The result of this study showed that the overall prevalence of bacteria isolated from septicemia suspected patients by blood culture method were 21%, which was in line with studies conducted in varieties of countries in the world outside and inside Ethiopia, in India, (16.08–22.3%) [22, 24, 34, 35], Jimma (18.2%)  Gondar (18.2%, 19.4%) [7, 12]. Similarly the same results were also found in Mozambique 15.1% , Iran 21.1% ,Ghanaian hospital (23.16%) , Nigeria 20.4% . On the other hand, the prevalence of bacteria isolated from this study was lower than articles conducted in India (90%,30.6) [29, 38], Iran (38%), Cameroon (28.3%) , Addis Ababa (27.9%,32.8%) [3, 40], Mekelle (28%)  and Bahir Dar (32%, 39.2%) [41, 42]. However, the current study was higher prevalence than from the studies done in Addis Ababa (13%) , Jimma (8.8%) , Arba Minich (11.3%) , Nepal (10.6%) , Kuwait (2.3%) , New Zealand (9.5%) , India (4.4%) , Bangladesh (13.2%) , Ghana (11.2%) , and Southern Africa (5.5%) . The possible reason for this variation might be explained by the fact that difference in the geographical location, the study population, method of study design, sample size, epidemiological variation, implementation of infection prevention and control, blood culture diagnostic system, study duration, and health care policy systems in these different countries.
In our present study, Gram-negative bacteria were more predominantly isolated than Gram-positive bacteria, which a prevalence of 55.6% and 44.4%, respectively. Similarly, concordant findings were observed in the study conducted in India (51.2%) , Iran (55.4%) , and Nepal (52.3%, 50.5%) of the isolates were gramnegative bacteria which was the same with the current study[15, 48]. However, our finding was lower than other studies conducted, where grampositive bacteria was the predominant isolated bacteria than gram negative bacteria in India (60.9%61.4%) [43, 49], Jimma (53.3%) ,Addis Ababa (77.4%) , Gondar (66%)  from this Grampositive bacteria were predominant. The varying percentages could be due to methodology difference and the patient safety practice, diagnostic method.
Among blood bacteria isolates in this study were S. aureus, K.pneumonia, E. coli, CONS, and P.aeruginosa were the most plentiful causative agent for septicemia, which had orderly put the prevalence(26.7%),(17.8%),(13.3%),(8.9%),(8.9%) and similar findings were found in Indonesia which was K.pneumonia (17%), P.aeruginosa (12%).Greater or a lesser amount of similar results have been seen in studies conducted in septicemia suspected patients from different regions in Ethiopia. Although, the frequency and bacterial magnitude were varied in Addis Ababa, Ethiopia 50%,26.21% 14.02% ,in Mekelle, Ethiopia37.5%,11.1% , and Southern Ethiopia 32.2%,14.3% .These bacteria are normal microbiota of the skin, and in the majority of the time found in the hospital areas as a result cause infection in immune susceptible patients.
In the present study, different associated factors were evaluated for various associations with septicemia. Nevertheless, not all of the socio-demographic variables in the current study were statistically associated with septicemia. However, this study indicated that males were more infected than females (23.8% vs 19.2%), respectively, but there was no a statistically significant difference in gender variation (p = 0.632). This slight variation has been reported in different studies [12, 13, 51]. Patients in under-five age group were more infected with septicemia as compared to the adult age group. The possible reason for this is the children have more susceptible to infection due to less immune system and colonized by normal microbiota. Nevertheless, in this study the category of age groups were not statistically significant (p = 0.186). This finding disagrees with the research done in southern Ethiopia ,Rwanda , Nepal , and the USA . The possible reason for this difference might be due to the none representative of samples in each ward from this study.
Previous hospitalization in this study was 2.9 times more likely associated with septicemia in the current study. This research finding is consistent with the study conducted in Ethiopia AOR = 5.54 and AOR = 3.2[2, 9]. This association might be due to health care associated infections, high number of patients admitted to hospitals for prolonged periods, delayed request for blood culture, and a weak patient safety, and management system in the hospital. As a result, fast intervention had a must to minimize the infection.
The assessment of antimicrobial susceptibility pattern that differs in isolated bacteria, and the rates of resistance were high in this study, this may cause a serious therapeutic challenge for the management of common blood bacterial infections. The overall resistance of gram-positive and gram-negative bacteria in this study were 25%-66.7% and 8%-92% respectively, which was similar to the result reported in Mekelle for gram-positive 0-83.3% and gram-negative 0–100% , Jimma, 0-85.7% for gram-negative and 0%to 100% Gram-positive .This condition gives severe anxiety to the population. This high rate of resistance might indicate misuse and inappropriate usage of the antibacterial drugs. However, the resistance is different from the study conducted in Gondar where the rate of resistance for Gram positive bacteria ranged from 23.5–58.8% and gram negative bacteria 20–100% . The observed variation may be due to the arbitrary use of antimicrobials in the study area, self-prescribing antimicrobials in private drug stores and empirical treatment of hospital acquired infections without real etiological agent identification.
In our study, S.aureus was 91.7% resistant to penicillin, but susceptible to 75% gentamicin, and 83.3% ciprofloxacillin. This was similar with reports conducted in different countries 40%-97% [22, 35, 40]. This is due to the unrestrained use of antimicrobials without sensitivity testing in the study area. (66.7%) of methicillin, resistance Staphylococcus aureus (MRSA) was detected. In order of, the same report was done in Indian and Ethiopian researches [7, 22, 45].The incident of MRSA is more common because of the haphazard use of higher antimicrobials empirically and genetic proficiency to obtain antibiotic resistance. S. pyogenes were 100% sensitive to cefepime, ampicilin, and vancomycin from our study. This was in line with the study reports, which had in 100%, 87% [14, 35]. Vancomycin was 100% sensitive to S. agalactia. This finding was similar to the reports done in Mekelle . The possible reason for the absence of vancomycin resistant bacteria in this study is that the use of vancomycin by clinicians is controlled in the management of patients in the study setting, and in the nation as an entire.
K.pneumonia. E.coli, P.aeruginosa, K.rhinose, and Serratia spp in our study were 50–100% sensitive to meropenem, imipenem, and amikacin. This finding was similar to the reports [2, 22, 38, 45]. K.pneumonia and E.aerogenes were 100% resistant to ampicillin, amoxicillin -clavulanic acid. In addition, E.coli (83.3%) for gentamicin and P.aeruginosa (75%) for gentamicin and ampicillin resistance were observed in this study. These were consistent to other studies conducted in Ghana 74%, Ethiopia 86% and 94%[9, 41] and India 93%, 87%[45, 54].The finding of this study indicates most of the gram-negative bacteria were 50–100% resistance to ciprofloxacillin, gentamicin, ceftriaxone, chloramphenicol, trimethoprin-sulfamethoxazole, and ceftazidime. This was concordant with the research done in Cameroon 45%-100%  and Ethiopia [9, 11, 43].
The limitation of this study was done in a single hospital, and only a phenotypic test was done, which may under estimate the prevalence of bacteria for the different geographical areas. This study has not detected fungemia and anaerobic bacteria due to scarcity of the media.