Table 2).
The most common organism isolated was Mycobacterium
tuberculosis (32%), followed by Streptococcus pneumoniae
(29%) and PCP (19%). Five patients had polymicrobial
pneumoniae. Other organisms isolated were Escherichia
coli (2), Gram-negative bacilli (2), Klebsiella pneumoniae
(1), Haemophilus influenzae (1) and Staphylococcus pneumo-
niae (1). A correct HRCT diagnosis of pulmonary
tuberculosis, bacterial pneumonia and pneumocystis
carinii pneumonia (PCP) was made in 80%, 84% and
100% of cases, respectively (Table 3). The sensitivities of
HRCT compared with microbiology for the diagnosis of
pulmonary tuberculosis, PCP and bacterial pneumonia
were 80%, 100% and 85%, respectively. However, HRCT
was not specific for any diagnosis compared with
microbiology (Tables 3).
Discussion
Few studies have been performed on the role of HRCT
scanning of the chest in HIV related pulmonary infec-
tions [13–21]. None of these studies have been from sub-
Saharan Africa, which currently has the highest HIV or
acquired immunodeficiency syndrome burden. Most
have focused on PCP [13, 15, 16, 19, 20]. To our
knowledge, only Diehl et al [21] from Germany have
investigated the clinical value of HRCT of the chest in
patients with known HIV infection and acute lung
disease, but the entry criteria were that of a normal or
non-specific chest radiograph. In our study, no patients
with normal chest radiographs were recruited. No
patients with normal chest radiographs had HRCT scans
performed.
Our study has shown that in HIV infected African
patients with CAP, an alarmingly high percentage of
patients (82%) have abnormalities that are not visible on
plain chest radiograph. This study supports the findings
of Guillemi et al [17], who undertook a study to
characterize the frequency of lung lesions in asympto-
matic HIV infected individuals with advanced disease.
All were homosexual males on assessment for initiation
of PCP prophylaxis. The results of their study demon-
strated that as many as 60% of HIV infected patients
have unexpected abnormalities on HRCT at the time of
starting PCP prophylaxis. Syrjala et al [22] compared
HRCT with chest radiography in 47 immunocompetent
303

Table 1. Lesions missed on plain chest radiography
Number of patients
with lesion
Pleural effusions 14
Mediastinal lymphadenopathy 23
Ground-glass opacification 20
Pericardial effusion 8 20
Cavitation 4 10
Cysts 4 10
Bullae 4 10
Abscess formation 1 2.5
Pneumothorax 1 2.5
patients with signs and symptoms of CAP. HRCT
identified all 18 cases apparent on chest radiography as
well as an additional 8 cases.
The superiority of HRCT is well recognized [12, 23–
25]. Superior contrast resolution and cross-sectional
display is achieved. HRCT not only improves character-
ization of parenchymal infections in terms of location
and extent of disease, but also surpasses chest radio-
graphy in the detection of complications [24]. The
present study demonstrated the sensitivity of HRCT in
detailing abnormalities not apparent on chest radio-
graphy in a cohort of HIV infected patients with CAP.
Mediastinal lymphadenopathy, ground-glass opacifica-
tion, pleural effusions and pericardial effusions were
among the most commonly missed lesions on chest
radiographs. Lack of identification of these abnormalities
may impact significantly on patient management and
clinical outcome. The detection of some abnormalities
may expand the options available to the clinician for
obtaining diagnostic specimens for microbiology and
histology. The attending physician had knowledge of
the HRCT findings. The HRCT findings in conjunction
with the clinical features aided the physician in commen-
cing therapy in patients in whom microbiological tests
were negative. The HRCT findings did not direct the
microbiological investigations as most tests were per-
formed within the shortest possible time on admission.
Some abnormalities contribute to significant morbidity
and may increase mortality. Bullae and cysts may rupture
causing a spontaneous pneumothorax. Over time, these
lesions may enlarge, compressing normal lung in patients
already afflicted with the ravages of HIV induced
pulmonary infections such as repeated bacterial pneumo-
nia, PCP and tuberculosis. A small pneumothorax may
rapidly expand causing compression of already diseased
Table 2. Organisms isolated in 26 of the 49 patients
Pathogen Number of isolates %
Mycobacterium tuberculosis 10 32
Streptococcus pneumoniae 8 26
Pneumocystis jirovecii 6 19
Escherichia coli 2 7
Gram-negative bacilli 2 7
Haemophillus influenzae 1 3
Klebsiella pneumoniae 1 3
Staphylococcus aureus 1 3
304
K Nyamande, U G Lalloo and F Vawda
% of total number % of total number Number missed
of patients with of missed lesions on chest
missed lesions (n540) (n576) radiography
35 18 14 (100%)
50 26 20 (87%)
50 26 20 (100%)
10 8 (100%)
5 4 (100%)
5 4 (100%)
5 4 (100%)
1 1 (100%)
1 1 (100%)
lung with consequent demise of a patient. A lung abscess
not treated for the appropriate duration may increase
morbidity and mortality.
The present study reflects the presence of significant
complications detected on HRCT scans in patients with
advanced HIV disease. The mean CD4 count in the
present study group was 184 cells ml21 and the lowest
was zero. Advanced disease, as determined by the CD4
count, is associated with HRCT abnormalities even in
asymptomatic patients [17]. The lower the CD4 count,
the greater the likelihood of both opportunistic and
non-opportunistic pulmonary infections. Co-infections
become common. Coinfections occurred in 20% of
patients (5 out of 26) in the present study.
The sensitivity of HRCT diagnosis compared with the
microbiological diagnosis was 100% for PCP, 85% for
bacterial pneumonia and 80% for tuberculosis. The
present study confirmed that no HRCT or chest radio-
graphic pattern is specific for any infection [7, 26–28].
The average positive predictive values attained in the
present study suggest that HRCT scans are relatively
good for screening in, but not for screening out. The
negative predictive values that were obtained in the
present study suggest that the technique cannot be used
to exclude PCP, tuberculosis, or bacterial CAP.
Radiologists rely on the clinical information provided
to come to a reasonably accurate differential diagnosis.
In the absence of clinical information it is difficult to
distinguish pneumonia and other pathological processes
[29]. Although the radiologist in the present study was
blinded to patients’ clinical information, correlation with
the microbiological diagnosis was good, as shown by the
high sensitivities obtained. In patients who are not
improving clinically, HRCT may assist with the decision
to switch therapy. However, the limitations of HRCT in
the management of HIV associated CAP need to be
recognized. Sub-Saharan countries are poor and under-
resourced. There are financial, technical, human resource
and logistical constraints.
In conclusion, we have shown that a large proportion
of HIV infected African patients presenting with CAP
have clinically important abnormalities on HRCT that
are not apparent on plain chest radiographs. This has
important implications for the management of HIV
associated CAP. When reported by an experienced
radiologist, the HRCT scan diagnosis has good correla-
tion with the microbiological diagnosis. This may be
useful in modifying or switching therapy.
The British Journal of Radiology, May 2007
Comparison of plain CXR and HRCT in HIV infected patients with CAP

Table 3. Radiologist’s HRCT scan diagnosis vs microbiological diagnosis

HRCT Total Number in Number Number with % correlation of Sensitivity Specificity Positive Negative
diagnosis number agreement diagnosed on negative HRCT with predictive predictive
with microbiology microbiology microbiology value value
microbiology but not HRCT
diagnosis

Mycobacterium 17 8 2 7 80 % 80 % 0 53 % 0
tuberculosis
Pneumocystis 14 3 0 11 100 % 100 % 0 21 % 0
pneumonia
Bacterial 20 11 2 7 84 % 85 % 0 61 % 0
pneumonia
HRCT, high-resolution computed tomography.

Acknowledgments 12. Tomiyama N, Muller NL, Johkoh T, Honda O, Mihara N,

We would like to thank Chandrika Vedalankar,
Department of Medicine, Nelson R Mandela School of
Medicine, University of KwaZulu-Natal, for secretarial
assistance; the Department of Microbiology, Nelson R
Mandela School of Medicine, University of KwaZulu-
Natal, for analysis of microbiological specimens; and
Tonya Esterhuizen of Biostatistics for the statistical
analysis.
References
1. Marrie TJ, Durant H, Yates L. Community-acquired
pneumonia requiring hospitalization: 5 year prospective
study. Rev Infect Dis 1989;11:586–9.
2. Fang GD, Fine M, Orloff J, Arisumi D, Yu VL, Kapoor W,
et al. New and emerging aetiologies for community-
acquired pneumonia with implications for therapy: a
prospective multicenter study of 359 cases. Medicine
(Baltimore) 1990;69:307–16.
3. Niederman MS, Bass JB, Campbell GD, et al. Guidelines for
the initial management of adults with community-acquired
pneumonia: diagnosis, assessment of severity and initial
antimicrobial therapy. Am Rev Respir Dis 1993;148:1418–26.
4. Macfarlande J. An overview of community-acquired pneu-
monia with lessons learned from the British Thoracic
Society study. Semin Respir Infect 1994;9:153–65.
5. Bartlett JG, Mundy LM. Community-acquired pneumonia.
N Engl J Med 1995;333:1618–24.
6. Albaum MN, Hill LC, Murphy M, Li YH, Fuhrman CR,
Britton CA, et al. Interobserver reliability of the chest
radiograph in community-acquired pneumonia. Chest
1996;110:343–50.
7. Katz DS, Leung AN. Radiology of pneumonia. Clin Chest
Med 1999;20:549–62.
8. Diefenthal HC, Tashjian J. The role of plain films, CT
tomography, ultrasound and percutaneous needle aspira-
tion in the diagnosis of inflammatory lung disease. Semin
Respir Infect 1988;3:83.
9. Sagel SS, Slone RM. Lung. In: Lee JKT, Sagel SS, Stanley RJ,
Heiken JP, editors. Computed body tomography with MRI
correlation, 3rd edition. Philadelphia, PA: Lippincott-
Raven,
1998:442.
10. Heussel CP, Kauczor HU, Heussel G, Fischer B,
Mildenberger P, Thelen M. Early detection of pneumonia
in febrile neutropaenic patients: use of thin section CT. AJR
Am J Roentgenol 1997;169:1347.
11. Boiselle PM, Crans CA, Kaplan MA. The changing face of
Pneumocystis carinii pneumonia in AIDS patients. AJR Am J
Roentgenol 1999;172:1301–9.
Kozuka T, et al. Acute parenchymal lung disease in
immunocompetent patients: diagnostic accuracy of high
resolution CT. AJR Am J Roentgenol 2000;174:1745–50.
13. Hildago A, Falco V, Mauleon S, Andreu J, Crespo M, Ribera E,
et al. Accuracy of high resolution CT in distinguishing between
Pneumocystis carinii pneumonia and non-Pneumocystis carinii
pneumonia in AIDS patients. Eur Radiol 2003;13:1179–84.
14. Becciolini V, Gudinchet F, Cheseaux JJ, Schnyder P.
Lymphocytic interstitial pneumonia in children with AIDS:
high resolution CT findings. Eur Radiol 2001;11:1015–20.
15. Richards PJ, Riddell L, Reznek RH, Armstrong P, Pinching
AJ, Parkin JM. High resolution computed tomography in
HIV patients with suspected Pneumocystis carinii pneumo-
nia and a normal chest radiograph. Clin Radiol
1996;51:689–93.
16. Bianco R, Arborio G, Mariani P, Pompili GG, Alineri S,
Cornalba GP. Pneumocystis carinii lung infections in AIDS
patients: a study with high-resolution computed tomogra-
phy. Radiol Med 1996;91:370–6.
17. Guillemi SA, Staples CA, Hogg JC, Le AN, Lawson LM,
Schechter MT, et al. Unexpected lung lesions in high
resolution computed tomography (HRCT) among patients
with advanced HIV disease. Eur Respir J 1996;9:33–6.
18. Kauczor HU, Schnutgen M, Fischer B, Schwickert HC,
Hartel S, Schandmand-Fischer S, et al. Pulmonary manifes-
tations in HIV patients: role of thoracic radiography, CT
and HRCT. Rofo 1995;162:282–7.
19. Bergin CJ, Wirth RL, Berry GJ, Castellino RA. Pneumocystis
carinii pneumonia: CT and HRCT observations. J Comput
Assist Tomogr 1990;14:756–9.
20. Lazzari Agli L, Viglietta G, Arlotti M, Ciammarhugi R,
Forlani S, Muratore F. Bullous pneumopathy and sponta-
neous pneumothorax. Atypical radiologic patterns of
Pneumocystis carinii pneumonia. Radiol Med 1993;86:826–32.
21. Diehl SJ, Lehman KJ, Thienel F, Georgi M. Value of high-
resolution CT of the lungs in acute pulmonary symptoms of
patients with HIV infections. Rofo 1997;167:227–33.
22. Syrjala H, Broaas M, Suramo I, Ojala A, Lahde S. High
resolution computed tomography for the diagnosis of com-
munity-acquired pneumonia. Clin Infect Dis 1998;27:358–63.
23. Webb WR, Muller NL, Naidich DP. Clinical utility of HRCT
and indications for its use. In: High resolution CT of the
lung, 2nd edition. Philadelphia, PA: Lippincott Raven, 1996:
271–94.
24. Wheeler JH, Fishman EK. Computed tomography in the
management of chest infections: current status. Clin Infect
Dis 1996;23:232–40.
25. Brown MJ, Miller RR, Muller NL. Acute lung disease in the
immunocompromised host: CT and pathologic findings.
Radiology 1994;190:247–54.
26. Farr BM, Kaiser DL, Harrison BD, Connolly CK. Prediction
of microbial aetiology at admission to hospital for

The British Journal of Radiology, May 2007 305

 K Nyamande, U G Lalloo and F Vawda

pneumonia from the presenting clinical features. Thorax 28. Tew J, Calenoff L, Berlin BS. Bacterial or nosocomial
1989;44:1031. pneumonia: accuracy of radiographic diagnosis.
27. MacFarlane JT, Miller AC, Roderick Smith WH, Morris AH, Radiology 1977;124:607.
Rose DH. Comparative radiographic features of community- 29. Primack SL, Muller NL. HRCT in acute diffuse lung disease
acquired Legionnaire’s disease, pneumococcal pneumonia, in the immunocompromised patient. Radiol Clin North Am
mycoplasma pneumonia and psittacosis. Thorax 1984;39:28. 1994;321:731–44.

306 The British Journal of Radiology, May 2007