Keywords

1 Introduction

The improvement of obstetrical and neonatal management has enabled the survival of preterm infants; however, these corrective actions have inadvertently promoted an increased incidence of healthcare-associated infections (HAIs) which are contracted in the hospital setting and appear during hospitalization or soon after. Decembrino et al. argue HAIs result in prolonged hospital stays, increased hospital costs and are one of the major causes of morbidity and mortality in neonatal intensive care units (NICUs) [1]. Another Italian research has documented an incidence of these infections between 1 and 4 per 1000 live births in developed countries (7–19% in Europe, 14% in the USA), while this incidence results 6.5–38 per 1000 live births in developing countries [2]. In Italy, according to the National Institute of Health, infections occur in 5–8% of the hospitalized patients [3]. In NICUs the frequency of nosocomial infections is 7–24.5% [4] and increases up to about 40% in newborns weighing less than 1,000 grams and gestational age below 28 weeks, compared to 0.3–3% for healthy full-term babies. A previous study indicates 50% of infection cases are represented by sepsis, while 25% and 15% by respiratory and urinary tract infections, respectively [5].

Neonates result susceptible hosts when several conditions occur e.g. prematurity of organ systems, immaturity of immune system, presence of malformations, low birth weight, application of invasive devices and administration of antibiotics [6, 7]. Ghiradi [8] and Giuffré [9] have provided evidences the erroneous use of devices and antibiotics predispose neonates to colonization by the more resistant strains, especially certain Gram-negative bacteria, coagulase-negative staphylococci and yeasts. Previous studies indicate microorganisms’ transmission in NICUs is promoted moreover by direct contact (e.g. with personnel’s hands), other sources of infections (for instance air, water and food) and incubator’s microclimate, characterized by high humidity and hot-humid air recirculation, which could even encourage growth and multiplication of these microorganisms [10].

The literature on HAIs shows a variety of multidisciplinary approaches of infection surveillance which have as purpose the control of outbreaks, suggesting possible procedures to implement [1, 8].

The University Hospital “Federico II” of Naples implements a system (which considers and involves patients, microorganisms and environment) to monitor HAIs in NICUs. This system can provide several epidemiological data extrapolated by the monitoring of invasive procedures, the environmental microbiological sampling for the identification of sentinel pathogens, the analysis of microorganisms strains resistant to antibiotics and the molecular typing of bacteria which cause infections.

Simultaneously, the Strategic Management of the University Hospital is committed in the implementation of a purely business management tool, the Lean Six Sigma (LSS), to provide an effective framework for producing systematic innovation efforts in healthcare. Controlling healthcare cost increases, improving quality and providing better healthcare are some of the benefits of this approach [11].

In healthcare, LSS allows to improve the overall performance of a process identifying and eliminating the causes of possible deviation from its ideal standards [12].

Several publications have appeared in recent years documenting LSS implementation to introduce new clinical pathways in hip [13, 14], femur [15, 16] and knee surgeries [17], evaluating pathways utility according to some clinical variables, e.g. the reduction of the length of hospital stay (LOS). Montella and co-workers have proposed this methodology to improve HAIs control aiming at reducing the number of patients affected by infections in a surgery department [18].

To the authors’ knowledge, however, a study based on the implementation of the LSS methodology for the control of HAIs in a NICU is missing.

Therefore, the present work aims to identify the causes that can foster possible deviations of the neonates’ care process (namely, an HAIs risk increment) from its ideal standards as well as to determine the influence of each variation and the appropriate corrective actions to be taken. The integrated application of the LSS methodology with a tailored Define, Measure, Analyze, Improve, and Control (DMAIC) strategy can improve the performance of the care process in terms of incidence reduction of patients’ colonization by sentinel germs and, consequently, the risk of HAIs.

2 Methods

A tailored DMAIC strategy was adopted to study and analyze the data collected from the printed medical records and the informative system at the University Hospital “Federico II” of Naples throughout the year 2010 for 271 patients.

The tools implemented in each of the DMAIC phases are described in the next sections.

2.1 Define Phase

During this stage, a project charter and a Suppliers-Inputs-Process-Output-Customers (SIPOC) diagram were drafted. In the former, the problem to be solved was defined, the goals to achieve were clarified, the critical-to-quality (CTQ) characteristics were identified, and project time frame was fixed. In the latter, a synthetic and precise summary of suppliers, customers and process phases was drafted.

2.2 Measure Phase

During this stage, the study data were extrapolated from the hospital informative software QUANI (BIM Italia, Italy) and from the data stream for the monitoring of sentinel germs. The QUANI data provided the measurement of variables such as the number of procedures and the Diagnostic Related Group (DRG) classification; process control charts were used to preliminary check several data trends. The information extrapolated by the data stream allowed, instead, the demographic, nosological and epidemiological classification of the CTQ parameters.

CTQ was identified and represented for newborns with at least one positive bio-logical sample per sentinel germs (as reported by the Microbiology Operative Unit) using different types of specimens such as tracheal suctions, urine, celebrospinal fluid and blood.

2.3 Analyze Phase

The analyze stage continues the diagnosis and involves an identification of possible causal relationships between the inputs and the CTQ parameters.

Firstly, the data discussed in the previous section were analyzed using control charts, histograms and statistical tests. The last ones (namely, Chi Square tests) were carried out with Origin (OriginLab, USA) to evaluate possible correlations between variables (level of significance equal to 0.05), as summarized in Table 1.

Table 1. Statistical analyses on patients’ data. Specifically, with the term “DRG score” it was indicated the greatest integer less than or equal to the DRG classification.
Full size table

Secondly, several histograms were generated using Microsoft Excel for Microsoft 365 (Microsoft Corporation, USA). These correlate the percentage of colonization with the DRG score, the number of procedures and the LOS, respectively.

Moreover, a 5 Whys analysis and a questionnaire to the staff were developed by the Six Sigma team. The former was performed to understand the root causes which determine the infections caused by the sentinel germs. The latter was administrated to privilege witnesses (HAIs experts belonging to Hospital Infections Committee) to analyze both structural and organizational aspects of patients’ hospitalization and the professional behavior of NICU’s operators (physicians and nurses). The data extrapolated from the questionnaire were considered also in the 5 Whys analysis to find possible corrective actions to reduce or potentially eliminate HAIs risk.

3 Results

3.1 Define Phase

The project charter (Table 2) summarizes the information acquired during the meetings.

Table 2. Project charter.
Full size table

The team presented the results of the SIPOC analysis, summarized in Table 3.

Table 3. SIPOC analysis overview.
Full size table

3.2 Measure Phase

QUANI and microbiological surveillance data have shown 20 out of 271 patients (7.38%) were infected by several sentinel germs.

The first information focused in this study was patients’ LOS. Infected patients were hospitalized on average 59.85 days, while the non-infected patients 26.31 days.

Figure 1 plots LOS for both infected and non-infected patients.

Fig. 1.
figure 1

Diagram illustrating the LOS for the 271 patients.

Full size image

3.3 Analyze Phase

QUANI and microbiological surveillance data allowed a thorough understanding of the collected information. Firstly, the overall LOS data results (introduced in the measure phase) were summarized in the following whisker plot (Fig. 2).

Fig. 2.
figure 2

Whisker plot illustrating LOS for the overall 271 patients (left), the non-infected patients (center) and the infected patient (right). Vertical bars represent the confidence interval (95%).

Full size image

Secondly, samplings on infected newborns allowed the identification of the sentinel germs. Figure 3 illustrates a pie chart which summarizes the percentage amount of bacteria found in the 20 infected newborns. The main evidences of sentinel germs have been found from the tracheal aspirates of newborns (85% of the infected patients).

Fig. 3.
figure 3

Overview chart of sentinel germs found in biological samples.

Full size image

The 5 Whys analysis resulted in the 5 root causes summarized in Table 4.

Table 4. 5 Whys analysis output.
Full size table

Finally, the results of the chi square tests demonstrated the comparison between patients’ DRG score and their clinical status (infected/non-infected) did not show statistically significant results (p-value = 0.223). On the other hand, tests 2 and 3 (see Table 1) demonstrated statistically significant results (p-values of 0.034 and 0.018, respectively).

Plots showing raw data of such analyses are depicted in the following Fig. 4.

Fig. 4.
figure 4

Bubble plots illustrating trends of the raw data studied by the chi square statistical analyses (see Table 1). a) Test n°1. b) Test n°2; c) Test n°3.

Full size image

4 Discussion and Conclusions

This paper presents a study concerning HAIs for a population of 271 neonates in a NICU considering the criterions of the LSS methodology and a tailored version of the DMAIC problem-solving strategy. This guided to firstly define the issues and the aims of the research, then to carry out an analysis of the overall process (supported by the representation of the problems data with tools of visual management). Finally, the strategy helped to perform some statistical tests to characterize the indicators to consider to potentially raise the process performance.

The presented results show as possible enhancement the reduction of the mean LOS, as shown in Fig. 2 and 4c). In fact, the latter picture reports how patients hospitalized for 2–30 days presented a low colonization percentage (around 5%), while how several patients hospitalized for longer times presented a higher predisposition of colonization (5 to 10 times the former value). Another improvement indicated by the data analysis provides evidences that the reduction of procedures number could effectively reduce the risk of HAIs. Figure 4b) shows newborns who underwent a small number of procedures had a lower risk of colonization, while the highest number of infected (16/20) was neonates who underwent a higher number of procedures, maybe because previous ones have appeared inconclusive. Another implication suggested by the results concerns LOS and procedures together; it has been observed newborns whose preoperative LOS and hospitalization/first procedure LOS was reduced had a smaller risk of colonization; nevertheless, the results of the 5 whys analysis have demonstrated potential bottlenecks to be solved to consequentially guide an improvement from this point of view.

In contrast to some reports in the literature, many sentinel germs were found analyzing tracheal suctions samplings. These results do not agree with the findings of Zingg and Kumar who report bloodstream infections as the prevalent kind of infections in children [19, 20]. The most likely explanation of these conflicting results might be motivated by a smaller newborns population respect to the other two studies (2138 and 595, respectively). Nevertheless, it is still missing an international study reporting accurate cross-national point-prevalence surveys; currently, surveys are generally developed in different countries/areas according to non-standardized procedures. This limitation is pushing scientists to organize in-depth cross-national studies which could extend the knowledge in this field with greater clarity.

This paper presents a pilot study to methodologically find the necessary enhancements to be applied in NICUs to reduce HAIs risk. To the authors’ best knowledge, LSS have been used in the context of NICUs, for instance, to minimize the waste because of stocked supplies at the bedside cabinet [21] and, in combination with DMAIC, to promote the decrease of the incidence of intraventricular hemorrhage in newborns [22] and to reduce or eliminate the risk for errors in breast milk administration [23].

In conclusion, this pilot study showed the application of a tailored DMAIC approach focused on the reduction of the HAIs in a NICU. In this work, data of 271 patients were analyzed to identify possible corrective actions related to control several sentinel germs colonization. Since the HAIs are an important and sensitive theme in hospitals, further researches appear fully justified by the potential advantages for both hospitals and patients: these investigations could guide the clinical staff to improve the management of patients in NICUs reducing the number of infected newborns, their LOS and the costs for the hospital.

As an example, future developments could focus the analysis of the process considering process deviations characteristic for each single germ (and the relative improvements to bring).