Globally, 825 patients were included in the study. Median age was 49 years with extreme values of 15 and 94 years. The most represented age group was 45 - 59 years (33.5%). Men were more represented: 422 men (53.6%) and 383 women (46.4%), sex ratio was 1.1.

Out of 825 patients included in the study, 208 (25.2%) presented one or several signs suggestive of HAI and they received microbiological tests. 9.8% (IC_95% 7.8% - 11.8%) were confirmed or potential HAI.

As part of patients care, invasive devices such as urinary catheters were sometimes used. Distribution of these invasive devices is highlighted in Table 1. Respondents had one or two invasive devices. Among them, 821 were using venous catheter and 109 bladder catheter.

825 patients who participated in the study had a catheter. Among them, 81 patients (9.9%) had a Healthcare-Associated Infections.

The second invasive device used was the urinary catheter and 19.3% of the patients with urinary catheter had a Healthcare-Associated Infections.

Bacteremia were the most frequent healthcare-associated infections as highlighted in Figure 2.

Globally, germs most often identified irrespective of the site were respectively: K. pneumonia (38.5%), S. aureus (23.1%) and E. coli (20.0%) (Figure 3).

Germs were more often identified in blood than urine. The following germs were found in order of frequency in blood, K. pneumonia, S aureus, E. coli and in urine, K. pneumonia, E. coli, S. aureus (Table 2).

Frequency of germs identified according to admission ward is highlighted in Table 3.

K. pneumonia was predominant in internal medicine and S. aureus in nephrology.

Healthcare-associated infections are associated with significantly higher case fatality rate than other ailments. This case fatality rate is estimated 37.0% (Table 4).

P < 0.001.

HIV+ status, Internal Medicine Department, Nephrology and Endocrinology, duration of admission and the use of urinary catheter represent factors statistically associated with the onset of healthcare-associated infections in this study (Table 5).

Among the risk factors for Healthcare-associated infections, HIV infection, hospital stay, and the port of urinary catheter are significantly associated with urinary infection onset.

This study could not analyse respiratory healthcare-associated infections and surgical site infections which are also indicators of hospital hygiene; the diagnostic means of respiratory infections are not available in our context. However, the method of study having taken into account all patients followed in a given service and the realization of a cohort study was to minimize this bias.

The median age of patients included in our study was 49 years with extreme values of 15 and 94 years, and sex-ratio 1.1. These values are higher than those

reported by Amazian and al in 2010 [9] in a multicenter study conducted in the mediterranean region within 27 hospitals (median age 41.1 years and sex ratio 0.99). This gap may be related to the difference in the study populations. Amazian and al included pediatric patients, while our study concerns exclusively patients aged 15 and above.

Prevalence of HAI was 9.8%. This value is similar to records found by other authors in the sub-region; DIA and al in Senegal 10.9% [10] , TRAORE and al in Mali 8.5% [11] and OUENDO and al in Benin 9.8% [13] . However, our prevalence is lower than that of AHOYO and al who recorded 19.1% in 2012 in Benin [14] . This gap may be due to the fact that AHOYO’s team conducted a national cross-sectional study across 39 out of 45 hospitals including surgery department, gynecology and obstetrics unit, and internal medicine, while ours was conducted only in a single medical unit including 825 patients.

HAI prevalence is generally high in African hospitals while it records 5.7% on average in Europe according to ECDC in 2012 [5] [7] . As a matter of fact, non-compliance with the minimum standards of hygiene (care and hands hygiene practices), isolation procedures, biomedical waste management and rational use of antibiotics in our hospitals could explain the high rate recorded.

Respondents had one or two invasive devices, thus suggestive of gateway for germs. The study of healthcare-associated infections distribution according to the site revealed that: 25.9% of patients with bladder catheter developed urinary tract infection, and 65% of those with central venous catheter developed healthcare-associated infection. In AHOYO and al study [14] , several infectious sites were identified, but the most frequent were urinary tract (37.0%), venous catheter (27.0%), and surgical site (19.2%). Thus, invasive procedures are significant risk of healthcare-associated infections in healthcare settings. This aspect was taken into account in BIAOU and al study in CNHU in 2011 with 6.3% prevalence of surgical site infections.

Major germs identified during our study were K. pneumonia, S. aureus, and E. coli. This result is similar to data obtained by OUENDO and al in Cotonou [13] , and WALELEGN and al in Ethiopia [15] . However, in RAZINE and al study in Morocco [16] , S. aureus was the most identified germ. In Italy, MANCINI and al identified E. coli as the most frequently found germ [17] . In Senegal, DIA and al study revealed E. cloacae as major germ [10] . These gaps may be due to the study population and microbial ecology which varies from one geographical area to another. However, S. aureus, K. pneumonia and E. cloacae characteristics are favorable ground for KAI. In fact, these are either commensal flora S. aureus germs which can develop rapidly especially in vulnerable patients, or biofilms which can be found on poorly sterilized or unsterilized equipment likely to play a role in bacteria transmission from a patient to another.

HAI prevalence was 14.4% in internal medicine. This study shows that there is a relationship between the admission ward and onset of HAI (p = 0.027). This result is similar to findings of AHOYO and al who reported that internal medicine was most affected behind surgery department. MANCINI and al made the same observation. Nevertheless, OUENDO and al and AMAZIAN and al identified respectively burn care center and intensive care unit as most-at-risk services [9] [13] . This could be due to the type of patients included in the study. Given that internal medicine admits most immunocompromised patients, the high prevalence can also be related to other factors especially non-compliance with hygiene measures. As regards nephrology, it provides care for patients with kidney injury and dialysis patients with arterio-venous fistula. Bladder catheter installation and handling may be source of hand-borne contamination which could justify S. aureus prevalence in this unit. This assumption was corroborated by KHANAL and al in his study which indicates that the rate S. aureus transmission is high among doctors and nurses [18] .

HAI appears to be significantly associated with the deaths recorded in wards A and B of Medicine department (p < 0.001). This was the finding of WHO and ROSENTHAL and al [1] [19] . In fact, more often than not, HAI germs are multi-resistant; this leads to delayed treatment and increased fatality. The ever alarming situation in low-income countries such as Benin with limited financial resources, makes it difficult to secure appropriate antibiotics because they are too expensive for most patients with no medical cover.

There is no statistically significant relationship between factors such as age, gender, diabetes, kidney injury, dialysis, use of catheter and HAI. However, WALELEGN and al indicated that 1 - 14 age range patients were more exposed, while RAZINE and al reported that patients above 60 years were more exposed to risk of HAI. In fact, children and the elderly are most vulnerable to infections, and this was corroborated by these authors.

In our study, HIV infection is significantly associated with onset of HAI as reported by AMAZIAN and al [9] . However, unlike this author, we did not find statically significant association between HAI and diabetes, kidney injury and dialysis [9] . These pathologies are immunosuppressive conditions likely to foster onset of HAI. The absence of association could be due to the fact that the vast majority of diabetic patients in our study had normal blood sugar, with controlled immunity.