Our results indicate that CVC-ABSIs rates are increasing and more than 85% of CVC-ABSIs occur in the non-ICU setting, accounting for 33% of largely preventable CVC-ABSIs with the implementation of a multifaceted infection control programme. The decrease achieved in CVC-ABSIs rates in the first year of the intervention was higher and significant in medical wards, adjusted per catheter and per patient-days.
The prevention and control of CVC-ABSIs has rarely been investigated outside the ICU setting, and to our knowledge, there is only one study conducted at two small community hospitals demonstrating 57% CVC infection rate reduction in non-ICU CVC-ABSIs, with the use of antibiotic impregnated catheters for some patients . In this series the use of CVC, as we also reported, was higher in non-ICU setting, and considering that the vast majority of all hospitals are community hospitals with fewer than 200 beds, these results reinforce the necessity and importance of restoring measures of prevention and control in non-ICU inpatient [6, 13].
The results in non-ICU floors were lower than 65%-70% decrease for ICUs. However, the results are not comparable because of differing patient characteristics and nurse and physician to patient staffing ratios . The introduction of the bundle from January 2009 did not result in the levels falling below those achieved before 2006–2008, possibly because in those years the majority of oncology patients and patients included in haemodialysis programme were not admitted to our hospital.
In our study, the outside-ICU estimated CVC-ABSIs/1,000 catheter-days incidence is higher than 1–2.9/1,000 catheter-days of laboratory-confirmed central line associated bacteremia rates from data reported by hospitals participating in the National Healthcare Safety Network (NHSN) between January and December 2009 . Our incidence is also higher than 4.3/1,000 inserted CVC-days from Internal Medicine and Surgical wards in German hospitals , higher than 4.4 in an General Medical service , higher than 6.6 in burn patients  and than 0.87 in patients with solid tumours receiving chemotherapy in an ambulatory setting . Our greater incidence could be due to: our inclusion of all cases of CVC-ABSIs clinical sepsis, we have less dedicated infection control personnel than in other centers, the existence of best results publication bias and the method we used to calculate the incidence. Collecting information on central line-days is labor intensive and the rate per 1,000-catheter-days in our study was based on an estimation of catheter use . The size of the prevalence study sample and the number of cases of CVC-ASBIs were small and it is possible that the incidence rates are not very reliable and could be overvalued. However, having used the same method of estimation over time should not subtract solidity to the evolution of the incidence before and after bundles, as is proven achieving similar results when analyzing the evolution of CVC-ABSIs rate per patient days.
The strengths of our study are that surveillance was performed by staff with formal infection control training and who have years experience working in infection control , that the study provides hospital-wide surveillance information and reports secular trends over two decades. Our results indicate that there are temporal variations in the incidence and differences between surgical and medical wards. The incidence increased possibly due to an increasing use of these catheters in a progressively ageing population, with greater comorbidity and more invasive treatments, as well as due to better detection of blood stream infections.
CVC-ABSIs were associated with parenteral nutrition use and CVC-ABSIs rate was higher in surgical rather than medical wards. CVC and parenteral nutrition use was similar over time, but was greater in surgical wards and this could explain the difference in rate between admission wards . On the other hand, it is necessary to consider that patients sometimes need to be transferred from one ward to another, which can make it difficult to allocate CVC-ABSIs to a specific unit, and the results of one ward could show the effect of measures of prevention implemented in another unit [13, 20]. In this way, the ICU of our hospital participates from 2009 in the national program called “bacteremia zero” which was developed to reduce the CVC-ABSIs in patients with higher risk of this complication, and it is possible that the higher and significant rate reduction that was achieved in Internal Medicine could be explained in part because 65% of all patients with CVC followed there (with and without CVC-ABSI) were patients with CVC inserted by ICU physicians.
Another issue that could have contributed to different results is the difference in staff training and their motivation to participate in the prevention and control programme . This seems to be reflected in the different alcohol-based hand-rub consumption and by the increased CVC-ABSIs rate in the summer. This increase of CVC-ABSIs numbers in summer, a period of time with no apparent difference in the use of CVCs than in the rest of year and with a smaller number of admissions and inpatient-days (administrative data), makes us to formulate the hypothesis that this could be explained by the temporal recruitment of nurses with less experience in the handling of the catheters to work for the hospital.
We are unable to determine whether specific interventions led to the observed decrease in CVC-ABSIs rates because several measures were started simultaneously. Although a standard control group for comparison purposes was not feasible, the fact that a progressive and largely consistent increase in CVC-ABSIs rates from 1991 to 2008 and a decrease in CVC-ABSIs rates was observed for three years after starting the intervention programme, could support a cause and effect relationship between the beginning of a programme called “bacteremia zero” and a decrease in outside ICU CVC-ABSIs rates in Internal Medicine wards. The effects of this intervention and the increase in CVC-ABSIs rates during summertime reinforce the importance of allocating only trained and competent personnel for insertion and maintenance of central intravascular catheters, and ensuring appropriate nursing staff levels in non-ICU settings [2, 12].
Improving the measures of asepsis on insertion and after care of catheter, and removing the unnecessary catheters could have a very important impact [21–23], as it seems to be evidenced by a 64.1% reduction in the catheter-utilization ratio of CVC inserted in the last year. This decrease in CVC exposure could in part explain a decrease in CVC-ABSI incidence.
There are several limitations to this study. First, we did not collect data on pre and post intervention process measure compliance, and this lack of data decreased our ability to measure the full impact of the bundles. Secondly, this is an single-centre study and we have estimated incidence/1,000 catheter-days  because we do not know the incidence and duration of inserted catheters, which makes it difficult to generalise practice patterns to other centres, given that infection control policies may differ significantly between different hospitals. Thirdly, catheter insertion sites were not documented, therefore, infection rates attributable to a specific insertion site could not be analysed. Fourthly, cultures were not obtained from all CVCs after removal.