In our cohort of 238 patients with intestinal atresia or gastroschisis receiving a central venous catheter, the overall incidence of CLABSI was 35%. Out of the patients treated for intestinal atresia, 30% developed a CLABSI and 46% of the gastroschisis patients developed CLABSI. A catheter related thrombosis developed in 7% of all patients. In those treated for intestinal atresia 7% developed thrombosis whilst the same occurred in 6% of the patients treated for gastroschisis. Patients experiencing CLABSI stayed in the hospital significantly longer than those that did not. Moreover, children treated with an enterostomy and non-tunnelled catheters were significantly more at risk of CLABSI development Also, catheter dwell time was shorter and CLABSIs developed faster in non-tunnelled compared to tunnelled catheters. A catheter related thrombosis occurred significantly more often in non-tunnelled catheters, whilst there was no correlation with direct insertion into a central vein or peripheral insertion.
Previous studies in a general neonatal population showed that CLABSIs lengthen duration of hospital stay, increase costs and lead to severe complications including mortality [12]. Yet, there are just two studies that evaluated CLABSI and catheter related thrombosis as a primary outcome in neonates treated for intestinal atresia and gastroschisis. These studies retrieved their data from national databases using coding systems such as ICD-10. These studies report a CLABSI occurrence between 2–4% for those treated for gastroschisis, which is much lower than our cohort [6, 7]. Other studies report a higher occurrence, namely 18% in patients treated for intestinal atresia and 13% in children treated for gastroschisis [13, 14]. These studies were single centre retrospective cohort studies that reported on CLABSI as secondary outcome and which collected the data manually. National databases might suffer from underreporting of CLABSI and possibly coding errors since the care for patients suffering from these congenital birth defects is complex and complications are common. It is for instance remarkable that the database study reports a high CLABSI-rate per 1000-line days of 24 in patients treated for gastroschisis, whilst the numerator of this fraction (the incidence of CLABSI) is comparatively low (2%) and the denominator (catheter dwell time) is similar to this study (25 days). Other explanation for our higher incidence might be the definition used for CLABSI. The official CDC definition for CLABSI in young patients demands two separate positive blood cultures in case of a common commensal organism such as CNS, next to signs of infection and no other site of infection than the central venous catheter [11]. At our hospital, a CLABSI is presumed following a single positive culture combined with clinical suspicion of CLABSI in case of no other cause of infection. In case of a common commensal positive blood culture, antibiotic treatment is mostly started without central venous catheter removal. When clinical deterioration occurs, the central venous catheter is eventually removed. If these common commensal positive cultures were caused by contamination instead of a real CLABSI, patients could be expected to improve significantly more without removal of the catheter. Yet, the rate of central venous catheter removal was similar when comparing the CLABSIs caused by common commensal and other positive cultures. Still, it might be that we over-diagnosed patients with CLABSI leading to catheter removal in patients that might have improved by watchful waiting which would be a form of action bias. Better adherence to the existent guidelines on pediatric parenteral nutrition might both decrease central venous catheter associated complications as overly cautious removal of the inserted central venous catheter[15]. Furthermore, prevention could maybe furthermore be realized by the usage of line-lock infusions. For instance, the usage of taurolidine has been shown to decrease the rate of CLABSIs in children who receive TPV at home and is therefore included in the guidelines for this specific cohort[15]. For the patients concerning our study, a well-designed study should be started to investigate the efficacy and safety of different line locks including taurolidine in preventing central line infections.
Previous studies described complex gastroschisis as risk factor and, interestingly, prematurity and low birthweight as protective for CLABSI in patients treated for gastroschisis[6, 7]. They suggested that this could be related to a longer NICU submission in premature and/or low birthweight infants which, due to CLABSI prevention protocols in place at the NICU, could lead to a decrease in CLABSIs. In our cohort, we could not confirm these risk factors. Therefore, it seems that, as the previous studies too suggested, these factors are not general risk factors but rather local ones differing from hospital to hospital. Their influences might be limited if all hospitals would adhere to the same guidelines. Studies in neonates and PICU patients in general described male sex, higher birthweight and centrally placed catheters as risk factors which we could also not confirm [4, 16].
The increased risk of CLABSI in patients who are treated by enterostomy formation, which in this cohort mostly consisted of patients with an intestinal atresia receiving an ileo- or jejunostomy, might be partly related to the risk of high-output enterostomies which occurs in 50–60% of these patients [17, 18]. This in turn could lead to malnutrition which will make these patients more susceptible for systemic infections originating from bacteria which, particularly in neonatal ill patients with intestinal obstructions, seem to translocate from the gut [19, 20] Since the amount of stoma output was not well described in our cohort, we could not evaluate the influence of high-output enterostomies on CLABSI.
When it comes to CLABSI prevention in our cohort, we show that tunnelled central venous catheters seem to be preferred over non-tunnelled central venous catheters most specifically when a long duration of parenteral nutrition is expected. However, previous studies show conflicting results. Some show that dwell time does not affect CLABSI-risk and that it occurs less often in non-tunnelled central venous catheters in NICU patients in general [21]. We focussed specifically on a cohort of patients with congenital gastrointestinal diseases, which necessitates long-term parenteral nutrition. In the previous study in NICU patients, almost 50% of the non-tunnelled catheters were removed without complications within a week of dwell time, whilst this was the case in nearly 10% of the tunnelled catheters [21]. It thus seems that in this study patients who were less prone to long-term feeding difficulties were more likely to be treated by a non-tunnelled catheter. This could have resulted in selection bias which in turn could explain the difference with our cohort. Although some form of selection bias might have occurred in our cohort based on the patient’s clinical status or the preference of the treating physician, it seems plausible that this bias is more profound in studies including multiple diseases.
The central venous catheter related thrombosis incidence in patients treated for intestinal atresia was found to be 8% in a previous study including 47 patients treated for jejunoileal atresia, which is similar to our results [13]. To our knowledge, there is no study describing the incidence of this complication in patients treated for gastroschisis. Our incidences are lower than the 6–30% reported in cohorts of children up to 16–18 years of age. This could be caused by the amount, commonly near 20%, of patients treated for cancer in these cohorts as well as older age. Treatment for cancer is in itself a risk factor for thrombosis [3, 22]. Furthermore, the inclusion of non-symptomatic thrombosis in these studies will have led to an increase in incidence whilst the clinical significance of these thrombosis is debatable. Still, it seems that this complication is not uncommon. It might be that lowering the threshold for giving prophylactic low molecular weight heparin, which at the moment is set at an expected catheter dwell time of six weeks in our clinic, might decrease the occurrence of thrombosis specifically in this cohort. However, the efficacy and safety of prophylactic low molecular weight heparin should be assessed in a well-designed trial in this population.
In older children, up to 18 years of age, it has been suggested that catheters inserted directly into a central vein decrease the risk of symptomatic thrombosis due to the comparatively larger lumen of the central veins [3]. In our cohort this hypothesis could not be proven.
Although the catheter was removed in all patients who experienced thrombosis, some suggest that spontaneous regression of the thrombosis within 28 days can occur in 50% of the children [23]. For this reason, an ongoing multicentre Dutch prospective observational cohort study (Trial registration: Dutch Trial Register. Registered 24 December 2013), which protocol has been published in 2018, aims to evaluate a new treatment strategy. In this protocol non-occlusive thrombosis in neonates admitted to the NICU is treated by “watchful waiting”. In case of non-extension after 5 days, a thrombosis will be followed-up by ultrasound without anticoagulant treatment [23].
This study could not evaluate compliance to preventative measures (such as hand hygiene and skin preparation) which are in place to prevent CLABSI, due to the retrospective design [24]. This is a limitation of our study. Compliance to these measures has shown to able to decrease the incidence of CLABSI in neonates [25, 26]. Secondly, as explained previously, the definitions used for CLABSI might have resulted in different outcomes as would have been the case for catheter related thrombosis if we would have included non-symptomatic thrombosis. In both cases we tried to use a definition which resulted in the clinically most relevant results. If we would have followed the official CDC definition for CLABSI more strictly our results would not have been a true reflection of the number of patients treated due to a presumed CLABSI. Although incidence might have differed, our sensitivity analysis showed that the same risk factors would have been identified if we had chosen to only include CLABSIs that led to catheter removal. The definition for catheter related thrombosis could have been widened by including non-symptomatic thrombosis. This would have increased our incidence, though presumably only slightly due to the rarity of non-symptomatic thrombosis in this cohort. Yet, the clinical relevance of non-symptomatic thrombosis is debatable which is why we chose not to include them. Lastly, some factors, such as catheter size, which specialist inserted the catheter, experience of the specialist, were not clearly described and could thus not be evaluated. Taking these limitations into account when interpreting our results, we still were able to approach the incidence and evaluate risk factors in a specified cohort which increases the interpretability of our results compared to studies including multiple age groups or diseases.