Antibiotic resistance profile of multidrug-resistant Pseudomonas aeuriginosa from clinical isolates: a three-year retrospective study in two tertiary hospitals of Yaoundé, Cameroon

¹Faculty of Medicine and Pharmaceutical Sciences Sangmelima, University of Ebolowa, Ebolowa, Cameroon, ²Faculty of Medicine and Biomedical Sciences, University of Yaounde I, Yaounde, Cameroon, ³Faculty of Sciences of University of Douala, Douala, Cameroon, ⁴Military Hospital of Yaounde, Yaounde, Cameroon, ⁵Faculty of Medicine and Pharmaceutical Sciences of University of Douala, Douala, Cameroon, ⁶The University Teaching Hospital of Yaounde, Yaounde, Cameroon, ⁷Higher Institute for Scientific and Medical Research (ISM), Yaounde, Cameroon, ⁶Center for Development of Best Practices in Health (CDBPH), Yaounde, Cameroon

^&Corresponding author
Serge Eyebe, Faculty of Medicine and Pharmaceutical Sciences Sangmelima, University of Ebolowa, Ebolowa, Cameroon

Introduction    

Pseudomonas aeruginosa (PA) is a ubiquitous gram-negative bacterium with minimal requirements for survival in the environment. It has a remarkable ability to colonise surfaces even in the most restricted environments. Pseudomonas aeruginosa can cause infection in immunocompromised individuals [1]. It is one of the most important pathogens involved in healthcare-associated infections in intensive care units (ICUs). Due to natural and intrinsic resistance, infections caused by PA are difficult to treat [2]. A major global health concern is the growing resistance of PA to commonly used antibiotics and the increasing number of multidrug-resistant (MDR) strains [3]. Infections caused by these MDR strains have a significant impact on patient health, including treatment failure, increased healthcare costs, and high mortality. These infections are also associated with high mortality rates [4]. The inappropriate use of antibiotics and inadequate containment procedures in hospitals exert a selective pressure on the bacteria and are conducive to the emergence of MDR strains [3]. Pseudomonas aeruginosa possesses a wide range of virulence factors and different mechanisms for developing resistance to several classes of antibiotics. Biofilm synthesis, modification of outer membrane protein channels and efflux pumps, and production of several drug-inactivating enzymes (metallo-beta-lactamases, extended-spectrum beta-lactamases, and carbapenemases) have enabled some PA strains to effectively evade host defence and antibiotic therapy [5]. Increasing PA resistance to beta-lactams, fluoroquinolones and aminoglycosides has been reported worldwide for years [6].

In a recent study, the percentage of resistance to imipenem was 14%, 24.48% and 20.99% for the years 2013, 2014 and 2015. In the same study, resistance to fluoroquinolones ranged from 84.67% (2013) to 78.27% (2015) and resistance to ceftriaxone tended to increase from 80% (2013) to 92.59% (2015) [7]. According to US and EU surveillance, the prevalence of carbapenem-non-susceptible (CnS) PA increased from 4% in the 1990s to 14-36% in the 2000s [8]. In China, the same prevalence is over 30% [9]. The resistance profile of bacteria varies from country to country, hospital to hospital, and community to community [10]. In Africa, data are scarce and not available for 42.6% of the countries on the continent. In some countries, such as South Africa, Senegal, and the Central African Republic, the prevalence of infection by MDR-PA to commonly used antibacterial agents was low [11,12]. However, this is in contrast to the higher resistance rate of clinical isolates from Côte d'Ivoire and Nigeria [12].

Data are very scarce in Cameroon [13]. The antibiotic resistance pattern of organisms, including PA, varies from region to region. This could be due to differences in antibiotic prescription policies and infection control measures. Therefore, a better understanding of local trends in antibiotic resistance through local and regional surveillance, regular detection and reporting of antibiotic resistance patterns among bacteria, is critical for controlling the threat of drug resistance [10,14]. It would also be very helpful for local clinicians to have up-to-date knowledge to help them select appropriate antibiotics for empirical therapy. Identification of patient risk factors would also help the infection control team to monitor and update infection control measures [14]. The current study reviewed over 3 years the antibiotic resistance patterns of PA isolated in laboratory routine diagnostic in two tertiary care hospitals. The objectives were to estimate the frequency of colonisation of clinical isolates and the prevalence of infection by MDR-PA, to estimate antibiotic resistance rates of usual antibiotics, and to determine risk factors associated with infection by MDR-PA.

Methods     

Study design: this was an analytical, cross-sectional study using retrospective data collected.

Setting: the study was conducted in two tertiary hospitals, the Yaounde Military Hospital at GPS coordinates 3.854481, 11.517663, and the University Teaching Hospital at GPS coordinates 33.86264, 11.50008, both located in Yaoundé, the capital of Cameroon. These are referral hospitals according to the Cameroonian health system [15].

Participants and clinical isolates

Inclusion criteria: all patients who had a routine bacteriological laboratory test between 1^st January 2018 and 31^st December 2021 were selected. Individuals were identified by their unique medical record number in the laboratory registry, which, according to the hospital system in place, is considered to be unique regardless of the number of tests performed by the patient or the number of visits over time. The study was performed on all clinical isolates collected from these patients.

Exclusion criteria: patients for whom antibiotic susceptibility testing had been requested but no results were found in the laboratory.

Definition of variables: an infected case was defined as a patient with at least one colonised clinical isolate. A clinical isolate was any biological specimen collected from a patient for which antibiotic susceptibility testing was routinely performed. It was therefore possible for more than one clinical isolate to be associated with a single patient. A clinical isolate was considered to be colonised if a germ was identified in it. Polycolonisation, multi-colonisation or polymicrobial colonisation were defined in this context as clinical isolates that tested positive for more than one bacterial species. Co-infection was defined in this context as an individual whose clinical isolate(s) tested positive for more than one bacterial species. There could be two pathogens in the same sample. It could be two different pathogens in two different samples belonging to the same patient. Resistant phenotype or non-susceptibility is when the isolate is intermediate or resistant to an antibiotic. MDR: if an isolate was "resistant to three or more antimicrobial classes of antibiotics" [16,17]. This classification was based on the results available in the laboratory register. Resistance phenotypes were analysed according to EUCAST criteria. The dependent variable was infection with multidrug-resistant Pseudomonas aeuriginoa (MDR-PA), coded yes/no. Dependent variables were cross-tabulated with independent variables including: sex (female/male), age (years in categories), hospitalisation (yes/no), polycolonisation or polymicrobial colonisation (yes/no) and co-infection (yes/no).

Data sources/measurement: the following data were collected from the laboratory registers: age, sex, clinical isolate, pathogens identified, and susceptibility to antibiotics. The clinical isolates collected were blood, urine, stool, puncture liquid of acute, joint, and cerebrospinal fluid (CSF). The data on catheter tips were also collected, and they were provided from central line, oropharyngeal, and bladder. Culture and susceptibility reports were collected from laboratory registers. The susceptibility of antibiotics to which sensitivity was tested included piperacillin, ticacillin, piperacillin-tazocilline, ticacillin-clavulinique, amikacin, tobramycin, ceftazidime, cefepime, imipenem, and ciprofloxacin. The classification of isolates as MDR or non-MDR was performed by an experienced clinical biologist.

Potential bias: the main bias in this study was measurement bias. The information collected from the different isolates on which antibiograms were performed was routine. To reduce this bias, two hospitals with reference laboratories with qualified staff were selected, and the data were collected by two people to ensure their quality.

Study sample size: the following formula was used to calculate the minimum sample size of the study [18]. This was based on a similar study to ours, carried out in Karachi by Khan et al. which found an MDR-PA prevalence of 30% in clinical isolates [14]. According to all of this, we would need a minimum of 277 samples from clinical isolates.

Where, n = sample size; z_α/2 = confidence level according to the standard normal distribution (for a 95% confidence level, z_α/2 = 1.96); p = estimated proportion of the population with the characteristic; i = tolerated margin of error (5%). To achieve this required sample size, all patients registered during the period of the study were included. Applying this formula, the required sample size was 277. Exhaustive sampling was carried out. All samples identified in the laboratory records were selected. Using this sampling method, 5,195 samples were obtained.

Data analysis: the data were analysed using STATA 18 (Texas, US Corp). Qualitative variables were described as frequencies and percentages with medians and interquartiles. Prevalence was calculated as the number of persons infected with PA/MDR-PA divided by the number of persons admitted during the study period. Frequency of colonisation (FC) was calculated as the number of clinical isolates colonised with MDR-PA over all clinical isolates collected. The relative frequency of colonisation was calculated as the number of clinical isolates colonised by PA/MDR-PA in a specific collection (e.g. stool) over all clinical isolates in that specific collection. Antibiotic resistance rate (ARR) was the number of clinical isolates not susceptible (intermediate or resistant) to an antibiotic over all clinical isolates tested to the same antibiotic. Multidrug-resistant Pseudomonas aeruginosa rate was the number of clinical isolates classified as MDR-PA over the number of clinical isolates collected and positive for PA. Quantitative variables were described as mean and standard deviation. After checking the normality of the distributions, univariate analysis was performed between the dependent variable and each of the independent variables. Depending on the case, χ² or Fisher's exact test was used to compare categorical data, and Student's t-test or Mann-Whitney U-test for continuous data. Multivariate analysis, consisting of stepwise logistic regression, was performed to assess the association between infection with MDR-PA and each of the independent variables. The significance level was set at 5%.

Ethical consideration: this study was conducted in accordance with ethical principles. Ethical approval was obtained. The data transcribed on the electronic form were anonymised to ensure confidentiality. An ethical authorization from the Regional Ethics Committee (May 2022) of the Centre region and a research authorization No. 4750/AR/CHUY/DG/DGA/CAPRC of August 3, 2021 have been issued.

Results     

Participants and clinical isolates: a total of 5195 samples were collected from 4155 people. This is well above the minimum required sample size.

Descriptive data: first, with regard to clinical isolates, 1008 pathogens were identified in the 5195 clinical isolates (Table 1). A total of 891 (17.15%) clinical isolates were positive for at least one bacterium, and 117 (2.25%) were positive for two or more bacteria. The main clinical isolates collected were urine 46.67% (2421), followed by blood 25.77% (1339). Less than 2% of clinical isolates were collected from catheters (Figure 1). Secondly, in terms of individuals, the mean age was 34.75 years with a standard deviation of 26.12, ranging from 1 to 98 years. In this study, 21.52% (894) of the individuals were 60 years and older. The male gender predominated at 54.32% (2257). The majority of people, 55.23% (2295), were admitted to the University Teaching Hospital, and 56.00% (2327) of people were hospitalised (Table 2).

Outcome data: among the 5195 clinical isolates examined, 1008 pathogens were identified. The three most common organisms were E. coli with 20.83% (210), K. pneumoniae 12.70% (128), and S. aureus 11.90% (120). The fourth most common germ was PA with 6.35% (64) (Table 1). These 64 clinical isolates were distributed as follows: on pus 43.75% (28), in catheter stick 29.69% (19), in urine 9.38% (6), in blood 6.2% (4), and in stool 4.69% (3). None of the 316 cerebrospinal fluid and 13 synovial fluid samples were found to be colonised with PA (Figure 1). The clinical isolates most commonly colonised by PA were catheter sticks with a relative frequency of 16.10% (19), followed by ENT swabs with 15.38% (2) and pus with 12.73% (28) (Figure 1). Of the total 64 Pseudomonas aeruginosa strains, 37.5% (24) were MDR-PA. The clinical isolates most commonly colonised by MDR-PA were ENT swabs with a relative frequency of 15.38% (2), followed by catheter tips with 6.78% (8) and pus with 5% (11) (Figure 1). The urine sample and ascites fluid were colonised by PA but not by MDR-PA (Figure 1). Of the 4155 individuals, 16.68% (693) were infected with at least one bacterium. 1.97% (82) had a co-infection.

Main results: the frequency of colonisation of the 5195 clinical isolates by PA was 1.23% (64) and 0.46% for MDR-PA. The prevalence of persons infected with PA was 1.11% (46) and with MDR-PA 0.46% (19). Regarding antibiotic resistance rates, some antibiotics were found with resistance rates above 50%. This was the case for tobramycin with an ARR of 71.43% with 95% confidence interval ( 95% CI)= (56.74-83.42), for ticarcillin ARR= 62.96% with 95% CI= (48.74-75.71), for piperacillin ARR= 53.45 with 95% CI= (39.87-66.66) and for Aztreonam ARR=52.73% with 95% CI= (38.80-66.35). However, the beta-lactamase inhibitors, such as clavulanic acid and tazobactam, in combination with ticarcillin and piperacillin increased their susceptibility and decreased their resistance rates to 33.33% with 95% CI=(20.76-47.92) for ticarcillin and 27.08% with 95% CI= (15.28-41.25) for piperacillin. Ceftazidime had an ARR=27.87% with 95% CI= (17.15-40.83), Imipenem an ARR=23.81% with 95% CI= (13.98-36.21), ciprofloxacin an ARR=40.82% with 95% CI= (27.00-55.79) and amikacin an ARR=37.50% with 95% CI (24.92-51.45). The antibiotic with the lowest resistance rate was cefepime, ARR=18.97% with 95% CI (9.87-31.41) (Figure 2).

In univariate analysis, demographic characteristics associated with infection with MDR-PA were age over 60 years (p=0.000) and male gender (p=0.035), co-infection (p=0.000), hospitalisation (p=0.006), and persons from whom pus (p=0.000), catheter tip (p=0.000), urine (p=0.000), ENT swab (p=0.002) were collected (Table 3). Logistic regression confirmed the variables found in the univariate analysis. For example, patients aged 60 years and older with OR= 11.31 and 95% CI= (3.14; 40.68), male with OR= 4.59 and 95% CI= (1.16; 18.23), patients from whom pus specimens were collected with OR= 19.33 and 95% CI= (2.93; 127.45), patients from whom clinical isolates from catheter tip specimens were collected with OR= 13.16 and 95% CI= (1.78; 97.17), patients from whom ENT swabs were collected OR=261.87 and 95% CI= (1.84; 3724.95), patients infected with more than one bacterium OR=41.61 and 95% CI= (8.66;199.85), patients from whom pleural fluid samples were collected OR=118.76 and 95% CI= (4.75; 2970.84) (Table 4).

Discussion     

Demographics characteristics of persons infected with Pseudomonas aeruginosa and multidrug-resistant Pseudomonas aeruginosa: in terms of individuals, the mean age was 34.75 years with a standard deviation of 26.12, ranging from 1 to 98 years, with 21.52% of individuals being 60 years and older. This finding is similar to that reported by Nabina et al. [19]. Dash et al. in Saudi Arabia found a proportion of hospitalised individuals of 70%, lower than 56% in our study [19].

Characteristics of clinical isolates: a total of 891 (17.15%) of the 5195 clinical isolates tested were positive. This is consistent with the results of Sakr et al. [20]. Of the 64 clinical isolates of Pseudomonas aeruginosa, 43.75% were found in pus, 29.69% in catheter samples, 9.38% in urine, 6.2% in blood, and 4.69% in stool. These results are similar to those of Dash et al. who performed a similar study to ours with over 6000 samples [19].

Prevalence of infection by Pseudomonas aeruginosa and multidrug-resistant Pseudomonas aeruginosa: in this study, the prevalence of infection by PA is 1.11% and 0.46% for MDR-PA. This is lower than that of the vast majority of authors. For example, Restrepo et al. reported 4.2% and 2%, respectively [21].

Frequency of colonisation of clinical isolates by Pseudomonas aeruginosa and multidrug-resistant Pseudomonas aeruginosa: of all the 64 strains of PA, 37.5% were MDR-PA. Compared to some studies such as Dash et al. who reported 84.7%, this MDR rate seems to be very low [19]. It´s similar to values found by several authors such as Micek et al. [22], Senthamarai et al. [23] who reported 41.5%. In contrast, it´s high compared to some authors such as Gill et al., who found 23% [24]. Our study identified clinical isolates colonised with MDR-PA. Some clinical isolates had a high rate of MDR-PA colonisation (catheter, ENT swab, pleural and pus). Others contained a significant proportion of MDR-PA (blood, stool and ascites fluid). Cerebrospinal fluid and joint synovial fluid did not contain any PA strains. This distribution seems to be coherent as these clinical isolates are related to the organs commonly affected by Pseudomonas aeruginosa (the skin, the lung, the digestive system, the haematopoietic system and the urinary system) [25]. However, these results contradict many studies carried out elsewhere, where urine is one of the clinical isolates from which MDR-PA strains are most commonly found [10,14,26].

Antibiotic resistance rates: our rate of resistance to Imipenem (23.81%) seems to be lower than that of most authors in the world. For example, Tumbarrello et al. found 37.7% [27]. Tircacillin alone had a resistance rate of 62.96% (95% CI: 48.74-75.71), and piperacillin 53.45% (95% CI: 39.87-66.66). However, when combined with clavulanic acid and tazobactam, respectively, the resistance rates dropped dramatically to 33.33% CI 95% (20.76-47.92) and 27.08% CI95% (15.28-41.25). This trend was also observed by Tumbarello et al. who found resistance rates to ticarcillin, tircacillin+clavulanic acid, piperacillin and piperacillin+tazobactam of 55.4%, 39.6%, 28.8% and 25.5% respectively [27]. Resistance rates were lowest for cephalosporins. For example, ceftazidime had a resistance rate of 27.87% (95% CI 17.15-40.83), and the antibiotic with the highest susceptibility rate was cefepime (18.97% (95% CI 9.87-31.41). Tumbarello et al. found much higher rates for ceftazidime, 48 (45.3%), and cefepime, 47 (44.3%) [27]. The aztreonam resistance rate was 52.73% (95% CI 38.80-66.35), which differs from 32.6% found by Dash et al. and 25.4% found by Gasink [19,28]. However, our result is similar to that found in Iran by Rabiei et al. [29].

Aminoglycopeptide resistance was very high for tobramycin 71.43, 95% CI: 56.74; 83.42) and lower for amikacin (37.50%, 95% CI: 24.92; 51.45). Several authors found very high rates of resistance to tobramycin. On the other hand, others found rather low rates, such as Lila et al. or Ndip et al. in Cameroon, in Buea [30,31]. This finding contrasts with that of Tumbarello et al. in 2013, who reported a higher sensitivity of amikacin to PA of 7.5% [27]. However, our value follows the same trend as Javiya et al. who found 66% tobramycin resistance [32]. The resistance rate of ciprofloxacin (40.82%, 95% CI (27.00-55.79)) found in our study is similar to many studies, such as the study by Tumbbarello et al. who found 41.5% [27], and Lila et al. who found 45% [31]. However, this rate was much higher than those found in an African study by Tadesse et al. who found a resistance rate of 16.1% (95% CI 2-38.4) in a review in Africa [13].

Associated risk factors for infection by MDR-PA: in univariate analysis, hospitalisation was found to be a risk factor in our study. This was also the case in Aloush et al. [33]. However, this variable did not emerge in the multivariate analysis. The other associated risk factor was the patients from whom pus samples were taken, OR 19.33, 95% CI (2.93; 127.45). The patients from whom pus samples were taken were likely to be those with a wound and chronic hospitalisation as surgical patients, diabetic patients, and patients with pressure sores. These findings are in line with many authors. We could cite Ertugrul et al. and Alhussain et al. who showed that diabetic or surgical wounds are risk factors for MDR-PA infection [34,35]. The patients from whom a catheter tip was collected were at risk of infection with MDR-PA, with an OR 13.16, CI 95% (1.78; 97.17). This result is in agreement with Willmann et al. Burnham et al. and Kaur et al. [36-38].

Another risk factor is patients from whom an ENT swab was taken, with OR 261.87, 95% CI (1.84; 3724.95). These patients are those with mainly ENT infections. In the same vein, patients from whom pleural fluid was collected were associated with MDR-PA infection with OR 118.76, 95% CI (4.75; 2970.84). These last two factors are related to respiratory diseases. These results are in line with some authors such as Idigo et al. [39]. Patients infected with more than one bacterium were more likely to be infected with MDR-PA, OR 41.61, 95% CI (8.66; 199.85). Co-infection could be related to patients with another infection or sepsis in addition to infection by MDR-PA or PA. This finding is similar to that reported by Hoang et al. [40]. Age 60 years and older was associated with MDR-PA infection with an OR of 11.31, 95% CI (3.14; 40.68). This was also the case for male gender. These findings are similar to those found by Montero et al. [41].

Limitations: a strength of retrospective review is that it can collect data for many patients. It is also a strength that patients are not selected (i.e., this is the 'real world') [42]. Our study was retrospective, with very poor archiving and poorly kept records, which did not allow us to have complete data to identify other variables. For example, we do not have data on the services to which patients were admitted or the reason for their consultation. The other major problem is that we can't ignore the possible influence of readmission cases. On the other hand, we can't avoid the problem of misidentification of Pseudomonas species in routine testing in clinical laboratories. For example, Tohya et al. found that most clinical isolates identified as P. putida or P. fluorescens were misidentified in clinical laboratories [43]. Another limitation of our study is that we were not able to include certain cofactors that would have allowed us to examine the effects of social factors, such as social level, educational level, and hospital services. This is because these data were not always available or reliably recorded in the registers we used.

Conclusion     

Colonisation/infection by MDR-PA is feared in our context. The antibiotics with the lowest resistance rates were cefepime, piperacillin/tazobactam and imipenem, and the highest resistance rates were tobramycin and ticarcillin. Catheter-tipped patients and those from whom pus was collected, hospitalisations, age 60 years and older, and male gender were the main risk factors identified for infection with MDR-PA. This fact, found in a retrospective study, gives rise to two hypotheses. Firstly, the USI and the surgical unit are services where there is a high incidence of MDR-PA. Secondly, patients with chronic diseases, bedridden patients, and patients undergoing invasive procedures are the most exposed. To this end, the routine surveillance system should be strengthened to better collect and collate data on antimicrobial resistance, especially in patients with these risk factors.

Competing interests     

The authors declare no competing interests.

Authors' contributions     

Serge Eyebe and Arthur Mbida initiated the protocol. Serge Eyebe designed the study and drafted the final protocol. Serge Eyebe, Célestin Ayangma, Carolle Nsaamang Eyebe, Amani Adjidja, and Stéphane Kona performed the critical review, adapted the methodology, and drafted the literature review. Carolle Nsaamang Eyebe, Célestin Ayangma, Christella Iroume, Ferdinand Ndom, and Stéphane Kona performed the data collection. Carolle Nsaamang Eyebe and Célestin Ayangma checked the quality of the microbiological data. Serge Eyebe checked the quality of the data. Serge Eyebe performed data analysis, interpretation, and drafted the article. Serge Eyebe, Ferdinand Ndom, Carolle Nsaamang Eyebe, Serge Eyebe, Emilia Lyongha, Christella Iroume, Amani Adjidja, and Hugues Nana-Djeunga reviewed the draft article. Arthur Mbida, Alain Bertrand Dongmo, and Pierre Ongolo-Zogo supervised the publication and carried out the quality assessment before submission. Arthur Mbida and Alain Bertrand Dongmo critically revised the article for important intellectual content. Arthur Mbida made the final approval of the manuscript and revisions. All the authors have read and approved the final version of this manuscript.

Acknowledgments     

The authors would like to thank the health workers, the administrative staffs and the Directors of the University Teaching Hospital and the Military Hospital of Yaoundé, Cameroon.

Tables and figures     

Table 1: frequency distribution according to bacterial species of the 1008 bacterial strains found in the 5195 clinical isolates studied between 1^st January 2018 and 31^st December 2021 in two tertiary hospitals in Yaoundé

Table 2: demographic characteristics and prevalence of Pseudomonas aeruginosa and multidrug-resistant Pseudomonas aeruginosa infection among 4155 patients admitted to two Yaoundé tertiary hospitals between 1^st January 2018 and 31^st December 2021, with at least one antibiogram performed

Table 3: univariate analysis of 4155 patients who were admitted to two tertiary hospitals in Yaoundé between the 1^st of January 2018 and the 31^st of December 2021, and of whom at least one antibiogram was performed

Table 4: logistic regression performed on 4155 patients infected with or without multidrug-resistant Pseudomonas aeruginosa who were admitted to two tertiary hospitals in Yaoundé between 1^st January 2018 and 31^st December 2021 and who had at least one antibiogram.

Figure 1: frequency of Pseudomonas aeruginosa and multidrug-resistant Pseudomonas aeruginosa colonisation (%) of 5195 clinical isolates collected between 1^st January 2018 and 31^st December 2021 in two Yaoundé tertiary hospitals

Figure 2: antibiotic resistance rates (%) of Pseudomonas aeruginosa in clinical isolates colonised by Pseudomonas aeruginosa (n=64) among 5195 clinical isolates collected in two tertiary hospitals in Yaoundé between 1^st January 2018 and 31^st December 2021

References