Nosocomial infections prevalence study in a Serbian university hospital

Background/Aim. Nosocomial infections (NI) are a serious health problem resulting in an enromous burden of excess morbidity and mortaliti rates, and health care costs. The aim of this study was to assess the prevalence of NI and to identify groups of patients at special risk for NI in the University Clinical Center, Kragujevac, Serbia. Methods. A period prevalence study design was used in this study. A survey of NI included all patients hospitalized in all departments in the University Clinical Center, Kragujevac. Results. Among 764 patients surveyed, the global prevalence rate of patients with at least one NI was 6.2% (95%CI = 5.6–6.8), while the prevalence of NI was 7.1%. The most frequent infections were surgical site infections (14.1%; 95%CI = 12.9–15.3), followed by pneumonia (2.3%; 95%CI = 2.1–2.5) in surgical patients. In medical wards, the most common NI were skin and subcutaneous tissue infections (1.6%l 95%CI = 1.4–1.8), and urinary infections (1.4%; 95%CI = 1.3–1.5). Overall, 85.1% NI were culture-proven; the leading pathogens were Pseudomonas species (40.0%), followed by Staphylococcus species (25.0%), Escherichia coli (22.5%), Proteus mirabilis (17.5%) and Klebsiella-Enterobacter (12.5%). Multivariate logistic regression analysis identified 3 risk factors independently associated with NI appearance: hospital stay ≥ 8 days (p = 0.0015), urinary catheter (p = 0.0022) and antibiotic use (p < 0.001). Conclusion. This study showed that NI are a serious health problem in our hospital. The most common infections were surgical site infections, followed by skin and subcutaneous tissue infection and urinary tract infections. Nosocomial infections were most common in patients in urological and orthopedic departments, and then in intensive care units. Prolonged hospital stay, urinary catheter and antibiotic exposure were risk factors independently associated with NI appearance.


Introduction
Nosocomial infections (NI) are a serious health problem in hospitals all over the world [1][2][3][4] .Nosocomial infections result in an enormous burden of excess morbidity and mortality rates, and health care costs [4][5][6] .
The frequency of NI differs from country to country 1- 4, 6, 7 .It is estimated that in developed countries 5-10% patients get one of these infections during hospitalizations, whereas in developing countries rates are higher up to 25% 8 .An international study covering 47 hospitals in 14 countries (Europe, Eastern Mediterranean, Southeastern Asia and Western Pacific Region) over the period from 1983 to 1985 showed that an average prevalence rate was 8.7%, ranging from 3 to 21% 9 .The national programmes of prevalence studies were conducted in the most developed countries over the last decade of the 20th century (United States of America, France, Spain, Italy, German, Swiss, Sweden).In these countries the decline in prevalence of NI was evident -except in the intensive care wards and in the long-term-care hospitals 1,2,[10][11][12][13][14] .The occurrence of nosocomial infections differs in different hospitals and different wards, and in patients diagnosed differently.The highest prevalence rates of NI were observed in intensive care units and surgery wards 12,13 .Most investigations suggest that surgical site infections are the most prevalent, followed by infections of urinary tract, lower respiratory tract and bloodstream infections [12][13][14][15][16][17] .Etiology of NI shows that the causes of infections have changed over the last decades -possibly due to the change in antibiotic therapies in clinical treatment 18 .Resistance to antibiotics (especially the appearance of methicillin resistant Staphylococcus aureus) is one of the leading characteristics of the microbial agents of nosocomial infection 1,19 .
In spite of the huge effort, the concept of NI is still unclear to the Serbian medical community.This is the first time that this type of survey has been conducted in Kragujevac.The main aim of this study was to assess the prevalence of NI and to identify groups of patients at special risk for them.

Methods
The Clinical Center, Kragujevac, a 1,240-bed tertiary care University Center, is divided in 25 departments.A survey of NI included all patients hospitalized in all departments: surgical (general and endocrine surgery, gastrointestinal, biliopancreatic, colorectal, chest surgery, vascular, neurosurgery, plastic surgery, orthopaedic and traumatologic surgery, obstetric and gynaecologic, urologic, otorinolaringologic and ophtalmologic surgery, paediatrician surgery), with surgical intensive care unit, and medical departments (internal, pediatric, psihiatric, neurologic, pulmologic, oncologic, dermatovenereologic, infectious).
A period prevalence study design was used in this study.A 1-week prevalence survey was conducted from 15 to 19 December, 2003.Data were collected by the detailed uniform questionnaires.The trained medical doctors filled the questionnaires for each patient with the collected data from clinical records, temperature charts, laboratory reports, and information provided by physicians and nurses in each ward.Study variables included patients demographics (sex, age), primary diagnosis (primary diagnostic group: diseases of circulatory system, neoplasms, diseases of genitourinary, respiratory, musculoskeletal and central nervous system, infectious diseases, and new-born), comorbidities (diabetes, arterial hypertension, trauma, etc), and factors related to health care, including surgery procedures, mechanical ventilation, central and peripheral venous catheter, urinary catheter, and the use of antimicrobials.Information on variables associated with surgery (type of surgical site, duration of surgery, antibiotic prophylaxis) were also gathered.Microbiologic data were recorded as microorganisms identified in cultures, as well as antimicrobial sensitivity tests.
The definition of the Centers of Disease Control and Prevention (CDC) for NI was used 20 .All the patients who got infected and developed an overt form of NI, that was not present or incubating at the time of admission to the hospital, were recorded.Infections of more than one site in the same patient were considered as separate infections.
Surgical site infection was taken into consideration even though the diagnosis was made in some other ward than surgery department, for example in the internal medical ward.If the patient had more than one surgery procedure, the questionnaire was filled in with the data of the most recent one.
Only nosocomial infections that were active on the day of the study were taken into account, i.e. infections for which the prescribed antibiotics therapy had not yet started or was in progress at the moment.
All the patients hospitalized in a certain ward on the day of the study were included, that is the patients hospitalized in the ward at the moment of this study conduction, not those in the operation theatre or those taken to some other wards for a diagnostic procedure.We used to come back to the wards only to complete the data on the patients included in the study at our first visit.
The primary diagnoses on admission were classified according to International Classification of Diseases, Tenth Revision (ICD-10) codes 21 .
All samples were analyzed in the same laboratory.Identification of isolates was done using the routine methods 22 .All samples were processed in the Torlak System (Torlak, Belgrade, Serbia).Each specimen was screened for antibiotic-resistant organisms by antimicrobials susceptibility test disc (Bioanalyse, Ankara, Turkey).In cases of material for microbiological examination taken on the day of the study or earlier and no results available, we collected microbiological analyses within not longer than the next 72 hours.
Statistical analysis was performed using the Statistical Package for Social Sciences Software (SPSS Inc, version 7.50, Chicago, IL).The results were expressed as the mean ± SD or as a proportion of the total number of patients.Relations between categoric variables and NI were first evaluated using contingency table analysis and χ² test or Fisher's Exact Probability Test and method univariate logistic regression analysis.The Student's t-test was used for the comparison of parametric continuous variables.The odds ratios (OR) and corresponding 95% confidence intervals were computed for overall site infection rate.
A multivariate stepwise logistic regression model was used to identify variables that were significantly associated with the occurrence of NI, while the effects of other potentially confounding risk factors were simultaneously controlled.This model introduced variables with significant difference or association by univariate analysis.A difference was considered statistically significant for p < 0.050.

Results
At the Clinical Center, Kragujevac, 900 patients were hospitalized from 15 to 19 December, 2003.During the investigation, all hospitalized patients were included in the study.A total of 107 patients (11.9%) not at their wards for diagnostic or therapeutic procedures, were not included in the research.Twenty-nine (3.2%) patients were excluded from the study because of the incomplete chart review.Our prevalence study comprised 764 patients (response 84.9%).
The most frequent disorders on admission were as follows: genitourinary diseases (19.6%), diseases of central nerv-ous (16.5%), circulatory (15.2%), respiratory (13.4%) digestive (12.4%) and musculosceletal system (9.6%), as well as the infectious diseases (6.8%).Ninety-two patients (12.0%) had neoplasm.Trauma was recorded in 45 (5.9%).Fewer than 3% were diabetics.Other risk factors for NI (artherial hypertension, chronic respiratory, cardiovascular and renal diseases, presence of other infections, etc.) were recorded in 12.8% of all the patients.There were 50 (6.5%)newborn children and suckpigs.A total of 16.8% (128/764) patients went through a surgery procedure with skin incision.The other invasive procedures frequencies were urinary catheter -13.5%, peripherial intravenous device -46.7%.Totally 389 (51.0%) of all the patients received antibiotics at the time of the survey.This included: 278 cases in primary disease therapy, 43 with NI, and 68 patients without any signs of infection.Sixty-two (48.4%) of the patients submitted to surgery had antibiotic prophylaxis.
A total of 6.2% (47/764) of the patients were reported to have nosocomial infections.The overall prevalence of nosocomial infections was 7.1% (54/764), since in 4 patients 11 infections were detected simultaneously.
The prevalence of nosocomial infections was highest in the patients with diseases of the musculosceletal system, but as compared with other patients the differences were not statistically significant.The surgery patients had more frequently NI than the patients with no surgical procedure (p < 0.001).Nosocomial infections were more frequently found in the patients with trauma (p = 0.016).The patients admitted to intensive care units showed an even higher prevalence of NI than those admitted to other units (p = 0.002).In 10% of the pa- tients with intravenous devices at least one NI was registered (p < 0.001).Nosocomial infections were detected in 18.8% of the patients with urinary catheters (p < 0.001).It was more likely that the patients with NI had received antibiotics than patients without NI (p < 0.001).In comparison with other patients, the prevalence of NI was not higher in the patients with diabetes or neoplasms, while in the patients with other comorbidities a statistically significantly higher prevalence of NI was found (p = 0.001).
The NI prevalence was highest in the patients aged 75 or more.The NI were not found in patients aged 2-19 years.The differences were statistically significant (p = 0.039) (Figure 1).Gender had no influence on getting NI.At the Clinical Center Kragujevac, prevalence of NI was highest in surgery wards (urology -21.7%, orthopedic -20.0%, general -9.0% and other surgical wards -2.6%), followed by critical care unit (16.7%), medical internal department (4.1%) and department of obstretics and gynaecology (2.2%) (Table 2).The differences were statistically significant (p < 0.001).

S t a p h y l o c o c c u s s p e c ie s E s c h e r ic h i a c o l i P r o t e u s m i r a b i l i s K le b s i e l la -E n t e r o b a c t e r O t h e r
Frequency (%)

Fig. 2 -Prevalence of nosocomial infections in the Clinical Center, Kragujevac in the function of the isolated microbial agents frequency
According to the results of the univariate logistic regression analysis, age (years ≥ 65), hospital stay (days ≥ 8), trauma, other comorbidities, surgical interventions, history of intensive care unit stay, intravenous devices, urinary catheter and antibiotic use, were statistically significantly linked to NI risks (Table 4).The results of the univariate regression analysis showed that as compared to noninfected patients, the infected ones were significantly more frequently the elders and longer hospital stay, trauma, other comorbidities, surgical interventions, intravenous devices, urinary catheter, had used antibiotic, or were in intensive care unit.

Discussion
In our hospital the prevalence of NI was 7.1%.Nosocomial infections were mostly recorded in our youngest and oldest patients.Surgical site infections were most prevalent.At surgical departments, especially in patients with invasive diagnostic-therapeutic procedures (surgery, intravenous devices, urinary catheter), as well as in patients with cancer and trauma and other comorbidities, higher frequency of NI infections was reported.Risk factors independently associated with NI appearance were the duration of hospital stay, urinary catheter and antibiotic use.
The prevalence of NI in this study is within the range reported by investigators from developing countries [23][24][25][26] .However, it is very difficult to compare the prevalence studies and their results in different countries because of different patients' characteristics, different medical experience and, in many cases, different methodology.The first prevalence study in Serbia was done in 1985 at the Military Medical Academy, Belgrade, and 967 patients then showed the prevalence rate to be 5.5% 27 .Surveillance data on NI in an emergency surgery unit in Belgrade showed that prevalence was 11.2% 28 .Highest rates were recorded in the intensive care unit (25%), and neurosurgery ward (16.7%).Surgical site infections were the most prevalent (33.3%), followed by respiratory (25%) and urinary tract infections (16.7%).
The first point prevalence study in two Latvian hospitals 23 reported that prevalence rate of NI was 5.6% (72/1291), surgical site infection being the most common NI (62%), followed by respiratory tract infection (7.5%), and urinary tract infection (6.4%).The multicentric study in Greece showed that participation of urinary tract infections was 22.4-38.2%,respiratory tract infections 21.1-32.6%,surgical site infections 14.6-22.7%,and bloodstream infections 9-13.2% 17 .In seven Swiss pediatric hospitals in 2000 the survey of 520 patients proved the prevalence of patients with at least one NI was 6.7% (range per hospital 1.4-11.8%) 13.Bacteremia was most frequent (2.5%), followed by urinary tract infection (1.3%) and surgical site infection (1.1%).The rates of NI varied in different hospital wards.According to the results given by Pittet et al. 12 , the highest prevalence of NI was in intensive care (25%), surgery ward (12%), internal medicine (9%).Gastmeier et al. 2 cited that the overall prevalence rate was 3.5% (CI 3.1-3.9)and the highest prevalence rate (15.3%) was found in intensive care ward patients, followed by surgery (3.8%), general medicine (3.0%) and gynecology/obstetrics (1.4%).The prevalence rate of NI in general surgery departments was 14.0%, in orthopedic surgery departments 8% and in urological departments 20.9%.In Thailand 26 the prevalence of NI in 9.865 patients was 6.5%, while the intensive care unit had the highest infection rate of 22.6%, followed by pediatrics (6.8%).
In this study, the surgical site infections accounted for the majority of NI (78.3%) in surgical patients.Surgical site infections accounted for 60% in departments of general surgery, for 47% in orthopedic surgery departments and for 37% in urological departments.In four Swiss university hospitals, Pittet et al. 12 recorded that the most frequent NI was surgical site infection (30% of all NI), followed by urinary tract infection (22%), lower respiratory tract infection (15%), and bloodstream infection (13%).Prevalence of NI was higher in critical-care units (25%) than in medical (9%) and surgical wards (12%).Also, in Thailand, Latvian hospitals, and our study the most common site of NI was surgical site infections, while urinary tract and lower respiratory tract came second 26,23 .In contrast, some authors reported that the most frequent site of NI was the respiratory tract, followed by surgical site wound infection and urinary tract infection 4,29 .
Increasing numbers of persons older than 65 form a population that is particularly at risk of nosocomial infections.This result is not surprising, and it is similar to that reported by other authors.Recent data from the National Nosocomial Infections Surveillance study indicate that 54% of infections in adults appeared in patients 65 years of age or older 1 .On the basis of a prevalence survey in Switzerland acute hospitals, 2001, authors found that infection rates are higher for female patients and increase with age for both genders 30 .The susceptibility to infections of this age group is related to impaired host defenses, chronic underlying disease (especially frequency of diabetes mellitus), immune dysfunction, poor tolerance to diagnostic and therapeutic procedures, increased time of hospitalization and use of antimicrobial agents 12,15,[30][31][32][33] .In Brazil, elderly patients were more likely to develop nosocomial infections (16.1% prevalence), while surgical site infections accounted for the majority of the nosocomial infections 15 .On the other hand, the North American studies indicated that urinary tract infections were the most common 1,3 .
All the bacterial causes were isolated in our study of the prevalence of NI (they were predominant in many other studies as well, although the order of their appearance is more or less changed) 12,13,16,23,24 .As in many other authors who did some researches of NI problem, our study also identified Pseudomonas species as the most common cause of the infections 26,34 .A relatively high occurrence of this cause is specifically prominent in surgery and intensive care wards, where Pseudomonas species can cause very serious infections in patients with weaken immunity.In contrast, in four Swiss university hospitals, the most frequently isolated microorganisms were Enterobacteriaceae (28%), Staphylococcus aureus (13%), Pseudomonas species (11%), and Candida species (10%) 12,13 .А meaningful part of all episodes of NI wаs not microbiologically documented in several studies.
Risk factors independently associated with NI appearance аre hospital stay, urinary catheter and antibiotic use.These results are not unexpected, being similar to those reported by other authors 35,36 .The mean interval between admission to the hospital and our study was 11 days.This is probably due to the fact that patients who stay longer at hospitals are exposed to a greater number of reservoirs and sources of microorganisms.Also, a serious health disorder is always in relation to a longer hospitalization.Urinary catheter provide a portal entry and potential overgrowth of microbes in the urinary tract.Many of the authors showed that various antibiotic (therapeutic, prophylactic) strategies, as Ilić M, Marković-Denić Lj.Vojnosanit Pregl 2009; 66 (11): 868-875.
well as unreasonable excessive antibiotic use, were associated with raised risk of NI 35,36 .

Conclusion
This study shows that NI are a serious health problem in our hospital.The most common infections are surgical site infections, followed by skin and subcutaneous tissue infection and urinary tract infections.Nosocomial infections are most common in urological and orthopedic department, and then in intensive care units.Prolonged hospital stay, urinary catheter and antibiotic exposure are the risk factors independently associated with NI appearance.

Fig. 1 -
Fig. 1 -Age-related prevalence of nosocomial infections in the Clinical Center, Kragujevac, * p -probability value (χ² test and Fisher's exact test, 2-tailed) indicates statistical significance of the differences in infected and noninfected patients, age-related

Table 1 Prevalence of nosocomial infections (NI)* in the Clinical Center, Kragujevac; patients and health care characteristics related
*NI -

Table 2 Prevalence of nosocomial infections (NI) in the Clinical Center, Kragujevac by hospital units
-1.8) of the patients in medical departments.

Table 4 Results of univariate analysis of potential risk factors for appearance of nosocomial infections
*B -coefficient logistic regression analysis; † S.E. -