Patterns of infections, aetiological agents and antimicrobial resistance at a tertiary care hospital in northern Tanzania

To determine the causative agents of infections and their antimicrobial susceptibility at a tertiary care hospital in Moshi, Tanzania, to guide optimal treatment.

well as other chronic conditions such as cancer and kidney disease.
In Africa, infectious diseases constitute a much higher burden in people of all age groups [8][9][10][11][12][13] than in Europe and North America. Active surveillance systems on infectious disease control are inadequate in most African countries [14][15][16][17], and the absence of data on disease causes and the lack of control measures add to the disease burden [6]. Obtaining adequate and timely data on infectious diseases in African countries is difficult, whereas much more data are available from developed countries. As a result, empirical treatment in Africa regarding pathogenic bacteria might be based on data from clinical laboratories in developed countries [18]. The pattern of bacterial infections in Africa differs from those observed in Western Europe and North America; hence, empirical treatment should be adjusted to fit the different need. In Africa, as elsewhere, regular local data collection about patterns of infections, and their response to antimicrobial agents, coupled with a long-term commitment to providing adequate health information systems, is key to effective planning and policy formulation. The availability of data on disease patterns, etiological agents and their antimicrobial resistance is a prerequisite for the establishment of empirical treatment regimens and for avoiding escalation of the global trend of bacterial resistance to the current antimicrobial regimens [19][20][21].
This study was conducted to determine the most frequent bacterial species isolated from samples submitted for microbiological examinations at a tertiary care hospital in Moshi, Tanzania. We also determined antimicrobial susceptibility patterns of the isolates and associated comorbid conditions.

Study design, location and sample collection
This study is a descriptive analysis of culture, bacterial identification and antimicrobial susceptibility testing conducted at Kilimanjaro Christian Medical Centre (KCMC), which hosts a tertiary healthcare facility for the northern zone of Tanzania. The study was conducted from August 2013 to August 2015. Clinical samples were collected from inpatients admitted to the medical or surgical wards of the hospital. Informed consent was obtained from all participants. Specimens including stool (56), sputum (122), blood (56) and wound/pus swabs (286) were collected from 575 patients. Twenty specimens were not included due to insufficient quantity. Blood, stool or sputum samples were collected if the patient was diagnosed to have septicemia, diarrhoea or upper respiratory infection, respectively. Blood samples were also collected from patients with fever of unknown cause. Wound or pus swabs were collected from wounds due to burns, surgical procedures, diabetes mellitus, animal bites, motor traffic accidents and other injuries.
Patient hospital files were used to obtain socio-demographics and clinical characteristics of the study participants admitted at KCMC wards. Data were recorded on designated case report forms (CRFs). All samples were transported immediately after collection to the Kilimanjaro Clinical Research Institute (KCRI), to be processed by the microbiology unit of the biotechnology laboratory department.

Culture and identification
Wound/pus swabs and sputum specimens were cultured onto trypticase soy agar with 5% sheep blood (BD BBL TM ), chocolate agar and McConkey agar plate media (BD BBL TM ) and incubated at 37°C with 5% CO 2 for 18-24 h. Stool samples were cultured onto McConkey agar and incubated at 37°C overnight. Blood samples were collected in 40 ml BD BACTEC standard 10 Aerobic/F, followed by incubation in BD BACTEC for a maximum of 5 days. Gram stain was used to differentiate Gram-positive and Gram-negative bacteria from culture. Positive blood cultures were inoculated on blood agar, chocolate agar and McConkey agar and incubated for 18-24 h at 37°C with 5% CO 2 . Bacteria were isolated from culture by picking single colonies and subculturing onto purity plate (Blood agar) overnight. From the purity plate, morphology determination and gram stain was made followed by microscopic examination.
Based on Gram staining results, Gram-negative bacteria were identified using API 20E and/or API NE 20 (bioM erieux) for Enterobacteriacae and other non-fastidious Gram-negatives. Catalase and coagulase tests were used to identify Gram-positive cocci. Optochin susceptibility testing (BD BBL Taxo TM Benex Limited, Shannon County Clare, Ireland) was used to confirm Streptococcus pneumoniae, and BD BBL Streptocard â Enzyme Latex test was used to identify streptococcal group A, B, C, D, F and G.

Antimicrobial susceptibility testing
Antimicrobial susceptibility testing (AST) was performed using disc diffusion on M€ uller-Hinton II Agar (MHA) according to Clinical Laboratory Standards Institute (CLSI, 2013) guidelines. Gram-negative bacterial isolates were tested for ampicillin 10 [22].

Data analysis
Data were double-entered in OpenClinica (OpenClinica LLC, Waltham, MA, USA). Data extracts were exported to STATA 13 (StataCorp LP, Texas 77845, USA). This tool was used for all analyses. Proportions of comorbidities and bacterial isolates were calculated and presented as column or row percentages. Table 1 summarises characteristics of 575 patients with different medical conditions enrolled in the study from August 2013 to August 2015. Five patients were excluded during analysis due to incomplete patient information. The median age (IQR) was 43 (30-57) years; 61% (n = 348) were males, 39% females (n = 227).

Culture results
KCMC is a referral hospital serving approximately 11 million people in the northern zone of Tanzania. We conducted a 2-year study from August 2013 to August 2015 to determine the pattern of bacterial pathogens associated with different health conditions among patients who were admitted at medical and surgical departments in this hospital.
The culture results revealed a higher positive growth rate on wound swab specimens (n = 180, 72.3%), compared to others. Both Gram-negative and positive species were observed. Our findings are in agreement with other studies in Africa where wound infections tend to manifest with a variety of bacterial pathogens [23,24]. This is explained by the complexity of wound specimens that will tend to have a variety of bacterial pathogens, depending on the way wounds were acquired. Wounds acquired from a community environment will have more diverse organisms than hospital-acquired wounds [25, Table 4a summarises the association between commonly isolated groups of bacteria and absence or presence of three health conditions diabetes, HIV and cancer as common comorbidities identified in patients participated in this study. While Table 4b summarises association between common bacterial isolates and presence or absence of wound infection since most of the patients were diagnosed to have wound infections. NB: Numbers described as; Proportion (95% confidence interval).     26]. In our study, there were few positive blood specimens (n = 15, 6%) on culture, a finding which is in agreement with other studies that showed a low positive growth of blood cultures [27]. Possibly this is because most septicemia cases we tested were not due to bacterial infections, a finding that has been reported from other studies in Tanzania [28]. The use of antibiotic prior to coming to the hospital, which hinders the detection of susceptible organisms [29], may be another reason. This is a practice which is very common in Tanzania and other developing countries [30][31][32]. KCMC being a tertiary care hospital receives patients referred from health facilities in Kilimanjaro region and other parts of Tanzania. The majority of these patients have been treated with antibiotics in the referring hospital. However, the patients could also have self-medicated as antibiotics are easily available and policies to control improper use of antibiotics are non-existent in Tanzania, as in other developing countries [30,[32][33][34][35].

Bacterial spectrum
The bacterial spectrum observed from this study showed a high diversity of Gram-negative bacilli such as Proteus spp (12.7%), E. coli (11.7%), Pseudomonas spp (10.6%) and Klebsiella spp (10.1%), with Staphylococcus aureus (9.3%) being the predominant Gram-positive isolate. The majority of these Gram-negative bacilli were from wound infections rather than from other disease conditions. This predominantly Gram-negative infection pattern, as also observed in other studies [24,36,37], is different from that most commonly reported from Western Europe and North America [38,39]. The reasons for this are not entirely known, but recent studies have also shown that higher temperatures are correlated with increased numbers of infections caused by Gram-negative bacterial species [40]. Our results do however emphasise that we cannot use knowledge obtained in Western Europe and North America directly for clinical care and empirical treatment in sub-Saharan Africa.

Infection pattern
Like any other resource-constrained countries, Tanzania is still experiencing the burden of infectious diseases. In the current study, we observed presence of wound infections (n = 263, 45.9%), septicemia (n = 46, 8.03%), diarrhoea (n = 20, 3.5%), HIV (n = 81, 14.1%), and tuberculosis (n = 60, 10.5%) and many other respiratory infections. As also indicated in other studies [13,41], these diseases are still health challenges in low-income countries. We also recognised the presence of noncommunicable diseases such as cancer (n = 52, 9.1%) and diabetes (n = 122, 21.3%). This suggests co-existence of communicable and non-communicable diseases in lowincome countries as has been indicated in other studies [42,43]. The increase of diabetes in Tanzania may be the effect of little knowledge about the risk factors associated with the disease [4], which may be related to the majority (n = 339, 59.2%) of the participants having only primary level education. There was no relationship between the most frequently isolated bacteria and conditions like diabetes and cancer. However, a non-statistically significant association between diabetes and Proteus spp (19%) and E. coli (14%) (P = 0.07) was observed. The finding was more or less the same with HIV status. It was noted that prevalence of Proteus spp (16%) and Pseudomonas spp (13%) was higher in wound infections than in those with no wound infection (9% and 6%, respectively), and the difference was statistically significant (P = 00.3). This is in agreement with other studies [25] despite the fact that the design of the studies was different. Again we can generally attest that aetiology of a wound infection is broad, depending on environment and nature of the wound.

Antimicrobial susceptibility testing
Proper identification and determination of antimicrobial resistance of the bacterial pathogens is crucial to help physicians to provide proper treatment promptly. The advancement in pharmaceutical industries has lead to discoveries of many antibiotics, which has facilitated physicians' efforts into providing quality effective medical care. Despite of this increase, prudent use is essential in controlling antimicrobial resistance, which has now become one of the major challenges for medical progress. The current study demonstrated that a majority (n = 412, 71.9%) of patients had sought treatment prior to coming to the hospital. It is likely these patients received antibiotics during their previous treatment. The most frequently used drugs were ceftriaxone (46.2%), metronidazole (39.4%), cloxacillin (17.1%), ciprofloxacin (6.6%) and co-trimoxazole (5.5%). Along with this, the observed resistance patterns of Gram-negative bacterial isolates tested on drugs such as amoxicillin-clavulanic acid ampicillin, gentamicin, trimethoprim-sulfamethoxazole and chloramphenicol were relatively high in ampicillin (82.4%) and trimethoprim-sulfamethoxazole (60%), and below 50% in the rest. This is probably due to the fact that these drugs are used in Tanzania as firstline antibiotics in treatment of gastrointestinal diseases, respiratory diseases, obstetrical/gynaecological, cardiovascular and nervous system diseases [44]. Their easy availability from hospitals causes these drugs to be commonly used for treatment by medical practitioners as well as for self-medication, factors which play a great role in drug resistance [33,34,45]. The finding is in agreement with other two studies in Tanzania which indicated resistance of E. coli and Klebsiella to trimethoprim-sulfamethoxazole, ampicillin, amoxicillin-clavulanic acid and gentamicin (100%, 96%, 88%, 60%) and (85%, 95%, 70%, 70%), respectively [46,47]. Resistance ranging from 9.1% to 78% also has been noted to ciprofloxacin and nalidixic acid. These are categorised as second-line drugs in Tanzania, yet they show a high frequency of resistance. As ceftriaxone is a third-generation cephalosporin, we expected its use to be controlled, yet it was the most used. Moreover, convincing percentages of resistant strains of E. coli and Klebsiella to first and third-generations of cephalosporins have been broadly noted, ranging from 36.8% to 61.5%. This finding suggests existence of extended spectrum beta lactamase (ESBL) bacteria, as has been addressed in other studies in Tanzania [23] where 64.3% of E. coli and 80% of Klebsiella pneumoniae were ESBL producers. The presence of ESBL producers reduces treatment options, resulting in higher morbidity and mortality due to severe infections and sepsis. Staphylococcus aureus was the predominant spp among Gram-positive isolates. It accounted for 6 (27.3%) of the observed resistances to cefoxitin and other antibiotics such as erythromycin, trimethoprim-sulfamethoxazole and penicillin G, a characteristic that we would postulate to indicate methicillin-resistant Staphylococcus (MRSA) clones, as has been suggested in other studies [48].

Conclusion
The study has revealed a wide range of causative agents, with an alarming rate of resistance to the commonly used antimicrobial agents. Furthermore, the bacterial spectrum differs from those observed in high-income countries. This highlights the imperative of regular generation of data on aetiological agents and their antimicrobial susceptibility patterns especially in infectious disease endemic settings.
The key steps would be to ensure the diagnostic capacity at a sufficient number of sites and routine exchange, comparison, analysis and reporting of data. As the minimum, sentinel sites (hospitals) across the country and region should report on a representative subset of bacterial species and their susceptibility to drugs at least once a year. A central organising body should collate the data and report to all relevant national and international stakeholders [49].