Prevalence of antibiotic-resistant bacteria amongst dogs in Africa: A meta-analysis review

Antimicrobial resistance (AMR) is a global public health threat for both human and veterinary medicine. Increasing evidence suggests that animals are important sources of AMR to humans; however, most of these studies focus on production animals. In order to determine the pattern of AMR in pets, mainly in dogs in Africa, a meta-analysis was performed with AMR studies conducted in African countries and published between January 2000 and January 2021 in four databases: Medline (PubMed), Scopus, Cab abstract and Google Scholar. Seven bacterial strains, namely Staphylococcus aureus, Escherichia coli, Salmonella spp., Pseudomonas aeruginosa, Streptococcus pyogenes, coagulase-negative Staphylococcus (SNC) and Staphylococcus pseudintermedius were included in this study. A total of 18 out of 234 indexed articles met the study criteria. The results revealed that multiple bacteria were resistant to various commonly used antibiotics including enrofloxacin, ciprofloxacin, gentamicin, amoxicillin, clavulanic acid, cotrimoxazole, streptomycin, tetracycline and chloramphenicol. Concerning multidrug resistance, E. coli strains came first with the highest prevalence of 98%, followed by P. aeroginosa (92%) and Salmonella spp. (53%). In contrast, the overall prevalence of multidrug resistance was low for S. aureus (18%) and S. pseudintermedius (25%). It is therefore urgent to find, as soon as possible, alternatives to replace these antibiotics, which have become ineffective in controlling these bacteria in dogs in Africa. Moreover, further metagenomic studies are needed to describe the full resistome and mobilome in dogs regardless of the bacteria.


Introduction
The importance of the bond between people and their pets is increasingly recognised and many owners consider their pets to be a member of their family (WSAVA 2020). This is particularly the case with dogs that are highly valued amongst all other pets because of their many benefits to humans such as sheep guarding, detection of drugs and explosives, hunting and safety, breeding and companionship (Daodu et al. 2017). For years, the dog was the most widely employed scentdetector tool for civilian and military purposes. Recently, many studies highlighted the role of canine olfactory ability in the medical field, specifically in detecting different infectious, metabolic and neoplastic conditions including the coronavirus disease 2019 (COVID-19) (Sakr et al. 2021). Trained biodetection dogs are already being used for detecting illicit substances and for forensic purposes. They have also been used for helping to detect cancer (Sharun et al. 2021). A study conducted by Oliva and Johnston (2020) found a buffering effect of dog ownership against loneliness. Despite the benefits of dog ownership, dogs are susceptible to many infectious agents responsible for viral diseases such as rabies, parvovirosis, canine distemper and bacterial diseases, namely leptospirosis, pasteurellosis, skin diseases and parasitic illness such as piroplasmosis, and so on (Ghasemzadeh & Namazi 2015). As a result of their importance, most owners try to take care of their animals to the best of their ability and strive to keep their animals healthy and 'happy' (WSAVA 2020). This has led to a high use of antimicrobials such as antibiotics to treat diseases and injuries in their animals and the use of vaccines for the prevention of certain diseases (Gwenzi et al. 2021). Antibiotic resistance is a global public health problem that could be responsible for more than 10 million deaths per year and thus become the leading cause of mortality by 2050, with an economic cost of $100 billion if left unchecked (O'Neill 2016). The antibiotic agents used are often more closely related to those used in human medicine and antibiotic resistance risks should not be ignored as the close relationship between dog and human subjects presents an Antimicrobial resistance (AMR) is a global public health threat for both human and veterinary medicine. Increasing evidence suggests that animals are important sources of AMR to humans; however, most of these studies focus on production animals. In order to determine the pattern of AMR in pets, mainly in dogs in Africa, a meta-analysis was performed with AMR studies conducted in African countries and published between January 2000 and January 2021 in four databases: Medline (PubMed), Scopus, Cab abstract and Google Scholar. Seven bacterial strains, namely Staphylococcus aureus, Escherichia coli, Salmonella spp., Pseudomonas aeruginosa, Streptococcus pyogenes, coagulase-negative Staphylococcus (SNC) and Staphylococcus pseudintermedius were included in this study. A total of 18 out of 234 indexed articles met the study criteria. The results revealed that multiple bacteria were resistant to various commonly used antibiotics including enrofloxacin, ciprofloxacin, gentamicin, amoxicillin, clavulanic acid, cotrimoxazole, streptomycin, tetracycline and chloramphenicol. Concerning multidrug resistance, E. coli strains came first with the highest prevalence of 98%, followed by P. aeroginosa (92%) and Salmonella spp. (53%). In contrast, the overall prevalence of multidrug resistance was low for S. aureus (18%) and S. pseudintermedius (25%). It is therefore urgent to find, as soon as possible, alternatives to replace these antibiotics, which have become ineffective in controlling these bacteria in dogs in Africa. Moreover, further metagenomic studies are needed to describe the full resistome and mobilome in dogs regardless of the bacteria.
Keywords: prevalence; antibiotic resistance; dogs; Africa; meta-analysis; antimicrobial resistance. opportunity for the two-way transfer of bacterial (commensal and pathogen) or genetic determinants of resistance with associated potential for morbidity and mortality on both sides (Argudín et al. 2017;Rendle & Page 2018). The misuse of antibiotics has been reported in several regions of the world and is at the origin of the antibiotic resistance phenomenon (Kavanagh, Mitra & Basu 2021;Le Huy et al. 2020). In Africa, numerous recent studies in different countries have revealed high prevalence of resistant bacterial strains in dogs Zewdu et al. 2019). A survey conducted in Nigeria revealed that 82% of dog owners had, at different times, body-to-body contact of less than 50 cm between their face and the dog's body (Daodu et al. 2017). This close contact can promote the exchange of resistant pathogens via saliva, urine, faeces, aerosols, skin and thus amplify the phenomenon of antibiotic resistance in humans and exchange of antibiotic resistance between humans and dogs. Given the increase prevalence of antibiotic-resistant bacteria and the risk of bacterial transmission between dogs and their owners, it is important to understand the overall level of antibiotic resistance in the dog population. With this knowledge, veterinarians can then make recommendations to protect the health of both their patients and clients. This meta-analysis aims to assess the overall prevalence of antibiotic-resistant bacterial strains in dogs in Africa to develop and implement local and continental control programmes for drug-resistant bacterial strains and their spread.

Literature search
A systematic search was conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) protocol items on the phenomenon of antibiotic resistance in companion animals in Africa between January 2021 and March 2021. Searches were conducted in four databases, Medline (PubMed), Scopus, Cab abstract and Google Scholar by indexing several keywords in the title and abstract of the articles stored in these different databases. The search terms included 'Companion animals' and 'Antimicrobial resistance ' and 'Africa' or 'dog' and 'Antimicrobial resistance' and 'Africa' or 'dog' and 'Antibacterial resistance' and 'Africa' or 'dog' and 'Antibiotic resistance' and 'Africa' or 'Multidrug resistance' and 'Africa' or 'dog' and these term with each of 54 African countries. All studies (in English or French) on bacterial antimicrobial resistance (AMR) in dogs published between January 2000 and January 2021 were included.

Data extraction
After excluding irrelevant studies (non-African countries, antibiotic resistance in other animal species) and duplicates, the remaining studies were fully read in order to extract the data needed for this review: name of the first author, year of publication, country where the study was carried out, sources of samples, sample size, methods of analysis (disc diffusion method or molecular method), number of positive cases, number of bacterial strains tested and prevalence of antibiotic resistant strains. Figure 1 presents the flow chart of the process of identifying studies.

Statistical analyses
Data extracted from all selected publications were directly inserted into the Excel spreadsheet and subjected to metaanalysis in R version 3.1 (R Core Team 2016) using the Meta and Metafor packages (Schwarzer 2007;Viechtbauer 2010). The Mantel-Haenszel fixed-effects model (FEM) and the Der Simonian and Laird random-effects model (REM) were used to generate an overall estimate of the prevalence of antibiotic resistance.
Heterogeneity between the different studies was assessed using the I 2 statistic of Higgins et al. (2003). I 2 values of 25%, 50% and 75% were considered to have a low, moderate and high degree of heterogeneity, respectively. When heterogeneity between studies is low, the fixed effect model estimates were considered, whilst the random effect model was used to generate the overall prevalence in case of high heterogeneity. Finally, to summarise the results of the metaanalysis, a forest plot showing the different prevalence estimates of multidrug resistance (MDR) bacteria across studies, a confidence interval and an overall prevalence was generated for each of the two models used.

Ethical considerations
This article followed all ethical standards of research without direct contact with human or animal subjects.

Characteristics of eligible studies
The characteristics of the studies selected for the metaanalysis are presented in Table 1. A total of 18 eligible studies were selected for the meta-analysis out of 234 indexed articles. The selected studies took place in countries such as Nigeria (8 studies), South Africa (3), Kenya (2), Tanzania (1), Ethiopia (2), Egypt (1) and Tunisia (1) (Figure 2). The sources of contamination used in these studies are mainly, in order of importance, rectal, nasal, oral, skin, vaginal, ear canal and wound swabs. The methods used by these authors to study antibiotic resistance are disk diffusion and molecular methods.

Prevalence of antibiotic-resistant bacterial strains isolated from dogs in Africa
Many bacterial strains isolated from dogs and identified either biochemically or genotypically have been tested for antibiotic efficacy in eligible studies, however only seven strains with major public health impacts are included in this meta-analysis. These are Staphylococcus aureus, Escherichia coli, Salmonella spp., P.aeruginosa, Streptococcus pyogenes, coagulase-negative Staphylococcus and Staphylococcus pseudintermedius. Table 2 shows the overall prevalence of antibiotic-resistant S. aureus isolates from dogs in Africa. Analysis of this table reveals a higher prevalence (> 52%) of S. aureus isolates resistant to the following antibiotics: nalidixic acid, streptomycin, methicillin, cotrimoxazole, ampicillin, amoxicillin, kanamycin, tetracycline, penicillin G and pefloxacin. These isolates showed a low prevalence of resistance to the other antibiotics tested such as oxacillin, gentamycin, enrofloxacin, norfloxacin, erythromycin and a higher sensitivity to ofloxacin and ciprofloxacin. The prevalence of multidrug resistant strains was 18%. Table 3 shows the overall prevalence of antibiotic resistant E.coli strains isolated from dogs in Africa. An analysis of this table shows that the E. coli isolates tested were resistant to cefuroxime, cefotaxime, ceftazidime, ceftriaxone and clindamycin, all with 100% of prevalence. These strains were also resistant to kanamycin, ampicillin, amoxicillin, penicillin G, cefalotin, nalidixic acid, gentamicin, cotrimoxazole, tetracycline, doxycycline, chloramphenicol, nitrofurantoin and tylosin with more than 50% of prevalence. In addition, less than 50% of the strains tested were resistant to the other antibiotics evaluated in the selected studies. Beta-lactams, macrolides and

Prevalence of antibiotic-resistant Salmonella strains
The overall prevalence of antibiotic resistant Salmonella strains isolated from dogs in Africa is summarised in Table 4.    An analysis of this table shows that a significant proportion of the Salmonella strains tested were resistant to ampicillin, penicillin G, ceftazidime, cefotaxime, gentamicin, lincomycin, cotrimoxazole, oxytetracycline, streptomycin and vancomycin, with prevalences ranging from 53% to 100%. Regarding sulfamethazine, chloramphenicol, tetracycline, cefalotin, doxycycline and amoxicillin andclavulanic acid, less than half of the strains tested were resistant. However, none of the Salmonella strains tested were resistant to the quinolone antibiotics, particularly norfloxacin, ciprofloxacin and enrofloxacin. The prevalence of multidrug resistant Salmonella strains is 53%. Table 5 shows the prevalence of antibiotic-resistant P. aeruginosa strains isolated from dogs in Africa. From analysis of this table, all P. aeruginosa isolates tested are resistant to ampicillin, amoxicillin, cloxacillin, cephalexin, flucloxacillin, cefuroxime, gentamicin, neomycin, norfloxacin, nalidixic acid, erythromycin, cotrimoxazole and nitrofurantoin. Prevalences of more than 50% in P. aeruginosa strains are resistant to amoxicillin and clavulanic acid, penicillin G, carbenicillin, piperacillin, ceftazidime, enrofloxacin, orbifloxacin, lincomycin, tetracycline, doxycycline, chloramphenicol and tylosin. Also, there is a low prevalence of P. aeruginosa strains detected to be resistant to imipenem, tobramycin and ciprofloxacin. However, these strains are sensitive to ofloxacin. Resistance of P. aeruginosa is marked in all families of antibiotics tested and the prevalence of multidrug resistant strains is 92%. Table 6 shows the overall prevalence of antibiotic resistant S. pyogenes strains isolated from dogs in Africa. Streptococcus. pyogenes isolates are resistant to ampicillin, amoxicillin, penicillin G, flucloxacillin, cefuroxime, streptomycin, neomycin, nalidixic acid, erythromycin, tetracycline, chloramphenicol and nitrofurantoin with an overall prevalence of 100.00%. A high prevalence of norfloxacin-resistant S. pyogenes (87.50%) is also recorded. But these isolates are susceptible to cloxacillin, cephalothin, ciprofloxacin, enrofloxacin and ofloxacin. In addition, resistance in S. pyogenes affected all families of antibiotics tested.

Prevalence of antibiotic resistant coagulase negative Staphylococcus strains
The overall prevalence of antibiotic resistant coagulase negative Staphylococcus (SCN) strains isolated from dogs in Africa is shown in Table 7. Analysis of this table shows that all isolates tested are resistant to ampicillin, penicillin G, cotrimoxazole and tetracycline. Furthermore, high prevalences of resistant SCN were also recorded for amoxicillin, amoxicillin and clavulanic acid, cefuroxime, nalidixic acid and chloramphenicol. For methicillin, streptomycin, ceftriaxone, gentamicin and enrofloxacin, less than half of the isolates tested were found to be resistant. In addition, isolates were found to be susceptible to ciprofloxacin, norfloxacin and pefloxacin. Table 8 shows the overall prevalence of antibiotic-resistant S. pseudintermedius strains isolated from dogs in Africa. Analysis of this table shows that high prevalences of antibiotic resistant S. pseudintermedius isolates are found only in ampicillin (66.07%), penicillin G (53.19%) and clindamycin (51.79%). On the other hand, these strains are mainly susceptible to tobramycin, teicoplanin, vancomycin and mupirocin. However, a low prevalence  Figure 3 summarises the prevalence of the seven bacterial strains according to the antibiotics they have in common.   (25.00%), Salmonella (20.00%) and S. pseudintermedius (7.16%), which were sensitive to chloramphenicol, the other strains were resistant to this antibiotic.

Prevalence of multidrug resistant strains in dogs
The prevalences of multidrug resistant strains isolated from dogs in Africa are presented in Figure 4. From the analysis of this table it is obvious that the E. coli strain has the highest prevalence of multidrug resistance followed by Streptococcus CN first then P. aeruginosa then S. pyogenes. In contrast, the prevalence is low for S. aureus and S. pseudintermedius (Appendix 1).

Discussion
Antibiotic resistance is a growing phenomenon in African countries and in companion animals, notably dogs. Many antibiotics used in veterinary medicine are similar to those used in human medicine. The present systematic review assessed the overall prevalence of antibiotic-resistant bacterial strains isolated from dogs in Africa and found that many strains were multidrug resistant to most common and current used antibiotics (ampicillin, ampicillin/cloxacillin, tetracycline, penicillin, amoxicillin, cotrimoxazole, chloramphenicol and ciprofloxacin) used in humans (Okpara et al. 2018). The high resistance of S. aureus strains to ampicillin found in dogs was also observed in humans (100.0%) and pigs (100.0%) in Tanzania by Katakweba et al. (2016). But for tetracycline, resistance is moderate in humans (45.5%) and pigs (50.0%) in Tanzania, whilst it is high in dogs. In the Democratic Republic of Congo, resistance prevalences of 33.00% -72.00% against tetracycline, 5.00% -54.00% against cotrimoxazole, 31.00% against gentamicin, 26.00% -69.00% against erythromycin and 20.00% -59.00% against ciprofloxacin have been reported in humans (Lupande-Mwenebitu et al. 2020). Our study showed contrasting findings: whilst the results are similar to those obtained in Europe by Moyaert et al. (2019), the prevalence of penicillin G-resistant isolates in the present analysis (56.23%) is lower than that obtained in Europe (65.20%). Furthermore, strains from the seven studied African countries were resistant to cotrimoxazole (73.77%) and tetracycline (60.09%), whilst those isolated from European countries were found to be sensitive to the same antibiotics, with low resistance  prevalences equal to 0.00% and 13.00%, respectively (Moyaert et al. 2019). Staphylococcus aureus strains were resistant to nalidixic acid, streptomycin, methicillin, cotrimoxazole, ampicillin, amoxicillin, kanamycin, tetracycline, penicillin G and pefloxacin.
With regard to multidrug resistance, the prevalence in dogs in the present study (18.0%) is lower than that recorded in humans in Nigeria (68.0%) (Ogundipe et al. 2020) and in cats (28.6%) in South Africa (Qekwana et al. 2017).
In this study, E. coli strains were resistant to 18 antibiotics. These were cefuroxime, cefotaxime, ceftazidime, ceftriaxone, clindamycin, kanamycin, ampicillin, amoxicillin, penicillin G, cefalotin, nalidixic acid, gentamicin, cotrimoxazole, tetracycline, doxycycline, chloramphenicol, nitrofurantoin and tylosin. These results differ from those obtained in Europe by Moyaert et al. (2019), where E. coli strains tested were susceptible to ampicillin, cotrimoxazole and tetracycline, marked by low prevalences of resistance of 36.40%, 21.20% and 18.2%, respectively. However, the sensitivity of E. coli strains to orbifloxacin observed in this study is also reported in the study conducted in Europe by Ludwig et al. (2016) where the prevalence of resistant isolates was low, 3.7%. Furthermore, in the meta-analysis study conducted by Emami, Javanmardi and Pirbonyeh (2020), 73.00% of E. coli strains isolated from pregnant women in Africa were resistant to ampicillin and 52.00% of E. coli strains from Asia. These prevalences are lower than the prevalence in dogs in the present study (86.49%). Compared with gentamicin (30.54%), amoxicillin and clavulanic acid (45.11%) and tetracycline (70.27%), the prevalences of resistant E. coli in dogs are higher than those in humans in Africa, which are 23.00%, 44.00% and 60.00%, respectively, but the prevalence of ciprofloxacin resistant E. coli in dogs (6.05%) is very low compared with that in humans (26.00%).
Regarding antibiotic resistance of Salmonella strains, the results are not in agreement with those obtained in Europe by Bataller et al. (2020) who reported that Salmonella serovars are susceptible to ceftazidime, cefotaxime, gentamicin and cotrimoxazole. Nevertheless, the results obtained for the sensitivity of Salmonella strains to ciprofloxacin are similar to those reported by the same authors.
For P. aeruginosa strains, the results observed differ from those obtained in the studies conducted in Europe by Ludwig et al. (2016), where the prevalence of P. aeruginosa strains resistant to gentamicin and enrofloxacin are low, 18.80% and 18.20%, respectively.
Regarding coagulase negative Staphylococcus strains, the results recorded in this study are contrary to those reported by Wedley et al. (2014) who observed the sensitivity of United Kingdom (UK) coagulase negative Staphylococcus isolates to cotrimoxazole and tetracycline, with resistance prevalences of 22.50% and 27.50%, respectively. In contrast, methicillin, which is effective on African isolates with a low prevalence of resistance (33.33%), was found to be ineffective on UK strains, with a resistance prevalence equal to 100.00% (Wedley et al. 2014).
As for S. pseudintermedius, the result observed for penicillin G activity on isolates in this study is different from that obtained in Europe by Moyaert et al. (2019) where tested S. pseudintermedius isolates were susceptible to this antibiotic, with a resistance prevalence equal to 20.0%. However, this result is similar to the one obtained in Japan by Bardiau et al. (2013), but the prevalence recorded in our study for this antibiotic (53.19%) is lower than that observed in Japan (> 95.0%). Compared with tetracycline and enrofloxacin, the susceptibility of S. pseudintermedius strains observed in this study is also found in European isolates by Moyaert et al. (2019). For clindamycin, the prevalence obtained (51.79%) is lower than that obtained in work carried out in European and North American regions by Perreten et al. (2010), where the prevalence of resistant strains is equal to 89.3%.
There are several reasons for the differences in these results: the level of knowledge about antimicrobials and their use, the attitudes of owners towards dog healthcare, the spatial epidemiology of AMR, the level of enforcement of existing legislation on antimicrobials in veterinary medicine, and the conditions of access to veterinary services in given regions. In a study conducted in Nigeria by Okpara et al. (2018), a low level of knowledge on antimicrobials and their use was observed amongst the respondents; 64.4% of the respondents administered antimicrobials to their animals themselves and 60.6% used antimicrobials without veterinary prescriptions. Also, more than half of the respondents (51.5%) had never used veterinary services for their animals whilst 23.1% rarely used them. From the same study, 71.3% of the respondents mentioned that they did not have veterinary practices in their communities.
This can lead to problems such as misdirection of antimicrobials, over-or under-dosing, and poor storage and handling of antimicrobial agents. Regarding the spatial epidemiology of AMR, the study conducted in South Africa by Qekwana et al. (2019) revealed the similarity of prescribing practices amongst veterinarians in the study areas. Given that the realities of African countries differ from those of other continents in several respects, the differences in prevalences may, therefore, be related to differences in prescribing practices. The lack of surveillance programmes on antimicrobial use, as well as the lack of enforcement of existing legislation with appropriate sanctions for violators in African countries may account for the differences in results.
Such high prevalences of multidrug resistant strains amongst the investigated microbial strains constitute a real health problem that could lead to therapeutic failures in these animals and humans, especially their owners. Indeed, the sharing of the resistance genes of these strains between these animals and their owners could be a source of treatment failure of these targets in case of diseases (Feiyang et al. 2021).
Educating dog owners about the impact of antibiotic resistance on animals and nearby humans is essential. In addition, appropriate integrated programmes to control AMR in Africa, especially for companion animals, should be implemented.

Conclusion
Antibiotic resistance is a serious problem worldwide. This phenomenon is increasing in Africa, in pets and particularly in dogs, which are relatively closer companions to humans. All strains investigated are resistant to ampicillin and Penicillin G, but are sensitive to ciprofloxacin. In relation to the other antibiotics, the level of resistance varies according to the strain. Multidrug resistance is much higher in E. coli, P. aeruginosa and Salmonella strains compared with S. aureus and S. pseudintermedius strains. The high prevalence of resistance observed for strains of E. coli, Salmonella, P. aeruginosa, S. pyogenes, coagulase-negative Staphylococcus, S. aureus and S. pseudintermedius, which are zoonotic strains, should not be overlooked. As more than one of these bacteria can be found in the same individual and the antibiotics used in animals are practically used in humans for the treatment of infections, it is important to use these antibiotics rationally, so that control of one bacterial strain does not lead to resistance in the other. In view of the multidrug resistance observed in each of these strains, it is important and immediate need to implement measures to prevent the exchange of pathogens between dogs and their owners and to apply good hygienic practices and promote alternatives such as appropriate medicinal plants to combat these different strains in Africa.