Co-infections among COVID-19 Patients in Ugandan Intensive Care Units

Samuel Olowo (  samuel28olowo@gmail.com ) Mbarara University of Science and Technology John K. Obwiin Mulago National Referral Hospital Florence Pido Mulago National Referral Hospital Martha Alupo Mulago National Referral Hospital Charles Ssendikadiwa Mulago National Referral Hospital Jane Nakibuuka Mulago National Referral Hospital Edrin Jjuuko UMC Victoria Hospital Erasmus E. Okello Makerere University Lydia VN. Ssenyonga Busitema University Faculty of Health Sciences Jacob S. Iramiot Busitema University Faculty of Health Sciences

The study accessed records of all COVID-19 patients admitted in the ICU at Mulago national referral hospital, UMC-Victoria and TMR international private hospitals during the rst, second and third waves of the pandemic in Uganda who met the inclusion criteria. The study considered patients admitted from March 2020 to March 2022.

Sampling method
A non-probability purposive sampling method was employed to select only those patients with culture and sensitivity of one or more of blood, urine, tracheal aspirate and pus swab taken within 48hours of admission.

Data collection and management
Data was collected retrospectively from the patient records, speci cally the patients' les for Mulago national referral hospital and microbiology electronic databases for the private facilities. The collected data were entered into standardized case report forms. The collected data was stored in a locked le kept by the investigators and no unauthorized person had access to it.

Inclusion Criteria And Exclusion Criteria
All COVID-19 patients admitted in the intensive care unit at Mulago national referral hospital, UMC-Victoria and TMR international private hospitals with culture and sensitivity tests done on any body uid during the rst, second and third waves of the COVID-19 pandemic in Uganda. The study included only culture and sensitivity results of samples taken within 48hours of admission.
The study excluded COVID-19 patients with mixed culture growths or contaminated samples.

Data Analysis And Presentation
Collected data was veri ed to ensure completeness. It was then entered in Epi-Data version 3.1, cleaned and edited for inconsistency, coded then exported to Stata Version 17.0 for statistical analysis. The participants' base line characteristics like age and sex, were categorized and presented as frequencies and percentages and their respective proportions in tables and texts. The rate of having co-infection among COVID-19 ICU patients and the antimicrobial susceptibility patterns were also presented as frequencies, proportions and graphs.

Ethical considerations
The study was approved by Mulago hospital research and ethics committee (reference no. 2176). Adherence to the ethical principles for medical research involving human subjects as stipulated in the Declaration of Helsinki was ensured. Mulago hospital research and ethics committee also provided a waiver of informed consent (reference no. 2176) as the study possessed minimal risk to the participants. The researchers obtained administrative clearance from the executive director Mulago national referral hospital and directors of the two private hospitals. Permission was also sought from the o cers in charge of the different COVID-19 intensive care units. To maintain participants' anonymity, unique numbers were allocated to each of the study participants and no other identifying information was obtained.

Results
Participants' demographic characteristics Two hundred twenty ve patients' record were reviewed. Nine patients were excluded due to mixed culture growths or contaminated samples. Two hundred sixteen participants were recruited and all included in the nal analysis as illustrated in Fig. 1.
Among the 216 participants, the majority 133(61.57%) were males. The mean age of the participants was 54.93 ± 14.48 years. Majority of the participants, 119(53.72%) were from private facilities, Table 1.   Table 3.  Table 4. about 48(80%) demonstrated multidrug resistance and 12(20%) were labeled as possible extensive drug resistant as detailed in Table 5.

Co-fungal Infections Among The Participants
About 31(27.93%) of the organisms isolated were Candida albicans. A few of them were tested for their susceptibility to the different antifungal agents, Table 8.

Susceptibility of Candida albicans to various antifungal agents
Candida albicans, the fungi isolated was susceptible to most antifungal agents, Table 8

Overall Susceptibility Patterns Of Organisms Isolated Among The Participants
Among antibiotics that had at least 4 isolates tested for susceptibility against. Amoxclave which had 6/10(60%) different species of isolates tested against had the highest overall susceptibility that is 33/54(61.11%). This was followed by tetracycline 9/16(56.25%), chloramphenicol and imipenem which shared the same proportion, 20/39(51.28%), and then amikacin, 7/14(50%) overall susceptibility. Ceftriaxone had the lowest overall susceptibility 4/44(9.09%) among antibiotics that had at least 4 isolates tested for susceptibility against. The results of the overall susceptibility is shown in the Fig. 2 below.

Discussion
The novel Corona Virus Disease 2019 (COVID-19) pandemic whose etiological agent is severe acute respiratory syndrome coronavirus2 (SARS-CoV-2) has signi cantly strained healthcare systems worldwide [11]. One of the emerging concerns is the potential long term impact of SARS-CoV-2 on antimicrobial resistance [12]. Therefore, we set out to determine the prevalence of co-infections and their antimicrobial susceptibility pattern among COVID-19 Patients in Ugandan Intensive care units.
Most of the participants in this study were from private facilities. Mulago national referral hospital handled large volumes of COVID-19 patients in critical care units far beyond the private hospitals. However, the private hospitals were more likely to perform culture and sensitivity for their patients both at admission and during the course of hospital stay. Therefore, this explains why most of our participants were from private hospitals.

Prevalence Of Co-infections
This study found a high prevalence (51.39%) of co-infections (bacterial and fungal) among COVID-19 patients admitted in Uganda intensive care units. Higher prevalence of co-infections among ICU COVID-19 patients have been documented in related studies in China [13] where a 58.3% prevalence of bacterial respiratory pathogens was found and Belgium [14] where a 40.6% prevalence of co-infections among COVID-19 patients in the intensive care unit was reported. Similarly studies in England [2] found a high proportion (30.3%) of potential co-pathogens 48hrs after admission in to the intensive care unit. The high prevalence of co-infections among ICU COVID-19 patients observed in this study could be because it has been observed that critically ill COVID-19 patients have a high risk of secondary infections [15]. Some of the microbial organisms involved in co-infections have been reported to develop resistance to commonly used antibiotics resulting into superinfections among COVID-19 patients during intensive care unit stay [16,17]. Super-infection has been cited as a signi cant contributor to mortality among ICU COVID-19 patients after refractory respiratory failure [7]. It could also have resulted from the strain imposed by the pandemic on health care system as there was an increased demand for manpower in the intensive care units. This resulted into relocation of non-critical care health workers especially nurses and medical o cers to serve in the intensive care units and therefore probably paid less attention to infection control measures in relation to ICUs. Previous studies [18] have reported that the spread of bacteria in the intensive care unit environments are usually due to cross contamination mediated by health workers. Speci cally, the transmission of pathogens are fostered by direct contact between health workers and patients [19]. Moreover, it has been postulated that infection control measures during COVID-19 pandemic [20] are strongly directed at avoiding the transmission and cross-infection by SARS-CoV-2 and that during the height of the pandemic, limited attention is geared to simple infection control measures, and care bundles for invasive catheters. Respiratory tract infection was the most prevalent co-infection among ICU COVID-19 patients (51.35%) in this study. This was reported in other related studies in China [21]. Another study in China [13] similarly reported a higher proportion of respiratory co-infection among ICU COVID-19 patients. However, ndings from Spain [7] reported primary blood stream infections as the commonest infection among ICU COVID-19 patients followed by pneumonia. This difference could be due to difference in healthcare settings and infection control policies. This is of concern because positive respiratory and blood cultures (but not urine cultures) have been reported to demonstrate association with mortality and days spent in the intensive care unit [22]. And furthermore, positive tracheal aspirate cultures have been found to be more strongly associated with mortality than positive blood cultures [22]. Primary blood stream infection and urinary tract infections almost shared same proportion of 21.62% and 22.52 respectively and were therefore the second most common co-infections after respiratory tract infection. Similar ndings was reported in China [23] were primary blood stream infection with a proportion of 24.6% was the second most common nosocomial infection after pneumonia. This study found a greater proportion of gram positive organism compared to gram negative organisms. This could be because studies have shown greater trends towards gram positive organisms [24]. The commonest gram negative organisms in this study were; Citrobacter freudii, Pseudomonas aureginosa, and Klebsiella pneumoniae while the commonest gram positive bacteria species was Staphylococcus aureus. This nding is in agreement that in related studies in Italy [8] where Pseudomonas aureginosa, Klebsiella pneumonia and Staphylococcus aureus were the most isolated pathogens and with that done locally in Uganda [24] among the non-COVID-19 intensive care unit patients in Mulago hospital and international hospital Kampala in which Pseudomonas aureginosa and Klebsiella pneumonia were among the commonest gram negative organisms after Acinetobacter spp, and Staphylococcus aureus was the commonest gram positive organism. Considering the cause of respiratory tract infection, gram negative organisms were the most frequent cause where Citrobacter freudii was the most isolated organisms. This contribution of gram negative organisms has been reported in other related studies [7] though Pseudomonas aureginosa was the most implicated organism in contrast to this current study. Acinetobacter spp was among the least isolated organisms in this study together with Staphylococcus saprophyticus, Streptococci viridians, Proteus mirabilis and Leuconostoc mesenteroid spp with only one isolate of each discovered. This is contrary to other reports from Bangladesh [25] where Acinetobacter spp was found to be the dominant cause of hospital acquired infection and the most frequently isolated organisms in COVID-19 patients compared to the non-COVID-19 patients. Another study in South Africa [22] also reported Acinetobacter spp as the most frequently isolated organisms among ICU COVID-19 patients followed by Klebsiella spp. Acinetobacter spp has been reported to be in water supply in hospitals [26] and contaminates basic life support and advanced life support equipment in the intensive care unit [27]. This shows that commonly used infection control measures like sterilization of some of the life support equipment can be employed to prevent the transmission of Acinetobacter spp and ensuring aseptic techniques during nasal gastric tube feeding [27,28]. Therefore the limited number of Acinetobacter spp isolated in this study could be because of effective and e cient sterilization of resuscitation equipment and reusable ventilator circuits or use of non-reusable ventilator circuits especially in the private facilities in this study. It could further be due to the use of pre-packaged mineral water to ush nasogastric tube and patients' drinking as opposed to use of hospital supplied tape water.
Concerning fungal co-infections, Candida albicans was the only isolated fungal species with a prevalence of 27.93% considering all isolated organisms. This nding is similar to those from Egypt [29] where Candida albicans was isolated among severely ill COVID-19 patients. However, this Egyptian study [29] also isolated Candida glabrata Similar nding on Candida albicans was also reported in United kingdom [30] where a 12.6% proportion of yeast (candida spp) infections was reported. Reports from a systematic review [31] also highlighted candidemia as the commonest fungal co-infection among COVID-19 patients.
Yeast infections mainly caused by Candida spp have been reported as one of the major complication of COVID-19 in the intensive care units [31] because of the prolonged ICU stay, urinary catheterization and use of broad spectrum antibiotics among patients which has been implicated as risks factors for nosocomial candidemia [32]. Other species of fungal infections especially invasive pulmonary aspergillosis have been reported in other studies elsewhere [33,34]. Fungal infections have been associated with worst outcome among patients contributing to a high mortality rate [35]. We should therefore not only focus on bacterial co-infections but also pay attention to fungal forms of the infection.

Antimicrobial Susceptibility Pattern
This study found a high prevalence of multidrug resistant bacterial specis and possible extended drug resistant bacterial speciesis. A local strudy done at Mulago hospital on surgical site infections [36] also showed a high proportion of MDR and ESBL. However, this was done in obsteritics and gynecology, general surgery and othorpedic wards. Therefore, the high prevalence of MDR observed in this study is expected as ICU is a con uence of these wards as some COVID-19 patients were identi ed from these wards and transfers made to the intensive care unit. This is because some of ICU COVID-19 patients are trapped from these wards during screening. Among the MDR gram negative organisms, Pseudomonas aureginosa, Citrobacter freudii and Klebsiella pneumoniae showed the highest proportion of multdrug resistance. This has been reported in other related studies [7]. None of the Staphylococus aureus isolated was methicilin resistant. This is contray to other reated studies [37] were methicillin resistant Staphylococus aureus were isolated among ICU COVID-19 patients. The absence of methicilin resistant Staphylococus aureus (MRSA) in this study could be because of low proportion of samples from wounds and surgical sites [38]. The only Acinetobacter spp isolated was multidrug resistant and was further labled as possible extensive drug resistant. Yet, ICU acquired multdrug resistant Acinetobacter spp has been associated with high mortality [39] among patients. This study also identi ed four of the six bacterial speceis commonly labeld as potential pathogengs in the intensive care units. These are Pseudomonas aureginosa, Klebsiella pneumoniae, Staphylococus aureus, and Acinetobacter spp. These six speceisis are usually labelled under the acronym ESKAPE [40] that is Enterococcus faecium, Staphylococus aureus, Klebsiella pneumoniae, Acinetobacter baumannii, Pseudomonas aureginosa, and Enterobacter coacae. The members of ESKAPE identi ed in the study were all labeled us multidrug resistant and others; Pseudomonas aureginosa and Klebsiella pneumoniae were further labled as possible extended drug resistant. Molecular evidence has shown that the ESKAPE group has acquired various resistance mechaninsims to alter commonly used therapeutic options. "Reistance cassettes, antibiotic e ux pumps, plasmids, transposons, and integrons have been recognised as the genetic elements responsible for multi-resistance to antibiotics in ESKAPE bacteria" [41]. More than half of Klebsiella pneumoniae were ESBL producers and the only one isolate of Proteus mirabilis was a penicillinase producer. Augumentin/Amoxclave, was the antibiotic with highest proportion of susceptibility, preceeded by tetracycline and chloramphenicol, and further imipenem and amikacin which shared proportions of overal suceptibility among agents that had at least 4 isolates tested for susceptibility against.
Ceftriaxone had the lowest overall susceptibility among antibiotics that had at least 4 isolates tested for susceptibility against. We selected atleast 4 agents because some organisims like Acinetobacter spp and Proteus mirabilis, only one organisim was isolated and therefore their overall susceptibility does not represent the entire picture of antibiotics tested. The highest susceptibility rate recorded for Citrobacter freudii was to imipenem, augmentin and chloramphenicol, the highest susceptibility for Klebsiella pneumoniae was to colistin and ceftazidime (similar to ndings from Kenya [42]), and the highest susceptibility recorded for Pseudomonas aureginosa was to ceftriaxone and amikacin. This nding is in agreement with that in studies done in United States [43] among non COVID-19 ICU paatients. Comparing this nding with local data, some agreement exist, for example a study conducted at Mulago hospital and internation hosiptal kampala [24] among the non COVID-19 ICU patients also found that Klebsiella pneumoniae was susceptible to amikacin and that Pseudomonas aureginosa was not susceptible to amikacin. Though in this study, augmentin demonstrated the highest overall susceptibiltiy among the agents which had atleast 4 isolates tested against, it showed less suscepitibility to Pseudomonas aureginosa and Klebsiella pneumoniae which are among the top three common bacterial speecisi isolated in this study after Citrobacter freudii. The susceptibility for piperacillin and tazobactum were on average about 50% less than expected. This nding is similar to that carried out in Uganda among non COVID-19 patients [24]. This could be probably due to its increasing use [44]. Staphylococci aureus had the highest susceptibility to vancomycin, tetracycline, augumentin, clindamycin and chlorampheniocal and nonsusceptible to erythromycin, penicillin and oxacillin. Similar ndings were reported in studies done in Mulago hospital [24,36] among non-COVID-19 patients.
This may demobstrate commonality in origin of the organisims. A study done at Mulago hospital [45] "to characterise the lineages of Staphylococci aureus among patients with surgical site infection found two predominate lineages clonally circulating on the surgical wards and three others con ned to obstetric wards". Therefore, it's plausible that the pattrens of susceptibillity seen in the ICU maybe due to single clones. Candida albicans, the only fungal agent isolated in this study showed good overal suceptibility to most of the antifungal agents. However, emerging antifungal resistance especially for Candida auris [46] has been reported. Therefore, antifungal resistance needs critical attetion.

Conclusion
This study found a high prevalence of co-infections (bacterial and fungal). Respiratory tract infection was the most prevalent. There was an overwhelming burden of multidrug resistant infections with some extended spectrum drug resistant organisms isolated among COVID-19 patients admitted in the Ugandan ICUs. There is need for establishment of stronger policy measures in regards to antibiotic stewardship, antimicrobial surveillance and infection control to inform empirical antibiotic therapy and mitigate the spread MDR bacteria and antibiotic drug resistance among COVID-19 patients.

Limitations Of The Study
The results of this study should be interpreted with consideration of some limitations higlighted below; Our study was soley based on micribiology culture to determine co-infections. Most of our patients were already on atleast one empirical broad spectrum antibiotics before culture samples were taken off. This is because COVID-19 intensive care unit in most cases received patients from lower care units especially high dependency units where the patients in most cases are initiated on broad spectrum antibiotics. Microbiological culture has been reported to be relatively insenstive techneque especially during antibioitic treatment [47]. It was further reported that "it can be di cult to distingusih between infection and colonisation in a non-sterile state and even with sepsis only 30-50% will have a positive blood culture" [48]. We cannot therefore soley base on positive microbiology as an indicator of bacterial infection. Therefore, it's possible that ndings from this study does not represent the actual picture of bacterial coinfections and therefore antimicrobial susceptibiltiy parttern among ICU COVID-19 patients in Uganda.
Another limitation of the study was that since in some cases patients are to pay for the culture and sensitivity of the samples due to resource constrains, some patients' samples may have not been tested because they could not afford. Ethics approval and consent to participate The study was approved by Mulago hospital research and ethics committee (reference no. 2176). Adherence to the ethical principles for medical research involving human subjects as stipulated in the Declaration of Helsinki was ensured. Mulago hospital research and ethics committee also provided a waiver of informed consent (reference no. 2176) as the study possessed minimal risk to the participants. The researchers obtained administrative clearance from the executive director Mulago national referral hospital and directors of the two private hospitals. Permission was also sought from the o cers in charge of the different COVID-19 intensive care units. To maintain participants' anonymity, unique numbers were allocated to each of the study participants and no other identifying information was obtained. Figure 1 Study pro le. A total of two hundred twenty ve patients' records were reviewed. Of the two hundred twenty ve patients' records reviewed, two hundred sixteen participants were nally recruited in the study. All the 216 participants recruited were included in the nal data analysis.

Figure 2
Overall susceptibility patterns of organisms isolated.