Catheter-related bacteraemias involve increased hospital stays, costs, morbidity, and mortality .
Catheter-related bacteraemias are one of the most frequent cases of nosocomial bacteraemias and are classically associated with the use of central venous catheters, mainly in intensive care units.
In recent years, the incidence of catheter-related bacteraemias in other hospital units, as well as their relationship to the use of peripheral venous catheters, has also increased .
It has been estimated in different studies that the use of peripheral venous catheters is 60 times more common than that of central catheters; however, the rates of bacteraemia related to peripheral catheters are lower than those of bacteraemia related to central catheters. In our study, central catheters were responsible for 77% of catheter-related bacteraemias compared with 23% for peripheral catheters. These data are similar to those of the study on the prevalence of nosocomial infections in Spain (EPINE, 2016), in which data recovered from 294 hospitals and 59,016 patients showed that among all catheter-related bacteraemias (306 bacteraemias), 73.85% (226 bacteraemias) were related to central venous catheters and 26.14% (80 bacteraemias) were related to peripheral catheters . In our study, the cumulative incidence of catheter-related bloodstream infections per 1,000 patients/day of hospital stay was more than 3 times higher for central catheters than that for peripheral catheters. This finding concurs with the data reported in other studies, such as that by Pujol et al , whose estimated rate was 0.18 episodes of bacteraemia per 1,000 days of peripheral catheter use at a university hospital; a 5 times higher rate, 0.9 episodes per 1,000 days of central catheter use, was recorded simultaneously in another study .
This difference may occur because the initial appearance of peripheral phlebitis is usually a physicochemical phenomenon [17, 18]. This initial peripheral phlebitis is associated with a low risk of initial infection, although it generally entails a change in the venous access route, decreasing the dwell times of peripheral venous catheters, which could lead to a lower incidence of bacteraemias associated with peripheral catheters. In our study, there were significantly higher catheter dwell times for central catheters compared with those for peripheral catheters (a median of 7. 4 more days). In the study by Targer et al., an increased risk of phlebitis occurred between the second and the third day of catheter insertion and remained stable thereafter, which is the reason why these authors recommend changing peripheral lines every 48 to 72 hours . However, subsequently, the data from randomised controlled prospective studies have been published in which systematic replacement every 72 hours was compared with that performed when clinically indicated. There were no differences in terms of the number of cases of phlebitis, dysfunctional catheters, and local infection or bacteraemia rates [20, 21].
Regarding clinician's diagnostic accuracy, clinicians have a lower diagnostic accuracy of suspected sources when bacteraemias are due to venous catheters than if bloodstream infections are due to other sources of bacteraemias.
Our study showed a clinician's diagnostic accuracy of over 85% when the bacteraemia source was respiratory, urinary, or gastrointestinal and a lower accuracy when the bacteraemia source was intravascular (congruencies of approximately 68%). This may explain why only 53.9% of the CB patients and 62.5% of the PB patients received appropriate empirical antimicrobial treatment against microorganisms associated with venous catheter-related bacteraemias. There were more accuracy for PB origin than for CB origin probably because of macroscopic manifestations of flebitis are more frequently associated to peripheral venus catheter bacteremias than with central ones.With respect to the microorganisms associated with catheter-related bacteraemias, coagulase-negative staphylococci, followed by Staphylococcus aureus, were the most frequent aetiological agents. Staphylococcus aureus was 2.5 times more common for bacteraemias associated with PB than for those associated with CB. This is congruent with another published studies. Akihiro Sato et al.  described 62 peripheral venous catheter related bloodstream infeccions. Gram positive microorganism were responsable of 58% (S aureus 17%) peripheral venous catheter related bacteremias (in our study S aureus is responsable of 23% of PB). In different studies a mean of 38% (range, 12–64%) of S. aureus catheter related bacteremias were related to peripheral venous catheter and a mean of 19% (7.6–35%) of S. aureus bacteremias were due to infected peripheral venous catheter .
About gram negatives PBs, Ripa el al  described that gram negative microrganisms were responsible of 22.8% of peripheral venous catheter related bloodstream infections. In our study is next to 19%. And in the study of Tsuboi M et al  gram negative rods were more frequently identified in peripheral venous catheter related bloodstream infections than in central ones (33%-18.8%) . In this study peripheral catheters were regularly replaced at least once every 96 hours and it could explain a relative higher amount of gram negative related bacteremias in front of longer peripheral catheter duration. In our study gram negative bloodstream infection are also more frequent in peripheral venous catheter related infections than in central ones ( 18.7% and 6. 3%). In our study 50% of PBs were related to more than 96 hours placed peripheral venous catheter, and it could explain a lowest relative gram negative bloodstream infections in advance of gram positives. There are studies that relates peripheral catheter dwell time and risk of S. aureus bacteremia. A study found a significant higher risk of S aureus bacteremias related to peripheral venous catheter if the median dwell time was over 3 days .. In this sense a study of 137 S. aureus peripheral catheter related bacteremias noted that 45% involved peripheral venous catheter in situ beyond 4 days and that 61% had been inserted by the ambulance service or in the emergency department .
The reason for this association is not clearly defined, although it may be related to the modification of the cutaneous microflora as the days of hospitalisation are extended as well as to simultaneous antibiotic therapy that the patient receives . In this sense, it is also striking that 15% of PBs and 8% of CBs were caused by enterobacteria, which should be taken into account when planning empirical antibiotic treatment.
In terms of the inpatient units where bacteraemias occurred, those associated with peripheral catheters mainly occurred in internal medicine inpatient units, while bacteraemias associated with central venous catheters occurred in those specialised units where interventions predominate or prolonged parenteral treatment is required, as is the case for ICUs, oncology, or haematology. In this respect, we observed that active neoplasia was present in 70. 4% of the patients with central venous catheter-related bacteraemias.
In general, patients with catheter-related bacteraemias have higher central catheter dwell times and prolonged hospital stays compared with patients with bacteraemias associated with a peripheral catheter.
The clinical status of the patients who suffer from a bloodstream infection associated with a catheter is similar for PB and CB, as we showed in our study, in which both groups had similar Pitt scores (0.89 in CB and 0.84 in PB) without differences in arterial blood pressure or creatinine, urea, and haematological values at the time of bacteraemia.
As we mentioned before, 53.9% of the CB patients and 62.5% of the PB patients received appropriate empirical antimicrobial treatment. The percentage of adequate empirical treatments was lower for CB than for PB, a nearly 10% difference with no statistical significance, probably due to the lower number of peripheral venous catheter-related bacteraemia cases. These cases, nevertheless, demonstrated survival rates close to 90% (CB, 91.1% and PB, 90. 2%).
Adequate empirical antibiotic treatment is a well-defined factor in bacteraemic survival. Therefore, the question is how it is possible that patients with different proportions of adequate empirical antibiotic treatment (CB, 53.9% and PB, 62.5%) have similar survival rates.
And in this sense in our study CB have more Hospital stay, active neoplasia, esteroids and others inmunosupressant treatments than PB. CB have also longer time to receive specific adequate treatment than PB. And CB are more frequently diagnosed in Intensive Care Units, Surgical departments, Haematology and Oncology deparments . But although all these ideas could make us to think about a worse prognosis in CB, they have the same prognosis than PB. The probable answer is that survival rates are influenced by several factors. The first is that at the time of bacteraemia, both groups had similar Pitt scores, without differences in blood test parameters such as creatinine, urea, or haematological values. The second factor is that microorganisms, such as coagulase-negative staphylococci, were more frequent in CB, at 141 (64.1%), than in PB, at 26 (40.6%). This factor could explain a better prognosis for the CB patients, even though they received fewer adequate empirical antibiotic treatments. In this sense there is a tendency toward higher mortality among patients with PB and who did not received adequate empiric therapy . It may be in part explained by the higher frequency of S. aureus in PB. It is described that PB due to S. aureus are significantly more likely to have a metastatic focus of infection , and greater mortality  and a significantly duration bacteremia compared to cases of non-PVC-related S. aureus bacteremia [23, 28].
And the third factor is catheter removal and the proportion of early catheter removal.that is probably one of the most important factors for outcome of catheter-related bloodstream infection.
About this point it would be interesting to know the number of days since the bacteremia to catheter removal but this item was not recordered.
Establishing a therapeutic recommendation thus requires further studies as well as preventive measures to reduce the incidence of bacteraemia associated with this type of device, which is so frequently used in hospitalised patients [15, 23].
About anothers limitations of the study , the first one is the retrospective nature of the study.
During the study period there were 101,690 184 hospital admissions with 185 609,686 hospital stays. It would be of interest knowing how many patients had central lines in this period, and not only in the bacteremic patients. Usually in the Hospital the most part of patients use peripheral lines and we only use central ones if long periods of time are needed for antibiotic tratment, chemotherapy, parenteral nutrition, or incapacity to use peripheral venous lines.
We found that the mortality was higher in patients who received inadequate empirical antibiotic, particularly in PB patients. It would be of interest knowing if all the patients were treated adequately thereafter (antibiotics adjusted according to susceptibility of the organism), how long were they treated if there was any persistent bacteremia or any resistant organism. Also it would have been of interest to know how was the catheter managed not only if it was retired or not . Finally as we have mentioned before it would be of interest knowing the number of days since the bacteremia to catheter removal because early catheter removal is probably one of the most important factors for outcome of catheter-related bloodstream infection.The study group with 285 catheter-related bacteraemia patients may seem relatively small and highly unbalanced, with 65 PB and 220 CB bacteraemia patients; however, this is a populational study, which means that variables with clinical and statistical differences are very strong from a statistical point of view. The proportion is also congruent with that in the EPINE 2016 study .