In Pediatric Intensive Care Units (PICU), the prevalence of nosocomial infection (NI) is variable, ranging from 10 to 33%1,2, and incidence rates from 5.9 to 15 per 1,000 patient-days3,4. This is explained by the greater complexity of these patients affected by a variety of medical and surgical pathologies1.

NI is also considered one of the most costly problems for hospitals. The most important direct NI-associated cost comes from the increase in the excess of length of stay (LOS)5. Therefore, most reports focus on NI-related costs due to LOS, as it is quite easy to translate such costs to currency value, applying the local cost per hospitalization day6. Further studies also have taken into account the costs of diagnostic and medical procedures, derived from or due to the infection, as the most accurate way to calculate the NI-attributable cost7–10.

The study of NI-associated cost has evolved in time with the introduction of new cost estimation methods11–13. The most common comparative method used to analyse the cost of NI, is the prospective or retrospective observational study of the hospital environment14.

Few studies have assessed the impact of NI on total costs in PICU15, some of them have analysed the cost attributable to any type of NI in PICU7,11,12,16, others have focused only on nosocomial primary bloodstream infection8,9. The average cost attributable to NI described in literature ranges between €1,51017-53,0859 per patient.

The Spanish Public Healthcare System is a tax-financed system that offers universal coverage, which is different from privately-funded healthcare systems, where most of the published reports on NI-related cost have been performed5,7. The lack of previous Spanish reports about NI costs in PICU and the actual economical crisis in this country, justifies the present paper. The aim of the study is to estimate the cost attributable to nosocomial infection in a pediatric intensive care unit and its related factors.

The number of patients enrolled in the study was 443 (57% were boys), with a median age of 1.5 years (percentile 25: 5.3 months- percentile 75: 6.2 years). The cause of admission was surgery in 58.5% of patients, being cardiovascular surgery the most common intervention (46.33%). Global mortality in the Unit was 6.3%. The most frequent procedures were: use of antibiotics, central venous catheterization and invasive mechanical ventilation (85%, 64% and 55%, respectively). Fifty patients suffered, at least, one episode of NI, which means a prevalence of patients with NI of 11.3%. The total number of NIs was 82 in 443 patients (14.71%). Table 2 shows the number of infections according to site and infection rates which ranged between 7.9% (bloodstream infections incidence) to 1.4% (other infections). 60.6% of the infected patients had a single episode of infection, 18.2% had two infections, and 16.7% had three infections. Two patients had up to 4 infections during admission to the Unit.

Non significant differences were found between patients who had NI and those who didn’t according to gender, patient origin, presence of malformation and reason for admission. On the contrary, those with a younger age had a higher risk of NI in comparison to those who were older (Table 3).

Table 4 shows the relationship between the total cost and independent variables for the patients admitted in the PICU. It highlights the higher median cost in children with less than 2 months of age (€6,903.30) and the difference in median cost between bloodstream infection and pneumonia (€37,356.68 and €34,912.96, respectively). Furthermore, the cost increase was proportional to the number of NI diagnosed, with a difference in median cost between having a single infection and 3 or 4 infections of €38,202.57.

Table 5 shows the differences between infection and non-infection regarding cost. It is worth noting that about 70% of the total cost was due to LOS expenses. In this sense, median LOS of patients with NI was 24 days versus 3 days for those without NI (p<0.001). The difference in the median cost of LOS in NI-patients, was statistically different comparing it to those without NI (€18,695.35 and €2,386.64, respectively). Regarding the total cost of the patients with NI included, a difference in median cost of €20,947.94 (€30,791.41 average cost) was observed between patients with and without NI (p<0.001).

The multivariate analysis shows that those with bloodstream infections (p<0.001) and pneumonia (p= 0.008) have a greater cost than those without NI. It is worth noting that a higher cost in PICU was associated with cardiovascular surgery (p<0.001), urgent type of admission (p= 0.004), higher Pediatric Risk of Mortality index (p<0.001) and the presence of inmunosuppression (p<0.001). By contrast, older children and surgical type of admission (p= 0.009) presented a lower cost (Table 6).

The analysis of NI-related costs started in countries with privately-funded health care systems, followed by others with a hybrid system of private and public insurance, like Spain10,24–26. In these studies, the type of patients included and the healthcare systems where they were performed were very different, which hinders any possible comparison. However, all of them report a remarkable cost associated to NI7.

In the present study, estimated costs have been based upon prolonged PICU LOS and variable costs of different medical procedures, the latter being a direct attribution resulting from the NI presence, which is a better estimation of attributable costs. In this sense, Pittoni and Scatto27 remark the importance of taking into account direct variable costs to understand the real costs generated from a patient admitted in the intensive care unit, including non complex medical procedures as they may affect the result of cost analysis in an accumulative way.

The authors have performed a multivariate analysis using multiple linear regression models in order to assess factors associated with the total cost, as it is difficult to choose the matching factors between cases and controls without committing a selection bias. This follows the current trend in NI-related cost analysis (from 1990 to 1999, 14.3% of the studies performed multiple linear regression and from 2000-2006, 34.5% did so28).

In this study, we observed a difference of 21 days in PICU LOS between patients with and without NI. Although, the authors did not adjust by other variables, there was a difference in costs between both groups of more than €30,000. In spite of being statistically significant, it may not be exclusively due to NI. This analysis, although it has been the most frequently used in studies of costs for NI, has limitations, as it results from an average measure. Also, it does not reflect the real expenditure, considering that this is not uniform throughout the stay, and that, as the NI is time-dependent, it should measure the extension of the stay only after its diagnosis, and not for the entire stay12. Nevertheless, other studies carried out in PICU, despite of using a different methodology, and adjusted due to other factors like patients’ age or cause of PICU admission, show similar results, like, Slonim et al.8, who found a difference of 14 days, or Elward et al.9, with a mean hospital LOS cost of €24,619 in patients with NI.

It also must be taken into account the association between LOS and NI, which may be contradictory, since, although it is recognized by the literature that NI increases hospital stay29, some authors also consider LOS as a risk factor for the NI30 (longer hospitalizations leading to colonization of patients by pathogenic microorganisms, longer duration of invasive procedures and pharmacotherapy which favours the development of NI). However, the cost study presented here was part of a prospective study which also analyzed the risk factors associated with NI, not finding a significant association between LOS and NI. Therefore, in this paper, the authors have assumed that NI has influenced the increase in LOS, considering what referred by the literature and the results of the study for risk factors of NI.

Average cost of medical procedures in the current study was higher in the case of NI patients when compared with non-infected patients, where the use of antibiotherapy, feeding and imaging tests should be highlighted. In the case of bloodstream infection, the cost of catheters and antibiotics doubled the observed in patients with pneumonia, as in Pittoni and Scatto's case27, which remarks that one of the most difficult challenges is the bloodstream infection management, because of its wide array of possible treatments and elevated associated cost.

Globally, the cost of NI is within the range described in other studies performed in PICU5,7,9,11,17,31 (from €1,51017 to €53,0859). The last study showed higher costs, which may, in part, be justified by the use of a cost-estimation method based upon cost to charge ratios and the inclusion of more complex patients. The BI was the most costly NI, showing significant difference when compared to pneumonia, which it agrees with other studies5,32.

Nevertheless, the results from other studies which only report average values must be interpreted cautiously. Their costs estimations were obtained in different countries with a healthcare management which may not be equivalent to the Spanish Health System. Likewise, other costs should be considered, such as those derived from patients who die as a result of the NI; the cost of the medical procedures required by the treatment of NI beyond 48hours after discharge from PICU, or the indirect or intangible costs referred to productivity loss, and more specifically what the patient loses as a result of the infection and/or treatment (pain, suffering, loss of leisure time), which affects their quality of life.

Some of the variables that were associated with an increase of total cost (presence of NI, emergency admission, previous cardiovascular surgery and immunosuppression), were also been reported by other studies9,16,29. Other factors that may also increase the cost of the Unit were the cause of non surgical admission and comorbidities such as severe lesions in primary organs7,9,12,33. In the present study, only immunosuppression, as a patient's intrinsic factor, contributed in a statistically significant way to the model. In this sense, Gemke et al.34 stated that non previously programmed admissions did increase the use of resources and, consequently, total costs.

Another factor included in our model as a risk factor is cardiovascular surgery, which could be due to the high number of patients undergoing such surgery in the PICU, postoperative LOS, frequent infectious complications and prolonged use of medical procedures, which means an increase of total economic costs. The inclusion of this factor is supported by a recent report published by González et al.35, which states that prolonged admissions convey a greater use of medical resources, higher risk of NI and higher mortality rates among children admitted to PICU35. Other studies mention hospital stay1,14 and non survival in the Unit7,9,24 as factors capable of rising costs. These variables were only statistically significant in the bivariate analysis.

In conclusion, this paper has approached to NI-related economic cost, where most of the expenses are derived from the medical procedures made to the patient during PICU stay and its complications. The higher total cost per patient requires the implementation of specific cost-effective preventive measures to control NI, in order to minimize its effects in the health of pediatric patients and in medical costs.

Miguel Ángel Negrín

The project's original idea was provided by R. López. A. Morillo and JM Aldana adapted the idea for the research project and its implementation in a Pediatric Intensive Care Unit. R. Valencia developed the project's methods, and performed the database and its depuration. The clinical data collection and the attribution of medical procedure-related costs to nosocomial infection were made by A. Morillo and R. López. A. Olry performed the economic analysis. M. Conde and M. Loscertales requested the costs information from the Departments of Pharmacology and Economic Management. A. Morillo wrote the first draft of the article, and was subsequently reviewed and approved by all authors.

This project has been funded by the Andalusian Regional Ministry of Health.

None