Introduction
Nosocomial infections (NIs) are an important cause of preventable morbidity and mortality; they also result in significant socioeconomic cost. Surveillance of NI is an essential part of the infection control programme (1). Despite their limitations, point prevalence surveys are usually preferred to determine the magnitude of NI when resources are limited. These studies are inexpensive, easy to perform, and do not require major human or technical resources. National prevalence surveys of NI have been performed in many European and developing countries (2-12) and were used as a tool to define NI control priorities.
Despite national regulations requiring the reporting of all NI cases, no reliable data are available at this time due to lack of coordination of measures between national and hospital authorities. During the past few years, however, there has been an increase of interest in the control of hospital acquired infections, and some hospitals have developed individual programmes.

This paper presents data on prevalence of infection obtained during a point prevalence survey conducted in two Latvian hospitals. This is the first time that this type of survey has been conducted in Latvia. The hospitals were chosen because they had an infectious diseases consultant and a hospital epidemiologist who were involved in infection control activities.

Materials and methods
A point prevalence study design was used. Methods, infection criteria, and definitions with minor modifications were those used by Meers in the national survey of infections in the United Kingdom (7) and also in Lithuania (12). Length of hospital stay was not a definition criterion. No significant outbreaks of influenza or other acute respiratory infections were reported in Latvia at the time of the survey. The survey was carried out in Stradins University Hospital, which has 1263 beds, and in the State Hospital of Traumatology and Orthopaedics, which has 320 beds. The University hospital had most types of medical unit, but no psychiatric or paediatric department. The neonatal unit was closed during the time of the survey due to renovation. The Orthopaedics hospital had surgical wards only.

We sought to collect data on all inpatients hospitalised before 8 am on the first day of the survey. The sampling units were wards. A coordinator was responsible for managing an investigation team in each hospital. Trained medical doctors collected data from clinical records, temperature charts, laboratory reports, and information provided by physicians and nurses in each ward. Infections of more than one site in the same patient were counted as separate infections. Antibiotics prescribed at the time of the survey were recorded. Preoperative and perioperative doses of antibiotics were registered separately as prophylactic. Approximately 90% of the questionnaires were completed during the first day of the survey. Records were computer analysed using the WHO/CDC EpiInfo 2000 software. The data were entered in duplicate to minimise transcription errors.

Results
All hospitalised patients from both hospitals were included in the survey. A total of 1291 patients was investigated. The median age of the patients was 56 years (range 1-98), 57 years for women (range 1-90), and 55 years for men (range 4-98), and the median length of hospitalisation prior to the study was six days (range 1-99). In 226 (17%) patients, signs of infection were detected, or these patients were receiving treatment for already diagnosed infection.

A total of 5.1 percent (66/1291) patients were reported to have NI. The overall prevalence of NI was 5.7% (72/1291) because in 6 patients, 2 infections were detected simultaneously (table 1). Surgical site infection (SSI) was the most frequent site of NI, with an overall prevalence 3.5% (45/1291) (table 2) accounting for 62% of all NI, followed by respiratory tract infection (RTI) (7.5%) and urinary tract infection (UTI) (6.4%). Hospital acquired UTI was reported in 12% (11/90) of patients with urinary catheters. Evidence of nosocomial pneumonia was found in 50% (4/8) of mechanically ventilated patients. In 53 (19.7%) patients with intravenous devices a least one NI was registered. Microbiological examination in 29% (21/72) of the patients with NI yielded positive results.

The prevalence of NI increased with the increasing age of the patients (p=0.02; c 2 test for the trend) and was higher in intensive care (p<0.001; c 2 test) and surgical departments (p<0.001; c 2 test) (table 1). Higher prevalence was also associated with mechanical ventilation (p<0.001; c 2 test), the presence of urinary catheters (p<0.001; c 2 test) and intravenous devices (p<0.001; c 2 test), with previous surgical intervention (p<0.001; c 2 test) and longer hospital stay. In comparison with other patients, the prevalence of NI did not increase in patients with diabetes.

Community acquired infection (CAI) was detected in 164 (12.7%) patients (table 3). RTIs accounted for 50.6% of all CAIs followed by gastrointestinal tract infections (GTIs) and UTIs (9.8% and 7.3% respectively).

Antibiotics were given to 288 patients (22%) on the day of the survey, among whom 66 received antibiotics without having any signs of infection. Cefazolin was administred to 182 patients, accounting for 50% percent of all antibiotics used. Other most commonly used antibiotics were aminoglycosides (64 patients), metronidazole (46), ampicillin/amoxicillin (46), ciprofloxacin (38), ceftriaxone (14), cefuroxime (13), doxycycline (8), amoxicillin/clavulanate (6), and trimethoprim/sulfamethoxazole (6). Combination therapy with two drugs were given to 69 patients, and 3 patients were treated with three drugs. The three most common antibiotic combinations were: cefazolin with metronidazole (25 patients), cefazolin with gentamicin (10), and ciprofloxacin with metronidazole (8).

Discussion
This pilot study on NI is the first of its kind reported in Latvia. The two hospitals surveyed are national referral centres in which highly specialised or risky medical manipulations are required. Thus the results of this survey cannot be extrapolated to other Latvian hospitals.
The definitions of the first UK National Prevalence Survey (6) were used, as opposed to CDC or the second National Prevalence Survey, because we found them to be simpler and more relevant to our clinical setting. More advanced definitions sometimes require clinical and laboratory information which is not available in our hospitals on a regular basis. The methodology of the first UK National Prevalence Survey also provided a better opportunity to analyse the prevalence of community acquired infections and antibiotic use. Our main objective was to recognise the problem of NIs in Latvia by using a simple survey approach.

The prevalence of NIs was found to be 5.6%, which is within the range reported by investigators from other European and developing countries (2-12). This result should be interpreted with caution because it is difficult to compare studies with different NI definitions used.
We do not think that information bias could significantly affect the outcome of the study because the hospital staff were not fully aware of the aims of the study. Detection bias could be taken into account because certain essential laboratory investigations were not available for the team of investigators.
In Latvian hospitals it is quite common for patients to stay hospitalised for some time after their clinical symptoms have improved, so they may finish the initiated course of treatment, or wait for some specific investigation. This is partially due to limited primary care or to specific socioeconomic situation.

Laboratory and radiological investigations may also take more time than recommended due to a lack of financial and personnel resources. It is likely that the NI rate would be higher if patients who were present in the wards but who had already recovered at the time of the survey had been excluded. Thus, the actual burden of the hospital acquired infection among very sick individuals may be higher than these results show.
RTI and SSI were the two principal types of infection found in this study. SSI accounted for the majority of hospital acquired infections, and RTI were most commonly community acquired. The low prevalence of hospital acquired UTI found in catheterised patients in the University hospital differs from the data published so far (2-12). This may be due to insufficiencies in our methodology. It is very likely that the data collection methods used that were based on doctors' diagnosis were not sensitive enough to record all UTIs. Symptoms of UTI in patients with a serious major disease of other organs may have been ignored. As a result, urine analysis and microbiological investigation were not ordered, and therefore the surveillance team was not able to identify these patients.
With patients' increasing age, previous surgical intervention, mechanical ventilation, and the presence of intravenous devices and urinary catheters were significantly associated with the rate of NI. Surprisingly, diabetes was not found to be associated with increased prevalence of NI, probably due to the small number of patients surveyed.

Only 29% of infections were microbiologically documented. The study design did not allow us to analyse the actual number of specimens cultured. Therefore it is difficult to comment whether this low rate was due to insufficient laboratory capacity. We can only speculate that a large proportion of patients with infection were not cultured before initiation of antibiotic therapy.
Data on antibiotic use clearly indicated the lack of antibiotic prescribing policy in the hospitals selected. On the day of the survey, 22% of hospitalised patients were receiving antimicrobial treatment. This could mean that approximately 4% of the patients received antibiotics without a previously documented infection. The questionnaire did not include questions regarding the reason for the use of antibiotics.
There may be various explanations for the alarming rate of cefazolin use. First, this drug is relatively inexpensive and its dosing interval is convenient for nurses who are often overloaded with work. Furthermore, the use of cephalosporins has recently been encouraged by active marketing by pharmaceuticals companies, and doctors' positive experience of their clinical efficiency. The problem of cephalosporin resistance has also only recently appeared. Second and third generation cephalosporins are more expensive, and there are frequent financial restrictions. For all the above reasons, cefazolin is considered to be an acceptable substitute to guarantee a positive clinical outcome. Secondly, in surgical departments the administered course of prophylactic antibiotics is sometimes prolonged for up to five days without clear clinical indications. And finally, the increased use of cefazolin might be due to the decline in penicillin use, which, in turn, has been caused by concerns about resistance as well as the view that this group of antibacterials may be outdated. An apparent lack of marketing by pharmaceutical companies due to the low cost of cefazolin adds to the problem. Furthermore, their dosing interval is an additional strain for nurses. Our major concern was the wide use of cefazolin for the treatment of respiratory tract infections. Other antibiotics that have been used extensively in treatment of various patients are metronidazole and gentamicin, probably because of their low cost and presumed excellent clinical efficacy, and physicians' personal experience of prescribing them.

In conclusion, the results of our survey show that the techniques used were practical, and that it was possible to obtain the information required for the purposes of the study. This pilot study clearly showed that nosocomial infections are more prevalent in Latvian hospitals than can be seen from the official reporting system, and that a national prevalence survey is necessary to obtain information on the level of the whole country. Our study also identified problems related to the excessive use of antibiotics. Importantly, we identified high risk departments that should be targeted by infection control measures. The following studies should be focused on prevalence of NI only. CDC definitions should be used in the national survey to make data comparable to other countries and geographical areas.

Acknowledgements

We thank the administrative, medical, and nursing staff of Stradins University Hospital and the State Hospital of Traumatology and Orthopaedics, Latvia.
We particularly acknowledge the important contributions of :
Jolanta Bärbale, Anda Jëca, Ivars Geldners, Lïga Liepina, Vita Manjakova, Viktorija Mihailova, Linda Micule, Evija Rudzïte, Tatjana Save|jeva, Kalvis Straupmanis, Dins Sumerags, Natälija Vinogradova, Indra Vi_umsone, and Ieva Ziediea to this study. The study was supported by a grant from the Latvian Science Council.