Dynamics and numerical simulations to predict empirical antibiotic treatment of multi-resistant Pseudomonas aeruginosa infection

DYNAMICS AND NUMERICAL SIMULATIONS TO PREDICT EMPIRICAL ANTIBIOTIC TREATMENT OF MULTI-RESISTANT Pseudomonas aeruginosa INFECTION Javier L´ opez-de-la-Cruz∗,∗∗ Departamento de Ecuaciones Diferenciales y An´alisis Num´erico C/ Tarfia s/n, Facultad de Matem´aticas, Universidad de Sevilla 41012 Sevilla, Spain Mar ´ ıa P´ erez-Aranda∗∗ Departamento de Qu´ Imica Org´anica y Farmac´eutica and Departamento de Microbiolog´ıa y Parasitolog´ıa C/ Profesor Garc´ıa Gonz´alez, s/n, Facultad de Farmacia. Universidad de Sevilla 41012 Sevilla, Spain Ana Alcudia∗and Bel´ en Begines Departamento de Qu´ımica Org´anica y Farmac´eutica C/ Profesor Garc´ıa Gonz´alez, s/n, Facultad de Farmacia. Universidad de Sevilla 41012 Sevilla, Spain Tom´ as Caraballo Departamento de Ecuaciones Diferenciales y An´alisis Num´erico C/ Tarfia s/n, Facultad de Matem´aticas, Universidad de Sevilla 41012 Sevilla, Spain Elo ´ ısa Pajuelo Departamento de Microbiolog´ıa y Parasitolog´ıa C/ Profesor Garc´ıa Gonz´alez, s/n, Facultad de Farmacia. Universidad de Sevilla 41012 Sevilla, Spain and Pedro J. Ginel Departamento de Medicina y Cirug´ıa Animal Edificio Hospital Cl´ınico Veterinario, Campus de Rabanales, Facultad de Veterinaria, Universidad de C´ordoba 14071 C´ordoba, Spain (Communicated by the associate editor name) 2010 Mathematics Subject Classification. Primary: 34K28, 34A12 Secondary: 92B05. Key words and phrases. numerical simulations, SIRI model, multi-resistant, Pseudomonas aeruginosa. ∗Corresponding authors: [email protected], jlop[email protected]. ∗∗ Both authors contributed equally to this study. This work has been partially supported by Project 2004/0001203 from Ministerio de Educaci´on y Ciencia (Spanish government), Project PGC2018-096540-B-I00 from Ministerio de Ciencia, Innovaci´on y Universidades (Spanish government) and Project US-1254251 from Consejer´ıa de Econom´ıa y Conocimiento (Junta de Andaluc´ıa). 1 2 PREDICT TREATMENT OF MULTI-RESISTANT Pseudomonas aeruginosa INFECTION Abstract. This work discloses an epidemiological mathematical model to predict an empirical treatment for dogs infected by Pseudomonas aeruginosa. This dangerous pathogen is one of the leading causes of multi-resistant infections and can be transmitted from dogs to humans. Numerical simulations and appropriated codes were developed using Matlab software to gather information concerning long-time dynamics of the susceptible, infected and recovered individuals. All data compiled from the mathematical model was used to provide an appropriated antibiotic sensitivity panel for this specific infection. In this study, several variables have been included in this model to predict which treatment should be prescribed in emergency cases, when there is no time to perform an antibiogram or the cost of it could not be assumed. In particular, we highlight the use of this model aiming to become part of the convenient toolbox of Public Health research and decision-making in the design of the mitigation strategy of bacterial pathogens. 1. Introduction Multi-drug resistant infections, typically caused by a bacterium who is resistant to more than one antibiotic, are an emerging problem and constitute an important threat to public health (see [1,2,3]), in fact, more than 700,000 people die per year due to this kind of infections currently. Latest 2019 World Health Organization report claims for urgent action to avert the antimicrobial resistance crisis and highlights that, if no action is taken, drug-resistant diseases could cause 10 million deaths each year by 2050 and damage the economy as catastrophic as the 2008-2009 global financial crisis. In this sense, these types of infections are a defiant therapeutic challenge in both human and veterinary medicine, due to the complex mechanisms that bacteria have developed to resist to antibiotics and their capacity to transmit this resistance both vertically (to their descendants) and horizontally (to other bacteria) (see [1,2]). In particular, Pseudomonas aeruginosa is one of the most frequent pathogen that acquires or develops antibiotic multi-resistance. This Gram-negative rod is very ubiquitous in the environment, can persist in water and soil with minimal nutritional presence and can tolerate several conditions of humidity and temperature (see [4,5]). Figure 1. Microscopic image of an exudate smear from the ear canal of a dog with otitis externa. The presence of numerous rods as well as inflammatory cells can be observed PREDICT TREATMENT OF MULTI-RESISTANT Pseudomonas aeruginosa INFECTION 3 As revealed, this capacity to resist extreme situations as well as the high capacity to develop multi-resistance after exposure to antimicrobials and cross-resistance between agents, makes Pseudomonas aeruginosa a leading cause of nosocomial infections, responsible for 10% of all hospital acquired infections. In addition, these are often severe and life threatening, and very often very difficult to treat because of the limited susceptibility to antimicrobial agents (see, for instance [1,3,4]). In particular, in dogs this pathogen is often a cause of otitis and pyoderma. Due to the close physical contact with their owners , dogs are a potential reservoir to multiresistant Pseudomonas aeruginosa, which could lead to zoonotic (animal to human) transmission (see [5,6,7]). This work compiles 40 cases of Pseudomonas aeruginosa otitis in dogs from the south of Spain in the last five years and their susceptibility to an extended panel of antibiotics (see Table 8in Appendix). Although the best procedure to treat these infections is to perform an antibiogram study, unfortunately that takes five to seven days (see [2,3,4]) and not always economic resources are available, specially in Third World Countries where sometimes the sanitary system has limited funding. Significantly, there are some critical clinical circumstances in which the treatment cannot be delayed due to the severity of the symptoms and the possible risk that may lead to an irreversible situation to the patient. Figure 2. Purulent otitis externa caused by Pseudomonas aeruginosa in a dog In view of the facts described before, it would be crucial to have a powerful tool that could help practitioners to predict the best empirical treatment possible for this type of infection. Therefore, a professional could prescribe an antibiotic treatment as soon as the infection is detected, in case there is no time available to perform the antibiogram. In this sense, the goal of the present work is to predict, using a mathematical model, the best empirical antibiotic treatment to avoid the bacterial development of resistance mechanisms. The classical reference about epidemics in Mathematics is due to Kermack and McKendrick (see [8,9,10]) who proposed a compartment model based on simple assumptions on the flow rates between different groups of members in some population. Basically, the population is divided in three groups: susceptible, the 4 PREDICT TREATMENT OF MULTI-RESISTANT Pseudomonas aeruginosa INFECTION individuals who are not infected yet but susceptible to the disease; infected, who are the infected individuals; and recovered, individuals who have been infected in the past and cannot be re-infected. From now on, both groups will be denoted by S,Iand R, respectively. Since Kermanck and McKendrick proposed the so-called SIR model, many authors have worked to develop innovations which improve the original SIR model in order to describe more complex situations [11,12]. Among these more sophisticated models, it is worth mentioning the SIR model with relapse, or the also called SIRI model, whose main difference compared to the SIR one is based on the the fact that recovered individuals could be re-infected. The SIRI model can be mathematically described with the following system of differential equations dS dt = Λ(S+I+R)−βSI −µS,(1.1) dI dt =βSI −(α+κ+µ)I+γR,(1.2) dR dt =κI −(µ+γ)R.(1.3) where S=S(t), I=I(t) and R=R(t) denote the number of susceptible, infected and recovered individuals, respectively. Λ and µare the birth and death rates of the populations, respectively, βis the transmission coefficient of the disease, αdenotes the death rate due to the disease, κdescribes the rate at which the infected individuals become recovered and γis the rate at which the recovered individuals become infected. We also remark that every constant is positive. The different flows between the three groups of the population can be represented as the transfer diagram in Figure 1. Figure 3. Transfer diagram of SIRI model To this end, we consider a population of 192 dogs of which 40 are infected and Matlab is used to make numerical simulations and understand the long-time dynamics of the different groups (susceptible, infected and recovered) of the population depending on time. In addition, we include a code in Matlab to compute the final number of susceptible, infected and recovered individuals in each case as well as the time at which it happens. Therefore, we display tables and bar diagrams compiling the information to predict the most effective antibiotics in every case. The present work is organized as follows: Section 2compiles the materials and methods used in this work. Section 3shows numerical simulations of every case. All the relevant information is depicted on tables and we use bar diagrams to reach a better understanding of the dynamics. Section 4provides a list of the most suitable antibiotics to treat infected dogs. Finally, Section 5includes conclusions Significantly, there is a strong evidence of the utility of this type of mathematical models performed in collaboration with other disciplines, such as health sciences. PREDICT TREATMENT OF MULTI-RESISTANT Pseudomonas aeruginosa INFECTION 5 2. Materials and methods For this work, forty Pseudomonas aeruginosa samples were aseptically collected from dogs with purulent otitis externa and were sent to a specialized laboratory to perform their culture and antibiogram analysis. We refer the reader to Table 8in Appendix to check the whole list of antibiotics that were tested and their rates of resistance. Antimicrobial susceptibility was tested using Kirby-Bauer method, also known as disk diffusion antibiotic sensitivity testing or disk diffusion test according to Clinical Laboratory Standards Institute (CLSI) guidelines (see [13]). Figure 4. Antimicrobial susceptibility assay using disk diffusions test in Pseudomonas aeruginosa culture In addition, Matlab was used to make numerical simulations and program some codes to obtain useful data about the behavior of the population. 3. Mathematical study. Numerical simulations and data In this section we present the mathematical study using a SIRI model (1.1)-(1.3) predict an empirical treatment. The complete panel of antibiotics tested in laboratory, as well as their corresponding resistance rate is presented in Table 8in Appendix. To organize all data obtained based on the resistance rate, the following groups are depicted: •Group 1- Antibiotics 100% resistant. •Group 2- Antibiotics with resistance close to 100%. •Group 3- Antibiotics with resistance between 30% and 50%. •Group 4- Antibiotics with resistance between 20% and 30%. •Group 5- Antibiotics with resistance between 10% and 20%. 6 PREDICT TREATMENT OF MULTI-RESISTANT Pseudomonas aeruginosa INFECTION •Group 6- Antibiotics with resistance between 1% and 10%. •Group 7- Antibiotics with resistance 0% resistant. The percentages corresponding to the groups cited before are represented in Figure 5. 29% 6% 6% 12% 8% 12% 27% Resistance rate 0% Resistance rate 1%-10% Resistance rate 10%-20% Resistance rate 20%-30% Resistance rate 30%-50% Resistance rate close to 100% Resistance rate 0% Figure 5. Resistance rates of antimicrobials For the implementation of the model, three tools are considered essential: •Figures (Figures 6,8,10,12,14,16,18) containing numerical simulations of the SIRI model (1.1)-(1.3) with a code developed in Matlab, which will allow to determine the dynamics of the susceptible, infected and recovered individuals for an estimated period of time. Moreover, it allows observing how the numbers of the different groups of our population change and its asymptotic behavior. •Tables (Tables 1,2,3,4,5,6,7) where the limit number of each group of the population is collected for different values of the transmission coefficient of the disease, as well as the time to reach the corresponding value. These tables are very useful to convert the dynamics information in numerical data. This second tool was implemented due to a code programmed in Matlab. •Bar diagrams (Figures 7,9,11,13,15,17,19) where the final number of each group of the population is presented for different values of the transmission coefficient of the disease. Once the mathematical study is accomplished, it could provide practitioners an appropriate and very convenient panel of the most effective antibiotics, which will provide essential information in case the antibiogram is not available. In order to develop the mathematical study, a population of 192 dogs were included, where 40 of them are infected by Pseudomonas aeruginosa, 152 are susceptible and none of them are recovered at the initial time, considering this initial value as zero. Concerning the parameters in the SIRI model (1.1)-(1.3) the birth and death rates of the population were Λ = 0 and µ= 0, assuming that no dogs were born or died during our study. Since the average duration of this infection is considered around 14 weeks, α= 0 since dog mortality null in this pathology. The recovery PREDICT TREATMENT OF MULTI-RESISTANT Pseudomonas aeruginosa INFECTION 7 rate κand the reinfection rate γdepend on the groups of antibiotics in study (see Section 3). We remark that the transmission coefficient of the disease is unknown but it is a rate. Therefore, every case will be studied using different values of this parameter between 0.1 and 1. 3.1. Resistance 100%. This subsection corresponds to antibiotics with a resistance rate of 100%. In Figure 6we display three different panels showing the dynamics of susceptible, infected and recovered individuals, respectively. In every panel the number of members in each group are plotted depending on time. As the initial number of dogs in the experiment is 192 and only 40 are infected, the initial value of susceptible is 152 and the initial value of infected as 40. Noticeable, there are no recovered dogs at the initial time. 0 days 1 days 2 days 3 days 4 days 5 days 6 days 7 days 0 dogs 50 dogs 100 dogs 150 dogs Dynamics of susceptible individuals 0 days 1 days 2 days 3 days 4 days 5 days 6 days 7 days 0 dogs 50 dogs 100 dogs 150 dogs Dynamics of infected individuals 0 days 1 days 2 days 3 days 4 days 5 days 6 days 7 days 0 dogs 50 dogs 100 dogs 150 dogs Dynamics of recovered individuals Figure 6. Dynamics of susceptible, infected and recovered individuals depending on time (in days) when using an antibiotic with a resistance rate of 100% In Table 1the number of susceptible, infected and recovered individuals at the final time is depicted for different values of the transmission coefficient of the disease β. For instance, when β= 0.1 the limit of susceptible individuals is 0.9892 and this limit is reached after 6.1075 days. The limit of infected individuals is 186.0153 and it is reached after 5.3200 days and, finally, the number of susceptible individuals remains constant zero during all the time. In Table 1we can notice that most of dogs will be infected in approximately 5 days and no dogs will be recovered corresponding to a 100% resistance rate. 8 PREDICT TREATMENT OF MULTI-RESISTANT Pseudomonas aeruginosa INFECTION Susceptibles Infected Recovered Time (days) Individuals Time (days) Individuals Time (days) Individuals β= 0.1 6.1075 0.9892 5.3200 186.0153 0.0000 0.0000 β= 0.2 3.0450 0.9997 2.8175 185.0218 0.0000 0.0000 β= 0.3 2.0300 0.9911 1.8200 183.0880 0.0000 0.0000 β= 0.4 1.5225 0.9821 1.4525 182.0727 0.0000 0.0000 β= 0.5 1.2250 0.9368 1.1025 180.0496 0.0000 0.0000 β= 0.6 1.0150 0.9635 0.9800 179.0448 0.0000 0.0000 β= 0.7 0.8750 0.9183 0.7875 177.0125 0.0000 0.0000 β= 0.8 0.7700 0.8748 0.7350 176.0281 0.0000 0.0000 β= 0.9 0.6825 0.8819 0.6125 174.0724 0.0000 0.0000 β= 1 0.6125 0.8888 0.5950 173.1766 0.0000 0.0000 Table 1. Data for different values of the transmission coefficient of the disease βin case of using antibiotics with a resistance rate of 100% Based on Table 1we present in Figure 7a bar diagram to to illustrate the results obtained. β= 0.1β= 0.2β= 0.4β= 0.6β= 0.8β= 1 0 2 4 6 6.11 3.05 1.52 1.02 0.77 0.61 5.32 2.82 1.45 0.98 0.74 0.60 0.00 0.00 0.00 0.00 0.00 0.00 Transmission coefficient of the disease Time (days) Susceptibles Infected Recovered Figure 7. Time (days) to reach the limit number of susceptible, infected and recovered individuals for different values of the transmission coefficient of the disease in case of using antibiotics with a resistance rate of 100% In order to avoid redundancies, in the following cases we will present the different figures following the structure presented before. 3.2. Resistance close to 100%. This subsection corresponds to antibiotics with a resistance rate close to 100%. Due to the similarities of the rates among the antibiotics of this group, Amoxicillin and Clavulanic acid has been considered to be representative, with a resistance rate of 97.50%. PREDICT TREATMENT OF MULTI-RESISTANT Pseudomonas aeruginosa INFECTION 9 0 days 1 days 2 days 3 days 4 days 5 days 6 days 7 days 0 dogs 50 dogs 100 dogs 150 dogs Dynamics of susceptible individuals 0 days 1 days 2 days 3 days 4 days 5 days 6 days 7 days 0 dogs 50 dogs 100 dogs 150 dogs Dynamics of infected individuals 0 days 1 days 2 days 3 days 4 days 5 days 6 days 7 days 0 dogs 50 dogs 100 dogs 150 dogs Dynamics of recovered individuals Figure 8. Dynamics of susceptible, infected and recovered individuals depending on time (in days) when using an antibiotic with resistance 97.50% Susceptibles Infected Recovered Time (days) Individuals Time (days) Individuals Time (days) Individuals β= 0.1 6.2475 0.9857 5.5475 182.0012 1.9775 3.0085 β= 0.2 3.1150 0.9875 2.9400 181.0199 1.0675 3.0005 β= 0.3 2.0825 0.9541 1.8725 179.0799 0.7700 3.0049 β= 0.4 1.5575 0.9571 1.4875 178.0629 0.6300 3.0395 β= 0.5 1.2425 0.9603 1.1200 176.1075 0.5425 3.0513 β= 0.6 1.0325 0.9637 0.9800 175.0559 0.4900 3.0900 β= 0.7 0.8925 0.8982 0.7875 173.1549 0.4375 3.0372 β= 0.8 0.7700 0.9709 0.7175 172.1004 0.4025 3.0154 β= 0.9 0.6825 0.9746 0.5950 170.2078 0.3850 3.0559 β= 1 0.6125 0.9783 0.5425 169.0158 0.3675 3.0651 Table 2. Data for different values of the transmission coefficient of the disease βin case of using antibiotics with a resistance rate of 97.50% 16 PREDICT TREATMENT OF MULTI-RESISTANT Pseudomonas aeruginosa INFECTION β= 0.1β= 0.2β= 0.4β= 0.6β= 0.8β= 1 0 10 20 30 40 34.74 33.43 16.66 9.89 6.67 4.81 25.66 15.49 3.34 1.63 1.96 5.95 34.27 30.45 15.47 9.26 5.81 4.17 Transmission coefficient of the disease Time (days) Susceptibles Infected Recovered Figure 17. Time (days) to reach the limit number of susceptible, infected and recovered individuals for different values of the transmission coefficient of the disease in case of using antibiotics with a resistance rate of 7.14% 3.7. Resistance 0%. This subsection corresponds to the antibiotics with total effectiveness. 0 days 1 days 2 days 3 days 0 dogs 50 dogs 100 dogs 150 dogs Dynamics of susceptible individuals 0 days 1 days 2 days 3 days 0 dogs 50 dogs 100 dogs 150 dogs Dynamics of infected individuals 0 days 1 days 2 days 3 days 0 dogs 50 dogs 100 dogs 150 dogs Dynamics of recovered individuals Figure 18. Dynamics of susceptible, infected and recovered individuals depending on time (in days) when an antibiotic with a resistance rate of 0% is used PREDICT TREATMENT OF MULTI-RESISTANT Pseudomonas aeruginosa INFECTION 17 Susceptibles Infected Recovered Time (days) Individuals Time (days) Individuals Time (days) Individuals β= 0.1 0.4375 145.9988 0.7000 0.9824 0.5950 39.0306 β= 0.2 0.8225 136.9863 0.8400 0.9809 0.8050 48.0938 β= 0.3 0.9100 125.9459 1.0150 0.9564 0.9450 59.0773 β= 0.4 1.4875 110.9878 1.1900 0.9695 1.3125 74.0443 β= 0.5 1.3825 94.9995 1.3300 0.9970 1.4525 90.0654 β= 0.6 1.6800 77.9836 1.4175 0.9886 1.4875 106.0461 β= 0.7 1.6275 62.9900 1.4350 0.9982 1.4175 120.0633 β= 0.8 2.6425 49.9998 1.4175 0.9869 1.4175 132.0735 β= 0.9 1.7850 39.9951 1.3825 0.9673 1.3300 141.0661 β= 1 1.6975 31.9943 1.3300 0.9899 1.2775 148.0393 Table 7. Data for different values of the transmission coefficient of the disease βin case of using antibiotics with resistance 0% β= 0.1β= 0.2β= 0.4β= 0.6β= 0.8β= 1 0 1 2 3 0.44 0.82 1.49 1.68 2.64 1.70 0.70 0.84 1.19 1.42 1.42 1.33 0.60 0.81 1.31 1.49 1.42 1.28 Transmission coefficient of the disease Time (days) Susceptibles Infected Recovered Figure 19. Time (days) to reach the limit number of susceptible, infected and recovered individuals for different values of the transmission coefficient of the disease in case of using antibiotics with a resistance rate of 0% 4. Results and discussions This study includes a group of 192 dogs affected by otitis externa living in the south of Spain, in which 40 individuals were infected by Pseudomonas aeruginosa. All samples were aseptically collected and sent to a laboratory to to perform an antibiogram that included 52 different antibiotics. We recall that the goal of this work was to develop an antibiotic panel that could be used as an empirical treatment for dogs infected with Pseudomonas aeruginosa, a pathogen transmissible to humans, with high multi-resistance capacities and that is considerate the leading cause of nosocomial infections in human medicine. To this 18 PREDICT TREATMENT OF MULTI-RESISTANT Pseudomonas aeruginosa INFECTION end, the well-known SIRI model (1.1)-(1.3) was used which helped us to provide a priority antibiotic treatment. In this way, the behavior of the different groups of individuals of the population of dogs infected by Pseudomonas aeruginosa was studied using the SIRI model to observe the dynamics of every group. In addition a code was programmed in Matlab to obtain the most relevant information from these simulations and then, they were represented in a table where the limit value of susceptible, infected and recovered individuals were included as well as the time at which the corresponding limits were reached. The SIRI model (1.1)-(1.3) reunited all the necessary characteristics and, after comparing with the possible application of other mathematical models, this is the one that was the most suitable and accurate relating to all the epidemiological data compiled as well as to clinical behavior of the infection. To our knowledge, this is the firs time this specific application of the model is described relating to pathogenic microorganisms dynamics. Especially, it is worth mentioning that this work is quite relevant for mathematical science, since it proves that mathematical modelling should become part of the convenient toolbox of Public Health research and decision-making, specially in the design of models of empirical treatment choices based on epidemiological data. Interestingly, results show that a significant number of antibiotics tested (27,45%) have null action to Pseudomonas aeuruginosa infection. In addition, it is remarkable to add that some of them are third generation Cephalosporins (more specific and expensive, with a more complex development). Moreover, approximately 12% of the antibiotics included had a resistance rate of almost 100% and 8% of them between 50% and 30%. Hence, when an empirical antibiotic treatment is provided, the chance of ineffectiveness or partial effect (which is one of the causes that increases the resistance rate) is, at least, 30%. This fact highlights the potential health problem that presents this multi-resistant microorganisms and the necessity of an accurate tool to prescribe the most specific and effective treatment. On the other hand, almost a third part of the total of antibiotic tested were always effective, however the use of a high number of them are hospital restricted. Regarding the results obtained, we consider that the SIRI model implemented in the present work can be a very useful tool to predict the dynamic of Pseudomonas aeruginosa infection in dogs, since it provides an antibiotic panel that in case is empirically prescribed, it has an accuracy of efficacy of more than 90% (Figure 20) . Subsequently, it can be used as an efficient model to formulate potential personalized treatment in both human and veterinary medicine, in order not only to save cost, but also to avoid future problems of development of multi-resistance by this bacteria in both humans and animals, because its high level of specificity and efficacy against the infection. PREDICT TREATMENT OF MULTI-RESISTANT Pseudomonas aeruginosa INFECTION 19 Although all the antibiotics included in the recommended panel have an efficacy of more than 90%, the use of most of them are hospital-restricted, which could present, in this concrete infection and microorganism, a problem in prescribing the treatment in some clinical situations. The results obtained are in consonance with the epidemiological data and the information that the World Health Organization, as well as numerous Epidemiological Health Services, in human and veterinary medicine, are providing. Consequently, the present work highlights the severity of the multi-resistance development in pathogens and reinforces the necessity of new tools to implement in the strategies that have to be applied. Figure 20 provides a selective antibiotic panel in case is necessary to prescribe an efficient empirical treatment to Pseudomonas aeruginosa infection. 20 PREDICT TREATMENT OF MULTI-RESISTANT Pseudomonas aeruginosa INFECTION Miscellaneous Colistin Tetracyclines Tigecyclin Quinolones Maxifoxacin Lome foxacin Oxonilic acid Levofoxacin Aminoglycosides Amykacin Estrptomycin Sisomycin Kanamycin Beta lactam antibiotics Penicilins Piperacilin Ticarcilin Penicilin G Carbapenem Meropenen Imipenem Pipe-tazobactam Cefquinome Cefoperazone Ceftazidime Antibiotic families Antibiotic subfamilies Treatment Figure 20. Antibiotic panel recommended to prescribe an empirica treatment against Pseudomonas aeruginosa infection in Spain 5. Conclusions Multi-resistant infections are an unequivocal public health problem and are responsible for an increasing number of deceased patients in hospitals. In this sense, companion animals can act as a reservoir of this type of microorganisms that can be transmitted to humans. PREDICT TREATMENT OF MULTI-RESISTANT Pseudomonas aeruginosa INFECTION 21 At this point, since no studies of these characteristics have been previously described, it is urgent and critical to develop an efficient tool to minimize the antimicrobial resistance process. Therefore, we have applied an adapted SIRI model as a compatible and potential tool based on Matlab software. Remarkably, this model allows predicting the behavior of Pseudomonas aeruginosa infection and is essential to select an effective empirical treatment when culture and antibiogram techniques are no available or the severity of the infection obliges an imminent treatment, providing an antibiotic panel with an accuracy of efficacy of more than 90%. Our study suggest that this method could be suitable to be used in other infections therapies in other microorganisms and in other countries based on their epidemiological data. Furthermore, it is likely to reinforce the accuracy to provide a more effective treatment of choice, using a model that behaves as a very useful tool to be implemented in Public Health research and decision-making strategies. Appendix In Appendix we provide the table of antibiotics tested as a treatment. Group Antibiotic Resistance (rate) Beta lactam Cephalosporins Cefalotin 100% antibiotics Ceftazidime 16,67% Cefotaxime 47,37% Cefepime 21,05% Cefoxitin 100% Cefuroxime 100% Cefazolin 100% Ceftazidime/CA 82,35% Cefovecin 95,24% Cefalexin 100% Cefotaxime/CA 100% Ceftiofur 100% Cefpodoxime 100% Cefoperazone 0% Cefquinome 0% Beta lactamase Amoxicilin clav. 97,50% inhibitors Ampicilin 100% Piperacillin-tazobactam 5,88% Carbapenem Imipenem 7,14% Ertapenem 53,85% Meropenem 0% Monobactams Aztreonam 20,69% Penicilins Piperacillin 0% Ticarcillin 0% Penicillin 0% Aminoglycosides Tobramycin 15,15% Gentamicin 23,68% Amikacin 6,25% Neomycin 50% Continue in the next page 22 PREDICT TREATMENT OF MULTI-RESISTANT Pseudomonas aeruginosa INFECTION Group Antibiotic Resistance (rate) Kanamycin 0% Sisomicin 0% Streptomycin 0% Quinolones Ciprofloxacin 28,57% Enrofloxacin 34,21% Marbofloxacin 23,81% Nalidixic acid 93,33% Pradofloxacin 28,57% Levofloxacin 0% Oxolinic acid 0% Maxifloxacin 0% Lomefloxacin 0% Tetracyclines Doxycycline 92,86% Tetracycline 83,33% Tigecycline 0% Afenicol Chloramphenicol 100% Florfenicol 100% Miscellaneous Polymyxin B 15,79% Colistin 0% Fusidic acid 100% Fosfomycin 100% Macrolide Erythromycin 33,33% Sulfamide Sulfonamides+Trimethoprim 100% Table 8: Table of all the antibiotics assayed in the antibiograms included in the study References [1] J. Rubin, R. Walker, K. Blickenstaff, S. Bodeis-Jones, S. Zhao, Antimicrobial resistance and genetic characterization of fluoroquinolone resistance of Pseudomonas aeruginosa isolated from canine infections, Veterinary Microbiology 131 (1-2) (2008) 164–172. doi:10.1016/j. vetmic.2008.02.018. [2] V. Aloush, S. Navon-Venezia, Y. Seigman-Igra, S. Cabili, Y. Carmeli, Multidrug-resistant Pseudomonas aeruginosa: Risk factors and clinical impact, Antimicrobial Agents and Chemotherapy 50 (1) (2006) 43–48. doi:10.1128/aac.50.1.43-48.2006. [3] D. van Duin, D. L. Paterson, Multidrug-resistant bacteria in the community, Infectious Disease Clinics of North America 30 (2) (2016) 377–390. doi:10.1016/j.idc.2016.02.004. [4] V. M. Paz-Zarza, S. Mangwani-Mordani, A. Mart´ınez-Maldonado, D. ´ Alvarez-Hern´andez, S. G. Solano-G´alvez, R. V´azquez-L´opez, Pseudomonas aeruginosa: patogenicidad y resistencia antimicrobiana en la infecci´on urinaria, Revista chilena de infectolog´ıa 36 (2) (2019) 180– 189. doi:10.4067/s0716-10182019000200180. [5] J.-E. Hyun, T.-H. Chung, C.-Y. Hwang, Identification of VIM-2 metallo-β-lactamase- producing Pseudomonas aeruginosa isolated from dogs with pyoderma and otitis in korea, Veterinary Dermatology 29 (3) (2018) 186–e68. doi:10.1111/vde.12534. [6] C. Pomba, M. Rantala, C. Greko, K. E. Baptiste, B. Catry, E. van duijkeren, A. Mateus, M. A. Moreno, S. Py¨or¨al¨a, M. Ruzauskas, P. Sanders, C. Teale, E. J. Threlfall, Z. Kunsagi, J. Torren-Edo, H. Jukes, K. Torneke, Public health risk of antimicrobial resistance transfer from companion animals, Journal of Antimicrobial Chemotherapy (2016) 957–968doi:10. 1093/jac/dkw481. PREDICT TREATMENT OF MULTI-RESISTANT Pseudomonas aeruginosa INFECTION 23 [7] C. C. Pye, A. A. Yu, J. S. Weese, Evaluation of biofilm production by Pseudomonas aeruginosafrom canine ears and the impact of biofilm on antimicrobial susceptibility in vitro, Veterinary Dermatology 24 (4) (2013) 446–e99. doi:10.1111/vde.12040. [8] W. O. Kernack, A. G. McKendrick, A contribution to the mathematical theory of epidemics, Proceedings of the Royal Society of London 115 (1927) 700–721. [9] W. O. Kernack, A. G. McKendrick, A contribution to the mathematical theory of epidemics, Proceedings of the Royal Society of London 115 (1932) 700–721. [10] W. O. Kernack, A. G. McKendrick, A contribution to the mathematical theory of epidemics, Proceedings of the Royal Society of London 141 (1933) 94–112. [11] T. Caraballo, R. Colucci, A comparison between random and stochastic modeling for a sir model, Communications on Pure and Applied Analysis 16 (1) (2017) 151–162. doi:10.3934/ cpaa.2017007. [12] T. Caraballo, X. Han, Applied Nonautonomous and Random Dynamical Systems, Applied Dynamical Systems, Springer International Publishing, 2016. doi:10.1007/978-3-319-49247-6. [13] Clinical and Laboratory Standards Institute, Performance Standards for Antimicrobial Susceptibility Testing, 29th Edition (2019). E-mail address:[email protected] E-mail address:[email protected] E-mail address:[email protected] E-mail address:[email protected] E-mail address:[email protected] E-mail address:[email protected] E-mail address:[email protected]