Very low birth weight (VLBW) infants (birth weight ≤ 1500g) are particularly vulnerable to nosocomial infections. The immaturity of their immune system, the fragility of their skin and the multiple life support procedures used in their care are among the factors associated with the high incidence of these infections and the associated morbidity and mortality.1,2 Bloodstream infections (BSIs) are the most frequent nosocomial infections in neonatal intensive care units (NICUs). Epidemiologic surveillance systems may contribute to the prevention of infection through the continuous assessment of its incidence and the associated risk factors.3–5 Such systems must regularly produce standardised indicators for self-assessment of health care quality and benchmarking among participating NICUs.6–8 In addition, the implementation and management of a surveillance system requires a sound methodology and an effective organisational strategy to obtain accurate and reliable data on an ongoing basis for the purpose of assessing the impact of newly introduced preventive and quality improvement measures.9,10

In Europe, the German National Reference Center for Surveillance of Nosocomial Infections, which has implemented a specific system for VLBW infants (NEO-KISS), has pioneered the epidemiological surveillance of BSI.5 A decrease in the incidence of BSI was observed in the early years of operation of this system in Germany.4 During this time, there was no nationwide system for prospective data collection that would have allowed Spanish NICUs to document and track BSIs in VLBW infants. This posed challenges to the standardised measurement of incidence rates and of associated risk factors, especially those that are actually modifiable (catheter and antibiotic use).

The purpose of this article was to describe our experience in the implementation of the Spanish NeoKissEs system for the surveillance of BSIs in VLBW infants. NeoKissEs is based on the German NEO-KISS, whose definitions and methodology have been validated and are currently applied in some of the surveillance protocols of the European Centre for Disease Prevention and Control.11

The implementation of NeoKissEs started with the collaboration of professionals from 50 NICUs from 15 administrative regions in Spain (Figure 2). Eight workshops were held to train implementation teams on the management of the system and attended by a total of 210 health professionals (neonatologists and nurses).

Figure 2.

Location of the neonatal intensive care units (NICUs) participating in the NeoKissEs surveillance system in Spain.

(0.21MB).

The 45 units that ultimately started the local implementation process received a questionnaire asking about the details of the implementation, the perceived barriers and the strategies used to address them. The survey response rate was 71.1% (32 units). The strategies used to inform the staff were: 1) informative meetings for the entire NICU staff; 2) series of specific meetings for each type of professional on staff; and 3) meetings attended only by staff that expressed an interest in participating in the initiative. Twelve champion stated that in spite of their efforts to inform and motivate their colleagues, they ended up being the only staff directly involved in data collection and quality assurance. The problems reported most frequently by facilitators as having a negative impact on the implementation were: difficulty in establishing the NeoKissEs team due to lack of motivation among the staff, difficulty ensuring knowledge of the initiative by all NICU staff, a high turnover rate in the NICU, especially among nurses, and poor collaboration between nurses and physicians.

Several facilitators felt that the work of nurses in data collection for NeoKissEs would benefit from the support of professionals from outside the unit experienced in infection prevention and control, yet only two NICUs reported an ongoing involvement of preventive medicine colleagues in infection assessment and control in close cooperation with the nurse and neonatologist. This collaboration was greatly valued in both instances.

By the beginning of 2015, the NeoKissEs website was fully operational, allowing for direct data entry and hosting a collection of streaming video tutorials developed to resolve possible questions concerning the correct registration of cases and BSI episodes and to explain how to produce reports. The data recorded to that point were transferred to definitive electronic case report forms.

Real-time statistical reports could be obtained through the website to provide feedback to the champion. These reports offered: 1) general patient information, such as dates of admission, birth weights, etc, 2) statistics for their particular unit (rates of infection and device use) and 3) comparative analyses of their NICU and other units based on the summary statistics calculated by the system.

Participating NICUs gradually started to recruit patients, depending on the date that the facilitators attended the training and the time required to set up and launch the system in their hospitals (Figure 3). Of the 50 NICUs that initially joined the system, 45 had registered cases by December 31, 2015 (Figure 1). NeoKissEs continues to gather data at the time of this writing. All participating NICUs were classified as level III based on the level of care they offered: Level III-A (15 NICUs, 660 infants), level III-B (23 NICUs, 1863 infants) and level III-C (7 NICUs, 1115 infants).14

Figure 3.

Follow-up period in the NeoKissEs surveillance system by administrative region in Spain.

(0.22MB).

During the pilot phase, the two participating NICUs included 260 VLBW infants that were monitored for a total of 9324 days; of these infants, 81 (31.2%) developed CABSIs, with an overall incidence rate of 19.6 CABSIs per 1000 catheter-days (95% CI, 15.5-24.0). From the time that NICU recruitment started nationwide (February 8, 2013) through the end of 2015, 3750 infants were registered in the system, of who 112 (3.0%) were excluded from the analysis because they did not meet the inclusion criteria.

Table 1 summarises the perinatal characteristics of the 3638 infants that met the inclusion criteria. Overall, 966 VLBW infants (26.6%) had at least one episode of BSI during the followup in the system. Compared to VLBW infants without BSI, those with BSI had been born at lower gestational ages with lower birth weights and had a higher mortality during the surveillance period (relative risk, 1.8; 95% CI, 1.4-2.3). Furthermore, the device use rates (for vascular catheters and antibiotics) and length of stay recorded in the system were higher in this group (Table 1).

A total of 1108 episodes of CABSI were registered, corresponding to an incidence rate of 18.4 BSIs per 1000 catheter days (95% CI, 17.8-19.1). We found considerable variability in the CABSI rate between the 45 NICUs that registered cases during the study period (Figure 4). In the analysis by level of care, CABSI rates ranged from 4.7 to 34.9 in class III-A units, from 3.9 to 29.0 in class III-B units, and from 18.6 to 34.0 in class III-C units.

Figure 4.

Catheter-associated bloodstream infection (CABSI) rates in participating neonatal intensive care units (NICUs) by level of care.

(0.15MB).

Our results reflect the successful implementation in a substantial number of Spanish NICUs of the German system (NEO-KISS) validated for the surveillance of nosocomial BSIs in the VLBW infant population. Our study also produced initial data on standardised indicators (statistics on BSI, CABSI, and antibiotic and device use) and identify significant challenges and barriers that need to be addressed in the implementation and management of the surveillance system and of potential interventions for quality improvement.

Previous research has shown that the implementation of a BSI surveillance system specifically focused on VLBW infants may help reduce the impact of sepsis in NICUs.4 This positive effect may be due to the systematic collection of data through specific methods and the application of standardised definitions that facilitate the analysis and interpretation of statistics, the comparison between NICUs and feedback to clinicians.4,10,18

Our experience demonstrates that it is possible to adapt and implement a prospective surveillance system for BSI in the population of VLBW infants admitted to Spanish NICUs. With its standardised procedures and outcomes, NeoKissEs may well become the Spanish reference system for the collection and use of BSI data in VLBW infants for benchmarking and quality improvement purposes. The data from NeoKissEs supplements the information produced since 1994 by the Neonatal Network of the Castrillo Hospital Group in Spain,19 whose data collection is restricted to microbiologically-confirmed cases of neonatal BSI, but periodically provides useful reports on the antibiotic susceptibility of the isolated microorganisms. Our initial cumulative incidence (26.6%) was only slightly higher than the one described by the Castrillo Group (21.3%) (Fernández-Colomer B, Coto-Cotallo G.D. 27 October 2017, unpublished raw data) and in early reports of the German NEO-KISS system (21.3%).5 On the other hand, the BSI incidence rate per 1000 patient days and CABSI rate per 1000 catheter days found by NeoKissEs are much higher compared to the earliest benchmarking dataset obtained from the German system (9.2 vs 6.5 and 18.4 vs 9.2, respectively).5 This gap was even larger when we compared our data to the NEO-KISS values for the same time period (2013-2015) (3.8 BSI episodes per 1000 patient days and 6.5 CABSI episodes per 1000 catheter days) (Piening B, 24 November 2017, unpublished raw data), which highlights the role of NeoKissEs as a useful tool to stimulate research on potentially modifiable risk factors.

Our preliminary findings also brought to light other issues that need considering. First of all, we found considerable heterogeneity in the approaches used to implement the surveillance system at each site, despite the general consensus achieved in the implementation workshops. Some NICUs reported difficulties forming multidisciplinary teams to direct the implementation process, ending with a single person being responsible for all data collection activities and the impeding independent data validation at the NICU level. Further, accurate measurement and reporting is of critical importance but of limited effect if not accompanied by organised efforts to address the main factors associated with an increased risk of BSI. The reported difficulties in maintaining stable implementation teams could be a major barrier to these units benefitting from benchmarking information to drive the design and implementation of better practices.20

Secondly, the responses to the survey showed that there was substantial variability in the resources available in the NICUs, their organisational characteristics, their caseloads, staffing, levels of motivation, reward systems and quality assurance policies. All these factors should be taken into account in the near future, when, as we hope, NeoKissEs becomes the national framework for the introduction and evaluation of evidence-based best practices and quality improvement initiatives focused on the prevention and control of BSI in VLBW infants.

Thirdly, the huge variability observed so far in the frequency of BSI episodes, even in NICUs with the same level of care (Figure 4), suggests the existence of potentially modifiable health care-related factors. This calls for in-depth examination of these factors and the development of an appropriate framework for targeted actions aimed at reducing the rates of preventable infection in the future. Many NeoKissEs units are currently participating in a new interventional study to test the systematic implementation of a set of bundles of measures concerning the insertion and maintenance of central vascular catheters in VLBW infants with the aim of reducing CABSI rates. NeoKissEs will continue to support the collection of BSI statistics for these units and will be used to assess the impact of the bundle interventions.

The creation of a web-based electronic support system has greatly facilitated the consecutive registration of patients and BSI episodes and the monitoring of data quality. It allows NICUs to consult their statistics, for their own unit or in comparison with others, at any time. It also helps identify areas for improvement by comparing their results with the overall data provided by NeoKissEs and the data from the German NEO-KISS. Comparisons may be biased if there is heterogeneity in the compared populations (case mix bias).21 To address this, the system carries out adjusted analyses, providing comparisons that are more valid by taking into account factors such as birth weight, gestational age and the complexity of the provided care.

The number of NICUs that participate in NeoKissEs ensures a sufficiently large sample of individuals and BSI episodes to provide accurate and reliable statistics and facilitate the performance of studies of appropriate scope to evaluate the safety and effectiveness of specific interventions and policies. That said, expanding the NeoKissEs system would make the data more representative of the whole of Spain. There is the limitation that NeoKissEs was developed in the framework of a research project funded for a limited period of time and that it has yet to be permanently integrated into the Spanish National Health System, unlike NEO-KISS, in which participation has been mandatory for all German neonatal units since 2006. It would also be advisable for NeoKissEs to establish collaborations with other similar initiatives, registries and neonatal networks to share data, expand benchmarking activities and encourage the systematic analysis of the risks associated with the development of BSI in this population.

NeoKissEs has been implemented in 45 Spanish NICUs and has proven to be a useful source of information on the incidence of BSI and important associated factors. This standardised system is easy to use and widely accepted by users, and provides standardised definitions, procedures and data collection instruments that allow comparative assessments. We believe that this system would be a good tool for evaluating ongoing interventions aimed at improving health care quality and to promote the implementation of safe practices in participating NICUs, individually or as a group. It would no doubt benefit from being used by a larger number of NICUs, greater integration in the daily activity of health care staff and sustained institutional support.

This project was funded by the Carlos III Research Institute through grant PI13/00587 and by the Red SAMID through grant RD12/0026 (both co-funded by the European Regional Development Fund/European Social Fund).

MM was the main author responsible for writing this article and its critical edition, and also contributed to the implementation of the surveillance system.

MCL was the neonatologist responsible for the NeoKissEs surveillance system and its implementation. She collaborated in drafting the manuscript.

JP collaborated in writing the manuscript, producing tables and graphs and organising the final version.

JE worked on the development and analysis of databases.

ES collaborated in the implementation of the surveillance system, and helped write the methods section of the manuscript.

BP made substantial contributions to the adaptation and interpretation of the definitions and instruments of the original German NEO-KISS. He was involved in the comparative analysis of NeoKissEs indicators with the German NEO-KISS reference population.

JIV was the principal investigator of the NeoKissEs system (PI13/00587) and led the implementation of the surveillance system. He collaborated in the writing of the manuscript.

JIP was responsible for analysing and interpreting the data obtained through the NeoKissEs system. He designed the methodology of the study and coordinated the writing of the manuscript.

All authors participated in the critical review of the manuscript and approved the final version.

The professionals of the NeoKissEs Group collected patient data in their respective hospitals for the surveillance system. They also participated in the completion of surveys and contributed to the accuracy of the final data by addressing the concerns brought up by the coordination team.