52 AJTCCM VOL. 31 NO. 2 2025
ORIGINAL RESEARCH: ARTICLES
Background. Community-acquired pneumonia (CAP) remains an important cause of morbidity and mortality in people with HIV (PWH),
and antimicrobial resistance (AMR) leads to poor treatment outcomes. Better tests are required to overcome the low sensitivity of sputum
Gram stain and culture for pneumonia diagnosis. Molecular diagnostic tests rapidly detect respiratory pathogens and markers of AMR, but
few studies have examined their role in PWH.
Objectives. To investigate the additional yield of the Biore FilmArray Pneumonia Panel plus (FilmArrayPN-PCR), an automated nested
multiplex polymerase chain reaction system, over culture for diagnosis of CAP, and determine clinical predictors of AMR in PWH.
Methods. We enrolled adult PWH hospitalised with cough <2months in a prospective cohort in Kampala, Uganda. Participants provided
expectorated sputum samples for testing by FilmArrayPN-PCR and culture. We performed drug susceptibility testing of cultured sputum
isolates and detection of genetic markers of AMR on sputum by FilmArrayPN-PCR.
Results. e 107 participants enrolled had a median (interquartile range) age of 40 (31-46)years, 50.5% (n=54/107) were female, and
74.8% (n=80/107) had recent antibiotic use. e median duration of cough was 3 (1-4)weeks. FilmArrayPN-PCR increased the detection
of respiratory pathogens by 64.5% (95% condence interval (CI) 54.8-73.1; p<0.001) and detected AMR in 25.2% (n=27/107). Baseline
room air oxygen saturation <92% (adjusted odds ratio (aOR) 9.20; 95% CI 2.52-33.57; p=0.001) and prior antibiotic use (aOR 4.14; 95%
CI 1.04-16.51; p=0.04) were independent predictors of AMR.
Conclusion. FilmArrayPN-PCR increased the diagnostic yield of pathogens, and a low baseline oxygen saturation (<92%) and prior
antibiotic use were associated with an increased risk of AMR in hospitalised PWH with CAP.
Keywords. FilmArray, pneumonia, hospitalised, HIV, antimicrobial resistance.
Afr J Thoracic Crit Care Med 2025;31(2):e2415. https://doi.org/10.7196/AJTCCM.2025.v31i2.2415
Molecular diagnostics improve the yield of diagnosis of
community-acquired pneumonia and multidrug-resistant
pathogens in hospitalised patients with HIV in a low-income
setting
W Worodria,1,2,3,4 MB ChB, MMed (Int Med), MSc, PhD ; A Andama,2,4 BBLT, MSc, PhD ; I Sanyu,2 BSc, MPH ;
D Orit,4 MB ChB, MMed (Int Med) ; R Kwizera,3 MSc, PhD ; A Sessolo,2 MB ChB, MSc ; P Byanyima,2 BBLT, MHI ;
J Zawedde,2 BSc, MPH; S Kaswabuli,2 BBLT, MSc ; E Mande,3 BSc ; C Mukashyaka,3 BSc ;
F Semitala,2,4 MB ChB, MMed (Int Med), MPH ; A Cattamanchi,5 MD ; D R Boulware,6 MD ; L Huang,7 MD
1 Department of Medicine, Mulago National Referral Hospital, Kampala, Uganda
2 Infectious Disease Research Collaboration, Kampala, Uganda
3 Infectious Disease Institute, Kampala, Uganda
4 Department of Medicine, School of Medicine, Makerere College of Health Sciences, Kampala, Uganda
5 Department of Medicine, University of California Irvine, Calif., USA
6 Department of Medicine, University of Minnesota, Minneapolis, Minn., USA
7 Department of Medicine, University of California, San Francisco, Calif., USA
Corresponding author: W Worodria (worodria@yahoo.com)
Study synopsis
What the study adds. e Biore FilmArray Pneumonia Panel plus detected 64.5% more respiratory pathogens compared with culture, and
detected antimicrobial resistance (AMR) genes in 25.2% of patients with HIV hospitalised with community-acquired pneumonia (CAP).
Baseline room air oxygen saturation <92% and prior antibiotic use were associated with nine times and four times increased odds of AMR,
respectively.
Implications of the ndings. Multiplex polymerase chain reaction (PCR) assays increase the speed of detection and diagnostic yield of
respiratory pathogens and may be useful for diagnosis of AMR in hospitalised patients with HIV and CAP. e clinical implications of
these ndings should be evaluated further in prospective studies and cost-eectiveness studies to dene the role of multiplex PCR tests in
the patient care pathway.
AJTCCM VOL. 31 NO. 2 2025 53
ORIGINAL RESEARCH: ARTICLES
Community-acquired pneumonia (CAP) is a common cause of
morbidity and mortality in people living with HIV (PWH), especially
those with advanced immune suppression.[1] The reduction in the
frequency of pneumonia in PWH on antiretroviral therapy (ART)
does not decline to levels in HIV-seronegative individuals.[2] As a
result, PWH remain susceptible to pneumonia from a wide range of
pathogens, which may need specic treatment.[3] is predisposition of
PWH to a broad spectrum of lung infections underscores the need for
a specic aetiological diagnosis.
Aetiological diagnosis of CAP remains a challenge because sputum
Gram stain is nonspecic, and conventional culture-based techniques
that can provide a specic diagnosis have low sensitivity and are not
universally available.[4] e long turnaround time for sputum culture
makes it impractical for initial antibiotic selection.[5] e sensitivity of
culture-based tests is further compromised by prior antibiotic use,[6] yet
early initiation of antibiotics is vital to prevent clinical deterioration of
patients with severe CAP.[7] Furthermore, the infrastructure and capacity
to perform sputum culture and drug susceptibility testing are limited
in low-income settings. In Uganda, a low-income country with a high
HIV burden, the last comprehensive hospital-based study on causes of
CAP in PWH was published in 2001.[8] e epidemiology of CAP may
have changed signicantly, and updated studies are urgently needed. In
addition, the increased emergence of antimicrobial resistance (AMR),
which is estimated to cause 700 000 deaths globally peryear, is a major
public health challenge contributing to increased healthcare costs.[9] In
many low-income settings where the vast majority of PWH reside, there
are limited data on the epidemiology of CAP and AMR in PWH.
Molecular tests, such as the polymerase chain reaction (PCR),
show promise in detecting pathogens causing CAP, as the PCR is less
affected than culture by prior antibiotic use. In addition, the PCR
identifies specific mutations associated with AMR. Multiplex PCR
panels designed for pneumonia can diagnose several bacterial and
viral pathogens simultaneously, leading to rapid diagnosis of pathogens
causing pneumonia[10] that can be used to triage hospitalised patients,
and can also identify AMR to optimise antibiotic selection. We therefore
investigated the additional yield of the Biore FilmArray Pneumonia
Panel plus (FilmArrayPN-PCR) (bioMérieux, France) over culture-
based methods for detection of lower respiratory tract pathogens and
AMR in PWH, and determined the clinical predictors of AMR in
hospitalised patients with CAP.
Methods
Study design and participants
We conducted a prospective cohort study at two referral hospitals in
Kampala, Uganda (Kiruddu Hospital and China-Uganda Friendship
Hospital), from 7 May 2019 to 22 February 2021. We performed laboratory
testing at the Infectious Disease Institute Translational Laboratory and the
Makerere University Medical Microbiology Laboratory.
Adult patients who were hospitalised with clinician-diagnosed CAP,
had provided written informed consent, were PWH, and had cough
<2months were enrolled. Patients were excluded if they were unable
to consent, had been hospitalised during the previous 4weeks, had
no chest radiograph to conrm radiographic pneumonia, could not
produce sputum, or had a conrmed diagnosis of tuberculosis.
We administered a standard questionnaire to participants to obtain
demographic information and a clinical history regarding their
respiratory symptoms, risk factors for pneumonia, history of prior
antibiotic use, and presence of comorbidities. A physical examination
was conducted, and vital signs were measured and recorded. Oxygen
saturation on room air was measured using pulse oximetry. Empirical
antibiotic treatment was provided by the clinicians based on existing
clinical care guidelines provided by the Ministry of Health, and
switching of antibiotics was at the discretion of the attending clinicians.
Sample collection and testing
Two expectorated sputum samples were collected. One sample was
sent to the laboratory for microscopy, culture and drug susceptibility
testing as the standard of care, and the other sample was tested
using FilmArrayPN-PCR. FilmArrayPN-PCR is a multiplex PCR
assay approved by the US Food and Drug Administration. e assay
identies 18 commonly occurring bacterial pathogens, 9respiratory
viruses, and 7 antimicrobial resistance genes for extended beta-
lactamases, carbapenemases, and Staphylococcus resistance
(Supplementary Table1, available online at http://coding.samedical.
org/file/2340). Sputum quality was assessed using the Bartlett
score prior to inoculation on culture media.[11] A sputum sample
with <10epithelial cells per low-power eld and >25 pus cells per
low-power eld was considered good quality. Sputum culture was
done on chocolate, blood and MacConkey agar. Isolates grown from
sputum were suspended in Mueller Hinton broth and incubated in
the automated BD Phoenix AST system (Becton Dickinson, USA) for
rapid identication and antimicrobial susceptibility testing.
Study outcomes
We dened a potential pathogen likely to cause CAP as any positive
result from sputum culture or FilmArrayPN-PCR. We defined
AMR as present based on the results of sputum culture drug
susceptibility testing or presence of molecular markers of resistance
on FilmArrayPN-PCR.
Data management and statistical analysis
Double data entry was performed in Microso Access 2016 (Microso,
USA), and statistical analysis was performed using Stata version 9.0
(StataCorp, USA). We calculated the proportions of patients with
a potential pathogen likely to cause CAP and AMR by standard of
care and FilmArrayPN-PCR and compared the concordance and
incremental yield of FilmArrayPN-PCR relative to sputum culture
and sensitivity. Participant demographic and clinical characteristics
were summarised as means and medians for continuous variables, and
categorical data were summarised as frequencies. Group comparisons
for categorical variables were made with Fisher’s exact test and for
continuous variables using the rank-sum test. To investigate the
association of clinical and demographic factors with drug resistance,
we used bivariate and multivariate binary logistic regression analysis.
A multivariate model was generated using variables with a p-value
≤0.20. Back stepwise regression procedures were used to develop the
nal multivariate model. A p-value <0.05 was considered signicant.
Ethical considerations
All participants provided written informed consent to participate
in the study. Ethics approval was obtained from the Makerere
University School of Medicine Research and Ethics Committee
54 AJTCCM VOL. 31 NO. 2 2025
ORIGINAL RESEARCH: ARTICLES
(ref. no. 2006-017). The study was also
approved by the Uganda National Council of
Science and Technology (ref. no. 2006-19).
Results
Study participants
Of 303 adults with presumptive CAP who
were screened, 107 were enrolled (Fig.1).
e baseline characteristics of the enrolled
participants are shown in Table 1. Their
median (interquartile range) age was 40
(31-46)years, and 54 (50.5%) were female.
The median CD4 lymphocyte count was
167 (85-331) cells/μL. All the participants
had a cough, with a median duration of 3
(1-4)weeks, and the cough was productive of
sputum in 100 (93.5%). e median baseline
oxygen saturation was 96% (92-98%) on room
air. At baseline, 21 (19.6%) of the 107 study
participants had oxygen saturation <92%,
31 (29.0%) had oxygen saturation 92-95%,
and 55 (51.4%) had oxygen saturation
>95%. Overall, 80 (74.8%) had recent use of
antibiotics prior to hospitalisation (Table1).
e median duration of antibiotic use before
hospitalisation was 5 (3-7) days. Overall, 68
participants (71.6%) reported ART use. On
admission, 30 participants (28.0%) were on
co-trimoxazole prophylaxis for Pneumocystis
jirovecii pneumonia. Reported comorbidities
included asthma (n=5; 4.7%), chronic
obstructive lung disease (n=3; 2.8%), liver
disease (n=2; 1.9%) and diabetes mellitus
(n=1; 0.9%).
Microbiological diagnoses
In 94 participants (87.9%), FilmArrayPN-
PCR (and sputum culture in 25 cases)
revealed a possible aetiological agent for
pneumonia (Table2). Participants who had
a shorter duration of cough were more likely
to have a positive sputum culture than those
with a longer duration of cough, although this
did not apply to the FilmArrayPN diagnostic
approach (Table1), and participants with a
positive culture were more likely to have low
oxygen saturation compared with those with
a negative culture (Table1).
Standard of care (sputum culture)
Twenty-five patients (23.4%) had bacteria
identified on sputum culture (Table 2).
Klebsiella pneumoniae was the most frequent
pathogen found in patients with a culture-based
diagnosis (n=15/25; 60.0%). is was followed
in frequency by Streptococcus pneumoniae
(n=3/25; 12.0%) and Acinetobacter baumannii
(n=3/25; 12.0%). Allbut 3 of these participants
had received antibiotics (A. baumannii n=1, S.
pneumoniae n=1 and Streptococcus pyogenes
n=1) prior to hospitalisation.
FilmArrayPN-PCR
FilmArrayPN-PCR identified a possible
bacterial aetiological diagnosis in 89 study
participants (83.2%) and viral aetiology in
53 (49.5%) (Table 2). The most frequent
bacterium identified on FilmArrayPN-
PCR was Haemophilus inuenzae (n=50/89;
46.7%). e semi-quantitative determination
of DNA for H. inuenzae was >104 genomic
copies/mL in 46/50 (46.7%). Other bacterial
pathogens identied by FilmArrayPN-PCR
included K. pneumoniae (n=29/107; 27.1%),
S. pneumoniae (n=27/107; 25.2%) and
Staphylococcus aureus (n=24/107; 22.4%).
e majority of viral infections identied
were due to human rhinovirus/enterovirus
(n=31/107; 29.0%) and non-SARS-CoV-2
coronaviruses (n=14/107; 13.1%) (Table2).
Forty-eight participants (44.9%) had mixed
infections with both bacterial and viral
pathogens.
Comparison of FilmArrayPN-PCR
with sputum culture
FilmArrayPN-PCR increased the detection
of potential pathogens associated with CAP
by 64.5% (95% confidence interval (CI)
54.8-73.1) (Table3). All the participants with
a positive sputum culture also had bacteria
identified by FilmArrayPN-PCR; however,
only 80.0% (n=20/25) of the results were
concordant. In the discordant 5participants,
sputum culture identied S. pyogenes in 2, A.
baumannii in another 2, and Pseudomonas
aeruginosa in 1. FilmArrayPN-PCR identied
H. influenzae in 3 participants (1 also had
P. aeruginosa), 1 had both S. pneumoniae
and Mycoplasma pneumoniae, and 1 had S.
aureus. Overall, FilmArrayPN-PCR identied
more than one pathogen in 14 participants
(56.0%) with a positive sputum culture. In
13participants (12.1%), no pathogens were
detected by either diagnostic test.
Antimicrobial resistance
Of all the study participants, 34 (31.8%) had
evidence of AMR on sputum culture or on
FilmArrayPN-PCR (Supplementary Table2,
available online at http://coding.samedical.
org/le/2341). Twenty of these participants
(58.8%) had AMR detected on sputum culture
and susceptibility testing, while 27 (79.3%)
had resistance gene mutations detected on
FilmArrayPN-PCR. Thirteen participants
(38.2%) had resistance detected by both
methods. Of the 13 who had AMR detected
on both tests, 12 were diagnosed with K.
pneumoniae and 1 with A.baumannii. All
12 K. pneumoniae organisms identied had
CTX-M (cefotaxime-Munich beta-lactamase)
resistance mutations and 2additionally had
carbapenemase (NDM (New Delhi metallo-
beta-lactamase) and OXA-48-like)-associated
Potential participants screened,
N=303
HIV seropositive,
n=157
Participants enrolled,
n=107
Conrmed tuberculosis, n=36
Cough >2 months, n=6
No chest X-rays, n=2
No sputum, n=2
Withdrew consent, n=2
Poor-quality sputum, n=1
Unable to consent, n=1
HIV seronegative,
n=146
Fig.1. Study prole.
AJTCCM VOL. 31 NO. 2 2025 55
ORIGINAL RESEARCH: ARTICLES
resistance mutations. e A. baumannii organism diagnosed with
resistance had an OXA-48-like mutation. All K. pneumoniae organisms
identied had multidrug resistance (MDR) for up to seven antibiotics.
All the K.pneumoniae MDR organisms had phenotypic resistance to
cefotaxime, and 11 had resistance to cefuroxime. Imipenem was the
only antibiotic to which these organisms were universally susceptible.
Fourteen participants had genotypic resistance mutations but did
not have phenotypic resistance demonstrated on sputum culture.
The resistance mutations identified were CTX-M in 13 of these
participants and mecAC/MREJ in 1 participant. ree participants
with CTX-M had additional mutations due to NDM (n=2) and
mecAC/MREJ (n=1). Seven participants had phenotypic resistance,
but had no genotypic resistance mutations detected. ree of these
participants had S. pneumoniae, 2 K. pneumoniae, 1 A. baumannii
and1 P. aeruginosa.
Factors associated with antimicrobial resistance
Using multiple logistic regression analysis, we examined the
association of demographic and baseline clinical characteristics with
AMR (Table4). Age and gender were confounders of the association
between prior antibiotic use and AMR. Factors independently
associated with increased odds of AMR included baseline oxygen
Table1. Baseline characteristics of study participants with community-acquired pneumonia compared by diagnostic approach
Characteristic
All (N=107),
n (%)‡
SoC*
p-value
FilmArrayPN-PCR†
p-value
Negative (n=82;
76.6%), n (%)‡
Positive (n=25;
23.4%), n (%)‡
Negative (n=13;
12.1%), n (%)‡
Positive (n=94;
87.9%), n (%)‡
Sociodemographic features
Age (years),
median (IQR)
40 (31.0-46.0) 40.2 (31.0-46.0) 37.4 (31.1-41.2) 0.31 40.2 (32.1-46.3) 39.7 (31.0-45.2) 0.50
Female 54 (50.5) 42 (51.2) 12 (48.0) 0.82 7 (53.9) 47 (50.0) 1.00
Smoking 22 (20.6) 19 (23.2) 3 (12.0) 0.27 3 (23.1) 19 (20.2) 0.73
Alcohol use 76 (71.0) 58 (70.7) 18 (72.0) 1.00 8 (61.5) 68 (72.3) 0.52
Biomass fuel use 74 (69.2) 68 (82.9) 6 (24.0) <0.001 11 (84.6) 63 (67.0) 0.34
Clinical symptoms
Duration of cough
(weeks), median
(IQR)
3.0 (1.0-4.0) 3 (2-4) 2 (1-4) 0.05 2 (2-4) 3 (1-5) 0.51
Sputum production 100 (93.5) 75 (91.5) 25 (100.0) 0.20 10 (76.9) 90 (95.7) 0.04
Fever 83 (77.6) 63 (76.8) 20 (80.0) 1.00 13 (100) 70 (74.5) 0.04
Chest pain 74 (69.2) 54 (65.9) 20 (80.0) 0.22 9 (69.2) 65 (69.2) 1.00
Diculty breathing 62 (57.9) 44 (53.7) 18 (72.0) 0.11 8 (61.5) 54 (57.5) 1.00
Wheeze 21 (19.6) 19 (23.2) 2 (8.0) 0.15 3 (23.1) 18 (19.2) 0.72
Weight loss 90 (84.1) 68 (82.9) 22 (88.0) 0.76 12 (92.3) 78 (83.0) 0.69
Physical examination
Temperature (oC),
median (IQR)
36.8 (36.4-37.5) 36.8 (36.3-37.2) 37.1 (36.5-37.8) 0.29 36.9 (36.4-37.1) 36.8 (36.4-37.5) 0.71
Heart rate (bpm),
median (IQR)
92 (82-104) 93 (79-105) 89 (86-96) 0.68 99 (92-106) 89 (82-102) 0.09
Respiratory rate
(/min), median
(IQR)
24 (20-26) 23.5 (20-26) 24 (21-28) 0.36 24 (20-28) 23.5 (20-26) 0.52
Oxygen saturation
(%), median (IQR)
96 (92-98) 96 (94-98) 92 (90-96) 0.004 97 (94-98) 95 (92-98) 0.41
Laboratory result
CD4 count (cells/
µL), median (IQR)
167 (85-331) 196 (75-340) 125 (94-280) 0.45 172 (125-484) 167 (75-331) 0.61
Treatment history
Receiving ART 68 (71.6) 52 (73.2) 16 (66.7) 0.60 6 (50.0) 62 (74.7) 0.09
Pneumocystis
prophylaxis
30 (28.0) 25 (34.3) 5 (25.0) 1.00 4 (30.8) 26 (34.5) 1.00
Prior antibiotic use 80 (74.8) 58 (70.7) 22 (27.5) 0.12 9 (69.2) 71 (75.5) 0.73
SoC = standard of care; FilmArrayPN-PCR = Biore FilmArray Pneumonia Panel plus; IQR = interquartile range; ART = antiretroviral therapy.
*SoC (positive or negative) refers to potential pathogens causing community-acquired pneumonia detected (positive) or not (negative) on sputum culture.
†FilmArrayPN-PCR (positive or negative) refers to potential respiratory pathogens detected (positive) or not (negative) on the FilmArrayPN-PCR.
‡Except where otherwise indicated.
56 AJTCCM VOL. 31 NO. 2 2025
ORIGINAL RESEARCH: ARTICLES
Table2. Aetiology of bacterial and viral causes of community-acquired pneumonia identied on sputum culture and
FilmArrayPN-PCR
Pathogens identied Sputum culture FilmArrayPN-PCR*
Bacteria identied n=25 patients n=89 patients
Haemophilus inuenzae 1 50
Klebsiella pneumoniae 15 29
Streptococcus pneumoniae 3 27
Staphylococcus aureus 0 24
Enterobacter cloacae 0 14
Moraxella catarrhalis 0 13
Acinetobacter baumannii 3 12
Escherichia coli 0 6
Streptococcus agalactiae 0 5
Streptococcus pyogenes 2 4
Pseudomonas aeruginosa 1 2
Klebsiella oxytoca 0 2
Proteus spp. 0 2
Atypical bacteria n=2 patients
Mycoplasma pneumoniae Not done 1
Chlamydia pneumoniae Not done 1
Legionella pneumoniae Not done 0
Viruses identied n=53 patients
Rhinovirus Not done 31
Coronavirus Not done 14
Respiratory syncytial virus Not done 6
Inuenza A virus Not done 3
Parainuenza virus Not done 3
Adenovirus Not done 3
Human metapneumovirus Not done 2
Inuenza B virus Not done 0
Middle East respiratory syndrome coronavirus Not done 0
FilmArrayPN-PCR = Biore FilmArray Pneumonia Panel plus.
*Some patients had more than one pathogen identied.
Table3. Comparison of SoC v. FilmArrayPN-PCR results for the detection of pathogens causing community-acquired pneumonia
Test n (%) (95% CI)
SoC
Negative 82 (76.6) (67.5-83.8)
Positive 25 (23.4) (16.2-32.5)
FilmArrayPN-PCR
Negative 13 (12.1) (7.1-20.0)
Positive 94 (87.9) (80.0-92.9)
Levels of agreement/disagreement
FilmArrayPN-PCR positive and SoC positive 25 (23.4) (16.2-32.5)
FilmArrayPN-PCR positive and SoC negative 69 (64.5) (54.8-73.1)
FilmArrayPN-PCR negative and SoC positive 0
FilmArrayPN-PCR negative and SoC negative 13 (12.1) (7.1-20.0)
Overall level of agreement
All are positive or all are negative 38 (35.5) (26.9-45.2)
Not all agree 69 (64.5) (54.8-73.1)
SoC = standard of care; FilmArrayPN-PCR = Biore FilmArray Pneumonia Panel plus; CI = condence interval.
AJTCCM VOL. 31 NO. 2 2025 57
ORIGINAL RESEARCH: ARTICLES
saturation <92% (adjusted odds ratio (aOR) 9.20; 95% CI 2.52-33.57;
p=0.001), baseline oxygen saturation 92-95% (aOR 3.16; 95% CI
0.99-10.10; p=0.05) and prior antibiotic use (aOR 4.14; 95% CI
1.04-16.51; p=0.04).
Discussion
Our study has several ndings with important implications for the
clinical care of PWH and presumptive CAP. Using FilmArrayPN-
PCR led to a 64.5% increase in the detection of potential respiratory
pathogens compared with traditional sputum culture-based methods.
Accurate diagnoses of pathogens causing pneumonia and accompanying
antibiotic resistance are critical in people with HIV and advanced
immune suppression, who are at increased risk of poor outcomes.[3]
e increased yield of FilmArrayPN-PCR and faster turnaround for
identication of respiratory pathogens can lead to instant treatment
decision-making. An additional advantage of the multiplex PCR assays
is the ability to detect pathogens despite prior antibiotic use.[12]
Murphy etal.[13] conducted a multicentre evaluation of FilmArrayPN-
PCR and found a sensitivity of >95% and a specicity of >91% compared
with culture. Similar increased yields of multiplex PCR assays have also
been reported in other studies, especially in HIV-negative populations.[14]
A few studies have addressed the usefulness of multiplex PCR assays
in PWH. In one of these studies, Maartens etal.[15] in Cape Town,
South Africa, performed multiplex PCR of sputum in PWH with
World Health Organization danger signs and cough, and found a high
prevalence of tuberculosis (52%), CAP (32%) and P. jirovecii pneumonia
(9.2%). Probable bacterial infection (>105 copies/mL) was detected in
47% of participants.
The epidemiology of pathogens causing CAP varies according
to geographical region. In the present study, H. influenzae was the
bacterial pathogen most frequently detected by FilmArrayPN-PCR,
while multidrug-resistant K. pneumoniae was most frequent on
sputum culture. The most common pathogens in CAP globally are
S.pneumoniae and H.inuenzae.[3] However, in the context of HIV-related
immunosuppression, there are higher rates of invasive disease due
to H.influenzae, which may be associated with high morbidity and
Table4. Logistic regression analysis for factors associated with antimicrobial resistance in HIV-positive patients with community-
acquired pneumonia
Characteristic OR (95% CI) p-value aOR (95% CI) p-value
Age (years)
≤29 1.00 1.00
30-39 1.08 (0.36-3.29) 0.89 1.26 (0.32-4.99) 0.75
40-49 0.65 (0.21-2.01) 0.45 0.64 (0.16-2.57) 0.53
≥50 0.48 (0.10-2.24) 0.35 0.58 (0.09-3.68) 0.57
Female 1.16 (0.51-2.61) 0.73 1.24 (0.46-3.36) 0.68
Smoking 0.76 (0.27-2.16) 0.61
Subjective fever 1.53 (0.55-4.28) 0.42
Any weight loss 4.14 (0.89-19.25) 0.07
Haemoptysis 3.85 (1.52-9.77) <0.001
Chest pain 2.74 (1.01-7.46) 0.05
Diculty in breathing 1.51 (0.65-3.51) 0.33
Sputum production 2.96 (0.34-25.6) 0.33
Wheeze 0.08 (0.01-0.63) 0.02 0.12 (0.01-1.00) 0.05
Oxygen saturation (%)
96-100 1.00 1.00
92-95 3.16 (1.11-9.05) 0.03 3.16 (0.99-10.10) 0.05
<92 14.38 (4.29-48.12) <0×001 9.20 (2.52-33.57) 0.001
Temperature (oC)
<36.4 1.00
36.4-37.5 1.45 (0.50-4.26) 0.50
>37.5 2.62 (0.78-8.75) 0.12
Heart rate >100 bpm 0.43 (0.16-1.18) 0.10
Respiratory rate >20/min 2.03 (0.83-4.98) 0.12
Receiving ART 1.21 (0.46-3.19) 0.69
Recent antibiotic use 5.06 (1.40-18.24) 0.01 4.14 (1.04-16.51) 0.04
Pneumocystis prophylaxis 1.35 (0.56-3.29) 0.51
CD4 count (cells/µL)
<200 1.00
200-350 0.83 (0.25-2.75) 0.76
>350 0.40 (0.09-1.77) 0.23
OR = odds ratio; aOR = adjusted odds ratio; ART = antiretroviral therapy.
58 AJTCCM VOL. 31 NO. 2 2025
ORIGINAL RESEARCH: ARTICLES
mortality.[16] A fundamental question is how many of these potential
pathogens were actual pathogens. It is usually dicult to dierentiate
between colonisation and infection, since some of these pathogens are
also commensals in the oropharynx. e majority of the patients with
a diagnosis of H. influenzae had >104 genomic copies, indicating a
relative abundance of this organism in sputum. e signicance of this
semi-quantitative measure when compared with colony-forming units
on standard culture is still not well established. However, in a study by
Park etal.,[17] the colonisation density of the upper respiratory tract was
associated with a conrmed microbiological diagnosis of H. inuenzae
infection. Itis not surprising that on culture, drug-resistant K. pneumoniae
was the dominant species identied, while fastidious organisms such as
S. pneumoniae and H. inuenzae were less frequent because 74.8% of the
participants had received prior antibiotics.
On FilmArrayPN-PCR, rhinovirus infections were the commonest
identified, followed by coronaviruses (non-SARS-CoV-2). The
significance of these viral infections is not well understood, but
rhinovirus/enterovirus and endemic coronavirus infections commonly
present as self-limited diseases in immune-competent individuals.
However, they may be associated with severe infections, including
CAP and intensive care unit infections, in immunosuppressed hosts,
the elderly, and patients with signicant underlying conditions.[18] ey
have also been reported as mixed infections with bacteria. Inthe EPIC
multicentre pneumonia study, rhinovirus infections were also the
commonest cause of pneumonia.[19] Multiplex PCR assays therefore
enable identication of viruses in the respiratory samples, to provide
insight into alternative diagnoses or co-pathogens.
Another significant finding in the present study was detection
of AMR in one-third of the patients with CAP. FilmArrayPN-PCR
detected resistance gene mutations to commonly used antibiotics in
79.3% of the study participants. Only 38.2% had resistance identied
by both culture and FilmArrayPN-PCR. is discordance between
phenotypic and genotypic resistance has also been observed in
other studies. Lee etal.[20] investigated the role of FilmArrayPN-
PCR for the detection of determinants of AMR compared with the
minimum inhibitory concentration (MIC) method, and found similar
discrepancies. e determinants of AMR are multifactorial. For the
participants who had AMR detected by the MIC method but had no
resistance mutations detected on FilmArrayPN-PCR, this failure to
detect resistance mutations may be due to the limited selection of
mutation targets in the FilmArrayPN-PCR, or because they had fewer
pathogens than the detection limit of the assay. On the other hand, for
those who had determinants of resistance identied on FilmArrayPN-
PCR but had no phenotypic resistance, the determinants detected may
not be ascribed to the pathogens cultured, since these mutations can
be associated with a wide range of pathogens.[21] Multidrug-resistant
pathogens were identied, especially K. pneumoniae, A. baumannii
and S. aureus. is worrying trend of AMR, which has been observed
globally, is of public health importance as it threatens to negate the
progress in antimicrobial treatment for CAP.[22,23] e presence of
such infections rapidly reduces treatment options for patients with
pneumonia, hence the need for regular antimicrobial surveillance in
hospitalised patients.
Hypoxia (oxygen saturation <92%) was associated with nine times
increased odds of AMR. Hypoxia may reect severe pneumonia as
a result of non-response to treatment. Prior antibiotic use was also
associated with increased odds of AMR. Based on these ndings,
patients with CAP and hypoxia should be targeted for sputum
culture and drug susceptibility testing, and PCR testing to detect
AMR. A study by Shindo etal.[24] found that prior antibiotic use and
immune suppression were independent predictors of drug-resistant
pathogens in patients with CAP. Similarly, Prina etal.[25] found that
previous antibiotic use was associated with an increased risk of
infection with P. aeruginosa, extended-spectrum beta-lactamase-
positive Enterobacteriaceae and methicillin-resistant S. aureus (PES)
organisms, increasing 30-day mortality.[25]
Study limitations
Limitations of this study include the lack of longitudinal follow-up
of the participants to assess the clinical outcomes of those who had
AMR identied but were on standard therapy. Secondly, 74.8% of
participants in our study had taken antibiotics prior to hospitalisation,
which may select for a population with suboptimal response to
initial antibiotics in an ambulatory setting and therefore skew the
aetiological prole of CAP, but reects the reality of hospitalised CAP.
irdly, the reference standard for the diagnosis of CAP is imperfect,
since sputum culture, the traditional gold standard, has low sensitivity
and blood cultures were not performed. On the other hand, multiplex
PCR assays may have high sensitivity but can also have high false-
positive rates in the real world. Finally, FilmArrayPN-PCR does not
detect opportunistic infections such as P. jirovecii, Mycobacterium
tuberculosis and Cryptococcus, which occur commonly in PWH, and
only detects a selected number of antimicrobial resistance mutations
that may not fully represent important infections and common
resistance patterns in a high HIV burden setting.
Conclusion
e FilmArray pneumonia panel increased the yield for diagnosis of
both potential bacterial and viral infections causing CAP in hospitalised
PWH in Uganda. In addition, prior antibiotic use and hypoxia were
associated with an increased risk of AMR. Further evaluation of
the signicance of these ndings and the cost-eectiveness of the
molecular tests should be done in prospective studies.
Data availability. e datasets generated and analysed during the present
study are available from the corresponding author (WW) on reasonable
request. Any restrictions or additional information regarding data access
can be discussed with the corresponding author.
Declaration. None.
Acknowledgements. We acknowledge the contributions of Magezi
Yusuf and Kate Nabakiibi with data collection, and Edson Mwebesa
with statistical analysis. We thank the Infectious Disease Research
Collaboration, Medical and Molecular Laboratories Ltd, China-Uganda
Friendship Hospital and Kiruddu Hospital for enabling this research.
Author contributions. WW, AA, FS, AC, DRB and LH contributed to the
design of the work. All authors contributed to data acquisition, analysis,
or interpretation of data for the work. All authors contributed to draing
the work or reviewing it critically for important intellectual content, and
all authors approved the nal version to be published.
Funding.is study was supported by funding from Biore Diagnostics
LLC and bioMérieux SA, the European and Developing Countries Clinical
AJTCCM VOL. 31 NO. 2 2025 59
ORIGINAL RESEARCH: ARTICLES
Trials Partnership (TMA2018SF-2465), and the National Institutes of
Health (NIH R01 HL128156 and NIH R01 HL143998).
Conicts of interest.None.
1. Hirschtick RE, Glassroth J, Jordan MC, etal. Bacterial pneumonia in persons
infected with the human immunodeciency virus. Pulmonary Complications of HIV
Infection Study Group. N Engl J Med 1995;333(13):845-851. https://doi.org/10.1056/
NEJM199509283331305
2. Garcia Garrido HM, Mak AMR, Wit F, et al. Incidence and risk factors for
invasive pneumococcal disease and community-acquired pneumonia in human
immunodeciency virus-infected individuals in a high-income setting. Clin Infect Dis
2020;71(1):41-50. https://doi.org/10.1093/cid/ciz728
3. Cilloniz C, Garcia-Vidal C, Moreno A, Miro JM, Torres A. Community-acquired
bacterial pneumonia in adult HIV-infected patients. Expert Rev Anti Infect Ther
2018;16(7):579-588. https://doi.org/10.1080/14787210.2018.1495560
4. Gleckman R, DeVita J, Hibert D, Pelletier C, Martin R. Sputum Gram stain assessment
in community-acquired bacteremic pneumonia. J Clin Microbiol 1988;26(5):846-849.
https://doi.org/10.1128/jcm.26.5.846-849.1988
5. Kobyakova OS, Deev IA, Vinokurova DA, etal. Is there a real need for sputum culture
for community-acquired pneumonia diagnostics? Results from a retrospective study
in Russia. Diagnosis (Berl) 2021;8(3):377-381. https://doi.org/10.1515/dx-2020-0027
6. Harris AM, Bramley AM, Jain S, etal. Inuence of antibiotics on the detection of bacteria
by culture-based and culture-independent diagnostic tests in patients hospitalized with
community-acquired pneumonia. Open Forum Infect Dis 2017;4(1):ofx014. https://doi.
org/10.1093/od/ofx014
7. Lee CC, Lee CH, Hong MY, Tang HJ, Ko WC. Timing of appropriate empirical
antimicrobial administration and outcome of adults with community-onset bacteremia.
Crit Care 2017;21(1):119. https://doi.org/10.1186/s13054-017-1696-z
8. Yoshimine H, Oishi K, Mubiru F, etal. Community-acquired pneumonia in Ugandan
adults: Short-term parenteral ampicillin therapy for bacterial pneumonia. Am J Trop
Med Hyg 2001;64(3-4):172-177. https://doi.org/10.4269/ajtmh.2001.64.172
9. World Health Organization. irteenth General Programme of Work, 2019-2023. WHO/
PRP/18.1. Geneva: WHO, 2019. https://iris.who.int/bitstream/handle/10665/324775/
WHO-PRP-18.1-eng.pdf (accessed 2 April 2025).
10. Ginocchio CC, Garcia-Mondragon C, Mauerhofer B, Rindlisbacher C; and the
EME Evaluation Program Collaborative. Multinational evaluation of the BioFire(R)
FilmArray(R) Pneumonia plus Panel as compared to standard of care testing. Eur J
Clin Microbiol Infect Dis 2021;40(8):1609-1622. https://doi.org/10.1007/s10096-021-
04195-5
11. Bartlett JG, Breiman RF, Mandell LA, File TM Jr. Community-acquired pneumonia in
adults: Guidelines for management. e Infectious Diseases Society of America. Clin
Infect Dis 1998;26(4):811-838. https://doi.org/10.1086/513953
12. Baudel JL, Tankovic J, Dahoumane R, etal. Multiplex PCR performed of bronchoalveolar
lavage fluid increases pathogen identification rate in critically ill patients with
pneumonia: A pilot study. Ann Intensive Care 2014;4:35. https://doi.org/10.1186/
s13613-014-0035-7
13. Murphy CN, Fowler R, Balada-Llasat JM, etal. Multicenter evaluation of the BioFire
FilmArray Pneumonia/Pneumonia Plus Panel for detection and quantication of agents
of lower respiratory tract infection. J Clin Microbiol 2020;58(7):e00128-20. https://doi.
org/10.1128/JCM.00128-20
14. Gilbert DN, Leggett JE, Wang L, et al. Enhanced detection of community-
acquired pneumonia pathogens with the BioFire(R) Pneumonia FilmArray(R)
panel. Diagn Microbiol Infect Dis 2021;99(3):115246. https://doi.org/10.1016/j.
diagmicrobio.2020.115246
15. Maartens G, Griesel R, Dube F, Nicol M, Mendelson M. Etiology of pulmonary infections
in human immunodeciency virus-infected inpatients using sputum multiplex real-
time polymerase chain reaction. Clin Infect Dis 2020;70(6):1147-1152. https://doi.
org/10.1093/cid/ciz332
16. Collins LF, Havers FP, Tunali A, etal. Invasive nontypeable Haemophilus inuenzae
infection among adults with HIV in metropolitan Atlanta, Georgia, 2008-2018. JAMA
2019;322(24):2399-2410. https://doi.org/10.1001/jama.2019.18800
17. Park DE, Baggett HC, Howie SRC, et al. Colonization density of the upper
respiratory tract as a predictor of pneumonia – Haemophilus inuenzae, Moraxella
catarrhalis, Staphylococcus aureus, and Pneumocystis jirovecii. Clin Infect Dis
2017;64(suppl_3):S328-S336. https://doi.org/10.1093/cid/cix104
18. Falsey AR, Walsh EE, Hayden FG. Rhinovirus and coronavirus infection-associated
hospitalizations among older adults. J Infect Dis 2002;185(9):1338-1341. https://doi.
org/10.1086/339881
19. Jain S, Self WH, Wunderink RG, etal. Community-acquired pneumonia requiring
hospitalization among U.S. adults. N Engl J Med 2015;373(5):415-427. https://doi.
org/10.1056/NEJMoa1500245
20. Lee SH, Ruan SY, Pan SC, Lee TF, Chien JY, Hsueh PR. Performance of a multiplex
PCR pneumonia panel for the identication of respiratory pathogens and the main
determinants of resistance from the lower respiratory tract specimens of adult patients
in intensive care units. J Microbiol Immunol Infect 2019;52(6):920-928. https://doi.
org/10.1016/j.jmii.2019.10.009
21. Hujer AM, Long SW, Olsen RJ, etal. Predicting β-lactam resistance using whole
genome sequencing in Klebsiella pneumoniae: The challenge of β-lactamase
inhibitors. Diagn Microbiol Infect Dis 2020;98(3):115149. https://doi.org/10.1016/j.
diagmicrobio.2020.115149
22. Holt KE, Wertheim H, Zadoks RN, etal. Genomic analysis of diversity, population
structure, virulence, and antimicrobial resistance in Klebsiella pneumoniae, an urgent
threat to public health. Proc Natl Acad Sci USA2015;112(27):E3574-E3581. https://doi.
org/10.1073/pnas.1501049112
23. Cao X, Xu X, Zhang Z, Shen H, Chen J, Zhang K. Molecular characterization of clinical
multidrug-resistant Klebsiella pneumoniae isolates. Ann Clin Microbiol Antimicrob
2014;13:16. https://doi.org/10.1186/1476-0711-13-16
24. Shindo Y, Ito R, Kobayashi D, etal. Risk factors for drug-resistant pathogens in
community-acquired and healthcare-associated pneumonia. Am J Respir Crit Care Med
2013;188(8):985-995. https://doi.org/10.1164/rccm.201301-0079OC
25. Prina E, Ranzani OT, Polverino E, etal. Risk factors associated with potentially
antibiotic-resistant pathogens in community-acquired pneumonia. Ann Am orac
Soc 2015;12(2):153-160. https://doi.org/l10.1513/AnnalsATS.201407-305OC
Submitted 11 July 2024. Accepted 14 August 2024. Published 2 June 2025.
