Introduction
Pneumonia, defined as infection and inflammation of the lung parenchyma [1], is the commonest lower respiratory infection [2, 3]. Systematic reviews of data from hospital systems, surveillance systems, and other sources, covering 471 million individual records or isolates and 7585 study-location-years, revealed that lower respiratory infections accounted for more than 1·5 million deaths associated with antimicrobial resistance, making it the most fatal infectious syndrome [3] out of a global estimated 4·95 million (3·62–6·57) deaths associated with bacterial antimicrobial resistance (AMR); with death rates in low-and-middle income countries at 19% in Senegal, 23% in Cambodia, 16% in Central African Republic and 18% in Uganda [4]. AMR has risen over time, and the advent of multi-drug-resistance (MDR) gram-negative bacteria (GNB) complicates therapy; causing increased morbidity and mortality in people of all ages around the world [2].
Moreover, emergence of MDR organisms poses tough challenges in deciding empirical as well as definitive therapy. Hence the need to prioritise development of local evidence on the cause and management, to improve outcomes of patients hospitalised with pneumonia [5].
GNB including Pseudomonas aeruginosa, Acinetobacter baumanii, Klebshiella pneumoneae and Stenotrophomonas maltophilia have frequently been isolated as causative pathogens in community-acquired pneumonias (CAP) [6, 7]. Similarly, there is an increase in the incidence of GNB such as Klebshiella pneumonia, Pseudomonas aeruginosa, and Acinetobacter baumani that are linked to the aetiology of hospital acquired pneumonias (HAP) [2, 5]. Most recently, multi-drug-resistant gram-negative bacteria (MDR-GNB) increasingly cause pneumonia [8] and majority (50–80%) HAP and ventilator-associated pneumonia (VAP) [9].
Antibiotic resistant Pseudomonas aeruginosa, Acinetobacter baumannii, Klebsiella pneumoniae, and Escherichia coli are responsible for 929,000 (660,000–1,270,000) deaths attributable to AMR and 3·57 million (2·62–4·78) deaths associated with AMR [3]. The global prevalence of GNB in CAP was reported between 49.7% to 83.1%, whereas in VAP patients ranged between 76.13% and 95.3%; with the most commonly reported MDR-GNB causes as Acinetobacter baumani, Klebshiella pneumonia, and Pseudomonas aeruginosa, with Acinetobacter baumani isolated particularly in VAP patients and the elderly. Prior MDR-GNB infection, older age, previous use of broad-spectrum antibiotics, high frequency of local antibiotic resistance, prolonged hospital stays, intensive care unit (ICU) admission, mechanical ventilation, and immunosuppression have been associated with the MDR-GNB colonization.
Similarly, S. maltophilia causes severe HAP/VAP among mechanically ventilated hospitalised patients due to its abilities of biofilm formation, site adhesion in respiratory devices, as well as its intrinsic and acquired drug resistance mechanisms [2]. In addition, the MDR- GN aerobic bacilli that cause pneumonia express several resistance phenotypes including resistance to extended spectrum beta lactamases (ESBL), cephalosporinases, and fluoroquinolone resistant phenotypes which render the most effective and safest antimicrobials ineffective in treatment of pneumonia, and unreliable as a choice for empiric treatment.
Laboratory results derived from respiratory tract specimens, however, are often unreliable for a pathogenic specific pneumonia diagnosis because of contamination by oropharyngeal bacterial flora [10], and better diagnostic yield is offered by lung tissue specimens obtained by direct means such as ultra sound guided lung biopsy [11]. Therefore, this study analyzed post-mortem lung specimens from the deceased patients with pneumonia to obtain microbiological, antibiotic susceptibility and epidemiological data that is essential for effective empirical therapy, and diagnosis of MDR gram negative bacterial pneumonia. Results from this study will contribute towards evidence-based recommendations for effective empiric therapy, particularly for high-risk individuals in critical settings where microbiological diagnosis may be inaccessible or delayed.
In addition, our results contribute to national and global data on MDR-associated mortality to inform surveillance and strategic interventions towards the rising burden of AMR.
Methods
Study design and population
This was a cross-sectional autopsy study in which lung autopsy specimens were collected aseptically from patients who died with a primary diagnosis of peumonia at Mulago national referral hospital mortuary during the period of June-December 2024. Specimen processing was conducted at the College of American Pathologists (CAP)-certified microbiology laboratory at Makerere University College of Health Sciences (MakCHS), following United Kingdom standards for microbiological investigation (UKSMI) protocols for investigating tissues and biopsies from deep-seated sites and organs [12]. Microbiology testing was later done to identify MDR aerobic gram- negative bacilli causing fatal pneumonia and factors associated with MDR- GNB pneumonia.
The Mulago national referral hospital mortuary is the clinical teaching facility for pathology practical sessions for both undergraduate and post graduate students of MaKCHS.
Study procedures
We included adult in-patients aged 18 years and above, irrespective of duration of hospital admission, who had an ante-mortem diagnosis of either pneumonia or lower respiratory tract infection and died not more than 20 hours before onset of the post-mortem examination at the mortuary and had legal next of kin (NOK) consent for autopsy. We excluded deceased patients who were missing patient files and clinical data, for whatever reason. Dependent variables that were extracted included the etiologic bacteria isolates, antibiotic susceptibility testing and MDR phenotypes.
Independent variables included socio-demographics, type of health facility where patient was referred from, clinical diagnosis and any underlying medical illness, antibiotic therapy and any other medication for co-morbidities, prescriptions, and any invasive devices the patient had received including pharyngeal, oxygen therapy by nasal prongs, and or endotracheal instrumentation.
Post-mortem procedures
The anterior chest wall was disinfected with alcohol and povidine iodine followed by dissection of the chest wall with high level disinfected knives with formaldehyde. The chest cavity and lungs were inspected for gross features of pneumonia then removed from the chest cavity and placed on specially modified board sterilized by formaldehyde. Using sterile gloves and a sterile surgical blade, the lungs were dissected and the parenchyma was inspected for edema, heptisation, congestion and palpated for pitting.
Lung tissue was aseptically dissected and immediately placed on MacConkey agar (MCA), Sheep Blood Agar (SBA), and Chocolate Blood Agar (CBA) agar plates which were immediately covered. Lung tissue fluid was also swabed to make smears that were air dried and gram stained. Agar plates were incubated according to laboratory SOP in the clinical microbiology lab of the department of medical microbiology of MakCHS.
This was followed by biochemical tests and susceptibility testing by BD PhoenixTM automated identification and susceptibility testing system and subsequently selection of isolates MDR isolates and detection of resistance phenotypes to betalactams was done using the European Committee on Antimicrobial Susceptibility Testing (EUCAST) guidelines 2017 [13].
Gross and histopathology
Lungs were inspected for features pathognomonic of pneumonia at gross examination and dissected to expose the internal lung parenchyma for further inspection. Slices of lung tissue were made and placed on agar plates of MacConkey, chocolate blood agar and sheep blood agar removed and covered the plates and later placed them in the icebox. Lung tissue fluid was aspirated or swabbed from the cut lung surface and used to immediately make a smear, which was air dried and then placed in sterile plastic bags for transportation to the clinical microbiology laboratory of MakCHS within two (2 ) hours of specimen collection.
Where features of inflammation were noted other areas of the body (e.g. the cranial, urinary bladder, intra-abdominal cavity), additional samples were taken to find out if the aetiological pathogens were similar (at least phenotypically).
Identification of bacteria and antimicrobial susceptibility testing
Blood and Chocolate blood agar plates that had been inoculated were incubated at 35 °C to 37 °C at 5% carbondioxide concentration for 18 to 24 h. In addition, inoculated MacConkey agar plates were incubated at ambient air for 18 to 24 h. At the end of the stated periods, the agar plates were removed from the incubators, and bacterial colonies were inspected for colony characteristics and recorded.
Discrete colonies were then gram stained and those that were gram negative rods had the oxidase test performed on them. Both oxidase negative and oxidase positive gram-negative rod bearing colonies were put in cryovials and dispensed using the BD PhoenixTM machine for identification and antimicrobial susceptibility testing. Bacterial colonies were also put in cryovials and stored at minus 80 °C using glycerol 99.5% (Loba Chermie ™) media.
Testing for resistance phenotypes to beta- lactam antibiotics
Resistance phenotype testing to beta lactam antibiotics was done using combination disk test according to EUCAST guidelines version 2.0 (2017). Resistance phenotypes to beta-lactam antibiotics that were tested included extended spectrum beta lactamases (ESBLs), AmpCs, and carbapenemases including classes A and B.
a) ESBL Detection: Screening Phase: Initial screening for ESBL-producing bacteria was performed using a screening breakpoint of >1 mg/L for both cefotaxime and ceftazidime. Confirmatory Testing: Phenotypic confirmation was conducted using the combination disk test method. This involved comparing the inhibition zones around cefotaxime-clavulanic acid and ceftazidime-clavulanic acid disks with the zones around cefotaxime and ceftazidime disks alone.
An increase in inhibition zone diameter of ≥5 mm with clavulanic acid compared to the antibiotic alone indicated a positive ESBL test.
b) AmpC Beta-Lactamase Screening: For E. coli and Klebsiella pneumoniae isolates, AmpC screening was performed using Cefoxitin MIC >8 mg/L OR zone size <19 mm combined with phenotypic resistance to ceftazidime and/or cefotaxime (breakpoint >1 mg/L).
c) Carbapenemase Detection: Screening Phase: Isolates showing a meropenem inhibition zone diameter of < 25 mm were selected for carbapenemase testing.
Phenotypic testing for carbapenemase classes: Carbapenemase detection was performed using combination disk tests on muller hinton agar to differentiate between enzyme classes; for Class A Carbapenemases,Meropenem-boronic acid disk and meropenem disk alone were placed on a bacterial lawn on Mueller-Hinton agar and for Class B Carbapenemases, Meropenem-EDTA disk and meropenem disk alone were placed on a bacterial lawn on Mueller-Hinton agar. Incubation and Interpretation: All inoculated plates were incubated at 35 °C to 37 °C for 18 to 24 h in ambient incubator. Results were read, interpreted according to EUCAST guidelines (2017) and recorded.
Quality control
i. Bacterial Control Strains: American Type Culture Collection (ATCC) bacterial control strains, including Escherichia. Coli ATCC 25,922, Pseudomonas. aureuginosa ATCC 27,853 were used for quality control procedures throughout this study.
These control isolates were utilized to validate media growth characteristics, antimicrobial sensitivity testing protocols, including carbapenemase and extended-spectrum beta-lactamase (ESBL) detection assays.
ii. Culture Media and Biochemical Testing Quality Control: Quality control of culture media and biochemical testing reagents was performed through systematic validation procedures. All prepared media was taken through sterility testing via overnight incubation at both room temperature (20-25°C) and optimal growth temperature (35-37°C) in ambient air conditions.
Growth characteristics were assessed through inoculation with standard reference organisms following media preparation, ensuring appropriate selectivity and nutritional adequacy.
iii. Gram Stain Quality Control: The reliability of Gram staining procedures was verified using established quality control protocols. Known gram-positive and gram-negative control strains were processed on glass slides according to standard Gram staining methodology.
Microscopic examination was conducted using oil immersion at 1000× magnification (100× objective lens) to confirm appropriate staining characteristics and procedural accuracy.
Data management
Data was checked for correctness and completeness and then entered into excel spreadsheets, validated, and analyzed using STATA version 14.0. Descriptive data was analyzed and presented in tables and figures. Multivariate analysis was done to determine the factors associated with resistance to carbapenems and ESBL.
Results
Demographic and clinical characteristics
Overall, 120 adults died due to a primary diagnosis of pneumonia during hospitalization between June -December 2024; of whom 60/100 (50%) were female. The mean age (standard deviation) of the deceased patients was 50.2 (16.6) years and the median duration of hospitalization before death was 42.5 (Inter-quartile range 19.0–80.0) hours. Over half of the patients [53/120 (44%)] were referrals from private health care facilities while 43 (35.8%) were from public health care facilities.
Up to 32 (26.7%) had a confirmed diagnosis of HIV (Table 1). Pathognomonic features of pneumonia and related complications were described grossly as illustrated in Fig. 1.
Almost half of the patients 59/120 (49%) had co-morbidities; and over 103(85.8%) received empirical antibiotic treatment without microbiological investigations (Table 2).
Table 1Demographics and clinical characteristics of 120 patients who died of pneumonia and related complications at Mulago National referral hospitalVariableFrequency (N = 120)Sex Female60 (50.0%) Age mean(sd) years50.2 (16.6)Health facility (Home, Private, Public) Home17 (14.2%) Private53 (44.2%) Public43 (35.8%) Not indicated7 (5.8%) Median hours of Hospital Stay; median(IQR)42.5 (19.0–80.0) Postmortem Interval (Hours) median(IQR)9.0 (6.0–13.0)HIV status Neg86 (71.7%) Positive32 (26.7%) Not indicated2 (1.7%)Risk factor for MDR colonisation† N = 9 Intraventricular Shunt Device2 (2.2%) Trauma2 (2.2%) Homelessness1 (11.1%) ICU Admission1 (11.1%) Intra Uterine Device1 (11.1%) Mechanical Ventilation1 (11.1%) Urinary Catheterization1 (11.1%)Clinical Differential Diagnosis† N = 131 Pneumonia40 (33.6%)Alcoholic Coma1 (0.8%) Sepsis21 (17.7%)Aplasitc Anaemia1 (0.8%) Pulmonary tuberculosis (PTB)12 (10.1%)COPD1 (0.8%) Severe Pneumonia8 (6.7%)Diabetic Encephalopathy1 (0.8%) Aspiration Pneumonia7 (5.9%)Dilated Cardiomypathy1 (0.8%) Urosepsis5 (4.2%)Diabetic ketoacidosis1 (0.8%) Cardiovascular accident4 (3.4%)Edematous Pressure Sores1 (0.8%) Malaria3 (2.5%)Heart Failure2 (1.7%) Diabetic Ketoacidosis2 (1.7%)HIV-associated complications2 (0.8%) HTN2 (1.7%)Ischaemic Stroke1 (0.8%) Post Operative Abdominal Sepsis2 (1.7%)Meningitis1 (0.8%) Septic Shock2 (1.7%)Neutropaenic Sepsis1 (0.8%) Stroke2 (1.7%)Obstructive Jaundice1 (0.8%) Alcohol intoxication2 (1.7%)Osteomyelitis1 (0.8%)†N = 131 because 11 patients had multiple clinical diagnosis
Fig. 1Post-mortem histological diagnosis for patients who died with primary diagnosis of pneumonia at Mulago National Referral Hospital between June-December 2024
Table 2Empirical antibiotic prescription for adults who died with a primary diagnosis of pneumonia at Mulago National referral hospital between June-December 2024Antibiotics prescribed† N = 135N (%)I.V Ceftriaxone30 (25.0%)I.V Metronidazole3 (2.5%)I.V Piperacillintazobactam26 (21.7%)Ehrz2 (1.7%)Ceftriaxone19 (15.8%)I.V Ciprofloxacin2 (1.7%)I.V Coamoxiclav14 (11.7%)I.V Gentamicin2 (1.7%)I.V Meropenem12 (10.0%)Amoxicillin1 (0.8%)I.V Amoxiclav5 (4.2%)Azithromicin1 (0.8%)I.V Ceftazidime4 (3.3%)Caps Doxycline1 (0.8%)Metronidazole4 (3.3%)Ceftazidme1 (0.8%)Coamoxiclav3 (2.5%)Gentamicin1 (0.8%)I.V Levofloxacin3 (2.5%)Tabsazithromicin1 (0.8%)Microbiology investigations n = 120 None103 (85.8%) MTB Gene Xpert8 (6.7%) Blood Culture3 (2.5%) Blood slide For Malaria Parasites2 (1.7%) Pus Swab for Culture and sensitivity2 (1.7%) Urinalysis2 (1.7%)Microbiology diagnosis, n(%) None made118 (98.3) Gram Negative Bacterial Pneumonia1 (0.8%) Disseminated Tuberculosis1 (0.8%)Comorbidities† N = 143, n(%) None61 (50.8%)Mental Illness2 (1.7%) Alcohol Abuse22 (18.3%)Prostate Cancer2 (1.7%) Diabetes Mellitus10 (8.3%)Burns1 (0.8%) HTN6 (5.0%)Cancer Head Of Pancreas1 (0.8%) COPD4 (3.3%)Cervical Cancer1 (0.8%) Substance Abuse4 (3.3%)Chronic Kidney disease1 (0.8%) HIV3 (2.5%)End Stage Renal Disease1 (0.8%) Lung Cancer3 (2.5%)Epilepsy1 (0.8%) Cigarette Smoking2 (1.7%)Intravenous Drug Abuse1 (0.8%) Dilated Cardiomyopathy2 (1.7%)Liver Cirrhosis1 (0.8%) Hydrocephalus2 (1.7%)Mental Illness2 (1.7%)
Bacteria isolates from lungs of patients who died of pneumonia
The most frequently prescribed antibodies were intravenous (i.v) Ceftriaxone among 30 (25%) of the patients, followed by intravenous (i.v) Piperacillin tazobactam 26 (21.7%), i.v Ceftriaxone 19 (15.8%), i.v Coamoxiclav 14 (11.7%) and Meropenem 12 (10.0%); as shown in Table 2.
Bacteria isolated from tissues and their antibiotic sensitivity profile
A Lung sample was collected from all patients who died with a primary diagnosis of pneumonia including 75 (54.4% lung tissue, 37 (26.8%) lung aspirate, 10 (7.3%) Lung pus and 3 (2%) pleural fluid (Tables 3 and 4). The bacteria isolated were Klebshiella pneumoniae species in 54 (45.0%) patients, E.coli in 33 (27.5%), Pseudomonas aeruginosa in 17 (14.2%) and Acinetobacter species in 12 (10.0%) of the patients (Tables 3 and 4).
Table 3Autopsy specimen collected for analysis of antimicrobial resistance profiles among individuals who died with a primary diagnosis of pneumoniaAutopsy specimen† N = 138n (%)Lung Tissue75 (54.4%)Lung Aspirate37 (26.8%)Lung Pus10 (7.3%)Pleural Fluid3 (2.2%)Turbid Cerebral spinal fluid3 (2.2%)Urinary Bladder Pus3 (2.2%)CSF2 (1.5%)Bone Tissue1 (0.7%)CSF Pus1 (0.7%)Mesenteric Nodes1 (0.7%)Peritoneal Pus1 (0.7%)†Some patients hand more than one sample
Table 4Bacterial isolated from autopsy specimen collected for analysis of antimicrobial resistance profiles among individuals who died with a primary diagnosis of pneumoniaIsolated bacteria at post-mortemN (%)SpecimenKlebshiella pneumoniae ssp. pneumoniae54 (45.0%)Lung pus, Lung aspirate, Lung tissue, Mesenteric Nodes, Pleural FluidE.coli33 (27.5%)Lung pus, Lung aspirate, Lung Tissue, Pleural FluidPseudomonas aeruginosa17 (14.2%)Lung pus, Lung aspirate, Lung Tissue, Peritoneal PusAcinetobacter ssp.12 (10.0%)Lung Aspirate, Lung TissueS.maltophia3 (2.5%)Lung TissueHyperviscous mucoid K.pneumoniae1 (0.8%)Lung Tissue
Antibiotic resistance profile
The bacterial isolates were more than 20% resistant to all the antibiotics used except tigecycline which was at 14.3% (Table 5).The bacteria isolated from the lungs of patients with pneumonia were resistant to most of the antibiotics that were used for empirical therapy Ceftriaxone (86.7%), Piperacillin tazobactam (44.2%), Coamoxiclav (46.7%) and Meropenem (21.7%). Amikacin and Tigecycline had the least resistance at 20% and 14.3%; respectively (Fig. 2).
Table 5Antibiotic resistance profiles among individuals who died with a primary diagnosis of pneumonia at Mulago National referral hospital between June-December 2004Antibiotic n (%)Sensitive (S)Resistant (R)Intermediate (I)Ampicillin4 (3.3%)116 (96.7%)0(0.0%)Cefazolin8 (6.7%)110 (92.4%)1 (0.8%)Cefuroxime11 (9.2%)109 (90.8%)0(0.0%)Trimethoprim–Sulphamethaxole11 (9.3%)107 (90.7%)0(0.0%)Ceftriaxone11 (9.2%)104 (86.7%)5 (4.2%)Cefepime21 (17.5%)91 (75.8%)8 (6.7%)Ciprofloxacin35 (29.4%)83 (69.7%)1 (0.8%)Ceftazidime34 (28.3%)76 (63.3%)10 (8.3%)Levofloxacin35 (29.4%)73 (61.3%)11 (9.2%)Gentamicin46 (38.3%)73 (60.8%)1 (0.8%)Ceftalozane-Tazobactam45 (37.5%)71 (59.2%)4 (3.3%)Amoxicillin-Clavulanic Acid44 (36.7%)56 (46.7%)20 (16.7%)Piperacillinn Tazobactam64 (53.3%)53 (44.2%)3 (2.5%)Ertapenem66 (55.5%)48 (40.3%)5 (4.2%)Imipenem71 (59.2%)42 (35.0%)7 (5.8%)Meropenem81 (67.5%)26 (21.7%)13 (10.8%)Amikacin96 (80.0%)24 (20.0%)0(0.0%)Tigecycline100 (84.0%)17 (14.3%)2 (1.7%)
Fig. 2Resistance Phenotype to Beta lactam Antibiotic for 104 patients who died with a primary diagnosis of pneumonia at a National Referral Hospital between June-December 2024
Factors associated with carbapenem resistance and Extended-Spectrum Beta-Lactamase (ESBL) resistance
Overall, Cabarpenem resistance was observed in 60/120 (50%) of patients; as defined by resistance to either Imipenem or Meropenem or Ertapenem. Patients who had pneumonia with co-morbidities were 2.5 times more likely to have Cabarpenem resistance; Odds Ratio, OR (95% Confidence interval (CI) 2.51(1.07 to 5.87); p value = 0.033 and patients referred from private health care facilities were almost 5 times more likely to have Cabarpenem resistance; OR 4.92 (CI 1.35 to 17.9); p-value = 0.016. Patients with a known diagnosis of HIV were twice more likely to have carbapenem resistance although the p-value was not statistically significant (Table 6).
Table 6Factors associated with carbapenem resistance among adults who died with a primary diagnosis of pneumonia at Mulago National referral hospital between June-December 2004Carbapenem Resistance (Imipenem R + Meropenem R and Ertapenem R)60 (50.0%)No carbapenem Resistance60 (50.0%)AOR (CI)P-valueAge51.4 (16.9)49.1 (16.5)0.99 (0.97–1.02)0.878Gender Female32 (53.2%)28 (46.7%)Ref. Male28 (46.7%)32 (53.3%)0.74 (0.34–1.61)0.447HIV status Positive13 (21.7%)19 (31.7%)Ref Negative47 (78.3%)39 (65.0%)2.27 (0.91–5.69)0.079Comorbidities Pneumonia without comobordities25 (41.7%)36 (60.0%)Ref. Pneumonia with comobordities35 (58.3%)24 (40.0%)2.51 (1.07–5.87)0.033Health Facility of primary care Home4 (6.7%)13 (21.7%)Ref Private30 (50.0%)23 (38.3%)4.92 (1.35–17.9)0.016 Public23 (38.3%)20 (33.3%)3.12 (0.82–11.9)0.096 Not indicated3 (5.0%)4 (6.7%)3.28 (0.43–25.330.254Duration of hospital stay62.0 (19.0,109.0)33.35 (18.0,71.5)1.00 (0.99–1.01)0.367AOR: Adjusted Odds Ratios; CI : Confidence Interval
Discussion
We reported a mean age of 50.2 years of deceased patients with a diagnosis of pneumonia who died within a median of 42.5 hours of hospitalization. Over half of the patients were referrals from private health care facilities while 43 (35.8%) were from public health care facilities. Almost half of the patients 59/120 (49%) had co-morbidities; and over 103(85.8%) received empirical antibiotic treatment.
We isolated Klebshiella pneumoniae species in 54 (45.0%) patients, Escherichia coli in 33 (27.5%), Pseudomonas aeruginosa in 17 (14.2%) and Acinetobacter species in 12 (10.0%) from lung samples of patients who died with a primary diagnosis of pneumonia. Our data is comparable to a recent global systematic review which showed that Gram-negative bacilli (GNB) were common causes of severe hospital-acquired pneumonia; and the six leading pathogens for deaths associated with resistance were Escherichia coli, followed by Staphylococcus aureus, Klebsiella pneumoniae, Streptococcus pneumoniae, Acinetobacter baumannii, and Pseudomonas aeruginosa [3]. Due to changes in the health care environment and selective antimicrobial pressure, MDR GNB are increasingly causing pneumonia even outside of the traditional hospital settings.
Similar to our finding, patients with serious comorbidities that need mechanical ventilation have high risk of infection with antibiotic-resistant Acinetobacter spp., representing a therapeutic challenge [14, 15]. Given that available diagnostic techniques have a limited capacity to accurately detect GNB pulmonary infection, as well as limited access to rapid culture and sensivity results, many patients with fatal peumonias are treated empirically similar to our findings where all patients who died with a diagnosis of pneumonia had received emperical antibiotics.
Unfortunately we reported that the causative bacteria isolated at autopsy were resistant to most of the antibiotics that were used presumptively during the antemortem period; exhibiting 86.7% resistance to Ceftriaxone, 44.2% resistance to Piperacillin tazobactam, 46.7% resistance to Coamoxiclav and 21.7% resistance to Meropenem; although Amikacin and Tigecycline had the least resistance at 20% and 14.3%; respectively. Moreover, patients who had pneumonia with co-morbidities were 2.5 times more likely to have Cabarpenem resistance and patients referred from private health care facilities were almost 5 times more likely to have Cabarpenem resistance. Our data is similar to previous reports where open lung biopsy provided a high diagnostic yield in kidney transplant patients with diffuse pulmonary infiltrates who did not have a definitive diagnosis and had failed to respond to empiric antibiotic treatment [11].
Similarly, patients with serious comorbidities that need mechanical ventilation were shown to have a high risk of infection with antibiotic-resistant Acinetobacter spp., representing a therapeutic challenge [14, 15]. For example, half of the infections due to Acinetobacter spp., particularly ventilator associated pneumonia in intenstive care units died within one month [9, 16]. Identification of these risk factors for MDR-GNB would guide clinicians to efficiently identify patients who are likely to have pneumonia due to these pathogens [10] and inform treatment algorithms for effective combination therapies targeting pan-drug resistant strains to improve treatment outcome and reduce AMR associated mortality [2].
For example, in settings where carbapenem resistance exceeded 20%, empiric colistemethate Sodium (CMS) in combination with a carbapenem (other than ertapenem), tigecycline, or sulbactam was recommended [17]. In our setting Amikacin or tigecycline would be recommended since they exhibited less resistance than carpapemens, although tigecycline may have limitations due to low plasma levels [18].
Our study is comparable with previous reports from Dessie Referral Hospital in Ethiopia where predominant pathogens in sputum of patients with pneumonia were Klebsiella pneumoniae (28.0%), Streptococcus pneumoniae (24.8%), Staphylococcus aureus (18.5%), and Pseudomonas aeruginosa (14.0%); and majority of the isolates exhibited resistance to ampicillin at 81.5%, followed by penicillin resistance at 75.9% and resistance to amoxicillin-clavulanate was at 61.2%. Resistance to ampicillin and penicillin was associated with repeated prescription and use [19] and ventilator-associated pneumonia [8]. A major strength of our study was the use of autopsy lung tissue from patients who died of pneumonia, which is more accurate than sputum studies that may include a number of commensal bacteria in the upper respiratory tract.
Limitations
We did not conduct molecular studies to determine presence and patterns of specific Gram-negative bacilli drug-resistant genes. Multiple drug resistance genes could be detected at the same time in most strains; for example KPC-2 was the most common carbapenemase-resistant gene in Klebsiella pneumoniae, whereas NDM-1 was common in Escherichia coli [19]. We recommend bacterial genotyping studies to further understand the expression of the Gram-negative bacilli drug-resistant genes to inform innovative efforts to expand AMR diagnostics through development of point of care PCR-based rapid diagnostic tests (RDT).
Conclusion
Fatal bacterial pneumonia was caused by multidrug-resistant gram-negative rods; that were resistant to most of the empirically prescribed antibiotics including ESBL resistance at 48% and Cabarpenem resistance (as defined by resistance to either Imipenem or Meropenem or Ertapenem) at 50% in this selected group; while Amikacin and tigecycline were still spared. Patients with co-morbidities (including Alcohol abuse, diabetes mellitus, hospitalisation intensive care unit, chronic lung diseases and prior antibiotic use) and referrals from private clinics had a higher risk of carbapenem resistance. We therefore recommend that tigecycline and Amikacin may be reserved for carbapenem-resistant cases of pneumonia.
Further research is needed to determine the prevalence of resistance genes and their potential transmission surveillance to inform strategic interventions to combat or delay the rapidly rising ESBL and carbapenem resistance in Uganda.